University of Southern California Dissertations and Theses

Personalized normative feedback applied to undergraduates with problem drinking: a comparison with psychoeducation and an examination of cognitive-affective change mechanisms via the articulated ...

(USC Thesis Other)

Personalized normative feedback applied to undergraduates with problem drinking: a comparison with psychoeducation and an examination of cognitive-affective change mechanisms via the articulated ...

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content ATSS AND PERSONALIZED NORMATIVE FEEDBACK 1

Personalized Normative Feedback Applied to Undergraduates with Problem Drinking: A
Comparison with Psychoeducation and An Examination of Cognitive-Affective Change
Mechanisms via the Articulated Thoughts in Simulated Situations Think-Aloud Paradigm

by
Justin Francis Hummer, M.A.
______________________________________________________________________________

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(PSYCHOLOGY)
August 2019

Committee:
Chair: Gerald Davison, Ph.D., Professor of Psychology and Gerontology
Darby Saxbe, Ph.D., Associate Professor of Psychology

David Walsh, Ph.D., Associate Professor of Psychology

Adam Leventhal, Ph.D., Professor of Preventive Medicine and Psychology

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 2
Acknowledgements
I humbly stand on the backs of giants. Most notably, my academic father, Jerry Davison, whose
invaluable advice and direction at every stage of the graduate school enterprise I am most
pleased to acknowledge. Beyond this study, your support over the past seven years has fostered
my professional and personal development in ways that I could not have imagined.

Sincere thanks are rendered to my advisory committee for their support during the development
and execution of this research idea. I am also grateful for the institutional support provided by
Paula Lee Swinford in the USC Office of Health and Wellness Promotion and Donna Turner in
the USC Office of Student Judicial Affairs and Community Standards.

I would like to express my deepest and most heartfelt gratitude to Joseph LaBrie. I am forever
grateful for the opportunity you provided to a bright eyed undergraduate. You took my nascent
curiosity of scientific research and skillfully molded it into a productive pursuit of creativity and
knowledge. Your unwavering confidence in my potential propelled me to uncover a world that
has rendered untold riches, discovery, and fulfillment. Thank you for your tremendous
mentorship and most importantly, your friendship.

A deep bow is warranted to my closest friends, who have displayed remarkable loyalty. You
taught me that boundaries of expression and achievement are but glass manifestations of our own
construction, meant to be shattered and rebuilt into limitless forms. I’ll see you at the top…

Finally, I would like to thank my family for their encouragement and support. To Daddy’s little
girl, Arya, you are the apple of my eye. Thank you for teaching me the true meaning of patience
and love. It is my untold daily pleasure to watch your warrior spirit soar to ever new heights. I
would like to thank especially my mother, father, and my wife, Robyn. Thank you, Mom and
Dad, for always challenging me to be the best version of myself and sacrificing to give me the
priceless gifts of fortitude, integrity, and perseverance. Your belief in me helped me to believe in
myself. Thank you, Robyn, my moon and stars, for being my ultimate companion, a source of
unyielding love, a wellspring of deep wisdom, and a goddess of timeless quality.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 3
Table of Contents
List of Tables 5
List of Figures 6
List of Appendices 7
Abstract 8
Chapter 1: Introduction 10
Adjudicated Students 10
Social Norms and College Student Drinking 12
Social Norms Alcohol Interventions for College Students 12
Exploring Change Mechanisms of PNF Intervention Efficacy 15
A Novel Methodological Approach: Articulated Thoughts in
Simulated Situations 19
ATSS and Depth of Processing 24
MA Thesis Extension 27
Reference Group Specificity 30
Moderators of Intervention Efficacy 31
Objectives of the Current Study 35
Specific Aims and Hypotheses 37
Chapter 2: Method 38
Participants and Recruitment 38
Procedures 41
Measures 45
ATSS Coding Strategy 49
Interrater Reliability for Codes 52
Chapter 3: Results 53
Analytic Plan 53
Missing Data, Outliers, and Assumptions for the General Linear Model 58
Preliminary Analyses 60
Correlations 62
Misperceptions: Perceived Peer Drinking vs. Actual Student Drinking 63
AIM I - Intervention Main Effects 64
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 4
AIM I - Regression Models Evaluating Intervention Efficacy with Contrasts 65
AIM II – Exploring Mechanisms of Change 70
Evaluating Explanatory Mechanism of ATSS Codes on Drinking 72
Evaluating Explanatory Mechanism of ATSS Codes on Perceived Norms 75
Evaluating Indirect Effect of Changes in Perceived Norms on Drinking 79
AIM III - Moderation Analyses 80
Evaluating Moderators of Intervention Effects on Drinking 80
Evaluating Moderators of Intervention Effects on Perceived Drinking Norms 82
Chapter 4: Discussion 85
ATSS and Mechanisms of Change 91
Moderating Effects 98
Limitations 99
Strengths and Future Directions 101
Concluding Remarks 103
References 104

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 5
List of Tables

Table 1. Descriptive Statistics for Demographics Characteristics and Primary Outcomes
by Study Condition 40
Table 2. Interrater Reliability (ICCs) Based on Absolute Agreement Criterion 52
Table 3. Interrater Reliability (ICCs) Based on Consistency Agreement Criterion 53
Table 4. Correlation Matrix of Variables for Males (n = 34) and Females (n = 36) 63
Table 5. Students' Perceptions of Typical Same-Sex USC Student Compared to
Students' Actual Drinking from Campus-Wide Survey 64
Table 6. Hierarchical Regression Analyses for Three Primary Outcomes Using
Contrast Coding 69
Table 7. Correlation Matrix of Outcome Changes and Coding Categories for
Control (n = 14) and PNF-ATSS (n = 31) 71
Table 8. Mean Differences of Coding Categories as a Function of Intervention
Condition 72

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 6
List of Figures

Figure 1. Conceptual model for mediation 56
Figure 2. Weekly drinking means pre- and post-intervention by condition 66
Figure 3. Perceived weekly drinking (DNRF) means pre- and post-intervention
by condition 67
Figure 4. Alcohol-related consequences means pre- and post-intervention by condition 68
Figure 5. Supported indirect effects model for follow/neutral on drinks per week
with regression coefficients and standard errors for all paths including the
(total effect) of X on Y 74
Figure 6. Supported indirect effects model of follow/neutral on perceived norms
with regression coefficients and standard errors for all paths including the
(total effect) of X on Y 77
Figure 7. Supported indirect effects model of believability on perceived norms
with regression coefficients and standard errors for all paths including the
(total effect) of X on Y 78
Figure 8. Autonomous orientation as a moderator of intervention efficacy on
perceived weekly drinking norms 85
Figure 9a. Histogram of normalized total drinks per week outcome variable 121
Figure 9b. P-P plot for normalized total drinks per week outcome variable 122
Figure 9c. Scatterplot plot for normalized total drinks per week outcome variable 123
Figure 10a. Histogram of normalized perceived weekly drinking outcome variable 124
Figure 10b. P-P plot of normalized perceived weekly drinking outcome variable 125
Figure 10c. Scatterplot plot for normalized perceived weekly drinking outcome variable 126
Figure 11a. Histogram of normalized alcohol-consequences outcome variable 127
Figure 11b. P-P plot of normalized alcohol-consequences outcome variable 128
Figure 11c. Scatterplot plot for normalized alcohol-consequences outcome variable 129
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 7
List of Appendices
Appendix A. Example Feedback Slides 130
Appendix B. PNF Intervention Script 133
Appendix C. PNF Intervention Script: ATSS Condition 134
Appendix D. Control Intervention Script and Content 136
Appendix E. Pre-Intervention (Baseline) Survey 140
Appendix F. One-Month Post-Intervention Survey 147
Appendix G. ATSS Coding Manual 151
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 8
Abstract

Objective: Problematic alcohol use among college students is a national health concern, with
students receiving sanctions for violating campus alcohol policy (adjudicated students) identified
as a particularly high-risk group. Personalized normative feedback (PNF) is a brief intervention
approach designed to correct normative misperceptions of peer drinking behavior which in turn
has been shown to reduce alcohol use. Employing a randomized longitudinal intervention design,
the present study sought to reduce individual drinking and alcohol-related consequences among
adjudicated undergraduates by reducing misperceived peer drinking norms through the provision
of a PNF intervention. The research explored explanatory mechanisms of change through an
adapted application of the Articulated Thoughts in Simulated Situation (ATSS) cognitive think-
aloud paradigm. Seven coding categories emerged and indirect effects of these codes on
intervention efficacy were considered. Questions about which subgroups of individuals benefit
most (or least) from the PNF intervention were investigated through analyses of moderation. The
chosen moderators are important on a theoretical level (i.e., group identity, controlled vs.
autonomous personality orientations) as well as pragmatically to adjudicated students (i.e., pre-
intervention defensiveness). Method: A sample of 70 (51% female) undergraduate students were
randomly assigned to one of three conditions: a PNF-ATSS condition, a PNF-Only condition
(without ATSS), and an active Control+ATSS condition which received psychoeducation about
alcohol use. Participants completed baseline and one-month post-intervention questionnaires.
Intervention content was delivered on a lab computer, with audio/visual synced components, and
with think-aloud segments in the two conditions that entailed ATSS. Think-aloud data were
recorded, transcribed, and content-analyzed. The seven cognitive-affective coding categories
were as follows: Sustain talk, skepticism, follow/neutral, believability, reflective analysis,
positive surprise and negative surprise. Results: The General Linear Model was used to answer
the specific aims of the study. Students in both the PNF and PNF-ATSS conditions reported
significant reductions in their misperceived peer drinking norms and alcohol-related
consequences at the 30-day follow-up, relative to students in the control condition, who were not
found to have significantly reduced their normative misperceptions or alcohol consequences.
These main effects were present whether the PNF conditions were evaluated independently or
collectively against the control. Moreover, there were no observed differences between the two
PNF conditions in magnitude or direction of the effects. With respect to self-reported alcohol
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 9
use, participants in the PNF-ATSS condition drank significantly fewer drinks per week at follow-
up than participants in the PNF-Only condition, but not less than participants in the control
condition. No differences were found between the control and PNF-Only conditions. There were
significant indirect effects from the intervention to drinking and perceived norms outcomes via
the follow/neutral code. Being in the PNF-ATSS condition was associated with lower levels of
neutrality regarding the intervention content, and lower levels of neutrality, in turn led to lower
drinking and lower perceived norms at follow-up. A significant indirect effect also emerged from
the intervention to perceived norms via believability. Being in the PNF-ATSS condition was
associated with lower levels of believability regarding the intervention content. However, the
greater the level of believability regarding the content, the lower perceived norms were at
follow-up. Lastly, autonomous orientation was found to moderate intervention efficacy such that
the intervention effect on perceived weekly drinking norms was stronger for PNF-Only
participants compared to Control, as level of autonomous orientation decreased. Conclusion: In
sum, the research provides important theoretical and practical contributions to the social norms
and alcohol intervention literature. Skepticism, believability, and neutrality remain issues
needing to be addressed if PNF is to be strengthened as an intervention component.
Understanding the key mechanisms by which PNF interventions work, why, and for whom, is
imperative for the evolution of norms-based intervention strategies and theory development.

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 10
CHAPTER 1: INTRODUCTION
Problematic drinking by college students is a national health concern. National data
indicate most young adults drink alcohol, with about one-third of young people between 18 and
25 engaging in heavy drinking (five or more drinks in a row in the past two weeks) and about
one in 10 young adults reporting consumption of 10 or more drinks in a row during just the past
two weeks (Schulenberg et al., 2018). The consequences of heavy drinking by young adults are
well documented and include academic problems, physical injuries and fights, risky sexual
behavior and sexual assaults, memory blackouts and passing out, sustained cognitive deficits,
alcohol poisoning, and even death (Welch, Carson, & Lawrie, 2013; White & Hingson, 2014;
Windle, 2016). Aside from the individual drinker, secondary consequences are imposed on
fellow students and the larger community, including physical aggression, disrupted
sleep/studying, sexual assault, and property damage, to name a few (Hingson, Heeren, Winter, &
Wechsler, 2005; Hingson, Zha, & Weitzman, 2009). Researchers and college student personnel
have designed and implemented a variety of individually focused and group-based interventions
to reduce alcohol harm (Larimer & Cronce, 2007). Despite significant resources and efforts
dedicated to this prevention goal, problematic alcohol use in college remains widespread.
Adjudicated Students
Adjudicated students are those who receive sanctions for violating the alcohol use policy
at their college or university. Tens of thousands of college students violate campus alcohol
policies and receive mandatory alcohol interventions each year (Porter, 2006). These students are
a particularly high-risk group and have more problems than the general student population as
they tend to have lower GPAs, consume more alcohol, and have a greater number of resulting
alcohol-related problems compared to their non-adjudicated student peers (Fromme & Corbin,
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 11
2004). It is worrisome then, that adjudicated students do not see their behavior as problematic
and are less prone to seek help for these issues (NIAAA, 2007). For example, adjudicated men
differ from other students in that they have greater intentions to drink alcohol and are less
concerned with their health, putting them at greater risk for alcohol-related problems (LaBrie,
Tawelbeh, & Earleywine, 2006).
Most if not all colleges have some policy prohibiting drinking in their under-age student
body as well as consequences for its violation that may include monetary fines, informing
parents, and mandated activities (Anderson & Gadaleto, 2006), but recidivism rates and risk
levels remain high (Hingson, 2010). In fact, there have been consistent increases in the number
of alcohol-related arrests, the number of students receiving alcohol citations, and the proportion
of students mandated to participate in a post-citation intervention on college campuses (Doumas,
McKinley, & Book, 2009; Porter, 2006; The National Center on Addiction and Substance Abuse,
2007), highlighting the need to focus on the development, effectiveness, and sustainability of
mandated intervention programs. When students become adjudicated, an opportunity exists to
intervene and prevent future alcohol abuse and associated negative or unwanted outcomes.
Brief Motivational Interventions (BMIs) are currently the individual intervention with the
strongest empirical support for use with mandated students (Carey, 2012). BMIs are often
delivered in one to two individual face-to-face meetings that are approximately 50 minutes long
(Carey, Scott-Sheldon, Carey, & DeMartini, 2007), use motivational interviewing (MI; Miller &
Rollnick, 2013) as the counseling approach, and often include personalized feedback to promote
less risky drinking. The growing body of literature on the efficacy of interventions with this
population (Larimer & Cronce, 2007; Mastroleo, Murphy, Colby, Monti, & Barnett, 2011)
reflects the need to identify key components that can be experimentally manipulated or that are
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 12
particularly educational and informative for use in the design of novel and effective
interventions. This study entails an intervention trial for adjudicated students with a focus on
mechanistic and intervening factors that may shed theoretical light and strengthen or weaken the
efficacy of a widely used intervention approach to curb college student drinking.
Social Norms and College Student Drinking
Social influences impact a variety of behavioral domains including heavy drinking
among college students (Borsari & Carey, 2003; Perkins, 2002). Student perceptions of the
prevalence of peer drinking (i.e., descriptive norms) factor heavily into personal decisions about
when and how much to drink (e.g., Neighbors, Lee, Lewis, Fossos, & Larimer, 2007; for reviews
and meta-analyses see Berkowitz, 2004, Borsari & Carey, 2003; Perkins, 2002). Epidemiological
research from 130 universities and colleges nationwide has shown that the vast majority of
college students overestimate how much and how frequently other students drink, and that those
perceptions are the strongest predictor of their own individual drinking (Perkins, Haines, & Rice,
2005). These perceived norms account for more variance in alcohol use than sex, Greek letter
organization membership (i.e., fraternities/sororities), alcohol expectancies, or drinking motives
(Neighbors et al., 2007). This strong social influence factor presents an ideal target for
intervention strategies seeking to reduce problematic drinking.
Social Norms Alcohol Interventions for College Students
Social norms-based interventions have been widely adopted over the past two decades at
higher education institutions as a means to correct students’ overestimations regarding the
prevalence of heavy drinking among fellow students (Larimer & Cronce, 2007). Discrepancies
between perceived and actual norms are consistently associated with alcohol use with larger
discrepancies related to higher rates of alcohol use (Larimer, Turner, Mallett, & Geisner, 2004;
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 13
Lewis & Neighbors, 2004; Reis & Riley, 2000). Because perceived norms have consistently been
shown to directly influence behavior (Rimal & Real, 2003; 2005), social norms interventions
seek to highlight the actual lower prevalence of drinking compared to most students’ perceived
norms.
An early form of social norms interventions was called social norms marketing. This
approach relied on mass communication methods for educating students regarding actual
drinking behaviors utilizing national or campus-specific drinking statistics, yet evidence for the
effectiveness of this approach was mixed at best (e.g., Perkins & Craig, 2006). Social norms
marketing efforts were hampered by questions of requisite attention and processing of the
messages by college students (DeJong et al., 2006; DeJong et al., 2009; Perkins, LinkenBach,
Lewis, & Neighbors, 2010). It was difficult to ascertain the degree to which students actually
viewed and processed the messages, and thus, whether failure to achieve desired outcomes was
related to a failure to reduce normative perceptions.
Personalized normative feedback. In contrast to mass marketing methods, personalized
normative feedback (PNF) is one-time individually delivered information designed to correct
normative misperceptions. Based on data demonstrating the strong association between
perceived descriptive norms and alcohol use in college populations, correction of normative
misperceptions using PNF is a prominent focus of many college drinking intervention studies
(for reviews see Carey, Scott-Sheldon, Carey, & DeMartini, 2007; Cronce & Larimer, 2011;
Lewis & Neighbors, 2006; Miller et al., 2013; Walters & Neighbors, 2005). PNF interventions
typically provide graphical and text-based feedback contrasting three pieces of information: (1) a
student’s own self-reported drinking; (2) a student’s perception of other students’ drinking; and
(3) actual drinking rates for a typical student on the same campus. When provided to heavier
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 14
drinking participants, PNF is designed to highlight two pieces of information regarding
normative beliefs known to influence drinking behavior, namely: (1) other students drink less
than the participant drinks (social comparison information provided the individual’s alcohol use
is heavier than the prevailing norm), and (2) other students drink less than the participant thinks
they drink (normative misperception correction). The idea is to correct normative misperceptions
by showing discrepancies between actual norms and students’ perceptions and behaviors, that is,
by informing participants that others like them drink less than what they believe them to drink.
Correcting normative misperceptions, the theory suggests, motivates behavior change because of
young people’s desire to conform to social norms (Rice, 2007).
Individual PNF is more effective than social norms marketing campaigns (Lewis &
Neighbors, 2006). PNF is also often used as part of several widespread, multicomponent alcohol
interventions that target adjudicated and other high-risk students (Walters & Neighbors, 2005),
such as Brief Alcohol Screening and Intervention for College Students (BASICS) (Dimeff, Baer,
Kivlahan, & Marlatt 1999), AlcoholEdu
®
(Outside the Classroom, 2010), and Heads Up (LaBrie,
2010). To date, both stand-alone and multicomponent computerized and Web-based
interventions that incorporate PNF have been found to reduce alcohol use in randomized clinical
trials, with effect sizes typically in the small range (e.g., Doumas, Haustveit, & Coll, 2010;
LaBrie et al., 2013; Lewis & Neighbors, 2007; Martens, Smith, & Murphy, 2013; Neighbors,
Larimer, & Lewis, 2004; Neighbors, Lewis, et al., 2010). In light of the relatively strong
evidence of the efficacy of computerized and Web-based PNF as an intervention strategy for
reducing college student drinking, the trend toward this approach replacing traditional SNM
approaches which are implemented on a more widespread basis, and the growth potential to
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 15
achieve stronger and longer-term outcomes, continued research seeking to better understand and
enhance PNF is warranted. The present study represents an effort to do so.
Exploring Change Mechanisms of PNF Intervention Efficacy

Despite the emerging evidence supporting the use of PNF, less is known about the
mechanisms driving its efficacy. It is unclear exactly which elements of PNF interventions are
responsible for the reductions in college student drinking that have previously been observed. It
is also unclear how the various elements are processed cognitively by those receiving them and
how those reactions function to produce behavior change. Change mechanisms are causal links
between treatment and outcome. Exploring possible mechanisms of change within PNF
interventions is important for several reasons. Identifying such mechanisms will clarify the
connections between what is done (PNF) and the diverse outcomes (e.g., reducing some drinking
behaviors but not others, typically unable to reduce composite measures of alcohol
consequences, mixed findings on most effective reference group to use in the feedback). Second,
by understanding the processes that account for change one ought to be better able to optimize
the effects of PNF. Indeed, without understanding what is critical to PNF and how it operates
(what leads to change and why), we are at a bit of a loss for how it might be improved. If we
know how changes come about, perhaps we can direct better, stronger, different, or more
strategies that trigger the critical change process(es). Third, more reliably extending treatments
from research to applied settings will be difficult without understanding how treatment works.
To optimize the generality of PNF effects to college settings and beyond, it is important to know
what the active ingredients are and what components must not be diluted (or should be
strengthened) to achieve change. Fourth, a better understanding of how PNF works can help
identify new moderators of intervention efficacy, such as the degree of resistance to the
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 16
intervention and personality characteristics that may differentially relate to outcomes.
Understanding the processes through which PNF operates can help sort through those facets that
might be particularly influential in intervention outcomes and permit better targeting of intended
recipients.
The current study seeks to evaluate several potential mechanisms of change, which may
occur and will be measured during treatment. To date, existing research has evaluated the
presence of only one mediator in PNF trials, which follows the theoretical foundation of the
approach; namely, that the impact of normative feedback on drinking is due to its impact on
perceived norms. Consistent with this hypothesized mechanism, several studies comparing PNF
to appropriate control groups have indeed shown that changes in perceived norms are associated
with changes in drinking. Neighbors, Lewis, Bergstrom, and Larimer (2006) conducted PNF in a
lab setting using the “typical student” referent and found mediational support for reductions in
perceived norms leading to changes in drinking. Lewis and Neighbors (2007) also conducted
PNF in a lab setting, but with same-sex norms as the reference group in the feedback, and they
found that changes in norms partially mediated intervention efficacy, but only for women. In a
large-scale study by LaBrie and colleagues (2013) comparing ‘typical student’ PNF to more
specific reference group feedback (e.g., a ‘typical male, Greek, student’), results indicated that
typical student PNF was associated with the greatest changes in typical weekly drinking, in part,
through an indirect effect on descriptive normative perceptions.
Of specific interest to the current study are two PNF intervention studies targeting
adjudicated students. Doumas and colleagues (2009) used multicomponent Web-based PNF with
typical student feedback and compared it to a Web-based alcohol education program. Although
both groups reported a reduction in drinking, mandated students who received Web-based PNF
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 17
reported greater reductions in weekly drinking, peak alcohol consumption, and frequency of
drinking to intoxication than the comparison group. Contrary to hypotheses, neither condition led
to a decrease in alcohol-related consequences. In addition, this was the first study to examine the
change in estimates of peer drinking as a mediator in the relationship between the intervention
effects and changes in drinking for mandated students. The effects of the intervention were
indeed accounted for by the changes in estimates of peer drinking, although it should be noted
that estimates of peer drinking and one's own drinking were reported at the same time rather than
temporally sequenced. Moreover, the PNF included additional alcohol intervention information,
such as approximate financial cost of drinking in the past year, calories associated with drinking,
how quickly the body processes alcohol, risk status for negative consequences associated with
drinking, and risk status for problematic drinking based on the participant's Alcohol Use
Disorders Identification Test (AUDIT) score.
Similarly, in a follow up study, Doumas, Workman, Smith, and Navarro (2011) targeted
adjudicated student drinking and compared multicomponent personalized feedback delivered by
a counselor to feedback delivered online. Where applicable, both interventions used non gender-
specific (i.e., typical student) norms in feedback. Results indicated that students in the counselor-
delivered feedback condition reported greater reductions in weekly drinking quantity and binge
drinking frequency than those in the Web-delivered group at follow up. As in their previous
study, changes in estimates of typical college student drinking from baseline to the follow up
partially mediated the effect of the intervention on changes in drinking. However, also like their
previous study, individualized graphed feedback was provided in the following domains:
Summary of quantity and frequency of drinking including graphical feedback such as the number
of cheeseburgers that are equivalent to alcohol calories consumed, graphical comparison of one's
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 18
own drinking to U.S. adult and college drinking norms, estimated risk status for negative
consequences associated with drinking and risk status for problematic drinking based on the
participant's AUDIT score, genetic risk, tolerance, approximate financial cost of drinking in the
past year, normative feedback comparing one's perception of peer drinking to actual university
drinking normative data, and referral information for local agencies. Again, it must be noted that
the estimates of peer drinking and one's own drinking used in the mediation analyses were
reported at the same time, not the ideal method for establishing true mediation over time.
The studies described above provide good support for the theoretically-derived
mechanism of change in PNF interventions, yet they are still limited by the focus on only one
change agent, which did not capture cognitive elements of participants’ reactions. In addition, the
studies on the efficacy of PNF among adjudicated students were done in the context of Web-
based multicomponent interventions that included PNF, rather than the traditional stand-alone
PNF, making it difficult to ascertain which components were most responsible for reductions in
drinking. They also assessed the mediator and outcomes concurrently, which deviates from the
recommended guidelines for establishing mediation. As with other PNF interventions, the studies
found no effects of the intervention on alcohol consequences.
Excessive drinking is associated with increased risk of fatal and non-fatal injuries,
academic failure, driving under the influence, violence and other crime and unsafe sexual
behavior (Hingson et al., 2005; 2009). Acute alcohol intoxication is associated with increased
accidental and self-inflicted injuries (Rehm, Gmel, Sempos, & Trevisan, 2003) and increases the
likelihood for alcohol-induced blackouts, which in turn confers its own risks for negative
consequences (LaBrie, Hummer, Kenney, Lac, & Pedersen, 2011). The negative impact that can
result from alcohol use and misuse among college students alone warrants continued research
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 19
that can help to strengthen intervention programs’ ability to reduce negative alcohol-related
consequences.
The present study will evaluate the efficacy of stand-alone PNF among adjudicated
students in a lab setting (AIM I). The first aim is elaborated upon in sections below. In addition
to evaluating main effects, we expect an indirect effect of reductions in perceived norms from
baseline to immediately post-intervention (i.e., reductions in normative misperceptions) on
changes in alcohol use one-month post-intervention (AIM II). Furthermore, as will be described
below, the present study will assess the presence of other change mechanisms by exploring
indirect effects of other constructs identified through the use of a novel think-aloud cognitive
assessment approach.
A Novel Methodological Approach: Articulated Thoughts in Simulated Situations
Despite the plentiful and still growing outcome research regarding PNF’s overall
effectiveness in reducing problem drinking, questions remain about the fundamental treatment
mechanisms. Furthermore, main effect sizes for drinking outcomes are typically in the small to
medium range and a substantial proportion of students who receive interventions continue to
drink heavily and experience negative consequences. Thus, additional research is needed to more
fully identify the limitations and active mechanisms of PNF if this approach is to continue to
evolve. New assessment strategies which attempt to identify mechanisms of change in PNF
interventions may aid in this endeavor. One such methodological approach with merit is the
Articulated Thoughts in Simulated Situations (ATSS) paradigm developed by Davison, Robins,
and Johnson (1983). The ATSS experimental paradigm is a laboratory-based think-aloud
approach designed to capture complex cognition and emotion through open-ended responding to
specific stimuli presented by the experimenter.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 20
Traditionally, ATSS involves having people imagine that they are participants in a
situation/vignette and to verbalize their thoughts and feelings during pauses in the story. The
scenario is typically presented in short segments or “doses”, after each of which the participant
articulates his thoughts and feelings. Because the ATSS paradigm is concerned with a person’s
immediate experience, responding does not rely on long-term memory and therefore avoids the
problems of retrospective self-reporting. The articulated thoughts themselves are recorded,
transcribed, and later content-analyzed. Among its advantages over other modes of assessment
such as questionnaires and interviews are its situational specificity, investigator control,
unconstrained response format, immediacy of assessment, unlimited choice of data coding
approaches, empirical flexibility, detection of group-specific cognitive-affective differences, and
the flexibility to study the cognitive aspects of a wide range of problems, including novel or
sensitive domains of inquiry (Davison et al., 1983; Zanov & Davison, 2010). In addition, since
cognitive processes are dynamic rather than static, the paradigm is an ideal assessment approach
for detecting ongoing cognition under the conditions in this study. Since the introduction of the
paradigm, over 100 published studies attest to the validity of ATSS in accessing complex
cognition across a wide array of domains.
Although PNF has been shown to improve drinking outcomes, empirical support for the
mechanism presumed to underlie the success of this approach has been more theoretically-driven
than practically-derived. Although such research is necessary and important, it is also limited by
its narrow focus on only one intervening variable – the reduction in normative misperceptions.
The ATSS provides a cognitive-affective and phenomenological framework that will be useful in
capturing participants’ impressions of the PNF intervention material and in so doing, potentially
uncover additional mechanisms of change. Rather than its traditional use within the context of a
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 21
“simulated situation”, the ATSS approach will be modified to consider the participants’
immediate reactions to each specific component (i.e., dose) encompassing the complete PNF
package as it is being assembled and displayed to the participant. The application of the ATSS to
studying how participants actively negotiate and construe information during an intervention is a
new and potentially fruitful direction for the paradigm.
As noted above, this paradigm assesses thoughts as they occur (rather than relying on
post factum reports) and does not limit response format while simultaneously allowing for
experimenter control. Because it allows for the participant to respond with any and all thoughts
that have occurred to her, ATSS seems to have particularly high utility when “little is known of
the cognitive terrain of interest” (Davison, Vogel, & Coffman, 1997, p.955). ATSS will allow us
to construct the intervention content for a participant while providing access to our participants’
real-time cognitions in response to the intervention. In this way, ATSS has the potential to
provide insight into the in-vivo cognitive processes that lead to behavior change. Taking the
cognitive and cognitive behavioral models described by both Aaron and Judith Beck (Beck, A.
1979; Beck, J. 1995) and Albert Ellis (Ellis & Dryden, 1997) as a theoretical basis and given the
variety of research supporting the validity of the cognitive behavioral model (and specifically the
support garnered for the contention that cognitions and emotions can influence behavior), we
expect that a detailed examination of the cognitions and emotions experienced during PNF
exposure will yield insight into the mechanisms that account for changes in outcome data (AIM
II).
Cognitive-affective change mechanisms. Although it is challenging to offer a priori
hypotheses for the emergence of specific cognitive-affective reactions that may indirectly
account for intervention efficacy, due to the dearth of research on the topic, some examples of
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 22
expected domains are as follows. First, if participants are skeptical regarding the source of the
normative feedback, and thus its validity, the effectiveness of social norms interventions may be
diminished (Fabiano, 1999; Granfield, 2002; Thombs, Dotterer, Olds, Sharp, & Raub, 2004). It is
generally understood by social norms practitioners that students experience varying degrees of
skepticism related to the normative information included in these interventions (Berkowitz,
2004). Participants may discredit feedback information if the norms are generated from a vague
source with which students are unfamiliar (e.g., “a survey of college students”). If a student
discredits or otherwise does not believe the normative feedback data to be accurate, then it is
unrealistic to expect that student to modify their own behavior to conform to a lower, more
modest true norm. Discounting the credibility of the normative feedback will allow heavy-
drinking students to continue their level of drinking without experiencing the sense of conflict
elicited by the knowledge that they are deviating from the prevailing norm, the presumed
mechanism of change within PNF interventions.
In one study that reported null findings for the impact of a SNM intervention to change
perceptions and thereby reduce alcohol use (Granfield, 2002), it was found via a post-
experimental questionnaire that the majority of students were skeptical of the marketed social
norms information. This skepticism was correlated with the variables of age, alcohol use, and the
perception of heavy use on campus, but was not evaluated in the context of the study’s main
effects. Based on their findings, the authors called for future efforts to assess measures of
believability and credibility in order to “assist in identifying and responding to the barriers that
may impede successful outcomes (Granfield, p.28).”
A similar study by Thombs et al. (2004) sought to investigate possible factors that could
help explain a failed SNM intervention. Most participants in a post-intervention sample reported
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 23
not believing the norms messages and higher levels of drinking were associated with lower levels
of perceived credibility. However, limitations associated with study design (non-experimental,
lack of a comparison group) precluded an explicit evaluation of the role that credibility issues
may have played in the intervention’s null effects (i.e., effect of feedback on perceived norms).
Despite the seemingly critical role of source credibility in the design and success of social
norms interventions, studies have not comprehensively explored this important factor and no
studies to our knowledge, aside from my MA thesis described below, have evaluated source
credibility or believability in the context of PNF, a more efficacious social norms intervention
than SNM. As with early SNM interventions, it is likely that some students may discount the
PNF data because they may not believe that the surveyed students represent a true sampling of
students or of their drinking behavior, or they may believe that the researchers manipulated the
data by providing incorrect information in an attempt to persuade the student to change. Students
may discredit the information based on these beliefs. Despite our intention to adequately
highlight the reliability and source of the data used in the feedback, we expect some variability in
the extent to which students find the information credible/believable and anticipate that this
coding theme will emerge in the ATSS data.
The current study will also assess for the presence of ATSS codes that capture
expressions of surprise/shock, neutrality, and self-reflection in response to the information
presented. It is expected that each of these codes, if present, will indirectly account for the effects
of PNF on perceived norms and alcohol use behaviors. The success of this study and its ability to
evaluate novel cognitive-affective variables related to mechanisms of change are strengthened by
its implementation in a laboratory setting, the real-time collection of cognitions and emotions,
and the unconstrained response format. As already noted, process research may provide a better
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 24
understanding of the key mechanisms of change. This will be of importance both theoretically
and in a broader applied realm. It may help improve the effectiveness of PNF and concurrently
evaluate the ability of an innovative methodological approach to provide important theoretical
and practical contributions to the larger social and clinical psychological literature on the
underpinnings of PNF for health behavior change.
ATSS and Depth of Processing
One potential concern with using the ATSS paradigm as a purely cognitive assessment
device within an active intervention is that it may impact an individual’s processing of the
information. That is, having participants pause periodically during the PNF to express what is on
their mind may either distract the participants or confer additive therapeutic effects beyond that
which arises purely from the content presented. The same may be true of the control group’s
alcohol education content. When one receives information, it can be processed with varying
levels of cognitive engagement. Depth of processing refers to the extent to which individuals
expend cognitive effort and think more carefully about the information provided in persuasive
communication. The elaboration likelihood model (Petty & Cacioppo, 1981, 1986) is based on
the assumption that attitude change and subsequent behavior change can occur as a result of two
differing routes of information processing—the central and peripheral routes—or a combination
of the two routes. When people utilize the central route of information processing, they expend
higher cognitive effort, think more carefully about information, and have greater depth of
information processing (e.g., explicitly weighing pros and cons).
One study on Web-based PNF sought to assess what college students are doing, whom
they are with, and where they are when viewing Web-based personalized feedback. Results
indicated that most students were engaged in other activities (32% in one activity and 30% in
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 25
two activities), were alone, and were at home when viewing the feedback. Although most
students were participating in one or more other activities when viewing feedback, most still
considered themselves to be attentive to the intervention content. This subjective measure of
attention was the only variable that was significantly and positively associated with time spent
viewing feedback, as more objective measures of attention (i.e., home, alone, number of
activities) were not associated with time spent viewing the content. Findings indicated that there
was a significant effect of the Web-based PNF on drinks per week, but only among those higher
in self-reported attentiveness. In sum, it appears that it was the quality of their attention when
viewing the feedback that moderated intervention effects on drinks per week, and not how many
other things participants were doing or how long they viewed the PNF.
Since self-reported attentiveness moderated the effect of Web based feedback on drinking
outcomes, it is worth considering whether the inclusion of the ATSS may lead to greater depth of
processing through heightened attention and thus confer additive intervention effects. Unlike
Web-delivered feedback in a students’ home, all three conditions in the current study will be
ensured adequate levels of attention to content given the experimenter-controlled setting of the
lab. Yet it remains uncertain as to whether the inclusion of ATSS will enhance the effectiveness
of conditions by deepening an individual’s processing of the content presented, beyond the
effects of content delivered sans ATSS.
In contrast to the consideration of central route of normal versus enhanced processing, it
is also important to consider that when people pay less attention to the content of persuasive
communication, they are more likely to be influenced by peripheral cues. Ability to pay attention
to persuasive communication can be influenced by distraction. Prior research has shown that any
factor that hinders the efficacy of processing the information of the message (e.g., distraction)
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 26
reduces the possibility of processing via the central route (Di Blasio & Milani, 2008; Petty,
Wells, & Brock, 1976). While the impact of distraction on task performance has been well
studied, its impact on cognitive assessment is less well-documented, particularly regarding think-
aloud assessment and emotion-provoking stimuli.
In a novel study by Hsu, Babeva, Feng, Hummer, and Davison (2014), the effects of task
disengagement on responding to the ATSS paradigm was examined by experimentally inducing
distraction. Participants were asked to verbalize their thoughts and feelings in response to three
commonly used types of hypothetical scenarios in the ATSS literature (i.e., a neutral, an anxiety-
provoking, and an anger-provoking scenario) in one of two experimentally manipulated
distraction conditions: (1) while playing Tetris or (2) while answering trivia questions. A third
condition involved no experimentally-induced distraction. Results revealed that distraction did
not impact indices of emotion, regardless of distractor modality, but did affect cognitive
processes. Specifically, the trivia questions distraction condition resulted in significantly higher
proportions of insight and causal words, and higher frequencies of non-fluencies (e.g., “uh” or
“umm”) and filler words (e.g., “like” or “you know”). Coder-rated content analysis found more
disengagement and more misunderstanding particularly in the trivia questions distraction
condition. The implications for the present study include a necessary focus on preventing
distraction during the ATSS to ensure that cognitive processes remain intact during the
intervention assessment.
Ensuring similar levels of attention to content delivered within each condition while also
mitigating against risks associated with distraction is not foreseen to be problematic in this
proposed study. In contrast to the Web-based feedback study and the study on distraction during
ATSS, participants in each of the conditions will receive the targeted information in the lab and
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 27
measures will be taken (described below) to enhance attention by guiding the participant through
the content on a computer. This will prevent participants from being distracted by other
activities. The computer-delivered content in a lab setting will be a more controlled environment
than a Web-based study, and participants will be asked to put away their belongings during the
study, thus eliminating potential distractions such as texting or phone calls. Therefore, it is
unlikely that participants will be distracted and that such distractions might harm the intervention
fidelity. On the other hand, because the ATSS has heretofore not been utilized in the fashion
proposed for this study, it is unknown if thinking aloud will result in added depth of processing
which might enhance the think-aloud conditions in some way.
To account for the potential confound of ATSS reactivity effects, we will include two
intervention conditions; one condition will include viewing standard PNF without ATSS and the
other condition will include PNF with ATSS (PNF-ATSS). Both intervention conditions will be
compared to an active control condition which will also include a think-aloud component and
visual presentation of alcohol-related psychoeducation on computer. This provides a
significantly more rigorous control comparison than an assessment-only condition. We
hypothesize that both PNF conditions will generate greater reductions in perceived norms,
alcohol use, and alcohol-related consequences compared to the control. Further, after considering
previous research on the nature of distractions during both Web-based PNF and ATSS, but given
the high level of experimenter control in the current study and low probability of distractions, we
anticipate no differences between the PNF-only condition and the PNF-ATSS condition on
intervention outcomes.
MA Thesis Extension
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 28
The current study represents, in part, a major extension and elaboration of my MA thesis
(Hummer & Davison, 2016). In that study, we evaluated a lab-based PNF experiment that used a
experimental manipulation to control the source credibility and reference group proximity of the
normative feedback data displayed in the intervention. Due to restricted timelines, subject pool
constraints, and lack of resources, the “actual” norm presented to participants in their PNF was
artificially produced to be about half of each individual’s perceived group norm, which was
consistent with findings reported in the literature using large sample sizes and multiple different
reference groups (e.g., Larimer et al., 2011; Neighbors, LaBrie, et al., 2010). As I will discuss
below, while the absence of population-wide drinking data and follow-up assessments of
behavioral data limited the inferences that we could from this study, it was designed that way to
examine specific elements by creating a laboratory-based analog of PNF presentations. The
primary aim was to gather a more thorough understanding of how the credibility and reference
group factors can moderate the effectiveness of PNF to reduce perceived descriptive norms and
decrease intentions to drink. The PNF was implemented via a PowerPoint presentation composed
of slides that graphically and with verbal explanations showed how the participant’s drinking
compared to that of other college students per the manipulations described above. Small groups
of 4-10 individuals participated at a time, each with their own computer and PNF.
Overall, results revealed large effect sizes for the reduction of perceived weekly drinking
norms of both a ‘typical American college student’ and a ‘typical same-sex, same-class year
USC student’. However, the intervention was not found to change participants’ intentions to
drink. Those students in the high credibility condition evidenced a greater reduction in perceived
weekly drinking norms of an American college student referent. However, no main effects for
the credibility factor were found for either of the other outcomes. Similarly, participants in the
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 29
high proximity condition evidenced greater reductions in perceived university-specific weekly
drinking norms than participants in the low proximity condition, although no main effect was
found for either of the other two outcomes.
The Hummer and Davison (2016) study had several limitations, which have been
carefully considered and addressed in the current study. First, the study lacked a comparison
group. As noted above, the present study will include an active comparison group. Second, the
use of manufactured data to create an “actual norm” that was approximately 50% lower than an
individual’s perceived norm is different from how PNF interventions operate in practice, in
which the actual norm is real data and held constant for all participants. Particularly for heavier
drinkers in the sample who typically held the highest estimates of others’ behavior, the use of a
manufactured norm may have resulted in an inaccurately smaller magnitude of normative
discrepancy, while for students who held low perceptions of others’ drinking, it may have
resulted in an inaccurately higher discrepancy. While it was indeterminable within the data as to
whether the effects may have been different with greater variability in the magnitude of
normative discrepancies, the current study will use real population-level data to improve upon
this limitation.
Second, the study was limited by its use of a convenience sample comprised of
psychology students. It is possible that some characteristic associated with that sample
contributed to the lack of a baseline relationship between perceived norms and individual
drinking behavior. Although students indeed modified downward their estimates of peer drinking
behaviors, the lack of an association between perceived norms and individual drinking behavior
at baseline suggested that students were not swayed by their perceived norms of their peers’
drinking when making their own behavioral decisions about when and how much to drink.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 30
Therefore, it makes conceptual sense that the observed reduction in perceived norms did not
coincide with changes in their intentions to drink. The current study will benefit from targeting a
known high-risk group of drinkers on college campuses (i.e., adjudicated students), while
providing important extensions of previous research targeting this population.
Third, we assessed drinking intentions immediately after the intervention as a proxy for
future alcohol use behavior. This was not unreasonable since intentions are widely accepted as a
precursor to behavior and behavior change (Fishbein & Ajzen, 1975). While it is not possible to
know whether there would have been a different pattern of results had actual drinking behavior
been collected, this potential limitation will be addressed in the current study, which will follow
participants for a short term follow up of one month to assess for self-reported behavior change
over time while providing an opportunity to assess for indirect effects.
Reference Group Specificity
Theoretical perspectives suggest that individuals look to those in their more immediate
community to make decisions about the appropriateness of behavior. Social Comparison theory
(e.g., Festinger, 1954) and Social Impact Theory (Latane, 1981) suggest that proximal reference
groups exert greater influence on beliefs and behavior than more distal social reference groups.
As applied to alcohol-related social norms, misperceptions of the drinking patterns of more
proximal and salient reference groups are more likely to influence drinking behavior than
misperceptions of more distal and less salient reference groups (Borsari & Carey, 2003;
Korcuska & Thombs, 2003; Larimer et al., 2009; Lewis & Neighbors, 2006; Neighbors, Lewis et
al., 2010).
Despite the stronger relationship between more proximal reference group norms and
drinking, relative to more distal reference group norms and drinking, evidence is mixed
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 31
regarding which reference group implicated in PNF interventions is more efficacious. Lewis and
Neighbors (2007) did not find any overall differences in the short-term efficacy of gender-
specific and nongender-specific PNF to reduce perceived norms and alcohol consumption, but
gender-specific feedback worked better for women who identified more closely with their gender
(identification as a moderator). In contrast, Neighbors, Lewis et al. (2010) demonstrated that
PNF delivered with gender-specific norms reduced weekly drinking, whereas nongender-specific
information did not. In a more recent multisite randomized controlled trial using 8 increasingly
proximal reference groups, the PNF intervention was found to be most effective at reducing
drinking when the typical student (i.e., least proximal normative referent) was used as the
normative reference group (LaBrie et al., 2013). In their studies of Web-based feedback among
adjudicated students, Doumas et al. (2009; 2011) utilized non gender-specific reference groups
for the normative feedback components. The current study will use a gender-specific normative
referent and evaluate whether identification with this group moderates intervention efficacy, as
noted below.
Moderators of Intervention Efficacy
Questions about which subgroups of individuals benefit most (or least) from an
intervention are investigated through tests of moderation. Moderation analyses are important for
informing the next generation of PNF trials and for directly informing practice. Identifying those
who respond differently to intervention may lead to further investigation of subgroups for whom
there may be distinct causal patterns or prognoses. Findings may reassure practitioners that a
particular intervention can be effective for groups traditionally thought to be hard to treat, such
as adjudicated students. Within an intervention trial such as PNF, the relevant question is
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 32
whether a correlated variable measured prior to intervention is differentially associated with
drinking outcomes for the treatment compared to control groups.
Group Identity. Social Identity Theory (SIT; Tajfel & Turner, 1979) proposes that
identification with group membership influences individual and group behaviors, norms, and
cognitions. Considerable work on SIT has explored between- and within-group processes. A
central issue of SIT is the overlap between one’s self-perceptions and one’s view of others with
whom one feels connected, versus others to whom one does not feel connected (Abrams &
Hogg, 1999). According to SIT, people view themselves and others as group members with a
common or shared social identity, which is derived primarily from group memberships (Abrams
& Hogg, 1999; Turner, Hogg, Oakes, Reicher, & Wetherell, 1987). The more one identifies with
a particular group, the more favorably he or she views other members of that group (Klar &
Giladi, 1997). Importantly, with respect to between-group influences, the more an individual
identifies with other members of his or her group (e.g., other students at one’s university), the
more influence perceptions of the group norms should have on the individual. In support of this
notion, several studies have shown that the more one identifies with a reference group the
stronger the relationship between perceived norms and drinking (e.g., Grossbard, Hummer,
LaBrie, Pedersen, & Neighbors, 2009; Hummer, LaBrie, Lac, & Louie, 2013; Hummer, LaBrie,
& Pedersen, 2012; Neighbors, LaBrie, et al., 2010).
Together, these findings suggest that the relationship between perceived drinking norms
and alcohol consumption may depend on how strongly students identify with other students in
the reference group. If the relationship between perceived norms and alcohol consumption
depends on the strength of students’ identification with group members, then differences in
identification should impact the effectiveness of PNF on perceived norms and alcohol
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 33
consumption (AIM III). This will be especially salient information to obtain within an
adjudicated student sample receiving PNF. It is hypothesized that the PNF intervention will be
most effective among students who identify more closely with a typical same-sex USC student.
Self-determination and causality orientations. Self-determination theory suggests that
the extent to which individuals are more or less autonomous or controlled, arises over time as a
function of exposure to autonomy-supportive versus controlled environments (Deci & Ryan,
1985a, 2002). As a result of prolonged exposure to environments that emphasize submission to
authority, living up to others’ expectations, inadequate opportunity for self-expression, and
pressure to engage in specific behaviors, individuals presumably develop a more controlled
motivational orientation (Williams & Deci, 1998; Williams, Deci, & Ryan, 1995). The controlled
orientation refers to a general tendency to perceive pressure from one’s environment and to
experience a lack of true choice in one’s behavior (Deci & Ryan, 1985a, 1985b). The autonomy
orientation involves a high degree of experienced choice with respect to the initiation and
regulation of one’s own behavior (Deci & Ryan, 1985a). When autonomy oriented, people seek
out opportunities for self-determination and choice, and accordingly they can be described as
having a generalized tendency toward what deCharms (1968) described as an internal perceived
locus of causality. Deci (1980) referred to the controlled versus autonomous tendencies as
causality orientations, suggesting that these broad motivational orientations can be usefully
characterized in terms of people’s understanding of the nature of causation of behavior.
A number of studies suggest that controlled orientation is strongly associated with
sensitivity to social expectations and pressures (e.g., social norms) (Deci & Ryan, 1985b; Lewis
& Neighbors, 2005; Zuckerman, Gioioso, & Tellini, 1988). Evidence does suggest that more
controlled students are more likely to drink as a function of perceived social pressure (Knee &
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 34
Neighbors, 2002). Only one study to our knowledge has evaluated self-determination and
controlled orientation, more specifically, in a PNF intervention, but found that it did not
moderate the effects of the intervention on drinking outcomes (Neighbors et al., 2006). However,
one’s orientation style may play a stronger role for heavier-drinking college students who are
sanctioned for violating the campus alcohol policy. Because controlled orientation is associated
with social influences on drinking, we expect that controlled orientation will moderate the effects
of PNF on drinking outcomes, such that the intervention will be more effective among more
controlled students (AIM III). In addition, not previously examined is the role of autonomous
orientation with respect to social influences on drinking. Individuals high in autonomy may be
more apt to react negatively to PNF, which seeks to leverage perceived social pressure to
conform one’s own behavior to that of a prevailing norm. We anticipate that autonomous
orientation will thus moderate the effects of PNF on drinking outcomes, such that the
intervention will be more effective among less autonomous students (AIM III).
Intervention defensiveness. When studying adjudicated students, it is important to
consider characteristics or reactions that may likely arise following the sanctioning process and
prior to the delivery of an alcohol intervention. Defensiveness is one such factor. Defensiveness
is defined here as the degree of resistance to the “alcohol-education session” measured prior to
attendance and includes aspects of perceived choice, perceived benefits, and openness to change
attempts. Defensiveness is prominent in students who receive an alcohol violation (Sharkin,
2007) and is higher in adjudicated students than non-adjudicated students (Palmer, 2004; Palmer,
Kilmer, Ball, & Larimer, 2010). Palmer et al. (2010) found that defensiveness moderated
drinking outcomes such that students lower in defensiveness had better outcomes after a brief
intervention than students high in defensiveness. Thus, mandated students lower in defensiveness
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 35
may react differently to PNF following a sanction than their more defensive peers, though this
has not yet been examined. We will evaluate defensiveness as a moderator of PNF intervention
efficacy on perceived norms and drinking (AIM III), expecting that the intervention would be
more effective among students who are lower in defensiveness prior to the intervention.
Objectives of the Current Study
Although there is a growing consensus regarding the efficacy of PNF interventions, both
theoretical and practical questions remain regarding how PNF influences drinking outcomes.
PNF approaches are designed to correct normative misperceptions by showing discrepancies
between actual norms and students’ perceptions and behaviors. Correcting normative
misperceptions then motivates behavior change (Rice, 2007). Consistent with this proposed
mechanism, reductions in perceived descriptive norms typically partially mediate the efficacy of
PNF interventions (e.g., Doumas et al., 2011; LaBrie et al., 2013; Walters, Vader, & Harris,
2007). But is this the whole story? We think not. Previous PNF trials have been limited by their
focus on only one mediator. Although exploratory in nature, we expect that a more qualitatively
detailed examination of the cognitions and emotions experienced during PNF exposure will yield
insight into other mechanisms that account for the positive outcome data. The ATSS
experimental paradigm was adapted from its more traditional applications in hypothetical and
simulated scenarios to a laboratory-based, process-oriented assessment of a PNF intervention in
which little is known of the cognitive terrain regarding how individuals process such
information. Since cognitive processes are dynamic rather than static, the paradigm is an ideal
assessment approach for capturing ongoing complex cognition and emotion through open-ended
responding to the specific stimuli under the conditions in this study. We anticipate that the ATSS
might uncover as of yet unidentified cognitive-affective change mechanisms within the PNF
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 36
intervention.
This study is further strengthened by its targeting of a high-risk group of college students
who have been sanctioned by the university for violating its alcohol policies. The two prior
studies (Doumas et al., 2009; 2011) that used PNF among adjudicated students were done in the
context of Web-based multicomponent interventions that included several additional intervention
components rather than the traditional stand-alone PNF, making it difficult to ascertain which
components were most responsible for reductions in drinking. Moreover, the mediator and
drinking outcomes were assessed concurrently rather than temporally over time. This study will
test a pure PNF intervention delivered individually on computer in a lab setting and use pre/post
assessments to monitor changes in study outcomes longitudinally. These measures address
limitations in previous research while allowing for a high degree of experimenter control and
increase the potential to identify mechanisms of change within the intervention. The study will
evaluate a PNF-only condition as well as a PNF-ATSS condition to assess for any additive
effects of the inclusion of the ATSS assessment during intervention. Both conditions will be
compared to an active comparison group which received information about college students and
alcohol use derived from the National Institute on Alcohol Abuse and Alcoholism (NIAAA) in a
format similar to how participants receive PNF intervention content (i.e., on a lab computer, with
audio/visual synced components, and with think-aloud segments).
Finally, questions about which subgroups of individuals benefit most (or least) from the
PNF intervention will be investigated through moderator analyses. The chosen moderators are
important on a theoretical level (i.e., identification with reference group) as well as potentially
relevant to the targeted sample of adjudicated students (i.e., controlled vs. autonomous
orientations and pre-intervention defensiveness). In sum, the research seeks to provide important
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 37
theoretical and practical contributions to the social norms and alcohol intervention literature.
Understanding the key mechanisms by which social norms interventions work, why, and for
whom, is imperative for the evolution of norms-based intervention strategies and theory
development.
The specific aims and major hypotheses of the current project are as follows:
AIM I: Examine main effects of the Personalized Normative Feedback (PNF) intervention in
reducing perceived alcohol use norms of other same-sex USC college students, individuals’
own drinking behavior, and individuals’ alcohol-related consequences.
Hypothesis 1: There will be main effects of PNF-only condition and PNF-ATSS
condition on intervention efficacy such that participants in both intervention conditions,
independently and collectively, will demonstrate greater reductions in perceived typical
same-sex student norms one-month post-intervention relative to control participants. No
differences are expected between the two PNF intervention conditions.
Hypothesis 2: There will be main effects of PNF-only condition and PNF-ATSS
condition on intervention efficacy such that participants in both intervention conditions,
independently and collectively, will demonstrate greater reductions in alcohol use one-
month post-intervention relative to control participants. No differences are expected
between the two PNF intervention conditions.
Hypothesis 3: There will be main effects of PNF-only condition and PNF-ATSS
condition on intervention efficacy such that participants in both intervention conditions,
independently and collectively, will demonstrate greater reductions in self-reported
alcohol-related consequences one-month post-intervention relative to control
participants. No differences are expected between the two PNF intervention conditions.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 38
AIM II: Investigate mechanisms of change within the PNF intervention.
Hypothesis 4: There will be indirect effects of ATSS codes on changes in perceived
norms and alcohol use one-month post-intervention.
Hypothesis 5: There will be an indirect effect of reductions in perceived norms from
baseline to immediately post-intervention (i.e., reductions in normative misperceptions)
on changes in alcohol use one-month post-intervention.
AIM III: Investigate moderators of PNF intervention efficacy.
Hypothesis 6: The PNF intervention will be most effective at reducing perceived norms
and drinking behavior from pre– to post-intervention among students who identify more
closely with a typical same-sex USC student.
Hypothesis 7: The PNF intervention will be most effective at reducing perceived norms
and drinking behavior from pre– to post-intervention among more controlled students.
Hypothesis 8: The PNF intervention will be most effective at reducing perceived norms
and drinking behavior from pre– to post-intervention among less autonomous students.
Hypothesis 9: The PNF intervention will be most effective at reducing perceived norms
and drinking behavior from pre– to post-intervention among students who are lower in
defensiveness prior to the intervention.
CHAPTER 2: METHOD
Participants and Recruitment
Male and female undergraduate students (N = 70, 51% female) ranging in age from 18 to
21 (M = 19.01 years), who had been issued a first offense citation for violating the campus
alcohol policy, completed all aspects of this study. Their class standings were 50.0% Freshmen,
28.6% Sophomore, 18.6% Junior, 2.9% Senior. Students self-identified as White (64.3%), Asian
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 39
(27.1%), and African American (8.6%). A third of the sample was a member of a fraternity or
sorority (32.9%), indicating an overrepresentation compared to the overall student body involved
in fraternities/sororities (approx. 25%). Table 1 provides descriptive information for
demographic variables as well as for baseline and follow-up measures of all outcome variables in
the overall sample and by study condition.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 40

Table 1
Descriptive Statistics for Demographics Characteristics and Primary Outcomes by Study Condition
Variable % (N) M (SD) % (N) M (SD) % (N) M (SD) % (N) M (SD)
Age (Years) 19.01 (.96) 19.08 (1.01) 18.82 (.88) 19.06 (.93)
Sex
Male 51.4 (36) 54.1 (20) 41.2 (7) 43.8 (7)
Female 48.6 (34) 45.9 (17) 58.8 (10) 56.3 (9)
Ethnicity
Hispanic 17.1 (12) 18.9 (7) 70.6 (12) 0 (0)
Non-Hispanic 82.9 (58 81.1 (30) 29.4 (5) 100 (16)
Race
Asian 27.1 (19) 24.3 (9) 35.3 (6) 25.0 (4)
Black or African American 8.6 (6) 8.1 (3) 5.9 (1) 12.5 (2)
White or Caucasian 64.3 (45) 67.6 (25) 58.8 (10) 62.5 (10)
Greek-Affiliation
Greek Member 32.9 (23) 32.4 (12) 29.4 (5) 37.5 (6)
Non-Greek Member 67.1 (47) 67.6 (25) 70.6 (12) 62.5 (10)
Baseline Outcomes
Weekly Drinks 8.32 (5.51) 9.14 (5.82) 6.74 (5.13) 8.13 (5.10)
Negative Consequences 30.13 (5.27) 30.11 (5.68) 29.29 (3.50) 31.06 (5.98)
Perceived Weekly Drinking 11.26 (5.62) 11.19 (5.52) 12.09 (7.16) 10.56 (4.00)
Follow-up Outcomes
Weekly Drinks 6.91 (5.16)** 6.78 (5.02)** 7.24 (5.45) 6.88 (5.49)
Negative Consequences 28.80 (5.34)** 28.00 (4.27)** 27.63 (3.56)* 31.81 (7.76)
Perceived Weekly Drinking 7.01 (4.06)*** 6.94 (4.35)*** 5.50 (3.17)** 8.67 (3.75)
*p < .05; **p < .01; *** p < .001
Note. Asterisks next to follow-up measures indicate significant changes from baseline to follow-up for a specific measure; there were no significant differences by
intervention condition for demographics or outcome variables at baseline.
Control (N = 16) PNF (N = 17) PNF ATSS (N = 37) Overall (N = 70)
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 41
In coordination with USC Student Judicial Affairs, students were offered the choice to
either participate in our research study or complete a separate alcohol awareness program
through Student Affairs. Participating in our study served to fulfill the students’ sanctions. Upon
receiving confirmation of expressed interest in participation, that is, after students enrolled in our
research study as an alternative sanction, participants were randomized to one of the three
conditions using a 2:1:1 allocation ratio: PNF-ATSS, PNF-only, or control. Students were given
information regarding informed consent, the research protocol, the risks and benefits of
participation, the voluntary nature of participation, measures to ensure confidentiality, how to
complete the pre-intervention survey, and how to sign up for the lab-based portion of the study.
Surveys were sent out and administered using Qualtrics online data collection software.
Recruitment and retention were further bolstered by the study’s incentive structure ($40 for
completing the one-month follow up survey).
Procedures
Recruitment, baseline (pre-intervention) assessment, the intervention, and one month
follow up assessment took place across two semesters in the 2016 - 2017 academic year. All data
were collected via self-report directly from participants. Both PNF and control interventions
occurred within two to three weeks following completion of a baseline assessment. At the end of
the intervention and while still in the laboratory, the participants completed a brief online post-
intervention assessment of their normative perceptions. This was followed by instructions for
how to complete the follow up assessment and receive the monetary incentive.
Baseline assessment. After agreeing to participate in the study, each student received an
email containing an electronic link to the online survey that served as the baseline assessment.
Upon clicking on the link, the participant viewed the informed consent form and provided
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 42
electronic consent before being directed to the survey itself. At the end of the survey, the
participant was given instructions for how to sign up for the lab-based intervention and
assessment. Participants were informed that they must complete the online assessment prior to
attending the lab portion of the study, with no exceptions.
PNF intervention protocol. Following completion of the baseline survey, participants
arrived during their scheduled appointment for the lab portion of the study and were given a brief
introduction to the agenda. Next, participants were seated at a computer and presented with an
example feedback slide to introduce the types of information that would subsequently be shown
(see Appendix B for PNF intervention script and Appendix C for PNF-ATSS condition script).
The slide was built for the participant, one piece at a time, and pre-recorded audio guided the
participant with an explanation about what each component of the slide represents. The audio
was synced with visual objects and arrows pointing at each part of the feedback graph, as it was
introduced, until it was a complete feedback slide.
Participants then progressed through the normative feedback designated by their assigned
condition (see Appendix A for example slides). Both PNF conditions contained four
individually-tailored feedback slides (each slide referring to a different drinking behavior). Text
as well as separate graphs, each including three bars, were used to present information regarding
the average number of drinks per drinking occasion, the total number of drinks consumed per
week, the maximum number of drinks consumed on one occasion in the past month, and binge
drinking frequency for (a) one’s own drinking behavior, (b) their reported perceptions of a
typical same-sex USC student, and (c) actual same-sex USC college student drinking norms. The
bottom of each respective slide contained source information for the data, noting that the
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 43
information came from a large, representative sample of USC students surveyed in 2016 (also
including the sample size for the survey).
The data used in constructing the electronic normative feedback graphs were derived
from two sources. An individual’s perceived group norms and individual alcohol use were
collected from responses on the baseline survey. The actual student drinking norms were
obtained from a 2016 representative campus wide survey of over 600 students that I conducted in
coordination with the Wellness and Health Promotion Office at USC. After viewing the example
feedback slide, participants were given an opportunity to ask any questions about the graphical
feedback display.
PNF-ATSS intervention protocol. The ATSS is a flexible paradigm that allows for an
in-vivo, “online” assessment of the content of cognitions and affective dynamics in a semi-
structured manner. In a review of numerous studies that utilized the ATSS paradigm, Davison et
al. (1997) provided support for the predictive, face, concurrent, and construct validity of the
paradigm. They also note that the reliability of coding (Pearson’s product moment coefficient)
can range between .75 and .86. This is an impressively high correlation coefficient that arguably
depends not only on the quality of the codes used, but on the extent of the coders’ training. With
an assessment tool like ATSS, psychometric issues are often brought up in response to the
coding of data rather than the data themselves (Heyman & Slep, 2004). In the case of ATSS,
validity and reliability, especially inter-rater reliability, are contingent on the coding scheme
selected and the training of coders. To the degree that the coding scheme has relevance to the
topic being discussed and inter-rater reliability is acceptable, ATSS yields valid and reliable data.
We invested considerable efforts in these undertakings. High achieving undergraduate research
assistants were recruited to aid in the development and piloting of intervention materials during
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 44
the summer of 2016 and continued to aid in all aspects of the study’s implementation, including
coding of ATSS-generated qualitative information. And, not to discount the primary resource,
my advisor has close to 40 years’ experience with the coding of ATSS data.
As in the PNF-only condition, participants arrived during their scheduled appointment for
the lab portion of the study and were given a brief introduction to the agenda. This time, in
addition to being presented with an example feedback slide to introduce the types of information
that would subsequently be shown, participants also received instructions for the think-aloud
portion of the intervention. We modeled the PNF on how complex scenarios are traditionally
divided into short segments in the ATSS so that think-aloud data could be obtained as the PNF
presentation unfolds. The experimenter informed the participant that as they listen and follow
along to the slide presentation, they would periodically hear a bell tone, which would be
followed by a 60-second pause in the presentation. During this pause, the participants were
instructed to speak into a small microphone their immediate thoughts and feelings at that
particular point of time in reaction to the graphical feedback unfolding in the presentation. Once
it was ensured that the participant understood the task at hand in the course of the practice trial,
the experimenter then began to record the audio in the microphone, cued the main PNF
presentation, and exited the room. The ATSS assessment was conducted twice during each of
four PNF slides that were shown. For a total of eight think-aloud segments. When finished, the
participant was instructed to knock on the door of the room for the experimenter to come in and
present further instructions for the immediate post-intervention survey.
Control condition. Both intervention conditions were compared to an active control
condition, in which participants received alcohol-related psychoeducation from the National
Institute on Alcohol Abuse and Alcoholism (NIAAA) and the Center for Disease Control (CDC)
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 45
(See Appendix D for control intervention script and content.) To control for setting and interface,
this information was presented in the same format as the two active conditions, a powerpoint
presentation with audio narration. To control for potential effects conferred by the ATSS think-
aloud portion of the PNF intervention, participants in the control condition also participated in
the think-aloud assessment during their presentation. An equal number of think-aloud
opportunities were presented in both conditions.
Post-intervention assessment. Immediately following completion of the presentation
within each respective condition, participants were asked to complete a brief online survey. To
determine the ability of the PNF intervention to immediately correct normative perceptions,
participants responded to the identical measure as assessed at baseline (Drinking Norms Rating
Form, see below).
Measures
Measures were carefully chosen to assess behavioral patterns, main outcome targets, and
theoretical constructs potentially related to intervention efficacy (see Appendices E and F for full
descriptions and complete scales that were used in the pre- and post-intervention surveys). Most
behaviors were reported over the past week or month to reduce problems with retrospective
recall of behaviors. The following measures were used:
Demographic information. Age, gender, ethnicity, class year, and Greek-membership
affiliation were assessed.
Alcohol consumption. Participants completed the Daily Drinking Questionnaire (DDQ:
Collins, Parks, & Marlatt, 1985), which is designed to assess typical daily alcohol consumption
over a specific time period (past 30 days in the present study). Participants are asked to estimate
the typical number of drinks consumed (based on a definition of a standard drink provided to
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 46
participants) on each day of the week, which, when summed together, provide a composite of
variable of the total number of drinks consumed in a typical week. The DDQ has been
commonly used in alcohol intervention studies and has been shown to be a reliable and valid
measure of college student drinking (e.g., Larimer et al., 2001; Marlatt et al., 1998). See Table 1
for pre- and post-intervention means and standard deviations across conditions.
Alcohol-related consequences. Alcohol problems were assessed using the Rutgers
Alcohol Problem Index (RAPI; White & Labouvie, 1989). The RAPI assesses the occurrence of
24 negative consequences resulting from one’s drinking over the past month (e.g., “Not able to
do your homework or study for a test” and “Had withdrawal symptoms, that is, felt sick because
you stopped or cut down on drinking”). Each item is rated on a scale from 1-4 with 1 indicating
“never” and 4 indicating “more than 10 times”. A summed composite was formed for use in the
study. See Table 1 for pre- and post-intervention means and standard deviations across
conditions.
Descriptive norms. The Drinking Norms Rating Form (DNRF: Baer, Stacy, & Larimer,
1991) was used to assess descriptive norms, or perceptions of actual drinking behavior of other
students. The format of the DNRF mirrors that of the DDQ, except participants provide estimates
of alcohol use for a particular reference group, in this case, a typical same-sex USC student. The
DNRF perceived total weekly drinking was used as a primary outcome measure. The DNRF and
modifications thereof have been used in numerous studies related to social norms and college
student drinking. It has consistently demonstrated good prospective and concurrent validity and
has good test-retest reliability (e.g., Neighbors, Dillard, Lewis, Bergstrom, & Neil, 2006). See
Table 1 for pre- and post-intervention means and standard deviations across conditions.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 47
Group identity. The Inclusion of Other in the Self (IOS) scale (Aron, Aron, & Smollan,
1992; Tropp & Wright, 2001) was used to measure identification of interrelatedness or closeness
with the chosen normative reference group. Participants were presented a series of seven Venn
diagrams ranging from non-overlapping circles to completely overlapping circles and asked to
select which diagram best represented their level of identification with a typical USC student of
the same sex as the respondent. The IOS has demonstrated good test-retest reliability, and good
concurrent, discriminant, and construct validity (Tropp & Wright, 2001). Level of identification
for females (M = 4.22, SD = 1.33) and males (M = 4.38, SD = 1.16) indicated moderate levels of
identification with USC students of their same sex.
General causality orientation. The General Causality Orientation Scale (Deci & Ryan,
1985b; revised: Ryan, 1989) was used to assess controlled and autonomous orientations. This
scale is derived from Self-Determination Theory (Deci & Ryan, 1985b). It is intended to assess
the strength of different motivational orientations within an individual. The orientations of
interest in the current study, labelled Autonomy and Controlled, are understood as relatively
enduring aspects of personality, and each is theorized to exist within each individual to a greater
or lesser extent.
The revised form of the General Causality Orientation Scale is comprised of 17 scenarios.
It has the original 12 vignettes and the original 36 items. Five vignettes and 15 items were added.
The newer vignettes and items refer to more social-interactions because the original vignettes
were heavily oriented toward achievement situations. Responses follow each scenario: an
autonomous response and a controlled response. Participants rated the extent to which each
response would be characteristic for him or for her. An example scenario is “Within your circle
of friends, the one with whom you choose to spend the most time is ______.” The controlled
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 48
orientation is then measured by the response “The one who is the most popular of them.” The
autonomous orientation is measured by the response “The one with whom you spend the most
time exchanging ideas and feelings”. Participants rate each response on a scale ranging from 1
(very unlikely) to 7 (very likely). Subscale scores are generated by summing the individual's 17
responses on items corresponding to each subscale. Higher scores indicate higher amounts of the
particular orientation represented by the response. The controlled orientation subscale had a
mean of 68.57 (SD = 12.17) and demonstrated adequate reliability (α = .76). The autonomous
orientation subscale had a mean of 98.79 (SD = 12.27) and demonstrated good reliability (α =
.86).
Intervention defensiveness/resistance. Defensiveness was assessed with an adapted 11-
item scale designed and validated by Palmer et al. (2010) to assess the degree of resistance to the
“alcohol-education session” that the students would be attending. Participants were asked to rate
the following items from 1 (strongly disagree) to 6 (strongly agree). Items #1,2,3,4,5,7, and 9
were reverse scored and a mean composite was calculated such that higher scores indicate
greater defensiveness. The scale had a mean of 3.11 (SD = .96) and demonstrated good reliability
(α = .85).
1) I am genuinely interested in the alcohol education session I will be attending.
2) It is my choice to attend the alcohol education session.
3) I am interested in knowing more about my drinking.
4) I would be interested in learning how my drinking compares to other students.
5) I am open minded about the alcohol education session.
6) Attending the alcohol education session will be a waste of my time.
7) The alcohol education session might benefit me.
8) I am not like the people the alcohol education session was designed for.
9) I might make some changes in my drinking as a result of attending the alcohol education
session.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 49
10) The alcohol education session may have information that will be useful for me.
11) I have no reason to think about how much I drink.
ATSS Coding Strategy
ATSS think-aloud data and the control condition think-aloud data were recorded and
analyzed iteratively using an a priori coding scheme. This coding scheme was created through
several months review of participant audiotapes and transcripts by Dr. Davison’s Lab for
Cognitive Studies in Clinical Psychology, which included several graduate students and several
undergraduate student research assistants. Multiple iterations of coding categories and code
definitions were generated, and seven coding categories were ultimately selected based on a
continuum of factors hypothesized to relate to behavior change or the absence thereof: Sustain
talk, skepticism, follow/neutral, believability, reflective analysis, positive surprise, negative
surprise. Thirty to forty example statements were then extracted from transcripts as
representative statements of specific scores within each coding category. The resulting coding
manual (Appendix G) explains in-depth the seven coding categories by defining their
characteristics (see below), describing how each code may appear within transcripts, and
providing 10-20 scoring examples for each code.
1. Sustain Talk, Resistance, Defensiveness, “Digging in one’s heels”
This coding category captures statements referring to the person's own arguments for not
changing, for sustaining the status quo. They may minimize their own drinking behaviors/levels
or give reasons why they do not need to change either their perceptions or their own behavior.
2. Discounting/Skepticism/Rejection of Data
This coding category can be viewed as the opposite of the believability category. It captures
statements of the various ways in which participants discount the data being presented. Examples
include skepticism regarding the source or accuracy of the data, justification for their original
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 50
beliefs, or anecdotal reflections of personal experience that lead them to reject the information
they are viewing.
3. Follow/Neutral
Statements within this category include simply restating or summarizing what they are viewing.
Tone and content reflect a degree of indifference to or disengagement from information being
presented. Scores reflect the extent to which participants are simply following along with the
presentation.
4. Believability of Data
This coding category captures statements referring to the degree to which a participant believes
the data that are presented. Statements may refer to the source of the data, the individual’s
reports of own drinking or estimates of others’ drinking, or the actual reported group norm. This
code is not to be scored unless there is explicit reference to the data in some fashion.
5. Reflective Analysis & Beliefs Modification
This coding category refers to the extent to which participants acknowledge that a shift in
perspective might be needed or attempt to actually modify existing beliefs in light of the
information being presented. Statements include a thoughtful analysis of the data in which
previously held beliefs/opinions/perspectives are reconsidered. Explanations may involve
recalling anecdotal examples that justify and are consistent with what the participant is viewing
or identifying specific reasons why original beliefs may be incorrect. Scored statements should
reflect a depth of processing that goes beyond simply summarizing what they are viewing (see,
follow/neutral) or reflect a degree of self-exploration in which participants view their own
behavior/perceptions or others’ behavior in a new light.
6. Surprise
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 51
a. The surprise (6a) coding category captures primarily positively-valanced
emotional expressions of surprise, shock, amazement, or astonishment in reaction
to the presented data. It may involve an emotional/affective reaction to a
difference between one’s perceived norm and the actual group norm.
b. The surprise (6b) coding category captures primarily negatively-valanced
emotional expressions of surprise, shock, amazement, or astonishment in reaction
to the presented data. Coded (negative) reactions of surprise may often
accompany skepticism of the data (e.g., “Wow, I definitely I don’t believe those
numbers.”). In this case, dual scores for both surprise (6b) and skepticism are
warranted.
Each 60-second ATSS think-aloud segment of participants’ verbalizations was
individually analyzed for the presence and intensity of the coding categories using a four-point
Likert scale [0 – not at all (complete absence of code); 1 – slightly/somewhat (low presence of
code); 2 – moderately (moderate presence of code); 3 – very (high presence of code; unequivocal
endorsement)]. These ratings were recorded for use in the reliability analyses (see below for
description of interrater reliability). The two coders were highly trained to ensure adequate
reliability and weekly coding meetings were held to protect against coder drift.
To resolve minor discrepancies between raters and arrive at a final score, a third-rater
adjudication procedure was used. Several transcripts were rated each week by the two
independent coders. Then during weekly coding meetings, each transcript was reviewed and the
rationale behind code assignment was discussed as a group by the two coders and the PI. The
ratings for all codes within all segments made by the two coders were reviewed with the PI until
consensus was reached between all three coders based on the definitions outlined in the coding
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 52
manual. These final scores were then used to generate a single summary score for each code, for
each participant, by summing the scores for a particular code across all segments. The same
coding procedure was applied to the control condition think-aloud segments.
Interrater Reliability for Codes
In order to evaluate the interrater reliability for the seven ATSS codes (sustain talk,
skepticism, follow/neutral, believability, reflective analysis, positive surprise, negative surprise),
intraclass correlation coefficients (ICCs) were calculated based on the coding scores obtained
from two independent RA coders after being summed across all segments. ICCs are typically
used to assess the consistency of continuous measurements and/or ratings made by two or more
observers reporting on the same quantity (Landers, 2015). Values reported in the tables below
are based on the two-way mixed model (average measures). This model was chosen due to the
fact that the two ratings came from the same two coders for each participant (Shrout & Fleiss,
1979). Table 2 reports values using the absolute agreement criterion, while Table 3 uses the
consistency criterion.
Table 2. Interrater Reliability (ICCs) Based on Absolute Agreement Criterion

Code PNF-ATSS CONTROL
Sustain 0.99 0.64
Skepticism 0.98 0.97
Follow/Neutral 0.95 0.83
Believability 0.98 0.76
Reflective Analysis 0.99 0.79
Positive Surprise 0.97 0.66
Negative Surprise 0.96 0.90

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 53
Table 3. Interrater Reliability (ICCs) Based on Consistency Agreement Criterion

Code PNF-ATSS CONTROL
Sustain 0.99 0.63
Skepticism 0.98 0.97
Follow/Neutral 0.95 0.82
Believability 0.98 0.82
Reflective Analysis 0.99 0.79
Positive Surprise 0.97 0.66
Negative Surprise 0.96 0.90

Cicchetti (1994) offers the following guidelines for interpretation of ICC inter-rater
agreement measures: Less than 0.40—poor; Between 0.40 and 0.59—fair; Between 0.60 and
0.74—good; Between 0.75 and 1.00—excellent. The results of the reliability analysis indicate
that all seven of the coding categories within the PNF-ATSS condition fell in the ‘excellent’
reliability range, while reliabilities in the control condition ranged from the ‘good’ to ‘excellent’
range. The lack of noticeable differences between the ICC values based on absolute agreement
and those based on consistency indicates that the coding disagreements were unsystematic (i.e.,
there was no main effect of coder).
CHAPTER 3: RESULTS
Analytic Plan
Means and standard deviations are described and preliminary paired-samples t-tests were
conducted to compare baseline levels of the three primary constructs to follow-up levels within
each of the conditions. Bivariate correlations were conducted between the primary variables of
interest (Total Drinks Per Week, Perceived Weekly Peer Drinking Norms, and Alcohol-related
Consequences), pre- and post-intervention. One-sample t-tests determined whether a significant
difference was present between students’ perceptions concerning other same-sex USC students’
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 54
alcohol use behaviors and the actual drinking behaviors reported by students of their same
gender.
AIM I - Overall main effects. One-way analyses of covariance (ANCOVA) were used
to examine overall main effects of the interventions with: (a) normalized post-intervention
outcome as the dependent variable (Total Drinks Per Week, Perceived Weekly Peer Drinking
Norms, and Alcohol-related Consequences); (b) the control and two intervention groups as levels
of the independent variable; and (c) the baseline outcome variable as the covariate. Follow-up
tests were conducted to evaluate pairwise differences among the estimated marginal means for
intervention conditions.
AIM I - Main effects using group coding system. Three multiple regression models
using contrast coding were specified to evaluate whether students in the intervention conditions
reported significantly greater reductions in the outcome measures, relative to each other and,
when combined, relative to the control participants from pre-intervention (baseline) to one-
month post-intervention. As discussed in many treatments of regression analysis (e.g., Cohen,
Cohen, West, & Aiken, 2003; Darlington & Hayes, 2017), a multicategorical variable with k
categories can be used as a predictor in a regression model if it is properly represented with k − 1
variables coding the groups represented by the multicategorical variable. Two contrast codes
were constructed. Contrast 1 (D1) was coded to evaluate whether the two intervention conditions
(PNF-Only, PNF+ATSS), averaged together, differed from the control condition, while contrast
2 (D2) was coded to evaluate whether the intervention conditions (PNF-ATSS, PNF-Only)
differed from one another, irrespective of control. In this model, the regression coefficient for
contrast D1 estimates the difference between the post-intervention outcome measure in the PNF-
Only (D1 = -2, D2 = –1) and PNF-ATSS (D1 = -2, D2 = 1) conditions relative to the Control
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 55
condition (D1 = 1, D2 = 0), and the regression coefficient for contrast D2 estimates the average
difference in outcomes between the PNF-only and PNF-ATSS conditions.
In each regression model, the pre-intervention score was entered on step 1 to statistically
control for baseline levels of the construct. Contrast coded condition status was entered on Step
2. On Step 3, the two, two-way interactions involving the contrast comparisons by baseline
constructs were entered. If neither interaction term emerged as significant on Step 3, any
significant contrast comparisons were interpreted on Step 2.
AIM II – Indirect effects with correlations and tests of mean differences. Prior to
conducting tests of indirect effects, correlations examined bivariate relationships between the
seven coding categories and changes from baseline to one-month follow-up on drinking and
perceived norms. Change scores for perceived norms and drinking were calculated by
subtracting baseline scores from follow up scores on the two variables, respectively. Positive
change scores indicated increases in perceived norms and/or drinking, whereas negative change
scores indicated decreases in these constructs. Next, a series of independent-samples t-tests were
conducted to compare mean scores on the various ATSS coding categories between the PNF-
ATSS and control condition. In the presence of a significant Levene’s F test, the t-test was
interpreted with equal variances not assumed.
Baron and Kenny (1986), Judd and Kenny (1981), and James and Brett (1984) discussed
four steps in establishing mediation (for reference, see Figure below):

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 56
Figure 1. Conceptual model for mediation

Step 1: Show that the causal variable is correlated with the outcome. Use Y as the criterion
variable in a regression equation and X as a predictor (estimate and test path c in the above
figure).
Step 2: Show that the causal variable is correlated with the mediator. Use M as the criterion
variable in the regression equation and X as a predictor (estimate and test path a).
Step 3: Show that the mediator affects the outcome variable. Use Y as the criterion variable in a
regression equation and X and M as predictors (estimate and test path b).
Step 4: To establish that M completely mediates the X-Y relationship, the effect of X on Y
controlling for M (path c') should be zero. The effects in both Steps 3 and 4 are estimated in the
same equation.
If all four of these steps are met, then the data are consistent with the hypothesis that
variable M completely mediates the X-Y relationship, and if the first three steps are met but the
Step 4 is not, then partial mediation is indicated. Following, Kenny, Kashy, and Bolger (1998),
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 57
one might ask whether all of the steps have to be met for there to be mediation. Most
contemporary analysts believe that the essential steps in establishing mediation are Steps 2 and 3.
Certainly, Step 4 does not have to be met unless the expectation is for complete
mediation. According to Kenny, in the opinion of most analysts, Step 1 is not required. As
described by Kenny and Judd (2014), the tests of c and c’ have relatively low power, especially
in comparison to the indirect effect.
An increasingly popular method of testing the indirect effect is bootstrapping (Bollen &
Stine, 1990; Shrout & Bolger, 2002). Bootstrapping is a non-parametric method based on
resampling with replacement which is done many times, e.g., 5000 times. From each of these
samples the indirect effect is computed and a sampling distribution can be empirically
generated. With the distribution, a confidence interval can be determined and it is checked to
determine if zero is in the interval. If zero is not in the interval, then one can be confident that the
indirect effect is different from zero.
To address the mechanisms of change research question under consideration in the
current study, an SPSS macro developed Andrew Hayes (i.e., PROCESS; see Hayes, 2013;
Hayes & Rockwood, 2017) was used to test for possible indirect effects of ATSS think-aloud
codes on intervention effects on drinking and perceived norms. Note that these confidence
intervals are "percentile" intervals which do not involve a bias correction ("accelerated
confidence intervals"), as the bias-corrected limits may have slightly elevated Type I error rates
(Fritz, Taylor, & MacKinnon, 2012; Hayes & Scharkow, 2013).
AIM III - Moderation analyses. A series of linear hierarchical regression models were
conducted to evaluate potential moderators of intervention efficacy, including pre-intervention
defensiveness, group identity, and the two orientation subscales. Due to the three-condition
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 58
nature of the IV in the current study, two variables D1 and D2 were constructed using indicator or
dummy coding: D1 = D2 = 0 for all cases in the control group (i.e., reference group), D1 = 1 and
D2 = 0 for all cases in the PNF-Only group, and D1 = 0 and D2 = 1 for all cases in the PNF-ATSS
group. Thus, the regression coefficients for D1 and D2 represent differences between group
means.
All moderation models specified the normalized outcome measure as the DV and
controlled for its respective baseline counterpart. Indicator variables and the moderator were
entered along with their product terms. Significant interactions in the presence of a statistically
significant increase in R
2
when the product terms are added to the model is affirmative evidence
for moderation. Models were run using the PROCESS macro Version 3 (Hayes, 2013) for SPSS.
Missing Data, Outliers, and Assumptions for the General Linear Model
In the overall sample, 94.8% of the participants assessed at baseline (N = 77) completed
the follow-up survey one month later (N = 73). The four non-completers were evenly split
between the PNF-Only and the control conditions. No differences were observed on primary
study variables between the students who dropped out and those who remained in the study
through follow-up (all ps > .05).
Data points with large residuals (outliers) and/or high leverage may distort the outcome
and accuracy of a regression. In particular, in regression analysis an influential point is one
whose deletion has a large effect on the parameter estimates. Cook's distance is a commonly
used estimate of the influence of a data point by measuring the effect of deleting a given
observation. Points with a large Cook's distance are considered to merit closer examination in the
analysis. Cook's distance was calculated for the main outcomes to determine the presence of
outliers prior to data analysis. Three influential data points were observed (Cook’s d > 1.3).
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 59
Upon closer examination, the three participants responded incorrectly to two attention-control
questions embedded in the survey and exhibited a pattern of random responding. Thus, they were
excluded from all analyses. The final overall sample used in analyses was N = 70. For analyses,
we did not impute missing values but rather used all available data for each specific analysis.
Thus, minor discrepancies in degrees of freedom reflect missing data.
Testing assumptions for the general linear model. Normality, homoscedasticity, multi-
collinearity, and linearity assumptions were evaluated before conducting analyses using the
General Linear Model. Appropriate transformations were explored for extreme departures from
normality.
Histograms revealed that all three outcome measures had slightly negatively skewed
distributions. In order to make valid inferences from the regression, the residuals of the
regression should follow a normal distribution. The residuals are simply the error terms, or the
differences between the observed value of the dependent variable and the predicted value.
Normal Predicted Probability (P-P) plots were used to determine if the residuals were normally
distributed. Again, slight deviations were observed. Thus, the three outcome measures were
normalized using Blom’s rank based normalization. Normality was again evaluated (see Figures
9a, 9b, 10a, 10b, 11a, 11b on Pgs. 120-127 for histograms and p-plots using the new normalized
outcomes). All three variables appeared normal following transformation, with residuals
conforming to the diagonal normality line indicated in the plot.
Next, homoscedasticity was evaluated to determine whether these residuals are equally
distributed, or whether they tend to bunch together at some values, and at other values, spread far
apart. Normalized predicted values and residuals were plotted on a scatterplot. No obvious
pattern was visible, as points were equally distributed above and below zero on the X axis, and to
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 60
the left and right of zero on the Y axis (see Figures 9c, 10c, and 11c on Pgs. 122-128), thus
meeting the assumption of homoscedasticity.
Linearity was graphically assessed by fitting linear, lowess, cubic, and quadratic lines to
the scatterplots. The predictor variables in the regression were found to be best fitted by a
straight-line with the outcome variable. Lastly, multicollinearity refers to when predictor
variables are highly correlated with each other. If this occurs, the regression model will not be
able to accurately associate variance in the outcome variable with the correct predictor variable,
leading to potentially incorrect inferences. Multicollinearity was assessed by checking variance
inflation factor (VIF) values. Each value was between 1 and 1.5 (well below the cutoff of 10),
indicating that the assumption was met.
Evaluation of normality for ATSS coding categories. Tests of normality were
evaluated for the seven final summary code scores prior to use in multi-level models. This was
done on the basis of the values for skewness and kurtosis. The results indicated that none of the
seven codes was normally distributed, as slight skewness and kurtosis were present in each of
their distributions. Thus, all seven code scores were normalized using Blom’s rank based
normalization. Normality was again evaluated and all values were found to be within the
acceptable range for skewness or kurtosis (below +1.5 and above -1.5; Tabachnick & Fidell,
2013).
Preliminary Analyses
Given the non-normality of the baseline variables, Kruskal-Wallis tests for independent
samples were conducted to evaluate differences among the three conditions (PNF-ATSS, PNF-
Only, control). Results revealed no significant differences among the three conditions on weekly
drinking, χ
2
(2) = 1.82, p = .40, alcohol-related consequences χ
2
(2) = .09, p = .96, or perceived
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 61
weekly drinking norms χ
2
(2) = .29, p = .86, whereby supporting the equivalency of intervention
and control groups at baseline.
Means and standard deviations for all of the primary outcomes are displayed in Table 1.
As denoted, asterisks next to follow-up measures indicate significant changes from baseline to
follow-up for a specific construct. Paired-samples t-tests revealed significant decreases from
baseline to follow-up for the primary outcomes within the overall sample. Notably, while the
same was true within the PNF-ATSS condition and to a slightly lesser extent for PNF-Only
participants, no mean differences were observed in the control condition.
To wit, participants within the PNF-ATSS condition reduced the amount of weekly
drinking from (M = 9.14, SD = 5.82) at baseline to (M = 6.78, SD = 5.02) at one month follow-
up, t(36) = 3.42, p = .002. They similarly reduced the amount of perceived weekly drinking
norms from (M = 11.19, SD = 5.52) at baseline to (M = 6.94, SD = 4.35) at one month follow-up,
t(34) = 3.81, p < .001. And they also demonstrated reductions in negative alcohol consequences
from baseline (M = 30.11, SD = 5.68) to follow-up (M = 28.00, SD = 4.27), t(36) = 2.97, p =
.005.
Participants within the PNF-Only condition were not found to reduce the amount of
weekly drinking from baseline (M = 6.74, SD = 5.13) to one month follow-up (M = 7.24, SD =
5.45), t(16) = -.53, p = 60. However they did evidence reductions in perceived weekly drinking
norms from (M = 12.09, SD = 7.16) at baseline to (M = 5.50, SD = 3.17) at follow-up, t(14) =
3.60, p = .003. Likewise, these participants demonstrated reductions in negative alcohol
consequences from baseline (M = 29.29, SD = 3.50) to follow-up (M = 27.63, SD = 3.56), t(15) =
2.37, p = .03.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 62
Control participants, on the other hand, were not found to have reduced their self-
reported weekly drinking from baseline (M = 8.13, SD = 5.10) to follow up (M = 6.88, SD =
5.49), t(15) = 1.91, p = .08. Nor were there significant changes from pre-intervention (M = 10.56,
SD = 4.00) to post intervention (M = 8.67, SD = 3.75) in perceived drinking norms, t(14) = 2.02,
p = .06. And lastly, no significant differences were observed on alcohol consequences from
baseline (M = 31.06, SD = 5.98) to follow-up (M = 31.81, SD = 7.76), t(15) = -.74, p = .47.
Of note, these preliminary paired t–test analyses are not a formal test of intervention
efficacy, since they does not take into account relative changes between all three conditions
controlling for baseline levels of the outcomes. Intervention main effects are evaluated elsewhere
but these initial tests reveal important trends pertaining to the interventions.
Correlations
Bivariate correlations were computed separately for males and females on the main
variables (see Table 4). Of particular interest, among males, Time 1 weekly drinking was
positively correlated with Time 1 perceived drinking norms of other male USC students (r = .32,
p < .05) and Time 2 weekly drinking (r = .73, p < .001). Among females, Time 1 weekly
drinking positively correlated with Time 1 alcohol-consequences (r = .60, p < .001), and Time 2
weekly drinking (r = .75, p < .001), but not with perceived drinking norms of other female USC
students at Time 1 (r = .16, p = ns).
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 63
Table 4
Correlation Matrix of Variables for Males (n = 34) and Females (n = 36)
Measure Mean (SD) Males Mean (SD) Females 1 2 3 4 5 6
1. T1 Weekly Drinking 8.07 (6.16) 8.56 (4.90) -- .16 .60*** .75*** .00 .41**
2. T1 Perceived Drinking Norm 11.72 (5.65) 10.83 (5.63) .32* -- .22 .25 .19 .04
3. T1 Alcohol Consequences 29.62 (5.27) 30.61 (5.30) .29 -.01 -- .43** .04 .70***
4. T2 Weekly Drinking 7.32 (5.88) 6.53 (4.43) .73*** .08 .13 -- .01 .36*
5. T2 Perceived Drinking Norm 8.69 (4.34) 5.47 (3.13) .24 .12 -.29 .45** -- .35*
6. T2 Alcohol Consequences 28.42 (5.56) 29.14 (5.18) .22 -.17 .70*** .33* -.16 --
Note . T1 refers to pre-intervention scores. T2 indicates post-intervention scores.
Correlations for males are below diagonal, correlations for females are above diagonal.
*p < .05; **p < .01; *** p < .001

Misperceptions: Perceived Peer Drinking vs. Actual Student Drinking
Male participants’ perceived norms were approximately double that of the actual male
student drinking norms at USC, across all four drinking variables (see Table 5): Average number
of drinks per occasion, one-sample t(33) = 6.67, p < .001, total number of drinks per week, one-
sample t(33) = 5.36, p < .001, maximum number of drinks consumed at any one time in the past
month, one-sample t(33) = 5.50, p < .001, and binge drinking frequency in the prior two weeks,
one-sample t(33) = 4.37, p < .001.
Female participants’ perceived norms were more than double that of the actual female
student drinking norms at USC, across all four drinking variables (see Table 5): Average number
of drinks per occasion, one-sample t(35) = 7.91, p < .001, total number of drinks per week, one-
sample t(35) = 6.93, p < .001, maximum number of drinks consumed at any one time in the past
month, one-sample t(35) = 8.58, p < .001, and binge drinking frequency in the prior two weeks,
one-sample t(35) = 6.00, p < .001.

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 64
Table 5.
Descriptive Norm
Perception of
Typical Male
Student (n = 38)
Actual
Drinking
(n = 259) t value
Perception of
Typical Female
Student (n = 37)
Actual
Drinking
(n = 379) t value
Average Number Drinks Per Occasion
4.54 (1.59) 2.76 6.67*** 3.81 (1.37) 2.00 7.91***
Total Number Drinks Per Week
11.72 (5.65) 6.53 5.36*** 10.83 (5.63) 4.33 6.93***
Maximum Number of Drinks 7.18 (2.70) 4.63 5.50*** 5.58 (1.84) 2.95 8.58***
Binge Drinking Frequency Two Weeks 2.15 (1.60) 0.95 4.37*** 2.06 (1.29) 0.77 6.00***
*p < .05; **p < .01; *** p < .001
Students' Perceptions of Typical Same-Sex USC Student Compared to Students' Actual Drinking from Campus-Wide
Survey
M (SD ) M (SD )

AIM I - Intervention Main Effects
Weekly drinking. The ANCOVA examining post-intervention weekly drinking revealed
a marginally significant main effect of condition after controlling for baseline weekly drinking,
F(2, 69) = 2.98, p = .05, ηp
2
= .08. The results showed that participants in the PNF-ATSS
condition (M = -.138) drank significantly fewer drinks per week post-intervention, controlling for
baseline drinking, than participants in the PNF-Only condition (M = .303), but not less than
participants in the control condition (M = -.009).
Alcohol-related consequences. The ANCOVA examining post-intervention alcohol-
related consequences revealed a significant main effect of condition after controlling for baseline
consequences, F(2, 68) = 3.23, p = .04, ηp
2
= .09. The results showed that participants in the
PNF-ATSS condition (M = -.067) and those in the PNF-Only condition (M = -.123) reported
fewer alcohol consequences post-intervention, controlling for baseline levels, than participants in
the control condition (M = .409).
Perceived weekly drinking norms. The ANCOVA examining post-intervention
perceived peer drinking norms revealed a significant main effect of condition after controlling
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 65
for baseline perceived norms, F(2, 64) = 3.70, p = .03, ηp
2
= .11. The results showed that
participants in the PNF-ATSS condition (M = -.061; marginal effect p = .05) and those in the
PNF-Only condition (M = -.395) held lower perceived weekly drinking norms post-intervention,
controlling for baseline perceived norms, than participants in the control condition (M = .471).
AIM I - Regression Models Evaluating Intervention Efficacy with Contrasts
See Table 6 for full details from regression models and Figures 2, 3, and 4 for
illustrations of change in primary outcomes from pre- to post-intervention using raw (non-
normalized) scores.
Model 1 – Total drinks per week. On Step 1, pre-intervention weekly drinking was
found to be significant (B = .76, p < .001), with an R
2
of .53. Step 2 revealed a significant effect
of contrast comparison 2 (B = -.22, p = .02), with an R
2
change of .04. No interactions emerged
as significant on Step 3. The final model, interpreted at Step 2, accounted for a total of 57% of
the variance in T2 student weekly drinking.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 66
Figure 2. Weekly drinking means pre- and post-intervention by condition

Model 2 – Perceived peer weekly drinking norm. On Step 1, pre-intervention
perceived peer weekly drinking norm was not found to be significant (B = .11, p = .36), with an
R
2
of .01. Step 2 revealed a significant effect of contrast comparison 1 (B = -.20, p = .02), with
an R
2
change of .09. No interactions emerged as significant on Step 3. The final model,
interpreted at Step 2, accounted for a total of 10% of the variance in T2 perceived norms.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 67
Figure 3. Perceived weekly drinking (DNRF) means pre- and post-intervention by condition

Model 3 – Alcohol-related consequences. On Step 1, pre-intervention alcohol-related
consequences was found to be significant (B = .61, p < .001), with an R
2
of .39. Step 2 revealed a
significant effect of contrast comparison 1 (B = -.20, p = .004), with an R
2
change of .07. No
interactions emerged as significant on Step 3. The final model, interpreted at Step 2, accounted
for a total of 46% of the variance in T2 alcohol-related consequences.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 68
Figure 4. Alcohol-related consequences means pre- and post-intervention by condition

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 69
Table 6. Hierarchical Regression Analyses for Three Primary Outcomes Using Contrast Coding
R
2
change
F
change R
2
total
R
2
Adj
F
total
b SE t
Model 1
T2 Weekly Drinking
Step 1 .53 .52 (1, 69) =78.06***
T1 Weekly Drinking 0.76 0.09 8.84 ***
Step 2 .04 (2, 67) = 2.83 .57 .55 (3, 67) = 29.29***
Contrast Comparison 1 0.05 0.06 0.88
Contrast Comparison 2 -0.22 0.09 -2.34 *
Model 2
T2 Perceived Weekly Drinking
Step 1 .01 .00 (1, 64) =.84
T1 Perceived Weekly Drinking 0.11 0.12 0.96
Step 2 .09 (2, 62) = 2.93 .10 .06 (3, 62) = 2.25
Contrast Comparison 1 -0.20 0.09 -2.31 *
Contrast Comparison 2 0.16 0.14 1.18
Model 3
T2 Alcohol-related Consequences
Step 1 .39 .38 (1, 68) = 43.63***
T1 Consequences 0.61 0.09 6.61 ***
Step 2 .07 (2, 66) = 4.45* .46 .44 (3, 66) = 18.99***
Contrast Comparison 1 -0.20 0.07 -2.96 **
Contrast Comparison 2 0.03 0.10 0.28
Note. T1 refers to pre-intervention scores. T2 indicates post-intervention scores.
*** p < .001; ** p < .01; *p < .05
Hierarchical Regression
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 70

AIM II – Exploring Mechanisms of Change
First, correlations were computed separately for PNF-ATSS and control conditions on
pre/post change scores for perceived norms and drinking along with scores on all seven coding
categories (see Table 7). Change scores for perceived norms and drinking were calculated by
subtracting baseline scores from follow up scores on the two variables, respectively. Positive
change scores indicated increases in perceived norms and/or drinking, whereas negative change
scores indicated decreases in these constructs.
Of particular interest, among control participants, changes in weekly drinking were
positively correlated with changes in perceived norms, such that as norms decreased, drinking
decreased (and vice versa) (r = .56, p < .05). Lower scores on negative surprise were correlated
with decreases in drinking (r = .68, p < .01). Also, among control participants, increases in
perceived norms were correlated with higher scores on the follow/neutral code (r = .66, p < .05),
whereas more believability was associated with decreases in drinking (r = -.69, p < .01).
Among participants in the PNF-ATSS condition, changes in weekly drinking was
positively correlated with changes in perceived norms, such that as norms decreased, drinking
decreased (and vice versa) (r = .38, p < .05). Higher scores on follow/neutral were associated
with increases in drinking (r = .36, p < .05), whereas higher scores on skepticism were found to
be related to decrease in drinking (r = -.57, p < .01).
Also, among PNF-ATSS participants, change scores in perceived norms was negatively
correlated with skepticism (r = -.48, p < .01) and with negative surprise (r = -.39, p < .05), such
that more skepticism and greater negative surprise were associated with decreases in perceived
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 71
norms, an unexpected finding. Conversely, higher scores on follow/neutral were associated with
increases in perceived norms (r = .44, p < .05).
Table 7
Correlation Matrix of Outcome Changes and Coding Categories for Control (n = 14) and PNF-ATSS (n = 31)
Measure 1 2 3 4 5 6 7 8 9
1. Change in total drinks per week - .38
*
.08
-.57
**
.36
*
.22 -.12 .20 -.31
2. Change in perceived norms .56
*
- .19
-.48
**
.44
*
.14 -.28 -.05
-.39
*
3. Sustain talk -.41 -.39 - .32 -.29 -.16
-.45
*
-.32 .11
4. Skepticism -.52 -.42 .20 - -.29
-.41
*
-.29 -.33
.63
**
5. Follow/Neutral .31
.66
*
-.36 -.46 - -.08
-.36
*
.03 -.05
6. Believability -.31
-.69
**
.15 .51
-.61
*
- .27 .21
-.49
**
7. Reflective analysis -.01 .22 -.10 .13 -.53 -.03 - .39
*
-.25
8. Positive surprise .11 -.12 -.36 -.38 .06 .25 -.36 - -.05
9. Negative surprise .68
**
.15 -.14 -.20 .08 -.20 -.11 .16 -
Note . Correlations for Control are below diagonal, correlations for PNF-ATSS are above diagonal.
*. Correlation is significant at the 0.05 level (2-tailed).
**. Correlation is significant at the 0.01 level (2-tailed).

Next, a series of independent-samples t-tests were conducted to compare mean scores on
the various ATSS coding categories between PNF-ATSS and control conditions (See Table 8).
Differences between the conditions on four of the seven coding categories emerged as
significant. Participants in the PNF-ATSS condition (M = 4.2, SD = 1.3) were found to articulate
greater skepticism than those in the control condition (M = 1.29, SD = 1.86), t(42.39) = 2.26, p =
.03. Participants in the PNF-ATSS condition (M = 3.06, SD = 2.59) exhibited lower scores on
follow/neutral than those in the control condition (M = 5.86, SD = 4.35), t(17.32) = 2.23, p = .04.
Participants in the PNF-ATSS condition (M = 3.10, SD = 2.53) were lower on believability
scores than those in the control condition (M = 6.64, SD = 3.89), t(18.16) = 3.12, p = .006. And
lastly, those in the PNF-ATSS condition (M = 1.13, SD = 1.61) were higher in negative surprise
than those in the control condition (M = .14, SD = .13), t(40.84) = -3.06, p = .004.

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 72
Table 8
Mean Differences on Coding Categories as a Function of Intervention Condition
ATSS Coding Category Mean SD Mean SD Independent t
Sustain Talk 3.26 5.08 1.29 2.67 -1.36
Skepticism 3.16 3.71 1.29 1.86 -2.26*
Follow/Neutral 3.06 2.59 5.86 4.35 2.23*
Believability 3.10 2.53 6.64 3.89 3.12**
Reflective Analysis 6.42 4.37 4.64 4.24 -1.27
Positive Surprise 2.00 2.46 2.00 1.84 0.00
Negative Surprise 1.13 1.61 .14 .53 -3.06**
Note . *p < .05. **p < .01. ***p < .001
Control (n = 14) PNF-ATSS (n = 31)

Evaluating Explanatory Mechanism of ATSS Codes on Drinking
Regression analysis was used to investigate the hypothesis of indirect effects from the
ATSS codes to intervention changes in drinking. The X predictor variable in all the models was
dummy-coded intervention condition, (0 = Control, 1 = PNF-ATSS). The Y criterion variable
was total drinks per week at follow up. The indirect effect (M) was the respective designated
ATSS code.
Sustain talk. Results indicated that while controlling for baseline drinking, intervention
condition was not a significant predictor of drinking at 1-month follow-up, B = -.06, SE = .19, p
= .74, or a significant predictor of sustain talk, B = .38, SE = .23, p = .10. With baseline drinking
and intervention condition included in the model, sustain talk was similarly not found to be a
significant predictor of post-intervention drinking, B = -.05, SE = .13 p = .69. These results do
not support the indirect effect hypothesis.
Skepticism. Results indicated that while controlling for baseline drinking, intervention
condition was not a significant predictor of drinking at 1-month follow-up, B = -.06, SE = .19, p
= .74, or a significant predictor of skepticism, B = .49, SE = .26, p = .06. With baseline drinking
and intervention condition included in the model, skepticism was found to be a significant
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 73
predictor of post-intervention drinking, B = -.33, SE = .10 p = .002. However, due to lack of
support for path a, these results do not support continuing the analysis for indirect effects.
Follow/Neutral. Results indicated that while controlling for baseline drinking,
intervention condition was not a significant predictor of drinking at 1-month follow-up, B = -.06,
SE = .19, p = .74, but was a significant predictor of follow/neutral, B = -.72, SE = .27, p = .01.
With baseline drinking and intervention condition included in the model, follow/neutral was
found to be a significant predictor of post-intervention drinking, B = .23, SE = .10 p = .03.
Despite lack of support for a direct effect (path c), these results support continuing the analysis
for indirect effects. Intervention condition was still not a significant predictor of post-
intervention drinking after controlling for follow/neutral, B = .10, SE = .19, p = .59.
Approximately 67% of the variance in post-drinking was accounted for by the predictors (R
2
=
.67). The indirect effect was tested using a percentile bootstrap estimation approach with 5000
samples (Shrout & Bolger, 2002), implemented with the PROCESS macro Version 3 (Hayes,
2013). These results indicated that the indirect coefficient was significant, B = -.16, SE = .07,
95% CI = -.30, -.03. Receiving the PNF-ATSS intervention was associated with less drinking at
follow-up compared to control through the indirect effect of follow/neutral. Being in the PNF-
ATSS condition was associated with lower levels of neutrality regarding the intervention
content, and lower levels of neutrality, in turn led to lower drinking at follow-up. Please see
Figure 5 below for an illustration of this indirect effects model.

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 74
Figure 5. Supported indirect effects model for follow/neutral on drinks per week with regression
coefficients and standard errors for all paths including the (total effect) of X on Y.

Believability. Results indicated that while controlling for baseline drinking, intervention
condition was not a significant predictor of drinking at 1-month follow-up, B = -.06, SE = .19, p
= .74, but was found to predict believability, B = -.92, SE = .26, p = .001. With baseline drinking
and intervention condition included in the model, however, believability was not found to be a
significant predictor of post-intervention drinking, B = -.02, SE = .11, p = .87. These results do
not support the indirect effect hypothesis.
Reflective analysis. Results indicated that while controlling for baseline drinking,
intervention condition was not a significant predictor of drinking at 1-month follow-up, B = -.06,
SE = .19, p = .74, or a significant predictor of reflective analysis, B = .48, SE = .31, p = .13.
With baseline drinking and intervention condition included in the model, reflective analysis was
similarly not found to be a significant predictor of post-intervention drinking, B = -.08, SE = .09
p = .42. These results do not support the indirect effect hypothesis.
Positive surprise. Results indicated that while controlling for baseline drinking,
intervention condition was not a significant predictor of drinking at 1-month follow-up, B = -.06,
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 75
SE = .19, p = .74, or a significant predictor of positive surprise, B = -.07, SE = .28, p = .82. With
baseline drinking and intervention condition included in the model, positive surprise was
similarly not found to be a significant predictor of post-intervention drinking, B = .17, SE = .10 p
= .092. These results do not support the indirect effect hypothesis.
Negative surprise. Results indicated that while controlling for baseline drinking,
intervention condition was not a significant predictor of drinking at 1-month follow-up, B = -.06,
SE = .19, p = .74, but was found to predict negative surprise, B = .50, SE = .23, p = .03. With
baseline drinking and intervention condition included in the model, however, negative surprise
was not found to be a significant predictor of post-intervention drinking, B = -.09, SE = .13, p =
.49. These results do not support the indirect effect hypothesis.
Evaluating Explanatory Mechanism of ATSS Codes on Perceived Norms
Regression analysis was used to investigate the hypothesis of indirect effects from ATSS
codes to intervention changes in perceived norms. The X predictor variable in all the models was
dummy-coded intervention condition, (0 = Control, 1 = PNF-ATSS). The Y criterion variable
was perceived norms at follow up. The indirect effect (M) was the respective designated ATSS
code.
Sustain talk. Results indicated that while controlling for perceived norms at baseline,
intervention condition was not a significant predictor of perceived norms at 1-month follow-up,
B = -.41, SE = .32, p = .20, or a significant predictor of sustain talk, B = .37, SE = .27, p = .17.
With baseline norms and intervention condition included in the model, sustain talk was similarly
not found to be a significant predictor of post-intervention norms, B = .06, SE = .19 p = .77.
These results do not support the indirect effect hypothesis.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 76
Skepticism. Results indicated that while controlling for perceived norms at baseline,
intervention condition was not a significant predictor of perceived norms at 1-month follow-up,
B = -.41, SE = .32, p = .20, or a significant predictor of skepticism, B = .45, SE = .25, p = .08.
With baseline norms and intervention condition included in the model, sustain talk was similarly
not found to be a significant predictor of post-intervention norms, B = -.23, SE = .20 p = .26.
These results do not support the indirect effect hypothesis.
Follow/Neutral. Results indicated that while controlling for perceived norms at baseline,
intervention condition was not a significant predictor of perceived norms at 1-month follow-up,
B = -.41, SE = .32, p = .20, but was a significant predictor of follow/neutral, B = -.62, SE = .28,
p = .04. With baseline norms and intervention condition included in the model, follow/neutral
was found to be a significant predictor of post-intervention norms, B = .48, SE = .16 p = .005.
Despite lack of support for a direct effect (path c), these results support continuing the analysis
for indirect effects. The direct intervention effect on post-intervention norms was found to
weaken after controlling for the indirect effect of follow/neutral, B = -.12, SE = .31, p = .71.
Approximately 22% of the variance in perceived norms at follow-up was accounted for by the
predictors (R
2
= .22). The indirect effect was tested using a percentile bootstrap estimation
approach with 5000 samples (Shrout & Bolger, 2002), implemented with the PROCESS macro
Version 3 (Hayes, 2013). These results indicated the indirect coefficient was significant, B = -
.30, SE = .18, 95% CI = -.71, -.01. Receiving the PNF-ATSS intervention was associated with
lower perceived norms at follow-up compared to control through the indirect effect of
follow/neutral. Being in the PNF-ATSS condition was associated with lower levels of neutrality
regarding the intervention content, and lower levels of neutrality, in turn led to lower perceived
norms at follow-up. Please see Figure 6 below for an illustration of this indirect effects model.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 77
Figure 6. Supported indirect effects model of follow/neutral on perceived norms with regression
coefficients and standard errors for all paths including the (total effect) of X on Y.

Believability. Results indicated that while controlling for perceived norms at baseline,
intervention condition was not a significant predictor of perceived norms at 1-month follow-up,
B = -.41, SE = .32, p = .20, but was a significant predictor of believability, B = -1.06, SE = .28, p
< .001. With baseline norms and intervention condition included in the model, believability was
found to be a significant predictor of post-intervention norms, B = -.44, SE = .17 p = .01. Despite
lack of support for a direct effect (path c), these results support continuing the analysis for
indirect effects. Intervention condition was found to be a significant predictor of post-
intervention norms after controlling for the indirect effect of believability, B = -.88, SE = .35, p =
.02. Approximately 19% of the variance in perceived norms at follow-up was accounted for by
the predictors (R
2
= .19). The indirect effect was tested using a percentile bootstrap estimation
approach with 5000 samples (Shrout & Bolger, 2002), implemented with the PROCESS macro
Version 3 (Hayes, 2013). These results indicated the indirect coefficient was significant, B = .46,
SE = .21, 95% CI = .10, .94. Receiving the PNF-ATSS intervention was associated with lower
perceived norms at follow-up compared to control through the indirect effect of believability.
Being in the PNF-ATSS condition was associated with lower levels of believability regarding the
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 78
intervention content, but in turn, higher levels of believability led to lower perceived norms at
follow-up. Please see Figure 7 below for an illustration of this indirect effects model.

Figure 7. Supported indirect effects model of believability on perceived norms with regression
coefficients and standard errors for all paths including the (total effect) of X on Y.

Reflective analysis. Results indicated that while controlling for perceived norms at
baseline, intervention condition was not a significant predictor of perceived norms at 1-month
follow-up, B = -.41, SE = .32, p = .20, or a significant predictor of reflective analysis, B = .39,
SE = .32, p = .23. With baseline norms and intervention condition included in the model,
reflective analysis was found to be a significant predictor of post-intervention norms, B = -.35,
SE = .15 p = .02. However, due to lack of support for path a, these results do not support
continuing the analysis for indirect effects.
Positive surprise. Results indicated that while controlling for perceived norms at
baseline, intervention condition was not a significant predictor of perceived norms at 1-month
follow-up, B = -.41, SE = .32, p = .20, or a significant predictor of positive surprise, B = -.02, SE
= .31, p = .96. With baseline norms and intervention condition included in the model, positive
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 79
surprise was similarly not found to be a significant predictor of post-intervention norms, B = -
.17, SE = .17 p = .30. These results do not support the indirect effect hypothesis.
Negative surprise. Results indicated that while controlling for perceived norms at
baseline, intervention condition was not a significant predictor of perceived norms at 1-month
follow-up, B = -.41, SE = .32, p = .20, but was found to predict negative surprise, B = .51, SE =
.21, p = .02. With baseline norms and intervention condition included in the model, however,
negative surprise was not found to be a significant predictor of post-intervention norms, B = .06,
SE = .24, p = .82. These results do not support the indirect effect hypothesis.
Evaluating Indirect Effect of Changes in Perceived Norms on Drinking
Regression analysis was used to investigate the hypothesis of an indirect effect from
changes in perceived norms immediately post-intervention to post-intervention drinking. The X
predictor variable in all the models was dummy-coded intervention condition, (0 = Control, 1 =
PNF-ATSS). The Y criterion variable was total drinks per week at follow up. The indirect effect
(M) variable was changes in perceived norms. We created change scores for perceived norms by
subtracting baseline perceived norms from perceived norms collected immediately post-
intervention and while still in the lab. Positive change scores indicated increases in perceived
norms, whereas negative change scores indicated decreases in perceived norms.
Results indicated that while controlling for baseline drinking, intervention condition was
not a significant predictor of drinking at 1-month follow-up, B = -.13, SE = .18, p = .46, but was
found to predict changes in norms immediately post-intervention, B = -.4.52, SE = .1.64, p =
.008. With baseline drinking and intervention condition included in the model, however, changes
in norms was not found to be a significant predictor of post-intervention drinking, B = .01, SE =
.02, p = .45. These results do not support the indirect effect hypothesis.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 80
AIM III - Moderation Analyses
As noted above, a series of linear hierarchical regression models were conducted to
evaluate potential moderators of intervention efficacy on total drinks per week and perceived
weekly drinking norms. Two variables D1 and D2 were constructed using indicator coding: D1 =
D2 = 0 for all cases in the control group (i.e., reference group), D1 = 1 and D2 = 0 for all cases in
the PNF-Only group, and D1 = 0 and D2 = 1 for all cases in the PNF-ATSS group. All
moderation models specified the normalized outcome measure (weekly drinking or perceived
norms) as the DV and controlled for its respective baseline counterpart. Indicator codes were
entered along with each moderator independently, followed by its product term with each
indicator variable. To avoid potentially problematic high multicollinearity with the interaction
term, the variables were centered prior to computing product terms. Moderators of interest
included pre-intervention defensiveness, identification with reference group, and the two
orientation subscales: Autonomy and Controlled.
Evaluating Moderators of Intervention Effects on Drinking
Intervention defensiveness. To test the hypothesis that pre-intervention defensiveness
moderates the relationship between study condition and post-intervention weekly drinking, a
hierarchical multiple regression analysis was conducted. The final model accounted for a
significant amount of variance in post-intervention drinking, R
2
= .592, F(6, 63) = 15.22, p <
.001. Baseline drinking was found to significantly predict post-drinking (b = .81, p < .001).
Neither regression coefficients for D1 or D2 emerged as significant (b = .99, p = .31 and b = .15,
p = .86 respectively), nor did intervention defensiveness (b = .14, p = .59). Similarly, neither
product term was found to be statistically significant (b = -.22, p = .48 and b = -.09, p = .73
respectively), ΔR
2
= .004, ΔF(2, 63) = .31, p = .73. This means that there were no significant
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 81
differences in post-intervention drinking between conditions as a function of pre-intervention
defensiveness.
Group identity. To test the hypothesis that level of identification with same-sex USC
students moderates the relationship between study condition and post-intervention weekly
drinking, a hierarchical multiple regression analysis was conducted. The final model accounted
for a significant amount of variance in post-intervention drinking, R
2
= .597, F(6, 63) = 15.55, p
< .001. Baseline drinking was found to significantly predict post-drinking (b = .84, p < .001).
Neither regression coefficients for D1 or D2 emerged as significant (b = .73, p = .34 and b = -.28,
p = .66 respectively), nor did group identification (b = .07, p = .57). Similarly, neither product
term was found to be significant (b = -.102, p = .56 and b = .03, p = .85 respectively), ΔR
2
=
.005, ΔF(2, 63) = .36, p = .70. Thus, there were no significant between-group differences in post-
intervention drinking as a function of group identity.
Controlled orientation. To test the hypothesis that controlled orientation moderates the
relationship between study condition and post-intervention weekly drinking, a hierarchical
multiple regression analysis was conducted. The final model accounted for a significant amount
of variance in post-intervention drinking, R
2
= .785, F(6, 63) = 16.80, p < .001. Baseline drinking
was found to significantly predict post-drinking (b = .80, p < .001). Neither regression
coefficients for D1 or D2 emerged as significant (b = -2.12, p = .61 and b = -.41, p = .73
respectively), nor did controlled orientation (b = .01, p = .63). Similarly, neither product term
was found to be significant (b = .04, p = .10 and b = .04, p = .80 respectively), ΔR
2
= .03, ΔF(2,
63) = 2.28, p = .11. This means that there were no significant between-group differences in post-
intervention drinking as a function of how controlled individuals were.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 82
Autonomous orientation. To test the hypothesis that degree of autonomous orientation
moderates the relationship between study condition and post-intervention weekly drinking, a
hierarchical multiple regression analysis was conducted. The final model accounted for a
significant amount of variance in post-intervention drinking, R
2
= .785, F(6, 63) = 16.84, p <
.001. Baseline drinking was found to significantly predict post-drinking (b = .78, p < .001).
Neither regression coefficients for D1 or D2 emerged as significant (b = -2.22, p = .26 and b = -
.87, p = .57 respectively), nor did autonomous orientation (b = -.02, p = .15). Similarly, neither
product term was found to be significant (b = .03, p = .20 and b = .01, p = .62 respectively), ΔR
2

= .01, ΔF(2, 63) = .92, p = .41. Thus, there were no significant between-group differences in
post-intervention drinking as a function of how autonomous individuals were.
Evaluating Moderators of Intervention Effects on Perceived Drinking Norms
Intervention defensiveness. To test the hypothesis that pre-intervention defensiveness
moderates the relationship between study condition and post-intervention perceived weekly
drinking norms, a hierarchical multiple regression analysis was conducted. The final model did
not account for a significant amount of variance in post-intervention norms, R
2
= .371, F(6, 58) =
1.54, p = .18. Baseline perceived norms was not found to significantly predict post-norms (b =
.13, p = .29). Neither regression coefficients for D1 or D2 emerged as significant (b = -1.27, p =
.45 and b = -.18, p = .90 respectively), nor did intervention defensiveness (b = .13, p = .77).
Similarly, neither product term was found to be statistically significant (b = .11, p = .84 and b = -
.12, p = .80 respectively), ΔR
2
= .008, ΔF(2, 58) = .26, p = .77. This means that there were no
significant differences in post-intervention perceived drinking norms between conditions as a
function of pre-intervention defensiveness.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 83
Group identity. To test the hypothesis that level of identification with same-sex USC
students moderates the relationship between study condition and post-intervention perceived
weekly drinking norms, a hierarchical multiple regression analysis was conducted. The final
model accounted for a marginally significant amount of variance in post-intervention norms, R
2

= .432, F(6, 58) = 2.21, p = .05. Baseline perceived norms was not found to significantly predict
post-norms (b = .14, p = .25). Neither regression coefficients for D1 or D2 emerged as significant
(b = .73, p = .54 and b = .99, p = .36 respectively). There was a marginally significant main
effect of group identification, such that when D1 and D2 are both zero (which is the code for
control) group identity significantly predicted post-intervention perceived norms (b = .07, p =
.049). Neither product term was found to be significant (b = -.38, p = .16 and b = -.36, p = .14
respectively), ΔR
2
= .036, ΔF(2, 58) = 1.30, p = .28. Thus, there were no significant between-
group differences in post-intervention perceived drinking norms as a function of group identity.
Controlled orientation. To test the hypothesis that controlled orientation moderates the
relationship between study condition and post-intervention perceived weekly drinking norms, a
hierarchical multiple regression analysis was conducted. The final model did not account for a
significant amount of variance in post-intervention norms, R
2
= .363, F(6, 58) = 1.47, p = .21.
Baseline perceived norms was not found to significantly predict post-norms (b = .16, p = .22).
Neither regression coefficients for D1 or D2 emerged as significant (b = -1.04, p = .69 and b = -
1.10, p = .63 respectively), nor did controlled orientation (b = -.002, p = .95). Similarly, neither
product term was found to be significant (b = .003, p = .95 and b = .01, p = .80 respectively),
ΔR
2
= .002, ΔF(2, 58) = .05, p = .95. Thus, there were no significant between-group differences
in post-intervention perceived drinking norms as a function of how controlled individuals were.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 84
Autonomous orientation. To test the hypothesis that degree of autonomous orientation
moderates the relationship between study condition and post-intervention perceived weekly
drinking norms, a hierarchical multiple regression analysis was conducted. The final model
accounted for a significant amount of variance in post-intervention norms, R
2
= .274, F(6, 58) =
16.84, p < .001. Baseline perceived norms did not significantly predict post-norms (b = .06, p =
.59). There was a significant main effect for D1 indicating a stronger intervention effect for the
PNF-Only participants compared to other study condition participants (b = -8.27, p = .01).
Neither D2 (b = -1.97, p = .40) nor autonomous orientation (b = -.04, p = .07) emerged as
significant. The interaction term D1Xautonomy was found to be significant (b = .08, p = .02)
while the product term for D2 was not (b = .02, p = .52), explaining an additional 10% of the
variance in post-intervention perceived norms, ΔR
2
= .103, ΔF(2, 58) = 4.12, p = .02. Figure 8
presents predicted values derived from parameter estimates where high and low values for
autonomous orientation were specified as being one standard deviation above and below the
mean, respectively (Aiken, West, & Reno, 1991; Cohen, Cohen, West, & Aiken, 2003). At high
levels of autonomy, the intervention effect on perceived norms was similar for control and PNF-
Only participants. However, the intervention effect on perceived weekly drinking norms was
stronger for PNF-Only participants compared to control, as level of autonomous orientation
decreased.

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 85
Figure 8. Autonomous orientation as a moderator of intervention efficacy on perceived weekly
drinking norms

CHAPTER 4: DISCUSSION
The present research project aimed to broaden our understanding of alcohol interventions
targeting groups at-risk for heavy and problematic alcohol use by exploring moderators of
intervention efficacy as well as applying a novel adaptation of a cognitive assessment strategy
(ATSS) to identify potential mechanisms of change. Utilizing prior research to inform
intervention content, this study included three intervention conditions: one condition involved
viewing standard PNF without ATSS and the second condition included PNF-ATSS. Both
intervention conditions were compared to each other and to an active control condition which
received alcohol-related psychoeducation derived from the NIAAA and that also included the
ATSS think-aloud component.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 86
Gordon Paul, an early influencer who helped transform clinical psychology into the
scientific practice that it is today, once asked one of the most widely cited questions about the
proper goal of a science of evidence-based interventions: “What treatment, by whom, is most
effective for this individual with that specific problem, under which set of circumstances, and
how does it come about?” (Paul, 1969, p. 44). This was an important factor in scientific
approaches to therapeutic intervention and inquiry. The question underscored the importance of
assigning and testing interventions for specific problem areas that fit the needs of individuals
based on known processes of change. As noted by Hoffman and Hayes (2019) in a recent
reflection on the future of intervention science, there has been a notable re-emphasis on the
centrality of the issue of process of change and of moderation to interventions and therapies. The
term “re-emphasis” is used here to denote the fact that ‘functional analysis’ was a foundational
element at the very heart of early behavior therapy research (see Davison 2018 for a review of
this topic) but had become de-emphasized as research on psychosocial interventions began
favoring comparison of treatment packages in their efficacy and effectiveness in alleviating
psychological problems; A.K.A. the “gold standard” of RCTs. There is nonetheless value in the
recognition that the identification of change mechanisms and moderators are first steps in
producing an adequate account of the relationships between all of the variables that can be
involved in change processes (Nock, 2007). In other words, knowledge about moderators and
mediators provides evidence from which causal and functional accounts can begin to emerge
(Kazdin, 2007; Kazdin & Nock, 2003).
Previous studies on the efficacy of PNF among adjudicated students were done in the
context of Web-based multicomponent interventions that included PNF among other intervention
content, rather than stand-alone PNF, making it difficult to ascertain which components were
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 87
most responsible for reductions in drinking. Findings from a recent meta-analysis (Dotson,
Dunn, & Bowers, 2015) suggest that computer-delivered stand-alone PNF is a promising
prevention approach for college student drinking but has minimal impact on alcohol-related
consequences. Overall, effect sizes were small but significant for alcohol use and less than small
for alcohol-related consequences. Importantly, the eight studies reviewed in the meta-analysis
were not targeting adjudicated student samples. The present study evaluated the efficacy of
stand-alone PNF provided to adjudicated students in a lab setting, compared to an active control
condition.
As anticipated, and consistent with prior studies involving general college students (e.g.,
Larimer et al., 2009) and adjudicated student samples (e.g., Doumas et al., 2009), the overall
adjudicated student sample demonstrated large misperceptions concerning the actual population-
level drinking behavior of fellow same-sex USC students, overestimating by at least double, four
different drinking behaviors: Average number of drinking per occasion, total number of drinks
per week, maximum number of drinks consumed at any one time in the past month, and binge
drinking frequency in the prior two weeks. This is particularly noteworthy given that these
students were sanctioned for receiving a campus alcohol policy violation. Despite their
involvement in alcohol consumption resulting in a sanction, these students still believe that a
typical student drinks twice as much as they drink themselves.
AIM I of the current study examined the main effects of the PNF intervention in
reducing perceived alcohol use norms of other same-sex USC college students, individuals’ own
drinking behavior, and individuals’ alcohol-related consequences. Hypothesis 1 stated the
expectation that participants in both PNF intervention conditions, independently and collectively,
would demonstrate greater reductions in their perception of typical same-sex student drinking
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 88
relative to control participants, who received psychoeducation, with no differences expected
between the two PNF conditions. Findings completely supported this hypothesis. The students
who received accurate information via PNF about typical same-sex college student drinking
reported a significant reduction in their misperceptions at the 30-day follow-up, relative to
students in the control condition, who were not found to have significantly reduced their
normative misperceptions. These main effects held whether the PNF conditions were evaluated
independently or collectively against the control. Moreover, there were no observed differences
between the two PNF conditions in magnitude or direction of the effect on perceived norms
when evaluated irrespective of control.
With respect to self-reported drinking, findings partially supported the hypothesis that
adjudicated students receiving PNF would reduce their drinking as a result of the intervention.
Hypothesis 2 stated that there would be main effects of the PNF-only condition and PNF-ATSS
condition such that participants in both intervention conditions, independently and collectively,
would demonstrate greater self-reported reductions in alcohol use one-month post-intervention
relative to control participants, again with no differences expected between the two intervention
conditions. The results showed that participants in the PNF-ATSS condition drank significantly
fewer drinks per week post-intervention, controlling for baseline drinking, than participants in
the PNF-Only condition, but not less than participants in the control condition. No differences
were found between the control and PNF-Only conditions. Similarly, when the two PNF
conditions were combined and then compared to control, no significant differences emerged.
Although reducing misperceived drinking norms is generally considered a pre-requisite to
accompanying changes in drinking, this did not occur with the PNF-only condition. If PNF alone
was sufficient to produce behavior change, then alcohol use should have been reduced in this
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 89
group as well, since there was an intervention effect on reducing overestimated normative
perceptions. Yet the intervention did not reduce alcohol use. In fact, those students who received
PNF without ATSS actually slightly increased drinking (albeit non-significantly) while control
participants slightly decreased drinking, though also with no significant within-group effects.
This pattern of findings, while at first somewhat counterintuitive, may relate to the shared
“think-aloud” methodology inherent in the PNF-ATSS and control participants’ interventions. It
is possible that stopping periodically to reflect aloud led to an added depth of processing
regarding the intervention content, thereby facilitating greater internalization and consideration
of the material presented and thus leading to a stronger influence on future behavioral decisions
regarding drinking, relative to those who were only presented PNF without such additional
opportunities/prompts. According to social norms theory, normative influence will lead to
behavior change only when “highlighted prominently in consciousness” (Cialdini & Goldstein,
2004). So, for the PNF-ATSS group, such added processing may have enabled those students to,
whether consciously or subconsciously, frame their behavior within the context of the more
modest norms being presented and allow for more active social comparisons to occur. The
control participants were confronted with rather commonplace facts about alcohol, which might
not otherwise occasion a serious reflection. But being ‘forced’ to pause and attend to their
genuine internal reactions to the information may have resulted in an intervention effect that
might otherwise not have been influential. Similarly, the inclusion of ATSS may have
occasioned greater attention and therefore greater retention of intervention material, laying the
foundation for a greater effect on the outcomes of under study. That ATSS may encourage better
memory of source material is an interesting area for future research.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 90
Finally, for Hypothesis 3 it was anticipated that participants in both PNF intervention
conditions, independently and collectively, would demonstrate greater reductions in alcohol-
related consequences relative to control participants, with no differences expected between the
two PNF conditions. In concordance with the findings on perceived norms, this hypothesis was
completely supported. Students who received PNF intervention content reported a significant
reduction in their alcohol-related consequences at the one-month follow-up relative to students in
the control condition, who were not found to have a significant decrease in alcohol problems.
The main effect on consequences from drinking was evidenced both when the PNF conditions
were evaluated independently as well as collectively against control. And, no differences were
observed between the two PNF conditions in magnitude or direction of the effect on negative
consequences, despite the fact that only participants from the PNF-ATSS condition were found
to reduce drinking.
Focusing on the shared nature of their intervention content, receiving normative feedback
may be uniquely associated with drinking in a less risky fashion. Perceived norms have been
shown to predict a composite of alcohol-related consequences even after controlling for level of
consumption (LaBrie, Hummer, Neighbors, & Larimer, 2010). This is an important finding given
the now widely-adopted focus on harm-reduction as a more realistic goal for problematic
drinking behaviors among young adults. A primary reason for the resources aimed at
understanding and intervening in student drinking is the alcohol-related problems that are
experienced by drinkers and, indirectly, their non-drinking peers and communities. Most current
alcohol treatment programs for college students include moderate drinking and harm reduction as
their primary goals. The sought-after reduction in harm, however, is nearly always viewed as a
function of reductions in alcohol use as the primary outcome variable. It is therefore clinically
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 91
significant that the results of this study support PNF as a means of reducing harm in the absence
of reductions in drinking (as is the case with the PNF-only participants).
ATSS and Mechanisms of Change
AIM II of the study entailed a wider investigation of potential change mechanisms
involved in intervention efficacy. My own advisor along with other writers argued many years
ago that research concerning itself solely with the comparison of treatments to one another takes
clinical psychological scientists away from what could and perhaps should be our primary
mission, namely, to explain why a given intervention effects beneficial change (e.g., Bandura,
1969; Davison, 1968, 1994, 1997, 2000, 2019; Goldfried, 1980; Rosen & Davison, 2003).
Understanding therapeutic processes via underlying change mechanisms can enhance attainment
of a desirable treatment goal. The ATSS procedure used in this study provided unique insight
into the phenomenology of college students being exposed to intervention content by capturing
moment-to-moment thoughts and feelings throughout the process.
Themes derived from preliminary analysis of audio recordings collected during think-
aloud segments were translated into seven coding categories, each of which was rated as being
present to a certain degree (as described below) following formal coding procedures. As noted in
the coding manual, the coding scheme was akin to a continuum of factors hypothesized to relate
to behavior change or the absence thereof. Five cognitively-oriented codes occurred on the
spectrum. If we were to view the five codes as a continuum, we anticipated that cognitions not
contributing to a change in alcohol-related outcomes would be on the low end (sustain talk,
skepticism), while cognitions that would be expected to form a commitment to reduce drinking
would be on the high end (believability, reflective analysis). The follow-neutral code was
thought to be on the middle of such a continuum.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 92
To remind the reader, the “sustain talk” coding category captured statements referring to
the person's own arguments for not changing, for sustaining the status quo. It reflected a
resistance or defensiveness to intervention content, in which participants may have minimized
their own drinking behaviors in their responses or given reasons why they need not change either
their attitudes or their drinking behavior. An example statement of sustain talk is as follows:
“I know that I am someone who, I am at that point in my life where when I go out, I want
to get drunk. I am not drinking to, I don’t like the taste of alcohol. I don’t drink wine. I
don’t even drink mixed drinks because I don’t think they taste good so there is no point in
drinking them. If I am drinking, I am drinking to get drunk so I am basically always
binge drinking and in the past two weeks that’s kind of been, and as I said I go out twice
a week, so that would be the 4 times.”
The “skepticism” coding category can be viewed as the opposite of the believability
category. It captured statements of the various ways in which participants discounted the data
being presented. Examples included skepticism regarding the source or accuracy of the data,
justification for their original beliefs, or anecdotal reflections of personal experience that led
them to reject the information they were viewing. Example statements are as follows: “I feel like
that number is skewed low because part of the 800 just don’t drink or don’t drink often” and
“people always lie on these surveys”.
These two codes were anticipated to act as barriers to changes in beliefs and drinking
behavior. Some students presented with PNF may believe that (1) the surveyed students do not
represent typical students and therefore they have little in common with them; (2) the surveyed
students were not being truthful in their responses; and/or (3) the researchers manipulated the
data by providing incorrect information in an attempt to persuade the student to change. Students
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 93
may discredit the information based on these beliefs. If a student does not believe the
intervention content that he or she receives or otherwise becomes resistant to its message, opting
instead to argue for maintaining the status quo (i.e., previous drinking patterns), then it is
unrealistic to expect positive change to occur.
When comparing the PNF-ATSS to the Control condition, no significant difference was
found between the two conditions on the degree of sustain talk evidenced across the intervention.
However, participants in the PNF-ATSS condition (M = 4.2) were found to articulate greater
skepticism of the content than those in the Control condition (M = 1.29). Interestingly, among
participants in the PNF-ATSS condition, greater skepticism was bivariately associated with
change scores representing movement toward reductions in weekly drinking as well as changes
in perceived norms, while no significant bivariate relationships emerged in the control condition.
While no indirect effects emerged from either of these two codes, it is nonetheless noteworthy
that skepticism remains a greater issue for the provision of PNF than other psychoeducational
information on drinking, but is actually negatively associated with intervention outcomes, such
that more skepticism predicted less drinking and lower perceived norms. Although speculative in
nature, it may be that skepticism articulated during the intervention was ‘healthy’ in nature,
reflective of a deeper consideration of the data and that later, following the intervention,
participants came to accept what was presented as truthful and accurate. In future research, the
construct of skepticism (as well as believability) could be added to follow-up assessments to
explore the degree to which they too may change over time to influence drinking outcomes.
The follow/neutral code was expected to reveal little in the way of understanding a
participant’s cognitive or emotional reactions to the intervention information being presented.
The coding manual suggested that the ambiguity of the articulated statements that fell in this
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 94
category did not allow for inferences regarding the extent to which a person believed the data or
what change, if any, the information would bring about in the individual. Statements within this
category included simply restating or summarizing what they were viewing. An example
statement is as follows: “As you can see I think I definitely drink more than the actual student
both in my perception and you know in actuality.” Participants in the Control condition (M =
5.86) exhibited higher scores on follow/neutral than those in the PNF-ATSS condition (M =
3.06), reflecting, perhaps, the novelty of more analytic information contained in PNF.
Interestingly, among both Control and PNF-ATSS participants, higher scores on the
follow/neutral code were predictive of less movement toward reductions in perceived norms
post-intervention, and the same was true for PNF-ATSS participants with respect to drinking.
Moreover, there were significant indirect effects from the intervention to drinking and perceived
norms outcomes via follow/neutral, in support of Hypothesis 4; Being in the PNF-ATSS
condition was associated with lower levels of neutrality regarding the intervention content, and
lower levels of neutrality, in turn led to lower drinking and lower perceived norms at follow-up.
These findings corroborate a recent study in which, during an MI-delivered brief motivational
intervention for alcohol use targeting mandated students, follow/neutral utterances (as coded by
the Motivational Interviewing Skill Code 2.0; Miller, Moyers, Ernst, & Amrhein, 2003) were
associated with increases in drinking at follow-up (Borsari et al., 2015). It is likely that, although
less subtle than sustain talk or skepticism, following along (“uh huh”) to intervention content
may represent resistance rather than engagement. Indeed, tone and content of the articulations
within the current sample reflected a degree of indifference to or disengagement from the
intervention content.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 95
Progressing forward on the continuum of factors hypothesized to relate to behavior
change are believability and reflective analysis. The believability code captured statements
referring to the degree to which a participant believed the information that was presented. An
example statement is as follows:
“Okay, yeah, this makes sense. I would say that yeah, other people who are going to the
football games probably took part in a little bit of binge drinking and other people
probably didn’t at all so the average makes sense. Yeah, I would have put a decimal, if it
half counted, because yeah, in the last two weeks I haven’t had more than 5 drinks in
awhile so this makes sense. I would say that my estimate is pretty accurate. Because
binge drinking is a little more intense than just, you know, drinking a glass of wine in the
week, so I would say this is about accurate.”
Participants in the PNF-ATSS condition (M = 3.10) articulated weaker explicit
statements of believability compared to those in the Control condition (M = 6.64). And among
control participants, more believability was bivariately associated with change scores
representing decreases in drinking. The more they believed the psychoeducational materials, the
greater the change. A significant indirect effect emerged from the intervention to perceived
norms via believability, also in support of Hypothesis 4; Being in the PNF-ATSS condition was
associated with lower levels of believability regarding the intervention content. However, the
greater the level of believability regarding the content, the lower perceived norms were at
follow-up. This finding highlights the importance of making PNF content (as well as other
psychoeducational content) as believable as possible in order to maximize intervention efficacy.
Reflective analysis was the final ‘cognitive’ code on the continuum we anticipated to
occasion a greater shift in post-intervention outcomes. Reflective analysis referred to the extent
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 96
to which participants shifted their beliefs or perspectives in light of the information presented.
Statements included a thoughtful analysis of the data in which previously held
beliefs/opinions/perspectives were reconsidered. Participants recalled anecdotal examples that
were consistent with what the participant was viewing or identified specific reasons why their
original beliefs may have been incorrect. Scored statements reflected a depth of processing that
went beyond simply summarizing what they were viewing or reflected a degree of self-
exploration between their own drinking and what they were learning from the intervention. An
example statement is as follows:
“Um… I estimated that I binge drink twice as much as the average female student. But
now… considering the information I received in this powerpoint, prior to this slide, I feel
like my perceived estimation of other people will still be over what the actual... average
is? Um ya four times binge drinking in two weeks, that’s unhealthy, like, and I definitely
should reconsider the drinking habits, my drinking habits, because binge drinking can
lead to like kidney failure, heart disease, and like um…that’s very worrisome to me cause
like I’m obviously trying to have a good time, but I don’t want that to compromise my
health later in life, or even sooner than I might expect. So…I should definitely put a limit
on how much I drink in a day, um, or a night. And yeah.”
Contrary to expectations, no differences were found between the two conditions on this
construct. Participants in both conditions articulated relatively high degrees of self-reflection
throughout the presentations. Although reflective analysis was found to negatively predict post-
intervention norms (though not drinking), the lack of an intervention effect on this construct
negated a test of indirect effects.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 97
Lastly, surprise was coded throughout the interventions. A code for “positive surprise”
attempted to capture more positively-valenced expressions of surprise, shock, amazement, or
astonishment in reaction to the presented data, while “negative surprise” captured primarily
negatively-valenced emotional expressions of surprise or shock. Negative reactions of surprise
often accompanied skepticism of the data (e.g., “Wow, I definitely don’t believe those
numbers.”) while positive reactions often accompanied believability or reflective analysis (e.g.,
“This is actually very surprising because I thought I would drink less than the average and this is
actually the opposite of what I was believing and it turns out others actually drink less than me
per occasion and that’s really shocking”).
No significant findings emerged with respect to positive surprise. Although negative
surprise was present only to a relatively small degree for both groups, participants in the PNF-
ATSS condition (M = 1.13) were higher in negative surprise than those in the Control condition
(M = .14). For control participants, less negative surprise was bivariately associated with change
scores representing decreases in drinking. Among PNF-ATSS participants, however, more
negative surprise was associated with decreases in perceived norms.
With respect to Hypothesis 5, results did not support an expected indirect effect of
reductions in perceived norms from baseline to immediately post-intervention on changes in
alcohol use one-month post-intervention. Despite the presence of large normative misperceptions
of peers’ drinking, among males, pre-intervention weekly drinking was positively correlated with
perceived drinking norms of other male USC students, but among females, this finding did not
emerge. Such a phenomenon may partially account for the smaller main effects between study
conditions as well as the null finding for an indirect effect that has been demonstrated in other
studies (e.g., Doumas et al., 2011; LaBrie et al., 2013; Walters et al., 2007).
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 98
Moderating Effects
AIM III involved investigating moderators of PNF intervention efficacy. Clinically, tests
of moderation are helpful in identifying with greater precision the types of individuals for whom
PNF may be particularly suitable, or conversely, subgroups for whom extra therapeutic effort
may be needed. Questions about which subgroups of individuals benefit most (or least) from the
interventions under consideration in the current study were investigated through tests of
moderation. Evaluated moderators of intervention effects on drinking and perceived norms were
important on a theoretical level (i.e., identification with reference group) as well as potentially
relevant to the targeted sample of adjudicated students (i.e., controlled vs. autonomous
orientations and pre-intervention defensiveness).
Contrary to expectations, results did not support that the PNF intervention would be most
effective among students who identified more closely with a typical same-sex USC student
(Hypothesis 6), among more controlled students (Hypothesis 7), or among students who were
lower in defensiveness prior to the intervention (Hypothesis 9). However, partial support was
found for Hypothesis 8, namely, that the PNF intervention would be most effective at reducing
perceived norms among less autonomous students (note, this effect was not found for drinking).
At high levels of autonomy, the intervention effect on perceived norms was similar for all
participants. However, the intervention effect on perceived weekly drinking norms was stronger
for PNF-Only participants compared to Control, as level of autonomous orientation decreased.
A high level of autonomy orientation is the initiator of goal-directed behavior. A
tendency toward greater autonomy leads people to interpret their existing situations as more
autonomy promoting, and to organize their actions on the basis of personal goals and interests
rather than controls and constraints. Being sanctioned for exercising one’s ‘autonomy’ in
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 99
alcohol-related decision making may have led to a subsequent reactance to efforts intended to
change one’s attitudes toward drinking. Conversely, individuals lower in autonomy may have
been more open to receiving information contradicting their own perceptions of peers’ behavior,
whereby fostering a readiness to adjust their perceptions following the presentation of more
modest norms.
Limitations
Careful consideration was given to the ways in which the research design strengthens
generalizability and reliability of conclusions to be drawn from the findings. Potential drawbacks
nonetheless remain. The nature of sensitive information such as alcohol use often comes with
attendant concern about the validity of self-report in this population. There is, however, a
substantial body of research supporting the validity of self-report measures of alcohol
consumption (e.g., Babor, Steinberg, Anton, & Del Boca, 2000; Chermack, Singer, & Beresford,
1998; Sobell & Sobell, 1990). Other external sources of information, such as collateral
information, are not readily available or pragmatic for assessing college drinking. Because
confidentiality enhances the reliability and validity of self-report data (Darke, 1998), participants
were continuously reminded that all data were kept in strict confidence. Despite attempts to
assure the students that their responses to the questionnaires were completely confidential, there
is always the possibility that perceived coercion to complete a mandated sanction resulted in
underreporting on drinking indicators which may influence the results, though prior research has
found little evidence of intentional bias for alcohol use reporting among mandated students
(Borsari & Muellerleile, 2009).
Secondly, students were given the option to participate in this study or receive the
standard sanction, Alcohol EDU, and it is unknown how the selected sample may have differed
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 100
in baseline characteristics from those who did not choose to participate. Comparatively small
samples sizes within two of the study conditions may have interacted with selection effects to
produce a pattern of findings that may differ if participation was mandated for all sanctioned
students.
Another potential confounding factor with respect to data validity is that adjudicated
students may have already made initial reductions in drinking before volunteering for our
research study due to having received a sanction (White, Mun, & Morgan, 2008). This outcome
would undoubtedly be favorable to student judicial affairs personnel but may have an unseen
impact in some of the statistical relationships between constructs. Having said that, this study
still produced positive change for many participants despite the potential effects of the
sanctioning process, also surely of import to university personnel. Relatedly, all incoming
students are required to take the online course Alcohol EDU, which contains a normative
feedback component. It is possible that the potency of effects to be generated by normative
feedback have been already mostly realized. Students, then, may be ‘onto the tricks’ of PNF
whereby weakening the potential for further intervention, such as those under the conditions of
the current study.
Finally, the intervention’s effects were measured for a relatively brief follow-up period of
30 days, consistent with prior studies using PNF with mandated students (Doumas et al., 2009;
2011). Because the intervention itself was brief, its effects were likely to be short-lived as well.
Although effects of Web-based personalized feedback have been shown to last for up to 6
months in college students (Neighbors et al., 2004), future longitudinal research could examine
the degree to which stand-alone PNF may attenuate drinking among adjudicated college students
over longer follow-up periods of at least 6 months. It is possible that more, not less, robust
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 101
findings may have been observed if longer-term follow-up data had been included. PNF
following a sanction may perhaps gradually accelerate a reduction in problematic drinking. Such
a maturation effect would not be apparent when outcomes are analyzed using only between-
group, post-test mean differences. To study the impact of PNF on drinking maturation,
researchers should incorporate statistical approaches that examine changes between and within
groups over repeated and longer-term follow-up periods.
Strengths and Future Directions
Despite the aforementioned areas of potential drawbacks, results of this study have
important implications for brief intervention programs targeting adjudicated college students.
Despite intervention efforts, adjudicated students remain a high-risk population for problematic
drinking on college campuses. Studies targeting change mechanisms have been underemphasized
in college-student drinking intervention literature, so the functional importance of processes of
change for many brief interventions has not been sufficiently well-established. More broadly,
associations that certify evidence-based intervention methods, such as Division 12 (clinical
psychology) of the American Psychological Association, have failed to require evidence of
processes of change linked to the underlying theoretical model and procedures deployed (Tolin,
McKay, Forman, Klonsky, & Thombs, 2015). The current study presents a proof of concept for
an adapted ATSS think-aloud methodology as a clinical science intervention tool. The ability to
capture cognitions in response to intervention content generated in real-time provides a fertile
opportunity to specify the processes of change linked to that intervention for particular problems,
persons, and contexts. Another strength of the current study design is that it included relatively
few intervention components, making it easier to infer what factors were responsible for the
observed effects.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 102
The present study also benefitted from a large sample size for the campus-wide survey
documenting accurate drinking norms of the student body. Future research may wish to
document norms among adjudicated students as well to explore whether including normative
feedback pertaining to this subgroup may confer additive effects to more general student
normative feedback as part of the sanctioning intervention.
A broad lingering question from this and other brief interventions targeting problematic
drinking among college students is how to make brief intervention content relevant enough for
someone to have the desired effect. A model of identity-based motivation (Oyserman et al.,
2017) assumes that because identity is multifaceted, which identities come to mind and what
they are taken to mean depends on the situation in which one finds themselves. People interpret
situations in ways that are consistent with whichever identity is currently on their mind and
prefer to act in ways that are identity-consistent. This premise is foundational in the provision of
PNF – invoke the behavior of one’s peers (fellow students) and hopefully this identity is salient
enough to the individual to manifest a desire to be consistent with other’s behavior. But what if
the “typical student” is not a strong enough identity to motivate sufficient movement towards
change? An alternative perspective would be to leverage the self-as-context. That is, students
may be more motivated to take steps toward a future self-goal, such as more moderate and less
risky drinking, when confronted with how heavy drinking in the present is discrepant with short-
term and longer-term goals for the self. Perhaps research showing the stability of heavy drinking
behaviors and trajectories towards such behaviors in the future might be worthwhile information
for a student. If the future identity feels connected to the current self and efforts are made to
reconcile the two, acute and enduring effects on drinking may be realized. This is of course not a
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 103
new concept, but heavily rooted in decades of therapeutic principles. Nonetheless, the idea of
self-as-context would be an interesting avenue for future research.
Concluding Remarks
Though PNF may be limited in clinical significance as a stand-alone intervention, the
observed effects on drinking are clinically relevant when PNF is examined from a public health
perspective as an approach for intervening with problematic or higher-risk students, such as
those who have been cited for violating their campus alcohol policy. These results add to the
literature examining stand-alone PNF by describing the efficacy of the intervention to reduce
alcohol-related outcomes among adjudicated college students, exploring explanatory
mechanisms and identifying moderating characteristics associated with efficacy. This can help to
inform future research and assist interventionists in designing and choosing prevention strategies
that will be maximally effective, while setting the stage for the adoption of think-aloud
methodology as a clinical science tool for not only assessment, but potentially as a contributor to
change itself.

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 104
References
Aiken, L. S., West, S. G., & Reno, R. R. (1991). Multiple regression: Testing and interpreting
interactions. Sage.
Abrams, D. E., & Hogg, M. A. (1999). Social identity and social cognition. Blackwell
Publishing.
Anderson, D. S. & Gadaleto, A. F. (2006). Results of the 2006 College Alcohol Survey.
Comparison with 2003 results, 1994 results, and baseline year. Fairfax, VA: George
Mason University, Center for the Advancement of Public Health.
Aron, A., Aron, E. N., & Smollan, D. (1992). Inclusion of other in the self scale and the structure
of the interpersonal closeness. Journal of Abnormal and Social Psychology, 63(4), 596-
612.
Babor, T. F., Steinberg, K., Anton, R., & Del Boca, F. (2000). Talk is cheap: Measuring drinking
outcomes in clinical trials. Journal of Studies on Alcohol, 61(1), 55-63.
Baer, J. S., Stacy, A., & Larimer, M. E. (1991). Biases in the perception of the drinking norms
among college students. Journal of Studies on Alcohol, 52(6), 580-586.
Bandura, A. (1969). Principles of behavior modification. New York, NY: Holt, Rinehart &
Winston.
Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social
psychological research: Conceptual, strategic and statistical considerations. Journal of
Personality and Social Psychology, 51, 1173-1182.
Beck, A.T. (1979). Cognitive theory of depression. New York: Guilford Press.
Beck, J.S. (1995). Cognitive therapy: basics and beyond. New York: Guilford Press
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 105
Berkowitz, A. D. (2004). The social norms approach: Theory, research, and annotated
bibliography. [Electronic Version] from http://www.edc.org/hec/socialnorms/theory.html.
Bollen, K. A., & Stine, R., (1990). Direct and indirect effects: Classical and bootstrap estimates
of variability. Sociological Methodology, 20, 115-40.
Borsari, B., Apodaca, T. R., Jackson, K. M., Mastroleo, N. R., Magill, M. Barnett, N. P., &
Carey, K. B. (2015). In-session processes of brief motivational interventions in two trials
with mandated college students. Journal of Consulting and Clinical Psychology, 83(1),
56-67.
Borsari, B., & Carey, K. (2003). Descriptive and injunctive norms in college drinking: A meta-
analytic integration. Journal of Studies on Alcohol, 64(3), 331-341.
Borsari, B. & Muellerleile, P. (2009). Collateral reports in the college setting: A meta-analytic
integration. Alcoholism: Clinical & Experimental Research, 33(5), 826 – 838.
10.1111/j.1530-0277.2009.00902.x.
Carey, K. B. (2012). Brief Motivational Interventions. In: Correia, C. J., Murphy, J. G., Barnett,
N. P., editors. College student alcohol abuse: A guide to assessment, intervention, and
prevention. New York, NY: Wiley; p. 218-245.
Carey, K. B., Scott-Sheldon, L. A., Carey, M. P., & DeMartini, K. S. (2007). Individual-level
interventions to reduce college student drinking: A meta-analytic review. Addictive
Behaviors, 32, 2469–2494.
Chermack, S. T., Singer, K., & Beresford, T. P. (1998). Screening for alcoholism among medical
inpatients: How important is corroboration of patient self-report? Alcoholism: Clinical
Experimental Research, 22(7), 1393-1398.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 106
Cialdini, R. B., & Goldstein, N. J. (2004). Social influence: Compliance and conformity. Annual
Reviews in Psychology, 55(1), 591–621. doi: 10.1146/annurev.psych.55.090902.142015
Cicchetti, D. V. (1994). Guidelines, criteria, and rules of thumb for evaluating normed and
standardized assessment instruments in psychology. Psychological Assessment, 6(4),
284–290.
Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2003). Applied Multiple
Regression/Correlation Analysis for the Behavioral Sciences (3rd edition). Mahwah, NJ:
Erlbaum.
Collins, R. L., Parks, G. A., & Marlatt, G. A. (1985). Social determinants of alcohol
consumption: The effects of social interaction and model status on the self-administration
of alcohol. Journal of Consulting and Clinical Psychology, 53(2), 189-200.
Cronce, J. M., & Larimer, M. E. (2011). Individual-focused approaches to the prevention of
college student drinking. Alcohol Research & Health: The Journal of the National
Institute on Alcohol Abuse and Alcoholism, 34, 210–221.
Darke, S. (1998). Self-report among injecting drug users: a review. Drug and alcohol
dependence, 51(3), 253-263.
Darlington, R. B., & Hayes, A. F. (2017). Regression analysis and linear models: Concepts,
applications, and implementation. New York: The Guilford Press.
Davison, G. C. (1968). Systematic desensitization as a counter-conditioning process. Journal of
Abnormal Psychology, 73(2), 91-99.
Davison, G. C. (1994). Issues in psychotherapy as the practice of psychology. Journal of
Psychotherapy Integration, 4, 39–45.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 107
Davison, G. C. (1997). The mutual enrichment of basic and applied research in psychological
science and practice. Psychological Science, 8, 194–197.
Davison, G. C (2000). Stepped care: Doing more with less. Journal of Consulting and Clinical
Psychology, 68, 580–585.
Davison, G. C. (2019). A return to functional analysis, the search for mechanisms of change, and
the nomothetic-idiographic issue in psychosocial interventions. Clinical Psychological
Science, 7, 51-53.
Davison, G. C., Robins, C., & Johnson, M. K. (1983). Articulated thoughts during simulated
situations: A paradigm for studying cognition in emotion and behavior. Cognitive
Therapy and Research, 7(1), 17-39.
Davison, G. C., Vogel, R. S., & Coffman, S. G. (1997). Think-aloud approaches to cognitive
assessment and the articulated thoughts in simulated situations paradigm. Journal of
consulting and clinical psychology, 65(6), 950.
deCharms, R. (1968). Personal causation: The internal affective determinants of behavior. New
York: Academic Press.
Deci, E. L. (1980). The psychology of self-determination. Lexington, MA: Heath (Lexington
Books).
Deci, E. L., & Ryan, R. M. (1985a). The General Causality Orientations Scale: Self-
determination in personality. Journal of Research in Personality, 19, 109–134.
Deci, E. L., & Ryan, R. M. (1985b). Intrinsic motivation and self-determination in human
behavior. New York: Plenum.
Deci, E. L., & Ryan, R. M. (Eds.). (2002). Handbook of self-determination research. New York:
University of Rochester Press.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 108
DeJong, W., Schneider, S. K., Towvim, L. G., Murphy, M. J., Doerr, E. E., Simonsen, N. R., et
al. (2006). A multisite randomized trial of social norms marketing campaigns to reduce
college student drinking. Journal of Studies on Alcohol, 67(6), 868-879.
DeJong, W., Schneider, S., Towvim, L., Murphy, M., Doerr, E., Simonsen, N., et al. (2009). A
multisite randomized trial of social norms marketing campaigns to reduce college student
drinking: A replication failure. Substance Abuse, 30(2), 127-140.
Di Blasio, P., & Milani, L. (2008). Computer-mediated communication and persuasion:
Peripheral vs. central route to opinion shift. Computers in Human Behavior, 24, 798–815.
Dimeff, L. A., Baer, J. S., Kivlahan, D. R., & Marlatt, G. A. (1999). Brief Alcohol Screening and
Intervention for College Students (BASICS): A harm reduction approach. New York,
NY: Guilford Press.
Dotson, K. B., Dunn, M. E., & Bowers, C. A. (2015). Stand-alone personalized normative
feedback for college student drinkers: A meta-analytic review, 2004 to 2014. PLoS ONE,
10(10), 17.
Doumas, D. M., Haustveit, T., & Coll, K. M. (2010). Reducing heavy drinking among first year
intercollegiate athletes: A randomized controlled trial of web-based normative feedback.
Journal of Applied Sport Psychology, 22, 247–261.
Doumas, D. M., McKinley, L. L., & Book, P. (2009). Evaluation of two Web-based alcohol
interventions for mandated college students. Journal of Substance Abuse Treatment,
36(1), 65-74. doi:10.1016/j.jsat.2008.05.009
Doumas, D. M., Workman, C., Smith, D., & Navarro, A. (2011). Reducing high-risk drinking in
mandated college students: Evaluation of two personalized normative feedback
interventions. Journal of Substance Abuse Treatment, 40(4), 376-385.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 109
doi:10.1016/j.jsat.2010.12.006
Ellis, A., and Dryden, W. (1997). The Practice of Rational Emotive Behavior Therapy (2
nd
ed.).
New York: US Springer Publishing Co.
Fabiano, P. (1999, July). Learning lessons and asking questions about college social norms
campaigns. Paper presented at the National Conference on the Social Norms Model:
Science Based Prevention, Big Sky, MT.
Festinger, L. (1954). A theory of social comparison processes. Human Relations, 7, 117-140.
Fishbein, M., & Ajzen, I. (1975). Belief, attitudes, intention, and behavior. An introduction to
theory and research. Massachussets: Addison-Wesley.
Fritz, M. S., Taylor, A. B., & MacKinnon, D. P. (2012). Explanation of two anomalous results in
statistical mediation analysis. Multivariate behavioral research, 47(1), 61-87.
Fromme, K. & Corbin, W. (2004). Prevention of heavy drinking and associated negative
consequences among mandated and voluntary college students. Journal of Consulting
and Clinical Psychology, 72(6), 1038-1049. doi: 10.1037/0022-006X.72.6.1038
Goldfried, M. R. (1980). Toward the delineation of therapeutic change principles. American
Psychologist, 35, 991–999.
Granfield, R. (2002). Believe it or not: Examining to the emergence of new drinking norms in
college. Journal of Alcohol and Drug Education, 47(2), 18-31.
Grossbard, J. R., Hummer, J. F., LaBrie, J. W., Pedersen, E. R., & Neighbors, C. (2009). Is
substance use a team sport? Team attitudes, perceived norms, and alcohol and marijuana
use among male and female intercollegiate athletes. Journal of Applied Sport Psychology,
21, 247-261.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 110
Hayes, A. F. (2013). Introduction to mediation, moderation, and conditional process analysis: A
regression-based approach. Guilford Press, New York, NY.
Hayes, A. F., & Rockwood, N. J. (2017). Regression-based statistical mediation and moderation
analysis in clinical research: Observations, recommendations, and implementation.
Behaviour research and therapy, 98, 39-57.
Hayes, A. F., & Scharkow, M. (2013). The relative trustworthiness of inferential tests of the
indirect effect in statistical mediation analysis: Does method really matter?.
Psychological science, 24(10), 1918-1927.
Heyman, R.E. & Slep, A.M. (2004) Analogue Behavioral Observation. In: Haynes, S. N. and
Heiby, E. (Eds.). Behavioral Assessment. New York: Wiley Hingson, R., Heeren, T.,
Winter, M., & Wechsler, H. (2005). Magnitude of alcohol-related mortality and
morbidity among US college students ages 18-24: Changes from 1998 to 2001. . Annual
Review of Public Health, 26, 259-279.
Hofmann, S. G., & Hayes, S. C. (2019). The future of intervention science: Process-based
therapy. Clinical Psychological Science, 7, 37–50. doi:10.1177/2167702618772296
Hingson, R. W. (2010). Magnitude and prevention of college drinking and related problems.
Alcohol Research and Health, 33(1-2), 45-54.
Hingson, R., Heeren, T., Winter, M., & Wechsler, H. (2005). Magnitude of alcohol-related
mortality and morbidity among US college students ages 18-24: Changes from 1998 to
2001. Annual Review of Public Health, 26, 259-279.
Hingson, R., Zha, W., & Weitzman, E. R. (2009). Magnitude of and trends in alcohol-related
mortality and morbidity among U.S. college students ages 18-24, 1998-2005. Journal of
Studies on Alcohol and Drugs, Suppl. 16, 12–20.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 111
Hsu, K. J., Babeva, K. N., Feng, M. C., Hummer, J. F., & Davison, G. C. (2014). Experimentally
induced distraction impacts cognitive but not emotional processes in think-aloud
cognitive assessment. Frontiers in Psychology, 5(474), 1-9.
Hummer, J. F. & Davison, G. C. (2016). Examining the role of source credibility and reference
group proximity on personalized normative feedback interventions for college student
alcohol use: A randomized laboratory experiment. Substance Use and Misuse.
Hummer, J. F., LaBrie, J. W., Lac, A., & Louie, B. (2013). The influence of reflective opposite-
sex norms and importance of opposite-sex approval on adjudicated student drinking:
Theoretical extensions and implications. Journal of Student Affairs Research and
Practice, 50(4), 373-392.
Hummer, J. F., LaBrie, J. W., & Pedersen, E. R. (2012). First impressions on the scene: The
influence of the immediate reference group on incoming first-year students’ alcohol
behavior and attitudes. Journal of College Student Development, 53(1), 149-162.
James, L. R., & Brett, J. M. (1984). Mediators, moderators and tests for mediation. Journal of
Applied Psychology, 69, 307-321.
Judd, C. M., & Kenny, D. A. (1981). Process analysis: Estimating mediation in treatment
evaluations. Evaluation Review, 5, 602-619.
Kazdin, A. E. (2007). Mediators and mechanisms of change in psychotherapy research. Annual
Review of Clinical Psychology, 3, 1-27. doi:10.1146/annurev.clinpsy.3.022806.091432
Kazdin, A. E., & Nock, M. K. (2003) Delineating mechanisms of change in child and adolescent
therapy: methodological issues and research recommendations. Journal of Child
Psychology and Psychiatry, 44, 1116–1129.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 112
Kenny, D. A., Kashy, D. A., & Bolger, N. (1998). Data analysis in social psychology. In D.
Gilbert, S. Fiske, & G. Lindzey (Eds.), The handbook of social psychology (Vol. 1, 4th
ed., pp. 233-265). Boston, MA: McGraw-Hill.
Kenny, D. A., & Judd, C. M. (2014). Power anomalies in testing mediation. Psychological
Science, 25, 334-339.
Korcuska, J. S., & Thombs, D. L. (2003). Gender role conflict and sex-specific drinking norms:
Relationships to alcohol use in undergraduate women and men. Journal of College
Student Development, 44(2), 204-216.
Klar, Y., & Giladi, E. E. (1997). No one in my group can be below the group's average: a robust
positivity bias in favor of anonymous peers. Journal of personality and social
psychology, 73(5), 885.
Knee, C. R., & Neighbors, C. (2002). Self-determination, perception of peer pressure, and
drinking among college students. Journal of Applied Social Psychology, 32, 522–543.
LaBrie, J. W. (2010). Heads UP, a harm reduction approach to problematic drinking prevention
and intervention: A manualized treatment program. Charleston, SC: BookSurge
Publishing.
LaBrie, J. W., Hummer, J. F., Kenney, S., Lac, A., & Pedersen, E. R. (2011). Identifying factors
that increase the likelihood for alcohol-induced blackouts in the prepartying context.
Substance Use and Misuse, 46, 992-1002.
LaBrie, J., Hummer, J., Neighbors, C., & Larimer, M. (2010). Whose opinion matters? The
relationship between injunctive norms and alcohol consequences in college students.
Addictive Behaviors, 35(4), 343-349.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 113
LaBrie, J. W., Lewis, M. A., Atkins, D. C., Neighbors, C., Zheng, C., Kenney, S. R., . . .
Larimer, M. E. (2013). RCT of web-based personalized normative feedback for college
drinking prevention: Are typical student norms good enough? Journal of Consulting and
Clinical Psychology, 81, 1074–1086.
LaBrie, J. W., Tawalbeh, S., & Earleywine, M. (2006). Differentiating adjudicated from
nonadjudicated freshmen men: The role of alcohol expectancies, tension, and concern
about health. Journal of College Student Development, 47(5), 521-533.
Landers, R. N. (2015). Computing intraclass correlations (ICC) as estimates of interrater
reliability in SPSS. The Winnower, 2:e143518.81744.
Larimer, M., & Cronce, J. (2007). Identification, prevention, and treatment revisited: Individual-
focused college drinking prevention strategies 1999-2006. Addictive Behaviors, 32(11),
2439-2468.
Larimer, M. E., Kaysen, D., Lee, C. M., Lewis, M. A., Dillworth, T., Montoya, H. D., et al.
(2009). Evaluating level of specificity of normative referents in relation to personal
drinking behavior. Journal of Studies on Alcohol and Drugs, Supplement, 70(1 suppl.),
115-121.
Larimer, M. E., Neighbors, C., LaBrie, J., Atkins, D. C., Lewis, M. A., Lee, C. M., . . . Walter, T.
(2011). Descriptive drinking norms: For whom does reference group matter? Journal of
Studies on Alcohol and Drugs, 72(5), 833-843.
Larimer, M. E., Turner, A. P., Anderson, B. K., Fader, J. S., Kilmer, J. R., Palmer, R. S., &
Cronce, J. M. (2001). Evaluating a brief alcohol intervention with fraternities. Journal of
Studies on Alcohol, 62(3), 370-380.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 114
Larimer, M., Turner, A., Mallett, K., & Geisner, I. (2004). Predicting drinking behavior and
alcohol-related problems among fraternity and sorority members: Examining the role of
descriptive and injunctive norms. Psychology of Addictive Behaviors, 18(3), 203-212.
Latane, B. (1981). The psychology of social impact. American Psychologist, 36(4), 343-356.
Lewis, M. A., & Neighbors, C. (2004). Gender-specific misperceptions of college student
drinking norms. Psychology of Addictive Behaviors, 18(4), 334-339.
Lewis, M. A., & Neighbors, C. (2005). Self-determination and the use of self-presentation
strategies. The Journal of Social Psychology, 145(4), 469-490.
Lewis, M., & Neighbors, C. (2006). Social norms approaches using descriptive drinking norms
education: A review of the research on personalized normative feedback. Journal of
American College Health, 54(4).
Lewis, M. A., & Neighbors, C. (2007). Optimizing personalized normative feedback: The use of
gender-specific referents. Journal of Studies on Alcohol and Drugs, 68(2), 228-237.
Marlatt, G. A., Baer, J. S., Kivlahan, D. R., Dimeff, L. A., Larimer, M. E., Quigley, L. A., et al.
(1998). Screening and brief intervention for high-risk college student drinkers: Results
from a 2-year follow-up assessment. Journal of Consulting and Clinical Psychology,
66(4), 604-615.
Martens, M. P., Smith, A. E., & Murphy, J. G. (2013). The efficacy of single-component brief
motivational interventions among at-risk college drinkers. Journal of Consulting and
Clinical Psychology, 81, 691–701.
Mastroleo, N. R., Murphy, J. G., Colby, S. M., Monti, P. M., & Barnett, N. P. (2011). Incident-
specific and individual-level moderators of brief intervention effects with mandated
college students. Psychology of Addictive Behaviors, 25(4), 616-624.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 115
doi:10.1037/a0024508
Miller, M. B., Leffingwell, T., Claborn, K., Meier, E., Walters, S., & Neighbors, C. (2013).
Personalized feedback interventions for college alcohol misuse: An update of walters &
neighbors (2005). Psychology of Addictive Behaviors,
Miller, W. R., Moyers, T. B., Ernst, D., & Amrhein, P. (2003). Manual for the motivational
interviewing skills code (MISC) Version 2.0. Retrieved April 19, 2009, from
http://www.motivationalinterview.org/training/MISC2.pdf
Miller, W. R., Rollnick, S. Motivational Interviewing, Third Edition: Helping People Change.
Guilford Publication; 2013.
Neighbors, C., Dillard, A. J., Lewis, M. A., Bergstrom, R. L., & Neil, T. A. (2006). Normative
Misperceptions and Temporal Precedence of Perceived Norms and Drinking. Journal of
Studies on Alcohol, 67(2), 290-299.
Neighbors, C., LaBrie, J. W., Hummer, J. F., Lewis, M. A., Lee, C. M., Desai, S., Kilmer, J. R.,
& Larimer, M. E. (2010). Group identification as a moderator of the relationship between
perceived social norms and alcohol consumption. Psychology of Addictive Behaviors,
24(3), 522-528.
Neighbors, C., Larimer, M. E., & Lewis, M. A. (2004). Targeting misperceptions of descriptive
drinking norms: Efficacy of a computer-delivered personalized normative feedback
intervention. Journal of Consulting and Clinical Psychology, 72(3), 434-447.
Neighbors, C., Lee, C., Lewis, M., Fossos, N., & Larimer, M. (2007). Are social norms the best
predictor of outcomes among heavy-drinking college students?. Journal of Studies on
Alcohol and Drugs, 68(4), 556-565.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 116
Neighbors, C., Lewis, M. A., Bergstrom, R. L., & Larimer, M. E. (2006). Being controlled by
normative influences: Self-determination as a moderator of a normative feedback alcohol
intervention. Health Psychology, 25(5), 571-579.
Neighbors, C., Lewis, M. A., Atkins, D. C., Jensen, M. M., Walter, T., Fossos, N., . . . Larimer,
M. E. (2010). Efficacy of web-based personalized normative feedback: A two-year
randomized controlled trial. Journal of Consulting and Clinical Psychology, 78, 898–911.
NIAAA, 2007. What Colleges Need to Know Now: An Update on College Drinking Research.
Bethesda, MD: National Institutes of Health, DHHS, (NIH publication no. 07–5010).
Nock, M. K. (2007). Conceptual and design essentials for evaluating mechanisms of change.
Alcoholism: Clinical and Experimental Research, 31 (Suppl. 3), 4S-12S.
doi:10.1111/j.1530-0277.2007.00488.x
Outside the Classroom. (2010). AlcoholEdu
®
for college. Needham, MA: Outside the Classroom.
Oyserman, D., Lewis, 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,
139–147. doi:10.1080/1047840X.2017.1337406
Palmer, R. S. (2004). Efficacy of the Alcohol Skills Training Program in mandated and non-
mandated heavy drinking college students.
Palmer, R. S., Kilmer, J. R., Ball, S. A., & Larimer, M. E. (2010). Intervention defensiveness as
a moderator of drinking outcome among heavy-drinking mandated college students.
Addictive behaviors, 35(12), 1157-1160.
Paul, G. L. (1969). Behavior modification research: design and tactics. In C. M. Franks (Ed.),
Behavior therapy: Appraisal and status (pp. 29-62). New York: McGraw-Hill.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 117
Perkins, H. (2002). Social norms and the prevention of alcohol misuse in collegiate contexts.
Journal of Studies on Alcohol Supplement, 14, 164-172.
Perkins, H. W., & Craig, D. W. (2006). A successful social norms campaign to reduce alcohol
misuse among college student-athletes. Journal of Studies on Alcohol, 67(6), 880-889.
Perkins, H., Haines, M., & Rice, R. (2005). Misperceiving the college drinking norm and related
problems: A nationwide study of exposure to prevention information, perceived norms
and student alcohol misuse. Journal of Studies on Alcohol, 66(4), 470-478.
Perkins, H. W., LinkenBach, J. W., Lewis, M. A., & Neighbors, C. (2010). Effectiveness of
social norms media marketing in reducing drinking and driving: A statewide campaign.
Addictive Behaviors, 35, 866-874.
Petty, R. E., & Cacioppo, J. T. (1981). Issue involvement as a moderator of the effects on
attitude of advertising content and context. Advances in Consumer Research, 8, 20–24.
Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of persuasion. Advances
in Experimental Social Psychology, 19, 123–205.
Petty, R. E., Wells, G. L., & Brock, T. C. (1976). Distraction can enhance or reduce yielding to
propaganda: Thought disruption versus effort justification. Journal of Personality and
Social Psychology, 34, 874–884.
Porter, J. R. (2006). Arrests for alcohol rise on campuses. Chronicle of Higher Education,
53(10), A40.
Rehm, J., Gmel, G., Sempos, C. T., & Trevisan, M. (2003). Alcohol-related morbidity and
mortality. Alcohol Res. Health, 140, C00-C97.
Reis, J., & Riley, W. (2000). Predictors of college students' alcohol consumption: Implications
for student education. Journal of Genetic Psychology, 161(3), 282.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 118
Rice, C. (2007). Misperception of college drinking norms: Ethnic/Race differences. Journal of
Human Behavior in the Social Environment, 14(4), 17-30.
Rimal, R., & Real, K. (2003). Understanding the influence of perceived norms on behaviors.
Communication Theory, 13(2), 184-203.
Rimal, R., & Real, K. (2005). How behaviors are influenced by perceived norms: A test of the
theory of normative social behavior. Communication Research, 32(3), 389-414.
Rosen, G. M., & Davison, G. C. (2003). Psychology should identify empirically supported
principles of change (ESPs)and not credential trademarked therapies or other treatment
packages. Behavior Modification, 27, 300–312.
Ryan, R. M. (1989). The Revised General Causality Orientation Scale. Unpublished manuscript,
University of Rochester.
Schulenberg, J. E., .Johnston, L. D., O'Malley, P. M., Bachman, J. G., Miech, R. A., & Patrick,
M. E. (2018). Monitoring the Future national survey results on drug use, 1975-2017:
Volume II, college students and adults ages 19-55. Ann Arbor, MI: Institute for Social
Research, The University of Michigan.
Sharkin, B. S. (2007). Against their will? Counseling mandated students. Special populations in
college counseling: A handbook for mental health professionals, 143-155.
Shrout, P. E., & Bolger, N. (2002). Mediation in experimental and nonexperimental studies: new
procedures and recommendations. Psychological methods, 7(4), 422.
Shrout, P., & Fleiss, J. (1979). Intraclass correlations: Uses in assessing rater reliability.
Psychological Bulletin, 86, 420-428.
Sobell, L. C., & Sobell, M. B. (1990). Self-report issues in alcohol abuse: State of the art and
future directions. Behavioral Assessment, 12(1), 77-90.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 119
Tabachnick, B .G. & Fidell, L. S. (2013). Using multivariate statistics. Pearson, Boston.
Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup conflict. The social
psychology of intergroup relations, 33(47), 74.
The National Center on Addiction and Substance Abuse (2007). Wasting the Best and the
Brightest: Substance Abuse at America’s Colleges and Universities. New York, NY:
Commission on Substance Abuse at Colleges and Universities, Columbia University.
Thombs, D. L., Dotterer, S., Olds, R. S., Sharp, K. E., & Raub, C. G. (2004). A close look at why
one social norms campaign did not reduce student drinking. Journal of American College
Health, 53(2), 61-68.
Tolin, D. F., McKay, D., Forman, E. M., Klonsky, E. D., & Thombs, B. D. (2015). Empirically
supported treatment: Recommendations for a new model. Clinical Psychology: Science
and Practice, 22, 317 – 338. doi:10.1111/cpsp.12122
Tropp, L. R., & Wright, S. C. (2001). Ingroup identification as the inclusion of ingroup in the
self. Personality and Social Psychology Bulletin, 27(5), 585-600.
Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987).
Rediscovering the social group: A self-categorization theory. Basil Blackwell.
Walters, S., & Neighbors, C. (2005). Feedback interventions for college alcohol misuse: What,
why and for whom?. Addictive Behaviors, 30(6), 1168-1182.
Walters, S. T., Vader, A. M., Harris, T. R. (2007). A controlled trial of web-based feedback for
heavy drinking college students. Prevention Science, 8, 83–88.
Welch, K. A., Carson, A., & Lawrie, S. M. (2013). Brain Structure in Adolescents and Young
Adults with Alcohol Problems: Systematic Review of Imaging Studies. Alcohol and
Alcoholism, 48(4), 433-444.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 120
White, A., & Hingson, R. (2014). The burden of alcohol use: Excessive alcohol consumption and
related consequences among college students. Alcohol Research: Current Reviews, 35(2),
201-218.
White, H. R., & Labouvie, E. W. (1989). Towards the assessment of adolescent problem
drinking. Journal of Studies on Alcohol, 50(1), 30-37.
White, H. R., Mun, E. Y., & Morgan, T. J. (2008). Do brief personalized feedback interventions
work for mandated students or is it just getting caught that works? Psychology of
Addictive Behaviors, 22, 107−116.
Williams, G. C., & Deci, E. L. (1998). The importance of supporting autonomy in medical
education. Annals of Internal Medicine, 129, 303–308.
Williams, G. C., Deci, E. L., & Ryan, R. M. (1995). Building health-care partnerships by
supporting autonomy: Promoting maintained behavior change and positive health
outcomes. In P. Hinton-Walker, A. L. Suchman, & R. Botehlo (Eds.), Partnerships,
power, and process: Transforming health care delivery. New York: University of
Rochester Press.
Windle, M. (2016). Drinking over the lifespan: Focus on early adolescents and youth. Alcohol
Research: Current reviews, 38(1), 95-101.
Zanov, M. V., & Davison, G. C. (2010). A conceptual and empirical review of 25 years of
cognitive assessment using the articulated thoughts in simulated situations (ATSS) think-
aloud paradigm. Cognitive Therapy and Research, 34(3), 282-291.
Zuckerman, M., Gioioso, C., & Tellini, S. (1988). Control orientation, self-monitoring, and
preference for image versus quality approach to advertising. Journal of Research in
Personality, 22, 89–100.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 121

Figure 9a. Histogram of normalized total drinks per week outcome variable

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 122

Figure 9b. P-P plot for normalized total drinks per week outcome variable

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 123

Figure 9c. Scatterplot plot for normalized total drinks per week outcome variable

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 124

Figure 10a. Histogram of normalized perceived weekly drinking outcome variable

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 125

Figure 10b. P-P plot of normalized perceived weekly drinking outcome variable

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 126

Figure 10c. Scatterplot plot for normalized perceived weekly drinking outcome variable

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 127

Figure 11a. Histogram of normalized alcohol-consequences outcome variable

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 128

Figure 11b. P-P plot of normalized alcohol-consequences outcome variable

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 129

Figure 11c. Scatterplot of normalized alcohol-consequences outcome variable

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 130

Appendix A

Example Feedback Slides

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 131

How do you compare to other USC students?

AVERAGE DRINKS PER OCCASION
You reported in the last
month…

• You have an average
of 5 drinks per
occasion.

• You think the typical
USC student has an
average of 7 drinks
per occasion.

According to students
surveyed…

• The typical USC
student has about
2.4 drinks per
occasion.*

You = Your drinking behavior.

Perceived = Your estimate of the typical USC student’s
drinking behavior.

Actual = Actual drinking behavior of USC students.*

*Note: This information comes from a 2015 campus-wide survey which
included a random sample of 6,000 USC students.
0
1
2
3
4
5
6
7
8
You Perceived Actual
Total # of Drinks per Occasion
Drinks per Occasion
5
7
2.4

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 132

How do you compare to other USC students?

DRINKS PER WEEK

You reported in the last
month…

• You have an average of
24 drinks per week.

• You think the typical
USC
student has an average
of 35 drinks per week.

According to students
surveyed…

• The typical USC
student has about 2.4
drinks per week.*

You = Your drinking behavior.

Perceived = Your estimate of the typical USC
student’s drinking behavior.

Actual = Actual drinking behavior of USC students.*

*Note: This information comes from a 2015 campus-wide survey which
included a random sample of 6,000 USC students.
0
5
10
15
20
25
30
35
40
You Perceived Actual
Total # of Drinks per Week
Drinks per Week
24
35
2.4

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 133

Appendix B

PNF Intervention Script

Introduction and Example Feedback Slide:

Audio:
Recently, you completed an online survey that asked about your own alcohol use, as well as your
estimates of other USC college students’ drinking behavior. We have compiled your responses
into graphical feedback that you will be viewing. In addition, we will provide you with feedback
about the actual drinking behavior of USC students.

Please take a moment to follow along with an example feedback slide. This slide will describe
how to interpret all the following slides.

As the example slide is built, one component at a time, arrows will point to the appropriate
object that the audio is referring to.

First, the top of the slide will provide you with information about how your responses on the
earlier online survey compare to other USC students. Next, notice the top of the graph. Here you
will see the drinking behavior that the graph is referring to. This will be one of three drinking
behaviors: the average number of drinks consumed on a typical drinking occasion, the total
number of drinks consumed per week, or the maximum amount of drinks consumed at any one
time in the past month.

Then, you will see two bars appear in the graphs. The first bar is your own drinking behavior,
exactly as you reported it on the recent online survey. The second bar represents your estimate
of a typical USC student, also exactly as you reported it on the survey.

Next, you will see a third bar that represents the actual drinking behavior of a typical USC
college student. At the bottom of the slide, you will see a note that informs you of the source of
the data for the actual group norm. The statistics were collected from a very large representative
sample of USC students who completed surveys during 2015. They were ensured that their
responses would be anonymous and never associated with their names. Thus, the statistics are
very accurate and reliable.

As you are viewing the complete feedback slide, please pay attention to how your own alcohol
use, and your perceptions, is similar to or different from the average behavior of other college
students. Consider whether your estimates were too high or too low and how your own alcohol
use compares to the average use of a typical USC student.

You will now be viewing feedback slides using data from yourself and other USC college
students.

Any participant questions will be answered.

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 134

Appendix C

PNF Intervention Script
ATSS Condition

(Differences from the main script are bolded/italicized):

Audio:
Recently, you completed an online survey that asked about your own alcohol use, as well as your
estimates of other USC college students’ drinking behavior. We have compiled your responses
into graphical feedback that you will be viewing. In addition, we will provide you with feedback
about the actual drinking behavior of USC students.

Please take a moment to follow along with an example feedback slide. This slide will describe
how to interpret all the following slides.

As the example slide is built, one component at a time, arrows will point to the appropriate
object that the audio is referring to.

First, the top of the slide will provide you with information about how your responses on the
earlier online survey compare to other USC students. Next, notice the top of the graph. Here you
will see the drinking behavior that the graph is referring to. This will be one of three drinking
behaviors: the average number of drinks consumed on a typical drinking occasion, the total
number of drinks consumed per week, or the maximum amount of drinks consumed at any one
time in the past month.

Then, you will see two bars appear in the graphs. The first bar is your own drinking behavior,
exactly as you reported it on the recent online survey. The second bar represents your estimate
of a typical USC student, also exactly as you reported it on the survey.

Next, you will see a third bar that represents the actual drinking behavior of a typical USC
college student. At the bottom of the slide, you will see a note that informs you of the source of
the data for the actual group norm. The statistics were collected from a very large representative
sample of USC students who completed surveys during 2015. They were ensured that their
responses would be anonymous and never associated with their names. Thus, the statistics are
very accurate and reliable.

Often, when people are presented with new or interesting information, they have a kind of
internal monologue going through their heads, a constant stream of thoughts or feelings,
which reflect their reactions to what they are viewing or reading. What we’d like you to do is
to verbalize these reactions as they are happening. Every so often the recording will stop, you
will hear a tone, and you will be asked to speak into a microphone for 30 seconds. Simply say
out loud whatever is going through your mind. Say as much as you can until you hear another
tone. Of course, there are no right or wrong answers, so please just say whatever comes to
mind without judging whether it is appropriate or not. The more you can tell us, the better.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 135

Everything that you say will be completely confidential. Your name will not be associated with
the recording in any way.

As you are viewing the complete feedback slide, please pay attention to how your own alcohol
use, and your perceptions, is similar to or different from the average behavior of other college
students. Consider whether your estimates were too high or too low and how your own alcohol
use compares to the average use of a typical USC student.

You will now be viewing feedback slides using data from yourself and other USC college
students.

Any participant questions will be answered.

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 136

Appendix D

Control Intervention
Script and Content

(Highlighted text appears on the slides for students to view as well)

College Drinking: What Students Need to Know
Part I: Information for College Students about Alcohol
*Next Slide (2)*
Voiceover: “The following information was obtained from the National Institute on Alcohol
Abuse and Alcoholism of the National Institutes of Health, and Centers for Disease Control and
Prevention.”
“As a young adult, you've probably had the opportunity to drink alcohol on many different
occasions.”
“The statistics are clear, consistent, and hardly surprising: People tend to drink the heaviest in
their late teens and early- to mid-20s. But here are the sobering facts: Just about every college
student is adversely affected in some way by alcohol.”
“College students are at higher risk for dangerous behaviors such as binge drinking that can
lead to tragic consequences. Each year in colleges and universities across the United States,
alcohol contributes to more than 1,800 student deaths.”
*Next Slide (3)*
Risks associated with college drinking and problems related to alcohol on college campuses
include the following:
“Academic and disciplinary problems—such as missed classes and assignments, poor or failing
grades and probation, or even expulsion. Impaired judgment, and physical and sexual assaults
can also result from alcohol abuse. Depression and other mental health issues, as well as health
problems--such as effects on the brain, liver, heart, stomach and other organs are other possible
problems related to alcohol.”
“Legal problems associated with assaults, DUIs, vandalism, and public drunkenness are also
associated with college drinking. Unsafe sex such as unprotected sex and inability to consent due
to intoxication; may lead to STDs including HIV, unwanted pregnancy and lower self-esteem.”
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 137

*Insert think aloud prompt here*
*Next Slide (4)*
Part II: Effects of Alcohol
Voiceover: “Intoxication occurs when alcohol is consumed at a rate faster than the body can
metabolize it and builds up in the bloodstream.”
“Alcohol affects many parts of the body and, especially when consumed in excess, some effects
are unpleasant, harmful, progressive—and even dangerous.”
*Next Slide (5)*
Voiceover: “Effects of Alcohol Intoxication Include: Dehydration, Erratic or unusual behavior,
Flushing of the face and neck, Impaired balance, Slurred speech, Uncoordinated movement and
Vomiting.”
*Insert think aloud prompt here*
*Next Slide (6)*
Part III: What Is Alcohol Poisoning?
Voiceover: “Alcohol acts as a central nervous system depressant. It affects breathing, heart rate
and involuntary muscle responses, such as the gag reflex. High levels of blood alcohol can lead
to a condition called "acute alcohol poisoning," which is a serious medical emergency that can
lead to brain damage, coma and death. Alcohol poisoning usually results from drinking a large
amount of alcohol in a short amount of time.”
*Next slide (7)*
Voiceover: “This slide lists some of the common signs of alcohol poisoning. If you suspect that
someone has alcohol poisoning, call for help immediately. Blood alcohol concentration (BAC)
in the bloodstream can continue to rise after the person has stopped drinking—even after he or
she has passed out. In rapid binge drinking, it's possible to ingest a fatal dose of alcohol before
becoming unconscious.”
*Insert think aloud prompt here*
*Next Slide (9)*
Part IV: College Drinking: Common Myths about Alcohol
Voiceover: “The following slides list some common myths about alcohol.”
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 138

“It is a common myth that Drinking alcohol isn’t that dangerous. But in
Fact, each year among college students, alcohol use is associated with the following: 1 out of
every 3 emergency room visits, Almost 600,000 injuries and more than 1,800 deaths. It is also
responsible for about 700,000 assaults, (97,000 of which are sexual assaults) and more than
150,000 health problems.”
*Next Slide (10)*
Voiceover: “Another common myth believed by many is that “I can "handle" alcohol—I’m in
control when I drink. However, Alcohol is a drug that affects your central nervous system
and changes how your brain functions. It affects your perception, thinking and coordination and
impairs your judgment. Drinking reduces inhibitions and increases high-risk behavior. Most
people who drink alcohol report having done something while drinking that they later regret.”
*Next Slide (11)*
Voiceover: “The myth that “I can drive safely after a couple drinks” is clearly false.
Approximately half of all fatal car crashes in young adults between the ages of 18 and 24
involve alcohol. Don't drink and drive. And never get into a car with a driver who has been
drinking. If you are under the age of 21, it is illegal to drive with any amount of alcohol in
your blood.”
*Next Slide (13)*
Voiceover: “The belief that If necessary, I can sober up from the effects of alcohol quickly.”
Is a myth, as it takes time for your body to metabolize alcohol. How fast that happens depends on
several factors, including your weight and how much alcohol is in your system. Generally, it
takes about 3 hours for your body to metabolize 2 drinks. Drinking coffee, water or energy
drinks; going for a walk; sleeping; or taking a cold shower will not speed up the process—it will
only result in a cold, wet, and alert drunk.
“Also, women metabolize alcohol differently than men do. A woman who drinks the same
amount as a man will be more intoxicated and more impaired—even if they both weigh about the
same.”
*Next Slide (15)*
Voiceover: “Some believe the myth that It's better if my body "gets used to" drinking.
However, really the opposite is true. If you need to drink more and more to feel the effects of the
alcohol, you're developing a tolerance. Tolerance is a warning sign of alcohol dependence and a
serious problem with drinking. Tolerance does not affect BAC.
*Insert think aloud prompt here*
*Next Slide (16)*
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 139

Part V: How to Avoid Problems with Alcohol in College
Voiceover: “If you’re in college, it’s important to get information about your school’s policies
regarding alcohol on campus. If you have concerns about alcohol, seek guidance from your
parents, your school’s counseling services, or another trusted person.”
*Insert think aloud prompt here*

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 140

Appendix E

Participant Pre-Intervention (Baseline) Survey

Age:
Sex: 0. Female 1. Male
Class: 1. Freshmen 2. Sophomore 3. Junior 4. Senior 5. Other:

Current Location of Residence:
1. On-Campus Housing

2. Off-Campus w/ Roommates
3. Off-Campus w/primary parent(s)/guardian(s)
4. Off-Campus Alone
5. Off-Campus Fraternity/Sorority House

Are you currently a member of a fraternity or sorority? 0 No 1 Yes

If no, do you intend to join a fraternity or sorority? 0 No 1 Yes

Ethnic Identification:
1. Asian
2. Hawaiian/Pacific Islander
3. Native American/Alaska Native
4. African American/Black
5. Caucasian
6. Hispanic/Lationo(a)
7. Mixed:
8. Other:

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 141

Group Identity (inclusion of in-group to the self):

If A represents yourself and B represents a typical [gender] USC student, please select the pair of circles
that you feel best represents your own level of identification with a typical [gender] USC student. For
example, if you don’t identify at all with a typical [gender] USC student, you would select 1. If you
identify a great deal, you would select 6 or 7.

1. 1
2. 2
3. 3
4. 4
5. 5
6. 6
7. 7

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 142

Alcohol Use:

The following questions ask you to recall some of your behaviors over the past month (30 days).
Please do your best to approximate your alcohol use during the time assessed. Use a calendar to help
aid your memory.

For the following two questions regarding # of drinks, one drink equals:
• 12 oz. beer (8 oz. Canadian beer, malt liquor, or ice beers or 10 oz. of microbrew)
• 10 oz. of wine cooler
• 4 oz. of wine
• 1 oz. of 100 proof or 1 1/4 oz. of 80 proof liquor (one shot)
• 1 cocktail with 1 oz. of 100 proof or 1 1/4 oz. of 80 proof liquor

Consider a typical week during the past month (30 days). How much alcohol, on average,
(measured in number of drinks), did you drink on each day of a typical week?

Drinks
Monday
Tuesday
Wednesday
Thursday
Friday
Saturday
Sunday

In the PAST MONTH (30 days):
On average, how many days per week did you drink alcohol?

On average, how many days during the month did you drink alcohol?

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 143

On average, how many drinks did you have each time you drank?

What is the maximum number of drinks you drank at any one time?

Binge Drinking
Think back over the last two weeks. How many times have you had 4 or more drinks in a two hour
period? (For females only)

Think back over the last two weeks. How many times have you had 5 or more drinks in a two hour
period? (For males only)

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 144

Rutgers Alcohol Problem Index (RAPI):

Use the scale below to rate how often you have had any of the following problems over the PAST ONE
MONTH as a result of drinking alcoholic beverages. Please select your answers.

Never
1-2
times
3-5
times 6-9 times
10 or
more
times
Not able to do your homework or study for a test
Got into fights, acted bad, or did mean things
Missed out on other things because you spent too much money on alcohol
Went to work or school drunk
Caused shame or embarrassment to someone
Neglected your responsibilities
Relatives avoided you
Felt that you needed more alcohol than you used to use in order to get the same
effect

Tried to control your drinking by trying to drink only at certain times
Had withdrawal symptoms, that is, felt sick because you stopped or cut down on
drinking

Noticed a change in your personality
Felt that you had a problem with school
Missed a day (or part of a day) of school or work

Tried to cut down on drinking

Suddenly found yourself in a place you could not remember getting to

Passed out or fainted suddenly

Had a fight, argument, or bad feelings with a friend

Had a fight, argument, or bad feelings with a family member

Kept drinking when you promised yourself not to

Felt you were going crazy

Had a bad time

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 145

Felt physically or psychologically dependent on alcohol

Was told by a friend or neighbor to stop or cut down on drinking

Had a hangover or felt sick {added to survey}

Perceived Norms

For the following two questions regarding # of drinks, one drink equals:
• 12 oz. beer (8 oz. Canadian beer, malt liquor, or ice beers or 10 oz. of microbrew)
• 10 oz. of wine cooler
• 4 oz. of wine
• 1 oz. of 100 proof or 1 1/4 oz. of 80 proof liquor (one shot)
• 1 cocktail with 1 oz. of 100 proof or 1 1/4 oz. of 80 proof liquor

Consider a typical week during the past month (30 days). How much alcohol, on
average, (measured in number of drinks), did you think a typical [gender] USC
drank on each day of a typical week?

Drinks
Monday
Tuesday
Wednesday
Thursday
Friday
Saturday
Sunday

Consider the occasion a typical [gender] USC student drank the most in the past month (30 days). How
many drinks did they consume on that occasion? ______ drinks

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 146

Intervention Defensiveness

Items
1. I am genuinely interested in the alcohol awareness research meeting I will be attending (R)
2. It is my choice to attend the alcohol awareness research meeting. (R)
3. I am interested in knowing more about my drinking. (R)
4. I would be interested in learning how my drinking compares to other students. (R)
5. I am open minded about the alcohol awareness research meeting. (R)
6. Attending the alcohol awareness research meeting will be a waste of my time.
7. The alcohol awareness research meeting might benefit me. (R)
8. I am not like the people the alcohol awareness research meeting was designed for.
9. I might make some changes in my drinking as a result of attending alcohol awareness
research meeting. (R)
10. The alcohol awareness research meeting may have information that will be useful for me.
(R)
11. I have no reason to think about how much I drink.

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 147

Appendix F

Participant Post-Intervention Survey

Alcohol Use:

The following questions ask you to recall some of your behaviors over the past month (30 days).
Please do your best to approximate your alcohol use during the time assessed. Use a calendar to help
aid your memory.

For the following two questions regarding # of drinks, one drink equals:
• 12 oz. beer (8 oz. Canadian beer, malt liquor, or ice beers or 10 oz. of microbrew)
• 10 oz. of wine cooler
• 4 oz. of wine
• 1 oz. of 100 proof or 1 1/4 oz. of 80 proof liquor (one shot)
• 1 cocktail with 1 oz. of 100 proof or 1 1/4 oz. of 80 proof liquor

Consider a typical week during the past month (30 days). How much alcohol, on average,
(measured in number of drinks), did you drink on each day of a typical week?

Drinks
Monday
Tuesday
Wednesday
Thursday
Friday
Saturday
Sunday

In the PAST MONTH (30 days):
On average, how many days per week did you drink alcohol?

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 148

On average, how many days during the month did you drink alcohol?

On average, how many drinks did you have each time you drank?

What is the maximum number of drinks you drank at any one time?

Binge Drinking
Think back over the last two weeks. How many times have you had 4 or more drinks in a two hour
period? (For females only)

Think back over the last two weeks. How many times have you had 5 or more drinks in a two hour
period? (For males only)

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 149

Rutgers Alcohol Problem Index (RAPI):
Use the scale below to rate how often you have had any of the following problems over the PAST ONE
MONTH as a result of drinking alcoholic beverages. Please select your answers.
Never 1-2
times
3-5
times
6-9 times 10 or
more
times
Not able to do your homework or study for a test
Got into fights, acted bad, or did mean things
Missed out on other things because you spent too much money on alcohol
Went to work or school drunk
Caused shame or embarrassment to someone
Neglected your responsibilities
Relatives avoided you
Felt that you needed more alcohol than you used to use in order to get the
same effect

Tried to control your drinking by trying to drink only at certain times
Had withdrawal symptoms, that is, felt sick because you stopped or cut down on
drinking

Noticed a change in your personality
Felt that you had a problem with school
Missed a day (or part of a day) of school or work

Tried to cut down on drinking

Suddenly found yourself in a place you could not remember getting to

Passed out or fainted suddenly

Had a fight, argument, or bad feelings with a friend

Had a fight, argument, or bad feelings with a family member

Kept drinking when you promised yourself not to

Felt you were going crazy

Had a bad time

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 150

Felt physically or psychologically dependent on alcohol

Was told by a friend or neighbor to stop or cut down on drinking

Had a hangover or felt sick {added to survey}

Perceived Norms:

For the following two questions regarding # of drinks, one drink equals:
• 12 oz. beer (8 oz. Canadian beer, malt liquor, or ice beers or 10 oz. of microbrew)
• 10 oz. of wine cooler
• 4 oz. of wine
• 1 oz. of 100 proof or 1 1/4 oz. of 80 proof liquor (one shot)
• 1 cocktail with 1 oz. of 100 proof or 1 1/4 oz. of 80 proof liquor

Consider a typical week during the past month (30 days). How much alcohol, on
average, (measured in number of drinks), did you think a typical [gender] USC
drank on each day of a typical week?

Drinks
Monday
Tuesday
Wednesday
Thursday
Friday
Saturday
Sunday

Consider the occasion a typical [gender] USC student drank the most in the past month (30 days). How
many drinks did they consume on that occasion? ______ drinks

ATSS AND PERSONALIZED NORMATIVE FEEDBACK 151

Appendix G

ATSS Coding Manual

The coding scheme is akin to a continuum of factors hypothesized to relate to behavior change or
the absence thereof. Five cognitively-oriented codes occur on the spectrum. If we were to assign
the five codes a numerical designation on the continuum, they might occur as follows: The
continuum goes from cognitions that would not contribute to a change in attitude towards
drinking and to a reduction in drinking to cognitions that would be expected to form a
commitment to reduce drinking.

1. Sustain Talk, Resistance, Defensiveness, “Digging in one’s heels”
This coding category captures statements referring to the person's own arguments for not
changing, for sustaining the status quo. They may minimize their own drinking behaviors/levels
or give reasons why they do not need to change either their perceptions or their own behavior.
2. Discounting/Skepticism/Rejection of Data
This coding category can be viewed as the opposite of the believability category. It captures
statements of the various ways in which participants discount the data being presented. Examples
include skepticism regarding the source or accuracy of the data, justification for their original
beliefs, or anecdotal reflections of personal experience that lead them to reject the information
they are viewing.
3. Follow/Neutral
Statements within this category include simply restating or summarizing what they are viewing.
Tone and content reflect a degree of indifference to or disengagement from information being
presented. Scores reflect the extent to which participants are simply following along with the
presentation.
4. Believability of Data
This coding category captures statements referring to the degree to which a participant believes
the data that are presented. Statements may refer to the source of the data, the individual’s
reports of own drinking or estimates of others’ drinking, or the actual reported group norm. This
code is not to be scored unless there is explicit reference to the data in some fashion.
5. Reflective Analysis & Beliefs Modification
This coding category refers to the extent to which participants acknowledge that a shift in
perspective might be needed or attempt to actually modify existing beliefs in light of the
information being presented. Statements include a thoughtful analysis of the data in which
previously held beliefs/opinions/perspectives are reconsidered. Explanations may involve
recalling anecdotal examples that justify and are consistent with what the participant is viewing
or identifying specific reasons why original beliefs may be incorrect. Scored statements should
reflect a depth of processing that goes beyond simply summarizing what they are viewing (see,
follow/neutral) or reflect a degree of self-exploration in which participants view their own
behavior/perceptions or others’ behavior in a new light.
6. Surprise
a. The surprise (6a) coding category captures primarily positively-valanced
emotional expressions of surprise, shock, amazement, or astonishment in reaction
to the presented data. It may involve an emotional/affective reaction to a
difference between one’s perceived norm and the actual group norm.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 152

b. The surprise (6b) coding category captures primarily negatively-valanced
emotional expressions of surprise, shock, amazement, or astonishment in reaction
to the presented data. Coded (negative) reactions of surprise may often
accompany skepticism of the data (e.g., “Wow, I definitely I don’t believe those
numbers.”). In this case, dual scores for both surprise (6b) and skepticism are
warranted.

Codes 1 (defensiveness, sustain talk) and 2 (discounting/skepticism of data) are
hypothesized to act as barriers to changes in beliefs and behavior. Some students presented with
PNF may believe that (1) the surveyed students do not represent typical students and therefore
they have little in common with them, (2) the surveyed students were not being truthful in their
responses, and/or (3) the researchers manipulated the data. Students may discredit the
information based on these beliefs. If a student does not believe the normative data that he or she
receives, then a presumed active mechanism of change (modifying one’s own behavior to
conform to the lower, more modest actual norm) is unrealistic to expect. The theory of cognitive
dissonance (Festinger, 1957) predicts that discounting the credibility of the normative feedback
allows heavy drinking students to continue their level of drinking without experiencing the sense
of conflict elicited by the knowledge that they are deviating from the prevailing norm.
Code 3 (F/N) reveals little in the way of understanding a participant’s cognitive or
emotional reactions to the information being presented. The ambiguity of the articulated
statements do not allow for inferences regarding the extent to which a person believes the data or
what change, if any, the information would bring about in the individual. It merely reflects that
the person understands/remembers the information.
Progressing forward on the continuum of factors hypothesized to relate to behavior
change are codes 4 (believability of data) and 5 (reflective analysis and beliefs modification).
Changing one’s beliefs regarding perceived drinking norms is considered a prerequisite for social
norms interventions to impact behavior. It is reasonable to expect that the more “cognitive shift”
we see on the continuum (i.e., articulated thoughts receiving scores on codes 4 and 5), as well as
higher scores on the positively emotionally-valanced code of “surprise” (6a), the greater
likelihood for an accompanying shift in behavior (i.e., reductions in perceived norms and
individual drinking behavior) one-month post-intervention. However, greater articulations of
negative surprise (6b), along with defensiveness and skepticism, have a greater likelihood of
being associated with the maintenance of one’s behavioral drinking patterns and beliefs (i.e.,
little to no change).
It should be noted that the eight segments are independently coded of one another, so it is
possible for participants to obtain scores for divergent codes at various points throughout the
presentation.

Please use the following 4-point Likert scale to rate the categories described below.

0 – not at all (complete absence of code)
1 – slightly/somewhat (low/minimal presence of code; may contain qualifying adjectives such as
“a little bit”, “maybe”, “I guess”, “probably”, “kind of”, etc.)
2 – moderately (moderate presence of code; more than minimal endorsement)
3 – very (high presence of code; unequivocal endorsement that may contain qualifying adjectives
such as “very” “a lot” “definitely”, etc.)]
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 153

Please take into consideration the tone of voice and other verbal cues of the verbalizations
when coding. These factors help us understand the meaning of what the person is saying.
The context in which a statement is made may affect scoring.

Coding Categories:

1. Sustain Talk, Resistance, Defensiveness, “Digging in one’s heels”
This coding category captures statements referring to the person's own arguments for not
changing, for sustaining the status quo. They may minimize their own drinking
behaviors/levels or give reasons why they do not need to change either their perceptions
or their own behavior.

Examples
• These include statements of the following:
i. Inability to change
ii. Commitment not to change
iii. Desire not to change
iv. Lack of need to change or need to not change
v. Reasons not to change
• I feel like 8 shots is normal but after 8 shots is too much (1)
• I mean I know that this sounds like I am making excuses. 2 extra drinks a week, I feel
like, probably aren’t necessary on my end, but not that scary of a difference. (1)
• I feel like the amount of drinks I have is pretty safe for how much my body can handle
and like I said, it is typically lower. (1)
• So yeah, that makes a lot of sense and I am not too worried about the difference between
mine and that because yeah, it’s like 1 drink of a difference or I will plan ahead for it. I
will eat more. I will drink a lot of water. (1)
• Yeah, I mean, I don’t know, I was anticipating higher than that and I don’t know if that is
indicative of how much I am drinking, but I mean I am drinking about as much as I think
that would be fun while also being safe. I mean, I have only really had one incident
before so it’s been working pretty well so far. (1)
• Just kind of knowing myself, I know, as someone who likes to go out the amount I said
before, 2, 3 times a week which isn’t even so so much, I know I am definitely on the
further end of that spectrum when there are so many people who don’t drink at all so if
we are talking about average that is going to play a huge role and as someone who does
take a lot more alcohol to reach the same point as many people when they take less (1)
• I would consider myself, it is weird to say heavy drinker because it is not like every time
I go out I get black out drunk and I never throw up from drinking and obviously, I am
here because I made a mistake and went too hard one time, but when I am taking like 6,
7, 8 shots, I am not getting wasted, I am literally the same amount as many people when
they are take 3, 4, 5 so my perception is the average person doesn’t take as much to get
drunk so if they are going out with the same amount, less, they are going to have less. (2)
• Um, well I feel like for me usually I go for about 5 drinks per occasion simply because
um, you know, every time I do end up going out I cannot really feel the effects until I’ve
had about 4 to 5 shots, any less, and I would feel as though I, you know, I didn’t really
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 154

drink, I would not usually drink if I were not pursuing a specific you know, like feeling
of tipsiness. (3)
• something for me is I don’t drink because I think other people are drinking and that’s
never been, I don’t drink as much I drink because I think other people are drinking that
much and honestly, I don’t think that’s a fair representation, but in terms of alcohol use,
the fact that I am taking, as I said, an average of 8 drinks per occasion when people are
drinking 2. They honestly, maybe not 8 and 2, but 8 and 4, are affecting us the same way
so my taking 8 drinks, I know personally, is the same as my friends taking 3, 4. (3)
• I mean yeah, obviously I am here because I drank too much and obviously I know my
maximum of that very high number is a lot, but again, I feel like that versus my
perception, you know, my body, I know, I have always been a huge, huge heavy weight,
between that and drinking there and drinking at a pre-game, whatever, depending, I eat a
lot of food, it takes me a lot to get drunk so I think me having a maximum of 10 drinks
and obviously that was really bad is the same amount as many people having 4, 5, 6, and
that would be the same amount for them. (3)
• I know that I am someone who, I am at that point in my life where when I go out, I want
to get drunk. I am not drinking to, I don’t like the taste of alcohol. I don’t drink wine. I
don’t even drink mixed drinks because I don’t think they taste good so there is no point in
drinking them. If I am drinking, I am drinking to get drunk so I am basically always
binge drinking and in the past two weeks that’s kind of been, and as I said I go out twice
a week, so that would be the 4 times, but I think most people don’t, especially girls, don’t
go out to get drunk especially for obviously general so I can’t put the number as low as 1.
(3)
• honestly, it is not really something that I am looking to change. It is more to the extent
and not having become problematic like again, why I am here, but besides that, twice a
week, 4 times in 2 weeks. That’s kind of it. (3)

2. Discounting/Skepticism/Rejection of Data
This coding category can be viewed as the opposite of the believability category. It captures
statements of the various ways in which participants discount the data being presented. Examples
include skepticism regarding the source or accuracy of the data, justification for their original
beliefs, or anecdotal reflections of personal experience that lead them to reject the information

Examples
• Yeah it’s a little bit hard to believe (1)
• Even though this is based on data, it’s hard to believe I guess (1)
• It seems as if some of the data is conflicting (1)
• I also think, in terms of all of these things, it is like the average of female USC
students drink a maximum of 2.95 students. That is averaging out all, in addition to
all of the people that don’t drink whatsoever and I think it is hard to kind of take that
into account, when, it is not that they are irrelevant, but if you are going to do a
survey about people drinking and talking about drinking behavior, you obviously
have to account for the people that don’t drink, but it is weird to implement the thing
as a whole because obviously the number is going to be lowered a lot as opposed to
for people who do drink, how much do they drink. (1)
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 155

• I’m gonna be honest, this was not within the last month, more like within the
semester, it’s definitely because ever since the incident, I have not been drinking all
that much, but since the beginning of the semester, and last year, my maximum
number of drinks in one occasion was about 9-10. And I played it safe a little assume
you know that the maximum number of drinks were uh your average student would
be about 5, so yeah. It might be more, it might be less (1)
• Of course, I would like to see like the standard deviation on this cause you know
there’s a lot of people who straight up just don’t binge drink. I don’t think it’s that
common. This might be like you know some people binge drinking like multiple
times a week, which is horrible, but you know that’s their loss. (1)
• Generally I have not seen male students drink that little. Maybe for female students
(2)
• I feel like the actual number isn’t that accurate because on the last slide it said
average is more than 4 shots and when I go to parties most people drink more than 3
shots in general so a max of 2.95 is too low. (2)
• I think a factor that this survey does not account for is who you are drinking with bc
you don’t want to drink 3 or 4 drinks with people you don’t know. It really depends
on the crowd of people you are drinking with. (2)
• When I go to other parties or events I see other people drinking more (2)
• I feel like that number is skewed low because part of the 800 just don’t drink or don’t
drink often (2)
• I think I drink less than usc female students bc im light weight and so I don’t drink a
lot compared to others (2)
• Once again, it would be really interesting to see the demographics of the sample and
like, how the survey was given to them (2)
• I’d like to know more about the sample they used – probably had a ton of nondrinkers
in it. (2)
• I have never seen anyone who goes out, like to clubs or like to a pre-game or
something, have only two drinks. Yeah, that’s pretty, pretty surprising to me. I would
say that that is maybe the fact that there are people that don’t drink in this study. (2)
• I am actually trying to think back to when I took this and I don’t remember there
being a time when I drank 5 drinks in a 2 hour period. I did pre-game a football game
so that might have been why I did 1 instead of 0. Yeah, because I typically don’t have
more than 5 in a 2 hour period because that gets a little dicey for me so I wouldn’t do
that, but yeah, I probably just didn’t feel fully comfortable putting 0 because I do
drink around that number and like a little longer of a time period and yeah. (2)
• People always lie on these surveys. (3)
• data makes sense depending on the sample that was surveyed because if there are
people that don’t drink, 3 drinks could be a lot, but for people who do drink I feel like
3 drinks is wayyyyy to low of an estimation (3)

3. Follow/Neutral
Statements within this category include simply restating or summarizing what they are
viewing. Tone and content reflect a degree of indifference to or disengagement from
information being presented. Scores reflect the extent to which participants are simply
following along with the presentation.
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 156

Examples
• In the last slide I was almost double the average student, but here I’m close to the same
• Uh huh
• I perceive that people drink more, and I definitely drink more than the average
• yeah
• As you can see I think I definitely drink more than the actual student both in my
perception and you know in actuality.
• My um, the number of drinks per occasion is higher than the actual, um not really
surprised on that, and my perceived is really close to the actual. I thought I had pretty
good ground in that. Um I feel like the actual there’s a lot of people all over the place,
but obviously they’re right around there.
• So yeah I mean I felt like that was pretty spot on and know that my drinking is higher
• my number of drinks per week is pretty high, um my perceived is right in the middle
between 0 and (inaudible). I feel like this is cause you normally drink on Friday
Saturday, have around the average of 2 and a half drinks (inaudible) so then they would
add up to 5. Mine’s 9 because I was had it at 4 and a half in two days so.
• Um, my maximum number of drinks is um pretty high relative to my perceived. Because
um, there was like one day where it was just a long day, it was pretty spread out, um kind
of like morning to middle of the day, all the way to the night. And so that’s why it’s
pretty high, but my perceived is at 4 just because I feel like that occasion doesn’t happen
often for other people, so it’s just kind of trying to generalize, what other people would
be at.
• My estimate of what other people drink is lower than mine. Um I thought this was
because when I drink it’s not often, but when I do it’s uh greater volumes, while I thought
the average of everyone else would be kind of spread out in um lower amounts
• Um my perceived is 4 and actual is 4.6. Um I think that’s pretty close. Um my view, I
feel like is pretty accurate. But my actual I feel, or my own, the actual data really doesn’t
make me feel any different about it. I know it’s just kind of like a one off occasion. That
normally doesn’t happen. So yeah, I don’t know. The actual is probably about what I
would normally be.

4. Believability of Data
This coding category captures statements referring to the degree to which a participant
believes the data that are presented. Statements may refer to the source of the data, the
individual’s reports of own drinking or estimates of others’ drinking, or the actual
reported group norm. This code is not to be scored unless there is explicit reference to the
data in some fashion.

Examples
• The researchers used a survey of a lot of USC students, so I guess the data are
accurate (1)
• 6.5 drinks per week – well if you go out once a week I guess that makes sense (1)
• I mean obviously it was a little lower than I predicted, but I am not so surprised by
the number. (1)
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 157

• Um so yeah that’s um actual average of about 5 seems about right, um happy about it,
nice to see, yeah (2)
• I would say that my estimate is pretty accurate. Because binge drinking is a little
more intense than just, you know, drinking a glass of wine in the week, so I would
say this is about accurate. (2)
• Okay, so this one seems about right from when I am going out. (2)
• The sample of 800 students campus wide obviously includes a wide eclectic group of
people (2)
• I do believe that I binge drink on more occasions (2)
• I think as a maximum as drinks go, I would say that this is a pretty accurate depiction
of what I think (2)
• this is kind of what I was predicting that most people, especially females, aren’t binge
drinking, again, so if you were averaging all the people who do drink and who don’t
drink it will kind of balance out and yeah. (2/3)
• Yeah, so, makes sense. (2/3)
• During the weekend people go out and drink, usually more than I drink. (starts
explaining math). Which makes sense to me. For an average this makes sense for
weekend occasions. (3)
• But yeah, that makes total sense. That is typically around what I would do so yeah (3)
• Okay, yeah, this makes sense. (3)
• So yeah, that makes a lot of sense. (3)
• Yeah I believe that (3)

5. Reflective Analysis & Beliefs Modification
This coding category refers to the extent to which participants acknowledge that a shift in
perspective might be needed or attempt to actually modify existing beliefs in light of the
information being presented. Statements include a thoughtful analysis of the data in which
previously held beliefs/opinions/perspectives are reconsidered. Explanations may involve
recalling anecdotal examples that justify and are consistent with what the participant is
viewing or identifying specific reasons why original beliefs may be incorrect. Scored
statements should reflect a depth of processing that goes beyond simply summarizing what
they are viewing (see, follow/neutral) or reflect a degree of self-exploration in which
participants view their own behavior/perceptions or others’ behavior in a new light.

Examples
• I don’t think binge drinking is that frequent. I think 1 is a good estimate. Maybe a
little less than 1. When I think of binge drinking it means people having a bad night.
(explains math). While I think a lot of people don’t know how to drink I think most
students know their limits. (1)
• This (discrepancy/misperception) could be because I don’t drink as much as I think
other people are drinking (1)
• Maybe I perceived that the average male student is your average frat star, when that
only really represents a small group, so I doubt the actual results will be as high as my
perception, but I will be surprised if the actual is less than mine, it’ll definitely make
me think (1)
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 158

• I think that most people at USC probably don’t drink that much so yeah, I probably
should have focused a little harder while taking this, but yeah, I would say that it is
probably closer to like 4 drinks per week or maybe 3. And yeah, I just put that I
perceived that everyone else would be around my number truly because I didn’t want
to assume somebody would be a lot worse than I was. Or yeah, so I don’t know. It’s a
little high. (1)
• I guess going out twice a week is more than the average usc student (1)
• that (perception) comes from the people I hang out with and I see at parties (2)
• Thinking back now, I think maybe 10-12 would be a good perceived estimate, but it
would probably be too high of an estimate, based on previous data. (2)
• The people I am surrounded with are above average and are bigger drinkers than
normal so that is what is making me perceive higher numbers than in general. (2)
• Like what I said before, I figured now looking back on this, that my estimates were a
little high because like I said, some people don’t drink at all and the people who do
drink, like 2 seems like a good number especially if you are just having casual drinks
when you are going to like a dinner or when you are just like home with friends. (2)
• Like what I was saying before I do think that my estimates were a little high and that
4 drinks sounds about right. I am surrounded by a lot of people who drink fairly
heavily so that might have been why I was estimating this high number because I am
friends with a lot of theatre people and a lot of times theatre people can drink a little
more than others because they have less actual homework because their schoolwork
revolves mostly around being present and in the moment. (2)
• I perceived it would be 1 because other students, because of like pre-games, we were
having a lot of football games right now, but yeah, I was assuming the actual is like .4
or something because I didn’t know that you could do decimals—I probably would
have done the decimal in there, but yeah, I am expecting the actual to be a little lower
because of people that don’t drink. (2)
• Maybe it’s (discrepancy/misperception) is because of the people im around, which
makes me think it happens much more than it actually does (2/3)
• Maybe im comparing myself to what im seeing in my direct surroundings, rather than
what I see in general (2/3)
• I mean I usually go out drinking about two times a week. My estimate of 8 was also
that the typical USC student also drinks about two times a week, meaning that you
know each time average 4 to 5 shots, um I thought it was pretty reasonable, but I may
not have been taking into account students who you know really don’t go out drinking
all that often, yeah because I do not have much contact with students that don’t go
out, or you know students you know period. (3)
• So I have personally felt that there was a perception amongst people who go out
drinking consistently that you know the more shots you’re able to take in one
occasion the better, maybe like 10. I’ve heard somebody say like oh yeah you can
only take 10 shots in a night that’s weak, but um you know I frankly think it’s you
know even though the perceptions there, I frankly think it’s kind of stupid. Um so
yeah that’s um actual average of about 5 seems about right, um happy about it, nice to
see, yeah. (3)
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 159

• I would say that yeah, other people who are going to the football games probably took
part in a little bit of binge drinking and other people probably didn’t at all so the
average makes sense. (3 for reflective analysis and 2 for believability)
• Um I feel like after looking at this my perceived is pretty high. My own was a 2
simply because there was just two times on the weekends where it was kind of like a
pregame situation, um but yeah. I feel like I misjudged the perceived on this one,
even without even seeing the actual I can already kind of feel like the actual is going
to be a lot lower than 2. (3)
• My perceived is uh, almost double the actual. I feel like I really missed on that one. I
noticed it right when it popped up between mine and the perceived that the actual was
gonna be a lot lower. Definitely makes me feel a little bit different about binge
drinking and how often um it actually happens. There are certain cultures where it
does happen a lot, and other cultures where it doesn’t, and I feel like the culture that
I’m in it happens a lot, so that’s why I, mine is pretty high. (3)

6. Surprise
a. The surprise (6a) coding category captures primarily positively-valanced
emotional expressions of surprise, shock, amazement, or astonishment in reaction
to the presented data. It will typically involve an emotional/affective reaction to a
difference between one’s perceived norm and the actual group norm.

Examples Surprise 6a
• Yeah I mean again, I am not, I mean obviously it was a little lower than I
predicted, but I am not so surprised by the number. (0)
• again, this doesn’t surprise me very much. (0)
• Um yeah again it was a little high, but I am not so surprised by the number. (0)
• Um yeah again, I am not surprised at all this was kind of exactly, sorry, I am
really not that surprised (0)
• So yeah, this is kind of what I thought it would be so I am not too surprised. (0)
• I am not so surprised with the actual percentage (0)
• But yeah, I am not surprised by this result. (0)
• Once again I am not surprised by the result bc my perceived is pretty close to the
actual. (0)
• I’m a little bit shocked (1)
• I guess im a little bit surprised. I didn’t think the actual would be this low. I
thought I would be a lot lower than the average male student. (1)
• So this one is actually surprising, a little bit. (1)
• Um yeah, though I am a little surprised, I might have expected somewhat lower
numbers for the actual considering many people might not drink at all. (1)
• Seeing that I’m double the actual drinking of another male student is actually
surprising (2)
• Damn, wasn’t expecting that. (2)
• Okay so this one is also kind of high. I don’t know what I was thinking when I
was taking this test, but yeah. (2)
ATSS AND PERSONALIZED NORMATIVE FEEDBACK 160

• This isn’t actually too surprising but what I am perceiving is actually way higher
than the actual and it looks like I am thinking that female usc students are heavier
drinkers than they are in general, but yeah, this is very eye opening. (2/3)
• Wow! (3)
• this is actually very surprising bc I thought I would drink less than the average
and this is actually the opposite of what I was believing and it turns out others
actually drink less than me per occasion and that’s really shocking (3)
• oh God. Shit. Oh my God. (3)

b. The surprise (6b) coding category captures primarily negatively-valanced
emotional expressions of surprise, shock, amazement, or astonishment in reaction
to the presented data. Coded (negative) reactions of surprise may often
accompany skepticism of the data (e.g., “Wow, I definitely I don’t believe those
numbers.”). In this case, dual scores for both surprise (6b) and skepticism are
warranted.

Abstract (if available)

Abstract Objective: Problematic alcohol use among college students is a national health concern, with students receiving sanctions for violating campus alcohol policy (adjudicated students) identified as a particularly high-risk group. Personalized normative feedback (PNF) is a brief intervention approach designed to correct normative misperceptions of peer drinking behavior which in turn has been shown to reduce alcohol use. Employing a randomized longitudinal intervention design, the present study sought to reduce individual drinking and alcohol-related consequences among adjudicated undergraduates by reducing misperceived peer drinking norms through the provision of a PNF intervention. The research explored explanatory mechanisms of change through an adapted application of the Articulated Thoughts in Simulated Situation (ATSS) cognitive think-aloud paradigm. Seven coding categories emerged and indirect effects of these codes on intervention efficacy were considered. Questions about which subgroups of individuals benefit most (or least) from the PNF intervention were investigated through analyses of moderation. The chosen moderators are important on a theoretical level (i.e., group identity, controlled vs. autonomous personality orientations) as well as pragmatically to adjudicated students (i.e., pre-intervention defensiveness). Method: A sample of 70 (51% female) undergraduate students were randomly assigned to one of three conditions: a PNF-ATSS condition, a PNF-Only condition (without ATSS), and an active Control+ATSS condition which received psychoeducation about alcohol use. Participants completed baseline and one-month post-intervention questionnaires. Intervention content was delivered on a lab computer, with audio/visual synced components, and with think-aloud segments in the two conditions that entailed ATSS. Think-aloud data were recorded, transcribed, and content-analyzed. The seven cognitive-affective coding categories were as follows: Sustain talk, skepticism, follow/neutral, believability, reflective analysis, positive surprise and negative surprise. Results: The General Linear Model was used to answer the specific aims of the study. Students in both the PNF and PNF-ATSS conditions reported significant reductions in their misperceived peer drinking norms and alcohol-related consequences at the 30-day follow-up, relative to students in the control condition, who were not found to have significantly reduced their normative misperceptions or alcohol consequences. These main effects were present whether the PNF conditions were evaluated independently or collectively against the control. Moreover, there were no observed differences between the two PNF conditions in magnitude or direction of the effects. With respect to self-reported alcohol use, participants in the PNF-ATSS condition drank significantly fewer drinks per week at follow-up than participants in the PNF-Only condition, but not less than participants in the control condition. No differences were found between the control and PNF-Only conditions. There were significant indirect effects from the intervention to drinking and perceived norms outcomes via the follow/neutral code. Being in the PNF-ATSS condition was associated with lower levels of neutrality regarding the intervention content, and lower levels of neutrality, in turn led to lower drinking and lower perceived norms at follow-up. A significant indirect effect also emerged from the intervention to perceived norms via believability. Being in the PNF-ATSS condition was associated with lower levels of believability regarding the intervention content. However, the greater the level of believability regarding the content, the lower perceived norms were at follow-up. Lastly, autonomous orientation was found to moderate intervention efficacy such that the intervention effect on perceived weekly drinking norms was stronger for PNF-Only participants compared to Control, as level of autonomous orientation decreased. Conclusion: In sum, the research provides important theoretical and practical contributions to the social norms and alcohol intervention literature. Skepticism, believability, and neutrality remain issues needing to be addressed if PNF is to be strengthened as an intervention component. Understanding the key mechanisms by which PNF interventions work, why, and for whom, is imperative for the evolution of norms-based intervention strategies and theory development.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Breaking it down to make it stronger: examining the role of source credibility and reference group specificity in the influence of personalized normative feedback on perceived alcohol use norms a...

PDF

Assessing therapeutic change mechanisms in motivational interviewing using the articulated thoughts in simulated situations paradigm

PDF

Measuring narcissistic cognitions in response to relationship threat using articulated thoughts in simulated situations

PDF

Reacting to the negative consequences of your decisions: an ATSS comparison of cognitive and affective reactions in older versus younger adults

Asset Metadata

Creator Hummer, Justin Francis (author)

Core Title Personalized normative feedback applied to undergraduates with problem drinking: a comparison with psychoeducation and an examination of cognitive-affective change mechanisms via the articulated ...

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Psychology

Publication Date 08/02/2019

Defense Date 06/17/2019

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

Tag adjudicated students,alcohol use,articulated thoughts in simulated situations,ATSS,College students,intervention,mechanisms of change,OAI-PMH Harvest,personalized normative feedback,PNF,problematic drinking

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Davison, Gerald C. (committee chair), Leventhal, Adam (committee member), Saxbe, Darby (committee member), Walsh, David (committee member)

Creator Email hummerjf@gmail.com,jhummer@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-204006

Unique identifier UC11663143

Identifier etd-HummerJust-7727.pdf (filename),usctheses-c89-204006 (legacy record id)

Legacy Identifier etd-HummerJust-7727.pdf

Dmrecord 204006

Document Type Dissertation

Format application/pdf (imt)

Rights Hummer, Justin Francis

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