University of Southern California Dissertations and Theses

A roadmap for changing student roadmaps: designing interventions that use future “me” to change academic outcomes

(USC Thesis Other)

A roadmap for changing student roadmaps: designing interventions that use future “me” to change academic outcomes

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Running head: ROADMAP FOR CHANGING STUDENT ROADMAPS

1

A Roadmap for Changing Student Roadmaps: Designing Interventions That Use Future
“Me” to Change Academic Outcomes

Eric Horowitz

A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(Psychology)

August 2019

ROADMAP FOR CHANGING STUDENT ROADMAPS

2
TABLE OF CONTENTS
A Roadmap for Changing Student Roadmaps: Designing Interventions That Use Future
“Me” to Change Academic Outcomes......................................................................................... 1
Dissertation Introduction ............................................................................................................. 7
References ................................................................................................................................. 10
Chapter 1: How, when, and why thinking about future “me” influences current choices .. 11
Abstract ..................................................................................................................................... 11
Introduction ............................................................................................................................... 12
Part 1: Existing Approaches - A New Approach is Needed ..................................................... 15
Overview ............................................................................................................................... 15
Possible Selves ...................................................................................................................... 17
Self-Gap and Self-Continuity................................................................................................ 18
Summary ............................................................................................................................... 21
Part 2: A New Approach – Identity-based Motivation Theory ................................................ 22
Overview ............................................................................................................................... 22
The Three Components of Identity-based Motivation (IBM) ............................................... 23
When Does Identity-based Motivation Yield Future-Focused Action?................................ 28
Summary ............................................................................................................................... 34
Part 3: Using Identity-based Motivation Theory to Synthesize the Future Self Literature ...... 35
Overview ............................................................................................................................... 35
Accessibility .......................................................................................................................... 53
Possible Selves: Valence, Balance Fit, Plausibility, Linked Strategies, and Efficacy .......... 56
ROADMAP FOR CHANGING STUDENT ROADMAPS

3
Self-Gaps: Gaps Between Current and Future Identities, Experienced Progress Addressing
These Gaps, and “Mental Contrasting” to Highlight and Address These Gaps ................... 67
Self-Continuity ...................................................................................................................... 78
Part 4: Broadening the Lens...................................................................................................... 89
Making Future “Me” Count .................................................................................................. 89
IBM and Other Approaches to Future Time and Goals ........................................................ 92
Final Remarks and Moving Forward .................................................................................... 99
References ............................................................................................................................... 103
Supplemental Materials .......................................................................................................... 121
Chapter 2: Do you need a roadmap or can someone give you directions: When school-
focused possible identities change so do academic trajectories ............................................ 127
Abstract ................................................................................................................................... 127
Introduction ............................................................................................................................. 128
Do Possible Identities Change Over The Course of The School Year?.............................. 129
How Might Context Affect Possible Identity Change? ...................................................... 129
Possible Identity Change and Intervention ......................................................................... 130
Current Study .......................................................................................................................... 133
Sample ................................................................................................................................. 133
Method ................................................................................................................................ 134
Human subjects and power ............................................................................................. 134
Possible identities and linked strategies .......................................................................... 134
Measures ......................................................................................................................... 136
Results ................................................................................................................................. 139
ROADMAP FOR CHANGING STUDENT ROADMAPS

4
Preliminary analyses ....................................................................................................... 139
Do school-focused possible identities change over the course of the school year? ........ 140
Do changes in school-focused possible identities predict changes in academic outcomes
and are these effects context-dependent? ........................................................................ 142
Secondary Analyses ................................................................................................................ 146
Sample ................................................................................................................................. 146
Method ................................................................................................................................ 146
Data collection and coding.............................................................................................. 146
Training the classifier ..................................................................................................... 147
Results ................................................................................................................................. 148
Discussion ............................................................................................................................... 151
References ............................................................................................................................... 154
Supplemental Materials .......................................................................................................... 158
Chapter 3: Teachers can do it: Scalable identity-based motivation intervention in the
classroom ................................................................................................................................... 163
Abstract ................................................................................................................................... 163
Introduction ............................................................................................................................. 164
Identity-based Motivation Theory ...................................................................................... 166
Implementation Fidelity Entails Fidelity of Delivery and of Receipt ................................. 172
Research Questions ............................................................................................................. 175
Materials and Methods............................................................................................................ 176
Sample ................................................................................................................................. 176
Procedure ............................................................................................................................ 177
ROADMAP FOR CHANGING STUDENT ROADMAPS

5
The Intervention .................................................................................................................. 180
Consent................................................................................................................................ 184
Fidelity ................................................................................................................................ 184
Computing Core GPA and Course Failure ......................................................................... 194
Computing Teaching Quality Inside and Outside the IBM intervention ............................ 194
Results ..................................................................................................................................... 195
Can Teachers Implement with Fidelity? ............................................................................. 195
Distinguishing Fidelity Specific to Pathways ..................................................................... 197
Effects of Fidelity................................................................................................................ 198
Teachers’ Perspectives ........................................................................................................ 206
Discussion ............................................................................................................................... 207
Summary of Results ............................................................................................................ 208
Theoretical Implications: Identity-based Motivation .......................................................... 210
Practical Implications .......................................................................................................... 211
Limitations .......................................................................................................................... 215
Future Directions................................................................................................................. 216
Conclusion .............................................................................................................................. 217
References ............................................................................................................................... 218
Supplemental Materials .......................................................................................................... 225
Conclusion and Future Directions ........................................................................................... 266
Future Theoretical Research ................................................................................................... 266
Applied Implications............................................................................................................... 269
Concluding Remarks............................................................................................................... 270
ROADMAP FOR CHANGING STUDENT ROADMAPS

6
References ............................................................................................................................... 272

ROADMAP FOR CHANGING STUDENT ROADMAPS

7
Dissertation Introduction
Every day people make choices that will only pay off in the future, whether it’s the
decision to study for a test, go to the gym, or put money aside for retirement. Researchers
attempting to understand these choices have increasingly focused on how one’s “future self” or
“possible identities”—their future “me”—is experienced, examining how these images or
thoughts of oneself in the future might influence choices in the present. Insights from this work
have been used by practitioners, corporations, and other researchers to craft programs and
interventions that aim to use images of one’s future “me” to motivate future-focused behaviors in
domains as varied as education (Oyserman, Bybee, & Terry, 2006), health (Sheeran, Harris,
Vaughn, Oettingen, & Gollwitzer, 2013), and finance (Bryan & Hershfield, 2012). Yet there are
a number of gaps in our understanding of future “me” that hinder efforts to design optimal
interventions. This dissertation attempts to address some of these gaps.
In Chapter 1, I outline a new unifying framework for understanding when a future “me”
will influence behavior. I begin by discussing the prevailing theories about when and why
thinking about a future “me” influences choices in the present, highlighting inconsistencies and
conflicts that emerge from these different approaches. I then use identity-based motivation
theory (Oyserman et al., 2017) to propose a new unifying framework that can account for
seemingly inconsistent or conflicting findings and allow for a more parsimonious understanding
of when a future “me” is likely to influence choices in the present.
Specifically, I propose that whether a future “me” matters for behavior depends on
whether it is experienced as relevant to current choices, and that relevance depends on the three
components of identity-based motivation—dynamic construction, action readiness, and
procedural readiness. These components suggest a number of different routes to relevance, many
ROADMAP FOR CHANGING STUDENT ROADMAPS

8
of which are based on characteristics of a future “me” that are highlighted by other theories (e.g.
psychological connection, Bartels & Rips, 2010; having strategies to attain a future “me”,
Oyserman, Bybee, Terry, & Hart-Johnson, 2004; having self-efficacy, Hooker & Kaus 1994,
etc.) Thus, the proposed identity-based framework provides a better starting point for
intervention designers—the goal must be to ensure that a future “me” is experienced as relevant,
and that actions to attain it are congruent with one’s current identity. This may entail programs or
activities that aim to create psychological connection or self-efficacy, for example, but by
grounding the intervention in the relevance of future “me” intervention designers can avoid
situations where specific future “me” characteristics are likely to fail to produce relevance.
In Chapter 2, I focus on filling in a specific gap in the literature that can hinder the design
of academic interventions based on changing school-focused possible identities. Specifically,
researchers have not quantified the normal development change in school-focused possible
identities over the course of 8
th
grade, or studied whether these changes matter for academic
outcomes. Understanding the nature of these changes—whether school-focused possible
identities tend to increase, decrease, or stay the same—is necessary for knowing how an
intervention should be designed. If there is no change, interventions should focus on getting
participants to imagine their futures in new ways. If school-focused possible identities tend to
decline, interventions would need to focus on helping students maintain these future identities. If
these possible identities tend to increase, interventions would need to focus on enhancing the
kinds of strategies students have, ensuring that students have a plausible roadmap to work toward
these possible futures. I find that, on average, school-focused possible identities decline over the
course of 8
th
grade. Furthermore, this decline predicts a decline in core GPA over the course of
the school year.
ROADMAP FOR CHANGING STUDENT ROADMAPS

9
Because coding whether possible identities are school-focused is resource intensive—it
requires training multiple coders to ensure reliability—in a follow-up analysis I develop and test
a machine-learning algorithm for computationally classifying the content of students’ possible
identities. A computational solution for coding possible identities would make it much easier to
scale the evaluation of interventions that attempt to change school-focused possible identities. I
find that the algorithm can classify possible identities with sufficient accuracy. In addition,
change in the number of these machine-coded school-focused possible identities predicts change
in core GPA, albeit effects are marginally weaker when compared with change in the number of
school-focused possible identities coded by researchers.
In Chapter 3, I focus on another gap in the literature on possible identity interventions
and their scalability. Specifically, I ask whether these interventions can be implemented by
middle school teachers, in a standard class period, and with minimal training. I evaluate the
implementation of Pathways-to-Success, an identity-based motivation intervention that aims to
change students’ possible identities. I find that teachers can implement with sufficient fidelity,
and that higher fidelity is associated with improvement in core GPA. These results suggest that
teachers are capable of effectively implementing possible identity interventions. Furthermore,
my findings suggest that putting these interventions in the hands of teachers may be one solution
to the difficulties of finding the time and staff necessary to bring interventions to scale.

ROADMAP FOR CHANGING STUDENT ROADMAPS

10
References
Bartels, D. M., & Rips, L. J. (2010). Psychological connectedness and intertemporal choice.
Journal of Experimental Psychology: General, 139(1), 49-69.
Bryan, C. J., & Hershfield, H. E. (2012). You owe it to yourself: Boosting retirement saving with
a responsibility-based appeal. Journal of Experimental Psychology. General, 431(3),
429-432.
Hooker, K., & Kaus, C. R. (1994). Health-related possible selves in young and middle adulthood.
Psychology and Aging, 9(1), 126-133.
Oyserman, D., Bybee, D., & Terry, K. (2006). Possible selves and academic outcomes: How and
when possible selves impel action. Journal of Personality and Social Psychology, 91(1),
188-204.
Oyserman, D., Bybee, D., Terry, K., & Hart-Johnson, T. (2004). Possible selves as roadmaps.
Journal of Research in Personality, 38(2), 130-149.
Oyserman, D., Lewis, N.A., Jr., Yan, V. X., Fisher, O., O'Donnell, S. C., & Horowitz, E. (2017).
An identity-based motivation framework for self-regulation. Psychological Inquiry, 28
(2-3), 139-147.
Sheeran, P., Harris, P., Vaughan, J., Oettingen, G., & Gollwitzer, P. M. (2013). Gone exercising:
Mental contrasting promotes physical activity among overweight, middle-aged, low-SES
fishermen. Health Psychology, 32(7), 802.

ROADMAP FOR CHANGING STUDENT ROADMAPS

11
Chapter 1: How, when, and why thinking about future “me” influences current choices

Abstract
People often fail to invest sufficiently in their future, starting too late and persisting too little. To
predict why, each of the three major theoretical future “me” approaches (possible selves, self-
gap, and self-continuity) focuses on a different aspect of future “me.” Our comprehensive review
(N=145 comparisons, 139 studies) of this literature reveals that these approaches make
contradictory predictions and yield inconsistent results: Sometimes people invest if they have
strategies to get going or feel efficacious or if their future “me” is positive (or negative).
Sometimes people invest if the person they might become feels different from the person they
are right now—or like the same person they are now. We make sense of these inconsistent
results by using the dynamic construction, action-readiness, and procedural-readiness
components of identity-based motivation (IBM) theory. In doing so, we shift focus from what
future “me” is—positive or negative, close or distant, continuous or discontinuous with current
“me”—to what future “me” does, allowing us to make three predictions, two regarding when
people maintain a present-focused course of action and one regarding when people take future-
focused action. People maintain present-focused action if future “me” is not on the mind. People
also maintain present-focused action if future “me” feels irrelevant to current choices and
difficulties taking future-focused action seem to imply low value or low odds of success. In
contrast, people take future-focused action if future “me” feels relevant to current choices and
difficulties taking future-focused action seem to imply the value of doing so.

ROADMAP FOR CHANGING STUDENT ROADMAPS

12
Introduction
People frequently think about the future (Baumeister, Hofmann, Summerville, Reiss, &
Vohs, 2018) but often fail to invest enough in their future “me”—the person they will be in the
future. They don’t do enough schoolwork (Oyserman & Destin, 2010), don’t save (half of
households lack retirement savings, Morrissey, 2016) and live recklessly (the most common
causes of death are lifestyle-related, Case & Deaton, 2015). One reason offered for why this
might be is that current “me” pushes hypothetical future “me” aside (Pronin, Olivola, &
Kennedy, 2008). However, it is not the case that everyone fails to invest in the future or that
people always fail to take future-focused action. People do sometimes take future-focused action:
some students invest deeply in school, half of households do have retirement savings, and many
people do lead healthy lifestyles.
Why some people invest in their future while others do not has been the subject of
decades of psychological research, and three major theoretical approaches have emerged to
attempt to explain the differences: possible selves, self-gap, and self-continuity. Each
individually makes plausible predictions about what triggers and sustains such future-focused
(in)action. Unfortunately, when one takes a step back to examine these approaches holistically,
not only do their predictions conflict rather than complement one another, so do their supporting
evidentiary bases. From a possible-selves frame, future-focused action occurs if people have a
“possible” future that provides a goalpost, a target to orient current action (Oyserman & James,
2011). From a self-gap frame, future-focused action occurs if an aspect of that future is
experienced as a positive standard to be attained or a negative standard to be avoided (control
theory, Carver & Scheier, 1982, 2016). From a self-continuity frame, future-focused action
occurs if that future is experienced as part of the same entity as current “me” (Hershfield &
ROADMAP FOR CHANGING STUDENT ROADMAPS

13
Bartels, 2018). In the current paper, we highlight conflicting and inconsistent findings arising
from each of these approaches. Then we use identity-based motivation theory (Oyserman, 2007,
2009a, 2015a) to synthesize findings and articulate when and how people’s course of action
becomes future- rather than present-focused.
Identity-based motivation theory is a situated social psychological theory of motivation
and goal pursuit that focuses on when and how self-regulation works (Oyserman, 2007, 2009b,
2015, Oyserman et al., 2017). Identity-based motivation theory starts with people’s everyday
experiences of knowing who they are and of making choices based on this self-knowledge,
predicting that people prefer to act and make sense of their experiences in ways that feel identity-
congruent—like “me” or “us” things to do. On the one hand, identity feels stable and
consequential (Quoidbach, Gilbert, & Wilson, 2013). But—to paraphrase William James
(1890)—thinking (about the self) is for doing, or put another way, people are sensitive to the
choices afforded and constrained by their immediate situation. The implication is that identities
are not stable but dynamically constructed by features of the immediate situation that shape
which identities come to mind and what these identities seem to imply for action and for
meaning making. We use these three elements of identity-based motivation theory (termed
dynamic construction of identity, action-readiness, and procedural-readiness) to predict when
considering one’s future “me” shifts people from present-focused to future-focused action. As
we will detail below, future “me” has to be on the mind to matter, but accessibility is not enough.
To matter, an on-the-mind future “me” must be experienced as relevant to the affordances and
contrasts facing current “me” and difficulties imagining future “me” or in starting and persisting
in future-focused action must be experienced as signaling that this is “for me” (“no pain, no
gain”). Even if their future “me” is on their mind, people are likely to continue on a present-
ROADMAP FOR CHANGING STUDENT ROADMAPS

14
focused path of action if future “me” does not feel relevant to the choices they currently face.
They are likely shift to a future-focused path only if their future “me” is on their mind and feels
relevant to the choices they face at the moment, otherwise, difficulties imagining future “me” or
in starting and persisting in future-focused action will be interpreted as meaning “not for me.”
We divide our paper into four parts. First, we clarify why a new approach is needed. To
do so, we present examples of the conflicting evidence current approaches yield. Second, we
articulate our new approach. To do so, we define and then use the three core components of
identity-based motivation (dynamic construction, action-readiness, and procedural-readiness). As
we show, this allows us to predict when future “me” or a specific possible identity will be
experienced as relevant to the choices facing current “me,” triggering future-focused rather than
present-focused action. Third, we summarize the extant literature. To do so, we operationalize
core constructs, review predictions arising from each approach and from IBM theory, highlight
the critical comparisons needed to understand if the predictions each approach makes are
supported, and discuss what results imply. Fourth, we look beyond possible-self based, self-gap,
and self-continuity approaches to other theories linking time, identity, and motivation to situate
our IBM-based predictions within an even broader context, and conclude by summarizing our
key results and articulating future directions.
A methodological note prior to proceeding: The possible self, self-continuity, and self-
gap literatures use a variety of different terms to describe the future self, mostly without
operationalizing what these terms entail. Since this foments lack of clarity as to what is predicted
and found and makes synthesis difficult, going forward we impose a uniform operationalization
to allow us to synthesize across studies and approaches. To avoid confusion, we use the general
term “self” only to refer to these literatures generally—i.e., a possible self-based approach, a
ROADMAP FOR CHANGING STUDENT ROADMAPS

15
self-gap approach, a self-continuity approach. Otherwise, we follow the operationalization of
Oyserman, Elmore, and Smith (2012) in which “self” is a superordinate term including an “I” or
reflective capacity and a “me” or that which is reflected on. The “me” is temporal (past, present,
future). It includes valenced content (specific identities including social roles and traits) and
valenced meta-perceptions (e.g., “I have value”—self-esteem; “I am or could be competent”—
self-efficacy). These temporal contents and meta-perceptions are structured in self-concepts (e.g.
independent self-concept, interdependent self-concept). Thus, throughout, if a study focuses on
temporality but not content, we describe this as an effect of future “me.” If a study focuses on
specific content, valence or meta-perception, we describe this as an effect of a possible future
identity or a “possible identity” for short.
Part 1: Existing Approaches - A New Approach is Needed
Overview
On the one hand, as we summarize here and show in detail in Part 3, the existing
literature suggests that future “me” can trigger future-focused action. Each approach (possible
self, self-gap, self-continuity) highlights a particular aspect of possible future identities or of
future “me” and predicts that this is the aspect that triggers future-focused action. To facilitate
comparison across approaches, we briefly summarize each approach’s predictions in Table 1. As
can be seen, each approach’s predictions are plausible when considered alone. At the same time,
the predictions do not take into account the rest of the literature because the literature is siloed,
compartmentalized by approach. When taken together, predictions not only diverge, they also
sometimes conflict, as does the evidence. As a result, when, why, and how future-focused action
is triggered remains unclear.
ROADMAP FOR CHANGING STUDENT ROADMAPS

16
Table 1. Predictions made by each of the three main future “me” approaches
Framework Associated future “me” or possible
identity characteristics
Prediction
Possible self Accessible, attainable Possible identities motivate when they are on the mind (accessible) and
represent a goal or outcome that can reasonably be attained or avoided
(attainable).
Self-gap Dissimilarity, disconnection, distance A positive possible identity motivates when it highlights a too-large gap
between the positive possible identity and one’s present identity, or
highlights too-slow progress towards attaining the possible identity. A
negative possible identity motivates when it highlights a too-small gap
between the negative possible identity and one’s present identity, or
highlights too-slow progress towards avoiding the possible identity. One
way to do this is to follow as specific procedure: To imagine the future,
imagine obstacles in the present that stand in the way, and ask yourself if
you are likely to be able to do these things and if you will succeed if you
do.
Self-continuity Connection, similarity, stability,
temporal proximity, vividness
Future “me” or possible identities motivate when they are experienced as
linked to current “me” or as part of the same entity as current “me.”

ROADMAP FOR CHANGING STUDENT ROADMAPS

17
Possible Selves
A possible self-based approach predicts that possible selves motivate future-focused
action (Markus & Nurius, 1986). The implication is that future “me” and specific possible future
identities such as “becoming an A-student” or “not becoming obese” are motivating in and of
themselves, either chronically or whenever they are on the mind. That the accessibility of a
future “me” is sufficient to yield future-focused action is a unique prediction, not rooted in either
of the other approaches (self-continuity, self-gap). That is, it does not entail an assumption that
the way possible selves matter is that they trigger a sense of continuity between a current and a
future “me” or an awareness of a particular kind of gap between a current and a possible future
identity.
On the one hand, both availability and accessibility predictions have some empirical
support. A number of correlational (measurement) studies support the availability prediction.
These studies show that specific possible identity content is correlated with future-focused action
(e.g., Aloise-Young, Hennigan, & Leong, 2001; Newberry & Duncan, 2001). A number of
experimental (manipulation) studies support the accessibility prediction. These studies show that
people have a more future-focused response after being directed to consider a possible identity
(King, 2001; Kuo, Lee, & Chiou, 2016) or future “me” generally (Hershfield, et al., 2011;
Hershfield, Cohen, Thompson, 2012, Study 5; van Gelder, Hershfield, & Nordgren, 2013, Study
2). In these latter studies, the experimental group is contrasted to a control group in which people
consider something other than the self (Hershfield et al., 2012, Study 5; King, 2001) or consider
their current “me” (Hershfield, et al., 2011; Kuo et al., 2016; van Gelder, Hershfield, &
Nordgren, 2013, Study 2).
On the other hand, as detailed in our review, most studies based on a possible selves
ROADMAP FOR CHANGING STUDENT ROADMAPS

18
approach show that availability and accessibility are necessary but not sufficient. To matter,
particular features of the possible self must be triggered. Indeed, even in their seminal paper,
Markus and Nurius (1986) actually focused on the valence of possible selves, showing an
association between self-esteem and having positive possible identities (though follow-up
research made the opposite case, Ogilvie, 1987). Since these initial publications, research has
continued to examine whether positive or negative valenced possible identities are associated
with future-focused action, yielding conflicting results. In some studies, having both positive and
negative possible identities (Lee et al., 2015) or having ‘balance’—positive and negative possible
identities in the same domain (Oyserman & Markus, 1990)—is associated with future-focused
actions (less delinquent engagement). In other studies, having positive possible identities, but not
negative ones, or having negative possible identities, but not positive ones, is associated with
taking future-focused action. Thus, healthy behavior is associated with having positive possible
identities, but not negative ones (Hoppmann, Gerstorf, Smith, Klumb, 2007), but also with
having negative possible identities, but not positive ones (Black, Stein, & Loveland-Cherry,
2001; Hoyle & Sherrill, 2006). Other studies suggest that what matters is having efficacy to
avoid a negative possible identity (Hooker & Kaus, 1994), having linked strategies to work on a
possible identity (Oyserman, Bybee, Terry, & Hart-Johnson, 2004), or “fit” between how context
and possible identities are considered (Oyserman, Destin, & Novin, 2015). In spite of the large
heterogeneity of findings, researchers using a possible self-based approach rarely address
inconsistencies across studies or how results might be synthesized (though see Oyserman et al.,
2004; Oyserman et al., 2015).
Self-Gap and Self-Continuity
While a possible self-based approach does not take a position as to whether it is
ROADMAP FOR CHANGING STUDENT ROADMAPS

19
motivating for future “me” to be experienced as part of or as separate from current “me,” self-
gap and self-continuity approaches do, taking opposing positions. Self-gap approaches highlight
the motivational power of contrasting, or using future “me” as a positive or negative benchmark
of what one could become. Self-continuity approaches highlight the motivational power of
assimilating, or including future “me” in one’s mental representation of current “me.”
Control theory, a classic self-gap approach, predicts that people automatically and
continuously monitor two things, gap and speed (e.g., Carver & Scheier, 1982, 2016). People
monitor the gap between their current identities and their desired possible identities and the gap
between their current identities and their undesired possible identities. They also monitor the
speed at which they are narrowing the former gap and widening the latter gap. From a control
theory perspective, people should invest in becoming a desired possible identity whenever the
gap between current and desired possible identities gets too big, and invest in avoiding becoming
like an undesired possible identity whenever the gap between current and undesired possible
identities gets too small. People will get more invested when their speed of progress requires it—
that is, if their progress narrowing or increasing the relevant gap is not moving at the expected
rate.
An emerging self-gap approach, mental contrasting theory (Oettingen, 2012), predicts
that gaps are only motivating when considered in a particular way and by particular people.
Instead of always mattering, the mental-contrast approach predicts that a gap matters if people
who are high in efficacy first elaborate on their positive desired possible identity and then
elaborate on obstacles that stand in the way of their current “me” attaining that possible future
identity. According to this approach, gaps between current and future identities do not trigger
future-focused action—they are not helpful when considered in other ways or when considered
ROADMAP FOR CHANGING STUDENT ROADMAPS

20
by people low in efficacy.
In contrast to possible self and self-gap approaches, a self-continuity approach does not
consider efficacy. Self-continuity approaches predict that future “me” is motivating when it is
experienced as continuous with or part of (assimilated into) current “me” (e.g., Bartels & Rips,
2010; Hershfield, 2011). Continuity is experienced in situations in which current and future “me”
seem to overlap or connect, when future “me” is experienced as proximal or even eminent, or
when it is experienced as vivid and clear. In these situations, investing in future “me” will be
experienced as benefiting current “me.”
The seemingly conflicting reasoning behind self-gap and self-continuity approaches
produces two sets of evidence that are often hard to reconcile. Consider the contradictory
evidence base: In some studies, better academic performance is associated with experiencing
connection between current and future “me” (Destin, 2017; Landau, Oyserman, Keefer, & Smith,
2014; Nurra & Oyserman, 2017). In other studies, better performance is associated with
considering obstacles that separate future “me” from better performance, among people who feel
efficacious about academic performance (Oettingen, Pak, & Schnetter, 2001, Study 4). In some
studies, acting in the interest of future “me” is predicted by considering why current and future
“me” are similar (Zhang & Aggarwal, 2015). In other studies, the opposite is found—rating
current and future “me” as less similar yields more future-focused action (Dalley & Buunk,
2011). In some studies participants act in the interest of their future “me” when it is experienced
as psychologically closer (Evan & Wilson, 2015; Peetz, Wilson & Strahan, 2009). In other
studies, the reverse is true, and participants act in the interest of their future “me” when it is
experienced as more distant (Rutchick et al., 2018; Peetz & Wilson, 2013).
Contradictory results even emerge across studies from self-gap and self-continuity
ROADMAP FOR CHANGING STUDENT ROADMAPS

21
approaches when they use the same measure of continuity, an overlapping circles measure,
depicted in Figure 1 (which is based on Aron, Aron, & Smollan, 1992). In studies using a self-
continuity approach, choosing circles with more overlap is associated with more future-focused
action (e.g., more retirement saving, Ersner-Hershfield, Garten, Ballard, Samanez-Larkin, &
Knutson, 2009). In studies using a self-gap approach, the reverse is found; choosing circles with
less overlap is associated with more future-focused action (e.g., more health investment, Peetz &
Wilson, 2013). The implication is that sometimes noticing lack of overlap (Top row, first circle
from left) is motivating and other times noticing more overlap is motivating. But neither self-gap
nor self-continuity approaches address what the moderator would be that would sometimes make
a gap motivating and sometimes make continuity motivating. Moreover, studies often omit the
kinds of comparisons that would clarify which of these motivational processes are actually
occurring.
Figure 1. Overlapping circles measure

Summary
Possible self-based, self-gap, and self-continuity approaches each make plausible
predictions and each has some supporting evidence. At the same time, for a number of reasons,
none of the current approaches is sufficient to understand when a future “me” matters for action.
First, possible self-based approach results are not well synthesized—neither self-gap nor self-
ROADMAP FOR CHANGING STUDENT ROADMAPS

22
continuity approaches predict or account for research showing that valence, balance, linked
strategies and efficacy as features of a future “me” sometimes matter. Second, self-gap and self-
continuity studies yield conflicting results and neither a self-gap nor a self-continuity approach
addresses the conditions in which evidence supporting the other approach would be found. Third,
self-gap and self-continuity study designs do not uniformly include an accessibility condition,
which means that whether effects are due to gap, continuity, or accessibility cannot be
disentangled. Fourth, within a particular approach, the empirical evidence is not fully consistent
when examined closely. Fifth, while having efficacy is sometimes noted to matter, what actually
would be the way to promote future-focused action when efficacy is low is never articulated.
The implication we draw from inconsistent results within approaches is that unmeasured
moderators are at play within the purview of each approach. The implication we draw from the
inability of current approaches to synthesize results across approaches is that none of the current
approaches is sufficient. Hence, a new model is needed to synthesis results, to articulate what the
unmeasured moderators might be, and to make novel predictions about how and when future
“me” matters for future-focused action. We articulate that model in Part 2.
Part 2: A New Approach – Identity-based Motivation Theory
Overview
Given the limitations of current approaches, a new approach is needed to synthesize
current research and make new predictions about when, why, and how future “me” triggers
future-focused rather than present-focused action. As we detail next, identity-based motivation
(IBM) theory allows us to do just that. IBM theory’s core prediction is that features of the
immediate situation dynamically shape identity accessibility, content, and implications of an
accessible identity for meaning making and action. For clarity, we first define each component of
ROADMAP FOR CHANGING STUDENT ROADMAPS

23
IBM and then use IBM to articulate when, why, and how an accessible future “me” triggers
future-focused as opposed to present-focused action.
The Three Components of Identity-based Motivation (IBM)
IBM theory starts with the premise that identities carry with them mental procedures that
guide people to act and to make sense of their experiences in ways that are congruent with those
identities. However, which identities come to mind and what they imply for meaning making and
action is dynamically constructed in the moment. Dynamic construction, the first component of
IBM, is the idea that identity content is shaped by features of the current situation the person is
in, while acknowledging that people’s previous situations can influence their perceptions of their
current situation (Lewis & Oyserman, 2016; Lewis & Yates, 2019; Oyserman, 2007; 2009;
Oyserman & Lewis, 2017; Oyserman et al., 2017). Dynamic construction differs from the more
standard notion of “working self-concept,” which refers to the effect of situational cues on which
identities are drawn from memory (Markus & Wurf, 1987). Both IBM theory and “working self-
concept” focus attention on the effect of the immediate situation on identity accessibility.
However, IBM theory goes further by predicting that situations dynamically shape identity
content and implications for action (action-readiness) and for meaning making (procedural-
readiness).
Consider for example, a male gender identity. Is doing schoolwork part of being a boy or
is it a girl thing to do? The dynamic construction component of IBM theory yields the prediction
that features of a situation can not only make identity as a boy accessible (working self-concept),
but also shape the content linked to being a boy—what being a boy means for behavior (action-
readiness) and for interpretation of experience (procedural-readiness). Elmore and Oyserman
(2012) tested this prediction by randomly assigning boys to one of four groups. Each group was
ROADMAP FOR CHANGING STUDENT ROADMAPS

24
shown a graph with U.S. Census data. In one group, the graph showed income broken down by
gender, with male earnings higher than female earnings. In the second group, the graph showed
high school graduation rates broken down by gender, with female graduation rates higher than
male graduation rates. Two other groups saw earnings and graduation graphs that did not
mention gender. The specific graph the boys saw influenced how they made sense of what being
a boy implied. On measures taken after seeing the graphs, boys who saw the graph implying that
men are more professionally successful than women listed more school-focused possible
identities and spent more time on an academic task than boys who saw the other graphs. These
findings illustrate that contexts do more than simply cue an identity from memory. Instead,
features of the immediate situation dynamically construct identity content, and hence, what an
identity implies for action. When the (experimental) situation showed that men succeed in
American society, boys understood that to mean they should work harder on academics in order
to become one of those successful men in the future. In contrast, when the (experimental)
situation showed that women succeed in American society, boys understood that to mean that
there was no need to work hard on academics in order to become a successful man in the future.
As these results highlight, what an identity implies about a particular action or course of action
depends on what contextual cues suggest is identity-relevant—the same identity can trigger
present or future-focused action depending on the situation (Oyserman, 2007, 2009a, 2009b).
The other two components of identity-based motivation, in addition to dynamic
construction, are termed action-readiness and procedural-readiness. IBM theory predicts that
situations shape which actions feel identity-congruent and the mental procedures (mindsets)
people use to make sense of their experiences. Action-readiness is the readiness to act in identity-
congruent ways, to do the things that, in context, seem to fit with on-the-mind identities—to do
ROADMAP FOR CHANGING STUDENT ROADMAPS

25
what “I” and “we” (people like “me”) do. Procedural-readiness is the readiness to use the mental
procedure that fits the content and structure of on-the-mind identities
1
. One key thing people try
to make sense of is their experiences of ease and difficulty. People have available to them two
difficulty mindsets and two ease mindsets, which together structure what experienced difficulty
and ease imply for identity (Fisher & Oyserman, 2017; Oyserman, 2009a; Oyserman et al.,
2017). Experienced difficulty can imply that succeeding at a task is identity-congruent, a “me” or
“us” thing to do, such that difficulty can be thought of as a signal of task importance (“no pain,
no gain”). Experienced difficulty can also imply that succeeding at a task is identity-incongruent,
a “not for me” or “not for us” thing, such that difficulty can be thought of as a signal of task
impossibility or hence irrelevance for “me” or for “us.”
If in context, taking action is identity-congruent—a “for me” or “for us” thing to do—
then experienced difficulty is likely to imply that engaging with the task is important and
valuable. Conversely, if in context, taking action is not identity-congruent—a “not for me” or
“not for us” thing to do—then experienced difficulty is likely to imply low odds of success or
even impossibility, and be interpreted as a sign that engaging with the task is a waste of time. As
depicted in Figure 2, IBM theory predicts that activating one component in the associative
knowledge network that includes identity content, action readiness, and procedural readiness
probabilistically activates the other components of the network. Consequently, when a particular
interpretation of experienced difficulty is on the mind (procedural-readiness) it can influence
what an identity seems to be about (dynamic construction of identity content) and what that

1
Recall that self-concepts are the ways identities are structured. IBM theory predicts that when
an individualistic self-concept is on the mind, people are more likely to use an analytic procedure
to make sense of their experiences and when a collectivistic self-concept is on the mind, people
are more likely to use a holistic procedure to make sense of their experiences (for a recent
review, Oyserman, 2017).
ROADMAP FOR CHANGING STUDENT ROADMAPS

26
identity implies for action (action-readiness). For example, if difficulty-as-importance comes to
mind, it implies that the task at hand has value for “me” or for “us.” Taking action to initiate,
persist, or creatively solve the task is then likely to be experienced as identity-congruent, cueing
readiness to engage in the task. Conversely, if difficulty-as-impossibility comes to mind, it
implies that the task at hand has low odds of success, is impossible for “me” or for “us.” Taking
action to initiate, persist, or creatively solve the task is then likely to be experienced as “not for
me” or “not for us,” undermining readiness to act.

Figure 2. Components of identity-based motivation (IBM) form an associative knowledge
network. Figure adapted from Oyserman & Fisher, 2017. Reprinted with permission.

ROADMAP FOR CHANGING STUDENT ROADMAPS

27
Four experiments highlight the effect of activating difficulty mindsets on identity content
and action readiness. Activated difficulty mindset affects identity certainty (Aelenei, Lewis &
Oyserman, 2017, Study 2) and centrality (Smith & Oyserman, 2015, Study 1b). It also affects
willingness to sacrifice to work on these identities (Aelenei, et al., 2017, Study 2), time devoted
to identity-relevant tasks (Smith & Oyserman, 2015, Study 2), and quality of performance
(Oyserman, Elmore, Novin, Fisher, & Smith, 2018). As detailed next, two methods were used to
activate difficulty mindsets—a biased scale method, and an autobiographical recall plus biased
scale method. Results were not dependent on activation method. Taken together, results support
the associative knowledge network prediction that activating procedural-readiness affects action-
readiness and identity content.
In the biased scale method (Aelenei et al., 2017, Study 2; Oyserman et al., 2018, Study 2)
students were randomly assigned to one of two groups (difficulty-as-importance, difficulty-as-
impossibility). Students in the importance group rated how much they agreed or disagreed with
statements reflecting the idea that experienced difficulty with schoolwork might imply task
importance. In the impossibility group, the statements reflected the idea that experienced
difficulty with schoolwork might imply task impossibility. The statements students considered
shaped identity content and action-readiness. Students randomly assigned to the importance
group were more certain that they could attain their academic possible identities (Aelenei et al.,
2017, Study 2), more willing to sacrifice to attain these possible identities (Aelenei et al., 2017,
Study 2), and actually performed better on an academic task (Oyserman et al., 2018, Study 2).
In the autobiographic recall plus biased scale method (Smith & Oyserman, 2015, Study
1b, Study 2) the researchers showed students response scales implying that their own frequency
was more than their peers or less than their peers. Students were randomly assigned to one of
ROADMAP FOR CHANGING STUDENT ROADMAPS

28
four groups (difficulty-as-importance is more common for me than for others, difficulty-as-
impossibility is more common for me than for others, difficulty-as-importance less common for
than for others, difficulty-as-impossibility is less common for me than for others). In the
‘importance is more common for me than for others’ group, students were instructed to recall
times that they experienced difficulty with schoolwork as implying its importance and were
given a low-frequency scale on which to respond (implying this difficulty mindset is rare for
others). In the ‘impossibility is less common for me than for others’ group, students were
instructed to recall times that they experienced difficulty with schoolwork as implying its
impossibility and were given a high-frequency scale on which to respond (implying that this
difficulty mindset is more common for others). Students randomly assigned to these groups
subsequently rated academics as more central to their identities (Study 1b) and spent more
time—and hence performed better—on a test of fluid intelligence (Study 2) than students
randomly assigned to the other groups.
When Does Identity-based Motivation Yield Future-Focused Action?
As detailed next, IBM theory makes two predictions about when identity-based
motivational processes should yield present-focused versus future-focused actions. First, it
predicts that future “me” may or may not be on the mind (accessible in the moment). When not
on the mind, future “me” is unlikely to affect current “me,” making present-focused rather than
future-focused action likely; in other words, if one is not considering their future self, there is no
reason to take action on that self’s behalf. Second, IBM theory predicts that an on the mind
future “me” may or may not feel relevant to the choices facing current “me.” IBM theory
predicts that people take present-focused action if future “me” is either not on the mind or is on
the mind but does not feel relevant to the choices they currently face. Thus, IBM theory predicts
ROADMAP FOR CHANGING STUDENT ROADMAPS

29
that people only take future-focused action if future “me” is on their mind and feels relevant to
the choices they currently face.
When we speak of the relevance and irrelevance of future “me” to the choices facing
current me, that is a shorthand way of talking about the result of triggering either future-focused
(Figure 3, top panel) or present-focused (bottom panel) IBM associative knowledge networks.
Both processes can start with identity content (dynamic construction), difficulty mindset
(procedural readiness), or with taking action (action readiness). In the next section, we articulate
how each element can be triggered by the characteristics of future “me” that are on the mind
(accessible) in the moment, yielding spreading activation to the other IBM elements.

ROADMAP FOR CHANGING STUDENT ROADMAPS

30

Figure 3. Components of identity-based motivation (IBM) form an associative knowledge
network, depending on contextual cues, the network may include future-focused (Top Panel) or
present-focused (Bottom Panel) content.

Figure 4 depicts our full process model, including recursive loops. As shown in Figure 4,
both future-focused action and present-focused action recursively affect identity-based
motivational processes. The implication is that having embarked on a future-focused course of
action may become self-reinforcing, dynamically constructing an identity as the kind of person
who is future-focused, increasing the subsequent experienced relevance of future “me.” In the
same way, having embarked on a present-focused course of action may also become self-
reinforcing, dynamically constructing an identity as the kind of person who lives for today,
decreasing the subsequent experienced relevance of future “me.”
ROADMAP FOR CHANGING STUDENT ROADMAPS

31

Figure 4. Full process model for understanding when and how an on the mind future “me” will influence behavior.
ROADMAP FOR CHANGING STUDENT ROADMAPS

32
A number of studies provide direct support for our prediction that an accessible future
“me” (or a specific future identity) can trigger either future-focused or present-focused action
depending on how it shapes dynamic construction of current “me” (Landau et al., 2014, Studies
1, 2, 3, 5, 7; Oyserman, Destin, & Novin, 2015, Studies 1, 2, 3, 4). In these studies, researchers
tested the effect of instructing participants to think about their possible identities in different
ways on participants’ subsequent future-focused versus present focused action. To do so, they
randomly assigned participants to different instruction groups and, as detailed next, the results of
these studies demonstrate that an on-the-mind possible identity may or may not be experienced
as relevant to the choices facing current “me.” Sometimes instructions focused only on identity
content, sometimes only on identity connection, and sometimes on both. Neither identity content
nor identity connection was a sufficient relevance cue without the other. Instead, for an on-the-
mind possible identity to feel relevant to the choices facing current “me,” it had to be envisioned
in a way that activated the elements of future-focused IBM.
First, consider studies employing metaphors to dynamically construct a possible identity
that is relevant to current “me.” In these studies, some students were randomly assigned to
consider their school-focused possible identities and write them down on an image of a path
leading to the future (Landau et al., 2014, Studies 1, 2, 3, 5, 7). The path instruction can be
considered a cue of connection. According to the authors, if future “me” is on the path to current
me, this metaphorically implies locomotion, with the consequence that taking action feels easy.
When students were assigned to consider and write about their academic possible identities on an
image of a path, their accessible possible identities felt relevant to the academic choices they
were currently. In contrast, other students were randomly assigned to consider and write about
ROADMAP FOR CHANGING STUDENT ROADMAPS

33
their social and relational future identities on a path, or to consider and write about their school-
focused possible identities and place them in separate boxes or houses. According to the authors,
if future “me” is in a different box or house than current “me,” this implies that future “me” has
to get out of the box first, making taking action feel implausibly difficult and triggering a
difficulty-as-impossibility mindset. Hence, these instruction set made accessible possible
identities that felt irrelevant to the academic choices facing current “me.” In one instruction set,
accessible possible identities felt irrelevant because their content was not relevant to the
academic choices facing current “me” in the task. In the other instruction set, accessible possible
identities felt irrelevant because their metaphoric link set the stage for participants to experience
taking action as too difficult. Indeed, students who were randomly assigned to the possible
identity relevant condition (thinking about school-focused possible identities on a path) took
more future-focused academic action than other students, including expressing more interest in
an academic workshop and performing better on an exam.
Next, consider studies employing context to dynamically construct a possible identity that
is relevant to current “me.” In these studies, students were randomly assigned to consider their
possible identities in ways that fit or did not fit how they were led to imagine the college context
(Oyserman, Destin, & Novin, 2015, Studies 1, 2, 3, 4). Specifically, students were assigned to
one of four instruction groups. In two groups, students received versions of fit instructions and in
the other two groups students received versions of misfit instructions. One fit group read about
college as a success-likely context in which they were likely to succeed and they were asked to
describe their desired possible identities over the college years. The second fit group read about
college as a failure-likely context in which they were likely to fall short of their aspirations and
they were asked to describe their undesired possible identities over the college years. One misfit
ROADMAP FOR CHANGING STUDENT ROADMAPS

34
group read about college as a failure-likely context but was asked to imagine and then write
about their desired possible identities. The second misfit group read about college as a success-
likely context but was asked to imagine and then write about their undesired possible identities.
Students assigned to the fit groups planned to study more, start studying sooner, and endorsed a
difficulty-as-importance mindset more than students assigned to the misfit groups. In these
studies, it was not valence or context that made a possible identity influential, but rather whether
valence and context interacted in a way that made the possible identity be experienced as
relevant to current academic choices that mattered.
Summary
IBM theory provides a synthesizing account of when, why, and how an on-the-mind
possible identity (or future “me”) can trigger either future-focused or present-focused action. In
doing so, as we will show in Part 3, IBM theory highlights the common process underlying the
different predictions made by possible selves, self-gap, and self-continuity approaches about
when a possible identity (or future “me”) triggers future-focused action. Specifically, our IBM-
based process model predicts that the characteristics of future “me” associated with possible
selves, self-gap, and self-continuity approaches trigger future-focused action when future “me”
feels relevant to the choices facing current “me” in the immediate situation. When experienced
as relevant to the affordances and contrasts facing current “me,” taking future-focused action
feels identity-congruent and a difficulty-as-importance mindset is used to make sense of
difficulties in getting going or in surmounting obstacles. Otherwise, future “me” will be
experienced as irrelevant to the affordances and contrasts facing current “me.” When this occurs
taking future-focused action will feel identity-incongruent and a difficulty-as-impossibility
mindset will be used to make sense of difficulties in getting going or in surmounting obstacles to
ROADMAP FOR CHANGING STUDENT ROADMAPS

35
taking future-focused action.
Part 3: Using Identity-based Motivation Theory to Synthesize the Future Self Literature
Overview
Having articulated the predictions IBM theory makes as to when and how future-focused
action is triggered, we turn to the otherwise compartmentalized possible selves, self-gap, and
self-continuity empirical literatures on the relationship between future “me” and behavior. We
wanted to obtain the full body of work rather than selected examples so that we could understand
what these literatures actually show. We chose the period from January 1985 to June 2018—
starting just before Markus and Nurius (1986) launched a renewed interest in future “me” with
their seminal paper on possible selves. To obtain studies, we searched the PsychINFO database
for abstracts with relevant key search terms including “future self/ves”, “possible identity/ies”,
“possible self/ves”, and “mental contrasting
2
.” We added ancestry searches and searches based
on authors known to research future “me.” We read abstracts and then full papers. We included
all studies designed to determine effects of future “me” on behavior by measuring or
manipulating future “me” or a specific possible identity and measuring a future-focused
behavioral intention or behavior
3
. This process yielded 139 experiments and measurement
studies that include 145 results (6 studies test two distinct independent measures) in 81
publications, which are listed in Table 2.
Our goal is to highlight what is known and to critically synthesize otherwise conflicting

2
To preserve our goal of inclusivity, we included all studies entailing mental contrasting that
were linked to a behavioral dependent variable even though in some of these studies, the future
and present are not always specified as being about the self. For clarity, we excluded studies in
which mental contrasting was combined with another intervention because in these studies, it
was impossible to disentangle the effect of mental contrasting.
3
To avoid confusion, we use the term experiment to describe studies that entail manipulation and
the term measurement study to describe studies that entail measurement.
ROADMAP FOR CHANGING STUDENT ROADMAPS

36
findings with a general model that will advance the field, set a path for future research, and help
practitioners predict when a future “me” or possible identity triggers future-focused behavior.
For a number of reasons, a quantitative synthesis is not feasible: both the comparisons and the
independent and dependent measures vary across studies in ways that undermine quantitative
synthesis. Hence, the current literature is not set up to obtain a quantified estimate of how much
future “me” matters or to test whether the average effect in possible selves, self-gap, and self-
continuity approaches differ. The point of our review is not to determine which approach is
better but rather to understand why none can address the findings from the others and to
articulate a model that can.
In Table 3, we operationalize the key constructs in each approach and highlight the IBM
theory predictions related to each approach. We start with accessibility and then organize our
results using the three main theoretical approaches (possible selves, self-continuity, self-gap).
We separated evidence about accessibility into a separate section because, although accessibility
of future “me” is a core part of the possible self-based approach, accessibility is also implicitly a
factor in each experiment and measurement study regardless of the approach used. A
shortcoming of the current literature is that researchers do not consistently rule out the possibility
that effects are driven by accessibility because they do not always include a condition in which
participants simply bring future “me” to mind. This omission reduces our ability to draw firm
conclusions as to whether it is accessibility of future “me” or other features of future “me” that
produce future-focused action, and if so, when each feature matters. In the sections below, we
present a summary of how each approach can be tested and articulate what the empirical
evidence base is and the ways this empirical base supports or fails to support the specific
predictions each approach makes. Then we describe how IBM theory accounts for these findings.
ROADMAP FOR CHANGING STUDENT ROADMAPS

37
Because designs and underlying assumptions regarding the role of accessibility differ, we present
experimental manipulations and measurement study results separately.
ROADMAP FOR CHANGING STUDENT ROADMAPS

38

Table 2. Studies included in our review
Reference Key future “me”
characteristic
Sample

Study Design Dependent variable
Studies providing evidence for the effect of future “me” accessibility
Austenfeld & Stanton, 2008 Best possible “me” Undergraduates Experiment Health center visits
Austenfeld et al., 2006 Best possible “me” Medical school
students
Experiment Health center visits
Hershfield et al., 2012, Study 5 Similar future “me” Adults Experiment Endorsement of unethical
business decisions and tactics
Hershfield et al., 2011, Study 1 Interaction with aged
avatar
Adults Experiment Hypothetical retirement
allocation decisions
Hershfield, et al., 2011, Study
2
Interaction with aged
avatar
Adults Experiment Hypothetical retirement
allocation; temporal
discounting
Hershfield, et al., 2011, Study
3a
Interaction with aged
avatar
Adults Experiment Hypothetical retirement
allocation decisions
King, 2001 Best possible “me” Undergraduates Experiment Health center visits
Kuo et al., 2016 Interaction with weight-
reduced avatar
Undergraduates Experiment Choice/amount of unhealthy
snack
Norman & Aron, 2003 Most hoped-for and feared
possible identities
Undergraduates Measurement Self-reported motivation
van Gelder et al., 2013, Study
2
Interaction with aged
avatar
Undergraduates Experiment Lab-decision to cheat for real
monetary reward
Studies providing evidence for the effect of future “me” valence or
balance
Aloise-Young et al., 2001 Positive valence, balance 6th-9th graders Measurement Cigarette and alcohol
consumption
Anderman et al., 1999, Study 1 Positive valence 7th graders Measurement Grades
Bi & Oyserman, 2015, Study 4 Positive and negative
valence
Chinese secondary
school students
Measurement Test scores
Black et al., 2001 Negative valence Women, aged 50- Measurement Breast-cancer screening
ROADMAP FOR CHANGING STUDENT ROADMAPS

39
75
Cho et al., 2015 Positive and negative
valence
Undergraduate
English language
learners
Experiment Essay editing
Comello, 2015 Negative valence Undergraduates Measurement Marijuana use
Hoppmann et al., 2007 Positive Valence Elderly Germans Measurement Activities in a given domain
Hoyle & Sherrill, 2006 Negative valence Female
undergraduates
Experiment Interest in health activities
Ko et al., 2014 Balance Older adults Measurement Progress toward self-reported
social goal
Lee et al., 2015 Positive and negative
valence
8th graders Measurement Self-reported drinking
behavior
Murru & Ginis, 2010 Positive and negative
valence
Young adults Experiment Exercise behavior
Newberry & Duncan, 2011 Positive and negative
valence
High school
students
Measurement Delinquency
Oyserman et al., 1995, Study 4 Balance African American
8th graders
Measurement Academic performance
Oyserman & Markus, 1990 Balance High school
students
Measurement Delinquency
Oyserman & Saltz, 1993 Balance High school
students
Measurement Delinquency
Ouellette et al., 2005 Positive and negative
valence
Undergraduates Experiment Exercise behavior
Pierce et al., 2015 Negative valence 7th Grade students Measurement Self-reported delinquency
Ruvolo & Markus, 1992, Study
1
Positive valence Undergraduate
women
Experiment Effort and persistence task
Seli et al., 2009 Balance Community
college students
Measurement Self-handicapping
Yowell, 2002 Negative valence 9th grade students Measurement Dropout risk status
Studies providing evidence for the effect of future “me” self-efficacy, plausibility, fit, or linked strategies
Bi & Oyserman, 2015, Study 3 Strategies Chinese secondary
school students
Measurement Test scores, class behavior
ROADMAP FOR CHANGING STUDENT ROADMAPS

40
Bi & Oyserman, 2015, Study 4 Strategies Chinese secondary
school students
Measurement Test scores
Black et al., 2001 Self-efficacy Women, aged 50-
75
Measurement Breast-cancer screening
Destin & Oyserman, 2010,
Study 1
Fit between domain of
possible identity and
domain of behavior
Low-income 8
th

graders
Measurement Grades
Hooker & Kaus, 1994 Self-efficacy Adults Measurement Self-reported health behaviors
Ko et al., 2014 Self-efficacy Older adults Measurement Progress toward self-reported
social goal
Norman & Aron, 2003 Self-efficacy Undergraduates Measurement Self-reported motivation
Perras et al., 2016 Self-efficacy New retirees Measurement Self-reported exercise
behavior
Perras et al., 2015 Self-efficacy New retirees Measurement Self-reported exercise
behavior
Oyserman et al., 2006 Plausibility Mostly minority,
middle school
students
Experiment Grades, attendance, behavior
Oyserman et al., 2004 Strategies Middle school
students
Measurement Time spent on homework,
summer school referral,
grades
Oyserman et al., 2015, Study 1 Fit between valence of
possible identity and
context
Undergraduates Experiment Academic behaviors
Oyserman et al., 2015, Study 2 Fit between valence of
possible identity and
context
Undergraduates Experiment Planned time allocated to
academics
Oyserman et al., 2015, Study 3 Fit between valence of
possible identity and
context
Undergraduates Experiment Study time
Oyserman & Saltz, 1993 Strategies High school
students
Measurement Truancy
Strachan et al., 2017 Strategies Adults Experiment Physical activity
ROADMAP FOR CHANGING STUDENT ROADMAPS

41
Studies providing evidence for the effect of future “me” gaps
Dalley & Buunk, 2011, Study
1
Similarity/ dissimilarity Female
undergraduates
Measurement Dieting intention
Dalley & Buunk, 2011, Study
2
Similarity/ dissimilarity Female
undergraduates
Measurement Snack choice
Peetz & Wilson, 2013, Study 4 Separated by temporal
landmark
Undergraduates Experiment Health motivation
Peetz & Wilson, 2013, Study 5 Separated by temporal
landmark
Undergraduates Experiment Fitness plans
Peetz & Wilson, 2013, Study 6 Separated by temporal
landmark
Undergraduates Experiment Choice of healthy cookbook
Sobh & Martin, 2011 Dissimilarity Adult women Measurement Health Motivation
Sobh & Martin, 2011 Lack of progress Undergraduates Experiment Health Motivation
Studies providing evidence for the effect of mental contrasting
Adriaanse et al., 2013 Contrasted with present
obstacles
Diabetes patients Experiment Dieting behavior
Gollwitzer, et al., 2011, Study
1
Contrasted with present
obstacles
Elementary school
students
Experiment Quiz Performance
Gollwitzer, et al., 2011, Study
2
Contrasted with present
obstacles
Middle school
students
Experiment Quiz Performance
Oettingen et al., 2000, Study 1 Contrasted with present
obstacles and high efficacy
Middle school
students
Experiment Effort; Grades
Johannessen, et al., 2012 Contrasted with present
obstacles
Undergraduates Experiment Calorie consumption; exercise
Kappes et al., 2011, Study 5 Contrasted with present
obstacles and high efficacy
Undergraduates Measurement Energization to address
academic concern
Kappes et al., 2011, Study 6 Contrasted with present
obstacles and high efficacy
Undergraduates Measurement Persistence on study habit
task
Kappes et al, 2013, Study 1 Contrasted with present
obstacles and high efficacy
Undergraduates Experiment Self-reported study effort
Kappes et al, 2013, Study 2 Contrasted with present
obstacles and high efficacy
Undergraduates Experiment Responsibility for getting into
graduate school
Kappes et al, 2013, Study 3 Contrasted with present 10-12 year old Experiment Solving chess problem
ROADMAP FOR CHANGING STUDENT ROADMAPS

42
obstacles and high efficacy chess players
Kirk et al., 2011 Contrasted with present
obstacles
Undergraduates Experiment Mutually beneficial
negotiation outcome
Oettingen et al., 2012, Study 1 Contrasted with present
obstacles and high efficacy
Undergraduates Experiment Creative problem solving
Oettingen et al., 2012, Study 2 Contrasted with present
obstacles and high efficacy
Undergraduates Experiment Creative problem solving
Oettingen, Mayer, &
Brinkmann, 2010
Contrasted with present
obstacles
Mid-level
personnel
managers
Experiment Management at work
Oettingen, Mayer, & Thorpe,
2010
Contrasted with present
obstacles and high efficacy
University
students; smokers
Experiment Smoking behavior
Oettingen et al., 2005, Study 1 Contrasted with present
obstacles and high efficacy
Female college
students
Experiment Willingness to exert effort in
self-efficacy training
Oettingen et al., 2005, Study 2 Contrasted with present
obstacles and high efficacy
High School
students
Experiment Willingness to collaborate
Oettingen et al., 2009, Study 1 Contrasted with present
obstacles and high efficacy
Undergraduates Experiment Goal commitment
Oettingen et al., 2009, Study 2 Contrasted with present
obstacles and high efficacy
Undergraduates Experiment Presentation quality
Oettingen et al., 2001, Study 3 Contrasted with present
obstacles and high efficacy
Students Experiment Energization; immediacy of
action
Oettingen et al., 2001, Study 4 Contrasted with present
obstacles and high efficacy
Males in computer
programming
vocational schools
Experiment Energization; teacher-reported
achievement
Oettingen, Stephens et al.,
2010, Study 1
Contrasted with present
obstacles and high efficacy
Undergraduates Experiment Academic help-seeking
Oettingen, Stephens et al.,
2010, Study 2
Contrasted with present
obstacles and high efficacy
Nurses Experiment Giving Help
Sheeran et al., 2013 Contrasted with present
obstacles
Middle-aged males Experiment Physical Activity
Sevincer et al., 2014, Study 2 Contrasted with present
obstacles and high efficacy
Undergraduates Experiment Performance on letter writing
task
ROADMAP FOR CHANGING STUDENT ROADMAPS

43
Sevincer & Oettingen, 2013,
Study 1
Contrasted with present
obstacles and high efficacy
Students Measurement Goal commitment
Sevincer & Oettingen, 2013,
Study 2
Contrasted with present
obstacles and high efficacy
Adults Measurement Self-reported performance
Sevincer & Oettingen, 2013,
Study 3
Contrasted with present
obstacles and high efficacy
Undergraduates
interested in
graduate school
Measurement Graduate school essay quality
Studies providing evidence for the effect of future “me” continuity, connection, similarity, and stability
Adelman et al, 2016, Study 2 Connection, similarity Undergraduates Measurement Course Grades
Bartels & Rips, 2010, Study 1 Connection, similarity Undergraduates Measurement Temporal discounting
Bartels & Rips, 2010, Study 2 Connection, similarity Undergraduates Measurement Willingness to wait for more
“good days” at work
Bartels & Urminsky, 2015,
Study 1a
Connection, stability Adults Measurement Hypothetical spending
decisions
Bartels & Urminsky, 2015,
Study 2
Connection, identity
stability
Mechanical Turk
workers
Measurement Hypothetical spending
decisions
Bartels & Urminsky, 2015,
Study 3
Identity stability Mechanical Turk
workers
Experiment Hypothetical spending
decisions
Bartels & Urminsky, 2015,
Study 4
Identity stability Mechanical Turk
workers
Experiment Hypothetical spending
decisions
Bartels & Urminsky, 2015,
Study 5
Identity stability Adults Experiment Spending decisions
Bartels & Urminsky, 2015,
Study 6
Identity stability Coffee shop
patrons
Experiment Spending decisions
Bartels & Urminsky, 2015,
Study 7
Identity stability Mechanical Turk
workers
Experiment Hypothetical spending
decisions
Bartels & Urminsky, 2011,
Study 1
Identity stability Senior
undergraduates
Experiment Willingness to wait for larger
monetary reward
Bartels & Urminsky, 2011,
Study 2
Identity stability Young adults Experiment Willingness to wait for larger
monetary reward
Bartels & Urminsky, 2011,
Study 3
Identity Stability Undergraduates Experiment Willingness to wait for laptop
price to drop; temporal
discounting
ROADMAP FOR CHANGING STUDENT ROADMAPS

44
Bartels & Urminsky, 2011,
Study 4
Identity Stability Adults Experiment Temporal discounting
Bartels & Urminsky, 2011,
Study 5
Connection, similarity,
identity stability
Undergraduates Measurement Willingness to wait for larger
monetary reward
Blouin-Hudon & Pychyl, 2015,
Study 1
Connection, similarity University
students
Measurement Academic procrastination
Blouin-Hudon & Pychyl, 2015,
Study 2
Connection, similarity University
students
Measurement Academic procrastination
Blouin-Hudon & Pychyl, 2015,
Study 3
Connection, similarity University
students
Measurement Academic procrastination
Bryan & Hershfield, 2012 Connection, similarity Adults Measurement Retirement savings
Burum et al., 2016, Study 1 Similarity Undergraduates Experiment Willingness to leave boring
task for future "me" to finish
Burum et al., 2016, Study 2 Similarity Undergraduates Experiment Willingness to leave boring
task for future "me" to finish
Hershfield, et al., 2011, Study
3b
Similarity Adults Measurement Hypothetical retirement
allocation decisions
Ersner-Hershfield, et al., 2009,
Study 1
Connection, similarity Undergraduates Measurement Temporal discounting task
Ersner-Hershfield, et al., 2009,
Study 3
Connection, similarity Adults Measurement Self-reported financial assets
Ersner-Hershfield et al., 2008 FMRI (rACC Activation) 18-23 year-olds Measurement Temporal discounting task
Hershfield et al., 2012, Study
1a
Similarity Adults Measurement Endorsement of unethical
business decisions and tactics
Hershfield et al., 2012, Study
1b
Similarity Adults Measurement Endorsement of unethical
business decisions and tactics
Hershfield et al., 2012, Study 2 Similarity Adults Measurement Endorsement of unethical
business decisions and tactics
Hershfield et al., 2012, Study 3 Similarity Undergraduates Measurement Lab-decision to lie or cheat
for real monetary reward
Hershfield et al., 2012, Study 4 Similarity Undergraduates Measurement Lab-decision to lie or cheat
for real monetary reward
Joshi & Fast, 2013, Study 2 Connection, similarity Undergraduates Measurement Temporal discounting
ROADMAP FOR CHANGING STUDENT ROADMAPS

45
Joshi & Fast, 2013, Study 3 Connection, similarity Undergraduates Measurement Temporal discounting
Landau et al., 2014, Study 1 Connection Freshman
undergraduates
Experiment Interest in an academic
workshop
Landau et al., 2014, Study 2 Connection Freshman
undergraduates
Experiment Performance on a math task
Landau et al., 2014, Study 3 Connection Undergraduates Experiment Exam scores
Landau et al., 2014, Study 5 Connection Freshman
undergraduates
Experiment Intention to prioritize
schoolwork
Landau et al., 2014, Study 6 Connection Mechanical Turk
workers
Experiment Goal commitment
Landau et al., 2014, Study 7 Connection Undergraduates Experiment Interest in an academic
workshop
Lewis & Oyserman, 2015,
Study 3
Connection Mechanical Turk
workers
Experiment Immediacy of saving
Lewis & Oyserman, 2015,
Study 4
Connection Mechanical Turk
workers
Experiment Immediacy of saving
Lewis & Oyserman, 2015,
Study 5
Connection Mechanical Turk
workers
Experiment Immediacy of saving
Lewis & Oyserman, 2015,
Study 7
Connection Mechanical Turk
workers
Experiment Temporal discounting
Nurra & Oyserman, 2017,
Study 4
Connection, similarity 12
th
graders Experiment Grades
Pietroni & Hughes, 2016 Similarity Undergraduates Measurement Temporal discounting
Rutchick et al., 2018, Study 1 Connection, similarity Mechanical Turk
workers
Measurement Self-reported health
Sheldon & Fishbach, 2015,
Study 2
Identity Stability Undergraduates Experiment Lab-decision to lie or cheat
for real monetary reward
Zhang & Aggarwal, 2015,
Study 3
Similarity Undergraduates Experiment Donation to charity future
"me" cares about
Studies providing evidence for the effect of future “me” proximity
Evans & Wilson, 2014 Closeness Adult exercisers Measurement Self-reported exercise
Joshi & Fast, 2013, Study 4 Closeness Mechanical Turk
workers
Measurement Self-reported financial assets
ROADMAP FOR CHANGING STUDENT ROADMAPS

46
Nurra & Oyserman, 2017,
Study 3
Near vs. far Elementary school
students
Experiment Performance on math task
Nurra & Oyserman, 2017,
Study 5
Near vs. far Middle school
students
Experiment Performance on concentration
task
Peetz et al., 2009, Study 2 Near vs. far Undergraduates Experiment Self-reported academic
motivation
Rutchick et al., 2018, Study 2 Near vs. far Undergraduates Experiment Exercise
van Gelder et al, 2013, Study 1 Near vs. far Young Adults Experiment Hypothetical decision to steal
Studies providing evidence for the effect of future “me” vividness
Dalley, 2016 Clarity University women Measurement Weight loss dieting
motivation
Ellen et al., 2012 Vividness Adults, 25-55 Measurement Self-reported retirement
preparedness
van Gelder et al., 2015 Imagined along with age-
progressed photo
High School
Students
Measurement Self-reported delinquency
Macrae et al., 2017, Study 1 Imagined from 3
rd
person
perspective
Undergraduates Experiment Hypothetical spending
decisions
Macrae et al., 2017, Study 2 Imagined from 3
rd
person
perspective
Undergraduates Experiment Hypothetical spending
decisions
Strauss et al., 2012, Study 1a Clarity, ease of imagining Adults Measurement Proactive career behavior
Strauss et al., 2012, Study 1b Clarity, ease of imagining Doctoral students Measurement Proactive career behavior
Strauss et al., 2012, Study 3 Clarity, ease of imagining Doctoral students Measurement Proactive career behavior
Taber & Blankenmeyer, 2015 Clarity, ease of imagining Undergraduates Measurement Proactive career skill
development and networking

ROADMAP FOR CHANGING STUDENT ROADMAPS

47
Table 3. Future “me” characteristics and associated identity-based motivation predictions
Characteristic Operationalization IBM-based Prediction
Accessibility
Accessibility is operationalized as the state of being
“on the mind” (e.g., Bargh, 2016; Schwarz &
Strack, 2016).

Accessibility is manipulated by randomly assigning
participants to instruction group. In the accessible
condition, participants are asked to consider a
specific future identity or their future “me”
generally. In the alternative condition, participants
are asked to consider something else.

Accessibility is measured (indirectly) as response
latency (how fast participants respond), what
participants say when asked to describe their future
“me,” and from how often or how recently they
report that a future “me” (or a specific future
identity) has been called to mind.
People prefer to act and make sense of experience in
identity-congruent ways but features of the
immediate situation influence which of a person’s
identities come to mind and what these identities
seem to entail, and hence, which actions feel identity-
congruent. An accessible future “me” can be a feature
of the situation, influencing the dynamic construction
of current “me.” Accessibility is a precondition for
relevance but is not sufficient—an on the mind future
“me” can feel irrelevant to the choices facing current
“me.”
Possible Self Valence
and Balance

Valence is operationalized as positive or negative
content of future “me” (Markus & Nurius, 1986;
Ogilvie, 1987) or a particular possible identity (e.g.,
Hoyle & Sherrill, 2006; Lee et al., 2015; Murru &
Ginis, 2010; Newberry & Duncan, 2001).

Balance is operationalized as having both a positive
and a negative possible identity in the same domain
(e.g., Oyserman & Markus, 1990; Oyserman &
Saltz, 1993; Seli, Dembo, & Crocker, 1999).

People prefer to act and make sense of experiences in
identity-congruent ways, but features of the
immediate situation influence which of a person’s
identities come to mind and what these identities
seem to entail, and hence, which actions feel identity-
congruent. For current “me,” an accessible future
“me” can be a feature of the situation, and like other
features of the situation, future “me” can shape the
content of current “me.” Current “me” will take
present-focused action unless future- future “me” is
experienced as relevant to the choices it faces. Hence,
ROADMAP FOR CHANGING STUDENT ROADMAPS

48
Experiments manipulate whether possible identities
with positive or negative valence are on the mind at
the moment a choice is made. Some assess the effect
of positive or negative valence, and others compare
the relative effect of both (e.g., Aloise-Young,
Hennigan, & Leong, 2001; King, 2001; Pierce,
Schmidt, & Stoddard, 2015).
valence and balance are not essential features that
predict whether an accessible future “me” triggers
future-focused IBM. Instead, valence and balance can
change how a possible identity is experienced in the
moment. The consequence of valence and balance for
future-focused action depends on whether in context,
they trigger a difficulty-as-importance mindset and
bolster future-focused action-readiness. Whether this
occurs depends on features of the situation other than
valence and balance.
Possible Self Fit,
Plausibility,
Strategies, and
Efficacy

Fit is operationalized as the match or lack of
match—in terms of content and valence—between
the immediate situation and an on-the-mind possible
identity.

Fit can be manipulated so that people attend to the
failure- or success-likely aspects of the situation
while considering their positive (to-be-attained) or
negative (to-be-avoided) possible identities in the
same content domain.

Plausibility is operationalized as the likelihood that
one’s possible identity in a domain provides a self-
regulating path (Oyserman et al., 2004). It is scored
from open-ended responses to possible self and
strategy questions, with scores calculated based on
the number and concreteness of possible identities,
linked strategies, and the extent that strategies take
social context into account.
People prefer to act and make sense of experience in
identity-congruent ways but features of the
immediate situation influence which of a person’s
identities come to mind and what these identities
seem to entail, and hence, which actions feel identity-
congruent. An accessible future “me” can be a feature
of the situation, shaping what an accessible identity
seems to imply for action, if it feels relevant to the
choices facing current “me.” Neither fit, nor
plausibility, nor strategies, nor efficacy per se is an
essential feature for whether an accessible future
“me” or specific possible identity is experienced as
relevant to the choices facing current “me.” Instead,
what should matter is whether the future “me” is
experienced as relevant to current “me” in context.
Fit, plausibility, strategies, and efficacy can change
the way a possible identity is constructed in the
moment and this can create relevance by triggering a
difficulty-as-importance mindset and future-focused
ROADMAP FOR CHANGING STUDENT ROADMAPS

49

Strategies are operationalized as the actions people
report taking to work on their possible identities.

Strategies are scored from open-ended responses as
a count score (Oyserman, Johnson, James, 2011) or
a binary metric (Oyserman & Saltz, 1993).

Self-efficacy is operationalized as one’s belief that if
one tried, then one could take action to attain a
positive or avoid a negative possible identity and
that one’s action would yield the desired outcome
(Hooker & Kaus, 1994).

Self-efficacy is measured through self-report.
action-readiness. Misfit, low plausibility, lack of
strategies, and low efficacy also change the way a
possible self is constructed in the moment and this
can create irrelevance by triggering a difficulty-as-
impossibility mindset and undermining future-
focused action-readiness. One way that possible
identities can be made to feel relevant to the choices
facing current me is to increase experienced efficacy,
which implies that if one tries, one can overcome
obstacles (difficulties) to attaining these possible
identities. Another way that possible identities can be
made to feel relevant to the choices facing current me
is to increase their link to concrete strategies for
action and their fit with the current context.
Gaps Between
Current and Future
Selves and Progress
Addressing These
Gaps

Self-gaps are operationalized as gaps between
current “me” and possible future identities, and as
gaps in the expected progress in working toward
positive and away from negative possible future
identities (Carver & Scheier, 1982, 2016).

Control theory predicts that people automatically
monitor the gap between their current and positive
possible identities, the gap between their current and
negative possible identities, and the gap between
their expected and actual progress addressing these
gaps.
People prefer to act and make sense of experience in
identity-congruent ways but features of the
immediate situation influence which of a person’s
identities come to mind and what these identities
seem to entail, and hence, which actions feel identity-
congruent. An accessible future “me” can be a feature
of the situation, shaping what an accessible identity
seems to imply for action, if it feels relevant to the
choices facing current “me.” Self-gaps per se are not
themselves essential features for whether an
accessible possible identity is experienced as relevant
to the choices facing current “me.” Instead, what
should matter is whether the possible future identity
ROADMAP FOR CHANGING STUDENT ROADMAPS

50
is experienced as relevant to current “me” in context.
Hence, results should vary depending on other,
typically not assessed, features of the situation that
make a possible identity or future “me” feel relevant
in context. Highlighting a gap between current and
future “me” can change the way a possible identity is
constructed in the moment, as can highlighting that
progress addressing this gap is slow or fast.
Highlighting a gap or progress addressing the gap can
create relevance by triggering a difficulty-as-
importance mindset and future-focused action-
readiness, but it can also trigger a difficulty-as-
impossibility mindset and undermine future-focused
action depending on the nature of the meaning of the
gap in context.
Mental Contrasting
As A Means to
Recognize and
Address Self-Gaps

Mental contrasting is operationalized as a multi-step
processes. First, people generate an image or aspect
of themselves or their situation in the future and
some aspect of the present that stands in the way.
Second, people elaborate on the positive future and
then elaborate on present obstacles in that order.
According to mental contrasting theory, engaging in
this ordered elaboration process induces people to
consider their efficacy for taking action; future-
focused action ensues if efficacy is high.

The “mental contrast” is usually a positive future
contrasted with obstacles in the present that stand in
the way of attaining that future (e.g. Oettingen,
People prefer to act and make sense of experience in
identity-congruent ways but features of the
immediate situation influence which of a person’s
identities come to mind and what these identities
seem to entail, and hence, which actions feel identity-
congruent. An accessible future “me” can be a feature
of the situation, shaping what an accessible identity
seems to imply for action if it feels relevant to the
choices facing current “me.” By including obstacles
(difficulties to be surmounted or gotten around),
mental contrasts change the way a future identity is
constructed in the moment. Mental contrasts per se
are not essential features for whether an accessible
possible identity is experienced as relevant to the
ROADMAP FOR CHANGING STUDENT ROADMAPS

51
Stephens et al., 2010). Occasionally, the “mental
contrast” is a negative future contrasted with
positive aspects of the present that may be lost if
current behavior is continued (Oettingen, Mayer, &
Thorpe, 2010).

Whether or not a mental contrast is undertaken is
typically manipulated experimentally and
experienced efficacy is typically measured.
Responses of people assigned to the mental contrast
condition are compared to the responses of people
assigned to one or more comparison conditions.
People in these other conditions are asked to
elaborate on the future but not on the obstacles, are
asked to elaborate only the present obstacles, are
asked to elaborate on present obstacles before
elaborating the future, or are asked to elaborate on
an unrelated, control topic.
choices facing current “me.” Instead, what should
matter is whether an on the mind possible future
identity is experienced as relevant to current “me” in
context. Mental contrasts can increase the
experienced relevance of future “me” to the choices
facing current “me” if considering obstacles triggers
a difficulty-as-importance mindset. However, a
mental contrast can also sustain current focused
action by triggering a difficulty-as-impossibility
mindset, undermining future-focused action
readiness. A difficulty-as-importance mindset can be
cued directly or indirectly—by having people first
consider their possible identities and then consider
obstacles, with the implication that these obstacles
can be gotten around if not surmounted. Of course,
people might not experience obstacles in this way.
Hence, the result of mental contrasting should vary
depending on other, typically not assessed, features
of the situation that make a possible identity or future
“me” feel relevant in context.
Self-continuity (self-
connection, self-
stability, self-
similarity, proximity
of future “me,” and
vividness of future
“me”)

Self-continuity is operationalized as experiencing
current “me” and future “me” as continuous, sharing
the same fate.

Self-connection is operationalized as experiencing
current “me” and future “me” as connected and
People prefer to act and make sense of experience in
identity-congruent ways but features of the
immediate situation influence which of their
identities come to mind and what these identities
seem to entail and hence which actions feel identity-
congruent. An accessible future “me” can be a feature
ROADMAP FOR CHANGING STUDENT ROADMAPS

52
related.

Self-stability and self-similarity is operationalized as
experiencing current “me” and future “me” as
sharing core features and similarities.

Though each can be separately operationalized, the
terms are often used interchangeably in the
literature, making it nearly impossible to draw
distinctions between these constructs in practice.

For example, “continuity” is measured by averaging
participant responses about the similarity and
connection between current “me” and future “me”
(e.g., Hershfield et al., 2009) and “connection” is
manipulated by inducing participants to experience
their future “me” as similar to their current “me”
(e.g., Zhang & Aggarwal, 2015).

Proximity is operationalized as experiencing future
“me” or a possible identity as being near to or far
from current “me.”

Vividness is operationalized as experiencing future
“me” or a possible identity as being detailed, clear,
and easy to see or to imagine.
of the situation, shaping what an accessible identity
seems to imply for action if it feels relevant to the
choices facing current “me.” Self-continuity, self-
connection, self-similarity, proximity, and vividness
per se are not themselves essential features for
whether an accessible possible identity is experienced
as relevant to the choices facing current “me.”
Instead, what should matter is whether the possible
future identity is experienced as relevant to the
choices facing current “me” in context. Self-
continuity, self-connection, self-similarity, self-
proximity, and vividness can trigger future-focused
action through each of the three components of
identity-based motivation. First, via dynamic
construction, because continuity, connection,
similarity, proximity, and vividness all imply that
future “me” is part of current “me,” which implies
future-focus is identity-congruent, a “for me” thing to
do. Second, via action readiness, because continuity,
connection, similarity, proximity, and vividness
imply that taking action for future “me” will benefit
current “me.” Third, via procedural readiness,
because continuity, connection, similarity, proximity,
and vividness imply that current and future “me”
share the same fate, increasing the likelihood that
difficulties starting or sustaining future-focused
action are understood as signaling the value of this
course of action (“no pain, no gain”). However, if
continuity, connection, similarity, proximity, or
vividness is difficult to imagine, it can sustain current
focused action by triggering difficulty-as-
impossibility or by undermining action readiness.

ROADMAP FOR CHANGING STUDENT ROADMAPS

53
Accessibility
Table 3, first panel presents a brief operationalization of accessibility and the IBM-theory
prediction about when and how accessibility triggers future-focused action.
Accessibility manipulated. The logic of accessibility is that future “me” has to be on the
mind, either momentarily or chronically, to matter. We found nine experiments testing
momentary (six experiments) and chronic (three experiments) effects of accessibility. To test
whether it is accessibility of future “me” that matters, it would be necessary to document that
future “me” matters when on the mind, and not otherwise. A simple test, for example, would be
to have participants engage in a task that would allow for future-focused or present-focused
behavior either before (control) or after (experimental) having considered their future “me.”
However, as detailed next, that is not the comparison that was typically made. Instead, the
comparison was to focusing on current “me” (four experiments), to focusing on someone or
something else (two experiments), or to focus on a combination of past and current “me” (three
experiments). Focusing on future “me” increased future focused action compared to focusing on
current “me,” past “me,” or something else in all six momentary-effects and one of three
chronic-effects experiments, as summarized next.
In six experiments, researchers tested the immediate effect of momentary accessibility by
randomly assigning people to one of two groups (a future “me” group, a comparison group). In
four of these experiments, current “me” was the comparison—people randomly assigned to this
current “me” comparison group interacted with an avatar based on photograph of their present
self. In contrast, people randomly assigned to the future group interacted with an avatar based on
a digitally aged (Hershfield et al., 2011, Study1, Study 3a; van Gelder, Hershfield, & Nordgren,
2013, Study 2) or a digitally slimmed (Kuo, Lee, & Chiu, 2016) photograph of their present self.
ROADMAP FOR CHANGING STUDENT ROADMAPS

54
People focused on future “me” were more future-focused than those focused on current “me.”
They were less likely to cheat on a quiz to earn money (van Gelder et al., 2013, Study 2) and ate
less ice cream in a taste test and added less sugar to a drink given as a reward for participating
(Kuo, Lee, & Chiu, 2016). They also allocated more money to retirement savings in a
hypothetical scenario (Hershfield et al., 2011, Study1, Study 3a).
In a fifth momentary-effect experiment, the comparison was to someone else (a digitally
aged photograph of another person (Hershfield, et al., 2011, Study 2). In a sixth experiment, the
future group listed ways their future self in ten years would be similar to how they are now and
the comparison group listed what the world would be like in ten years (Hershfield, Cohen, &
Thompson, 2012, Study 5). People with an accessible future “me” were more future-focused than
people focused on something else; participants in the future “me” groups were more willing to
wait for larger rewards (Hershfield et al., 2011, Study 2) and endorsed fewer inappropriate or
unethical negotiation strategies (Hershfield et al., 2012, Study 5).
The final three accessibility experiments focused on how making a best possible identity
chronically accessible through repeated writing exercises influenced health, operationalized as
health center visits. In the first of these experiments, undergraduates were randomly assigned to
one of four groups. Two of the groups were best possible identity groups and the other two
groups were comparison groups (King, 2001). One group was asked to write about their best
possible identity for four consecutive days (20 minutes each time). The second group was asked
to write about a traumatic life event the first two days and their best possible identities the next
two days. The third group was asked to write about a traumatic life event for all four days, and
the fourth group was asked to write about their plans for the day for all four days. Students
randomly assigned to the two possible identity groups were healthier five months later than
ROADMAP FOR CHANGING STUDENT ROADMAPS

55
students randomly assigned to either of the two comparison groups. This main effect on health of
making one’s best possible future identity chronically accessible through repeated writing
exercises was not replicated in the other two chronic accessibility intervention experiments (with
medical students, Austenfeld, Paolo, & Stanton, 2006; with undergraduates, Austenfeld &
Stanton, 2008).
Accessibility measured. We found one study measuring accessibility (Norman & Aron,
2003). In this study, participants were randomly assigned to one of two groups. In one group,
participants were instructed to focus on their hoped for, positive possible identities. In the other
group, participants were instructed to focus on their feared, negative possible identities.
Participants in each group were asked to rate whether certain features were important for their
possible identities. Accessibility was operationalized as speed of response (latency) in rating
importance. More accessibility (faster response) was associated with higher motivation to attain
positive and avoid negative possible identities.
Making sense of accessibility from an IBM perspective. Identity-based motivation
theory predicts that for a future “me” to trigger future-focused action, it needs to be on the mind.
Accessibility studies test the prediction that accessibility is sufficient to produce future-focused
action. However, IBM theory predicts that accessibility is not a sufficient condition for triggering
future-focused action. Rather, IBM theory predicts that for an on-the-mind possible identity to
trigger future-focused action, the identity must activate elements of a future-focused identity-
based motivation knowledge network (dynamic construction of current “me,” action readiness,
or procedural readiness). While 8 of 10 accessibility studies yield a positive effect of
accessibility, they mostly compare on-the-mind future “me” to on-the-mind current “me.” While
useful, knowing that on-the-mind future “me” is better than on-the-mind current “me” is not a
ROADMAP FOR CHANGING STUDENT ROADMAPS

56
direct test of the sufficiency of accessibility. Without documentation that future “me” matters
when it is on the mind and not otherwise, we cannot conclude that accessibility of future “me” is
sufficient to yield future-focused action. Furthermore, as we document in the next three sections,
possible selves-based, self-gap, and self-continuity approaches all imply that accessibility is not
enough. Each approach starts with accessibility and is distinguished by which other features of
future “me” are predicted to be necessary to yield future-focused action.
The implication is that studies investigating main effects of accessibility likely include
hidden moderators. As a result of these hidden moderators, some studies will find what appears
to be a main effect of accessibility and others will not. Moderators might be in the situation or
the individual. While individual differences in propensity to be future-focused might matter,
IBM theory does not predict that future-focused action can only be triggered among people with
certain traits. Instead, from an IBM theory perspective, the source of moderators is likely to be in
the context, such that in some situations, but not others, an accessible possible identity will
trigger future-focused identity-based motivation.
Possible Selves: Valence, Balance Fit, Plausibility, Linked Strategies, and Efficacy
Table 3, second panel presents brief operationalizations of valence and balance and the
IBM-theory prediction about when and how valence and balance aspects of possible identities
trigger future-focused action. Table 3, third panel presents brief operationalizations of fit,
plausibility, linked strategies, and efficacy, and the IBM-theory prediction about when and how
fit, plausibility, linked strategies, and efficacy aspects of possible identities trigger future-focused
action.
Valence manipulated. We found five experiments manipulating whether participants
brought to mind negatively or positively valenced possible identities. Results suggest that
ROADMAP FOR CHANGING STUDENT ROADMAPS

57
valence does not matter, though possible identities do. Two experiments found an effect of
valence—one found that a positive possible identity was more likely to trigger future-focused
action (Ruvulo & Markus, 1992, Study 1) and the other found that a negative possible identity
was more likely to trigger future-focused action (Hoyle & Sherrill, 2006, Study 1). In the
remaining three experiments, possible identity valence did not affect future-focused action at all
(Murru & Ginis, 2010; Ouellette et al, 2006; Cho, 2015). Moreover, the five experiments varied
considerably in their methodology.
Ruvolo and Markus (1992, Study 1) randomly assigned college students to one of three
instruction groups (best possible self, worst possible self, positive mood). Students assigned to
the two possible selves groups differed in their outcomes. Students assigned to the group
instructed to consider their best possible self persisted longer in a boring writing task and were
more accurate in a boring editing task than students assigned to the group instructed to consider
their worst possible self. Hoyle and Sherrill (2006, Study 1) randomly assigned college students
to one of three writing groups (healthy possible identity, unhealthy possible identity, writing
directions control). Participants who wrote about an unhealthy possible identity expressed more
interest in health workshops and took more health-related informational materials; this effect
disappeared when participants underwent an “ego-depletion” manipulation. Murru and Ginis
(2010) randomly assigned participants to one of three writing instruction groups (positive
possible health identity, negative possible health identity, and health quiz control). Eight weeks
later, participants reported on their minutes of physical activity. The two possible identity groups
did not differ from each other but, when combined, differed from the control group. Ouellette
and colleagues (2005) randomly assigned students to one of four writing instruction groups
(positive possible exerciser identity, negative possible non-exerciser identity, positive
ROADMAP FOR CHANGING STUDENT ROADMAPS

58
prototypical person exerciser identity, negative prototypical person exerciser identity). The
researchers also measured consideration of future consequences. Group assignment had no effect
but being assigned to a possible identity group combined with high consideration of future
consequences yielded more exercising. Cho (2015) randomly assigned English language learner
students to one of three writing instruction groups (positive successful possible identity, negative
unsuccessful possible identity, successful past identity). Compared to the successful possible
identity group, the successful past group spent more time revising their essay. No other between-
group differences were found.
Valence measured. We found ten studies that measured possible identity valence and
one or more future-focused outcomes. Negative valence mattered in six studies, positive valence
mattered in four studies, and in two studies valence did not matter (with two studies double
counted due to multiple outcomes or ways of coding). In each of these studies, participants
reported their possible identities. Except for this common feature, neither methodology nor
results were consistent. In some studies, participants self-reported their future-focused behaviors;
in other studies school records were used as indicators of behavior. In some studies, researchers
asked participants to rate their hoped for or to-be-avoided possible identities on valence or
related characteristics (e.g. expected attainment). In other studies, researchers asked participants
to describe their possible identities and content-coded valence or asked participants to do so.
One study seemed to only measure negatively valenced possible identities (Comello,
2015) and one seemed to only measure positively valenced identities (Anderman et al., 1999,
Study 1), while the other eight elicited both positive (hoped for, desired, expected) and negative
(to-be-avoided, feared) possible identities (Aloise-Young et al., 2001; Bi & Oyserman, 2015,
Study 4; Black et al., 2001; Hoppmann et al., 2007; Lee et al., 2015; Newberry & Duncan, 2001;
ROADMAP FOR CHANGING STUDENT ROADMAPS

59
Pierce, Schmidt, & Stoddard, 2015
4
; Yowell, 2002). Two studies suggested that content, not
valence of possible identities matter but in the other studies valence mattered, just not in any
consistent pattern—which valence mattered differed across studies even when examining the
same outcomes.
In some studies, identity content, not identity valence mattered for predicting the effect of
possible identities on school attainment (Bi & Oyserman, 2015, Study 4) and social engagement
(Hoppmann et al., 2007). Rural Chinese school children students who generated more school-
focused possible identities had better test scores the next marking period, but the valence of their
possible identities did not matter. Similarly, elderly Germans who generated more positive or
negative social-relational possible identities improved their subsequent social engagement.
In some studies, negative valence mattered for predicting school attainment (Yowell,
2002), delinquency (Newberry & Duncan, 2001; Pierce et al., 2015), healthy behavior (Black et
al., 2001), and unhealthy behavior (Comello, 2015; Lee et al., 2015). American high school
students who rated their feared possible identities as more negative were more at risk of school
dropout (Yowell, 2002). American high school students who thought that negative but not
positive possible identities described them were more involved in delinquency (Newberry &
Duncan, 2001). American 7
th
graders who reported more feared possible selves content-coded as
being about delinquency reported more delinquent behaviors, an effect that was stronger if they
reported that their peers were engaged in delinquent behavior (Pierce et al., 2015). American
women in their 60s who had a feared health-related possible identity, but not a hoped for health-
related possible identity, were more likely to get cancer screenings (Black et al., 2001).
American 8
th
and 9
th
graders whose first generated feared possible identity was about school

4
In this study only results for feared possible identities were reported though both hoped for and
feared-possible identities were elicited.
ROADMAP FOR CHANGING STUDENT ROADMAPS

60
consumed less alcohol (Lee et al., 2015). American undergraduates who rated their future “me”
two years after college as more likely to do dangerous things reported more marijuana use over
the previous three months (Comello, 2015).
However, in other studies it was positive valence that mattered for predicting improved
school grades (Anderman et al., 1999, Study 1), physical activity (Hoppmann et al., 2007), and
less unhealthy behavior (Aloise-Young et al., 2001). The grade point average of American
students went up from 6
th
to 7
th
grade if they rated their expected academic possible identities as
more positive in 7
th
grade (Anderman et al., 1999, Study 1). Elderly Germans who generated
positive rather than negative health-related possible identities subsequently showed improved
physical activity (Hoppman et al., 2007). American middle school students whose expected
possible identities were coded as positive smoked fewer cigarettes and consumed less alcohol
(Aloise-Young et al., 2001). American 8
th
and 9
th
graders who generated more desired possible
identities about school consumed less alcohol (Lee et al., 2015).
Balance measured. We found six studies that measured ‘balanced’ possible identities.
‘Balanced’ possible identities in a particular domain tends to predict better outcomes in that
domain, though researchers are not always able to compute balance from content coding open-
ended responses. Having balanced school-focused possible identities is associated with higher
tests scores (middle school students, Oyserman, Gant, & Ager, 1995, Study 4), reduced risk of
delinquent involvement (high school students, Oyserman & Markus, 1990; Oyserman & Saltz,
1993), and fewer self-handicapping strategies (community college students, Seli, Dembo, &
Crocker, 2009). Having balanced social-relational possible identities is associated with better
progress toward social goals (college students, Ko, Mejia, & Hooker, 2014). When coded across
domains and scored as the percentage of reported possible identities that are balanced (rather
ROADMAP FOR CHANGING STUDENT ROADMAPS

61
than as a raw count), balance is not associated with the smoking or drinking behavior of 6
th
to 9
th

graders (Aloise-Young, Hennigan, & Leong, 2001).
Making sense of valence and balance from an IBM perspective. The 21 studies testing
effects of valence and balance yield both conflicting (inconsistent) and non-overlapping (focused
on different issues) results. This heterogeneity implies that valence and balance alone are not
sufficient to trigger future-focused action. Some studies find that having negatively valenced
possible identities is associated with taking future-focused action. Some find that having
positively valenced possible identities is associated with taking future-focused action. Some find
that having both positively and negatively valenced possible identities in the same content
domain (balance) is associated with taking future-focused action. Some find that what matters is
the content of possible identities, not valence or the balance between positively and negatively
valenced aspects of the same content. These inconsistencies suggest that an unmeasured
moderating factor not predicted from possible self, self-gap, or self-continuity approaches is at
play. IBM theory can account for this heterogeneity in results by articulating the hidden
moderator. Specifically, IBM theory predicts that it is not the specific valence of a possible
identity or whether positive and negative aspects of the same content are considered that matters,
but the likelihood that the possible identity feels relevant rather than irrelevant to the choices
facing current “me.” As detailed in Figure 4, IBM theory predicts that an accessible possible
identity will activate a future-focused IBM network if it is experienced in the moment as relevant
to the choices facing current “me.” Otherwise, activation of a present-focused IBM network will
be maintained. As can be seen in the next section, a number of possible self-based studies begin
to address this idea by considering possible identity-context fit and by considering the extent that
possible identities provide a plausible roadmap for future-focused action by being concrete,
ROADMAP FOR CHANGING STUDENT ROADMAPS

62
detailed, and linked to strategies and the social context.
Fit, plausibility, linked strategies, and efficacy manipulated. We did not find any
experiments manipulating experienced efficacy. We did five experiments that manipulated
experienced fit, or plausibility, or linked strategies. Three experiments manipulated fit between
context and possible identity (Oyserman et al., 2015, Studies 1, 2, 3). One experiment
manipulated plausibility as part of a randomized control trial intervention (Oyserman, Bybee, &
Terry, 2006). One experiment manipulated whether people generated a possible identity with
linked strategies or just generated a possible identity (Strachan et al., 2017). The four
experiments that manipulated fit and plausibility found that these aspects of possible identities
increased future-focused action. The fifth experiment, which manipulated strategies, showed less
clear results—being led to generate a possible identity had the same effect as being led to
generate a possible identity with strategies to get there.
In the fit experiments, students were randomly assigned to one of four college-context
and possible identity writing instruction groups (Oyserman et al., 2015, Studies 1, 2, 3). Whether
writing about one’s positive or negative possible future identities resulted in investment in
schoolwork depended on fit with the context—that is, whether students were instructed to
consider the college context as a context in which success was likely or unlikely. Students led to
consider college as a success-likely context were more motivated by accessible positive possible
identities, and those led to consider college as a failure-likely context were more motivated by
accessible negative possible identities—these students planned more study time and planned to
study for finals sooner.
The plausibility experiment was a randomized control trial test of an IBM intervention
with middle school students (Oyserman, Bybee, & Terry, 2006). Eighth graders were randomly
ROADMAP FOR CHANGING STUDENT ROADMAPS

63
assigned to two groups. One group was randomly assigned to the 12-session (over six weeks)
intervention and the other group to school-as-usual. The researchers created a plausibility score
from a count of the combined number of school-focused possible identities that were linked to
strategies, with extra weight given to concrete strategies and strategies situated in the students’
social context (e.g., asking friends for help). Students in the IBM intervention group scored
higher on plausibility than students in the school-as-usual group by the end of the school year.
This higher plausibility predicted positive change in school grades and attendance at the end of
both eighth and ninth grade.
The possible identity and strategy generation experiment compared college students
randomly assigned to one of three instruction groups (Strachan et al., 2017). One group of
students was instructed to generate a healthy and active possible identity. The second group was
instructed to generate this possible identity as well as strategies to attain it. The third (control)
group took a physical activity quiz. At 8-week follow-up, students randomly assigned to the
healthy and active possible identity generation groups reported more physical activity than
participants in the control group. Students randomly assigned to the two possible identity groups
did not differ, implying that generating strategies does not necessarily increase future-focused
action compared to just generating the possible identity.
Fit, plausibility, linked strategies, and self-efficacy measured. We found eleven
studies that measured fit, plausibility, linked strategies, or efficacy and examined the association
between these aspects of possible identities and future-focused action. As detailed next, each of
the eleven studies yield support for the predicted positive relationship between these aspects of
possible identities and future focused action. At the same time, since only one of the studies
assessed more than one of these measures, it is not clear if each is a distinct attribute of possible
ROADMAP FOR CHANGING STUDENT ROADMAPS

64
identities or if these theoretically distinct measures are redundant or substitutable for one another
in practice.
One study measured fit and examined its behavioral consequences, finding that when the
way students thought about their possible identities fit the context, they engaged more in future-
focused action (Destin & Oyserman, 2010, Study 1). Specifically, at the start of eighth grade,
low-income American students were asked to think about the job they would be doing in 10
years. The researchers content-coded students’ open-ended responses for whether the student
listed their job as contingent on their schooling, labeling this kind of response education-
dependent and other responses as education-independent. The dependent variable was change in
school grades from fall to spring, which the researchers obtained from school records.
Controlling for prior grades, spring grades were better among students who described their
possible jobs as education-dependent than those of students who described their possible jobs as
education-independent. The implication is that fit between possible identities and the school
context matters.
Four studies measured strategies or plausibility and examined behavioral consequences,
each using a different operationalization but all supporting the prediction that strategies increase
future-focused action. In one study, researchers asked American 13-17 year-olds in inner-city
schools or youth detention centers to describe their next year expected and feared possible
identities and then to mark with a check any of these possible identities that they currently were
“doing something about” (Oyserman & Saltz, 1993). Truancy was lower among students who
were doing something about at least one of their possible identities (Oyserman & Saltz, 1993).
In another study, researchers built on this strategy but took the next step of asking what
students were doing to work on their possible identities (Oyserman et al., 2004). They asked
ROADMAP FOR CHANGING STUDENT ROADMAPS

65
inner-city, mostly minority American eighth graders to describe their next year and feared
possible identities, and then for each one to check whether they were doing something about it,
and if so, to describe what it is. Possible selves were coded for both balance and plausibility.
Higher fall plausibility scores predicted end-of year academic outcomes obtained by school
records (grade point average, referral to summer school) and teacher report (classroom
participation), even after controlling for fall grades and participation (Oyserman et al., 2004). In
this study, plausibility was a better measure than balance or a simple count of strategies.
Bi and Oyserman (2015) started with the open-ended stems but counted the number of
strategies participants (rural Chinese middle school students) had to attain their school-focused
possible identities. Number of strategies to attain school-focused possible identities predicted
subsequent teacher-rated problem behavior (six weeks later, Bi & Oyserman, 2015, Study 3) and
subsequent exam scores, controlling for prior performance (six weeks later, Bi & Oyserman,
2015, Study 3; one year later, Bi & Oyserman, 2015, Study 4).
We found six studies measuring the consequences of experienced efficacy to attain
possible identities on current (four studies) and downstream (two studies) action. As detailed
next, results support the prediction that feeling that you can attain a possible identity can trigger
future-focused actions. Students who felt efficacious about avoiding feared health-related
possible identities smoked less (Hooker & Kaus, 1994). Retirees who felt efficacious about
attaining an exerciser possible identity reported more physical activity (Perras, Strachan, &
Fortier, 2015). Students who felt efficacious about attaining or avoiding possible identities
reported more motivation, even after controlling for accessibility (Norman & Aron, 2003).
Women who felt efficacious about avoiding a feared health-related possible identity reported
higher likelihood of getting a cancer screening (Black et al., 2001).
ROADMAP FOR CHANGING STUDENT ROADMAPS

66
In one of the studies assessing downstream action, Ko and colleagues (2014) assessed
progress toward social goals over a 100-day daily diary among sixty-year-old American adults.
Participants first described their possible identities and their efficacy to attain their social
possible identities and then logged an online daily diary to mark progress on their social goals.
Efficacy to attain social identities was associated with more progress toward social goals. In the
other study assessing downstream action, Perras and colleagues (2016) assessed physical activity
over 12 weeks (three times at intervals of 4 weeks) among recently retired Canadians who also
rated their current exerciser identity and efficacy to attain such an identity. Close-ended rating of
efficacy to attain “being a physically active retiree” and close-ended rating of the importance and
likelihood of attaining this possible future identity had indirect effects on subsequent physical
activity, controlling for baseline physical activity.
Making sense of fit, strategies, plausibility and efficacy from an IBM perspective.
Fit, strategies, plausibility, and feeling efficacious all seem to matter; yet a possible self-based
approach does not articulate why they would matter or when they do. As we review in the next
sections, self-gap and self-continuity approaches do not provide a rationale for these findings
either. IBM theory provides a synthesizing explanation. IBM theory predicts that the way a
possible identity comes to mind in context can bolster or undermine its experienced relevance to
the choices facing current “me.” Relevance can be triggered by plausibility, strategies or “fit.” In
each case the actions a possible identity cues make sense given the choices facing current “me.”
People who have plausible possible identities with more concrete and actionable strategies and
people who feel able to successfully follow their strategies are more likely generate strategies
that fit the affordances and constraints their current “me” faces in context. When the way a
possible identity comes to mind fits important features of the affordances and constraints of the
ROADMAP FOR CHANGING STUDENT ROADMAPS

67
immediate situation, the possible identity is more likely to be experienced as relevant to current
“me,” yielding future-focused action. Low efficacy can trigger irrelevance since low efficacy
implies that even if one tries, one is unlikely to be able to take the actions needed to attain a
possible identity. High efficacy should mitigate this experience. At the same time, an IBM
perspective would not require efficacy—future “me” can feel important to attain, requiring
current action, even if I am not particularly good at the skills involved.
Self-Gaps: Gaps Between Current and Future Identities, Experienced Progress Addressing
These Gaps, and “Mental Contrasting” to Highlight and Address These Gaps
Table 3, fourth panel, presents brief operationalizations of current-to-future self-gap and
experienced progress addressing a self-gap, as well as the IBM-theory prediction about when and
how a current-to-future self-gap and experienced progress addressing such a self-gap trigger
future-focused action. Table 3, fifth panel, presents brief operationalizations of mental
contrasting to highlight and address a current-to-future self-gap and the IBM-theory prediction
about when and how engaging in mental contrasting triggers future-focused action.
Self-gaps manipulated. We found four self-gap experiments—three experiments
manipulated whether or not people experienced a self-gap, and one experiment manipulated the
perceived speed of progress in changing a self-gap. As detailed next, the results partially
supported the prediction made by a self-gap approach. The complicating factor across studies is
valence. Possible identity valence mattered in spite of it not being relevant to a self-gap
approach. Moreover, which valence (positive or negative) mattered varied across studies just as
it did in the studies testing valence directly.
In two of the three self-gap experiments, researchers randomly assigned German
undergraduates to one of two healthy possible identity instruction groups and asked them to
ROADMAP FOR CHANGING STUDENT ROADMAPS

68
consider their hoped for healthy possible identity (Peetz &Wilson, 2013, Studies 4, 5).
Undergraduates in one group (no gap) were shown a timeline ending in 7 weeks (Study 4) or
were shown a calendar ending in 6 months (Study 5). The other group (self-gap) of
undergraduates got the same instructions but saw a segmenting marker—Christmas marked in
the middle of the timeline (Study 4) or holidays and weekends colorfully marked on the calendar
(Study 5). In the third self-gap experiment (Peetz & Wilson, 2013, Study 6), researchers
randomly assigned undergraduate participants to one of four instruction groups by combining the
type of health-related possible identity (hoped for or feared) with whether or not a self-gap was
featured. In the no gap instructions, undergraduates saw a timeline ending in 6 months. In the
self-gap instructions, undergraduates saw the timeline with the end of the semester marked at the
midpoint. Across the three experiments, self-gap mattered when students considered positive
possible identities. Students randomly assigned to see a self-gap reported less similarity and
overlap between their current and possible healthy identities; they had more health motivation
(Study 4), were more likely to write fitness plans (Study 5) and to try to obtain a healthy
cookbook (Study 6). The effect of segmenting the future (seeing a self-gap) on motivation was
mediated by the discrepancy between health ratings of current and positive possible identities
(Study 4) and not found for negative possible identities (Study 6).
In the self-gap progress experiment, Sobh and Martin (2011, Study 2) randomly assigned
participants to one of four instruction groups by combining the type of appearance-related
possible identity (hoped for or feared) with feedback on progress working toward or avoiding
that possible identity. Results matched self-gap prediction for participants who were randomly
assigned to imagine a feared (negative) appearance-related possible identity. In these conditions,
poor progress (negative progress feedback) was associated with more health motivation than
ROADMAP FOR CHANGING STUDENT ROADMAPS

69
good progress. However, results mismatched self-gap prediction for participants who were
randomly assigned to imagine a hoped for (positive) appearance-related possible identity. In
these conditions, poor progress was less, not more, motivating than good progress.
Self-gaps measured. We found three studies, two that measured the size of an
experienced current-to-future self-gap and one that measured the speed of progress to address the
gap between a current and a future possible identity. Each study used a mixed design so that half
of participants were randomly assigned to consider a positive, hoped for possible identity, and
half of participants were randomly assigned to consider a negative, feared possible identity.
Results generally supported the prediction, though once again there were inconsistencies and
valence unexpectedly mattered, as detailed next.
In the size of self-gap studies, researchers randomly assigned Dutch female
undergraduates to one of two instruction groups (Dalley & Buunk, 2011, Studies 1, 2). One
group was asked to imagine their hoped for body type and the other group was asked to imagine
their feared body type. Both groups then rated how similar they felt to that imagined body type.
In line with a gap framework prediction, women reported stronger intention to diet (Study 1) and
chose a healthier snack (Study 2) when they imagined a hoped for body type and rated
themselves as currently less similar to that body type, and when they imagined a feared body
type and rated themselves as currently more similar to that body type.
In the study measuring progress addressing a self-gap, researchers used a similar design
(Sobh & Martin, 2011, Study 1). In this study, one group of participants was instructed to
imagine a hoped for appearance possible identity and describe their progress attaining it. The
other group was instructed to imagine a feared appearance possible identity and describe their
progress avoiding it. Results matched self-gap prediction for participants who were randomly
ROADMAP FOR CHANGING STUDENT ROADMAPS

70
assigned to imagine a feared (negative) appearance-related possible identity. Making less
progress toward avoiding a feared appearance possible identity was associated with more
motivation to do so. However, results mismatched self-gap prediction for participants who were
randomly assigned to imagine a hoped for (positive) appearance-related possible identity. Not
making progress toward attaining a hoped for appearance possible identity undermined
motivation to do so, the opposite of what a self-gap approach would predict.
Making sense of gaps from an IBM perspective. Self-gaps can matter. In four of the
seven studies and experiments results supported the self-gap prediction that experiencing a gap
or insufficient progress addressing a gap between a current and possible identity triggers future-
focused action. However, self-gaps do not always matter in the ways the self-gap approach
predicts. In three studies and experiments, results were not consistent with a self-gap prediction
because effects differed by valence of possible identity. In one experiment, a gap was not
motivating when a feared possible identity was involved; in one study and one experiment lack
of progress working toward a desired possible identity was not motivating.
Beyond the problem with valence, the self-gap approach has another limitation, which is
that it only attempts to address a particular piece of the motivational power of a future “me” or
possible identity. It focuses only on gaps between current and future “me” and hence cannot
address why self-continuity (described below) or possible self-based efficacy, plausibility,
strategies, and fit also sometimes trigger future-focused behavior. This suggests that a hidden
moderator is at work, such that experiencing a self-gap or insufficient progress addressing that
gap is neither sufficient nor necessary to yield future-focused action.
As detailed in Figure 4, IBM theory predicts that what an accessible possible identity
implies for meaning making and action depends on whether it is experienced as relevant to the
ROADMAP FOR CHANGING STUDENT ROADMAPS

71
choices facing current “me.” A self-gap is one way that a future “me” can be experienced as
relevant to current “me.” A gap or insufficient progress addressing a gap will be motivating if
strategies to take action now are experienced as congruent with current “me” or if difficulty
imagining, starting, or persisting in future-focused action are interpreted as implying the action’s
value or importance for current “me.” It is not a self-gap or lack of adequate progress to address
the gap per se that is motivating, but rather what the gap or lack of adequate progress seems to
imply in context.
Mental contrasting manipulated. We found 23 experiments in which participants were
randomly assigned to a mental contrasting condition or to one or more alternative conditions.
With two exceptions
5
, the mental contrasting condition entailed being guided to mentally
contrast a desired future with obstacles in the present by elaborating on aspects of the future and
then elaborating on aspects of present obstacles. Participants randomly assigned to the other
condition(s) were guided to elaborate only on the future (a version of an accessibility
manipulation), elaborate only on the present, elaborate on the present before elaborating on the
future, elaborate on an unrelated control topic, or not elaborate on anything. The dependent
variable was either future-focused action or motivation to take future-focused action and was in
the same content domain as the desired future (except in Sevincer, Busatta, & Oettingen, 2014,
Study 2).
In 17 of the 23 experiments, the same procedure was used in the mental contrasting
condition; participants reported their efficacy to attain a particular desired future, listed aspects
of that desired future and of present obstacles, elaborated on aspects of that desired future and

5
In two studies some participants mentally contrasted an undesired negative future and a positive
current situation (Oettingen, Mayer, & Thorpe, 2010; Oettingen, Mayer, Thorpe, Janetzke, &
Lorenz, 2005, Study 2).
ROADMAP FOR CHANGING STUDENT ROADMAPS

72
then elaborated on aspects of present obstacles. In the other six experiments, the efficacy step of
the procedure was varied: In one experiment, efficacy was regarding a behavior unrelated to the
dependent variable (Sevincer et al., 2014, Study 2). In two experiments, efficacy was
operationalized as receipt of positive feedback about performance (Oettingen, Marquardt, &
Gollwitzer, 2012, Studies 1, 2). In three experiments, efficacy was not assessed (Sheeran, Harris,
Vaughn, Oettingen, & Gollwitzer, 2013; Kirk, Oettingen, & Gollwitzer, 2011; Oettingen, Mayer,
& Brinkmann, 2010).
In each experiment, participants were randomly assigned to the mental contrast
instruction group or to alternative instructions. To test the prediction that mental contrast
instructions yield more future-focused action than accessibility alone, the critical comparison is
to being instructed to elaborate on one’s future “me.” Nineteen of the 23 experiments included
this instruction group
6
. Seven experiments yielded results that fully support a mental contrast
prediction and 12 yielded partially supporting results, with efficacy being the complicating
factor. That is, in seven of the 19 experiments people randomly assigned to the mental contrast
condition engaged in more future-focused action if they were high in efficacy compared to
people randomly assigned to elaborate only on the future or compared to people who first
elaborated on obstacles in the present and then elaborated on the future. In another six of the 19
experiments, the positive effect in the high efficacy group was mirrored by a negative
(undermining) effect of being in the mental contrasting group for participants low in efficacy,
implying that mental contrasting can be deleterious compared to just focusing on future “me.” In
another two of the 19 experiments, people randomly assigned to the mental contrasting condition

6
In three experiments, instructions were to elaborate on the future and also on obstacles, but not
in the order specified by mental contrast theory; in 16 experiments, instructions were to elaborate
on the future.
ROADMAP FOR CHANGING STUDENT ROADMAPS

73
engaged in more future-focused action than people randomly assigned to elaborate only on the
future, but efficacy was not measured. In the final four of the 19 experiments, efficacy was
measured but did not matter, implying that the role of efficacy in mental contrasting is not fully
understood.
First, we review the seven experiments in which mental contrasting was better than
focusing on the future for those high in efficacy and had no negative effects for those low in
efficacy: German and American college students who received positive feedback on their
creative ability and were led to mentally contrast performed better on a creativity task than
students who elaborated only on the future (Oettingen et al., 2012, Studies 1, 2). German college
students led to mentally contrast who felt efficacious about benefiting from a self-efficacy
training program were more likely to exert effort to participate in the training program than
students who elaborated only on the future (Oettingen et al., 2005, Study 1). German high school
students led to mentally contrast a negative future with a positive present and who felt more
efficacious about helping foreigners integrate reported more effort towards building relationships
with immigrants than students led to elaborate only on a negative future (Oettingen et al., 2005,
Study 2). German college students led to mentally contrast who felt efficacious about
successfully resolving an interpersonal problem reported more commitment to solve that
problem than students led to elaborate only on the future (Oettingen et al., 2009, Study 1).
American university students led to mentally contrast who felt efficacious about getting their
desired grade in a class reported more effort studying than students randomly assigned to
elaborate on the present before elaborating on the future (Kappes, Wendt, Reinelt, & Oettingen,
2013, Study 1). Youth chess players randomly assigned to mentally contrast who felt efficacious
about succeeding in chess were more likely to solve a chess problem than youth players assigned
ROADMAP FOR CHANGING STUDENT ROADMAPS

74
to elaborate on the present before elaborating on the future (Kappes et al., 2013, Study 3).
In six experiments, efficacy moderated the effect of being assigned to mentally contrast
rather than elaborate only on the future—mental contrasting benefited those high in efficacy and
harmed those low in efficacy. German undergraduates randomly assigned to mentally contrast a
positive (negative) future and negative (positive) aspects of the present were more likely to take
immediate steps to quit smoking if they felt efficacious about quitting and less likely to do so if
they felt inefficacious about it (Oettingen, Mayer, & Thorpe, 2010). German undergraduates
randomly assigned to mentally contrast gave better presentations about their professional skills
than those randomly assigned to elaborate on the future if they felt efficacious about doing the
presentation and worse if they felt inefficacious about it (Oettingen et al, 2009, Study 2). German
undergraduates randomly assigned to mentally contrast were more likely to seek academic help
than those randomly assigned to elaborate on the future if they felt efficacious about receiving
help and the reverse if they felt inefficacious (Oettingen, Stephens, et al., 2010, Study 1).
German nurses randomly assigned to mentally contrast made more effort to improve
communication with patients’ families if they felt efficacious about doing so and less effort if
they felt inefficacious about it than nurses assigned to elaborate on the future (Oettingen,
Stephens, et al., 2010, Study 2). American undergraduates randomly assigned to mentally
contrast on an intelligence test self-reported better empathic letter-writing than students assigned
to elaborate on the future if they felt efficacious about the task and worse performance if they felt
inefficacious (Sevincer et al., 2014, Study 2). American undergraduates led to mentally contrast
about getting into their desired graduate school felt more responsible for getting in compared to
students randomized to elaborate on the present before the future if they felt efficacious about
getting in and the reverse if they inefficacious than students who elaborated on a control topic
ROADMAP FOR CHANGING STUDENT ROADMAPS

75
(Kappes, et al., 2013, Study 2).
In two experiments efficacy was not assessed and being assigned to mentally contrast
rather than to elaborate only on the future predicted more future-focused action. American
undergraduates assigned to mentally contrast reported more mutually beneficial outcomes in a
negotiation task than students assigned to elaborate on the future (Kirk et al., 2011). German
mid-level managers assigned to mentally contrast were more likely to have good time and project
management than those assigned to elaborate on the future (Oettingen, Mayer, & Brinkmann,
2010).
In four experiments being assigned to the mental contrast rather than the elaborating only
on the future condition was associated with more future-focused action, but assessed efficacy did
not matter. German elementary school students (Gollwitzer et al., 2011, Study 1) and American
middle school students (Gollwitzer et al., 2011, Study 2) led to mentally contrast performed
better on a vocabulary quiz that students who elaborated only on the future. Dutch diabetes
patients randomly assigned to mentally contrast were more likely to adopt a healthy diet than
participants who elaborated only on the future (Adriaanse, De Ridder, & Voorneman, 2013).
American undergraduates interested in eating healthier randomly assigned to mentally contrast
reported consuming fewer calories and being more physically active over the next two weeks
than those assigned to elaborate only on the future (Johannessen, Oettingen, & Mayer, 2012).
The final four of the 23 experiments omitted a condition in which participants elaborated
only on the future and so it is not possible to tell if positive effects of being assigned to the
mental contrasting condition were due to mental contrasting per se or simply due to accessibility
of future “me.” In three of these experiments, participants randomly assigned to mentally
contrast were compared to a single pooled comparison group combining participants randomized
ROADMAP FOR CHANGING STUDENT ROADMAPS

76
to different conditions (Oettingen et al., 2001, Study 3; Oettingen et al., 2001, Study 4; Oettingen
et al., 2000). This pooling makes it impossible to know if positive effects are due to mental
contrasting per se or simply to accessibility of future “me.” In the fourth of these experiments,
participants assigned to mentally contrast were compared to a control group that did not
elaborate on the future or present obstacles (Sheeran et al., 2013). These experiments
(summarized in Section 1 of Supplemental Materials) can be understood as showing that
accessibility of future “me” matters.
Mental contrasting measured. We found five studies that measured whether
participants mentally contrasted and also assessed their efficacy to attain their positive desired
future (Kappes, Oettingen, Mayer, & Maglio, 2011, Studies 5, 6; Sevincer & Oettingen, 2013,
Studies 1, 2, 3). In all five, participants who mentally contrasted were compared to a pooled
group of participants who engaged in other processes, including focusing only on the future,
focusing only on obstacles, or writing about obstacles before the future. This pooling makes it
impossible to know if positive effects are due to mental contrasting per se or simply to
accessibility of future “me.” Hence, these studies (summarized in Section 2 of Supplemental
Materials) can be understood as showing that accessibility of future “me” matters.
Making sense of mental contrasting gaps from an IBM perspective. Mental
contrasting can matter. In 13 experiments, people high in efficacy who mentally contrasted were
more future-focused than people who focused only on their futures. Though the approach
specifies that high efficacy is necessary, the role of efficacy is underspecified since it does not
always matter and low efficacy can sometimes make mental contrasting worse than other
approaches that aim to trigger future-focused action. Thus, in six of these 13 experiments, people
low in efficacy who mentally contrasted became less future-focused. Moreover, in six additional
ROADMAP FOR CHANGING STUDENT ROADMAPS

77
experiments, people who mentally contrasted were more future-focused than people who focused
only on their future, but in spite of the prediction that mental contrasting only helps people high
in efficacy, efficacy was not assessed or did not matter. Taken together, these 19 experiments
imply that 1) for some people high in efficacy mental contrasting can be better than just focusing
on the future, and 2) for other people low in efficacy, mental contrasting can be worse than just
focusing on the future.
In the remaining four experiments and in all five of the measurement studies, results are
significant but what they mean is unclear. That is because the results for people assigned to the
mental contrasting group are compared to a pooled mean of people who were assigned to focus
on the present as well as people who were assigned to focus on the future, or compared to people
who did not focus on the future at all. Note that documenting that being assigned to mentally
contrast increases future-focused motivation compared to not thinking about the future at all
simply shows that thinking about future “me” matters and not that mental contrasting per se
mattered.
Beyond the problems of when efficacy matters and inconclusive comparisons, the mental
contrast approach has another limitation, which is that it only attempts to address a particular
piece of the motivational power of a future “me.” It focuses on considering gaps between current
and future “me” in a certain way and for a certain group of people (the efficacious). It cannot
address why self-continuity (described below) or possible self approaches involving efficacy,
plausibility, strategies, and fit also sometimes trigger future-focused behavior without a mental
contrast. This suggests that a hidden moderator is at work, such that mental contrasting is neither
sufficient nor necessary to yield future-focused action.
As detailed in Figure 4, IBM theory provides a synthesis to bridge from the findings of
ROADMAP FOR CHANGING STUDENT ROADMAPS

78
mental contrasting to the rest of the literature. IBM theory predicts that what an accessible
possible identity implies for meaning making and action depends on whether it is experienced as
relevant to the choices facing current “me.” One way that a future “me” is experienced as
relevant to current “me” is that the interpretation of experienced difficulty it cues creates the
sense that taking action now is important for current “me,” no matter how much effort those
actions may require (a difficulty-as-importance interpretation). However, a future “me” can also
be experienced as relevant to current “me” if the interpretation of difficulty it cues creates a
sense that taking future action is not impossible for current “me” (a low difficulty-as-
impossibility interpretation). The implication is that it is not the mental contrast per se that is
motivating but what the mental contrast process seems to imply in context. In the context of
considering a possible identity, considering obstacles to attaining that identity is likely to trigger
a motivating interpretation of difficulty as a signal that the odds are not low—if one feels
efficacious about overcoming those obstacles. In contrast, considering obstacles to attaining that
identity is likely to trigger a demotivating interpretation of difficulty as signaling low odds of
success if one feels inefficacious about overcoming those obstacles. Moreover, IBM theory
predicts that an interpretation of difficulty as importance can be triggered directly; hence, neither
efficacy, nor the mental contrast process is necessary for future-focused action.
Self-Continuity
Table 3, bottom panel, presents brief operationalizations of self-continuity, self-connection, self-
stability, self-similarity, proximity and vividness of future “me,” and the IBM-theory prediction
about when and how they trigger future-focused action.
Self-continuity, self-connection, self-stability, and self-similarity manipulated. We
found 24 “self-continuity” experiments that manipulated whether current “me” and future “me”
ROADMAP FOR CHANGING STUDENT ROADMAPS

79
were experienced as continuous, connected, stable, or similar to current “me”.
7
The self-
continuity approach prediction that it is an experience of continuity between current and future
“me” (rather than the simple accessibility of future “me”) that yields future-focused action. To
test this prediction—that self-continuity instructions yield more future-focused action than
accessibility alone—researchers would need to compare results from participants who were
randomly assigned to consider future “me” and those randomly assigned to consider current and
future “me” as continuous, connected, stable, or similar. In practice, this critical comparison was
made in only seven of the 24 self-continuity experiments (Landau et al., 2014; Lewis &
Oyserman, 2015; Nurra & Oyserman, 2018). In the other 17 experiments, researchers randomly
assigned some participants to experience continuity, connection, stability, or similarity but
omitted the accessible future “me” condition and replaced it with a self-discontinuity condition
in which current and future “me” were presented as disconnected or future “me” was presented
as unstable.
In six of the seven experiments that included an accessible future “me” comparison
group, people assigned to future “me” continuity conditions engaged in more future-focused
action than people assigned to the accessibility condition alone. In one of these experiments,
connection was not significantly better than accessibility.
In the first two of these experiments, researchers primed connection with a path image,
disconnection with images of houses or boxes, and accessibility with no image (Landau et al.,
2014). Specifically, American first year college students were randomly assigned to imagine

7
Though each can be operationalized separately, the terms are often used interchangeably in the
literature, making it nearly impossible to draw distinctions between these constructs in practice.
For example, in some studies “continuity” is measured by averaging similarity and connection
between current “me” and future “me” (e.g., Hershfield et al., 2009) and in some “connection” is
manipulated by inducing participants to experience future and current “me” as similar (e.g.,
Zhang & Aggarwal, 2015).
ROADMAP FOR CHANGING STUDENT ROADMAPS

80
their best academic possible selves during the college years in the context of a connecting
metaphor (writing on a picture of a path), a disconnecting metaphor (separate boxes), or no
metaphor (regular lined paper). Students assigned to write about academic identities on a picture
of a path reported more interest in an academic workshop (Study 1) and greater intention to
prioritize schoolwork (Study 5) than students assigned to write on lined paper. Study 5 measured
experienced connection, revealing that the path manipulation worked by increasing experienced
connection to a possible identity, as measured on a 5-item scale.
In the next four of these seven experiments, researchers randomly assigned participants to
use the default (accessibility) or a more fine-grained (connection) time metric when thinking
about the future (Lewis & Oyserman, 2015, Studies 3, 4, 5, 7). Specifically, American adults
were randomly assigned to imagine their newborn being ready for college in 6,570 days or in 18
years (Studies 3, 7). Other American adults were randomly assigned to imagine wanting to retire
in 10,950 days or in 30 years (Studies 4, 7). Other American adults were randomly assigned to
imagine wanting to retire in 14,600 days or in 40 years (Study 5). Results supported the
prediction that connection is a better predictor of future-focused action than accessibility. People
randomly assigned to the connection (days) condition said they would start saving sooner
(Studies 3, 4, 5) and were more willing to wait for larger rewards (Study 7) than people
randomly assigned to the accessibility alone (years) condition. Connection increased future-
focused action even though it did not affect the importance or closeness of the future identity
(Study 6). The researchers also measured connection to determine whether it was a mediator
(Study 7). People who were randomly assigned to the connection (days) condition felt more
connected to their future identity as a retiree or as a parent of a college student than did people
who were randomly assigned to the control (years) condition. Experienced connection mediated
ROADMAP FOR CHANGING STUDENT ROADMAPS

81
the effect of being in the connection (days) rather than the accessibility alone (years) condition
on willingness to wait for larger monetary rewards.
In the last of these seven experiments, French high school students were randomly
assigned to one of three groups (Nurra & Oyserman, 2017, Study 4). One group saw a set of
overlapping circles labeled “what I am now” and “what I want to be.” They were instructed to
consider connection and similarities between their current and adult selves. A second group saw
a set of separated circles labeled “what I am now” and “what I want to be.” They were instructed
to consider disconnection and dissimilarity between their current and adult selves. A third group
saw the full set of circles pairs that ranged from very separated to very overlapping. They were
instructed to choose the pair that best represented themselves and then to describe their adult self.
Because they were not told to consider their future “me” as either connected or disconnected, the
third group is the critical comparison group. The dependent variable was course grades in the
subsequent marking period. Content analysis revealed that all students wrote about the jobs they
would have, and while what they wrote was the same, what it implied differed. Students who
were randomly assigned to the control condition had grades midway between students assigned
to the connection condition and students assigned to the disconnection condition; connection was
significantly better than disconnection but control did not differ significantly from either. The
implication is that without being primed, some students considered their future and current “me”
as connected and some students considered their future and current “me” as disconnected.
The other 17 experiments suggest that continuity is better than discontinuity (ten
experiments) or sometimes better (seven experiments). In these experiments, participants were
assigned to a continuity instruction or a discontinuity instruction. While interesting, this
comparison cannot shed light on whether continuity increases future-focused action compared to
ROADMAP FOR CHANGING STUDENT ROADMAPS

82
accessibility alone, making this an inconclusive test of the predictive power of self-continuity.
Without an accessible future “me” condition, it is not possible to tell if results are due to an
undermining effect of the discontinuity instruction or to a bolstering effect of the continuity
instruction. The few studies that include accessibility, continuity, and discontinuity suggest that
people do not chronically experience an accessible future “me” as discontinuous with current
“me.” The interested reader can find summaries of each of these experiments in Section 3 of the
Supplemental materials (self-continuity is better than self-discontinuity, Landau et al., 2014,
Studies 2, 4, 6; self-stability is better than self-instability, Bartels & Urminsky, 2011, Studies 1,
2, 3, 4; self-similarity is better than self-dissimilarity, Zhang & Aggarwal, 2015, Study 3;
dissimilarity is worse than an accessible future “me,” Burum et al., 2016, Studies 1, 2; self-
continuity sometimes matters, Bartels & Urminsky, 2015, Studies 3, 4, 5, 6; Bartels & Urminsky,
2015, Study 7; Landau et al., 2014, Study 7; Sheldon & Fishbach, 2015, Study 2).
Self-continuity, self-connection, self-stability, and self-similarity measured. We
found 23 studies that measured the relationship between experienced continuity of current and
future “me” and future-focused action. Assessed continuity mattered—20 studies reporting a
main effect and three reporting an effect if people’s attention was directed to continuity.
Continuity was measured similarly across studies—only two used a measure that was not partly
based on Aron, Aron and Smollan’s (1992) overlapping circles measure (see Figure 1). Eighteen
of the 20 studies finding a main effect of continuity used a variant of this measure; in 14 of these
studies, the measure was one or two sets of overlapping circles; in the other four of these studies,
other items were added to the overlapping circles measure
8
.

8
In eight studies participants were shown a single series of circle pairs that varied in overlap and
asked to choose the pair describing how similar they felt to their future self (seven studies) or to
describe how similar and connected they felt to their future self (one study). In six studies
ROADMAP FOR CHANGING STUDENT ROADMAPS

83
Participants who selected a circle pair with more overlap (self-similarity) endorsed fewer
unethical decisions in hypothetical business scenarios (Hershfield, Cohen, & Thompson, 2012,
Studies 1a, 1b, 2). They lied and cheated less to earn monetary rewards in lab tasks (Hershfield et
al., 2012, Studies 3, 4). They chose fewer sooner but smaller rewards in a temporal discounting
task (Pietroni & Hughes, 2016) and allocated more hypothetical income to retirement savings
(Hershfield et al., 2011, Study 3b). Adults who selected a circle pair with more overlap (self-
similarity and self-connection) were less likely to choose a few immediate good days at work
over more good days at work at a later time (Bartels & Rips, 2010, Study 2). College students
who selected two circle pairs with more overlap (self-similarity, self-connection) had higher
course grades (Adelman et al., 2016, Study 2) and procrastinated less on their academic work
(Blouin-Hudon & Pychyl, 2015, Studies 1, 2, 3). Adults who selected two circle pairs with more
overlap were more willing to wait for larger rewards in a temporal discounting task (Joshi &
Fast, 2013, Studies 2, 3). Adults who selected two circle pairs with more overlap and cared about
and liked their future “me” more were willing to wait for larger rewards in a temporal
discounting task (Ersner-Hershfield et al., 2009, Study 1). They reported having more retirement
savings, (Ersner-Hershfield et al., 2009, Study 3) and they reported better health (Rutchick,
Slepian, Reyes, Pleskus, & Hershfield, 2018, Study 1). Adults who selected a circle pair with
more overlap, rated themselves as more similar and connected to their future self on a 0-100
scale, and chose more overlap in a binary response item were more willing to wait for larger

participants were asked to use one set of circle pairs to indicate similarity and a second set to
indicate connection. The additional items added in the other four studies were also similar: In
one study, there were three items: overlapping circles measuring future self connection, a variant
of the circles as a bipolar item with the most overlapping pair on one side and the fully separated
pair on the other, and an item that asked about future self similarity and connection on 0-100
response-scale. In the other three studies, the two overlapping circles items were combined with
two other items: 1) how much do you care about and 2) how much do you like future “me.”
ROADMAP FOR CHANGING STUDENT ROADMAPS

84
rewards in a temporal discounting task (Bartels & Urminsky, 2011, Study 5).
In the other 2 of the 20 main effect studies, continuity was assessed with other measures.
Bartels and Rips (2010, Study 1) asked participants to rate how connected and similar they felt to
their future self on a 0-100 scale. Higher connection and similarity were associated with more
willingness to wait for larger rewards in a temporal discounting task. Ersner-Hershfield and
colleagues (2008) measured continuity using fMRI. In this study continuity was assessed by the
difference in patterns of rostral anterior cingulate (rACC) activation when participants thought
about current “me” relative to future “me”—that is, continuity was assessed based on the extent
to which thinking about future “me” activated neural patterns similar to thinking about current
“me” or similar to thinking about another person. When the future “me” activated pattern was
more similar to the current “me” activated pattern, participants were more willing to wait for
larger rewards in a temporal discounting task.
In the final three of the 23 studies, reporting greater self-continuity was associated with
more future-focused behavior only under particular conditions. In two of these studies people
who scored higher in connection (measured as connection circle overlap and connection to future
self on a 0-100 scale) were only more likely to avoid spending money if the opportunity cost of
spending was made salient (Bartels & Urminksy, 2015, Studies 1a, 2). In the third study, people
who scored higher in continuity (measured with two circle overlap items) were later more likely
to increase their retirement saving rates if they were led to consider saving as social
responsibility to future “me” rather than as self-interest (Bryan & Hershfield, 2012).
Making sense of self-continuity, self-connection, self-stability, and self-similarity
from an IBM perspective. Self-continuity can matter. Being assigned to consider current and
future “me” as continuous triggers more future-focused action than being led to consider future
ROADMAP FOR CHANGING STUDENT ROADMAPS

85
“me” alone in six of seven experiments in which accessibility was separately manipulated.
People who experience more continuity between their present “me” and their future “me” report
taking more future-focused action in 20 of the 23 measurement studies. Because 17 of 23
experiments did not include an accessible future “me” group, this large body of research does not
yet address the conditions in which people take future-focused action when future “me” is on the
mind but a specific cue of continuity is not presented. As detailed in Table 3, IBM theory
provides a synthesis, articulating why continuity cues matter in the context of a broader
formulation of when future “me” is experienced as relevant to the choices facing current “me.”
Specifically, IBM theory predicts that continuity cues could activate future-focused IBM in one
of three ways: First, these cues can shape dynamic construction of current “me” to include
elements of future “me.” Second, these cues can make future-focused action feel “for me” in the
current context. Third, these cues can make difficulty imagining, starting, or sustaining future-
focused action be experienced as a signal of the value of future-focused action for current “me.”
Proximity manipulated. We found five experiments that manipulated experienced
proximity of future “me.” Two of these five experiments (Nurra & Oyserman, 2018, Studies 3,
5) included an accessibility condition to test the difference between being assigned to consider
future “me” (accessibility) and being assigned to consider future “me” as near or as far
(proximity). In these studies, French children were randomized to three groups. One group was
instructed to imagine their “near” adult future self, another group was instructed to imagine their
“far” adult future self, and the third group was instructed to imagine their adult future self
without a modifier. All groups wrote about jobs that they wanted to have as adults. Proximity
mattered: Students in the future “me” is “near” condition solved more problems on a subsequent
math task (Study 3) and performed better on a timed concentration task (Study 5) compared to
ROADMAP FOR CHANGING STUDENT ROADMAPS

86
students in the future “me” is “far” condition. This latter effect only occurred among students
who agreed that school was the path to attaining their adult future self. However, whether
proximity mattered more than accessibility was not clear because students in the accessibility
condition performed midway between students in the near and far conditions and their
performance did not significantly differ from either of those two groups (Studies 3, 5).
The other three experiments did not include an accessible future “me” condition, which
means that they address the question of whether proximity is better than distance, but not the
question of whether either of these conditions is better than accessibility alone. These
experiments are summarized in Section 4 of the Supplemental Materials (proximity triggers
future-focused action compared to distance Peetz, Wilson, & Strahan, 2009; distance triggers
future-focused action compared to proximity Rutchick et al., 2018, Study 2; van Gelder,
Hershfield, Nordgren, 2013, Study 1).
Proximity measured. We found two studies that measured proximity to future “me.”
Evans and Wilson (2014) instructed Canadian students to imagine their ‘exerciser’ self one year
into the future, report how close in time they felt to that future identity, and then report their
exercise intentions. Four weeks later, participants who intended to exercise reported exercising
more, but only if they felt close to their next year future “me.” Joshi and Fast (2013, Study 4)
asked participants to rate how close they felt to their future “me” by rating the extent to which
their future felt temporally close to the present. Feeling that the future was closer was associated
with more self-reported retirement savings.
Making sense of proximity from an IBM perspective. Three experiments and two
measurement studies provide evidence that experiencing future “me” as proximal (near) rather
than distal (far) from current “me” enhances future-focused action and two experiments seem to
ROADMAP FOR CHANGING STUDENT ROADMAPS

87
show the reverse. Proximity was not significantly different from accessibility in the two
experiments in which this was tested. In these experiments, people assigned to consider future
“me” as near took more future-focused action than people assigned to consider future “me” as
far. People assigned to consider future “me” without a proximity cue were midway between the
two proximity groups and did not significantly differ from either. The implication is that once
future “me” is on the mind, proximity cues may not be necessary to trigger future-focused action.
The two experiments that seemed to show that proximity undermines future-focused action used
a different method (Rutchick et al., 2018, Study 2; van Gelder, Hershfield, Nordgren, 2013,
Study 1). Rather than randomly assign participants to consider the same future identity as near or
far in the future, people were randomly assigned to consider themselves in twenty years or in
three months. In this case, imagining a more distal future “me” was more motivating of future-
focused action. Why might both considering future “me” as near and a temporally distal future
“me” both increase future-focused action, at least sometimes?
Identity-based motivation theory provides a synthesis across these results and the results
of studies from possible self and self-gap approaches by suggesting that it is not how near or far
away future “me” is or is experienced to be per se that matters. What matters is that future “me”
be experienced in a way that makes it relevant to the choices facing current “me.” Proximity can
make future “me” feel relevant by implying that the future is now so now is the time to act.
However, distance can also make future “me” feel relevant because task-value can be triggered
by distance, as described in temporal construal theory, which we describe in detail in Part 4.
Vividness manipulated. We found two experiments that manipulated the vividness with
which future “me” is experienced (Macrae et al., 2017, Studies 1, 2). Participants were randomly
assigned to one of two instruction groups. Both groups were asked to imagine themselves in 40
ROADMAP FOR CHANGING STUDENT ROADMAPS

88
years walking on the beach. In the accessible future “me” group the instructions did not provide
further specification and in the vivid future “me” group the instructions were to imagine that they
could see their future self taking this walk. Being assigned to the vivid condition increased the
amount of a hypothetical $1,500 windfall participants earmarked for savings.
Vividness measured. We found seven measurement studies that assessed the relationship
between the vividness with which future “me” is experienced and future-focused action. In one
study, adults rated how vividly they experienced their “retiree” possible future identity and then
reported how financially prepared they were for retirement (Ellen, Weiner, & Fitzgerald, 2012).
Participants who experienced future “me” more vividly reported being more financially prepared
for their retirement. In three other studies (Strauss, Griffin, & Parker, 2012, Studies 1a, 1b, 3),
adults (Mechanical Turk, Study 1a) and doctoral students (Studies 1b, 3) were asked how easy it
was to imagine their “professional” possible future identities and then asked about their career-
focused behaviors (e.g., networking, sharing career goals with supervisors). People who reported
that it was easier to imagine their professional future identities reported more career-focused
behaviors. In a fifth study, undergraduates reported how easy it was to imagine their “ work”
possible future identities and then reported on their skill development and career networking
activities (Taber & Blankenmeyer, 2015). Students who reported that it was easier to imagine
their work future identities reported doing more to develop their work skills and more career-
focused networking. In a sixth study, female undergraduates reported how clearly they could
imagine their hoped for or feared future “body” and how often it came to mind on 2-item scale
(Dalley, 2016). Higher clarity and frequency of feared or hoped for “body” future identity was
associated with more motivation to engage in weight-loss dieting.
ROADMAP FOR CHANGING STUDENT ROADMAPS

89
In the seventh study Dutch high school students were randomly assigned to one of two
groups—an accessible future “me” group and current “me” group (van Gelder, Luciano,
Kranenbarg, & Hershfield, 2015). Change in vividness of future “me” and change in delinquent
involvement were measured. In the future “me” group, researchers set up Facebook pages using
aged digital avatars that sent student messages asking about what they would be doing on future
days. In the current “me” group, researchers set up Facebook pages using an avatar that was not
digitally aged and sent messages asking students about their behavior in the present. After seven
days of messages, students reported the vividness of their future “me” and then a week later
reported on their delinquent involvement that week. Group assignment had an indirect effect;
students who had an increase in experienced vividness reported less delinquency and students in
the future “me” condition were more likely to experience this change in experienced vividness.
Making sense of vividness from an IBM perspective. The results of two experiments
and seven studies suggest that vividness, like proximity, can trigger future-focused action.
Though the specific process by which vividness should influence future-focused action is not
addressed in these studies, vividness, like proximity, can be a cue of nearness. IBM theory
provides a synthesizing framework in which any feature of an accessible future “me” that implies
that future “me” is relevant to the choices facing current “me” is likely to trigger future-focused
action. Things that are close can be seen vividly and are more likely to be relevant to current
“me,” thereby requiring action even if that action is difficult to start or to sustain. But it is not
nearness or vividness itself that matters.
Part 4: Broadening the Lens
Making Future “Me” Count
Failure to take future-focused action is consequential. It results in short-term strategies
ROADMAP FOR CHANGING STUDENT ROADMAPS

90
that undermine long-term well-being, health, and happiness. When people are present-focused,
they are less likely to experience current difficulties and suffering as meaningful investments,
and hence they are more likely to miss chances to make progress toward future “me” goals that
require persistent investment. When people are present-focused, they exercise too little and eat
too much; they invest too little in studying, and save too little for retirement. At the same time,
focusing on the present is not an error. People need to be sensitive to the affordances and
constraints of their immediate situations. After all, the present is certain and its demands must be
attended to. Unlike the present, the future is uncertain and probabilistic—it may or may not
unfold in a given way. The problem is that if people are to attain their long-term goals—whether
of learning, preparing for a career, maintaining health, or saving money—people have to take
future “me” into account and experience future “me” as relevant to the choices current “me”
faces.
We conducted a systematic review of the literature to understand when that happens,
starting with Markus and Nurius’ (1986) seminal formulation of a possible selves approach,
which implied that positive possible identities are all people need. We asked whether this insight
is supported in the English-language literature that reports on a measured or manipulated future
“me” and a measured behavioral outcome. We found a diverse literature, 81 papers describing
139 experiments or measurement studies and 145 relevant analyses, about a third of which
seemed rooted in a possible self-based approach and the rest rooted in what we labeled self-gap
based and self-continuity based approaches. Just as the possible self-based approach started with
the idea that positive valence is key, each approach makes a distinct set of predictions as to
which features of future “me” predict future-focused action.
When read in isolation, each approach and the set of predictions it makes as to when
ROADMAP FOR CHANGING STUDENT ROADMAPS

91
future “me” matters for future-focused action is plausible. Thus, research rooted in a possible-
self approach predicts (and finds) that future-focused action is more likely when people link a
future identity to strategies, especially when these strategies are concrete and address aspects of
the current context. Research rooted in self-gap approaches predicts (and finds) that future-
focused action is more likely when people consider gaps between their current and their future
“me,” especially if they feel efficacious about addressing these gaps and consider gaps and
progress addressing them in a certain way. Research rooted in “self-continuity” approaches
predicts (and finds) that future-focused action is more likely when people experience future “me”
as close, connected, or continuous with current “me.”
However, when read together, it becomes clear that current approaches are insufficient
for three reasons: the approaches yield contradictory predictions, none is broad enough to include
both of the alternative approaches, and results are not fully consistent even within a single
approach. This has likely gone unnoticed because the empirical literature and reviews of this
literature are typically siloed, focusing only on a particular approach and not taking into account
that other approaches make conflicting predictions and find opposing results (e.g., self-
continuity, Hershfield, 2019; possible selves, Oyserman & James, 2011; self-gap, Oettingen,
2012). Even reviews that include more than one approach assume more consistency across
approaches than warranted (e.g. Oyserman & James, 2009). This implies that none of the current
approaches is sufficient to synthesize the empirical evidence; a new theoretical framework is
needed to synthesize approaches, predictions, and results. We used identity-based motivation
(IBM) theory to do just that.
IBM theory predicts that thinking (about the self) is for doing; people act in ways that fit
the identities that are on their mind and feel relevant to the affordances and constraints of their
ROADMAP FOR CHANGING STUDENT ROADMAPS

92
immediate situation (Oyserman, 2015a, 2015b). Though the self (and the identities it includes) is
experienced as stable if not fixed, which identities come to mind and what these identities imply
for meaning making and action are dynamically constructed in context (Oyserman, 2007). A
future “me” is like any other identity—it affects action, directing attention to future-focused
action if it is (implicitly or explicitly) experienced as relevant and not otherwise.
Using an IBM lens clarifies that what possible-self, self-gap, and self-continuity
approaches do is highlight specific features of future “me” that may increase the likelihood that
future “me” is experienced as relevant to the choices facing current “me” in a particular situation.
What these approaches do not do is provide a general model of the underlying process. However,
using an IBM lens, it becomes clear that the underlying process is future-focused identity-based
motivation. First, consider dynamic construction, that what current “me” entails is dynamically
constructed in context. In some contexts, future “me” feels like part of current “me,” implying
that future-focused action is relevant to current “me.” Next, consider action-readiness. Finding
oneself taking future-focused action implies that future “me” is relevant to current “me.” The
reverse is also true—experiencing future “me” as part of current “me” should trigger readiness to
act in future-focused ways. Finally, consider procedural-readiness. People are more likely to
interpret their experiences of difficulty as implying the importance of investing in future “me”
and not the impossibility of attaining that future if future “me” feels like part of current “me.”
The reverse should also hold—finding oneself interpreting difficulty as importance implies that
future “me” is relevant to the choices facing current “me.”
IBM and Other Approaches to Future Time and Goals
Using an IBM lens implies that future-focused action is context sensitive but accessible
to everyone. That is, depending on the situation, all people can take future-focused action. In this
ROADMAP FOR CHANGING STUDENT ROADMAPS

93
regard, an IBM lens contrasts with an individual differences approach (which we detail next).
From an individual differences perspective, people differ in the extent that they take future-
focused action, either because, as described until now, some people are efficacious and others are
not, or as we describe below, some people are higher in future orientation, or are individualistic
or reason abstractly. At the same time, an IBM-based approach does not predict that
developmental, socialization, chronic context, or individual differences do not exist in the
propensity to experience future “me” as relevant to the choices facing current me. Indeed, the
empirical literature we reviewed included participants living in the U.S., Canada, England,
Germany, France, the Netherlands, and China who varied in age from elementary school to post-
retirement, although little is known about children under ten. Though the current future “me”
literature is not set up to test for differences in culture or in development, developmental changes
in reasoning about the future likely matter (e.g., Hoerl & McCormack, in press). As we detail in
the next section, chronic differences in ability to make choices (e.g., Fisher, O’Donnell, &
Oyserman, 2017) and individual differences in future orientation (e.g., Strathman et al., 1994;
Zimbardo & Boyd, 1999) may make it more likely that future “me” is on the mind and more
likely that future “me” feels relevant to choices facing current “me.”
Social structural and cultural differences in future time perspective. Our English-
language based review focused on the effects of bringing future “me” to mind on future-focused
action. The literature we reviewed included participants who varied in their socioeconomic status
and who lived in different countries. We tagged our summaries so that life phase, socioeconomic
status, and country were highlighted. Even though the literature is not set up for quantitative
assessment of these factors as moderators of effects, our visual inspection of results does not
suggest that effects are limited to particular life phases, countries or social status of participants.
ROADMAP FOR CHANGING STUDENT ROADMAPS

94
However, an emerging literature on social structural and cultural differences in future time
perspective does attempt to test for this possibility and it suggests that differences exist (e.g.,
Ashkanasy Gupta, Mayfield, & Trevor-Roberts, 2004; Carter, McCollough, Kim-Spoon,
Corrales, Blake 2012; Dahl, 2000; Lee, Liu, & Hu, 2017). For example, religious people are, on
average, more future focused than non-religious people (Carter et al., 2012), while poverty has
been found to increase present focus (Bertrand, Mullainathan, & Shafir, 2006; Griskevicius,
Tybur, Delton, & Tobertson, 2011; Lawrence, 1991). It also may be more difficult to experience
future “me” as relevant to the choices facing current “me” when these choices are limited by
stigma and structural barriers (e.g., in the domain of health Oyserman & Fisher, 2017; Lewis &
Oyserman, 2016; in the domain of education Oyserman & Lewis, 2017; Lewis & Yates, 2019).
Indeed, students whose families are low in social-economic status are just as likely to have
school-focused possible identities (Azmitia et al., 2018; Destin & Svoboda, 2017; Stephens et al.,
2014, 2015; Oyserman et al., 2002; 2006), but less likely to have linked strategies that are
concrete and address barriers in their social context (Oyserman, Johnson, & James, 2011). At the
same time, having strategies to attain school-focused possible identities is just as effective for
these children (Bi & Oyserman, 2015).
With regard to other aspects of culture, some studies suggest that cultural axes of
individualism and collectivism are associated with experiencing the future as close or far
(Spassova & Lee, 2013). However, the mechanism is unclear. It is possible that a collectivistic
mindset, by triggering a connect-and-relate procedural mindset, makes experiencing current and
future “me” as connected more fluent than an individualistic mindset, which triggers a separate-
and-distinguish mindset, and is therefore likely to relegate future “me” to a different bin (for a
review, Oyserman, 2017). The future orientation literature strongly suggests that between-
ROADMAP FOR CHANGING STUDENT ROADMAPS

95
country differences are not simply due to country-level differences in wealth, development, or
axes such as individualism-collectivism (Dahl, 2000; Lee, et al. 2017). Instead, this literature
points to language differences in whether the future must be distinguished from the present or
can be described in a more continuous way (Chen, 2013). This linguistic difference, for example,
yields a contrast between high future time orientation Germany and low future time orientation
France (Dahl, 2000; Lee, et al. 2017).
Our IBM-based prediction is that people can take future-focused action if future “me” is
on the mind. This requires that future “me” be experienced as relevant to the choices facing
current “me.” The social structural, cultural, and language literatures suggest that experiencing
relevance will be easier in some contexts than in others. As we outline next, the same may be
true when considering individual differences.
Individual differences in future time perspective. Individual difference approaches ask
for whom future “me” affects future-focused action and make two predictions. First, people
higher in future time perspective will be more likely to have future “me” on their mind. Second,
people higher in future time perspective will be more likely to engage in future-focused action
when future “me” is on their mind. Though researchers conceptualize and measure future time
perspective in slightly different ways (for a review, see Andre, van Vianen, Peetsma, & Oort,
2018), each of these measures seeks to quantify how much a person plans for and achieves their
future goals (Zimbardo & Boyd, 1999) or bases their decisions on future (rather than immediate)
consequences of their actions (Strathman et al., 1994).
Studies measuring future time perspective suggest that people higher in future time
perspective do take more future-focused action. People who score higher on measures of future
time perspective are also more likely to eat healthy (van Beek, Antonidies, & Handgraaf, 2013)
ROADMAP FOR CHANGING STUDENT ROADMAPS

96
and are less at risk of alcohol and substance abuse (Keough, Zimbardo, & Boyd, 1999). They
spend more time studying (Zimbardo & Boyd, 1999) and report more proactive career behavior
(Fouarge, Schils, & De Grip, 2013). They report less impulsive buying and more willingness to
wait for larger monetary rewards (Joireman, Sprott, & Spangenberg, 2005), and they experience
shorter durations of homelessness after losing their housing (Epel, Bandura, & Zimbardo, 1999).
An individual differences perspective differs from our IBM-based model because the
former focuses on individuals while the latter focuses on context. That is, an individual
differences perspective highlights the possibility that only those high in future time perspective
are likely to take future-focused action, while our IBM-based model highlights the possibility
that people will take future-focused action when future “me” is on the mind and experienced as
relevant to the choices facing current “me.” However, the two models only yield conflicting
predictions at the extreme— if context or individual differences are never found to matter. To the
extent that people low in future orientation may require clearer contextual cues than those high in
future orientation, the two models can be seen as compatible. Hence, individual differences and
IBM perspectives can be integrated by considering individual difference measures as quantifying
individual variation in the tendency to chronically experience future “me” as relevant to the
choices facing current “me.” An IBM lens predicts three ways in which people higher in future
time perspective might more frequently experience future “me” as relevant to the choices facing
current “me.” First, future “me” may be on the mind more often. Second, future “me” may be
more likely to be experienced as part of current “me.” Third, difficulty imagining future “me”
and starting or persisting in future-focused action may be more likely to be seen as a signal of
importance and not of impossibility. That is, people higher in future orientation are more likely
to interpret their experiences of difficulty as implying the importance of investing in future “me”
ROADMAP FOR CHANGING STUDENT ROADMAPS

97
and not the impossibility of attaining that future “me.”
We found one study supporting our prediction that individual difference and IBM
approaches are compatible. Ouellette and colleagues (2005) measured future time perspective
and randomly assigned participants to two groups. In one group people were asked to imagine
their healthy future “me” and in the other group people were asked to imagine the prototypical
healthy person. The researchers found that people higher in future time perspective benefited
more from being assigned to imagine their healthy future “me” than people low in future time
perspective.
Goal highlighting vs. goal balancing. While future time perspective focuses on
individual differences in goal striving, there is another obstacle to taking future-focused action,
which is that people have more than one goal (more than one possible identity) and working on
one necessarily means not working on another. Fishbach and colleagues (2009) focus on this
question of how people toggle between various goals—when do they choose to focus on one
goal and when do they choose to shift to another goal. Rather than differentiate between people
who strive and those who do not, they ask what situational forces affect the choices people make
about which of their goals to focus in the moment. In a series of studies, Fishbach and colleagues
(2009) document that people can infer from their progress working on a goal that they are
committed to the goal (and hence should keep going) or that they can shift to another goal (since
they have already made progress on this one). This perspective, is congruent with self-gap
approaches like control theory (Carver & Scheier, 2016), which assume that people are paying
attention to their goal progress. However, unlike control theories, this perspective does not focus
on speed of progress but rather on the meaning people infer from progress with regard to whether
to shift to another goal or commit to the current one. By providing a more nuanced analysis of
ROADMAP FOR CHANGING STUDENT ROADMAPS

98
how people might interpret goal progress, this approach is relevant to an understanding of self-
gap studies, though it does not make specific predications about when future “me” is likely to be
experienced as relevant to the choices facing current “me.”
Construal level. Construal level theory (Trope & Lieberman, 2003) proposes that
psychological distance has cognitive and motivational consequences. Specifically, people think
more instrumentally about psychologically close events (How should I do this?) and are value-
driven when considering distal events (Why should I do this?). Construal level itself does not
make predictions about when future “me” will be relevant to the choices facing current “me.” It
does predict and show that people experience their distal future self as more like their true self
(Wakslak, Nussbaum, Liberman, & Trope, 2008), but this does not predict that people will take
action to attain their distal selves. However, distance itself is unlikely to be the active ingredient.
Indeed, IBM theory predicts that people are more likely to take future-focused action when
future “me” is experienced as relevant to the choices facing current “me.” This can occur when
future “me” feels psychologically close, when strategies to attain a future “me” come to mind,
and when difficulties starting and keeping going are understood as implying importance rather
than impossibility. Empirically, abstract construal is associated with endorsement of difficulty-
as-importance at a low level (e.g., correlation of about .3), implying that the two constructs may
(or may not) conceptually overlap, potentially via a higher tendency to take a future time
perspective (Fisher & Oyserman, 2017).
Self-regulatory focus. Higgins’ (1998) self-regulatory focus model outlines two different
ways that people can imagine and pursue their goals. They can imagine the self-goals they desire
to attain or the self-goals they ought to attain. They can strive to avoid failures and preserve what
they have (prevention focus) or they can strive to have more successes and build on what they
ROADMAP FOR CHANGING STUDENT ROADMAPS

99
have (promotion focus). People experience value from the fit between the way they
conceptualize their self-goals and the way that they strive to attain them (Higgins, 2000). That is,
people prefer to work on their prevention self-goals (the person they ought to become) in a
prevention way (avoiding failures) and to work on their promotion goals (the person they desire
becoming) in a promotion way (striving for successes). Value from fit is compatible with our
IBM-based prediction that people take future-focused action when future “me” is experienced as
relevant to the choices facing current “me” because “fit” is a way a strategy can feel identity-
congruent. When a person is prevention-focused, future “me” should be experienced as relevant
when it cues strategies geared toward vigilance or avoidance. When a person is promotion-
focused, future “me” should be experienced as relevant when it cues strategies that reflect
eagerness or an approach orientation. At the same time, our IBM-based prediction is broader
than self-regulatory focus theory in that experiencing the future “me” as relevant to the choices
facing current “me” is not limited to a match between promotion and prevention goals and goal-
striving. Self-regulatory focus and fit between goal and goal striving does not provide a means to
synthesize the results of possible self, self-continuity and when self-gap predictions.
Final Remarks and Moving Forward
We used IBM theory to synthesize the literature on when future “me” affects future-
focused action. We showed that our IBM-based synthesis is congruent with social structural,
individual difference, temporal-construal, and self-regulatory focus predictions. Going forward,
identity-based motivation theory can help make progress on the practical challenges of utilizing a
person’s future “me” to help improve their behavior. For example, consider the IBM prediction
that a future “me” will be experienced as relevant when it cues strategies that are congruent with
current “me.” The implication is that intervention should support experience of future-focused
ROADMAP FOR CHANGING STUDENT ROADMAPS

100
strategies as being identity-congruent. Otherwise health- and education-promoting behaviors
may feel incongruent with social class or racial-ethnic identities (e.g., Lewis & Oyserman, 2016;
Oyserman, Fryberg, & Yoder, 2007; Stephens et al., 2015; Walton & Cohen, 2011). Our IBM
framework highlights the need to focus on which strategies are likely to come to mind when a
person is encouraged to picture their future “me” and what experienced difficulty imagining
future “me” or strategies to get going is interpreted to imply about oneself.
A number of IBM-based interventions document positive effects of helping students
experience their far (adult) and near (next year) possible identities as connected to the present
and interpret their difficulties in starting, persisting, and making progress as implying importance
rather than impossibility (Horowitz, Oyserman, Yoder, & Sorensen, 2018; Oyserman et al.,
2006; Oyserman, Terry, & Bybee, 2002). These interventions affect students’ possible identities
and strategies to work on them, affecting school outcomes (Oyserman et al., 2006; Oyserman et
al., 2002). Other interventions that affect future-focused behaviors can be understood as
addressing barriers to experiencing future “me” as relevant to the choices facing current “me”
(e.g., Destin & Svoboda, 2017; Stephens, Hamedani & Destin, 2014; Stephens, Townsend,
Hamedani, Destin, & Manzo, 2015). Barriers can be in whether future “me” feels impossible to
attain due to lack of fit (low belonging) or due to a belief that people cannot really change—a
fixed mindset (e.g., Goyer et al., in press). Using an IBM approach can explain why, depending
on the intervention, intervening to change belongingness or fixed mindsets only sometimes
matters, or why effects are found for some but not for other students (Goyer et al., in press). That
is, it is not belongingness or fixed mindset per se that matters, but rather what these imply for the
relevance of future “me” to the choices facing current “me” in the moment.
Using an IBM-based framework highlights the role of relevance and yields predictions
ROADMAP FOR CHANGING STUDENT ROADMAPS

101
about when the features of future “me” highlighted in possible self-based, self-gap, and self-
continuity approaches are likely to matter. To experience future “me” as relevant to the choices
facing current “me,” a possible self-based approach predicts that what needs to be on the mind is
a possible identity linked to strategies for current action. A self-continuity approach suggests that
all that is needed is that future “me” feel close to or part of current “me.” In contrast, a self-gap
approach predicts that what is necessary is to hold future “me” in mind as a standard, attend to
the gap between that standard and future “me,” consider what can be done and whether one is
capable of doing it. Our IBM-based synthesis is congruent with cognitive approaches that
suggest that experiencing future “me” as occurring at the same time as current “me” or at the
same time as current action may require fewer cognitive resources than mentally simulating gaps
to attaining future “me” (Hoerl & McCormack, in press; Oyserman & Dawson, in press). The
implication is that a self-gap approach may be more prone to failure under cognitive load or
other pressures such that future “me” will simply be experienced as separate from current “me”
and irrelevant to the choices current “me” faces.
Using an IBM lens highlights that what matters is whether future “me” is experienced as
relevant or irrelevant to the choices facing current “me.” If an accessible future “me” feels
irrelevant to the choices facing current “me,” difficulties taking or imagining taking future-
focused action will imply that this is not “for me,” and present-focused action will continue.
However, if an accessible future “me” feels relevant to the choices facing current “me,”
difficulties taking or imagining taking future-focused action will imply that this is “for me,” and
future-focused action will be triggered. By tying the influence of a future “me” to whether or not
actions to attain it fit with the current identity on one’s mind, our approach illuminates potential
moderators and allows for the smooth integration of other identity-relevant factors—including
ROADMAP FOR CHANGING STUDENT ROADMAPS

102
race-ethnicity, social class, and culture—into theorizing about when a future “me” will be most
influential. This integration is likely to ultimately produce more stable findings and more
effective behavioral interventions.

ROADMAP FOR CHANGING STUDENT ROADMAPS

103
References
Adelman, R. M., Herrmann, S. D., Bodford, J. E., Barbour, J. E., Graudejus, O., Okun, M. A., &
Kwan, V. S. (2017). Feeling closer to the future self and doing better: Temporal
psychological mechanisms underlying academic performance. Journal of Personality,
85(3), 398-408.
Adriaanse, M. A., De Ridder, D. T., & Voorneman, I. (2013). Improving diabetes self-
management by mental contrasting. Psychology & Health, 28(1), 1-12.
Aelenei, C., Lewis, N.A., Jr., & Oyserman, D. (2017). No pain no gain? Social demographic
correlates and identity consequences of interpreting experienced difficulty as importance.
Contemporary Educational Psychology, 48, 43-55.
Aloise-Young, P. A., Hennigan, K. M., & Leong, C. W. (2001). Possible selves and negative
health behaviors during early adolescence. The Journal of Early Adolescence, 21(2), 158-
181.
Anderman, E. M., Anderman, L. H., & Griesinger, T. (1999). The relation of present and
possible academic selves during early adolescence to grade point average and
achievement goals. The Elementary School Journal, 100(1), 3-17.
Andre, L., van Vianen, A. E., Peetsma, T. T., & Oort, F. J. (2018). Motivational power of future
time perspective: Meta-analyses in education, work, and health. PloS one, 13(1),
e0190492.
Aron, A., Aron, E. N., & Smollan, D. (1992). Inclusion of other in the self scale and the structure
of interpersonal closeness. Journal of Personality and Social Psychology, 63(4), 596-612.
Ashkanasy N, Gupta V, Mayfield M, & Trevor-Roberts E. (2004). Future orientation. In R.
House, P. Changes, M. Javidan, P. Dorfman & W. Gupta (Eds). Culture, leadership, and
ROADMAP FOR CHANGING STUDENT ROADMAPS

104
organizations: The GLOBE study of 62 societies. (pp. 282–342). Thousand Oaks, CA:
Sage.
Austenfeld, J. L., Paolo, A. M., & Stanton, A. L. (2006). Effects of writing about emotions
versus goals on psychological and physical health among third-year medical students.
Journal of Personality, 74(1), 267-286.
Austenfeld, J. L., & Stanton, A. L. (2008). Writing about emotions versus goals: Effects on
hostility and medical care utilization moderated by emotional approach coping processes.
British Journal of Health Psychology, 13(1), 35-38.
Azmitia, M., Sumabat ‐Estrada, G., Cheong, Y., & Covarrubias, R. (2018). “Dropping Out is Not
an Option”: How Educationally Resilient First ‐Generation Students See the Future. New
directions for child and adolescent development, 2018(160), 89-100.
Bargh, J. A. (2016). Awareness of the prime versus awareness of its influence: implications for
the real-world scope of unconscious higher mental processes. Current Opinion in
Psychology, 12, 49-52.
Bartels, D. M., & Rips, L. J. (2010). Psychological connectedness and intertemporal choice.
Journal of Experimental Psychology: General, 139(1), 49-69.
Bartels, D. M., & Urminsky, O. (2011). On intertemporal selfishness: How the perceived
instability of identity underlies impatient consumption. Journal of Consumer Research,
38(1), 182-198.
Bartels, D. M., & Urminsky, O. (2015). To know and to care: How awareness and valuation of
the future jointly shape consumer spending. Journal of Consumer Research, 41(6), 1469-
1485.
Baumeister, R. F., Hofmann, W., Summerville, A., Reiss, P. T., & Vohs, K. D. (2018).
ROADMAP FOR CHANGING STUDENT ROADMAPS

105
Everyday thoughts in time: Experience sampling studies of mental time travel.
Manuscript under editorial review.
Bertrand, M., Mullainathan, S., & Shafir, E. (2006). Behavioral economics and marketing in aid
of decision making among the poor. Journal of Public Policy & Marketing, 25(1), 8–23.
Bi, C., & Oyserman, D. (2015). Left behind or moving forward? Effects of possible selves and
strategies to attain them among rural Chinese children. Journal of Adolescence, 44, 245-
258.
Black, M. E., Stein, K. F., & Loveland-Cherry, C. J. (2001). Older women and mammography
screening behavior: do possible selves contribute?. Health Education & Behavior, 28(2),
200-216.
Blouin-Hudon, E. M. C., & Pychyl, T. A. (2015). Experiencing the temporally extended self:
Initial support for the role of affective states, vivid mental imagery, and future self-
continuity in the prediction of academic procrastination. Personality and Individual
Differences, 86, 50-56.
Burum, B. A., Gilbert, D. T., & Wilson, T. D. (2016). Caught red-minded: Evidence-induced
denial of mental transgressions. Journal of Experimental Psychology: General, 145(7),
844-852.
Carter E. C., McCollough M. E., Kim-Spoon J., Corrales C., & Blake A. (2012). Religious
people discount the future less. Evolution and Human Behavior, 33(3), 224–231.
Carver, C. S., & Scheier, M. F. (1982). Control theory: A useful conceptual framework for
personality–social, clinical, and health psychology. Psychological Bulletin, 92(1), 111-
135.
Carver, C. S., & Scheier, M. F. (2016). Self-regulation of action and affect. In Vohs, K. D. &
ROADMAP FOR CHANGING STUDENT ROADMAPS

106
Baumeister, R. F. (Eds.). Handbook of Self-regulation: Research, Theory, and
Applications (pp. 3-23). New York, NY: Guilford Press.
Case, A., & Deaton, A. (2015). Rising morbidity and mortality in midlife among white non-
Hispanic Americans in the 21st century. Proceedings of the National Academy of
Sciences, 112(49), 15078-15083.
Chen, M. K. (2013). The effect of language on economic behavior: Evidence from savings rates,
health behaviors, and retirement assets. American Economic Review, 103(2), 690-731.
Cho, M. (2015). The effects of working possible selves on second language performance.
Reading and Writing, 28(8), 1099-1118.
Comello, M. L. G. (2015). How does a risk-oriented “future-self” influence behavior?: A
structural-equation-modeling approach with marijuana-related outcomes. Journal of
Health Psychology, 20(1), 37-47.
Dahl, Ö. (2000). The grammar of future time reference in European languages. In Tense and
Aspect in the Languages of Europe (Vol. 6, pp. 309-328). Berlin: Mouton de Gruyter.
Dalley, S. E. (2016). Does the salience of possible selves mediate the impact of approach and
avoidance temperaments on women's weight-loss dieting?. Personality and Individual
Differences, 88, 267-271.
Dalley, S. E., & Buunk, A. P. (2011). The motivation to diet in young women: Fear is stronger
than hope. European Journal of Social Psychology, 41(5), 672-680.
Destin, M. (2017). An open path to the future: Perceived financial resources and school
motivation. The Journal of Early Adolescence, 37(7), 1004-1031.
Destin, M., & Svoboda, R. C. (2017). A brief randomized controlled intervention targeting
parents improves grades during middle school. Journal of Adolescence, 56, 157-161.
ROADMAP FOR CHANGING STUDENT ROADMAPS

107
Ellen, P. S., Wiener, J. L., & Fitzgerald, M. P. (2012). Encouraging people to save for their
future: Augmenting current efforts with positive visions of the future. Journal of Public
Policy & Marketing, 31(1), 58-72.
Elmore, K. C., & Oyserman, D. (2012). If ‘we’ can succeed, ‘I’ can too: Identity-based
motivation and gender in the classroom. Contemporary Educational Psychology, 37(3),
176-185.
Epel, E. S., Bandura, A., & Zimbardo, P. G. (1999). Escaping homelessness: The influences of
self-efficacy and time perspective on coping with homelessness. Journal of Applied
Social Psychology, 29, 575–596.
Ersner-Hershfield, H., Garton, M. T., Ballard, K., Samanez-Larkin, G. R., & Knutson, B. (2009).
Don’t stop thinking about tomorrow: Individual differences in future self-continuity
account for saving. Judgment and Decision Making, 4(4), 280-286.
Ersner-Hershfield, H., Wimmer, G. E., & Knutson, B. (2008). Saving for the future self: Neural
measures of future self-continuity predict temporal discounting. Social Cognitive and
Affective Neuroscience, 4(1), 85-92.
Evans, M. B., & Wilson, A. E. (2014). Subjective temporal proximity to future selves moderates
the link between exercise intentions and behavior. Sport, Exercise, and Performance
Psychology, 3(3), 184-190.
Fishbach, A., Zhang, Y., & Koo, M. (2009). The dynamics of self-regulation. European Review
of Social Psychology, 20(1), 315-344.
Fisher, O., & Oyserman, D. (under editorial review). When the going gets tough, social class
matters: Social class predicts difficulty mindsets.
Fisher, O., & Oyserman, D. (2017). Assessing interpretations of experienced ease and difficulty
ROADMAP FOR CHANGING STUDENT ROADMAPS

108
as motivational constructs. Motivation Science, 3(2), 133-163.
Fouarge, D., Schils, T., & De Grip, A. (2013). Why do low-educated workers invest less in
further training? Applied Economics, 45, 2587–2601.
Gollwitzer, A., Oettingen, G., Kirby, T. A., Duckworth, A. L., & Mayer, D. (2011). Mental
contrasting facilitates academic performance in school children. Motivation and Emotion,
35(4), 403-412.
Goyer, J. P., Cohen, G. L., Cook, J. E., Master, A., Apfel, N., Lee, W. Henderson, A. G., Reeves,
S. L., Okonofua, J. A., & Walton, G. M. (in press). Targeted identity-safety interventions
cause lasting reductions in discipline citations among negatively stereotyped boys
Journal of Personality and Social Psychology.
Griskevicius, V., Tybur, J.M., Delton, A.W., & Tobertson, T.E. (2011). The influence of
mortality and socioeconomic status on risk and delayed rewards: A life history theory
approach. Journal of Personality and Social Psychology, 100, 1015-1026.
Hershfield, H. E. (2011). Future self continuity: How conceptions of the future self transform
intertemporal choice. Annals of the New York Academy of Sciences, 1235(1), 30-43.
Hershfield, H.E. (2019). The self over time. Current Opinion in Psychology, 26, 72-75.
Hershfield, H. E., & Bartels, D. (2018). The future self. In G. Oettingen, A. T. Sevincer, & P. M.
Gollwitzer (Eds), The psychology of thinking about the future (89-109). New York, NY:
The Guilford Press.
Hershfield, H. E., Cohen, T. R., & Thompson, L. (2012). Short horizons and tempting situations:
Lack of continuity to our future selves leads to unethical decision making and behavior.
Organizational Behavior and Human Decision Processes, 117(2), 298-310.
Hershfield, H. E., Goldstein, D. G., Sharpe, W. F., Fox, J., Yeykelis, L., Carstensen, L. L., &
ROADMAP FOR CHANGING STUDENT ROADMAPS

109
Bailenson, J. N. (2011). Increasing saving behavior through age-progressed renderings of
the future self. Journal of Marketing Research, 48, S23-S37.
Higgins, E.T. (2000). Making a good decision: Value from fit. American Psychologist, 55, 1217–
1230.
Higgins, E. T. (1998). Promotion and prevention: Regulatory focus as a motivational principle.
Advances in Experimental Social Psychology, 30, 1–45.
Hoerl, C. & McCormack, T. (in press). Thinking in and about time: A dual systems perspective
on temporal cognition Brain and Behavioral Sciences.
Hooker, K., & Kaus, C. R. (1994). Health-related possible selves in young and middle adulthood.
Psychology and Aging, 9(1), 126-133.
Hoppmann, C. A., Gerstorf, D., Smith, J., & Klumb, P. L. (2007). Linking possible selves and
behavior: Do domain-specific hopes and fears translate into daily activities in very old
age? The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences,
62(2), P104 –P111.
Horowitz, E., Sorensen, N., Yoder, N., & Oyserman, D. (2018). Teachers can do it: scalable
identity-based motivation intervention in the classroom. Contemporary Educational
Psychology, 54, 12-28.
Hoyle, R. H., & Sherrill, M. R. (2006). Future orientation in the self-system: Possible selves,
self-regulation, and behavior. Journal of Personality, 74(6), 1673-1696.
Johannessen, K. B., Oettingen, G., & Mayer, D. (2012). Mental contrasting of a dieting wish
improves self-reported health behaviour. Psychology & Health, 27(2), 43-58.
Joireman, J., Sprott, D., & Spangenberg, E. (2005). Fiscal responsibility and the consideration of
future consequences. Personality and Individual Differences, 39, 1159–1168.
ROADMAP FOR CHANGING STUDENT ROADMAPS

110
Joshi, P. D., & Fast, N. J. (2013). Power and reduced temporal discounting. Psychological
Science, 24(4), 432-438.
Kappes, H. B., Oettingen, G., Mayer, D., & Maglio, S. (2011). Sad mood promotes self-initiated
mental contrasting of future and reality. Emotion, 11(5), 1206-1222.
Kappes, A., Wendt, M., Reinelt, T., & Oettingen, G. (2013). Mental contrasting changes the
meaning of reality. Journal of Experimental Social Psychology, 49(5), 797-810.
Keough, K. A., Zimbardo, P. G., & Boyd, J. N. (1999). Who’s smoking, drinking, and using
drugs? Time perspective as a predictor of substance use. Basic and Applied Social
Psychology, 21, 149–164.
King, L. A. (2001). The health benefits of writing about life goals. Personality and Social
Psychology Bulletin, 27(7), 798-807.
Kirk, D., Oettingen, G., & Gollwitzer, P. M. (2011). Mental contrasting promotes integrative
bargaining. International Journal of Conflict Management, 22(4), 324-341.
Ko, H. J., Mejía, S., & Hooker, K. (2014). Social possible selves, self-regulation, and social goal
progress in older adulthood. International Journal of Behavioral Development, 38(3),
219-227.
Koriat, A. (1993). How do we know that we know? The accessibility model of the feeling of
knowing. Psychological Review, 100(4), 609-639.
Kuo, H. C., Lee, C. C., & Chiou, W. B. (2016). The power of the virtual ideal self in weight
control: Weight-reduced avatars can enhance the tendency to delay gratification and
regulate dietary practices. Cyberpsychology, Behavior, and Social Networking, 19(2), 80-
85.
Landau, M. J., Oyserman, D., Keefer, L. A., & Smith, G. C. (2014). The college journey and
ROADMAP FOR CHANGING STUDENT ROADMAPS

111
academic engagement: How metaphor use enhances identity-based motivation. Journal of
Personality and Social Psychology, 106(5), 679-698.
Lawrance, E. C. (1991). Poverty and the rate of time preference: evidence from panel
data. Journal of Political Economy, 99(1), 54-77.
Lee, C. K., Corte, C., Stein, K. F., Park, C. G., Finnegan, L., & McCreary, L. L. (2015).
Prospective effects of possible selves on alcohol consumption in adolescents. Research in
Nursing & Health, 38(1), 71-81.
Lee, S., Liu, M., & Hu, M. (2017). Relationship between Future Time Orientation and Item
Nonresponse on Subjective Probability Questions: A Cross-Cultural Analysis. Journal of
Cross-Cultural Psychology, 48(5), 698-717.
Lewis, N.A., Jr., & Oyserman, D. (2015). When does the future begin? Time metrics matter,
connecting present and future selves. Psychological Science, 26(6), 816-825.
Lewis, N. A., Jr., & Oyserman, D. (2016). Using identity-based motivation to improve the
nation's health without breaking the bank. Behavioral Science & Policy, 2(2), 24-38.
Lewis, N. A., Jr., & Yates, J. F. (2019). Preparing disadvantaged students for success in college:
lessons learned from the preparation initiative. Perspectives on Psychological Science, 14
(1), 54-59.
Macrae, C. N., Mitchell, J. P., Golubickis, M., Ho, N. S., Sherlock, R., Parlongo, R., ... &
Christian, B. M. (2017). Saving for your future self: The role of imaginary experiences.
Self and Identity, 16(4), 384-398.
Markus, H., & Nurius, P. (1986). Possible selves. American Psychologist, 41(9), 954-969.
Markus, H., & Wurf, E. (1987). The dynamic self-concept: A social psychological perspective.
Annual Review of Psychology, 38(1), 299-337.
ROADMAP FOR CHANGING STUDENT ROADMAPS

112
Morrissey, M. (2016). The State of American Retirement. Washington, DC: Economic Policy
Institute. Retrieved from http://www.epi.org/files/2016/state-of-american-retirement-
final.pdf
Murru, E. C., & Ginis, K. A. M. (2010). Imagining the possibilities: The effects of a possible
selves intervention on self-regulatory efficacy and exercise behavior. Journal of Sport
and Exercise Psychology, 32(4), 537-554.
Newberry, A. L., & Duncan, R. D. (2001). Roles of boredom and life goals in juvenile
delinquency. Journal of Applied Social Psychology, 31(3), 527-541.
Norman, C. C., & Aron, A. (2003). Aspects of possible self that predict motivation to achieve or
avoid it. Journal of Experimental Social Psychology, 39(5), 500-507.
Nurra, C., & Oyserman, D. (2018). From future self to current action: an identity-based
motivation perspective. Self and Identity, 17(3), 343-364.
Ogilvie, D. M. (1987). The undesired self: A neglected variable in personality research. Journal
of Personality and Social Psychology, 52(2), 379-385.
Oettingen, G. (2012). Future thought and behaviour change. European Review of Social
Psychology, 23(1), 1-63.
Oettingen, G., Hönig, G., & Gollwitzer, P. M. (2000). Effective self-regulation of goal
attainment. International Journal of Educational Research, 33(7), 705-732.
Oettingen, G., Marquardt, M. K., & Gollwitzer, P. M. (2012). Mental contrasting turns positive
feedback on creative potential into successful performance. Journal of Experimental
Social Psychology, 48(5), 990-996.
Oettingen, G., Mayer, D., & Brinkmann, B. (2010). Mental contrasting of future and reality.
Journal of Personnel Psychology, 9(3), 138-144.
ROADMAP FOR CHANGING STUDENT ROADMAPS

113
Oettingen, G., Mayer, D., Sevincer, A., Stephens, E. J., Pak, H. J., & Hagenah, M. (2009).
Mental contrasting and goal commitment: The mediating role of energization.
Personality and Social Psychology Bulletin, 35(5), 608-622.
Oettingen, G., Mayer, D., & Thorpe, J. (2010). Self-regulation of commitment to reduce cigarette
consumption: Mental contrasting of future with reality. Psychology and Health, 25(8),
961-977.
Oettingen, G., Mayer, D., Thorpe, J. S., Janetzke, H., & Lorenz, S. (2005). Turning fantasies
about positive and negative futures into self-improvement goals. Motivation and
Emotion, 29(4), 236-266.
Oettingen, G., Pak, H. J., & Schnetter, K. (2001). Self-regulation of goal-setting: turning free
fantasies about the future into binding goals. Journal of Personality and Social
Psychology, 80(5), 736-753.
Oettingen, G., Stephens, E. J., Mayer, D., & Brinkmann, B. (2010). Mental contrasting and the
self-regulation of helping relations. Social Cognition, 28(4), 490-508.
Ouellette, J. A., Hessling, R., Gibbons, F. X., Reis-Bergan, M., & Gerrard, M. (2005). Using
images to increase exercise behavior: Prototypes versus possible selves. Personality and
Social Psychology Bulletin, 31(5), 610-620.
Oyserman, D. (2007). Social identity and self-regulation. In A. W. Kruglanski and E. T.
Higgins (Eds.), Social Psychology: Handbook of Basic Principles (432-453). New
York, NY: Guilford Press.
Oyserman, D. (2009a). Identity-based motivation: Implications for action-readiness, procedural-
readiness, and consumer behavior. Journal of Consumer Psychology, 19(3), 250–260.
Oyserman, D. (2009b). Identity-based motivation and consumer behavior. Journal of Consumer
ROADMAP FOR CHANGING STUDENT ROADMAPS

114
Psychology, 19(3). 276–279.
Oyserman, D. (2015a). Identity-based motivation. In R. Scott & S. Kosslyn (Eds.), Emerging
Trends in the Social Sciences. Hoboken, NJ: John Wiley and Sons.
Oyserman, D. (2015b). Pathways to success through identity-based motivation. New York:
Oxford University Press.
Oyserman, D (2017). Culture three ways: culture and subcultures within countries. Annual
Review of Psychology, 68, 435–463.
Oyserman, D., Bybee, D., & Terry, K. (2006). Possible selves and academic outcomes: How and
when possible selves impel action. Journal of Personality and Social Psychology, 91(1),
188-204.
Oyserman, D., Bybee, D., Terry, K., & Hart-Johnson, T. (2004). Possible selves as roadmaps.
Journal of Research in Personality, 38(2), 130-149.
Oyserman, D. & Dawson, A. (in press). Identity-based motivation and the paradox of the future
self: Getting going requires thinking about time (later) in time (now). Brain and
Behavioral Sciences.
Oyserman, D., & Destin, M. (2010). Identity-based motivation: Implications for intervention.
The Counseling Psychologist, 38(7), 1001-1043.
Oyserman, D., Destin, M., & Novin, S. (2015). The context-sensitive future self: Possible selves
motivate in context, not otherwise. Self and Identity, 14(2), 173-188.
Oyserman, D., Elmore, K., Novin, S., Fisher, O., & Smith, G. C. (2018). Guiding people to
interpret their experienced difficulty as importance highlights their academic possibilities
and improves their academic performance. Frontiers in psychology, 9:781.
Oyserman, D., Elmore, K., & Smith, G. (2012). Self, self-concept, and identity. In M. Leary & J.
ROADMAP FOR CHANGING STUDENT ROADMAPS

115
Tangney (Eds.), Handbook of self and identity, 2nd edition (pp. 69-104). New York:
Guilford.
Oyserman, D. & Fisher, O. (2017). Social Stigma and Health: An Identity-Based Motivation
Perspective. Chapter 17 In B. Major, J. F. Dovidio, and B. G. Link. The Oxford
Handbook of Discrimination, Stigma and Health, (pp 317-334). NY: Oxford University
Press.
Oyserman, D., Fryberg, S. A., & Yoder, N. (2007). Identity-based motivation and health. Journal
of Personality and Social Psychology, 93(6), 1011-1027.
Oyserman, D., Gant, L., & Ager, J. (1995). A socially contextualized model of African American
identity: Possible selves and school persistence. Journal of Personality and Social
Psychology, 69(6), 1216-1232.
Oyserman D., James L. (2011) Possible Identities. In: Schwartz S., Luyckx K., Vignoles V.
(Eds.) Handbook of Identity Theory and Research. (pp 117-145). New York, NY:
Springer.
Oyserman, D., Johnson, E., & James, L. (2011). Seeing the destination but not the path: Effects
of socioeconomic disadvantage on school-focused possible self content and linked
behavioral strategies. Self and Identity, 10(4), 474-492.
Oyserman, D. & Lewis, N.A., Jr. (2017) Seeing the destination AND the path: Using identity-
based motivation to understand racial disparities in achievement and what we can do
about them. Social Issues and Policy Review, 11(1), 159-194.
Oyserman, D., Lewis, N.A., Jr., Yan, V. X., Fisher, O., O'Donnell, S. C., & Horowitz, E. (2017).
An identity-based motivation framework for self-regulation. Psychological Inquiry, 28
(2-3), 139-147.
ROADMAP FOR CHANGING STUDENT ROADMAPS

116
Oyserman, D. & Markus, M. (1990). Possible selves and delinquency. Journal of Personality
and Social Psychology, 59(1), 112-125.
Oyserman, D., & Saltz, E. (1993). Competence, delinquency, and attempts to attain possible
selves. Journal of Personality and Social Psychology, 65(2), 360-374.
Oyserman, D., Terry, K., & Bybee, D. (2002). A possible selves intervention to enhance school
involvement. Journal of Adolescence, 25, 313-326.
Peetz, J., & Wilson, A. E. (2013). The post-birthday world: Consequences of temporal landmarks
for temporal self-appraisal and motivation. Journal of Personality and Social Psychology,
104(2), 249-266.
Peetz, J., Wilson, A. E., & Strahan, E. J. (2009). So far away: The role of subjective temporal
distance to future goals in motivation and behavior. Social Cognition, 27(4), 475-495.
Pierce, J., Schmidt, C., & Stoddard, S. A. (2015). The role of feared possible selves in the
relationship between peer influence and delinquency. Journal of Adolescence, 38, 17-26.
Pietroni, D., & Hughes, S. V. (2016). Nudge to the future: capitalizing on illusory superiority
bias to mitigate temporal discounting. Mind & Society, 15(2), 247-264.
Perras, M. G., Strachan, S. M., & Fortier, M. S. (2015). Back to the future: Associations between
possible selves, identity, and physical activity among new retirees. Activities, Adaptation
& Aging, 39(4), 318-335.
Perras, M. G., Strachan, S. M., & Fortier, M. S. (2016). Possible selves and physical activity in
retirees: The mediating role of identity. Research on Aging, 38(8), 819-841.
Pronin, E., Olivola, C. Y., & Kennedy, K. A. (2008). Doing unto future selves as you would do
unto others: Psychological distance and decision making. Personality and Social
Psychology Bulletin, 34(2), 224-236.
ROADMAP FOR CHANGING STUDENT ROADMAPS

117
Quoidbach, J., Gilbert, D. T., & Wilson, T. D. (2013). The end of history illusion. Science,
339(6115), 96-98.
Rutchick, A. M., Slepian, M. L., Reyes, M. O., Pleskus, L. N., & Hershfield, H. E. (2018).
Future self-continuity is associated with improved health and increases exercise behavior.
Journal of Experimental Psychology: Applied, 24(1), 72-80.
Ruvolo, A. P., & Markus, H. R. (1992). Possible selves and performance: The power of self-
relevant imagery. Social Cognition, 10(1), 95-124.
Schwarz, N (2015). Metacognition. In M. Mikulincer, P.R. Shaver, E. Borgida, & J. A. Bargh
(Eds.), APA Handbook of Personality and Social Psychology: Attitudes and Social
Cognition (pp. 203-229). Washington, DC: APA
Schwarz, N., & Strack, F. (2016). Beyond ‘what’ comes to mind: Experiential and
conversational determinants of information use. Current Opinion in Psychology, 12, 89-
93.
Seli, H., Dembo, M. H., & Crocker, S. (2009). Self in self-worth protection: The relationship of
possible selves and self-protective strategies. College Student Journal, 43(3), 832-842.
Sevincer, A. T., Busatta, P. D., & Oettingen, G. (2014). Mental contrasting and transfer of
energization. Personality and Social Psychology Bulletin, 40(2), 139-152.
Sevincer, A. T., & Oettingen, G. (2013). Spontaneous mental contrasting and selective goal
pursuit. Personality and Social Psychology Bulletin, 39(9), 1240-1254.
Sheeran, P., Harris, P., Vaughan, J., Oettingen, G., & Gollwitzer, P. M. (2013). Gone exercising:
Mental contrasting promotes physical activity among overweight, middle-aged, low-SES
fishermen. Health Psychology, 32(7), 802.
Sheldon, O. J., & Fishbach, A. (2015). Anticipating and resisting the temptation to behave
ROADMAP FOR CHANGING STUDENT ROADMAPS

118
unethically. Personality and Social Psychology Bulletin, 41(7), 962-975.
Smith, G. C., & Oyserman, D. (2015). Just not worth my time? Experienced difficulty and time
investment. Social Cognition, 33(2), 85-103.
Sobh, R., & Martin, B. A. (2011). Feedback information and consumer motivation: The
moderating role of positive and negative reference values in self-regulation. European
Journal of Marketing, 45(6), 963-986.
Spassova, G. & Lee, A. (2013). Looking into the future: a match between self-view and temporal
distance. Journal of Consumer Psychology, 40, 159-171.
Strachan, S. M., Marcotte, M. M., Giller, T. M., Brunet, J., & Schellenberg, B. J. (2017). An
online intervention to increase physical activity: Self-regulatory possible selves and the
moderating role of task self-efficacy. Psychology of Sport and Exercise, 31, 158-165.
Strathman, A., Gleicher, F., Boninger, D. S., & Edwards, C. S. (1994). The consideration of
future consequences: weighing immediate and distant outcomes of behavior. Journal of
Personality and Social Psychology, 66(4), 742-752.
Strauss, K., Griffin, M. A., & Parker, S. K. (2012). Future work selves: How salient hoped-for
identities motivate proactive career behaviors. Journal of Applied Psychology, 97(3),
580-598.
Stephens, N. M., Hamedani, M. G., & Destin, M. (2014). Closing the social-class achievement
gap: A difference-education intervention improves first-generation students’ academic
performance and all students’ college transition. Psychological Science, 25(4), 943-953.
Stephens, N. M., Townsend, S. S., Hamedani, M. G., Destin, M., & Manzo, V. (2015). A
difference-education intervention equips first-generation college students to thrive in the
face of stressful college situations. Psychological Science, 26(10), 1556-1566.
ROADMAP FOR CHANGING STUDENT ROADMAPS

119
Taber, B. J., & Blankemeyer, M. (2015). Future work self and career adaptability in the
prediction of proactive career behaviors. Journal of Vocational Behavior, 86, 20-27.
Trope, Y., & Liberman, N. (2003). Temporal construal. Psychological Review, 110(3), 403-421.
Trope, Y., Liberman, N., & Wakslak, C. (2007). Construal levels and psychological distance:
Effects on representation, prediction, evaluation, and behavior. Journal of Consumer
Psychology, 17(2), 83-95.
van Beek, J., Antonides, G., & Handgraaf, M. J. (2013). Eat now, exercise later: The relation
between consideration of immediate and future consequences and healthy behavior.
Personality and Individual Differences, 54, 785–791.
van Gelder, J. L., Hershfield, H. E., & Nordgren, L. F. (2013). Vividness of the future self
predicts delinquency. Psychological Science, 24(6), 974-980.
van Gelder, J. L., Luciano, E. C., Weulen Kranenbarg, M., & Hershfield, H. E. (2015). Friends
with my future self: longitudinal vividness intervention reduces delinquency.
Criminology, 53(2), 158-179.
Wakslak, C. J., Nussbaum, S., Liberman, N., & Trope, Y. (2008). Representations of the self in
the near and distant future. Journal of Personality and Social Psychology, 95(4), 757-
773.
Walton, G. M., & Cohen, G. L. (2011). A brief social-belonging intervention improves academic
and health outcomes of minority students. Science, 331(6023), 1447-1451.
Yowell, C. M. (2002). Dreams of the future: The pursuit of education and career possible selves
among ninth grade Latino youth. Applied Developmental Science, 6(2), 62-72.
Zhang, M., & Aggarwal, P. (2015). Looking ahead or looking back: Current evaluations and the
effect of psychological connectedness to a temporal self. Journal of Consumer
ROADMAP FOR CHANGING STUDENT ROADMAPS

120
Psychology, 25(3), 512-518.
Zimbardo, P. G., & Boyd, J. N. (1999). Putting time in perspective: A valid, reliable individual
difference metric. Journal of Personality and Social Psychology, 77(6), 1271-1288.

ROADMAP FOR CHANGING STUDENT ROADMAPS

121
Supplemental Materials
Section 1: Summary of mental contrasting experiments that do not compare participants in
the mental contrasting condition to participants in an accessible future “me” condition
In four experiments, participants who mentally contrasted where not compared to a group
in which all participants elaborated on future, often because the comparison group consisted of
participants randomly assigned to multiple conditions. Specifically, German middle school
students led to mentally contrast who felt efficacious about doing well in English class got higher
grades than a pooled mean of students randomly assigned to elaborate only on the present or only
on the future (Oettingen et al., 2000). German vocational school students led to mentally contrast
who felt efficacious about improving in mathematics performed better, by teacher report, than a
pooled mean of students who were randomly assigned to elaborate only on the present or only on
the future (Oettingen, Pak, et al., 2001, Study 4). German college students led to mentally
contrast who felt efficacious about successfully resolving an interpersonal problem reported
more commitment to solving an interpersonal problem compared to the pooled mean of
participants randomly assigned to elaborate only on the present, elaborate only on the future, or
elaborate on the present before the future (Oettingen et al., 2001, Study 3). Low SES, English
adults led to mentally contrast as part of a randomized controlled trial reported more physical
activity at 1-month and 7-month follow up compared to participants randomly assigned to the
control condition that did no elaboration (Sheeran, Harris, Vaughn, Oettingen, & Gollwitzer,
2013).

ROADMAP FOR CHANGING STUDENT ROADMAPS

122
Section 2: Summary of mental contrasting measurement studies that do not compare
participants in the mental contrasting to participants in an accessible future “me”
condition
In three studies, whether or not a person mentally contrasted was content-coded from
written responses about attaining future goals (Sevincer & Oettingen, 2013, Studies 1, 2, 3). In
two other studies, mental contrasting was content-coded from written responses about academic
concerns (Kappes, Oettingen, Mayer, & Maglio, 2011, Studies 5, 6). One study did not find an
effect of mental contrasting regardless of efficacy (Kappes, Oettingen, Mayer, & Maglio, 2011,
Study 6) and the other four report effects for participants who felt high in efficacy.
Specifically, German college students whose writing exhibited more mental contrasting
about an interpersonal goal and who also reported high efficacy about attaining the goal reported
more commitment to attaining the goal than a pooled comparison group of students whose
writing reflected something else—a focus only on the future, a focus only on obstacles, or a
focus on obstacles before a focus on the future (Sevincer & Oettingen, 2013, Study 1). Online
participants whose writing exhibited more mental contrasting about an academic or professional
goal and who also reported high efficacy about attaining that goal made more progress toward
the goal than a pooled comparison group of students whose writing reflected the other three
processes (Sevincer & Oettingen, 2013, Study 2). American college students whose writing
about being admitted to their favorite graduate exhibited more mental contrasting and who also
reported feeling high efficacy about writing a good graduate school essay wrote essays that were
more highly rated by outside observers than a pooled comparison group of students whose
writing reflected the other three processes (Sevincer & Oettingen, 2013, Study 3). American
college students whose writing about an academic concern exhibited more mental contrasting
ROADMAP FOR CHANGING STUDENT ROADMAPS

123
and who also reported high efficacy about overcoming this concern were more energized to
address this concern than a pooled comparison group of students whose writing reflected the
other three processes (Kappes, Oettingen, Mayer, & Maglio, 2011, Study 5). American college
students whose writing about ways to improve their study habits reflected mental contrasting
were no more persistent on a study habit improvement task than a single comparison group of
students whose writing reflected the other three processes, regardless of whether they felt
efficacious about improving their study habits or not (Kappes, Oettingen, Mayer, & Maglio,
2011, Study 6).
Section 3: Summary of self-continuity experiments that do not compare a continuous
future “me” to an accessible future “me”
Three experiments compared participants in a connection condition to participants in a
disconnection condition. In two of these experiments, college students were randomly assigned
to write about academic possible identities on a picture of a path (connection condition) or to
write about their academic possible identities on pictures of separate boxes or containers
(disconnection condition). College students in the connection condition put more effort into an
academic task (Landau et al., 2014, Study 2) and performed better on a subsequent exam
(Landau et al., 2014, Study 3) than did students in the disconnection condition. In the third
experiment, Mechanical Turk workers were randomly assigned to imagine a successful possible
identity and shown a picture of a path (connection condition). The other group was instructed to
imagine a successful possible identity and shown a picture of a container (disconnection
condition). Participants randomly assigned to the connection group reported more motivation to
work on their possible identity. Experienced connection was measured and mediated the effect of
being in the connection condition (Landau et al., 2014, Study 6).
ROADMAP FOR CHANGING STUDENT ROADMAPS

124
Four experiments compared participants in a self-stability condition to participants in a
self-instability condition. In these experiments participants in the self-stability condition either
read that research reveals that identity is stable and unlikely to change (Bartels & Urminsky,
2011, Studies 1, 2) or were asked how easy it would be to come up with two examples of self-
stability (Bartels & Urminsky, 2011, Studies 3, 4). Participants in the self-instability condition
read that research reveals that identity is unstable and likely to change (Studies 1, 2) or were
asked how easy it would be to come up with ten examples of self-stability (Studies 3, 4).
Participants randomly assigned to the stability condition were more willing to wait for a larger
monetary reward (Bartels & Urminsky, 2011, Studies 1, 2), more willing to wait for the price of
a laptop to drop before making a hypothetical purchase (Bartels & Urminsky, 2011, Study 3),
and more willing to wait for larger later rewards (Bartels & Urminsky, 2011, Studies 3, 4).
One experiment compared participants in a self-similarity condition to participants in a
self-dissimilarity condition. In this experiment participants in the self-similarity condition were
instructed to write down aspects of their present identity that would be similar to whom they
would be in five years and provide supporting evidence (Zhang & Aggarwal, 2015, Study 3).
Participants in the self-dissimilarity condition were instructed to write down aspects of their
present identity that would be different from whom they would be in five years and provide
supporting evidence. Self-similarity condition participants were more likely to donate to an
organization their future “me” cared about.
Two other experiments compared participants in a self-dissimilarity condition to
participants in an accessible future “me” condition. In these experiments participants prepared
for a debate that would take place at a later time (Burum et al., 2016, Studies 1, 2). Half of
participants received no further information, but half were told that due to a scheduling error,
ROADMAP FOR CHANGING STUDENT ROADMAPS

125
they would be placed on the opposing team for the debate (dissimilarity condition). The
dependent variable was how much of a boring task participants chose to complete in the current
session and how much of it the put off to the next session when the debate would occur.
Participants in the dissimilarity condition (future “me” would be on an opposing team) left more
work for their future “me” to do.
In seven experiments, no main effect of self-continuity, self-connection, self-stability or
self-similarity was found, and there was no test against an accessible future “me” condition. One
experiment showed a moderator for self-continuity or connection and the other six showed
moderators for self-stability. Specifically, participants randomly assigned to write about
academic possible identities on a picture of a connecting path had greater interest in an academic
workshop than participants randomly assigned to write about academic possible identities on
pictures of separate containers, but only when they previously reported being low in connection
to their possible identity (Landau et al., 2014, Study 7). In five of the self-stability and self-
instability experiments, participants either read about research showing that identity is stable or
read about research showing that identity is unstable. People randomly assigned to the self-
stability condition were less likely to cheat to gain monetary rewards in a lab task, but only when
tempted to do so (Sheldon & Fishbach, 2015, Study 2). They were more likely to save money in
hypothetical spending decisions, but only when the opportunity cost of spending money was
made salient (Bartels & Urminsky, 2015, Studies 3-6). In the sixth self-stability experiment,
participants randomized to provide two examples of identity as stable spent less money in
hypothetical purchases than participants randomized to provide ten examples, but only when the
opportunity cost of spending money on expensive products was made salient (Bartels &
Urminsky, 2015, Study 7).
ROADMAP FOR CHANGING STUDENT ROADMAPS

126
Section 4: Summary of proximity experiments that do not compare a near future “me” to
an accessible future “me”
One experiment focused on the motivating effect of experiencing future “me” as
psychologically close (Peetz, Wilson, & Strahan, 2009). In this experiment, undergraduates were
randomly assigned to two groups. One group was asked to think about their future “me” at
college graduation and mark graduation on a timeline ending 5 years in the future. The other
group was given the same instruction but given a timeline ending 25 years in the future. Four
years in the future (graduation) was marked closer to the present on the 25-year timeline than on
the 5-year timeline and this translated into more academic motivation in the 25-year condition
compared to the 5-year condition. In two other experiments participants were randomly assigned
to one of two letter-writing groups. One group wrote a letter to their future “me” three months
into the future. The other group wrote a letter to their future “me” 20 years into the future.
Compared to participants in the three months condition, participants in the 20 years condition
were less likely to make hypothetical delinquent choices (Van Gelder, Hershfield, Nordgren,
2013, Study 1), and subsequently exercised more (Rutchick et al., 2018, Study 2).

ROADMAP FOR CHANGING STUDENT ROADMAPS

127
Chapter 2: Do you need a roadmap or can someone give you directions: When school-
focused possible identities change so do academic trajectories

Abstract
Interventions seeking to improve the academic outcomes of at-risk students are often based on an
implicit theoretical process model in which academic trajectories are changed by changing what
students imagine is possible for themselves and/or the ways they imagine attaining their possible
identities. However, there is almost no empirical research documenting normally occurring
changes in school-focused possible identities or linking changes in these possible identities to
change in academic outcomes. Moreover, theoretical models are unclear about whether possible
identities are best understood as content or as content linked to strategies. Hence, intervention
researchers cannot know if they need to focus on strategies or not, if they can capitalize on large
normally-occurring shifts, or if they need to create change where none would otherwise occur.
We address these gaps, assessing normally-occurring changes in school-focused possible
identities over the school year and testing whether these changes predict changes in academic
outcomes (n=247 Chicago eighth graders). We find that school-focused possible identities
decline over the school year, that these changes predict changes in academic outcomes, and that
whether linked strategies matter depends on school context. We then use a separate data set
(n=1006 eighth graders) of open-ended responses to possible identity and strategies probes to
train a machine-learning algorithm to code students' possible identities. The codes generated by
the algorithm predict academic outcomes in our original sample. Establishing machine-coding
capacity to count school-focused possible identities is important for scaled-up research because it
provides a method to automate coding of idiographic data.
ROADMAP FOR CHANGING STUDENT ROADMAPS

128
Introduction
September :“[Next year I expect to be]...more punk rock (I'm listening to music that's emo, I
have a lot of street cred and I never leave the house without my choker and looking super
emo.)”.....“[Next year I want to avoid]...talking to [name] (I ignore him at all costs and pray he'll
stop texting me on Kik all the time.)”
May: “[Next year, I expect to be]…getting good grades (Currently, I'm making sure I
don't fall out of habits that include paying attention and doing my homework on a daily
basis.) improving my vocabulary (Currently, I'm reading a lot of books and such in order
to improve my vocabulary.) healthier (I'm currently working out frequently and cutting
out bad foods.) more rational (I'm trying to be sure to look more on the rational side of
things, what's more beneficial for me and what isn't and so on and so forth.) …”[Next
year I want to avoid]…drama (Currently, I mind my own business rather than doing
things that would cause drama like talking about other people or asking about their
private lives.) getting bad grades (In general I avoid getting bad grades by making sure I
understand what's being taught and hopefully I do carry on with that in high school.)
certain people (I hope to be able to avoid certain people next year that haven't necessarily
helped me grow as a person/be a better version of myself, in which they haven't done any
of that in the time that we have known each other so in high school, I'd like to be more
social with people who are going to be good for me rather than be negative and tear me
down.) falling into the influence of others (Now, I don't do things that everyone else does
and I want the same thing to happen later on.)”
— female 8th grader, September and May possible identities (strategies in parentheses).

As our opening quote highlights, students have rich images of their possible identities for
the coming year—the identities they might attain—and they can describe an array of strategies—
things they are doing now to work on these possible identities. Both can change over the school
year. Indeed, the assumption that changes occur and also matter is one reason why many
interventions to improve youth outcomes implicitly evoke change in school-focused possible
identities as their active ingredient (Ansong et al., 2018; Destin & Svoboda, 2017; Lewis &
Yates, 2019; Stephens, Hamedani & Destin, 2014; Stephens, Townsend, Hamedani, Destin, &
Manzo, 2015) or do so explicitly (Elliott, Choi, Destin, & Kim, 2011; Lee, Husman, Scott, &
Eggan-Wiggins, 2015; Rinaldi & Farr, 2018; Wooley et al., 2013). Surprisingly, we found that
intervention studies rarely assess whether change in possible identities occurs or matters for
student academic outcomes, and even basic research on developmental change in possible
ROADMAP FOR CHANGING STUDENT ROADMAPS

129
identities over the course of the school year is lacking. In the current paper we address these gaps
by examining how school-focused possible identities change over the academic year without
intervention and the extent that these changes matter for changes in academic outcomes. Having
done that, we then develop a machine-learning algorithm to test whether changes in
computationally coded school-focused possible identities and linked strategies can predict
changes in academic outcomes. We show that this new method is effective in predicting change
in academic outcomes. Next, we consider the current state of the literature.
Do Possible Identities Change Over The Course of The School Year?
We found little attention to normal developmental change in school-focused possible
identities over the course of the school year. Instead, what the field knows about change over
time in possible identities is limited to the context of health and aging (Frazier, Hooker, Johnson,
& Kaus, 2000; Smith & Freund, 2002) rather than the context of academics or careers and
adolescence. As we summarize next, extant empirical literature has a number of gaps. It has not
addressed the effect of context on the motivational power of possible identities nor has it
addressed the likelihood of change or the direction of change in school-focused possible
identities over the course of the school year. Finally, it has not addressed whether normally-
occurring change in possible identities matters by affecting academic trajectories.
How Might Context Affect Possible Identity Change?
We looked for studies examining context effects on the changes in school-focused
possible identities or in the relationship between change in school-focused possible identities and
changes in academic outcomes. We did not find systematic empirical research on context.
Moreover, the studies we found linking school-focused possible identities at one point in time to
subsequent academic outcomes took place in resource-scarce contexts (e.g., Bi & Oyserman,
ROADMAP FOR CHANGING STUDENT ROADMAPS

130
2015; Oyserman, Gant, & Ager, 1995; Oyserman, Bybee, & Terry, 2004). The same was true for
the randomized control intervention study we found (Oyserman, Terry, & Bybee, 2006).
Participating in intervention to bolster school-focused possible identities and strategies to work
on them changed student possible identities and strategies and this changed students’ academic
trajectories. Moreover, being randomly assigned to participate in the intervention rather than the
school-as-usual control group compensated for the effect of low parental involvement in school
on children’s school grades (Oyserman, Brickman, & Rhodes, 2007). In each of these studies,
children were drawn from high poverty contexts—for example, in rural China (Bi & Oyserman,
2015) or Detroit (Oyserman et al., 2004; 2006; 2007).
There is some evidence that contextual resources matter. In one study with data from four
states, the children of low-income families and the children of families living in low-income
neighborhoods were less likely to have strategies to work on their school-focused possible
identities than children living in higher resourced contexts (Oyserman, Johnson, & James, 2011).
Unfortunately, studies to date have not assessed the possibility that being in a higher or a lower
resourced context affects the relationship between possible identities and academic outcomes.
Potentially, having possible identities is sufficient in contexts in which others can serve as
models or provide direction as to how to proceed, while having strategies (one’s own roadmap
forward) is particularly important in low resource contexts in which others cannot provide the
necessary directions (e.g., Oyserman, et al., 2006).
Possible Identity Change and Intervention
The idea that changes in possible identities occur and matter for academic performance is
implicitly or explicitly the basis for a large number of youth interventions, including those that
focus on middle school (Destin & Svoboda, 2017; Woolley et al, 2013), high school (Rinaldi &
ROADMAP FOR CHANGING STUDENT ROADMAPS

131
Farr, 2018), and college (Stephens et al., 2014; 2015). However, these interventions do not
typically measure change in possible identities and whether these changes are associated with
changes in academic outcomes. In fact, Horowitz and Oyserman’s (2019) methodic review of
decades of research on the effects of future or possible identities on current action found only
one intervention study linking change in possible identities to academic outcomes (Oyserman et
al., 2006) and none linking normally occurring developmental changes in possible identities to
academic outcomes.
It seems that intervention developers largely draw on two literatures in assuming that
changes in possible identities occur and matter for academic outcomes. The first literature links
school-focused possible identities
9
at a single time point to academic outcomes (e.g., Bi &
Oyserman, 2015; Oyserman, Gant, & Ager, 1995; Oyserman, Bybee, Terry, & Hart-Johnson,
2004). While useful, these studies do not provide evidence that the proposed active ingredient—
change in school-focused possible identities—occurs or matters for academic outcomes. The
second literature documents that interventions or lab manipulations can change possible
identities. However, only one study—a randomized trial of an intervention—documents a change
in possible identities and that this change is associated with change in academic trajectories
(Oyserman et al., 2006). Six additional studies document that intervening to change possible
identities affects these identities, but none show that these changes actually matter for changing
academic trajectories, and none can shed light on normal developmental changes. Specifically,

9
Operationalizations differ in whether they focus on possible identities alone or in combination
with strategies. The former includes counting school-focused identities (Bi & Oyserman, 2015),
counting ‘balanced’ positive and negative school-focused identities (Oyserman, et al., 1995), and
coding for possible identity concreteness (Rathbone et al., 2016). The latter includes counting
whether students say they were doing something, without asking them to write down strategies
(Oyserman & Saltz, 1993), counting school-focused possible identities and strategies to attain
them (Oyserman, et al., 2004), or a scoring ‘plausible’ possible identity roadmaps (Oyserman, et
al., 2004).
ROADMAP FOR CHANGING STUDENT ROADMAPS

132
two lab experiments document that students given negative feedback about the likelihood of
attaining their academic or career possible identities reframe these identities (Carroll, Shepperd,
& Arkin, 2009; Kerpelman & Pittman, 2001). Four other intervention studies provide some
evidence that specific programs or activities can change school-focused possible identities, but
none link these changes to changes in academic outcomes (Kortsch, Kurtines, & Montgomery,
2008; Lee et al., 2015; Oyserman, Terry, Bybee, 2002; Stake & Nickens, 2005).
As result, it is unclear how school-focused possible identities are likely to change over
time, and whether this change is associated with academic outcomes. This is problematic
because understanding this process of change is critical for intervention design. If school-focused
possible identities and strategies tend to be stable over time, interventions would need to focus
on getting participants to imagine their futures in new ways and to articulate new ways to get
going working on these possible futures. If school-focused possible identities tend to decline,
interventions would need to focus on stemming this decline. If they tend to increase,
interventions would need to focus on enhancing the kinds of strategies students have, ensuring
that students have a plausible roadmap to work toward these possible futures. Possible identities
could be motivating in each of these three scenarios so none can be ruled out. Thus, if possible
identities and strategies are stable, then imagining one’s future identities may be motivating
because future identities can stay the same even in the face of changes in the present. That is,
even students struggling in 8
th
grade can still believe that in 9
th
grade they will be on the honor
roll and that doing homework and asking for help are strategies to get there. Similarly if possible
identities and strategies change, then imagining one’s future identities may be motivating
because of this change. As 8
th
graders get closer to high school they may envision a new set of
positive academic possibilities associated with being older and having more academic options.
ROADMAP FOR CHANGING STUDENT ROADMAPS

133
For example, a struggling 8
th
grader may decide that excelling in English is impossible but that
focusing on Physics is a reasonable goal.
Current Study
We addressed three open questions: do school-focused possible identities change over the
course of the school year, does change matter by affecting academic outcomes, and does context
affect whether students need strategies in addition to possible identities. We focused on students
who might be at risk of poor academic outcomes due to their minority background and
socioeconomic status, but who attend schools varying in resources. In attempting to answer these
questions we used operationalizations of school-focused possible identities empirically linked to
academic outcomes.
Sample
Our sample constituted Chicago Public School 8
th
-graders (N=461) attending one of 7
schools participating as the control group in a larger intervention development grant. We
obtained written parental consent, excluding children whose parents refused consent (n=50) and
or who entered the classrooms after consent forms were collected (n=91)
10
. Of the remaining 320
children, some were missing possible identity data (n= 40) or school records (n=33), reducing
our final analytic sample to n=247 (55% female, 92% low-income—receiving free or reduced
price lunch—83% self-described as Latinx, 14% self-described as African American, 2% White,
1% a different race-ethnicity).

10
To insure that all forms were handed in entailed a lengthy process and had to be done in a way
that fit the school’s requirements. Students who returned as signed form (whether parents signed
‘no’ or ‘yes’) were given movie tickets. Once the process was complete, it was too disruptive to
the classrooms to start the process again.
ROADMAP FOR CHANGING STUDENT ROADMAPS

134
Method
Human subjects and power
We obtained IRB approval (approval number 03635 American Institutes for Research,
project name IES SCHOOL JOBS DEVELOP). Our planned sample was sufficient to detect a
small effect (f
2
=.02; Cohen, 1988) with a power of .80 at p < .05, for which, according to our
sensitivity analysis, we needed a sample of 394.
Possible identities and linked strategies
In September and May students logged onto an online questionnaire in their classrooms.
At the start of the questionnaire, students saw the prompts for expected and feared possible
identities that are detailed in Figure 1. At each time point (September, May) students could write
four expected and four feared possible identities and strategies for a maximum of 8 expected and
8 feared responses (16 total). The first author and a research assistant, blind to other student
information, used an on-line coding scheme (https://dornsife.usc.edu/daphna-
oyserman/measures/). They coded four operationalizations of school-focused possible identities:
Count of school-focused possible identities, count of ‘balanced’ feared and expected school-
focused identity pairs, count of school-focused possible identities with linked strategies, and
plausibility score of school-focused possible identities and linked strategies. Each
operationalization is detailed next.
ROADMAP FOR CHANGING STUDENT ROADMAPS

135

Figure 1. Full instructions for possible identity and strategies, adapted from Oyserman et al.,
2006.

ROADMAP FOR CHANGING STUDENT ROADMAPS

136
Measures
School-focused Possible Identity Count. After training to reach 90% agreement, our
coders double-coded all n=1,527 expected and n=1,473 feared possible identity responses by
reading the full entry, including strategies if any. They agreed 89.6% of the time and discussed
disagreements to agreement. As Table 1 reveals, school and achievement was the dominant focus
(70% of fall, 56% of spring) responses; the only other categories with 10% or more of responses
were off-track (11% fall, 16% spring) and interpersonal relationships (10% fall, 18% spring).
Table 1.
Percentage of responses reflecting each possible identity domain
Possible Identity
Domain
Fall Spring
School-focused 70% 56%
Off-track 11% 16%
Interpersonal
Relationships
10% 18%
Material/Lifestyle 5% 3%
Personality Traits 2% 3%
Health/Physical 2% 3%
Negative but
expected
<1% <1%
Note: School-focused include possible identities focused on school
and on achievement.

Balanced School-focused Possible Identities Count. Next, the coders counted the
number of ‘balanced’ school-focused possible identities (Aloise-Young, Hennigan, & Leong,
2001; Oyserman, et al., 1995). ‘Balance’ is a count of corresponding positive and negative
school-focused possible identity pairs (e.g. “on the honor roll” and “getting bad grades”). Coders
again double coded and discussed to agreement any disagreements.
ROADMAP FOR CHANGING STUDENT ROADMAPS

137
School-focused Possible Identities with Strategies. Coders also counted the number of
school-focused possible identities with at least one linked strategy (Oyserman & Saltz, 1993)
using the same double coding procedure described above.
School-focused Possible Identity Plausibility Score. The first author coded plausibility
(Oyserman et al., 2004) using the rubric in Table 2. Plausibility, the extent school-focused
possible identities and linked strategies formed a roadmap forward, is meant to reflect more than
just whether there are strategies, but to account for whether the possible identities and strategies
are concrete and involve interpersonal aspects of the school context. To obtain a reliability score,
a second research assistant double coded a random sample of 10% responses. Our Cohen’s
Kappa (Fleiss & Cohen, 1973) was 0.85, reflecting substantial agreement following Landis and
Koch’s (1977) Cohen’s Kappa rule-of-thumb that scores between 0.61 and 0.8 reflect substantial
agreement. As percentage agreement, reliability was 88%, with no disagreements differing by
more than a single point on the 0-5 plausibility scale.

ROADMAP FOR CHANGING STUDENT ROADMAPS

138
Table 2.
Metric for coding school-focused possible self plausibility, adapted from Oyserman et al., 2004.
Plausibility
Score
Count of
Academic
Expected or
Feared Possible
Identities (API)
Count of
Strategies
attached to
these API
Codes noted with * mean code at this level only
if at least one of the possible selves and/or
strategies that are provided are
detailed/concrete, that is if specific action is
implied and possible selves are not redundant,
otherwise code at the next lower level of
plausibility.

0

0 EITHER 0 academic possible selves (API)
OR
1 API that is vague or general AND 0 API
strategy
1

0

1
1 1 EITHER 1 API and 1 API strategy
OR
2 APS but no APS strategies 2 0

2
1 2* or more EITHER 1 API and 2 or more API strategies*
OR
2 APS and 1- 2 API strategies
OR
3 APS and 0*-1 API strategies
OR
4 or more API and 0 API strategies
2 1- 2
3 0*-1
4 or more 0

3

2 3* or more EITHER 2 API and 3 or more API strategies*
OR
3 API and 2-3 APS strategies
OR
4 or more APS and 1*-2 API strategies
3 2-3
4 or more 1*-2

4
3 4 or more EITHER 3 API and 4 or more API
OR
4 API and 2*-4 API strategies
4 or more 2*, 3-4

5
4 or more 4-5+ 4 or more API AND 4 or more strategies API at
least one strategy for an academic possible
identity is focused on interpersonal aspects of
school context.

Academic Outcomes. Chicago Public Schools provided 6
th
, 7
th
grade, and 8
th
grade
course grades as part of a data sharing agreement with the American Institutes for Research. We
ROADMAP FOR CHANGING STUDENT ROADMAPS

139
computed final 6
th
, 7
th
, and 8
th
core grade point average (GPA) by computing the average final
grades in Math, Science, English, History, and Social Studies (0=F, 1=D, 2=C, 3=B, 4=A).
School-level Context. We used the Common Core of Data
(https://nces.ed.gov/ccd/pubschuniv.asp) to obtain school-level enrollment percentage of students
eligible for free or reduced price lunch, enrollment of minority students, and number of full-time-
equivalent classroom teachers. We used these data to calculate each school’s student to teacher
ratio, percentage of students in poverty (students who were eligible for free or reduced price
lunch), and percentage of students who identified as Latinx or African American.
Results
Preliminary analyses
In preliminary analyses we examined the association between our child-level
demographics and school-level variables and our possible identity and academic outcome
variables. Child-level and school-level poverty were significantly but not highly correlated
(r=.19 p < .01). Child-level and school-level race-ethnicity metrics were highly correlated
(Latinx, r=.76 p <.01; Black, r=.83 p <.01) and the school-level metrics of the two most common
racial-ethnic groups—Black and Latinx, were almost perfectly negatively correlated (r=-.99 p
<.01). This meant that we could include child-level and school-level poverty rates but only one
race-ethnicity variable (we chose the most common, percentage Latinx).
Regarding academic outcomes, being female was associated with 6
th
, 7
th
, and 8
th
grade
final core GPA. Being from a poor family (receiving free or reduced-price lunch) and school-
level poverty were each associated with 6
th
and 8
th
grade core GPA. School-level poverty was
also associated with change in core GPA from 6
th
to 7
th
and from 7
th
to 8
th
grade. Being Latinx
and school-level Latinx enrollment were each associated with 7
th
grade core GPA and change in
ROADMAP FOR CHANGING STUDENT ROADMAPS

140
core GPA from 6
th
to 7
th
grade. School-level Latinx enrollment was also associated with change
in core GPA from 7
th
to 8
th
grade. School-level student-teacher ratio was associated with 6
th

grade core GPA and change in core GPA from 6
th
to 7
th
grade and 7
th
to 8
th
grade.
Regarding school-focused possible identity measures, being female was associated with
both balance and plausibility at the end of 8
th
grade and with change in balance and plausibility
over the course of 8th grade. Being Latinx was associated with change in the count of school-
focused possible identities over the course of 8
th
grade. None of the school level variables were
associated with our school-focused possible identity measures.
Given these associations, we included in our analyses child gender, child-level and
school-level measures of free and reduced-price lunch, school-level student-teacher ratios, and
Latinx enrollment.
Do school-focused possible identities change over the course of the school year?
Changes in students’ school-focused possible identities are common, no matter how these
school-focused possible identities are operationalized. Table 3 presents Sept-May paired t-tests
with 95% confidence intervals. As detailed in Table 3, on average, scores decline significantly
(d=.25 balance, d=.17 count of school-focused possible identities with strategies, d=.18 school-
focused roadmap plausibility score, and d=.11 count of school-focused possible identities).
Finding a decline over the course of the school year without intervention implies that a first task
of any intervention may be to stave off decline, especially in how school-focused possible
identities are structured and linked with strategies. Table 4 organizes results into three groups.
Students whose school-focused possible identity scores remained stable (nearly a quarter of
students), those whose scores declined from fall to spring (over 40% of students), and those
whose scores increased from fall to spring (the remaining students).
ROADMAP FOR CHANGING STUDENT ROADMAPS

141
Table 3.
Mean (SD), 95% confidence intervals, and paired-sample t-test results for change in school-focused possible identity metrics
Measure of School-Focused Possible Identities Fall Spring t-test of
Change
95% CI of
Change
p
M (SD) M (SD)
Simple Count (0=none, 8=all) 3.86 (2.09) 3.59 (1.85) -.1748 -.586, .035 .082
Balance Count (0=none, 4=all) 1.35 (1.04) 1.05 (0.95) -3.869 -.458, -.149 .000
With Strategies Count (0=none, 8=all) 3.42 (2.12) 3.00 (1.91) -2.614 -.738, -.104 .009
Plausibility Score (0=none, 5=all) 3.54 (1.49) 3.20 (1.58) -2.759 -.583, -.097 .006
Note: Count scores range from a theoretical minimum of 0 to a theoretical maximum of 8, balance relies on pairs of possible
identities so the theoretical maximum is 4, and plausibility is a score with a theoretical maximum of 5, coded for students with
multiple possible identities focused on school linked to concrete strategies for action, including at least one strategy focused
on the social context of school.

Table 4.
Percentage of participants who had an increase, no change, or a decrease in their
school-focused possible identities on each metric.
Measure of School-Focused Possible Identities Increase No Change Decrease
Simple Count 34% 19% 47%
Balance Count 23% 35% 42%
With Strategies Count 34% 16% 49%
Plausibility Score 30% 28% 42%
Average 30% 25% 45%
ROADMAP FOR CHANGING STUDENT ROADMAPS

142
Do changes in school-focused possible identities predict changes in academic outcomes and
are these effects context-dependent?
We created a change score for each of the four school-focused possible identity metrics
by regressing May scores on September scores and saving the residuals. We then standardized
these change scores and student GPA measures and created four hierarchical multiple regression
equations for each possible identity metric (16 equations in total) with 8
th
grade core GPA as our
dependent variable. In the first model for each possible identity metric we entered our
residualized school-focused possible identity change score. In the second model we added school
dummy codes to control for school effects. In the third model we added 6
th
and 7
th
grade core
GPA to control for prior academic performance. In the fourth model, we entered contrast codes
for gender, for being Latinx, and for receiving free or reduced price lunch in order to control for
student-level characteristics.
Table 5 details the effects of each possible identity metric in each model. As can be seen,
each possible identity metric significantly predicts 8
th
grade core GPA and these effects hold
even when controlling for school, prior core GPA, and individual demographics. As can be seen
in Model 1, possible identity plausibility score is the strongest predictor of 8
th
grade GPA,
explaining at least 25% more variance in 8
th
grade core GPA than any of the other possible
identity metrics. However, once school context is accounted for, the stronger effect of
plausibility is mitigated, and the effects of possible identity metrics that do not include strategies
are about the same as that of possible identity plausibility. As can be seen in Model 2, once
school context is accounted for, the lower end of the 95% confidence intervals increase for the
regression coefficients for the two possible identity metrics that do not account for strategies.
This suggests that schools differ in the extent to which they help students self-regulate and craft
ROADMAP FOR CHANGING STUDENT ROADMAPS

143
strategies to attain their school-focused possible identities. Some schools seem to provide
opportunities to engage in strategies for anyone with a possible identity, while other schools do
not. As can be seen in Models 3 and 4, possible identities matter even when prior academic
attainment and demographics are accounted for.
Overall, the real-world effect of change in school-focused possible identities is
meaningful when considered in the context of the stability of grades and size of the effect of
demographics. Grades are very stable over time—in our sample the average absolute value of the
change in students’ core GPA is .44, and average core GPA increases over the school year by
just over a quarter of a letter grade (0.2648). After controlling for prior core GPA, school, and
individual poverty and demographics, change in the count of students’ school-focused possible
identities explains about 3.8 % of the remaining variance in 8
th
grade core GPA. This compares
favorably to the predictive power of student gender, race-ethnicity, and poverty, which together
explain about 2.1% of the remaining variance in 8th-grade core GPA after school and prior GPA
are accounted for. Post-hoc sensitivity analyses suggest we would need a sample of n=209 to
detect effects of this size, so we are powered to find this effect, as well as the effect of count in
possible identities without these controls (Model 1), which would require a sample of n=218.
We ran a follow-up regression model for each possible identity score to investigate how
aspects of the school context might influence the effect of strategies. In this model, we entered
each possible identity score at the first step, but instead of entering school dummy codes at the
second step, we entered school-level measures of Latinx ethnicity, free or reduced price lunch
eligibility, and student-teacher ratios. These results are presented as Model 5 in Table 6. As can
be seen, the school-level resource factors we entered in Model 5 are significant predictors of 8
th

grade core GPA, but they do not fully account for school-level effects on whether it is necessary
ROADMAP FOR CHANGING STUDENT ROADMAPS

144
to account for student strategies in predicting the effect of change in school-focused possible
identities on change in academic outcomes.
ROADMAP FOR CHANGING STUDENT ROADMAPS

145
Table 5.
Effect of change in school-focused possible identity metrics on 8
th
grade core GPA
Model 1 Model 2 Model 3 Model 4
Predictor B 95% CI p B 95% CI p B 95% CI p B 95% CI p
School-focused possible
identity count
.190 .066, .313 .003 .210 .099, .321 .000 .088 .032, .145 .002 .087 .030, .144 .003
School-focused possible
identity balance count
.178 .054, .302 .005 .207 .097, .317 .000 .084 .028, .141 .003 .084 .027, .140 .004
School-focused possible
identities with strategies count
.183 .059, .307 .004 .178 .066, .291 .002 .090 .033, .146 .002 .089 .033, .146 .002
School-focused possible
identities plausibility score
.220 .097, .342 .001 .210 .099, .322 .000 .078 .021, .135 .008 .075 .017, .132 .011
Notes: In each model the dependent variable is 8
th
grade core grade point average; Model 1 includes no additional predictors; Model 2
includes school dummy codes; Model 3 includes prior core GPA in 6
th
and 7
th
grades; Model 4 includes student-level demographic
and poverty data; When B is positive, the possible identity predictor is associated with higher core grade point average in 8
th
grade.

Table 6.
Effect of change in possible identity metrics on 8
th
grade Core GPA, controlling for school level context.
Model 5
Predictor B 95% CI p
School-focused possible identity count .222 .101, .343 .000
School-focused possible identity balance count .205 .084, .325 .001
School-focused possible identities with strategies count .206 .085, .327 .001
School-focused possible identities plausibility score .242 .123, .361 .000
Notes: The dependent variable is 8
th
grade core grade point average; Model 5 controls for three measures of school context: the
school-level free and reduced-price lunch rate, the student-teacher ratio, and the percentage of the school that identifies as Latinx.

ROADMAP FOR CHANGING STUDENT ROADMAPS

146
Secondary Analyses
Our core finding, that normally-occurring changes in possible identities over the course
of the school year predict change in academic outcomes is an important finding for
developmental researchers and good news for interventionists who had assumed this to be the
case. However, coding open-ended data is time-consuming and resource intensive and may not
be possible in scaled up research or interventions. Hence, to create an alternative to time and
labor intensive coding, we developed a machine-learning classification algorithm that could
computationally code the content of students’ possible identities.
Sample
We used a separate training sample (n=1171) of Chicago Public School 8
th
graders to
develop our algorithm. We obtained written parental consent, excluding children whose parents
refused consent (n=58) and children who entered the classrooms after consent forms were
collected (n=77). In addition, some children were missing possible identity data (n= 30),
reducing our final sample (n=1006, 52% female, 87% free/reduced price lunch eligible, 65%
Latinx, 18% African American, 12% White, 4% Asian, 1% other race-ethnicity).
Method
Data collection and coding
We used the data collection and coding method we used in our main study, collecting
data in September and May, coding n=6,189 expected and n=6,060 feared possible identity
responses. The first author and two research assistants double-coded 80% of all responses,
discussing disagreements to agreements. The remaining 20% of responses were coded by the
first author and a research assistant double coded a subset of 20% of these responses (89.55%
agreement before discussion). Table 7 shows the distribution of responses.
ROADMAP FOR CHANGING STUDENT ROADMAPS

147
Table 7.
Content of possible selves: Percentage of responses in each domain in
algorithm development sample (as coded by researchers)
Content Domain Development Sample Fall
and Spring Combined
School/Achievement 59%
Off-track 16%
Interpersonal Relationships 14%
Material/Lifestyle 4%
Personality Traits 3%
Health/Physical 3%
Negative but expected <1%

Training the classifier
We trained the classifier on our researcher-coded responses. We made three decisions.
First, given that most responses were school-focused and only off-track, and interpersonal
categories garnered 10% or more of responses, we asked our machine algorithm to code into four
groups—school-focused, off-track, interpersonal, and “other.” Second, given that researchers use
both possible identity and strategy content to code possible identities, we combined these blocks
of text so the algorithm had access to both to code a given response. Third, given that expected
and feared possible identity responses had different content, we trained separate classifiers for
each (in case machine coding worked only for one or the other).
Our algorithm used support vector machine classification and Distributed Dictionary
Representations (DDR; Garten et al., 2018; Hoover, Johnson, Boghrati, Graham, & Dehghani,
2018). In DDR, words are represented as points in a low-dimensional space, generally 10-2000
dimensions. Psychological constructs are represented in the space as the average of the
ROADMAP FOR CHANGING STUDENT ROADMAPS

148
representations of all the words associated with it. Features of a new piece of text are then
classified by first averaging the locations of its words to create a representation of that text, and
then evaluating the distance between this representation and the representation of relevant
psychological constructs.
We used Word2Vec (Mikolov et al., 2013), trained on a corpus of Google News articles,
to generate word embeddings for each of the words in the responses. We then summed up the
word embeddings for all the words in the response (Garten et al. 2015) to generate a ‘response-
level’ representation for each data point. Next, we trained Support Vector Machine classifiers
(Joachims, 1998) on the response-level embeddings and the human code for that response. We
used 10-fold cross validation to evaluate how well our classifiers—one for expected possible
identities and one for feared—coded possible identities into the four domains (school-focused,
interpersonal, off-track, other). Our expected possible identity classification was accurate for
89.18% of expected possible identity responses. Our feared possible identity classification was
accurate for 85.64% of feared possible identity responses. We describe the algorithm
development in more detail and provide the python code in our Supplemental Materials. Using
the algorithm we developed with our “development” sample (n=1006), we classified each
possible identity response from our study sample (n=247) to generate a count of machine-coded
school-focused possible identities. In this sample, our machine codes matched researcher codes
for 90.9% of expected and 88.3% of feared possible identity responses.
Results
We tested the possibility that a machine-coded version of school-focused possible
identities could be substituted for researcher coding. First, we asked if the pattern of decline from
Fall to Spring that we saw with our researcher-coded school-focused possible identities was
ROADMAP FOR CHANGING STUDENT ROADMAPS

149
replicated. It did replicate, but the decline was not significant for the machine-coded version of
school-focused possible identities (fall M=3.95, SD=2.00, spring M=3.78, SD=1.89, paired t-test,
t=-1.09, 95% CI [-.477, .137], p=.276, d=.069). Second, we asked if the predictive power of
change in possible identities on change in academic outcomes replicates with our machine-coded
version of school-focused possible identities. To address this question we created a standardized
change score for the machine-coded school-focused possible identity scores by regressing May
scores on September scores and saving the residuals. As detailed in Table 8, we used this
variable in four regression equations. In Model 6 we tested whether change in machine coded
possible identities predicted 8
th
grade core GPA. In Model 7 we added dummy codes for school.
In Model 8 we added prior core GPA in 6
th
and 7
th
grade. In Model 9 we entered student-level
demographic and poverty measures. As Table 8 details, change in machine coded school-focused
possible identities predicted 8
th
grade core GPA, even when controlling for school, prior
academic performance, and student level demographics. Machine-coding, when compared to
researcher-coding (Table 4), yields slightly weaker but still useful effects.
ROADMAP FOR CHANGING STUDENT ROADMAPS

150
Table 8.
Effect of change in machine-coded school-focused possible identities on 8
th
grade Core GPA.
Model 6 Model 7 Model 8 Model 9
Predictor B 95% CI p B 95% CI p B 95% CI p B 95% CI p
Machine-coded school-focused
possible identity score
.162 .038, .286 .011 .154 .042, .266 .007 .073 .016, .129 .012 .073 .016, .130 .012
Notes: In each model the dependent variable is 8
th
grade core grade point average; Model 6 includes no additional predictors; Model 7
includes school dummy codes; Model 8 includes prior core GPA in 6
th
and 7
th
grades; Model 9 includes student-level demographic
and poverty data; When B is positive, the possible identity predictor is associated with higher core grade point average in 8
th
grade.
ROADMAP FOR CHANGING STUDENT ROADMAPS

151
Discussion
We addressed three open questions: do students’ school-focused possible identities
change over the course of the school year, is change in possible identities associated with change
in academic trajectories, and in what ways does school context matter. We found that most
students experienced change in their school-focused possible identities over the course of the
school year and that change mattered, affecting students’ academic trajectory even controlling
for two prior years of grades. We also found that each of the possible identity metrics we tested
were useful predictors of change in academic outcomes, even our newly developed machine-
coded version. Finally, we found that school context matters. Depending on school-context,
taking strategies into account can add to the motivational force of school-focused possible
identities. We found that effect of change in school-focused possible identities is consequential.
Indeed, the variance explained by change in school-focused possible identities is larger than the
variance explained by child gender, poverty, and race-ethnicity combined.
Our results imply that possible identities are malleable and open to change, that changing
them matters, and that school-level effects are significant. These results imply that interventions
targeting changing possible identities and strategies can matter, that students are open to
changing their possible identities, and hence, intervention focus can be on directing and
stabilizing school-focused identities and strategies. All of this is good news for interventions to
reduce students’ academic aspiration-attainment gaps because interventions often that assume
but do not test that changing possible identities is consequential, and that participation changes
students’ school-focused possible identities and strategies for action.
Moreover, since coding open-ended responses is a stumbling block for researchers,
especially those who wish to use large data sets and evaluate scaled-up interventions, we used a
ROADMAP FOR CHANGING STUDENT ROADMAPS

152
separate sample to develop a machine-coded version of school-focused possible identities. We
share our newly developed machine-coding algorithm code so that others can use it in their own
research. We document that this algorithm is effective. The implication is that researchers
wanting to study possible identity effects at scale can use an ideographic metric which allows
students to express their possible identities and quantify results without requiring costly coding.
Any study, of course, has limitations. We consider here three: lack of experimental
control, limited access to school context variables, and use of a single geographic region
(Chicago). First, with regard to experimental control, our results do address an important gap
regarding the temporal stability of possible identities and the consequences of change, we
document temporal change in possible identities over the course of the school year and effects of
this change on change in academic trajectories. However, we cannot make causal claims since
we since even though we controlled for two prior years of academic attainment, child-level
demographics, and school-level factors, we did not manipulate change in possible identities.
Hence our research is informative of normal developmental trajectories, not of causal processes.
Second, with regard to school context variables, our study obtained data from seven
schools, allowing us to begin to test school context effects. However, our school-level variables
were limited. That meant that although we could document that school context matters, we were
not able to fully unpack why. Future research is needed to better understand what about schools
differs such that in some schools, students need their own roadmap (the strategies to get going
and keep on track) and in other schools, someone else can provide directions as long as students
have the school focused possible identities. Our hunch is that in some schools, parents, teachers,
and classmates are able to provide students with needed directions, while in other schools,
students need to carry their own roadmap with them.
ROADMAP FOR CHANGING STUDENT ROADMAPS

153
Third, with regard to geographic region, our study documented effects in one geographic
region, Chicago, and developed a machine coding algorithm from students in the same large
urban school district. While important first steps, future research is needed to test stability of our
results in different settings and the ability of our algorithm to code responses from students in
different regions, schools, and settings.
Irrespective of these limitations, our results are important because they provide evidence
that changes in school-focused possible identities and strategies influence academic trajectories
over time. These results are congruent with the theorized process of behavior change underlying
numerous interventions, but our results are the first to show that this normal developmental
change process occurs and matters for academic outcomes. Our results support the idea that
inducing changes in school-focused possible identities is suitable path for influencing academic
trajectories.

ROADMAP FOR CHANGING STUDENT ROADMAPS

154
References
Ansong, D., Chowa, G., Masa, R., Despard, M., Sherraden, M., Wu, S., & Osei-Akoto, I. (2018).
Effects of youth savings accounts on school attendance and academic performance:
evidence from a youth savings experiment. Journal of Family and Economic Issues, 1-13.
Bi, C., & Oyserman, D. (2015). Left behind or moving forward? Effects of possible selves and
strategies to attain them among rural Chinese children. Journal of Adolescence, 44, 245-
258.
Carroll, P. J., Shepperd, J. A., & Arkin, R. M. (2009). Downward self-revision: Erasing possible
selves. Social Cognition, 27(4), 550-578.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd edition). Hillsdale,
NJ: Lawrence Erlbaum.
Destin, M., & Svoboda, R. C. (2017). A brief randomized controlled intervention targeting
parents improves grades during middle school. Journal of Adolescence, 56, 157-161.
Elliott, W., Choi, E. H., Destin, M., & Kim, K. H. (2011). The age-old question, which comes
first? A simultaneous test of children's savings and children's college-bound
identity. Children and Youth Services Review, 33(7), 1101-1111.
Fleiss, J. L., & Cohen, J. (1973). The equivalence of weighted kappa and the intraclass
correlation coefficient as measures of reliability. Educational and Psychological
Measurement, 33(3), 613-619.
Frazier, L. D., Hooker, K., Johnson, P. M., & Kaus, C. R. (2000). Continuity and change in
possible selves in later life: A 5-year longitudinal study. Basic and Applied Social
Psychology, 22(3), 237-243.
ROADMAP FOR CHANGING STUDENT ROADMAPS

155
Garten, J., Hoover, J., Johnson, K. M., Boghrati, R., Iskiwitch, C., & Dehghani, M. (2018).
Dictionaries and distributions: Combining expert knowledge and large scale textual data
content analysis. Behavior Research Methods, 50(1), 344-361.
Garten, J., Sagae, K., Ustun, V., & Dehghani, M. (2015). Combining distributed vector
representations for words. In Proceedings of the 1st Workshop on Vector Space Modeling
for Natural Language Processing (pp. 95-101).
Hoover, J., Johnson, K., Boghrati, R., Graham, J., & Dehghani, M. (2018). Moral framing and
charitable donation: Integrating exploratory social media analyses and confirmatory
experimentation. Collabra: Psychology, 4(1), 9.
Horowitz, E. & Oyserman, D. (under review). How, when, and why thinking about future “me”
influences current choices.
Joachims, T. (1998, April). Text categorization with support vector machines: Learning with
many relevant features. European Conference on Machine Learning (pp. 137-142).
Springer, Berlin, Heidelberg.
Kerpelman, J. L., & Pittman, J. F. (2001). The instability of possible selves: Identity processes
within late adolescents' close peer relationships. Journal of Adolescence, 24(4), 491-512.
Kortsch, G., Kurtines, W. M., & Montgomery, M. J. (2008). A Multistage Longitudinal
Comparative (MLC) Design Stage II: Evaluation of the Changing Lives Program (CLP):
The Possible Selves Questionnaire—Qualitative Extensions (PSQ-QE). Journal of
Adolescent Research, 23(3), 342-358.
Landis, J. R., & Koch, G. G. (1977). The measurement of observer agreement for categorical
data. Biometrics, 33(1), 159-174.
ROADMAP FOR CHANGING STUDENT ROADMAPS

156
Lee, J., Husman, J., Scott, K. A., & Eggum-Wilkens, N. D. (2015). Compugirls: Stepping stone
to future computer-based technology pathways. Journal of Educational Computing
Research, 52(2), 199-223.
Lewis Jr, N. A., & Yates, J. F. (2019). Preparing disadvantaged students for success in college:
Lessons learned from the preparation initiative. Perspectives on Psychological
Science, 14(1), 54-59.
Mikolov, T., Sutskever, I., Chen, K., Corrado, G. S., & Dean, J. (2013). Distributed
representations of words and phrases and their compositionality. In C. J. C. Burges, L.
Bottou, M. Welling, Z. Ghahramani, & K. Q. Weinberger (Eds.), Advances in neural
information processing systems (pp. 3111–3119). Redhook, NY: Curran Associates Inc.
Oyserman, D., Brickman, D., & Rhodes, M. (2007). School success, possible selves, and parent
school involvement. Family Relations, 56(5), 479-489.
Oyserman, D., Bybee, D., & Terry, K. (2006). Possible selves and academic outcomes: How and
when possible selves impel action. Journal of Personality and Social Psychology, 91(1),
188-204.
Oyserman, D., Bybee, D., Terry, K., & Hart-Johnson, T. (2004). Possible selves as roadmaps.
Journal of Research in Personality, 38(2), 130-149.
Oyserman, D., Gant, L., & Ager, J. (1995). A socially contextualized model of African American
identity: Possible selves and school persistence. Journal of Personality and Social
Psychology, 69(6), 1216.
Oyserman, D., Johnson, E., & James, L. (2011). Seeing the destination but not the path: Effects
of socioeconomic disadvantage on school-focused possible self content and linked
behavioral strategies. Self and Identity, 10(4), 474-492.
ROADMAP FOR CHANGING STUDENT ROADMAPS

157
Oyserman, D., & Saltz, E. (1993). Competence, delinquency, and attempts to attain possible
selves. Journal of Personality and Social Psychology, 65(2), 360-374.
Oyserman, D., Terry, K., & Bybee, D. (2002). A possible selves intervention to enhance school
involvement. Journal of Adolescence, 24, 313– 326
Rinaldi, R. L. & Farr, A. C. (2018). Promoting positive youth development: a psychosocial
intervention evaluation. Psychosocial Intervention, 27(1), 22-34.
Smith, J., & Freund, A. M. (2002). The dynamics of possible selves in old age. The Journals of
Gerontology Series B: Psychological Sciences and Social Sciences, 57(6), P492-P500.
Stake, J. E., & Nickens, S. D. (2005). Adolescent girls’ and boys’ science peer relationships and
perceptions of the possible self as scientist. Sex Roles, 52(1-2), 1-11.
Stephens, N. M., Hamedani, M. G., & Destin, M. (2014). Closing the social class achievement
gap: A diversity education intervention improves first-generation students’ academic
performance and all students’ college transition. Psychological Science, 25, 943-953.
Stephens, N. M., Townsend, S. S., Hamedani, M. G., Destin, M., & Manzo, V. (2015). A
difference-education intervention equips first-generation college students to thrive in the
face of stressful college situations. Psychological Science, 26(10), 1556-1566.
Woolley, M. E., Rose, R. A., Orthner, D. K., Akos, P. T., & Jones-Sanpei, H. (2013). Advancing
academic achievement through career relevance in the middle grades: A longitudinal
evaluation of CareerStart. American Educational Research Journal, 50(6), 1309-1335.

ROADMAP FOR CHANGING STUDENT ROADMAPS

158
Supplemental Materials
Section 1: File Setup and Python Code for Developing and Using Possible Selves Classifier

File Setup

The code is written to read possible self training data in the following format (The code ensures
that only the possible self code is used to train the algorithm.)

ID Possible Self CODE Strategy CODE
1 On honor roll 1 Study every day 1
2 More popular 2 Hang out after school 2
3 Playing my x-box 5 Do my chores 5

The code is written to classify or code uncoded possible self data in the following format:

ID Possible Self Strategy
1 On honor roll Study every
day
2 More popular Hang out after
school
3 Playing my x-box Do my chores

Running the code will output possible self classification in the format below. The ‘Combined”
column shows which words in the responses were used by the algorithm. The ‘Vecs’ column can
be ignored. The ‘Class’ column has the possible self code generated by the algorithm.

ID Possible Self Strategy Combined Vecs Class
1 On honor roll Study every day [‘honor’, ‘roll’, ‘study’,
‘every’, ‘day’]
[-0.0802 -
0.34582]
1
2 More popular Hang out after school [‘More’, ‘pouplar’,
‘Hang’, ‘after’, ‘school’]
[ 1.416e-01
5.060e-01]
2
3 Playing my x-box Do my chores [‘Playing’, ‘x-box’,
‘chores’]
[-0.02734
0.103027]
0

ROADMAP FOR CHANGING STUDENT ROADMAPS

159

Python Code

#Import the necessary packages

import pandas as pd
import gensim
from autocorrect import spell
import string
from nltk.corpus import stopwords
import numpy as np
from sklearn import svm
from sklearn.metrics import accuracy_score
from sklearn.model_selection import ShuffleSplit
from sklearn.metrics import confusion_matrix
from sklearn.metrics import f1_score
import xlrd
import openpyxl

#Defining classification functions

def CombineFiles(files):
t = pd.DataFrame()
list = []
for file in files :
data = pd.read_excel(file)
list.append(data)
return pd.concat(list,ignore_index=True)

def listCreation(X):
X_n = []
for i in X:
X_n.append(i)
return X_n

def combineClass(class_codes,t):
df = t.copy()
for i in class_codes:
df.ix[df.CODE == i, 'CODE'] = 0
return df

def classifyingResult(cv,classifier,X,Y):
accuracies = []
for train_index, test_index in cv.split(X):
X_tr = [X[i - 1] for i in train_index]
X_tes = [X[i - 1] for i in test_index]
ROADMAP FOR CHANGING STUDENT ROADMAPS

160
y_tr = [Y[i - 1] for i in train_index]
y_tes = [Y[i - 1] for i in test_index]
clf = classifier.fit(X_tr, y_tr)
y_pred = clf.predict(X_tes)
a = accuracy_score(y_tes, y_pred)
c =confusion_matrix(y_tes, y_pred)
accuracies.append(a)
print a
print c
print "Mean Accuracy : " , np.mean(accuracies)

def classifyingResultWithSeprateTrainTest(X_tr,y_tr,X_tes):
classifier_SVM = svm.SVC(kernel='linear',decision_function_shape='ovr')
classifier = classifier_SVM
clf = classifier.fit(X_tr, y_tr)
y_pred = clf.predict(X_tes)
t_test['Class']=y_pred
t_test.to_excel('output_cycle.xlsx')

#Defining stopwords and importing word representations
#Google doc2vec file location can be downloaded from https://github.com/mmihaltz/word2vec-
GoogleNews-vectors

stop = set(stopwords.words('english'))
model = gensim.models.KeyedVectors.load_word2vec_format('GoogleNews-vectors-
negative300.bin', binary=True)

#This section first combines files with training data (if there are multiple files) then preprocess
them (including ensuring there is a single code -- the possible self code -- for each response), and
combines necessary classes (here, 3, 4, 5, and 7) into one class

files = [[Names of Files with training data in them]]
t = CombineFiles(files)
t = t[pd.notnull(t['CODE'])]
t.loc[t['CODE'] != t['CODE.1'], 'Strategy'] = ''
t["Combined"] = t["Possible Self"] +' '+ t["Strategy"]
t['Combined'].fillna('', inplace=True)
t['Combined'] = t['Combined'].apply(lambda x:x.encode('utf-8'))
t['Combined'] = t['Combined'].apply(lambda x:str(x))
t['Combined'] = t['Combined'].apply(lambda x:x.lower())
t['Combined'] = t['Combined'].apply(lambda x:x.translate(None,string.punctuation))
t['Combined'] = t['Combined'].apply(lambda x: [spell(str(item)) for item in x.split() if
spell(str(item)) not in stop and spell(str(item)) in model.vocab])
ROADMAP FOR CHANGING STUDENT ROADMAPS

161
t['Vecs'] = t['Combined'].apply(lambda s:reduce(lambda x, y: x + y, map(lambda e:
np.array(model[e]), s)) if len(s)!=0 else np.zeros(300, dtype='float32'))

original = t.copy()
class_codes= [3,4,5,7] #combining classes 3, 4, 5, 7 (or whichever classes you wish to combine
into one class)
t = combineClass(class_codes,original)
X = t['Vecs']
Y = t['CODE']
X = listCreation(X)
Y = listCreation(Y)

print confusion_matrix(Y,Y) #To get an idea of the data distribution

#Split the data based on parameters provided
n_splits = 10
test_size = 0.15

#10-fold cross validation SVM classification
cv = ShuffleSplit(n_splits=n_splits, test_size=test_size, random_state=0)
classifier_SVM = svm.SVC(kernel='linear',decision_function_shape='ovr')
classifier = classifier_SVM
classifyingResult(cv,classifier,X,Y)

#this section of the code should be run if a seperate test file needs to be coded. First, it trains on
the combined pre-processed files from the previous portion, then it preprocesses the test file, and
runs the classification

files_test = [Name of file with data to be coded]
t_test = CombineFiles(files_test)
t_test['Strategy'].fillna('', inplace=True)
t_test['Possible Self'].fillna('', inplace=True)
t_test["Combined"] = t_test["Possible Self"] +' '+ t_test["Strategy"]
t_test['Combined'].fillna('', inplace=True)
t_test['Combined'] = t_test['Combined'].apply(lambda x:x.encode('utf-8'))
t_test['Combined'] = t_test['Combined'].apply(lambda x:str(x))
t_test['Combined'] = t_test['Combined'].apply(lambda x:x.lower())
t_test['Combined'] = t_test['Combined'].apply(lambda x:x.translate(None,string.punctuation))
t_test['Combined'] = t_test['Combined'].apply(lambda x: [spell(str(item)) for item in x.split() if
spell(str(item)) not in stop and spell(str(item)) in model.vocab])
t_test['Vecs'] = t_test['Combined'].apply(lambda s:reduce(lambda x, y: x + y, map(lambda e:
np.array(model[e]), s)) if len(s)!=0 else np.zeros(300, dtype='float32'))
original_test = t_test.copy()

X_test = t_test['Vecs']
ROADMAP FOR CHANGING STUDENT ROADMAPS

162
X_test = listCreation(X_test)

classifyingResultWithSeprateTrainTest(X,Y,X_test)

ROADMAP FOR CHANGING STUDENT ROADMAPS

163
Chapter 3: Teachers can do it: Scalable identity-based motivation intervention in the
classroom

Abstract
Classroom activities aimed at changing students’ identity-based motivation (IBM) improve
student outcomes by helping students experience school as the path to their adult future identities
and their difficulties along the way as signals of the importance of schoolwork. One way to scale
these effects would be to have teachers deliver IBM activities. Hence, we asked if, after a brief
two-day training, teacher-delivered IBM intervention could meet fidelity standards and if
attaining more fidelity matters. We trained all eighth grade teachers in two middle schools
(N=211 students). We compared attained fidelity (dosage, adherence, quality of delivery, student
responsiveness, fidelity of receipt) to Durlak and DuPre’s (2008) empirically derived standard
for fidelity. We found that most classrooms (88%) and students (89%) received IBM intervention
at-or-above threshold standard, implying that teacher-based IBM delivery is viable. Moreover,
investing in improving fidelity is worthwhile; above-threshold fidelity improved core grade-
point-average and reduced risk of course failure.

ROADMAP FOR CHANGING STUDENT ROADMAPS

164
Introduction
“In the beginning of the year we did a program called Pathways-to-Success. It was a program
about how we thought of our futures, and if something got in the way, how would we make plans
to overcome them. Something [my teacher] always told the class of 2023 was that if it’s difficult,
it’s important. I feel like this is true. In life if you find something difficult like school for instance
it is important.” (8
th
grader Middle School Graduation Speech)”

“Thank you for helping us with what will happen later in life…For giving us a pathway to
success and now it is our choice to take that path. You helped us find forks that we may have, the
decisions we have to make.” (8
th
grader receiving special education services, letter to teacher
delivering the Pathways-to-Success program)

“This was by far and away the best advisory program we have had and I’ve been here 10 years.
We have had some attempts at it with very little support that have fallen flat on their face. This
may be our third or fourth advisory program.” (8
th
grade Science teacher who delivered the
Pathways-to-Success program)

Students want to do well in school and go on to college, yet they often fail to attain their
high aspirations (Oyserman & Destin, 2010; Oyserman & Lewis, 2017). One way teachers can
harness students’ high aspirations is to use identity-based motivation to help their students
imagine school as the path to their future, generate strategies to succeed on that path, and see
obstacles and failures along the way as signaling importance and value (Oyserman, Johnson, &
James, 2011; Oyserman et al., 2017). As our opening quotes suggest, both students and teachers
appreciate the usefulness of the identity-based motivation (IBM) perspective. Students found the
main points of the IBM intervention useful enough to include in graduation speeches and even
felt an impulse to write thank you notes to teachers. Indeed, student academic outcomes improve
when classroom interventions target identity-based motivation. Analyses of two identity-based
motivation interventions revealed significantly improved student academic outcomes at end of
school year follow-up (Oyserman, Terry, & Bybee, 2002) and at end of a two-school-year
follow-up (Oyserman, Bybee, & Terry, 2006). In these tests of identity-based motivation theory,
pairs of college students (Oyserman et al., 2002) or staff holding undergraduate degrees
ROADMAP FOR CHANGING STUDENT ROADMAPS

165
(Oyserman et al., 2006) delivered the intervention. These prior tests were important because they
provided support for the robustness of IBM theory by showing significant effects in real-world
settings on important academic outcomes (core course grades and risk failing a class) via change
in IBM variables. However, they did not test if, after a brief training, teachers can deliver an
IBM intervention with sufficient fidelity to have its promised effects. This test is needed if
scaling via teacher implementation is to be possible. We take two steps to address this issue in
the current paper.
At step one we test the prediction that a brief 2-day in-service training yields sufficient
fidelity to likely have effects. At step two we test the prediction that achieving higher fidelity
matters for core grade-point average and course failure rates. We focus on a brief 2-day training
because teachers are unlikely to be given time for longer training. We focus on sufficient fidelity
because a large review suggests that interventions delivered with less than 60% fidelity are
unlikely to have their intended effects (Durlak & DuPre, 2008). We focus on implications of
higher fidelity because the Durlak and DuPre (2008) review also suggests that practitioners are
unlikely to deliver with more than 80% fidelity. Taken together, this range implies that analyses
should focus on whether the fidelity threshold of 60% is attained and whether fidelity above 60%
and closer to the 80% practical maximum improves targeted outcomes. To situate our results and
their implications, we divide the introduction into three sections. First, we review identity-based
motivation theory, the evidence that it predicts academic outcomes, its translation to
intervention, and the need for testing teacher-led IBM intervention. Second, we describe what
fidelity is and how to operationalize it. Third, we specify our research questions.

ROADMAP FOR CHANGING STUDENT ROADMAPS

166
Identity-based Motivation Theory
Operationalization. Identity-based motivation theory is a social psychological theory of
motivation and goal pursuit that explains when and in which situations people’s identities
motivate them to take action towards their own goals (Oyserman, 2015a; Oyserman et al., 2017).
Identity-based motivation theory starts with the assumption that people are sensitively attuned to
their immediate context and that this shapes identities (dynamic construction). People prefer to
act (action-readiness) and make sense of situations (procedural-readiness) in identity-congruent
ways—ways consistent with what ‘I’ and people ‘like me’ do. However, even though identity
(who one was, is, and might become) feels stable, identities are dynamically constructed in
context. Dynamic construction means that contexts shape which identities come to mind, what
these identities seem to imply for behavior, and how people interpret experienced difficulty. The
thing of interest is not that people can change how they regard themselves after sustained effort,
but rather the surprisingly large effects that small shifts in context can have on changing how
people regard themselves. As detailed next, each component of identity-based motivation
(dynamic construction, action-readiness, and procedural-readiness) has been operationalized and
its effect on academic performance empirically tested.
Experimental evidence of effects on academic outcomes. In this section we briefly
review experiments documenting effects of identity-based motivation on academic outcomes.
First, we consider studies showing that dynamic construction of identity cues action-readiness;
readiness to act in ways that fit constructed identity. In one study, researchers subtly shifted what
context implied about being a boy (Elmore & Oyserman, 2012). In this study, researchers
randomly assigned middle school boys into groups; each group was shown a different graph of
accurate statewide census data. One group—the “men succeed” group—saw a graph showing
ROADMAP FOR CHANGING STUDENT ROADMAPS

167
that men earned more money than women. This graph implied that academic success fits with
being a boy. Boys who saw the “men succeed” graph made more attempts to solve a math task
and imagined more school-focused possible identities than boys who saw other graphs. Boys in
these other conditions saw graphs that did not mention gender or graphs showing that women are
more likely to have graduated high school than men, implying “women succeed”.
In a second set of studies also examining the consequences of dynamic construction of
identity on action-readiness, researchers subtly shifted what the future self seemed to imply for
action by changing the fit between identity and context (Oyserman, Destin, & Novin, 2015). In
these studies, researchers randomly assigned students to think about school as a success-likely
context in which most students succeed or to think about school a failure-likely context in which
most students do not do as well as they hoped. The researchers then asked students to write about
their possible identities, with half of the students guided to consider desired possible identities
and half of students guided to consider undesired ones. Thus, half of students were led to
consider their future self and their current context as fitting together, either because in that
context people often fail and their to-be-avoided future self was on their mind, or because in that
context people often succeed and their to-be-attained future self was on their mind. The results
showed that the action-readiness component of the future self is context-sensitive. That is,
students planned to start studying sooner if the way they thought about their possible identities
and the way they thought about school fit together. They were more likely to take action after
thinking about undesired possible identities while thinking of school as a failure-likely context or
after thinking about desired possible identities while thinking of school as a success-likely
context.
In other studies examining the link between dynamic construction of identity and action
ROADMAP FOR CHANGING STUDENT ROADMAPS

168
readiness, researchers used small contextual cues to make the future self be experienced as
relevant to the present moment (Destin, 2017; Destin & Oyserman, 2009; Landau, Oyserman,
Keefer, & Smith, 2014; Nurra & Oyserman, 2018). Nurra and Oyserman (2018) randomized
students to consider their adult future self as occurring soon or occurring later, as connected to
their current self or as distinct from their current self. Experiencing one’s adult self as near and
connected is consequential for behavior. Across studies, if researchers led students to experience
their adult and present selves as connected, students worked harder on current assignments,
focused more on boring tasks, and actually attained better core course grades by the end of the
semester.
Destin and colleagues (Destin, 2017; Destin & Oyserman, 2009) randomized middle
school students to either learn about need-based financial aid (open path) or to estimate the cost
of college and report how they planned to cover this cost (closed path). Students who learned
that income is not a barrier had significantly higher school engagement compared to students
who asked to consider the cost of college and how they would pay for it. In contrast, students
asked to consider cost and how they would pay for college seemed to infer that cost was a barrier
and hence college was not likely for them, making hard work in eighth grade feel like a pointless
endeavor. Landau and colleagues (2014) randomized students to either think about their
academic possible identities in the context of an image that implied action (a path) or one that
did not (a container). Students led to list their academic possible identities on an image of a path
rather than an image of a container were more engaged with their schoolwork. Across studies,
these students were more likely to seek out academic help, worked harder on current
assignments, planned to study more for an upcoming quiz, and actually performed better on the
quiz.
ROADMAP FOR CHANGING STUDENT ROADMAPS

169
In addition to cuing readiness to act, dynamic construction of identity also cues
procedural-readiness—that is, how one makes sense of experienced ease and difficulty with
schoolwork as implying something about oneself. Interpretation of experienced difficulty matters
for downstream behavior and for identity. Across studies, once students considered that
experienced difficulty might be a sign that schoolwork is important, they saw academics as more
central to their identity (Aelenei, Lewis, Oyserman, 2017; Oyserman, Elmore, Novin, Fisher, &
Smith, 2018; Smith & Oyserman, 2015) and did better on a variety of school tasks (Elmore,
Oyserman, Smith, & Novin, 2016; Oyserman et al., 2018; Smith & Oyserman, 2015). Students
are also more likely to endorse the idea that difficulty means importance if they experience fit
between identity and context, as we previously described. That is, when student think about their
undesired future self and think of school as a context in which failures are likely, then they are
more likely to endorse a difficulty-as-importance mindset. The reverse is also true—students are
more likely to endorse a difficulty-as-importance mindset when they think about their desired
future self and think of school as a context in which success is likely (Oyserman et al., 2015).
Translating experiments to intervention. Evidence to date shows that identity-based
motivation intervention can matter by changing student engagement, effort, and grades. The
School-to-Jobs intervention translated the three core components of identity-based motivation
(dynamic construction, action-readiness, and procedural-readiness) into a set of activities
(Oyserman et al., 2002; Oyserman et al., 2006). Randomized control trial evaluation showed that
the intervention changed elements of identity-based motivation by the end of eighth grade
(Oyserman et al., 2002) and these changes mediated changes in core course grades and course
failures by the end of 9th grade (Oyserman et al., 2006). Finding effects on core course grades
and course failures provide important metrics because core course grades and course failure by
ROADMAP FOR CHANGING STUDENT ROADMAPS

170
9th grade increase likelihood of on-time high school graduation (Allensworth & Easton, 2005),
and failing even a single course reduces likelihood of high school graduation (Allensworth &
Easton, 2007).
In developing and delivering the intervention, the intention was that each activity provide
students with a different concrete experience of one or more components of identity-based
motivation in a way that made it likely that students would internalize the core ideas. To do so,
activities allowed students to discover and experience each component of identity-based
motivation on their own, rather than being told about identity-based motivation by their
instructor, thereby reducing the chances of reactance (Brehm & Brehm, 2013; Elmore et al.,
2016) and increasing likelihood of deep processing of messages (Petty & Cacioppo, 1986).
Activities were group-based rather than occurring alone to increase chances that student social
identities—as students, boys or girls, or members of their racial-ethnic group—were cued as ‘we
do school’, increasing chances that an identity-based motivation cycle would ensue (Oyserman,
2007). When delivered and received as intended, participating students should experience change
in their identity-based motivation, and this should improve academic outcomes. Specifically,
students should have more school-focused possible identities and strategies to get there, be more
likely to see difficulty with schoolwork as implying its importance, and be less likely to see
difficulty with schoolwork as implying that schoolwork is ‘not for them.’ They should be more
likely to see school as the path to their adult future self, see school-focused possible identities as
congruent with important social identities, and be more likely to experience their adult future self
as relevant to their current schoolwork. Over time, these changes should result in better school
grades and less likelihood of failing classes. That is what researchers found (Oyserman et al.,
2006).
ROADMAP FOR CHANGING STUDENT ROADMAPS

171
Moving from trainers to teachers. Initial tests of the identity-based motivation
intervention showed that researchers could train undergraduates or people with undergraduate
degrees to go into schools and deliver the intervention as intended after 40 hours of training.
However, these tests did not use teachers. That provided a clean test of the theory, but for
practical application teachers are needed so that IBM intervention can be rooted in a school
system. Outside trainers may come and go, but teachers can maintain an intervention over time.
Teachers’ time is limited and hence a test in which a brief training is provided to teachers and
fidelity is assessed is a first step in addressing whether IBM interventions might scale through
teacher-delivery. We chose a 2-day test as the briefest likely sufficient training for teachers to be
able deliver and their students receive an identity-based motivation intervention as intended. This
combination of teacher delivery as intended and student receipt as intended is termed
implementation fidelity.
Ascertaining implementation fidelity is critical for both theory-testing and pragmatic
reasons, as detailed next. From a theory-testing perspective, without knowledge of
implementation fidelity it is not possible to know whether any changes after intervention are due
to the theory on which the intervention is based. At the same time, pragmatically, as we outline
next, without implementation fidelity, interventions are empirically less likely to have their
intended longer-term effects (Century, Rudnick & Freeman, 2010; Durlak & DuPre, 2008).
Program differentiation is the aspect of implementation fidelity related to theory testing.
It is an assessment of whether the ingredients of an intervention are operationalizations of the
theoretical process model or theory of change the intervention is based on. If the program uses
ingredients that are not part of the theoretical process model or the same ingredients (perhaps
differently labeled) also are in other programs (or the control group), program differentiation is
ROADMAP FOR CHANGING STUDENT ROADMAPS

172
low. There is not much point in delivering multiple interventions with different names that
deliver the same intervention ingredients or in delivering an intervention with ingredients not
linked to an empirically validated process model of change. Program differentiation assessment
might be obtained from a school district, principal, teacher, or researcher, and is at the level of
the intervention itself—the program either is differentiated or is not differentiated from other
programs. In the case of identity-based motivation intervention, the question would be whether
other programs in the school use IBM theory or IBM ingredients otherwise labeled. Our
principals and teachers concluded that they were not already delivering an identity-based
motivation intervention or even another socio-emotional learning (SEL) program.
Implementation Fidelity Entails Fidelity of Delivery and of Receipt
Aside from program differentiation, the other components of implementation fidelity are
dosage, adherence, quality of delivery, student responsiveness and fidelity-of-receipt (Dane &
Schneider 1998; see also Bellg et al., 2004; Dusenbury et al., 2003; Crosse et al., 2011;
Mowbray, Holter, Teague, & Bybee, 2003; O’Donnell, 2008; Resnick et al., 2005). To be useful,
implementation fidelity operationalization (O’Donnell, 2008) and report (Hulleman & Cordray,
2009) should fit the intervention itself. For example, in the case of classroom-level delivery, it is
reasonable to expect classroom-level and student-level variation—some classrooms and some
students will experience more implementation fidelity (experience faithful delivery and message
uptake) than others. Rather than being independent, the components of fidelity are best
understood as interdependent building blocks that scaffold and support each other.
At the base of fidelity are dosage, the timing and number of sessions delivered compared
to plan, and adherence, the extent that each activity in each session is delivered in the sequence
and as the manual describes it. Dosage and adherence scaffold quality of delivery and student
ROADMAP FOR CHANGING STUDENT ROADMAPS

173
responsiveness. Quality of delivery entails teacher-managed session ‘feel,’ which teachers
produce via classroom and student-level emotional and organizational support and behavior
management and via their structuring of delivery of take-home points. High quality delivery
entails students experiencing take-home points as emerging from themselves rather than from
teachers and as easy to process and hence true, rather than as emerging from teachers, difficult to
process, and hence not necessarily true. Student responsiveness entails student response to
adherence and quality of delivery. When teachers deliver (dosage) the correct content
(adherence) in the correct way (quality of delivery), their students should respond with engaged
attention and productivity (student responsiveness) and hence internalize the take-home points
(fidelity of receipt). Given the interrelatedness of each element conceptually, teasing apart these
elements of fidelity analytically would require randomizing teachers to deliver varying doses or
to adhere in varying levels or to separately deliver take-home points with varying quality.
Researchers can use classroom observation to obtain classroom-level ratings of how
much of intended intervention intensity and duration was delivered (dosage), how much delivery
followed protocol (adherence), and how much participants responded as intended (student
responsiveness). Researchers can use classroom observation and student reports to obtain quality
of delivery ratings. Researchers can use student report to assess the extent that participants have
received and understood take-home points (fidelity of receipt).
Reasonable expectations for fidelity. Having defined fidelity, the next questions are
how much fidelity can be expected, and how much is sufficient for an intervention to have its
desired effects. To address these questions, Durlak and DuPre (2008) reviewed the meta-analytic
literature on interventions delivered in real world settings by non-researchers, adding 59
additional studies they found that were not included in the prior meta-analyses. They asked
ROADMAP FOR CHANGING STUDENT ROADMAPS

174
whether this literature pointed to a threshold at which, on average, an intervention yields its
desired effects and whether this literature provided guidance into how much fidelity non-
researchers could be expected to produce. They concluded that studies rarely show intended
effects unless fidelity reaches or surpasses a 60% threshold and that non-researchers rarely
implement with greater than 80% fidelity. These findings have a number of implications. First,
fidelity researchers should expect that a successful training would yield fidelity between 60%
and 80%. Second, researchers should test whether moving from 60% to 80% fidelity increases
likelihood of attaining intended impacts, and if so, if the increase is linear or looks more like a
step function, and if so, where the step-up occurs.
The 60% fidelity threshold in educational research. We examined school and
education-focused evaluation research published since the Durlak and DuPre (2008) threshold
and practical maximum estimates were published. We found that Durlak and DuPre’s (2008)
60% fidelity threshold is used across an array of community-based and school-based intervention
evaluations to document that sufficient fidelity is attained. For example, Fagan, Hanson,
Hawkins, and Arthur, (2009) used this threshold in a 12-community evaluation of Communities
that Care, a system for community partners to utilize prevention science. Riordan, Lacireno-
Paquet, Shakman, Bocala, and Chang (2015) used the 60% threshold as a marker of necessary
fidelity in examining the implementation of a teacher evaluation system in 15 schools.
Bloomquist and colleagues (2013) used the 60% threshold as an indicator of sufficient fidelity in
the implementation of an intervention that aimed to reduce conduct problems in 27 elementary
schools. Lindsay, Davis, Stephan, and Proger (2017) used this threshold in evaluating a college
readiness program in 25 schools. At the pre-school level, the 60% threshold is used in testing
implementation of the preschool-based Early Literacy and Learning Model in 28 preschool
ROADMAP FOR CHANGING STUDENT ROADMAPS

175
classrooms (Preschool Curriculum Evaluation Research Consortium, 2008). The 60% threshold
is also used by evaluators of Life Skills Training, a widely-used school-based substance use
prevention intervention evaluated in over 30 peer-reviewed studies of programs in over 300
schools involving 20,000 students (e.g. Velasco, Griffin, Botvin, Celata, & Lombardia, 2017;
Botvin & Griffin, 2015; Botvin, Baker, Dusenbury, Botvin, Diaz, 1995).
While Durlak and DuPre (2008) describe a linear relationship between fidelity and
outcomes, they did not separately examine whether outcomes improve as program fidelity shifts
from the 60% threshold to the 80% practical maximum. We did not find other papers addressing
this question either so it is not clear if investing in fidelity beyond the 60% threshold matters, and
if so, if improved outcomes are linearly associated with improved fidelity.
Research Questions
Our review of the fidelity literature led us to two research questions. First, can teachers
deliver and students receive an identity-based motivation intervention at or above the 60%
fidelity threshold? Second, does moving beyond the 60% fidelity threshold matter for student
core grade-point average and likelihood of course failure? We used the mean of the five
components of implementation fidelity (dosage, adherence, student responsiveness, quality of
delivery, fidelity of receipt) that could vary at the classroom and student levels to test our first
research question, and the relationship between implementation fidelity and school grades and
course failure rates to test our second research question. To further understand the nature of our
fidelity effects, we also compared one aspect of teacher quality—teacher-driven classroom
climate—in the teacher’s subject classroom (while teaching math, science, history, or language
arts) with the teacher’s classroom climate score while delivering the identity-based motivation
intervention. This comparison allowed us to begin to address whether the quality aspect of
ROADMAP FOR CHANGING STUDENT ROADMAPS

176
fidelity was capturing what the teacher did in the identity-based motivation intervention or
something more general about the teacher. Finally, because our goal was to learn how to improve
fidelity, we examined teacher responses to our queries about ways to improve usability and
feasibility and looked carefully at the classroom experience to build and revise for future
implementations.
Materials and Methods
Sample
In the first year of our grant, all eighth-grade teachers (eight classrooms) in two Chicago
K-8 public schools and their full cohort of eighth graders (N = 211, 50% female, 93% nonwhite,
94% free or reduced price lunch eligible) participated in the intervention
11
. Classroom size
ranged from 25 to 31 students except for one pullout classroom, with 12 students receiving
special education services
12
. Three teachers were female, seven were white; their main subjects
were Math (two teachers), Science (two teacher), English (two teachers), Special Education (one
teacher), and History (one teacher). All teachers had three or more years of teaching experience,
making them similar to the statewide average—88% of teachers in Illinois have three or more
years of experience. The schools themselves were at or below state average in terms of their
standardized test scores for 8th grade. Statewide 40% of students scored in the range labeled
“met or exceeded expectations” in English; in our schools the percentages were 37% and 19%.
Statewide 32% of students met or exceeded expectations in Math; in our schools the percentages
were 35% and 26%.

11
Students were from a variety of racial-ethnic backgrounds: 68% were of Hispanic background,
16% were of Asian background, and 9% were of African American background. The remaining
students were categorized as having White (7%) or multiracial-multi ethnic backgrounds (1%).
12
We present results that exclude this classroom in our Supplemental Materials. Inclusion or
exclusion of these data does not change our results.
ROADMAP FOR CHANGING STUDENT ROADMAPS

177
Analyses describing fidelity (mean of the five fidelity components, n = 184) employ
listwise deletion for students missing student-level data on fidelity. Student-level data were
missing if parents refused to consent to data collection (n = 4) or if the data were missing even
though parents had consented—presumably because the student was absent the day of data
collection (n = 23). Analyses describing how fidelity affects subsequent course grades (n = 209)
employ listwise deletion for missing data on 8th grade academic outcomes. Only 1% of 8
th
grade
academic data are missing as a result of students leaving the district, an additional five students
were missing 7
th
grade academic data, 19 were missing student-level fidelity data, and 7 students
were missing both 7
th
grade academic data and student-level fidelity data. To preserve the
analytic sample in our regression models, we imputed missing data and used a dummy variable
to adjust for missingness in covariates. Specifically, we imputed 7
th
grade academic data by
assigning students the average score in their classroom and imputed missing student-level
fidelity data by assigning students their classroom average.
Procedure
We maintained continuity in training; Oyserman, who led the School-to-Jobs training,
also led training and weekly check-in calls (Oyserman et al., 2006). The implementation manual
and training highlighted how to deliver with high quality. This included seven components: (1)
scaffolding activities to both be personal and generate a sense of group norms of engagement, (2)
keeping a good pace, (3) creating a positive emotional climate, (4) being well organized, (5)
delivering the content in a way that appropriately evokes participation, (6) delivering the content
clearly in a way that facilitates student experience of each session as naturally unfolding and
building on prior lessons, and (7) delivering the content clearly in a way that facilitates student
experience of take-home points as student-generated, not teacher taught. Delivered with quality,
ROADMAP FOR CHANGING STUDENT ROADMAPS

178
students should experience ease in the concrete activities; feel that they, their classmates, and
teacher are trustworthy, warm and knowledgeable; and that together they generate useful
knowledge.
We made a number of decisions based on our goal of enhancing scalability. First, we
made the School-to-Jobs intervention implementation manual applicable to teachers—instead of
referring to two trainers, all instructions referred to a single teacher—and one activity that the
trainers took two sessions to deliver was consolidated into a single session. Second, we used a 2-
day (including breaks for breakfast and lunch) abbreviated form of the 5-day training Oyserman
provided to trainers in the School-to-Jobs intervention. Training took place in a classroom in
each school on two consecutive days in August prior to the September start of the school year.
Third, we allowed schools some variability in pace of delivery—School-to-Jobs was delivered
twice weekly but we allowed each school to choose if they would deliver twice per week or once
per week. Fourth, we allowed schools some variability in time of day and where the weekly
check-in during implementation would occur—either in school with all teachers physically
present, or outside of school with teachers calling in. Finally, we followed the original model,
which suggested that the intervention be named something that felt meaningful in context (e.g.
Oyserman, 2015b). In consultation with participating teachers and schools, we named the
teacher-led intervention Pathways-to-Success, or Pathways for short.
Teachers implemented the intervention during a designated advisory period during the
school day, insuring that each student was assigned a single teacher. One school chose bi-weekly
delivery for six weeks. In this school, teachers gathered together after school for a video call and
finished delivery by Halloween (the end of October). Students receiving special education
services participated in the intervention in a separate pull out classroom. The other school chose
ROADMAP FOR CHANGING STUDENT ROADMAPS

179
weekly delivery for twelve weeks. In this school, teachers called into the weekly call in the
evening from their own homes and finished delivery prior to Thanksgiving (end of November).
Students receiving special education services participated in their regular classrooms.
We maintained continuity in measurement of fidelity: we did not change fidelity
materials from the original School-to-Jobs with one exception. In School-to-Jobs, we used an
intervention-specific measure of trainer quality of delivery coded by observers in addition to
student-level report, as detailed in Oyserman (2015b). As it turns out, this measure was similar to
a widely used standardized measure of teacher instructional quality measure, the Classroom
Assessment Scoring System-Secondary (CLASS-S; Pianta, Hamre, Hayes, Mintz, & LaParo,
2008; Allen, Gregory, Mikami, Lun, Hamre, & Pianta, 2011). Using the CLASS-S requires a
two-day training for initial certification in coding and an annual recertification test in its use—
both of our coders were compliant with these requirements. Given our goal of communicating
with schools, we replaced our prior observer-based measure of quality of delivery with the
CLASS-S since we assumed schools would respond more positively to information based in part
on a familiar metric.
To obtain high quality data to assess fidelity we video-recorded each intervention session
of each teacher. Pragmatically, this meant that immediately preceding each session an American
Institutes of Research (AIR) staff member came into the classroom and positioned and turned on
an iPad on a tripod. At the end of each session the staff member came to collect the equipment;
video was then loaded onto the AIR secure server for coding. We also obtained student reports at
the end of the intervention using an online questionnaire. Chicago Public School District
provided school grades for all students as part of a master data sharing agreement with AIR. We
computed fidelity as the mean of dosage, adherence, student responsiveness, quality of delivery,
ROADMAP FOR CHANGING STUDENT ROADMAPS

180
and fidelity of receipt to test whether teachers could deliver at or above the 60% threshold and
whether moving beyond threshold mattered for student core grade point average and likelihood
of course failure.
Immediately following the final session of the Pathways intervention, a member of the
AIR team interviewed each teacher asking a range of questions pertaining to usability and
feasibility of delivering the intervention (e.g. “How did the resources provided by the Pathways-
to-Success program help you implement the program given your other responsibilities and time
commitments?”; the full set of questions is located in Supplemental Materials, Section 1). The
goal of this interview was to help us learn what obstacles teachers encountered during
implementation regarding preparing for and delivering each session of the intervention. We then
used their feedback to guide our plans for improvements aimed toward increasing scalability, as
described in our discussion of practical implications in section 4.3.
The week after the final session of the Pathways intervention, students completed a brief
end-of-intervention survey focused on fidelity of receipt and their perceptions of aspects of
quality of delivery. This survey included a brief set of parallel questions about an element of
teaching quality (teacher-driven classroom climate) for the student’s math, science, English, and
history teachers to allow for analysis of whether the training itself or aspects of teachers
generally influenced quality of delivery in Pathways (the full student end-of-intervention
questionnaire is located in Supplemental Materials, Tables S2.13, S2.14, and S2.16).
The Intervention
The full intervention manual that the teachers in this study used is published (Oyserman,
2015b). As an overview, Table 1 provides each session’s thumbnail sketch, take-home point, and
core identity-based motivation active ingredient.
ROADMAP FOR CHANGING STUDENT ROADMAPS

181

Table 1
Thumbnail sketch of each Pathways session, take-home point and activated IBM ingredients

Session Classroom activity flow Take home point Activated
IBM
constructs
1. Setting the Stage &
Introductions
Students are paired up and briefly interview one
another on the skills or ability they each have that
will help them complete the school year
successfully (e.g., “well organized,” “positive
attitude”). Then each student introduces his or her
interview partner in terms of these skills.
We all care about school, and we have
a skill or ability to work on our
“successful in school” possible self.
DC, AR
2. Adult Images

Students pick photographs that fit their adult
“images.” Images of what their adulthood will be
like. Photographs include the four domains of
adulthood material (e.g., homes), job (e.g., working
at various jobs), relationships (e.g., family, friends),
and community engagement (e.g., volunteering,
voting). Photographs include both genders and
match the racial-ethnic makeup of the school.
Domains of adulthood emerges from clustering
student responses and having students name these
clusters.
We all have images of ourselves as
adults in the far future.
DC
3. Positive and
Negative Forces
Students draw or write about positive and negative
forces—people or things that energize them to work
toward their possible identities by showing what to
do or what not to do.
Everyone faces obstacles and
difficulties; positive and negative
forces help by laying out paths to take
or avoid and ways to handle obstacles
and examples of what not to do.
DC, PR
ROADMAP FOR CHANGING STUDENT ROADMAPS

182
4. Timelines Students draw timelines into the future, including
forks in the road and obstacles. Since students start
with the present, all timelines involve school.
Present and future are linked on a path.
There are choices and obstacles.
Current actions set up which futures
are possible. Obstacles must be gotten
around to get back on path.
DC, PR
5. Action Goals Students write action goals, linking next year and
adult possible selves with actions they can take right
away in a specific time and place to concretize the
plan. They do this using an easy to recall formula
(because… I will… when…).
We have some control over possible
selves, but not our hopes and dreams -
control happens when we link the
future with the present through
specific action paths.
DC, AR
6 Possible selves and
strategies
Students map out their expected and to-be-avoided
possible selves and strategies for next year on a
pathways board.
Strategies are actions you are taking
now or could take to become your next
year possible self.
DC, AR
7. Pathways to the
Future
Students complete pathways boards to concretize
the link between current strategies for action, next
year possible selves, and adult possible selves.
Strategies I’m doing (or could be
doing) now to get to my next year
possible self also help me get to my
adult possible self.
DC, AR
8. Puzzles

Students break down problems that seem impossible
and use strategies to solve them.

Difficult things can seem impossible,
not worth your time; but difficulty can
be a signal of importance. Use a
strategy like breaking down to parts.
PR
9. Solving Everyday
Problems

Students write about a school problem they have.
They use the IBM skills they have learned to
consider how to break the problem down.
Everyday problems can be broken
down using the skills you have to
consider what is the adult possible self,
what is the next year possible self,
what is the positive and negative force,
the choice point or obstacle and what
are strategies to get around it.
PR
ROADMAP FOR CHANGING STUDENT ROADMAPS

183
10. Everyday
problems: High
school and beyond
Students brainstorm what is needed to finish high
school, see the requirements, brainstorm what is
needed to get to college, see the requirements.
You can identify the steps to get from
8
th
grade to graduating high school.

DC, AR,
PR
11. Wrapping up and
Moving Forward
Students name the Pathways sessions, what each
was about, what they liked and what they would
improve, providing a bird’s eye view, closure, and
reinforcement of the three main IBM ingredients.
What I do now matters for attaining
my next year and adult possible selves.
Possible selves that are linked to
strategies and to time and place of
action become action goals. There are
forks (choices) and roadblocks
(failures) along the way. It will be
difficult and may feel impossible, but
asking questions helps break down
what I need to find out and helps me
connect to others – positive forces and
models – as well as to learn from
negative forces and models of what
not to do.
DC, AR,
PR

Note: DC=Dynamic construction, AR=Action-readiness, PR=Procedural-readiness
ROADMAP FOR CHANGING STUDENT ROADMAPS

184
Consent
The school district approved our human subjects’ protocol. We included in our fidelity
analyses only students with parental consent for survey collection; almost all (98%) provided it.
To reduce teacher burden and ensure that paperwork was complete, an AIR staff member handed
out and collected parental consent forms and each student was given two movie tickets after
returning a form, regardless of whether parents provided or withheld consent. All teachers signed
consent forms for video recording. Prior to coding, faces of students without parental consent for
video recording were blurred out.
Fidelity
We assessed dosage (see Supplemental Materials, Tables S2.1-S2.11), adherence (see
Supplemental Materials, Tables S2.1-S2.11), and student responsiveness (see Supplemental
Materials, Tables S2.1-S2.11; and Supplemental Materials Table S2.12
13
), from video records of
sessions. We assessed quality of delivery from session video records and from end-of-
intervention student report (see Figure 1, Supplemental Materials Tables S2.12-S2.14) obtained
the week after the intervention ended. We assessed fidelity of receipt with end-of-intervention
student report (see Supplemental Materials, Table S2.15). As detailed below, our coders coded
video of each session for each teacher using a structured protocol and we scaled student-report
data.

13
Table S1.14 provides the CLASS-S General Scoring Rubric. CLASS-S is a proprietary
instrument; hence we cannot include the full manual. www.teachstone.com provides more
information about the CLASS instruments.
ROADMAP FOR CHANGING STUDENT ROADMAPS

185
Did the pace, repetition and clarity together converge to create a fluent experience (must be true)?

Message feels untrue. Sometimes feels true, sometimes
feels untrue.
Must be true
Figure 1. Quality of Delivery Rating-Scale for Take Home Point Fluency

Reliability of video-based coding. The third author coded all elements of fidelity
obtained from video records using the structured protocols described next. We randomly selected
nine tapes for double coding the full protocol by another AIR staff member to assess coding
reliability using our structured protocols. As noted above, both coders were certified CLASS
coders. We randomly selected thirteen tapes for double coding of the CLASS-S. We used two
metrics for reliability: percent agreement and Cohen’s Kappa (Fleiss & Cohen, 1973). Cohen’s
Kappa is useful given that coding is categorical. For ease of comparison, we report average
reliability when we coded multiple measures for a given fidelity component. To provide some
rule of thumb for Cohen’s Kappa, Landis and Koch (1977) suggest that scores in the .4 to .6
range represent moderate agreement, while scores in the .61 to .8 range represent substantial
agreement. Taken together, coder agreement is sufficient: Dosage (85% agreement, Cohen’s
Kappa = .59), adherence (78% agreement, Cohen’s Kappa = .55), student responsiveness (88%
agreement, Cohen’s Kappa = .75), quality of delivery (75% agreement, Cohen’s Kappa = .60).
Computing dosage fidelity. Dosage (α = .8414) is a mean score across the 11 sessions.
In each session, it had two components: implementation and task percentage. Implementation
was the percentage of sessions that teachers implemented (counted from the video). Because all
teachers implemented each session, each teacher scored 100% on implementation. We calculated

14
Alpha reliability used only on the task percentage component because the implementation
component was invariant across teachers (all implemented all sessions) and hence cannot be used
to calculate alpha.
ROADMAP FOR CHANGING STUDENT ROADMAPS

186
Task Percentage by dividing the number of tasks the teacher actually facilitated by the number
that were to be facilitated in each session and multiplying by 100. To obtain this number, the
observer watched the video of each session and marked with a check if the task occurred or not.
We used a checklist to measure these tasks (Supplemental Materials, Tables S2.1-S2.11;
Oyserman, 2015b). For ease, Table 2 shows the checklist for Session 2 of the intervention; the
first column is what was counted for task percentage. The number of tasks in each session varied
from 9 to 20.
ROADMAP FOR CHANGING STUDENT ROADMAPS

187
Table 2:
Checklist for Coding Dosage, Adherence, and Student Responsiveness in Session 2 (Adult Images)
Task Y N Detailed Teacher Behavior Y N Student Behavior Y N
Agenda hung
Opening
Welcome Greet participants and latecomers Greet trainer
Say today is session 2, adult
images

Last session Ask for what happened last session
and why
Share ideas (Learned names
about each other,
expectations, concerns,
games as a team, adding and
building on each others’
skills)

Reinforce student participation
(Why: people have lots of different
skills that will help them succeed)

Bridging Teacher bridges last session and
this session (last session we
focused on skills and abilities to
succeed in school, today we want
to look towards the future and the
adults we want to become)
Listen
Images
Introduce the
concept of adult
images
Explain task – choosing pictures
that represent images of yourself
as an adult. Each to pick 3 to 5
pictures, what do they mean for
you and when these will be true of
you, afterwards share
Listen

Create personal
images
Make instructions clear/Ask for
questions
Ask questions/Clarifies
directions

Have participants begin

Move around room, picking
pictures

Mingle – check for understanding
Share Have everyone rejoin circle
ROADMAP FOR CHANGING STUDENT ROADMAPS

188
Explain task – show 1 picture and
explain to group, while group
listens and pays attention
Participate
Write participant responses on
newsprint, clustering by themes
Listen
Domains of
adulthood

Highlight various
domains
Explain task – participant to call
out what they thought was similar
about everyone’s adult images
Share ideas

Reinforce personal
competence in
noticing
connections, ability
to contribute to the
in group
Highlight themes that emerge
(e.g., jobs, family, friends,
community involvement, life style;
trainer need only mention domains
that did emerge)

Listen
Next session and
goodbyes

Summary Statement: adult images
can be about jobs, family, friends,
community involvement, and
lifestyle (only those group brought
up or implied) (adult images +
repeat themes)
Listen

Connecting statement: next session
we’ll identify models and forces
that help us work on those adult
images that are goals

Completed
necessary
components of
session in
appropriate time

ROADMAP FOR CHANGING STUDENT ROADMAPS

189
Computing adherence fidelity. Adherence (α = .95) is a mean score across the 11
sessions. In each session, it was the count of the number of specific teacher actions the teacher
actually took in the session divided by the number that were to be taken in each session and
multiplied by 100. We used a checklist to measure these actions (Supplemental Materials, Tables
S2.1-S2.11; Oyserman, 2015b). The observer watched the video of each session and marked with
a check if the action occurred or not. For ease, Table 2 shows the checklist for Session 2 of the
intervention; the second column is what was counted for adherence percentage. The number of
actions in each session varied from 15 to 3.
Computing quality of delivery fidelity. Quality of Delivery (α = .74) is a mean score
across components. In each session, we computed from two sources and each had multiple
components: observer coding of each session’s videotape and end-of-intervention student report
(Figure 1; Supplemental Materials, Tables S2.12-S2.14 respectively). We calculated a quality of
delivery score for each student in two steps. At step one we obtained a percentage of total
possible points in each metric. At step two we obtained a mean percentage across metrics.
For observer-based elements of quality, the observer, certified in the CLASS-S, watched
each session video and rated 11 dimensions of quality twice per session. The 11 dimensions were
organized in three domains: Emotional Support, Organizational Support, and Instructional
Support. Emotional Support includes three dimensions (Positive Climate, Teacher Sensitivity,
and Regard for Adolescent Perspectives). Organizational Support includes three dimensions
(Negative Climate [reverse coded], Productivity, and Behavior Management). Instructional
Support includes five dimensions (Instructional Learning Formats, Content Understanding,
Analysis and Problem Solving, Feedback, and Classroom Discussions). To code, the observer
stopped the video at approximately the 20-minute mark and the 40-minute mark (or end) of the
ROADMAP FOR CHANGING STUDENT ROADMAPS

190
session. The two scores on each dimension were averaged to a lesson score for each dimension;
lesson scores were averaged to obtain a final score for each teacher. Coding was on a 7-point
scale from 1 = not all characteristic to 7 = highly characteristic (negative items reverse-coded).
Dimensions were coded by observing classroom interactions. Each dimension has a unique
scoring rubric (Pianta et al., 2008) and specific behavioral indicators associated with low (1-2)
mid (3-5), and high (6-7) range scores.
For example, the Productivity dimension’s behavioral indicators include maximizing
learning time, routines, transitions, and preparation. In the low range, teachers are providing few
tasks for students to complete, the class is disorganized and the students do not appear what to
do, the students spend a significant time in transitions, and the teacher is not prepared for the
session. In the mid-range, the teacher provides tasks for students to complete the majority of the
time, but the learning is sometimes disrupted or there are inefficiencies in managerial tasks.
There are times of uncertainty and there are some inefficiencies in transitions but mostly there
are classroom routines. The teacher is mostly prepared, but has some last minute preparations. In
the high-range, the students have tasks to complete, are comfortable with the routines, transitions
are efficient, and the teacher is prepared to deliver the lesson. Interested readers can find the
general CLASS-S manual scoring rubric in Table S1.12 in the Supplemental Materials.
At the end of each session, observers read the session take-home point, and coded how
Fully (Table 3) and how Fluently (Figure 1) each session’s take home point was conveyed. For
the Fully component, take-home scores ranged from 0 to 2—the take-home point was: 0 = not
evoked at all by activities, 1 = partially evoked but with unclear or inconsistent framing, 2 =
clearly and consistently evoked with concepts connected to student-generated examples. We
coded Fluency as 0 (thumbs down) if the session pace, repetition and clarity of delivery together
ROADMAP FOR CHANGING STUDENT ROADMAPS

191
converged to create a disfluent experience in which the take-home point did not ring true. We
coded Fluency as 1 (sideways thumb) if pace, repetition and clarity converged to create some
points in which the take-home rang true and some points in which it rang false. We coded
Fluency as 2 (thumbs up) if session pace, repetition, and clarity converged to create a sense that
the take-home point must be true.
Table 3 Take-Home Point by Session

Session Take Home Point
1: Skills and abilities You have some skills and abilities to help you succeed in the coming
year and others do too.
2: Adult Images We all have images of ourselves as adults in the far future.
3: Positive and
Negative Forces
Positive and negative forces make some adult images possible selves.
Positive forces help you lay out paths for success and handling
difficulties and setbacks. Negative forces do the opposite. They lay out
paths for failure and examples for how not to do handle difficulties and
setbacks.
4: Timelines The future is a path, current actions set up which futures are possible
5: Action Paths We have some control over possible selves, but not hope and dreams.
That control happens when we link the future with the present through
specific action goals.
6: Possible Selves
and Strategies
Strategies are actions you are taking now or could take to become your
next year possible self.
7: Pathways to the
Future
Any strategies I’m doing (or could be doing) now to get to my next
year possible self, also help me get to my adult possible self.
8: Puzzles Things can seem impossible and difficult, but can be solved by
breaking them down looking for alternative ways to set up the problem.
9: Solving Everyday
Problems
There are everyday choice points and difficulties that are obstacles to
navigate on the path linking near and far possible selves.
10: Solving
Everyday Problems
II Graduation
You can identify the steps to get from 8
th
grade to graduating high
school.
11: Wrapping Up
and Looking
Forward
What I do now makes a big difference for attaining my possible selves
for next year, for the next few years, and farther as an adult. Possible
selves that are linked to strategies and to time and place of action
become action goals. There are forks (choices) and roadblocks
(failures) along the way. It will be difficult and may feel impossible,
but asking questions helps break down what I need to find out and
helps me connect to others – positive forces and models – as well as to
learn from negative forces and models of what not to do.
ROADMAP FOR CHANGING STUDENT ROADMAPS

192
At the end-of-intervention survey, students provided their quality ratings on four scales
that assess classroom quality on dimensions compatible with the CLASS-S. Students rated their
teacher’s sensitivity (1 = not at all, 5 = a lot; α = .75): how often the teacher understood their
problems, listened to their comments, negatively criticized their ideas (reverse-coded), used
specific examples, gave everyone an equal chance to participate, and gave students the chance to
answer one another’s questions. Students rated two aspects of classroom positive climate
(teacher-driven and classmate-driven). Students rated the teacher-driven classroom climate (1 =
strongly disagree, 5 = strongly agree; α = .84): how enthusiastic, warm, clear, and
knowledgeable their teacher was. Students rated their classmate-driven classroom climate (1 =
strongly disagree, 5 = strongly agree; α = .85): how enthusiastic, warm, clear, and
knowledgeable their classmates were. Finally, students rated classroom regard for adolescent
perspectives (1 = strongly disagree, 5 = strongly agree; α = .67): how often they felt comfortable
asking questions, they could trust others to listen to what they had to say, others shared their
experiences and difficulties working toward their futures, it seemed that other students had the
same problems they did, what they talked about was relevant to them, and they felt concerned
they would be negatively criticized by another group member (reverse-coded). The exact
wording of each item is Table S2.13 in Supplemental Materials.
Computing student responsiveness fidelity. Student responsiveness (α = .82) is a mean
score across the 11 sessions of the two components scored in each session: Student behavior, as
measured by the checklist, and the Student Engagement dimension of the CLASS-S. Observers
watched the video of each session and as the session unfolded, they marked with a check if an
expected student response occurred or not following the original School-to-Jobs checklist of
student responses in each session (Supplemental Materials Table S2.1-S2.11; Oyserman, 2015b).
ROADMAP FOR CHANGING STUDENT ROADMAPS

193
For ease, Table 2 shows the checklist for Session 2 of the intervention, the third column is
student responsiveness. The number of responses expected varied by session from a low of 9 to a
high of 24. We translated counts to percentage scores for each session. In addition, observers
also rated the Student Engagement dimension of the CLASS-S (see Quality of Delivery section
for a description of the properties of the CLASS-S). Here, observers rated the degree that
students were focused and attentive in the classroom and actively participating in each learning
activity. In the low-range of this code, the majority of the students are disengaged from the class
or distracted from learning. In the mid-range of this code, students appear to be passively
engaged in the learning, not actively participating in the classroom; or there is a mix of student
engagement in which some are engaged and others are not. In the high-range, the majority of the
students are actively participating in the lesson.
Computing fidelity of receipt. Fidelity of receipt (α = .87) was obtained from the end-
of-intervention student survey. We operationalized fidelity of receipt as student-reported
confidence (1 = not at all confident, 5 = very confident) that they could engage in or demonstrate
the skills highlighted in each session, and how much they agreed or disagreed (1 = strongly
disagree, 5 = strongly agree) with the identity-based motivation messages regarding
interpretation of difficulty and strategy development. The full set of items is provided in Table
S2.15 in Supplemental Materials.
Computing student-level and classroom-level fidelity scores. Classroom-level (α =
.85) fidelity is the mean of the five components of fidelity dosage, adherence, student
receptiveness, quality of delivery, and fidelity receipt, with data coming from individual students
for quality of delivery and fidelity of receipt averaged at the classroom-level. Student-level
fidelity (α = .73) is the mean of the five components of fidelity dosage, adherence, student
ROADMAP FOR CHANGING STUDENT ROADMAPS

194
receptiveness, quality of delivery, and fidelity receipt with data from individual students for
quality of delivery and fidelity of receipt maintained at the student-level.
Computing Core GPA and Course Failure
Chicago Public Schools provided student 7th and 8th grade grades for each marking
period (n = 197 students had full records, 12 students were missing 7th grade grades and an
additional 2 students were missing 8th grade grades, likely due to out-of-district moves). We
computed 7th grade core GPA as an average of final grades for core classes (Math, Science,
English, History, and Social Studies) in 7th grade, and 8th grade core GPA as an average of these
core classes for 8th grade. We computed 7th grade course failure as 0 = no course failures in any
marking period in 7th grade and 1 = at least one 7th grade course failure in a marking period. We
computed 8th grade course failure as 0 = no course failures in any marking period in 8th grade
and 1 = at least one 8th grade course failure in a marking period.
Computing Teaching Quality Inside and Outside the IBM intervention
We did not have the resources to video-record teachers in their subject classes outside the
Pathways intervention so that a full direct comparison of general teaching quality and teaching
quality within Pathways was not possible. However, in the end-of-intervention student survey we
asked students to report on an element of teaching quality (teacher-driven classroom climate) for
each of their subject teachers (math, science, English, and history). Students rated whether their
teacher in each subject was enthusiastic, warm, clear, and knowledgeable (1 = strongly disagree,
5 = strongly agree, α = .85). To preserve independence of judgment, in our analyses, we
compared ratings of each teacher from their Pathways students (describing them in Pathways) to
the ratings each teacher received from their non-Pathways students in their subject class. The
exact wording of each item is Table S2.16 in Supplemental Materials.
ROADMAP FOR CHANGING STUDENT ROADMAPS

195
Results
Can Teachers Implement with Fidelity?
On average, training appeared successful at attaining sufficient implementation fidelity:
our brief two-day training resulted in average fidelity satisfying the 60% threshold criteria with
some room for improvement, as detailed next. Overall, 89% of students experienced the
intervention with at least 60% fidelity; the mean student-experienced fidelity was M = 68.71%,
SD = 7.11. We present these results graphically in Figure 2 by displaying the cumulative
percentage of students at each level of fidelity. Results are consistent at the classroom level,
87.5% of classrooms (seven of eight) experienced fidelity above the 60% threshold and the
eighth classroom had near threshold fidelity at 59%. For ease, classroom level results are
presented in Table 4 from lowest to highest fidelity classroom, and in Figure 3 as a boxplot. The
boxes on the left are the classrooms and the final boxplot (colored gray) is the average across
classrooms. In each boxplot, the top of the box shows the highest 25% fidelity, the bottom of the
box shows the lowest 25%, and the dark line in the box shows median fidelity. The whiskers
show the lowest and highest fidelity. As a can be seen, Figure 3 graphically shows that fidelity
across classrooms fits the Durlak and DuPre (2008) 60% threshold to 80% practical maximal
range. On average, classrooms fidelity was 68.28% (SD = 5.95%). Student variation within
classrooms is presented in Figure 3 as whiskers representing the upper and lower limits of
fidelity experienced by students in each classroom. Following Tukey (1977), whiskers exclude
any extreme outliers that are beyond 1.5 times the size of the difference between the lowest 25%
and the highest 75% above or below those quartiles; these outliers are represented individually as
dots.

ROADMAP FOR CHANGING STUDENT ROADMAPS

196

Figure 2. Cumulative percentage of students by fidelity level

Table 4. Classroom Fidelity Ordered From Lowest to Highest Fidelity Classroom
Average Fidelity
Classroom M SD
1 59% 4%
2 64% 3%
3 66% 3%
4 67% 4%
5 69% 4%
6 69% 3%
7 76% 4%
8 77% 4%
ROADMAP FOR CHANGING STUDENT ROADMAPS

197

Figure 3. Distribution of fidelity experienced by students in each classroom.
Note: Classrooms are ordered from lowest to highest fidelity. The top and bottom of the box
represent fidelity attained at the top 75% and bottom 25% respectively. The narrower the box,
the more uniform the classroom experience is for students in this 25% to 75% range. The dark
line inside the box highlights the median (middle 50%) attained classroom fidelity. The whiskers
represent the full range of fidelity students experience in each classroom excluding outliers. We
follow Tukey’s (1977) definition of outlier scores. Outlier low scores are lower than the
difference between the bottom 25% and 1.5x the difference between the lowest 25% and highest
75%. Outlier high scores are higher than the sum of the top 25% and 1.5x the difference between
the lowest 25% and highest 75%. Outliers are plotted separately as individual dots (classroom 2,
6, and 7).

Distinguishing Fidelity Specific to Pathways
Before examining the effects of differences in Pathways fidelity on student outcomes, we
addressed the question of whether teacher quality in Pathways was distinct from teacher quality
in their subject matter classes. To do so, we used the only element of teacher quality outside of
Pathways that we had, which was student ratings of teacher-driven classroom climate in
Pathways and in subject matter classes. To preserve independence of assessment, we used
student ratings of their Pathways teacher as the teacher’s Pathways quality rating. We used
ROADMAP FOR CHANGING STUDENT ROADMAPS

198
students who did not have that teacher for Pathways to obtain the teacher’s subject class quality
rating. Then we examined the correlation between the two ratings. We found a non-significant
correlation r = .15, p = .73, which implies that teacher-level Pathways fidelity is not simply the
product of a teacher’s ability to create a positive classroom climate generally.
Further examination of the data revealed that half of teachers had higher classroom
climate scores in Pathways than in their subject classroom and half had lower classroom climate
scores in Pathways than in their subject classrooms (Pathways M = 78.51, SD = 4.04; subject M
= 80.53, SD = 5.43). A meta-analytic synthesis using a random effects model, showed no overall
significant pattern of differences, Cohen’s d = 0.16, SE = 0.15, 95% CI (0.45, -0.14), z = 1.023, p
= .31, as reflected in the nonsignificant difference and the fact that the 95% confidence interval
includes zero. The implication is that teacher inside-of-Pathways-quality (fidelity) is distinct
from teacher outside-of-Pathways-quality (as assessed by teacher-driven classroom climate).
Hence our assessment of Pathways fidelity is not simply a reflection of a teacher trait or
characteristic that is independent of training in Pathways. Having established that Pathways
fidelity is unique, we now turn to the question of whether delivering Pathways with fidelity
matters for student academic outcomes, as operationalized by core course grade point average
and course failure rates.
Effects of Fidelity
Preliminary analyses. We tested the effect of demographic variables on core course
grade point average (Core GPA) and likelihood of course failure prior to testing the effects of
fidelity on these variables. We did so in six regression equations testing different outcomes: (1)
Core GPA at the end of 7
th
grade. (2) Core GPA at the end of 8
th
grade. (3) Core GPA at the end
of 8
th
grade controlling for Core GPA at the end of 7
th
grade. (4) Any class failed in any marking
ROADMAP FOR CHANGING STUDENT ROADMAPS

199
period in 7
th
grade (1 = any failure, 0 = no failures). (5) Any class failed in any marking period in
8
th
grade (1 = any failure, 0 = no failures). (6) Any class failed in 8
th
grade controlling for any
class failed in 7
th
grade. To test for effects of demographics we followed Cohen, Cohen, West,
and Aiken (2003) and used contrast or effect codes in regression equations predicting Core GPA
(regression equations 1 to 3) and dummy codes in logistic regression equations predicting course
failure (regression equations 4 to 6). For regression equations, our contrast codes were (1 =
female, -1 = male) and (free or reduced price lunch status 1 = eligible, -1 = not eligible). We
created effect codes for each of the four racial-ethnic descriptors (Hispanic, Black, Asian,
multiracial-ethnic) with White serving as the base group. So for example, the Hispanic effect
code was Hispanic = 1, White = -1, Black = 0, Asian = 0, multiracial or multiethnic = 0. For
logistic regression equations (1 = course failure, 0 = no course failure), our dummy codes were
(1 = female, 0 = male), (1 = free or reduced price lunch eligible, 0 = not), (1 = identify as
Hispanic, 0 = not), (1= identify as African American, 0 = not), (1 = identify as Asian, 0 = not),
and (1= identify as multiracial or multi-ethnic, 0 = not). Each of these regression equations is
presented in Section 3 of our Supplemental Materials. These regressions show that being female
was associated with better outcomes and identifying as African American with worse outcomes.
In addition, identifying as Hispanic was sometimes associated with better outcomes and
receiving free or reduced lunch had a trend-level effect on 8th grade course failure. As a result,
we report all analyses with these covariates included.
Fidelity predicts core GPA. Fidelity predicted 8th grade end-of-year Core GPA,
whether or not demographic covariates were included. Specifically, each fidelity percentage
increase is associated with an increased Core GPA of 0.02. Consider what would happen if
fidelity increased from threshold level (60%) to practical maximum level (80%). This increase in
ROADMAP FOR CHANGING STUDENT ROADMAPS

200
fidelity would result in a .40 increase in Core GPA, the equivalent of moving from a C+ to
almost a B. We used 2-Step (no demographic covariates) and 3-Step (with demographic
covariates) hierarchical multiple regression analyses to test for effects of fidelity. In each
regression equation, we first controlled for prior grades by entering at Step 1 student’s final 7th
grade Core GPA and dummy codes for imputed data on fidelity and 7th grade Core GPA. Then
we asked if student-level fidelity mattered either by entering it at Step 2 in the 2-Step model or
by entering it at Step 3, after first controlling for being female, free and reduced price lunch
status, and identifying as Hispanic, as African American, as Asian, or as multiracial or multi-
ethnic at Step 2.
Both models revealed that fidelity mattered for 8th grade end-of-year Core GPA (Table 5, top
panel, 2-Step model B = .024, SE = .006, β = .205, p < .001, 95% CI [.013, .036]; Table 5,
bottom panel, 3-step model B = .020, SE = .005, β = .170, p < .001, 95% CI [.010, .030]). These
effects remain significant and virtually unaltered if we exclude data from the small special
education classroom or from students with imputed data, as detailed in Section 4 of the
Supplemental Materials.
Table 5. Effects of Student-level Fidelity on 8
th
Grade Core GPA
Predictor B SE β t p ∆R
2
p
Two Step Model .468 .00
Step 1
7
th
Grade Final Core GPA 0.59 0.05 0.62 12.12 .00
Imputed fidelity measures -0.53 0.13 -0.22 -3.99 .00
Imputed 7th Grade Core GPA 0.17 0.19 0.05 0.88 .38
Step 2 .041 .00
Fidelity 0.02 0.01 0.21 4.14 .00
Three Step Model
Step 1 .468 .00
7
th
Grade Final Core GPA 0.59 0.05 0.62 12.12 .00
ROADMAP FOR CHANGING STUDENT ROADMAPS

201
Imputed fidelity measures -0.53 0.13 -0.22 -3.99 .00
Imputed 7th Grade Core GPA 0.17 0.19 0.05 0.88 .38
Step 2 .140 .00
Hispanic 0.55 0.10 0.42 5.55 .00
Black -0.23 0.14 -0.11 -1.68 .10
Asian 0.02 0.11 0.01 0.17 .86
Multiracial-ethnic -0.30 0.31 -0.10 -0.98 .33
Free or reduced price lunch 0.12 0.08 0.07 1.47 .14
Female 0.07 0.04 0.09 1.77 .08
Step 3 .028 .00
Fidelity 0.02 0.01 0.17 3.90 .00
Note: Fidelity is computed at the student-level

Does moving from threshold fidelity improve core GPA? Our next analyses unpacked
these positive effects of fidelity on Core GPA. We examined whether the significant effect of
fidelity was due to the positive effect of fidelity for students and classrooms near the practical
maximum of fidelity (80%) or if positive effects could already be seen at the mid-range between
threshold and practical maximum. This is different from simply finding that higher fidelity
matters since it pinpoints more specifically what level of fidelity training should target. As
detailed next, we found that being near the practical maximum of fidelity mattered.
We addressed this question by splitting our students into three equal groups based on
their student-level fidelity scores. The bottom third (range 48.90% to 65.47%; M = 61.33%, SD =
3.72%) averaged at about what Durlak and DuPre (2008) described as threshold fidelity. The top
third (range 70.86% to 85.54%; M = 76.22%, SD = 3.50%) averaged at about what Durlak and
DuPre (2008) described as the ‘practical maximum’ of delivered fidelity. The middle third
(65.47% to 70.80%; M = 67.99%, SD = 1.55%) averaged about midway between these. We used
these three fidelity groups and ran two analyses of covariance (ANCOVA) models, one without
ROADMAP FOR CHANGING STUDENT ROADMAPS

202
demographic covariates (F(2, 203) = 14.07; p < .001, η
2
= .12), and one with demographic
covariates (F(2, 197) = 11.78; p < .001, η
2
= .11). Each showed a significant effect of fidelity
group on 8
th
grade end-of-year Core GPA, controlling for 7
th
grade core GPA and whether data
were imputed (dummy variables). As detailed in Section 5 of the Supplemental Materials, these
effects are significant and virtually unaltered if we exclude data from the small special education
classroom or from students with imputed data.
We followed up with three planned contrasts, contrasting the practical maximum group to
the lower threshold group and to the mid-range group, and contrasting the lower threshold group
to the mid-range group. Given multiple comparisons, we applied Bonferroni adjustments to all p-
values and confidence intervals. Being near the practical maximum mattered. The results of the
planned contrast showed that the practical maximum group diverged from the other two groups.
Being in the practical maximum fidelity group was associated with significantly higher Core
GPA than being in the mid-range fidelity group. This result was found both for analyses without
demographic covariates (F(1, 203) = 16.73, p < .001, 95% CI of the between-group difference
[.164, .634], η
2
= .08) and for analyses with demographic covariates (F(1, 197) = 8.98, p < .01,
95% CI of the between-group difference [.052, .487], η
2
= .04). Being in the practical maximum
fidelity group was also associated with significantly higher Core GPA than being in the near
threshold fidelity group. This result was found both for analyses without demographic covariates
(F(1, 203) = 24.73, p < .001, 95% CI of the between-group difference [.244, .703], η
2
= .11) and
for analyses with demographic covariates (F(1, 197) = 23.14, p < .001, 95% CI [.202, .610], η
2

=.11). Core GPA did not differ for students in the near threshold fidelity group compared to
students in the midrange fidelity group whether analyses were without demographic covariates
F(1, 203) = 0.60, p = 1.00, 95% CI [-.158, .307], η
2
=.00 or with demographic covariates
ROADMAP FOR CHANGING STUDENT ROADMAPS

203
F(1,197) = 2.49, p = .35, 95% CI [-.073, .346], η
2
=.01.
For ease, we also represented these results graphically in Figure 4, without demographic
covariates and only for students with non-imputed data. We plotted students’ 8
th
grade end-of-
year Core GPA as a function of their 7
th
grade end-of-year Core GPA separately for each fidelity
group. This allowed us to see the effect of fidelity over and above the effect of the prior year’s
academic outcome. Specifically, we plotted a dot for each student, with their 7th grade Core
GPA as the x-value and their 8th grade Core GPA as the y-value. We used green colored dots for
students who experienced near practical maximum fidelity (M = 76.31%, SD = 3.53%). The
green regression line shows the predicted 8th grade Core GPA given 7th grade GPA for students
experiencing near practical maximum fidelity. We used red colored dots for students who
experienced near threshold fidelity (M = 61.08%, SD = 3.84%). The red regression line shows
the predicted 8th grade Core GPA given 7th grade GPA for students experiencing near threshold
fidelity. We used gray colored dots for fidelity at the mid-range between these two (M = 68.10%,
SD = 1.48%). The gray regression line shows the predicted 8th grade Core GPA given 7th grade
GPA for students experiencing mid-range fidelity. The effect of being in the near practical
maximum group is easy to see by looking at the green colored regression line. As can be seen,
the green line is above both the gray (midrange group) and red (near lower threshold group)
regression lines. The gray and red regression lines almost fully overlap. While visually, the green
regression line is particularly divergent from the others for students who entered 8
th
grade with
poorer 7
th
grade core course grades, we do not find an interaction between prior grades and
fidelity using the continuous measure of fidelity and non-imputed data (B = -.01, SE = .01, β = -
.59, p= .22, 95% CI [-.018, .004]).

ROADMAP FOR CHANGING STUDENT ROADMAPS

204

Figure 4. The relationship between fidelity and Core GPA
Note: The green represents students receiving the intervention near practical maximum (top
third, M=76.31%), the red represents students receiving the intervention near threshold (bottom
third, M=61.08%), the gray represents students receiving the intervention in the mid-range
(middle third, M=68.10%).

Fidelity predicts course failure. Fidelity predicted course failure. Specifically, each
increase in a single percentage point of fidelity is associated with a 5.9% reduction in the odds of
having even a single course failure (or alternatively, a 6.3% increase in the odds of passing every
course in every marking period.) Consider what would happen if fidelity increased from
threshold level (60%) to practical maximum level (80%). This increase in fidelity is associated
with a reduction in the predictive probability of failing a course from about 28% to 10%. As
detailed next, fidelity mattered whether or not demographic controls were used.
We used 2-Step (no demographic covariates) or 3-Step (with demographic covariates)
logistic regression equations in these analyses. We used logistic regressions because course
failure is a binary variable (1 = course failure, 0 = no course failure). In each equation, we first
ROADMAP FOR CHANGING STUDENT ROADMAPS

205
controlled for prior course failure by entering at Step 1 a dummy variable for a student’s failure
in any course in any marking period in 7th grade and dummy codes for imputed data on fidelity
and 7th grade Core GPA. Then we asked if student-level fidelity mattered either by entering it at
Step 2 in the 2-Step model or by entering it at Step 3 after first controlling for being female, free
and reduced price lunch status, and identifying as Hispanic, as African American, as Asian, or as
multiracial or multi-ethnic at Step 2. Fidelity predicted course failure to the same extent in both
models: no demographic controls model B = -.058, SE = .025, Wald = 5.46, Exp(B) = .943, p <
.02, 95% CI [.898, .991] and demographic controls model B = -.061, SE = .026, Wald = 5.49,
Exp(B) = .941, p < .02, 95% CI [.894, .990]). Table 6 provides the details of both models (the 2-
Step model, top panel; the 3-Step model, bottom panel). Effects remain significant and virtually
unaltered in size when analyses exclude the smaller special education classroom or exclude
students with imputed data as detailed in Section 6 of Supplemental Materials.

ROADMAP FOR CHANGING STUDENT ROADMAPS

206
Table 6. Effects of Implementation Fidelity on 8
th
grade course failure
Predictor B SE Exp(β) Wald p ∆R
2
p
Two Step Model
Step 1 .138 .00
7
th
Grade Course Failure 1.68 0.34 5.35 23.92 .00
Imputed fidelity measures 0.80 0.49 2.23 2.70 .10
Imputed 7th Grade Course
Failure
-0.15 0.70 0.86 0.05 .83
Step 2 .023 .02
Fidelity -0.06 0.03 0.94 5.46 .02
Three Step Model
Step 1 .138 .00
7
th
Grade Course Failure 1.68 0.34 5.35 23.92 .00
Imputed fidelity measures 0.80 0.49 2.23 2.70 .10
Imputed 7th Grade Course
Failure
-0.15 0.70 0.86 0.05 .82
Step 2 .056 .03
Hispanic 0.32 0.76 1.38 0.18 .67
Black 1.44 0.90 4.21 2.56 .11
Asian -0.20 0.88 0.82 0.05 .82
Multiracial-ethnic 1.51 1.66 4.53 0.83 .36
Free or reduced price lunch -1.39 0.77 0.25 3.29 .07
Female -0.77 0.36 0.46 4.55 .03
Step 3 .022 .02
Fidelity -0.06 -.03 0.94 5.49 .02
Note: Fidelity is computed at the student-level; ∆R
2
=change in Cox & Snell R-squared

Teachers’ Perspectives
AIR staff met with each teacher separately to discuss usability (how easy it was to use the
program given the provided training, resources, and support) and feasibility (how well teachers
felt they could implement the program given other demands in their teaching context) using a
two-page set of structured open-ended probes (see, Section 1, Supplemental Materials). The first
ROADMAP FOR CHANGING STUDENT ROADMAPS

207
question asked teachers what they liked about Pathways before shifting to questions about
usability and feasibility that targeted areas for improvement. In response to the first question,
teachers said that they liked, loved, or enjoyed it, that it was well-designed and well thought out,
that students were disappointed when it ended, that “It was a good platform for the students to
start to see a structure to get them to look at where they are going in the future”, and that it
“opened minds and eyes to next steps.” When asked how to improve it, they were unanimous in
three ideas as to how to increase fidelity of implementation. First, they suggested that the
intervention manual itself be reformatted to look like other teacher materials so that it would be
easier for them to process the information. Second, that the materials for students should be
reusable (e.g., laminated worksheets rather than single use). Third, that they should be provided
PowerPoint rather than newsprint for each session and activity. Finally, though not mentioned by
each of the teachers, a number also suggested changing the training to three days to provide
teachers more time to practice and absorb the intervention. When asked to detail problems
needing improvement, teachers also gave a variety of idiosyncratic critiques—critiques that were
unique to a single teacher. Unlike the unanimity of the other responses, this variability led us to
consider whether what teachers said was related to their fidelity overall or their fidelity in any
particular session. We did not find a clear pattern; it was not that teachers singled out sessions
they delivered with lower fidelity or that teachers all had problems with the same sessions.
Hence, we also viewed videotape to understand sessions in which delivery was problematic and
consider ways to improve the manual and training to address these issues.
Discussion
We found that teachers can deliver identity-based motivation intervention with fidelity in
their classrooms and that higher fidelity matters. Fidelity changes students’ academic trajectories
ROADMAP FOR CHANGING STUDENT ROADMAPS

208
particularly when it is nearer the ‘practical maximum’ of 80% found by Durlak and DuPre’s
(2008) examination of meta-analyses of practitioner-delivered interventions. Our findings
suggest that feasible increases in fidelity (moving fidelity closer to 80%) have meaningful effects
on core grade point average and course failure rates. Teachers had consensus suggestions for
improving fidelity that we implemented for future use. Moreover, after viewing videotaped
classroom sessions, we could discern what future training effort should focus on. Our findings
are important for a number of reasons. First, harnessing students’ identity-based motivation
matters for academic outcomes. Second, our findings are necessary first steps for embarking on
future ‘gold standard’ randomized control tests of the effect of identity-based motivation
intervention. Third, our findings highlight the need to assess fidelity in ways that allow
unpacking it to understand how to improve fidelity in the future.
Summary of Results
We asked if a two-day abbreviated version of the five-day training used for non-teachers
was sufficient for teachers to attain threshold fidelity and if higher attained fidelity changed
students’ academic trajectories. Based on the literature, we operationalized fidelity as five
interdependent components of dosage, adherence, quality of delivery, student responsiveness,
and fidelity-of-receipt. When teachers deliver the planned number of sessions when planned
(dosage) and with the correct content (adherence) in the correct way (quality of delivery), their
students should respond with attention, engagement, and productivity (student responsiveness)
and hence internalize the take-home points (fidelity of receipt). We carefully assessed fidelity
with reliable, structured measures coded from videotape of each session (observer report) and
student report. To do so, we used the original fidelity instruments used in the trainer-led School-
to-Jobs and incorporated the CLASS-S given developments in the field of education and its high
ROADMAP FOR CHANGING STUDENT ROADMAPS

209
overlap with the original instruments.
We followed other educational intervention evaluations by operationalizing fidelity as
sufficient if it met Durlak and DuPre’s (2008) empirically derived threshold of 60% (e.g.
Bloomquist et al., 2013; Lindsay et al., 2017). Given that Durlak and DuPre (2008) also found
that non-researchers rarely deliver with fidelity above 80%, making 80% a practical maximum of
fidelity, we asked if getting closer to this practical maximum mattered. We found that our two-
day training was successful: Almost all students received the Pathways identity-based motivation
intervention with fidelity at or above threshold and average classroom-level fidelity was within
the Durlak and DuPre suggested range of 60% to 80%. Moreover, higher fidelity mattered. We
examined the differential effects of near threshold, near practical maximum, and mid-range
fidelity. Though Durlak and DuPre noted that higher fidelity matters, their analyses are general
and do not separately examine if moving from threshold (60%) to practical maximum (80%)
fidelity has an effect on intended outcomes. Our review of the literature since then did not
uncover anyone else examining this possibility. Our own results suggest that moving from
threshold to practical maximum fidelity does matter for academic performance. Students who
received Pathways at close to practical maximum fidelity had better academic outcomes than
those who did not. These analyses showed improved end-of-year 8th grade core grade point
average and reduced likelihood of course failure, controlling for grades and course failure in 7th
grade.
Results were robust to inclusion of demographic controls of gender, race, and free or
reduced price lunch and to inclusion or exclusion of special education classroom or students with
imputed data. Effects were found for students at every level of 7
th
-grade grade point average.
Visually, effects looked stronger for students with lower 7
th
-grade core course grade point
ROADMAP FOR CHANGING STUDENT ROADMAPS

210
averages, but we did not find an interaction between fidelity and prior academic performance in
our sample.
Our end of Pathways intervention feedback from teachers was positive. They liked
Pathways and found it usable, and had very useful and actionable suggestions to improve, which
we detail after considering theoretical implications of our results. Because we had videotape of
each session, we could closely examine quality of delivery and consider ways to improve
training to more fully engage teachers with core aspects of identity-based motivation theory (e.g.
interpretation of experienced difficulty as importance).
Theoretical Implications: Identity-based Motivation
Our results add to the literature on the importance and malleability of identity-based
motivation. Because identities are experienced as stable but are in fact dynamically constructed
in context, small contextual cues can trigger important changes. For example, student’s next year
and adult possible selves can be made to feel near and connected to what they are doing right
now rather than far away and irrelevant to right now (e.g. Nurra & Oyserman, 2018). Similarly,
students can be cued to use a difficulty-as-importance mindset in making sense of experiences of
difficulty with schoolwork and in considering whether school-focused possible identities are
really ‘for me’ or ‘for us’ (Oyserman et al., 2018; Smith & Oyserman, 2015). When led to
consider their future selves as relevant to right now and to interpret experienced difficulty as a
sign that these future selves are important and that failures along the way are normal, students
succeed.
Without intervention, students may experience their future selves as far and irrelevant to
right now, and may misinterpret difficulties along the way as implying that school-focused
identities are not for them (Oyserman, 2015a). Even one-time cues such as those used in
ROADMAP FOR CHANGING STUDENT ROADMAPS

211
experiments can be powerful, influencing these elements of identity-based motivation, and
through influencing elements of IBM, changing student focus and effort, and impacting grade
point average few months later. Of course, the kinds of one-time cues used in experiments are
not enough for effects to last over years. For that, intervention is needed so that students are
repeatedly exposed to cues, shaping their focus, and hence how they likely will make sense of
their experiences over time. Prior intervention research revealed that university students and
adults with undergraduate degrees can successfully turn on middle schoolers’ identity-based
motivational processes with a brief manualized identity-based motivation intervention
(Oyserman et al., 2006; Oyserman et al., 2002). Our results extend these findings by
demonstrating that classroom teachers can implement identity-based motivation as part of the
regular school day.
Practical Implications
Our fidelity analyses allowed us to move beyond documenting that we could attain
threshold levels of fidelity to more carefully unpack whether there is a benefit of moving beyond
threshold fidelity, and if there is, at what level this benefit accrues. We found that attaining
higher fidelity matters for students’ academic trajectories, and that the effect of higher fidelity
was concentrated at fidelity closer to the practical maximum of 80%. There are a number of
important practical implications of these results. First, for identity-based motivation researchers,
our results imply that it is worth investing in teacher professional development to support
increased fidelity of receipt and of delivery of identity-based motivation. Second, our careful
assessment of fidelity and our separate request for teacher feedback worked synergistically to
allow us to respond to both teacher-noted and researcher-noted points for improvement, although
it also highlighted the limits and strengths of knowledge source. Teachers can highlight what
ROADMAP FOR CHANGING STUDENT ROADMAPS

212
feels difficult but cannot know why something felt difficult. For this reason their experience of
ease may also be misleading. Just because things feel fine does not necessarily mean that the
intervention is delivered or received as intended (e.g., the so-called Dunning-Kruger effect,
Kruger & Dunning, 1999). Researchers can see where fidelity is higher or lower but not know if
teachers experienced difficulty or ease at these points. Improving training requires both insights.
After implementing Pathways, our teachers were encouraged to highlight any problems
they had with sessions. All reported that they liked Pathways, found it worth the time to get
trained, and also suggested practical ways to improve usability and feasibility. Teachers made
pragmatic suggestions to improve fidelity that unanimously focused on creating a manual and
delivery system (e.g. pre-prepared Powerpoint rather than pre-prepared newsprint) that felt more
similar to their current textbooks. These suggestions were unanimous. Teachers also articulated
what felt difficult and which session activities did not work well for them. These differed by
teacher. This feedback step was important because teacher suggestions were different from the
suggestions we as the research team had for improving fidelity after examining the videos.
Without teacher feedback we would not have known to change the manual or to switch to
PowerPoint and reusable materials. Without the videotape we would not have been able to
understand where exactly training should be improved because teachers, like all people, do not
know what they do not know, and this is particularly true for novices who have low expertise
(the so-called Dunning-Kruger effect; Dunning, 2011; Kruger & Dunning, 1999). Teachers can
articulate what feels difficult but cannot be expected to know if the source of experienced
problems is in their training, in their preparation, in their delivery, or in the session or activity
itself. The same holds for ease: teachers can report what felt easy but not if that was because they
delivered the activity as intended and everything worked. Hence the research team watched
ROADMAP FOR CHANGING STUDENT ROADMAPS

213
classroom videos of all sessions to learn where there were gaps in the training, where delivery
fell short, separate from teacher reported ease or difficulty.
Third, our results suggest that brief training can yield fidelity separate from
characteristics of teachers. We base this implication on lack of association between Pathways
fidelity and teacher core subject and lack of association between Pathways fidelity and teaching
quality outside of Pathways. Our teachers taught each of the core subjects and we did not see a
by-subject difference in fidelity. In addition, our analyses of the relationship between teacher
quality ratings in their subject class and in Pathways revealed that teacher’s quality ratings in
their subject classes did not predict their quality ratings in Pathways.
Fourth, our results might generalize to other educational intervention evaluation efforts.
Our review of the literature did not uncover other evaluation studies that unpacked the
association of fidelity with outcomes by examining whether moving from threshold (60%) to
practical maximum (80%) fidelity mattered for intended outcomes. This is a more sensitive
analyses than simply documenting that more fidelity is better than less fidelity because it takes as
a starting point that fidelity lower than 60% is insufficient and spotlights whether fidelity above
60% yields better effects than fidelity at 60%. It is unclear whether prior analyses address this
issue because segmented analyses were not presented. An implication of our finding is that
evaluators should ask if increasing fidelity above threshold has a linear effect on outcomes. If it
does, then a careful examination of how to improve fidelity to move it beyond 60% threshold
and closer to 80% practical maximum is warranted. If it does not, then there is no need to invest
resources to improve fidelity beyond threshold.
Fifth, our results also have implications for the intervention fidelity literature. In our
review of the literature we found both agreement as to what fidelity entails and diversity as to
ROADMAP FOR CHANGING STUDENT ROADMAPS

214
how fidelity is measured (e.g., O’Donnell, 2008; Durlak & DuPre, 2008). We documented that
each of the five components of fidelity (dosage, adherence, quality of delivery, student
responsiveness, and fidelity of receipt) can be clearly operationalized and reliably assessed with
multiple measures. We benchmarked our five-component operationalization of fidelity against
the empirically derived 60% threshold and 80% practical maximum of fidelity found by Durlak
and DuPre (2008). We present empirical support both for the 60% to 80% range and document
that higher fidelity, closer to the practical maximum matters.
Finally, our results have a number of practical implications for feasible and scalable
teacher training. To be feasible, training must be brief, yet teachers need opportunities to practice
and receive feedback and to have ongoing support while implementing (Darling-Hammond,
Hyler, & Gardner, 2017). To be scalable, training should not be limited to a few trainers. To be
useful, training should feel relevant to teacher practice. In the case of Pathways, teachers had
some opportunity to practice as part of the brief 2-day training and as part of the weekly call-ins,
but an additional training day focused on implementation and structured feedback might enhance
teacher fidelity and indeed, that is what some teachers asked for. Some teachers carried
Pathways terms and concepts into their classes throughout the remainder of the academic year,
suggesting that Pathways provided a new way for teachers to engage their students about
connecting school to their futures. To support scalability, the possibility of a teacher-trained as
well as a teacher-led Pathways needs to be tested. We used our teacher feedback and
examination of videotape to develop a web-based resource including preparation tips from
teachers who delivered with fidelity, teacher-viewable videotape of high fidelity delivery, and a
video-assisted structured training module that teachers who already delivered Pathways can use
to train other teachers.
ROADMAP FOR CHANGING STUDENT ROADMAPS

215
Limitations
As with any study, our study has a number of limitations. First, our analyses include eight
classrooms and about two hundred students. While not small by standards of psychological
research, replication of our results is necessary since any one study alone cannot provide a fully
stable estimate of effects. Hence replication of our basic test is important and a goal that we are
currently pursuing in our ongoing work. Second, our sample size meant we were not powered for
mediation analyses, or to test for possible moderation of the effect of the intervention for
previously lower versus higher performing students. Instead, as noted, our goal was to test the
prediction that we could train teachers to attain threshold fidelity and so our study design focused
on training. In doing so we addressed a gap in the literature on interventions, which is that
fidelity is often not addressed at all (e.g., Chao, Visaria, Mukhopadhyay, & Dehejia, 2017) or is
not assessed carefully enough to provide an empirical assessment of how much of an intended
intervention students received (e.g., de Jong, Jellesma, Koomen, & de Jong, 2016; Bradley,
Crawford, & Dahill-Brown, 2016). Our design highlights fidelity assessment, and addresses the
question of whether threshold fidelity is likely to be attained with brief training. This is a distinct
question from the typical evaluation research question, which ignores fidelity and focuses on
student outcomes: whether students randomized to an IBM intervention group outperform
students randomized to a no-IBM control group. We could not test this latter question in our
fidelity-focused design since all students received intervention and therefore we did not
randomize to experimental and control groups.
Separate from limitations to our sample and design, there were also a number of
limitations to our training. We did not randomize teachers to varying intensity of training;
instead we chose as our start-point what we thought was the minimal training likely to provide
ROADMAP FOR CHANGING STUDENT ROADMAPS

216
teachers with sufficient chances to learn. Teachers were provided a very brief, two-day training
at their school. This 14-hour training truncated the 5-day, 40-hour training that trainers received
in the Schools-to-Job intervention. While we were able to show that we could attain fidelity of at
least threshold level, we did not test what would have happened with longer training. While our
school-based method would likely be feasible were training to become teacher-led, we did not
test what would have happened if teachers came together across schools or received longer
training. These changes might have increased quality of delivery by exposing teachers to more
diversity of styles and giving them more chance to practice. Lastly, we provided a single training
and did not assign teachers to different combinations of training for adherence vs. quality of
delivery. This means that our study cannot shed light on how the components of fidelity might
interact with each other.
Future Directions
Our current results suggest three important future directions for research: randomized
control test of outcomes, test of mediation, and development of new platforms for intervention
scaling. A randomized control test would allow us to know whether students’ academic
trajectories change as a result of being randomly assigned to Pathways compared to school as
usual or to an alternative socio-emotional or motivational intervention. A mediation test would
allow us to know whether effects are due to changes in the three components of identity-based
motivation (dynamic construction, action-readiness, and procedural-readiness). That is, whether
changes are due to changes in the extent that students experience their future selves as connected
to the present via schoolwork (dynamic construction). Whether changes are due to the extent that
students take action to start and persist in their schoolwork (action-readiness). And finally,
whether changes are due to the extent that students are flexibly able to interpret their experiences
ROADMAP FOR CHANGING STUDENT ROADMAPS

217
of difficulty as signals of importance (procedural-readiness). Another future direction is to test
whether fidelity can be maintained when training is teacher-led rather than researcher-led as it
was in the current iteration. A teacher-led training paradigm is clearly more usable and feasible
for scaling as long as it yields adequate fidelity, something that future research should test.
Conclusion
Our results shed light on both the promise of scalability and the difficulty of scaling
promising tests of theory in schools. We show that an identity-based motivation intervention can
be delivered and experienced at above threshold fidelity after a feasibly brief 2-day intervention.
We also show that moving from a threshold level of fidelity to higher fidelity matters.
Specifically, an average shift from threshold to practical maximum fidelity is associated with a
shift translating from a C+ to almost a B core course grade point average and reduction in the
predicted probability of failing a class from about 28% to 10%. Students across the continuum
from high attaining through those with individualized educational programs benefit from the
intervention. The implication is that teachers can help students harness their own high aspirations
using identity-based motivation. When teachers help students imagine school as the path to their
future, conceptualize strategies to succeed on that path, and see obstacles and failures along the
way as signaling importance and value, they are helping students succeed academically. Given
the meaningful size of effects, future work on scaling is critical.

ROADMAP FOR CHANGING STUDENT ROADMAPS

218
References
Aelenei, C., Lewis, N.A., & Oyserman, D. (2017). No pain no gain? Social demographic
correlates and identity consequences of interpreting experienced difficulty as importance.
Contemporary Educational Psychology, 48, 43-55.
Allen, J.P, Gregory, A., Mikami, A., Lun, J., Hamre, B., & and Pianta, R.C. (n.d.) Predicting
Adolescent Achievement with the CLASS-S Observation Tool. Center for
Advanced Study of Teaching and Learning, University of Virginia Curry School of
Education. Retrieved from http://curry.virginia.edu/research/
centers/castl/projects/castl-research-briefs.
Allensworth, E. M., & Easton, J. Q. (2007). What matters for staying on track and graduating in
Chicago public high schools. Chicago, IL: Consortium on Chicago School Research.
Retrieved from http://files.eric.ed.gov/fulltext/ED498350.pdf
Allensworth, E. M., & Easton, J. Q. (2005). The on-track indicator as a predictor of high school
graduation. Chicago, IL: Consortium on Chicago School Research, University of
Chicago. Retrieved from
http://consortium.uchicago.edu/sites/default/files/publications/p78.pdf
Bellg, A. J., Borrelli, B., Resnick, B., Hecht, J., Minicucci, D. S., Ory, M., Ogedegbe, G., Orwig,
D., Ernst, D., & Czajkowski, S. (2004). Enhancing treatment fidelity in health behavior
change studies: best practices and recommendations from the NIH Behavior Change
Consortium. Health Psychology, 23(5), 443.
Bloomquist, M. L., August, G. J., Lee, S. S., Lee, C. Y. S., Realmuto, G. M., & Klimes-Dougan,
B. (2013). Going-to-scale with the Early Risers conduct problems prevention program:
Use of a comprehensive implementation support (CIS) system to optimize fidelity,
ROADMAP FOR CHANGING STUDENT ROADMAPS

219
participation and child outcomes. Evaluation and program planning, 38, 19-27.
Botvin, G. J., Baker, E., Dusenbury, L., Botvin, E. M., & Diaz, T. (1995). Long-term follow-up
results of a randomized drug abuse prevention trial in a white middle-class population.
Journal of the American Medical Association, 273(14), 1106-1112.
Botvin, G. J., & Griffin, K. W. (2015). Preventing tobacco, alcohol, and drug abuse through Life
Skills Training. In L. M. Scheier (Ed.), Handbook of drug abuse prevention research,
intervention strategies, and practice. Washington DC: American Psychological
Association.
Bradley, D. N., Crawford, E. P., & Dahill-Brown, S. E. (2016). Defining and assessing FoI in a
large-scale randomized trial: Core components of values affirmation. Studies in
Educational Evaluation, 49, 51-65.
Brehm, S. S., & Brehm, J. W. (2013). Psychological reactance: A theory of freedom and control.
New York, NY: Academic Press.
Century, J., Rudnick, M., & Freeman, C. (2010). A framework for measuring fidelity of
implementation: A foundation for shared language and accumulation of knowledge.
American Journal of Evaluation, 31(2), 199-218.
Chao, M. M., Visaria, S., Mukhopadhyay, A., & Dehejia, R. (2017). Do Rewards Reinforce the
Growth Mindset?: Joint Effects of the Growth Mindset and Incentive Schemes in a Field
Intervention. Journal of experimental psychology: General, 146(10), 1402-1419.
Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2003). Applied Multiple
Regression/Correlation Analysis for the Behavioral Sciences, 3
rd
ed. Mahwah, NJ:
Lawrence Erlbaum Associates.
Crosse, S., Williams, B., Hagen, C. A., Harmon, M., Ristow, L., DiGaetano, R., ... & Derzon, J.
ROADMAP FOR CHANGING STUDENT ROADMAPS

220
H. (2011). Prevalence and Implementation Fidelity of Research-Based Prevention
Programs in Public Schools. Final Report. Washington, D.C.: Office of Planning,
Evaluation and Policy Development, U.S. Department of Education. Retrieved From
http://files.eric.ed.gov/fulltext/ED529062.pdf
Dane, A. V., & Schneider, B. H. (1998). Program integrity in primary and early secondary
prevention: are implementation effects out of control? Clinical Psychology Review,
18(1), 23-45.
Darling-Hammond, L., Hyler, M. E., Gardner, M. (2017). Effective Teacher Professional
Development. Palo Alto, CA: Learning Policy Institute. Retrieved from
https://learningpolicyinstitute.org/sites/default/files/product-
files/Effective_Teacher_Professional_Development_REPORT.pdf
de Jong, E. M., Jellesma, F. C., Koomen, H. M., & de Jong, P. F. (2016). A values-affirmation
intervention does not benefit negatively stereotyped immigrant students in the
Netherlands. Frontiers in psychology, 7.
Destin, M. (2017). An open path to the future: Perceived financial resources and school
motivation. The Journal of Early Adolescence, 37, 1004-1031.
Destin, M., & Oyserman, D. (2009). From assets to school outcomes: How finances shape
children's perceived possibilities and intentions. Psychological Science, 20(4), 414-418.
Durlak, J. A., & DuPre, E. P. (2008). Implementation matters: A review of research on the
influence of implementation on program outcomes and the factors affecting
implementation. American Journal of Community Psychology, 41 (3-4), 327-350.
Dusenbury, L., Brannigan, R., Falco, M., & Hansen, W. B. (2003). A review of research on
fidelity of implementation: implications for drug abuse prevention in school settings.
ROADMAP FOR CHANGING STUDENT ROADMAPS

221
Health Education Research, 18(2), 237-256.
Dunning, D. (2011). The Dunning–Kruger effect: On being ignorant of one's own ignorance.
Advances in experimental social psychology, 44, 247-296.
Elmore, K. C., & Oyserman, D. (2012). If ‘we’ can succeed, ‘I’ can too: Identity-based
motivation and gender in the classroom. Contemporary Educational Psychology, 37(3),
176-185.
Elmore, K., Oyserman, D., Smith, G.C., & Novin, S. (2016). When the going gets tough:
Implication of reactance for interpretations of experienced difficulty in the classroom.
AERA Open, 2(3), 1-11.
Fagan, A. A., Hanson, K., Hawkins, J. D., & Arthur, M. W. (2009). Translational research in
action: Implementation of the Communities That Care prevention system in 12
communities. Journal of Community Psychology, 37(7), 809-829.
Fleiss, J. L., & Cohen, J. (1973). The equivalence of weighted kappa and the intraclass
correlation coefficient as measures of reliability. Educational and psychological
measurement, 33(3), 613-619.
Hulleman, C. S., & Cordray, D. S. (2009). Moving from the lab to the field: The role of fidelity
and achieved relative intervention strength. Journal of Research on Educational
Effectiveness, 2(1), 88-110.
Kruger, J., & Dunning, D. (1999). Unskilled and unaware of it: how difficulties in recognizing
one's own incompetence lead to inflated self-assessments. Journal of Personality and
Social Psychology, 77(6), 1121-1134.
Landau, M. J., Oyserman, D., Keefer, L. A., & Smith, G. C. (2014). The college journey and
academic engagement. Journal of Personality and Social Psychology, 106(5), 679-698.
ROADMAP FOR CHANGING STUDENT ROADMAPS

222
Landis, J. R., & Koch, G. G. (1977). The measurement of observer agreement for categorical
data. Biometrics, 33(1), 159-174.
Lindsay, J., Davis, E., Stephan, J., & Proger, A. (2017). Impacts of Ramp-Up to Readiness
after one year of implementation (REL 2017–241). Washington, DC: U.S. Department of
Education, Institute of Education Sciences, National Center for Education Evaluation
and Regional Assistance, Regional Educational Laboratory Midwest. Retrieved from
http://ies.ed.gov/ncee/edlabs/projects/project.asp?projectID=1461.
Mowbray, C. T., Holter, M. C., Teague, G. B., & Bybee, D. (2003). Fidelity criteria:
Development, measurement, and validation. American journal of evaluation, 24(3),
315-340.
Nurra, C. & Oyserman, D. (2018). From future self to current action: An identity-based
motivation perspective. Self and identity.
O’Donnell, C. L. (2008). Defining, conceptualizing, and measuring fidelity of implementation
and its relationship to outcomes in K–12 curriculum intervention research. Review of
Educational Research, 78(1), 33-84.
Oyserman, D. (2015a). Identity-based motivation. In R. Scott & S. Kosslyn (Eds.), Emerging
Trends in the Social Sciences. Hoboken, NJ: John Wiley and Sons.
Oyserman, D. (2015b). Pathways to success through identity-based motivation. NY, New York:
Oxford University Press, USA.
Oyserman, D. (2007) Social identity and self-regulation. In A. Kruglanski & T. Higgins (Eds.)
Handbook of Social Psychology, Second Edition. (pp. 432-453). NY: Guilford Press.
Oyserman, D., Bybee, D., & Terry, K. (2006). Possible selves and academic outcomes: How and
when possible selves impel action. Journal of Personality and Social Psychology, 91(1),
ROADMAP FOR CHANGING STUDENT ROADMAPS

223
188-204.
Oyserman, D., & Destin, M. (2010). Identity-based motivation: Implications for intervention.
The Counseling Psychologist, 38(7), 1001-1043.
Oyserman, D., & Destin, M., & Novin, S. (2015). The context-sensitive future self: Possible
selves motivate in context, not otherwise. Self and Identity, 14(2), 173-188.
Oyserman, D., Elmore, K., Novin, S., Fisher, O, & Smith, G.C. 2018). Guiding people to
interpret their experienced difficulty as importance highlights their academic possibilities
and improves their academic performance. Frontiers in Psychology.
Oyserman, D., Johnson, E., & James, L. (2011). Seeing the destination but not the path: Effects
of socioeconomic disadvantage on school-focused possible self content and linked
behavioral strategies. Self and Identity, 10(4), 474-492.
Oyserman, D., & Lewis, N. A. (2017). Seeing the destination AND the path: Using
identity-based motivation to understand and reduce racial disparities in academic
achievement. Social Issues and Policy Review, 11(1), 159-194.
Oyserman, D., Lewis Jr, N. A., Yan, V. X., Fisher, O., O'Donnell, S. C., & Horowitz, E. (2017).
An identity-based motivation framework for self-regulation. Psychological Inquiry,
28(2-3), 139-147.
Oyserman, D., Terry, K., & Bybee, D. (2002). A possible selves intervention to enhance school
involvement. Journal of adolescence, 25(3), 313-326.
Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of persuasion. In L.
Berkowitz (Ed.). Advances in experimental social psychology (Vol. 19, pp. 123–205).
New York: Academic Press
Pianta, R. C., Hamre, B., Hayes, N., Mintz, S., & LaParo, K. M. (2008). Classroom Assessment
ROADMAP FOR CHANGING STUDENT ROADMAPS

224
Scoring System - Secondary (CLASS-S). University of Virginia.
Preschool Curriculum Evaluation Research Consortium (2008). Effects of Preschool Curriculum
Programs on School Readiness (NCER 2008-2009). Washington, DC: National Center
for Education Research, Institute of Education Sciences, U.S. Department of Education.
Washington, DC: U.S. Government Printing Office.
Resnick, B., Bellg, A. J., Borrelli, B., De Francesco, C., Breger, R., Hecht, J., ... & Ogedegbe, G.
(2005). Examples of implementation and evaluation of treatment fidelity in the BCC
studies: where we are and where we need to go. Annals of Behavioral Medicine, 29(2),
46-54.
Riordan, J., Lacireno-Paquet, N., Shakman, K., Bocala, C., & Chang, Q. (2015). Redesigning
teacher evaluation: Lessons from a pilot implementation (REL 2015–030). Washington,
DC: U.S. Department of Education, Institute of Education Sciences, National Center for
Education Evaluation and Regional Assistance, Regional Educational Laboratory
Northeast & Islands. Retrieved from http://ies.ed.gov/ncee/edlabs.
Smith, G. C., & Oyserman, D. (2015). Just not worth my time? Experienced difficulty and time
investment. Social Cognition, 33(2), 1-18.
Tukey, J.W. (1977). Exploratory Data Analysis. Reading, MA: Addison-Wesley.
Velasco, V., Griffin, K. W., Botvin, G. J., Celata, C., & Lombardia, G. L. (2017). Preventing
adolescent substance use through an evidence-based program: effects of the Italian
adaptation of Adaptation of Life Skills Training. Prevention Science, 18(4), 394-405.

ROADMAP FOR CHANGING STUDENT ROADMAPS

225
Supplemental Materials
Section 1 – Teacher Interview Protocol
Usability of Pathways for Success
We are going to start our conversation today discussing the usability of the program. In other
words, we want to understand how easy it was to use the program given the training, resources,
and support provided to you; whereas in the second part of the interview we will discuss the
feasibility of the program, or how well you were able to implement the program given the other
demands being placed on you in your teaching context.

• How did you like the Pathways for Success program? Why do you say that?
o [Probe: What aspects of the program did you like? Thought could be improved?]

• In our focus group discussion, we asked how confident you were implementing the
program. Can you tell me how confident you were before you began implementing the
program? How confident are you now? Why do you think your confidence changed or
stayed the same?

• In our last discussion, we also talked about the training that you received on Pathways for
Success. How do you feel about the training you received after having implemented the
program?
o [Probe: What aspects of the training do you feel helped you the most? How could
the training have been improved?]

• We also provided you a weekly implementation manual. How did you use the manual?
o [Probe: What aspects of the manual did you find helpful? How could the manual
be improved?]

• Let’s move from the implementation manual to the actual activities. What did you think
about the Pathways for Success activities?
o [Probe: How useful do you think they were for students to accomplish program
goals? Were there any activities that were not useful for students to accomplish
program goals?]
o [Probe: What activities were easy to implement? What made those activities easy
to implement?]
o [Probe: What activities were difficult to implement? What made those activities
difficult to implement?]

• Throughout the six weeks you were implementing the program, we provided a weekly
check-in via phone. Did you use the weekly check-ins? If so, can you tell me how you
used the weekly check-ins?
o [Probe: How were the check-ins helpful? How could they be improved?]

Feasibility of Pathways for Success
We just talked about the various components of the program, and how useful they were and their
ROADMAP FOR CHANGING STUDENT ROADMAPS

226
ease of use to implement. Now we want to shift focus to the feasibility of implementation. In
other words, we want to understand easy or difficult it was to implement this program given your
other responsibilities and initiatives in the school.

• Can you tell me how well the program worked with other initiatives? How did it fit into
the overall structure of your school and its class schedule?
o [Probe: What school structures were in place that facilitated implementation?
What school structures were in place that hindered implementation?]
o [Probe: How did the Pathways for Success connect to other school-level
curriculum or initiatives, if any?]
o [Probe: We asked you to implement it in your homeroom or during an advisory
period; was that the best time and place to implement the program? Why or why
not?]

• We know that you have a lot on your plate, and you have limited time with your students
every day. What do you think about the time commitment of Pathways for Success?
o [Probe: How could the program be adapted to improve implementation, given the
constraints on your time and other responsibilities?]
o [Probe: Thinking back to our conversation about activities, how could the
activities be adapted to better fit the time constraints?]

• How did the resources provided by the Pathways for Success program help you
implement the program given your other responsibilities and time commitments?
o [Probe: Is there a way that the resources provided could be improved? What
additional resources or materials would make the activities more feasible?]

• How do you plan to use what you have learned from Pathways for Success in the future?
o [Probe: Do you incorporate any of the concepts or activities into other classes
(e.g. English or math)? Why or why not?]
o [Probe: Do you plan to implement Pathways for Success next year? Why or why
not?]

Conclusion
Before we finish up, is there anything we haven’t asked you today that you think would help us
better understand your experiences thus far with the Pathways for Success program?
ROADMAP FOR CHANGING STUDENT ROADMAPS

227

Section 2 – Fidelity Instruments
Table S2.1
Session 1 (“Setting the Stage & Introductions”) Fidelity Checklist

Task Y N Detailed Teacher Behavior Y N Student Behavior Y N
Agenda hung
Opening
Welcome Greet participants and latecomers Greet teacher
State program title, an overarching description of program
and how often and when it will meet.
Listen
Introductions Teacher introduces self (Name) Listen
States there is observer/videographer/camera in the room to
observe trainer (improve program not grade students)
Acknowledge
observer/
videographer if
present

Introduction
Introduce the concept
of introductions as goal
oriented
Ask what an introduction is Share ideas

Write student responses down Listen
Reinforce: is a way of saying who you are and what you can
contribute

Different goals for introductions
Show preprinted newsprint definition (Introduction)
Introduce Pathways to
Success as success
oriented
State that Pathways to Success will focus on working to
reach goals and being successful.
Listen
Ask about skills and abilities for succeeding in school Share ideas
Reintroduce self with a skill or ability Listen
Introductions task
ROADMAP FOR CHANGING STUDENT ROADMAPS

228
Group creation process Explain activity Listen
Pass out marbles Take marble/Find
Partner

Creating sense of
competence
Ask for questions Respond to teacher

Circulate, check for understanding Talk in pairs
Make big circle Form circle
Ask youth to introduce partners/ask for repetition of names
and skills
Introduce partner
State specific plan for who is speaking
Has students repeat the names/skills of ALL those who have
already been introduced
Repeat names and
skills

Teacher participates
Expectations &
Concerns

Elicit sense of self-
control
Introduce new task, explain concept
Ask for examples Participate
Use newsprint to write group expectations
Use newsprint to write group concerns
Reinforce and repeat 4 basic themes (seeing both my far and
near future/developing strategies to work toward my
future/seeing the path between now and my future/getting
help (parents, community members, and teachers can be
resources)
Listen
Rules
Provide a sense of
safety
Elicit group rules (everyone participates, no name calling) Participate
Write on newsprint
Goals
State goal Listen
Show prepared newsprint
Naming Group
ROADMAP FOR CHANGING STUDENT ROADMAPS

229
Explain activity
Give examples, elicit ideas Share ideas
Call for a vote Vote
Human knot task
Group creation process Explain task, stand in circle, cross arms in front and grab
hands of two people across the circle, then without letting
go of hands, get them uncrossed so that we are again in a
circle
Move, reform circle
Ensures all students participate
Teacher is part of the circle
Reinforce cooperation and congratulate
Next session and
goodbyes

Summary Statement: strengths to succeed Listen

Connecting Statement: Next session will work on adult
images

Completed necessary
components of session
in appropriate time

Table S2.2
Session 2 (“Adult Images”) Fidelity Checklist

Task Y N Detailed Teacher Behavior Y N Student Behavior Y N
Agenda hung
Opening
Welcome Greet participants and latecomers Greet trainer
Say today is session 2, adult images
Last session Ask for what happened last session and why Share ideas
(Learned names
about each other,

Reinforce student participation (Why: people have lots of
different skills that will help them succeed)

ROADMAP FOR CHANGING STUDENT ROADMAPS

230
expectations,
concerns, games as
a team, adding and
building on each
others’ skills)
Bridging Teacher bridges last session and this session (last session
we focused on skills and abilities to succeed in school,
today we want to look towards the future and the adults we
want to become)
Listen
Images
Introduce the concept
of adult images
Explain task – choosing pictures that represent images of
yourself as an adult. Each to pick 3 to 5 pictures, what do
they mean for you and when these will be true of you,
afterwards share
Listen

Create personal images Make instructions clear/Ask for questions Ask
questions/Clarifies
directions

Have participants begin

Move around room,
picking pictures

Mingle – check for understanding
Share Have everyone rejoin circle
Explain task – show 1 picture and explain to group, while
group listens and pays attention
Participate
Write participant responses on newsprint, clustering by
themes
Listen
Domains of adulthood
Highlight various
domains
Explain task – participant to call out what they thought was
similar about everyone’s adult images
Share ideas

Reinforce personal
competence in noticing
connections, ability to
contribute to the in
group
Highlight themes that emerge (e.g., jobs, family, friends,
community involvement, life style; trainer need only
mention domains that did emerge)

Listen
ROADMAP FOR CHANGING STUDENT ROADMAPS

231
Next session and
goodbyes

Summary Statement: adult images can be about jobs,
family, friends, community involvement, and lifestyle (only
those group brought up or implied) (adult images + repeat
themes)
Listen

Connecting statement: next session we’ll identify models
and forces that help us work on those adult images that are
goals

Completed necessary
components of session
in appropriate time

Table S2.3
Session 3 (“Positive and Negative Forces”) Fidelity Checklist

Task Y N Detailed Teacher Behavior Y N Student Behavior Y N
Agenda hung
Opening
Welcome Greet participants and latecomers Greet trainer
Say today is session 3, positive role models & negative
forces

Last Session Ask for what happened last session and why (Why:
everyone has adult images)
Share ideas (picked
pictures of adult
images, when
would happen and
what had in
common)

Ask for domains of adulthood that discussed Share ideas
(lifestyle, career,
relationship,
community
involvement)

ROADMAP FOR CHANGING STUDENT ROADMAPS

232
Reinforce student participation (reinforce what students
said: Both what happened, everyone has adult images, and
domains of adulthood)
Listen
Bridging Trainer bridges last session and this session (discuss
difference between dreams and possible selves)
Listen
Role models and
negative forces

Adults images come
from somewhere

Ask for what are positive models and negative forces. Share ideas

Reinforce participation.
Use Newsprint to write what they say
Define terms (positive role model – image of attained
goal/supports work toward it, negative model – image of
failure, undermines effort)
Share ideas
Show preprinted newsprint definition Listen
Those close to us, often
parents, can support or
tear down
Explain task/handout worksheets. Start with Job domain –
write/draw adult image and a positive and negative force for
that adult image
Write goals/role
models/force

Mingle, check for understanding
Have students organize into circle Students organized
in a circle

Discuss models and negative forces. Have students give
examples
Participate
Write on newsprint as they do (cluster similar)
Read through positive model list Listen,
Read through negative force list Listen,
Say close people in our lives can be supporting,
Say everyone has negative forces
Next session and
goodbyes

Summary Statement: we worked on, role models and
negative forces, everyone has both
Listen
Connecting Statement: next session timelines into the future
ROADMAP FOR CHANGING STUDENT ROADMAPS

233
Completed necessary
components of session
in appropriate time

Table S2.4
Session 4 (“Timelines”) Fidelity Checklist

Task Y N Detailed Teacher Behavior Y N Student Behavior Y N
Agenda hung
Opening
Welcome Greet participants and latecomers Greet trainer

Say today is session 4, time lines
Last Session Ask for what happened last session and why (everyone has
positive and negative forces, can help us see path or make
more difficult). Ask to repeat themes
Share ideas
(positive and
negative forces for
career possible
selves, examples;
themes – jobs,
family, community
participation,
friends, lifestyle)

Reinforce student participation (why: difficulties along way
are normal; working on difficult things is important)
Listen
Bridging Trainer bridges last session and this session (Today we will
begin to map out how to get from now to the future)
Listen
Timelines Activity
Create sense of linear
time
Ask what are timelines, ask for ideas Share ideas
Write ideas on newsprint
Repeat examples (linear, history, now in future and future
not for sure)
Listen
Reveal preprinted newsprint timeline (general)
Create sense of Explain fork in the road Listen
ROADMAP FOR CHANGING STUDENT ROADMAPS

234
competence to handle
choices, obstacles;
Reinforce naturalness
of obstacles
Ask students for examples Share ideas
Explain obstacles-barriers-road blocks Listen
Ask students for examples Share ideas
Reveal preprinted timeline into the future Listen
Create Timelines Explain the tasks – rough draft then on timeline, everything
from now as far as can go, in order (at least 1 fork and 1
obstacle).
Listen
Pass out materials – tell to spread out Move
Repeat instructions as needed (out loud, individually) Work
Circulate & provide help Work
Discuss Timelines
Regroup Students (I know you are not done but..) Regroup
Ask students to show timelines Show timelines
Ask students to point to their fork in the road Show forks
Ask students to point to their obstacle Show obstacles
Get one or two students to state their fork Share ideas
Get one or two different students to state their obstacle
Time permitting, suggest an additional step in the timeline
for another student

Next session and
goodbyes

Summary Statement: timelines, forks in the road, obstacles,
timelines into the future
Listen
Connecting Statement: next session is 5, Action Goals
Completed necessary
components of session
in appropriate time

ROADMAP FOR CHANGING STUDENT ROADMAPS

235
Table S2.5
Session 5 (“Action Goals”) Fidelity Checklist

Task Y N Detailed Teacher Behavior Y N Student Behavior Y N
Agenda hung
Opening
Welcome Greet participants and latecomers Greet trainer
Say today is session 5, action goals
Last session Ask for what happened last session and why Participate
(Timelines, forks,
obstacles)

Reinforce student participation (why: timelines help us
order things from next year to future)
Listen
Bridging Trainer bridges last session and this session (Say: action
goals, timelines, adult images)
Listen
Action Goals
Define term Before we talked about adult images, some were PS and
others were dreams. What is the difference between a PS
and a dream?
Share ideas
Reveal preprinted newsprint definitions (action goals)
Define terms (Goal, Action Goal, An action goal has an
adult PS, closer PS, strategy and when and where actions
occur; this takes the form of a sentence)

Use worksheet to
practice Action Goals
Explain task – write specific action goal for two domains of
adulthood, start with career (Because, I will, By, When-
Where)
Listen
Provide example of an action goal Share ideas
Circulate & provide help Work
Move back to circle Move
Have students read their action goals (starting with someone
who hasn’t yet participated),
Participate
Use newsprint to cluster common themes from the career
ROADMAP FOR CHANGING STUDENT ROADMAPS

236
“Because,” clustering by the “I will” statements (these are
the near possible selves, many of which will involve school)
Anyone who doesn’t have today action, trainer helps
problem solves suggests working on different peace
Participate
Summarize the cluster themes Listen
Next session and
goodbyes

Summary statement: today worked on action goals Listen
Connecting statement: next session we will work on
possible selves and strategies

Completed necessary
components of session
in appropriate time

Table S2.6
Session 6 (“Possible Selves and Strategies) Fidelity Checklist

Task Y N Detailed Teacher Behavior Y N Student Behavior Y N
Agenda hung
Opening
Welcome Greet participants and latecomers Greet
Say this is session 6, Possible selves and strategies
Last session Ask for what happened last session and why (helps narrow
down to focus on the possible selves we can actually start
working on now)
Share ideas (Action
goal, Because, I
Will, By, When,
Where, give
examples that show
these and the idea
of linking far and
near future goals)

Reinforce the concept of action goals (because, I will, by,
when-where) link more distal goals to closer ones with
Listen
ROADMAP FOR CHANGING STUDENT ROADMAPS

237
activities to be done in certain times and contexts
Bridging Trainer bridges last session and this session (last session we
linked adult images to the next few year with action that
could be done right now, this week). This session we will
focus on the really close future, next year.
Listen
Possible Selves
Defining PS and
strategies
Introduce new concepts (expected, to-be avoided possible
selves and strategies)
Listen
Reveal preprinted newsprint definitions (possible selves)
Reveal preprinted newsprint definitions (strategies)
Connecting next year
goals and strategies
Show blank Poster Board (left, middle, right) Listen
Provide instructions for next year PS (focus on left only)
Making poster boards

Explain Use of Next Year PS stickers (read and choose 5
expected, 5 to be avoided for you, can write your own, after
read, do not pull off backing until picked best 5, choose
expected for the left top and to be avoided for bottom)
Listen
Show with finger on board. Pass out sticker bag (repeat
instructions read before peeling, only 5, top expected,
bottom to be avoided)
Listen

Pass out boards (repeat instructions out loud) Move, Work
Circulate, check for understanding
Explain choosing strategies connecting to next year selves
Pass out strategy stickers (collect PS stickers)
Repeat Instructions (ask if are doing anything to work on a
PS and if so, use a sticker to say what and place on board)
Listen
Circulate, check for understand Work
Explains use of red markers, pass out red markers Listen
Explains use of red markers again out loud and individually Work
Walk through group continuously, helping, giving positive
reinforcement, clarifying instructions

Explains use of blue markers Listen
Ask for red back in exchange for blue Work
ROADMAP FOR CHANGING STUDENT ROADMAPS

238
Circulate, look at boards, remark out loud, some possible
selves have strategies we are using now, they are marked
with a red line, some possible selves have strategies we
could be using but are not now, they are marked with a blue
line, some possible selves have no strategies. Then the
strategy space is blank.

Next session and
goodbyes

Summary Statement: today worked on possible selves and
strategies boards)
Listen
Connecting Statement: next session we will finish our
poster boards by listing adult possible selves and seeing if
there are pathways from next year to adulthood through
current action

Completed necessary
components of session
in appropriate time

Table S2.7
Session 7 (“Pathways to the Future”) Fidelity Checklist

Task Y N Detailed Teacher Behavior Y N Student Behavior Y N
Agenda hung
Opening
Welcome Greet participants and latecomers Greet trainer
Say today is session 7, Pathways to the future
Last session Ask for what happened last session and why Share ideas (Poster
boards, stickers,
possible selves,
strategies, examples
showing idea that
possible selves can

Put up preprinted newsprint (PS & strategies) from Session
6

ROADMAP FOR CHANGING STUDENT ROADMAPS

239
link to strategies)
Reinforce concepts (possible selves are possible, not for
sure but not just hopes, can work on them with strategies
right now, strategies are the things we do or can do now to
make them happen).
Listen
Bridging Trainer bridges last session and this session (next year
possible selves and strategies, connected with red lines if
doing now, blue lines if could do).
Listen
Show poster board and point to right (adult possible selves)
Now we are going to do this part (point to right)
Connecting next year
and the future

Show baggies. Like last session, I will give you a bag of
stickers. These are expected and to be avoided adult selves.
Read the stickers. Pick the best 5 expected and the best 5 to
be avoided. Do not pull off backing until ready.
Listen
Repeat instructions while passing back boards and adult
stickers
Ask
questions/Clarify
instruction

Repeat instructions while circulating to look at boards,
positively reinforce effort
Work
Explain markers and distribute. Say: Put the stickers back in
the bag and I will trade you for a red marker. For each adult
or to-be-avoided possible self, if one of the strategies that
you are doing now can help you get to or avoid it, then
connect the strategy to the possible self with a red line
Work on own
Circulate, repeat instructions, look at boards and check for
understanding
Work on own
As students finish, offer to trade red markers with blue.
Instruction: Look at the strategies you could use but are not
using now. Any of these that could help with adult possible
selves, draw a blue line from the strategy you could use and
the possible self it would help.
Raise hands, swap
markers

ROADMAP FOR CHANGING STUDENT ROADMAPS

240
Circulate, repeat instructions, check boards work
Sharing pathways
Ask students to move chairs, reorient to see each other’s
work
Students are with
partners or in
groups

Ask students to show their work Show, Listen
Define pathway (Define connection as pathways (strategy
connects a next year self to adult self)
Listen
Ask students with a red pathway to read the current pathway
(strategy, next year, adult)
Show, Listen, Talk
Ask students with a blue pathway to read the potential
pathway (strategy could use, next year, adult).
Show, Listen, Talk
Reinforce individual participation in activity Listen
Next session and
goodbyes

Summary Statement: today worked on pathways Listen
Connecting Statement: next session we will work difficult
puzzles in life

Completed necessary
components of session
in appropriate time

Table S2.8
Session 8 (“Puzzles”) Fidelity Checklist

Task Y N Detailed Teacher Behavior Y N Student Behavior Y N
Agenda hung
Opening
Welcome Greet participants and latecomers by name Greet
Say today is session 8, puzzles
Last session Ask for what happened last session and why Share ideas
ROADMAP FOR CHANGING STUDENT ROADMAPS

241
Reinforce student participation (why: some strategies are
linking pathways, they can help us get from next year to
adulthood)
(Possible selves,
pathways,
strategies, red, blue;
explain idea of
strategies doing
now vs. those could
try)
Bridging Trainer bridges last session and this session (sometimes it
feels hard, impossible, so just have possible selves)
Listen
Elicit inoculation/vaccination discussion Share ideas
Use Newsprint to write student ideas
Today will work on inoculating from difficulty by solving
puzzles that feel impossible
Listen
Reveal preprinted newsprint (Inoculation)
Solving Puzzles,
Puzzle 1

Provides puzzle activity instructions Listen
Passes out puzzle 1
Reads out loud Listen
Asks students to work in groups to figure out how to solve
(move to groups)
Students are with
partners or in
groups

Reinforce cooperative participation, effort, ideas Work in groups
Have students regroup (orient to front) to give their plan of
action and talk through how far they got in trying to solve
the problem
Multiple groups
share out loud

Elicit discussion of feelings when solving something
difficult

Have multiple students walk through Puzzle 1 solution
Use Newsprint to write out student plans, possible solution
paths (or have students do it)

Reinforce many ways to solve
ROADMAP FOR CHANGING STUDENT ROADMAPS

242
Reinforce that seems impossible before trying
Solving Puzzles,
Puzzle 2

Get help passing out Puzzle 2 Help w/ passing out
puzzle 2

Read out loud Puzzle 2 Listen
Asks students to work in groups to solve
Reinforces cooperative participation, effort, ideas Work in groups
Reinforces responses that move toward problem solution
Have students regroup (orient to trainer) and give their plan
of action and talk through how far they got in trying to solve
the problem
Multiple groups
share out loud

Write out student plans, possible solution paths (or has
students do it)

Write out grid, as a possible solution path, solves out loud
with students
Share ideas
Reinforce multiple ways to solve the problem

Reinforce how impossible seems before trying Listen
Next session and
goodbyes

Summary Statement: everyday puzzles, seem impossible,
sometimes need trial and error
Listen
Connecting Statement: next session we will practice dealing
with everyday problems

Completed necessary
components of session
in appropriate time

ROADMAP FOR CHANGING STUDENT ROADMAPS

243
Table S2.9
Session 9 (“Solving Everyday Problems”) Fidelity Checklist

Task Y N Detailed Teacher Behavior Y N Student Behavior Y N
Agenda hung
Opening
Welcome Greet participants and latecomers by name Greet
Today is session 9 solving everyday problems
Last session Ask for what happened last session and why Share ideas
(Inoculation from
difficulty so won’t
get infected with
the idea that
difficulty means
impossibility,
difficult puzzles)

Reinforce student participation (the last session’s
definitions/concepts, inoculation from thinking if it is
difficult it is impossible. Can do it, often the really difficult
is not impossible, need to start and answer some questions.
Listen
Bridging Trainer bridges last session and this session (Today - I am
going to give you an everyday problem and you are going to
think of questions that you need to ask before solving it.
Listen
Then I am going to ask you for everyday problems you have
and we will use what we learned to map out questions to ask
before solving it.

Everyday Problem 1,
Math Problem

Have students move into groups Students are with
partners or in
groups

Ask for Student help to pass out math problem Help pass out math
problem

ROADMAP FOR CHANGING STUDENT ROADMAPS

244
Read out loud math problem Listen

Give students newsprint/sheets to write questions
Ask students to consider the questions they need to ask
themselves to solve this

Circulate, reinforce effort, asking questions Work in groups
Have students move to one big circle Move
Have students hang up their newsprint or elicit their ideas
and write it down
Multiple groups
share out loud

Reinforce many solutions could do this as action goals:
Because (adult image) I will (next year possible self) by
(strategy) when and where during the day____), could do as
timeline, with obstacles and forks, could do as asking for
adult help (positive models, negative forces).
Listen
Everyday Problems 2
Ask students to think about a school problem like the math
problem they have faced or are facing now in school.
Listen

Have students write down problem, crumple it up and throw
it on the floor in the middle of the room (should still be in
circle)
Writes, Throws
paper on floor/in
bag

Trainer provides reinforcement, says: so many problems,
everyone has at least one.

Trainer reads out 4 problems, group selects one Vote
Asks, what are questions to ask Participate
Trainer stands with newsprint to write questions, writes in
clusters by theme
Listen
Uses clusters to link back to the action goal, timeline and
possible self activities and for coming up with forks in the
road obstacles, models to ask for help, negative forces to
avoid

Next session and
goodbyes

Summary Statement: today everyday problems Listen
ROADMAP FOR CHANGING STUDENT ROADMAPS

245
Connecting Statement: more inoculation by looking at what
you need to finish high school and get more training-like
college

Completed necessary
components of session
in appropriate time

Table S2.10
Session 10 (“Everyday Problems: High School and Beyond”) Fidelity Checklist

Task Y N Detailed Teacher Behavior Y N Student Behavior Y N
Agenda hung
Opening
Welcome Greet participants and latecomers by name Greet trainer
This is session 10, ‘graduating’
Last session Ask for what happened last session and why Share ideas (Solved
math problem and
everyday problems
by asking, could
use the PTS
activities to do it.)

Reinforce student participation (why: we all have problems
in school that we need to solve, need to think of solution
paths to those problems)

Bridging Trainer bridges last session and this session (Last session
we began to think about everyday problems by asking
questions and using PTS activities like timelines with forks
in the road and obstacles, action goals, forks in the road and
positive and negative models. Today we are going to work
on another part of the inoculation, working on a plan for
graduating high school)
Listen
Graduating High
School

Ask students to divide into groups or turn chairs Students are in
partners/groups

ROADMAP FOR CHANGING STUDENT ROADMAPS

246
Explain activity (what does it take to graduate) Listen
Ask students types of questions they might answer Provide responses
Circulate and check on groups’ progress Work in groups
Move to one circle Participate
Elicit students responses, write on newsprint
Prompt more questions: What classes, anything else? How
many classes? Anything else? Attendance?

Prompt more questions: Anything else?
Behavior/citizenship

Reinforce how much students know
Ask for help to pass out high school graduation
requirements for own location
Students help pass
out

Read out loud Listen
Ask for help connecting this to what students already said. Share ideas
Repeats process with a second high school. Participate
Highlights how not all high schools are the same but
graduation requirements are the same
Listen
Facilitate connection of course names and content Participate/Group
discussion

Going to College
Explain task: So we figured out graduating high school, a
lot of you mentioned college, so we are going to do the
same thing, figuring out how to get from high school to
college.
Listen
Work with partner: What do you need for college? Classes?
Grades? What else
Work in groups
Elicits and write responses on newsprint about what need
for college
Participate
Passes out college entrance requirements Review college
entrance
requirements

Helps link student statements to college Share ideas
ROADMAP FOR CHANGING STUDENT ROADMAPS

247
Repeats with another college, reading out loud the
requirements and linking to student ideas
Participate
Makes connections between high school graduation
requirements and requirements to apply for colleges
Listen
Next session and
goodbyes

Summary Statement (today worked on what you need to
finish high school and get more training-like college)
Listen
Connecting statement to next session (we will have a wrap-
up session. Next session we will review all sessions and
have a party

Completed necessary
components of session
in appropriate time

Table S2.11
Session 11 (“Wrapping up and Moving Forward”) Fidelity Checklist

Task Y N Detailed Teacher Behavior Y N Student Behavior Y N
Agenda hung
Opening
Welcome Greet participants and latecomers Greet trainer
This is session 11 ‘wrapping up and moving forward’
Last session Ask for what happened last session and why Share ideas (Get
through high
school, graduate to
college)

Reinforce student participation (why: wanted to see what
was needed for high school and college, and begin to think
about what our next year selves and strategies will be to
succeed)

Bridging Trainer statement of connection from last session to this
session and overview of this session (last session we
worked on planning for - high school and college by
knowing the requirements. This is our last in school PTS
Listen
ROADMAP FOR CHANGING STUDENT ROADMAPS

248
session; this is our wrapping up and looking forward
session. I will ask you what we did each session and why,
what was best, what was worst and how to improve. This is
also our PTS party).
Party
Provide food and other party materials Eat
Ask to move desks/chairs to circle Move desks to
circle

What did we do at
PTS

Review Ask for what did in each of the previous sessions Share or participate

Ask for help getting the right order Share (should be
different students)

Write on newsprint
Elicit a reason for each session, how connect
Give voice to students Ask for: favorite sessions, least liked sessions, what to
change about program
Share (should be
different students)
or participate

Write responses on newsprint
Explain a connection between all sessions
Goodbyes
Goodbyes Say goodbyes
Completed necessary
components of session
in appropriate time

ROADMAP FOR CHANGING STUDENT ROADMAPS

249
Table S2.12
CLASS-S General Score Rubric

Low Mid High
1 2 3 4 5 6 7
The low
range
descriptio
n fits the
classroom/
teacher
very well.
All, or
almost all,
relevant
indicators
in the low
range are
present.
The low range
description
mostly fits the
classroom/teache
r but there are
one or two
indicators that
are in the mid
range.
The mid
range
descriptio
n mostly
fits the
classroom/
teacher
but there
are one or
two
indicators
in the low
range.
The mid
range
descriptio
n fits the
classroom/
teacher
very well.
All or
almost all,
relevant
indicators
in the mid
range are
present.
The mid
range
descriptio
n mostly
fits the
classroom/
teacher
but there
are one or
two
indicators
in the high
range.
The high
range
descriptio
n mostly
fits the
classroom/
teacher
but there
are one or
two
indicators
in the mid
range.
The high
range
descriptio
n first the
classroom/
teacher
very well.
All, or
almost all,
relevant
indicators
in the high
range are
present.

Table S2.13
Student-Level Report Quality of Delivery Items

During Pathways, my teacher was… (1-strongly disagree, 5-strongly agree):
…Enthusiastic
…Knowledgeable
…Warm
…Clear
During Pathways, my classmates were…(1-strongly disagree, 5-strongly agree):
…Enthusiastic
…Knowledgeable
…Warm
…Clear
In Pathways, my teacher… (1-not at all, 5-a lot):
…Listened to my comments
…Understood my problems
…Negatively criticized my ideas (reverse coded)
…Used specific examples
…Gave us all equal chance to participate
…Gave us the chance to answer questions other students raised
In Pathways to Success…(1-strongly disagree, 5-strongly agree):
…I felt comfortable participating and asking questions
…I could trust others to listen to what I had to say
…Others shared their experiences and difficulties working toward their futures
…Other students have the same problems I do
IDENTITY RELEVANCE ACROSS TIME

250
…What we talked about was relevant for me
…I felt concerned I would be negatively criticized by another group member
(reverse coded)

Table S2.14
Observer Coded Quality of Delivery Rating-Scale for Take Home Point for Each Session

Session 1: Setting the Stage and Introductions
Take Home Point: You have some skills and abilities to help you succeed in the coming year and
others do too.
0 1 2

Key point not
evoked at all by
activities
Key point was partially evoked
but framing was unclear or
inconsistent
Key point was evoked clearly and
consistently; concepts are connected to
student-generated examples.

Session 2: Adult Images
Take Home Point: We all have images of ourselves as adults in the far future.
0 1 2
Key point not
evoked at all by
activities
Key point evoked but framing
and connections were unclear or
inconsistent
Key point was evoked clearly and
consistently; concepts are connected to
student-generated examples.

Session 3: Positive and Negative Forces
Take Home Point: Positive and negative forces make some adult images possible selves.
Positive forces help us lay out paths for success and handling difficulties and setbacks; negative
forces do the opposite, layout paths for failure and examples for how not to handle difficulties
and setbacks.
0 1 2
Key point not
evoked at all by
activities
Key point evoked but framing
and connections were unclear or
inconsistent
Key point was evoked clearly and
consistently; concepts are connected to
student-generated examples.

Session 4: Timelines
Take Home Point: The future is a path, current actions set up which futures are possible
0 1 2
Key point not
evoked at all by
activities
Key point evoked but framing
and connections were unclear or
inconsistent
Key point was evoked clearly and
consistently; concepts are connected to
student-generated examples.

Session 5: Action Goals
Take Home Point: We have some control over possible selves, but not hope and dreams. That
control happens when we link the future with the present through specific action goals.
0 1 2
IDENTITY RELEVANCE ACROSS TIME

251
Key point not
evoked at all by
activities
Key point evoked but framing
and connections were unclear or
inconsistent
Key point was evoked clearly and
consistently; concepts are connected to
student-generated examples.

Session 6: Possible Selves and Strategies
Take Home Point: Strategies are actions you are taking now or could take to become your next
year possible self.
0 1 2
Key point not
evoked at all by
activities
Key point evoked but framing
and connections were unclear or
inconsistent
Key point was evoked clearly and
consistently; concepts are connected to
student-generated examples.

Session 7: Pathways to the Future
Take home point: Any strategies I’m doing (or could be doing) now to get to my next year
possible self, also help me get to my adult possible self.
0 1 2
Key point not
evoked at all by
activities
Key point evoked but framing
and connections were unclear or
inconsistent
Key point was evoked clearly and
consistently; concepts are connected to
student-generated examples.

Session 8: Puzzles
Take home point: Things can seem impossible and difficult, but can be solved by breaking them
down looking for alternative ways to set up the problem.
0 1 2
Key point not
evoked at all by
activities
Key point evoked but framing
and connections were unclear or
inconsistent
Key point was evoked clearly and
consistently; concepts are connected to
student-generated examples.

Session 9: Solving Everyday Problems
Take home point: There are everyday choice points and difficulties that are obstacles to navigate
on the path linking near and far possible selves.
0 1 2
Key point not
evoked at all by
activities
Key point evoked but framing
and connections were unclear or
inconsistent
Key point was evoked clearly and
consistently; concepts are connected to
student-generated examples.

Session 10: Everyday Problems: High School and Beyond
Take home point: You can identify the steps to get from 8
th
grade to graduating high school.
0 1 2
Key point not
evoked at all by
activities
Key point evoked but framing
and connections were unclear or
inconsistent
Key point was evoked clearly and
consistently; concepts are connected to
student-generated examples.

Session 11: Wrapping Up and Moving Forward
Take home point: What I do now makes a big difference for attaining my possible selves for next
IDENTITY RELEVANCE ACROSS TIME

252
year, for the next few years, and farther as an adult. Possible selves that are linked to strategies
and to time and place of action become action goals. There are forks (choices) and roadblocks
(failures) along the way. It will be difficult and may feel impossible, but asking questions helps
break down what I need to find out and helps me connect to others – positive forces and models
– as well as to learn from negative forces and models of what not to do.
0 1 2
Key point not
evoked at all by
activities
Key point evoked but framing
and connections were unclear or
inconsistent
Key point was evoked clearly and
consistently; concepts are connected to
student-generated examples.

Table S2.15
Student-Level Report Fidelity of receipt Items

Response scale for items 1-7: 1= Not at all Confident, 5=Very Confident.
1. I can introduce myself in a way that emphasizes my skills.
2. I can imagine myself as an adult (working, having family and friendships,
having a nice lifestyle, and participating in my community).
3. I can draw a timeline to get to my adult images, including obstacles and forks
in the road.
4. I can take action now to work toward my adult image.
5. I can break down everyday situations into problems to be solved.
6. I can ask for help making plans.
7. I can plan my class schedule to meet my future goals.
Response scale for items 8 to 10:1= Strongly Disagree, 5=Strongly Agree.
8. In the future I will experience difficulties and setbacks in my efforts to do well
in school.
9. In the future I have strategies to handle these difficulties so I know what to do
next.
10. In the future I can come up with alternatives when a setback happens.

Table S2.16
Student-Level Teacher-driven Classroom Climate for Subject Teachers

Did you have [Teacher] as a teacher for a class other than Pathways to Success?
(Yes or No)

[If yes…]

During classes other than Pathways to Success (for example, Science, Math,
English or History), [Teacher] is... (1-strongly disagree, 5-strongly agree):
…Enthusiastic
…Knowledgeable
…Warm
…Clear

IDENTITY RELEVANCE ACROSS TIME

253
Section 3 -- Preliminary analyses examining the effect of demographics on Core GPA and
course failure

Table 3.1 Effects of demographics on 7
th
grade Core GPA
Predictor B SE β t p
Hispanic -0.44 0.18 -0.31 -2.49 .01
Black -0.54 0.23 -.024 -2.39 .02
Asian 0.38 0.20 0.21 1.91 .06
Multi-racial Ethnic 0.36 0.63 0.11 0.57 .57
Free or reduced price lunch -0.05 0.14 -0.03 -0.39 .70
Female 0.27 0.06 0.31 4.81 .00
Note: Dependent variable is Core 7
th
grade GPA; ∆R
2
=0.24

Table 3.2 Effects of demographics on 8
th
grade Core GPA
Predictor B SE β t p
Hispanic 0.23 0.13 0.18 1.74 .08
Black -0.67 0.18 -0.32 -3.64 .00
Asian 0.22 0.16 0.13 1.39 .17
Multi-racial Ethnic 0.23 0.42 0.08 0.53 .59
Free or reduced price lunch 0.09 0.12 0.05 0.74 .46
Female 0.24 0.05 0.30 4.72 .00
Note: Dependent variable is Core 8
th
grade GPA; ∆R
2
=0.21

Table 3.3 Effects of demographics on change in Core GPA from 7
th
to 8
th
grade

Predictor B SE β t p ∆R
2
p
Step 1 .471 .00
Core 7
th
grade GPA 0.63 0.05 0.69 13.18 .00
Step 2 .184 .00
Hispanic 0.61 0.11 0.46 5.51 .00
Black -0.19 0.14 -0.09 -1.37 .17
Asian 0.02 0.13 0.01 0.19 .85
Multi-racial Ethnic -0.30 0.39 -0.10 -0.76 .45
Free or reduced price lunch 0.21 0.09 0.11 2.45 .02
IDENTITY RELEVANCE ACROSS TIME

254
Female 0.05 0.04 0.06 1.32 .19
Note: Dependent variable is Core 8
th
grade GPA

Table 3.4 Effects of demographics on 7
th
grade course failure
Predictor B SE Exp(β) Wald p
Hispanic 1.26 0.70 3.53 3.27 .07
Black 2.32 0.90 10.18 6.71 .01
Asian -0.19 0.81 0.82 0.06 .81
Multi-racial Ethnic -20.35 40192.93 0.00 0.00 1.00
Free or reduced price lunch 0.47 0.89 1.61 0.28 .60
Female -0.87 0.31 0.42 7.87 .01
Note: Dependent variable is 7
th
grade course failure; Change in Cox & Snell R
2
=0.13

Table 3.5 Effects of demographics on 8
th
grade course failure
Predictor B SE Exp(β) Wald p
Hispanic 0.60 0.72 1.73 0.69 .41
Black 1.94 0.86 6.95 5.05 .03
Asian -0.29 0.85 0.75 0.12 .73
Multi-racial Ethnic 0.55 1.63 1.73 0.11 .74
Free or reduced price lunch -1.18 0.70 0.31 2.81 .09
Female -0.99 0.34 0.37 8.52 .00
Note: Dependent variable is 8
th
grade course failure; Change in Cox & Snell R
2
=0.11

Table 3.6 Effects of demographics on change in course failure from 7
th
grade to 8
th
grade
Predictor B SE Exp(β) Wald p ∆R
2
p
Step 1 .158 .00
7
th
grade
course failure
1.97 0.37 7.20 29.14 .00
Step 2 .040 .14
Hispanic -0.15 0.75 0.87 0.04 .85
Black 1.08 0.92 2.94 1.38 .24
Asian -0.31 0.88 0.73 0.13 .72
Multi-racial
Ethnic
-19.562 40193.05 0.00 0.00 1.00
Free or reduced
price lunch
-1.13 0.85 0.32 1.77 .18
IDENTITY RELEVANCE ACROSS TIME

255
Female -0.68 0.37 0.51 3.38 .07
Note: Dependent variable is Core 8
th
grade course failure; ∆R
2
=change in Cox & Snell R-squared

Section 4 – Additional analyses examining the effect of fidelity on change in 8
th
grade Core
GPA

4.1 Analyses excluding students with imputed data

We used two-step and three-step hierarchical multiple regression analyses to test the
relationship between fidelity and 8th grade end of year Core GPA, excluding students with
imputed data. As detailed in the top panel of Table S4.1, the simpler two-step model excludes
covariates and showed a significant effect of fidelity (B=.026, SE=.006, β=.233, p<.001, 95% CI
[.014, .037]). As detailed in the bottom panel of Table S4.1, the three-step model includes
covariates and also showed a significant effect of fidelity (B=.020, SE=.005, β=.182, p<.001,
95% CI [.010, .030]).

Table S4.1
Effect of Student-level Fidelity on 8
th
Grade Core GPA, Excluding Students with Imputed Data
Predictor B SE β t p ∆R
2
p
Two Step Model

.455 .00
Step 1

7
th
Grade Final Core GPA 0.60 0.05 0.68 12.12 .00

Step 2

.053 .00
Fidelity 0.03 0.01 0.23 4.35 .00

Three Step Model

Step 1

.455 .00
7
th
Grade Final Core GPA 0.60 0.05 0.68 12.12 .00

Step 2

.166 .00
Hispanic 0.61 0.12 0.45 5.11 .00

IDENTITY RELEVANCE ACROSS TIME

256
Black -0.24 0.17 -0.10 -1.37 .17

Asian 0.07 0.13 0.04 0.50 .62

Multiracial-ethnic -0.29 0.40 -0.09 -0.74 .46

Free or reduced price lunch 0.21 0.09 0.12 2.38 .02

Female 0.05 0.04 0.06 1.22 .23

Step 3

.032 .00
Fidelity 0.02 0.01 0.18 3.91 .00

Note: Fidelity is computed at the student-level

4.2 Analyses excluding smaller, special education classroom

We used two-step and three-step hierarchical multiple regression analyses to test the
relationship between fidelity and 8th grade end of year Core GPA, excluding students from the
smaller, special education classroom. As detailed in the top panel of Table S4.2, the simpler two-
step model excludes demographic covariates and showed a significant effect of fidelity (B=.026,
SE=.006, β=.227, p<.001, 95% CI [.015, .038]). As detailed in the bottom panel of Table S4.2,
the three-step model includes demographic covariates and also showed a significant effect of
fidelity (B=.022, SE=.005, β=.189, p<.001, 95% CI [.012, .032]).

Table S4.2
Effect of Student-level Fidelity on 8
th
Grade Core GPA, Excluding Smaller Special Education
Classroom
Predictor B SE β t p ∆R
2
p
Two Step Model

.461 .00
Step 1

7
th
Grade Final Core GPA 0.57 0.05 0.60 11.23 .00

Imputed fidelity measures -0.58 0.14 -0.24 -4.24 .00

Imputed 7th Grade Core GPA 0.20 0.19 0.06 1.07 .29

Step 2

.051 .00
Fidelity 0.03 0.01 0.23 4.46 .00

Three Step Model

Step 1

.461 .00
IDENTITY RELEVANCE ACROSS TIME

257
7
th
Grade Final Core GPA 0.57 0.05 0.60 11.23 .00

Imputed fidelity measures -0.58 0.14 -0.24 -4.24 .00

Imputed 7th Grade Core GPA 0.20 0.19 0.06 1.07 .29

Step 2

.136 .00
Hispanic 0.53 0.10 0.41 5.26 .00

Black -0.23 0.14 -0.12 -1.64 .10

Asian 0.03 0.11 0.02 0.25 .80

Multiracial-ethnic -0.30 0.31 -0.11 -0.97 .33

Free or reduced price lunch 0.12 0.09 0.07 1.46 .15

Female 0.07 0.04 0.09 1.73 .09

Step 3

.034 .00
Fidelity 0.02 0.01 0.19 4.15 .00

Note: Fidelity is computed

Section 5 – Additional analyses examining the effect of fidelity group on change in 8
th
grade
Core GPA

5.1 Analyses excluding students with imputed data

To test the effects of fidelity group, excluding students with imputed data, we first ran an
ANCOVA model without demographic covariates (F(2, 174)=14.80; p<.001, η
2
=.15) and then an
ANCOVA that did include these covariates (F(2, 168)=11.60; p<.001, η
2
=.12). In each model,
fidelity group was the independent variable and final 8
th
grade core GPA was the dependent
variable. In both ANCOVA analyses, final 7
th
grade core GPA was a covariate. In the latter
ANCOVA model, gender, free or reduced price lunch status, and Hispanic, Black, Asian, and
Multiracial-ethnic ethnicities were also included as covariates.
We followed up with three planned contrasts, contrasting threshold with midrange and
practical maximum; and practical maximum with midrange groups. Given multiple comparisons,
we applied Bonferroni adjustments to all p-values and confidence intervals. Being near the
IDENTITY RELEVANCE ACROSS TIME

258
practical maximum mattered. Practical maximum fidelity was associated with significantly
higher core GPA than mid-range fidelity (without demographic covariates (F(1, 174)=20.16,
p<.001, 95% CI [.208, .693], η
2
=.10); with demographic covariates (F(1, 168)=10.48, p<.01,
95% CI [.075, .518], η
2
=.06)), and near threshold fidelity (without demographic covariates (F(1,
174)=22.80, p<.001, 95% CI [.228, .697], η
2
=.12); with demographic covariates(F(1,
168)=21.68, p<.001, 95% CI [.190, .601], η
2
=.11). In contrast, the difference between core GPA
for students experiencing near threshold fidelity was not significantly different from students
experiencing fidelity in the midrange (without demographic covariates F(1,174)=0.01, p=1.00,
95% CI [-.237, .262], η
2
=.00; with demographic covariates F(1,168)=1.17, p=.84, 95% CI [-.123,
.321], η
2
=.01).
5.2 Analyses excluding students in smaller, special education classroom

To test the effects of fidelity group, excluding students from the smaller, special
education classroom, we first ran an ANCOVA model without demographic covariates (F(2,
191)=16.05; p<.001, η
2
=.14) and then an ANCOVA that did include these covariates (F(2,
185)=13.21; p<.001, η
2
=.13). In each model, fidelity group was the independent variable and
final 8
th
grade core GPA was the dependent variable. In both ANCOVA analyses, final 7
th
grade
core GPA and dummy variables for imputed data were covariates. In the latter ANCOVA model,
gender, free or reduced price lunch status, and Hispanic, Black, Asian, and Multiracial-ethnic
ethnicities were also included as covariates.
We followed up with three planned contrasts, contrasting threshold with midrange and
practical maximum; and practical maximum with midrange groups. Given multiple comparisons,
we applied Bonferroni adjustments to all p-values and confidence intervals. Being near the
practical maximum mattered. Practical maximum fidelity was associated with significantly
IDENTITY RELEVANCE ACROSS TIME

259
higher core GPA than mid-range fidelity (without demographic covariates (F(1, 191)=18.46,
p<.001, 95% CI [.187, .668], η
2
=.09); with demographic covariates (F(1, 185)=9.60, p<.01, 95%
CI [.064, .516], η
2
=.05)), and near threshold fidelity (without demographic covariates (F(1,
191)=28.33, p<.001, 95% CI [.282, .751], η
2
=.13); with demographic covariates(F(1,
185)=25.99, p<.001, 95% CI [.234, .656], η
2
=.12). In contrast, the difference between core GPA
for students experiencing near threshold fidelity was not significantly different from students
experiencing fidelity in the midrange (without demographic covariates F(1,191)=0.80, p=1.00,
95% CI [-.151, .329], η
2
=.00; with demographic covariates F(1,185)=2.89, p=.27, 95% CI [-.065,
.374], η
2
=.02).

Section 6 – Additional analyses examining the effect of fidelity on 8
th
grade course failure

6.1 Analyses excluding students with imputed data

We used 2-step (excluding demographics) hierarchical logistic regression analyses to test
the relationships between fidelity and 8th grade course failure, excluding students with imputed
data. We only report results of the 2-step models because when we attempted to include
demographic controls (our 3-step models), the maximum number of iterations was reached
before converging on a solution. As detailed in the top panel of Table S6.1, the two-step model
excludes demographic covariates and showed a significant effect of fidelity on course failure
(B=-.071, SE=.027, Wald=6.77, Exp(B)=.932, p<.01, 95% CI [.884, .983]).
Table S6.1
Effects of Implementation Fidelity on 8
th
Grade Course Failure: Excluding Students With
Imputed Data
Predictor B SE Exp(β) Wald p ∆R
2
p
Two Step Model

Step 1

.164 .00
IDENTITY RELEVANCE ACROSS TIME

260
7
th
Grade Course
Failure
2.06 0.40 7.87 27.21 .00

Step 2

.034 .01
Fidelity -0.07 0.03 0.93 6.77 .01

Note: Fidelity is computed at the student-level; ∆R
2
=change in Cox & Snell R-squared

6.2 Analyses excluding students from smaller, special education classroom

We used two-step and three-step hierarchical logistic regression analyses to test the
relationships between fidelity and 8th grade course failure, excluding students from the smaller,
special education classroom. As detailed in the top panel of Table S6.2, the simpler two-step
model excludes demographic covariates and showed a significant effect of fidelity (B=-.054,
SE=.025, Wald=4.57, Exp(B)=.948, p<.05, 95% CI [.902, .996]). As detailed in the bottom panel
of Table S6.2, the three-step model includes demographic covariates and also showed a
significant effect of fidelity (B=-.056, SE=.026, Wald=4.64, Exp(B)=.945, p<.05, 95% CI [.898,
.995]).

Table S6.2
Effects of Implementation Fidelity on 8
th
Grade Course Failure, Excluding Special Education
Classroom
Predictor B SE Exp(β) Wald p ∆R
2
p
Two Step Model

Step 1

.137 .00
7
th
Grade Course Failure 1.64 0.36 5.14 21.30 .00

Imputed fidelity measures 0.89 0.50 2.43 3.19 .07

Imputed 7th Grade Course
Failure
-0.17 0.70 0.85 0.06 .81

Step 2

.020 .03
Fidelity -0.05 0.03 0.95 4.57 .03

Three Step Model

Step 1

.137 .00
IDENTITY RELEVANCE ACROSS TIME

261
7
th
Grade Course Failure 1.64 0.36 5.14 21.30 .00

Imputed fidelity measures 0.89 0.50 2.43 3.19 .07

Imputed 7th Grade Course
Failure
-0.17 0.70 0.85 0.06 .81

Step 2

.056 .04
Hispanic 0.32 0.76 1.38 0.18 .67

Black 1.44 0.90 4.22 2.47 .11

Asian -0.18 0.88 0.84 0.04 .84

Multiracial-ethnic 1.57 1.66 4.79 0.89 .35

Free or reduced price lunch -1.41 0.77 0.24 3.40 .07

Female -0.72 0.37 0.49 3.67 .06

Step 3

.020 .03
Fidelity -0.06 -.03 0.95 4.64 .03

Note: Fidelity is computed at the student-level; ∆R
2
=change in Cox & Snell R-squared

Section 7 – Additional analyses examining the effect of fidelity group on 8
th
grade course
failure

7.1 Analyses excluding students with imputed data

We used 2-step (excluding demographics) hierarchical logistic regression analyses to test
the relationships between fidelity group and 8th grade course failure, excluding students with
imputed data. We only report results of the 2-step models because when we attempted to include
demographic controls (our 3-step models), the maximum number of iterations was reached
before converging on a solution. As can be seen in Table S7.1.1, being in the near-practical
maximum group rather than the near-threshold group predicted less likelihood of course failure
in the no demographic controls model, B = -1.53, SE = .489, Wald = 9.794, Exp(B) = .216, p <
.01, 95% CI [.083, .564]. Being in midrange group rather than the near-threshold group did not
predict less likelihood of course failure: B = -.267, SE = .476, Wald = .314, Exp(B) = .766, p
=.58, 95% CI [.301, 1.948]. As can be seen in Table S7.1.2, being in the near-practical maximum
IDENTITY RELEVANCE ACROSS TIME

262
group rather than the midrange was associated with less course failure in the no demographic
controls model, B = -1.264, SE = .543, Wald = 5.411, Exp(B) = .283, p = .02, 95% CI [.097,
.820].
Table S7.1.1
Effects of Implementation Fidelity Group on 8
th
Grade Course Failure, Compared to Near-
Threshold Group
Predictor B SE Exp(β) Wald p ∆R
2
p
Two Step Model
Step 1 .164 .00
7
th
Grade Course Failure 2.06 0.40 7.87 27.21 .00
Step 2 .053 .00
Midrange Group -0.27 0.48 0.77 0.31 .58
Near-practical maximum
Group
-1.53 0.49 0.22 9.79 .00
Note: Fidelity is computed at the student-level; ∆R
2
=change in Cox & Snell R-squared
Table S7.1.2
Effects of Implementation Fidelity Group on 8
th
Grade Course Failure, Compared to Midrange
Group
Predictor B SE Exp(β) Wald p ∆R
2
p
Two Step Model
Step 1 .164 .00
7
th
Grade Course Failure 2.06 0.40 7.87 27.21 .00
Step 2 .053 .00
Near-threshold Group 0.27 0.48 1.31 0.31 .58
Near-practical maximum
Group
-1.26 0.54 0.28 5.41 .02
Note: Fidelity is computed at the student-level; ∆R
2
=change in Cox & Snell R-squared
7.2 Analyses excluding students from smaller, special education classroom

We used 2-step (no demographic controls) and 3-step (with demographic controls)
hierarchical logistic regression analyses to test the relationships between fidelity group and 8th
grade course failure, excluding students from the special education classroom As can be seen in
IDENTITY RELEVANCE ACROSS TIME

263
Table S7.2.1, being in the near-practical maximum group rather than the near-threshold group
predicted less likelihood of course failure in both the 2-step and 3-step models: no demographic
controls model, B = -1.014, SE = .434, Wald = 5.455, Exp(B) = .363, p < .05, 95% CI [.155,
.849]; demographic controls model, B = -1.134, SE = .455, Wald = 6.22, Exp(B) = .322, p < .02,
95% CI [.132, .784]. Being in midrange group rather than the near-threshold group did not
predict less likelihood of course failure: no demographic controls model, B = -.470, SE = .425,
Wald = 1.22, Exp(B) = .625, p =.27, 95% CI [.272, 1.439]; demographic controls model, B = -
.434, SE = .460, Wald = .889, Exp(B) = .648, p =.35, 95% CI [.263, 1.596]. As can be seen in
Table S7.2.2, being in the near-practical maximum group rather than the midrange group did not
predict less course failure in either the 2-Step model or the 3-step model: no demographic
controls model, B = -0.545, SE = .472, Wald = 1.33, Exp(B) = .580, p = .25, 95% CI [.230,
1.463]; demographic controls model, B = -0.701, SE = .521, Wald = 1.81, Exp(B) = .496, p =
.18, 95% CI [.179, 1.377]).
Table S7.2.1
Effects Of Implementation Fidelity Group on 8
th
grade course failure, compared to near-
threshold group
Predictor B SE Exp(β) Wald p ∆R
2
p
Two Step Model
Step 1 .137 .00
7
th
Grade Course Failure 1.67 0.36 5.14 21.03 .00
Imputed fidelity measures 0.89 0.50 2.43 3.19 .07
Imputed 7th Grade Course
Failure
-0.17 0.70 0.81 0.06 .81
Step 2 .024 .06
Midrange Group -0.47 0.43 0.63 1.22 .27
Near-practical maximum
Group
-1.01 0.43 0.36 5.46 .02
Three Step Model
IDENTITY RELEVANCE ACROSS TIME

264
Step 1 .137 .00
7
th
Grade Course Failure 1.67 0.36 5.14 21.03 .00
Imputed fidelity measures 0.89 0.50 2.43 3.19 .07
Imputed 7th Grade Course
Failure
-0.17 0.70 0.81 0.06 .81
Step 2 .056 .04
Hispanic 0.32 0.76 1.38 0.18 .67
Black 1.44 0.90 4.22 2.57 .11
Asian -0.18 0.88 0.84 0.04 .84
Multiracial-ethnic 1.57 1.66 4.79 0.89 .35
Free or reduced price lunch -1.41 0.77 0.24 3.40 .07
Female -0.72 0.37 0.49 3.67 .06
Step 3 .026 .04
Midrange Group -0.43 0.46 0.65 .89 .35
Near-practical maximum
Group
-1.13 0.46 0.32 6.22 .01
Note: Fidelity is computed at the student-level; ∆R
2
=change in Cox & Snell R-squared

Table S7.2.2
Effects Of Implementation Fidelity Group on 8
th
grade course failure, compared to midrange
group
Predictor B SE Exp(β) Wald p ∆R
2
p
Two Step Model
Step 1 .137 .00
7
th
Grade Course Failure 1.67 0.36 5.14 21.03 .00
Imputed fidelity measures 0.89 0.50 2.43 3.19 .07
Imputed 7th Grade Course
Failure
-0.17 0.70 0.81 0.06 .81
Step 2 .024 .06
Near-threshold Group 0.47 0.43 1.60 1.22 .27
Near-practical maximum
Group
-0.55 0.47 0.58 1.33 .25
Three Step Model
IDENTITY RELEVANCE ACROSS TIME

265
Step 1 .137 .00
7
th
Grade Course Failure 1.67 0.36 5.14 21.03 .00
Imputed fidelity measures 0.89 0.50 2.43 3.19 .07
Imputed 7th Grade Course
Failure
-0.17 0.70 0.81 0.06 .81
Step 2 .056 .04
Hispanic 0.32 0.76 1.38 0.18 .67
Black 1.44 0.90 4.22 2.57 .11
Asian -0.18 0.88 0.84 0.04 .84
Multiracial-ethnic 1.57 1.66 4.79 0.88 .35
Free or reduced price lunch -1.41 0.77 0.24 3.40 .07
Female -0.72 0.37 0.49 3.67 .06
Step 3 .026 .04
Near-threshold Group 0.43 0.46 1.543 .89 .35
Near-practical maximum
Group
-0.70 0.52 0.50 1.81 .18
Note: Fidelity is computed at the student-level; ∆R
2
=change in Cox & Snell R-squared

IDENTITY RELEVANCE ACROSS TIME

266
Conclusion and Future Directions
Every single day, teachers, doctors, and financial advisors counsel students, patients, and
future retirees to imagine how their choices will affect who they want to be in the future. As
research on the influence of future “me” increasingly appears in media stories and professional
development workshops, the way that people imagine themselves in the future—or are guided to
imagine themselves in the future—will take on even more prominence. Accordingly, it is of
growing importance that researchers and practitioners studying the future “me” focus their
efforts in the right areas. The question then, is where should the field go from here?
Most broadly, it may be useful for the field to move away from framing the big questions
as pertaining to whether or not a particular future “me” aspect or characteristic is present (e.g., is
there connection? is there a gap?, etc.). This type of focus draws attention away from the context
as a whole and can lead to important situational features being overlooked. Instead, researchers
should begin by asking whether a future “me” is likely to be experienced as relevant and whether
actions to attain it will feel congruent with one’s present identity in the moment. This perspective
does not rule out further research on factors like psychological connection, but by situating them
in a broader, identity-based motivation framework, research will paint a clearer picture of the
underlying process and potential moderators.
One a more fine-grained level, the work in this dissertation highlights a few key points
about future work involving future “me.” As detailed next, this includes directions for theoretical
research and important guidelines for intervention design.
Future Theoretical Research
From a theoretical standpoint, there is still scant empirical evidence about how one’s
interpretation of experienced difficulty influences the relevance of possible identities. To date,
IDENTITY RELEVANCE ACROSS TIME

267
research has not conclusively demonstrated that a difficulty-as-importance mindset can increase
the relevance of possible identities or that a difficulty-as-impossibility mindset can decrease the
relevance of possible identities. Understanding this link is critical given recent work linking
interpretation of difficulty to academic achievement (Oyserman, Elmore, Novin, Fisher, &
Smith, 2018) and social class (Fisher & Oyserman, under review).
In a series of studies, I attempted to examine these relationships, but results were
inconclusive, perhaps because of methodological difficulties related to priming difficulty
mindsets in the context of a future self. Consider what happens when a participant receives
instructions that aim to bring to mind the idea that difficulty with saving money for the future is a
sign that such actions are impossible. Participants often think of moments in which they were
forced to spend money—for example, because an urgent problem with a car or plumbing arose
and needed fixing. However, instead of reinforcing the idea that saving money is impossible,
bringing these difficult experiences to mind often has the opposite effect because they remind
people how important it is to have money squirreled away for emergencies. This cues the idea
that saving money is important, and when subsequently presented with a dependent variable
related to financial decision making, participants then make a more future-oriented financial
choice. A similar process can occur in the domain of health, although perhaps less frequently.
Thus, reminding participants of times when it was impossible to invest in a healthy future self—
due to not having enough money or time to exercise or buy healthy food, for example—may cue
the idea that such investment is important. The result is more motivation to make the healthy,
future-oriented choice when presented with a health-related dependent variable. As these
examples highlight, there are clear limitations in our capacity to learn about underlying processes
from online survey experiments reliant on self-reports about hypothetical decisions.
IDENTITY RELEVANCE ACROSS TIME

268
Designing experiments that can avoid these types of boomerang effects when dependent
variables are presented in novel, constraint-free contexts is a challenge for future research. A
solution requires new independent and dependent variables that create a context in which
difficulty truly signals impossibility, not merely impossibility in one moment that can be
mitigated by an increased sense of importance in another moment. Doing so may require
situating participants in more ecologically valid scenarios where the experience of difficulty is
more proximal, rather than in the laboratory or an online survey. For example, difficulty mindset
primes have been found to influence academic performance when students subsequently sit down
to take an actual exam (Oyserman et al., 2018). In this context a difficulty-as-impossibility
mindset is more likely to carry over to the task, though in these contexts it may be more difficult
to adequately bring a future “me” to mind.
Future research should also explore a key implication of the model outlined in Chapter
1—that seemingly meaningful factors like psychological connection between present and future
may fail to influence choices if other features of the context imply that future “me” is irrelevant
to judgments or decisions. This research would provide support for our proposed model and
paint a clearer picture of the boundary conditions for when future “me” characteristics like
psychological connection are likely to matter. This work would entail examining how the effect
of certain aspects of future “me”—psychological connection, for example—are moderated by
other factors likely to influence the relevance of future “me.” Research showing that it is not
connection per se, but relevance, would provide strong support for an identity-based motivation
model. These kinds of studies would also help shift the focus of the field from specific
characteristics of the future “me” to the broader situational context in which people are thinking
about future identities.
IDENTITY RELEVANCE ACROSS TIME

269
Applied Implications
Interventions to change outcomes in the real world are complex vehicles that are
vulnerable and may break down. One aim of this dissertation is to highlight some of moving
parts that are often overlooked but can determine an intervention’s success. My findings point to
three important guidelines for interventions designers.
First, a one-size-fits-all approach to changing school-focused possible identities may be
unwise because different populations may undergo different developmental changes at different
times with regard to how they think about the future. There may also be different levels of
heterogeneity within a given population. In addition, aspects of the social context, such as
poverty, may have different effects in different populations. Accordingly, researchers should
attempt to understand how their population of interest tends to think about their future “me” and
their strategies outside the context of intervention, as this can inform key design decisions. For
example, in high-resource contexts, strategies for attaining possible identities may be provided
(e.g. help from family members, after-school programs, etc.), but in low-resource context
students may need to learn to generate their own strategies.
Second, researchers should commit resources toward understanding exactly what is going
on during intervention implementation. In my work, I used a detailed, labor-intensive measure of
fidelity, and found that fidelity mattered. Though it may not always be possible to have multiple
observers quantifying what happened, the takeaway from both my research and my review of the
literature is that interventions are more likely to err on the side of not dedicating enough
resources toward quantifying fidelity. It may also be tempting to think that fidelity matters only
when the intervention revolves around an abstract psychological construct. That is, that course-
correcting is harder when students are led astray while building something personal and
IDENTITY RELEVANCE ACROSS TIME

270
subjective—for example, a psychological roadmap to a future identity—but easier when they are
doing something objective (e.g., solving algebra problems). In this latter case there are
conceivable other ways to make up for poor implementation (e.g. asking classmates for help;
looking in a textbook), but less so in the former case. However, this does not appear to be case—
all kinds of interventions are likely to suffer in the absence of sufficient fidelity.
Third, in addition to actually measuring changes in possible selves and linked strategies,
interventions involving a future “me” should also attempt to develop and use new methods to
measure relevance as a potential mediator. To date, the two most commonly used metrics that
can serve as a proxy for relevance are psychological connection and strategies linked to possible
identities. However, as previously discussed, connection is not the only factor that contributes to
relevance, and a future “me” may be experienced as irrelevant even when there is connection.
Similarly, it may not be feasible to collect data on possible identities and strategies, or it may not
be relevant in a given intervention—for example, if the intervention is focused on the abstract
future self rather than a specific possible identity. In these situations, having generalizable
metrics for measuring the relevance of future “me” can help determine whether an intervention
fails because it doesn’t make a future “me” feel relevant, or for some other reason.
Concluding Remarks
The results presented in this dissertation highlight how shifting what people think is
possible in future has small to moderate effects on real world outcomes. These small effects are
meaningful when considered in the context of how much these outcomes—in this case, core
GPA—generally change (not a lot), and how much these outcomes are influenced by
demographics. It is clear that the future “me” is a useful tool for changing social outcomes, and
that many obstacles for scaling interventions based on changing one’s future “me” can be
IDENTITY RELEVANCE ACROSS TIME

271
overcome. Researchers should continue their efforts to answer important questions about how
people think about the future because these answers are likely to have an impact on people’s
lives.

IDENTITY RELEVANCE ACROSS TIME

272
References
Fisher, O., & Oyserman, D. (under review). When the going gets tough, does social class matter?
Income and subjective social status but not education predict difficulty mindsets.
Oyserman, D., Elmore, K., Novin, S., Fisher, O., & Smith, G. C. (2018). Guiding people to
interpret their experienced difficulty as importance highlights their academic possibilities
and improves their academic performance. Frontiers in psychology, 781(9), 1-15.

Abstract (if available)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Bridging possible identities and intervention journeys: two process-oriented accounts that advance theory and application of psychological science

PDF

Creating a pathway to success: teachers can change their students’ identity-based motivation

PDF

People can change when you want them to: changes in identity-based motivation affect student and teacher Pathways experience

PDF

Difficulty-as-sanctifying: when difficulties build character, purify the self, and elevate the soul

PDF

Culture's consequences: a situated account

PDF

Difficulty-as-importance and difficulty-as-impossibility: unpacking the context-sensitivity and consequences of identity-based inferences from difficulty

PDF

Classrooms are game settings: learning through and with play

PDF

Can I make the time or is time running out? That depends in part on how I think about difficulty

PDF

The interaction of teacher knowledge and motivation with organizational influences on the implementation of a hybrid reading intervention model taught in elementary grades

PDF

Teacher perception on positive behavior interventions and supports’ (PBIS) cultivation for positive teacher-student relationships in high schools: an evaluation study

PDF

How perceived moral congruence shapes propensities to engage in pro-group behaviors

PDF

Bridging the empathy gap: a mixed-method approach to evaluating teacher support in bullying prevention and intervention at an urban middle school in India

PDF

Embedded academic support for high school student success: an innovation study

PDF

Affirmation and majority students: Can affirmation impair math performance?

PDF

The knowledge, motivation and organizational influences that shape a special education teacher’s ability to provide effective specialized academic instruction

PDF

Examining internal communication practices to improve teachers’ motivation to use ICTs in instruction: a case study of Peruvian secondary schools

PDF

Building data use capacity through school leaders: an evaluation study

PDF

Key leverage points for improving educational outcomes of black male students: Response to intervention as an effective strategy to support students in general education: An evaluation study

PDF

Gender beyond the binary: transgender student success and the role of faculty

PDF

The intersection of teacher knowledge and motivation with organizational culture on implementing positive psychology interventions through a character education curriculum

Asset Metadata

Creator Horowitz, Eric (author)

Core Title A roadmap for changing student roadmaps: designing interventions that use future “me” to change academic outcomes

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Psychology

Publication Date 07/27/2020

Defense Date 06/05/2019

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag Academic Achievement,future self,identity,intervention,Motivation,OAI-PMH Harvest,possible identities

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Oyserman, Daphna (committee chair), Dehghani, Morteza (committee member), Monterrosso, John (committee member), Schwarz, Norbert (committee member), Townsend, Sarah (committee member)

Creator Email ejhorowi@usc.edu,ericjhoro@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-199179

Unique identifier UC11663048

Identifier etd-HorowitzEr-7676.pdf (filename),usctheses-c89-199179 (legacy record id)

Legacy Identifier etd-HorowitzEr-7676.pdf

Dmrecord 199179

Document Type Dissertation

Format application/pdf (imt)

Rights Horowitz, Eric

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA