Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Motivational interviewing with adolescent substance users: a closer look
(USC Thesis Other)
Motivational interviewing with adolescent substance users: a closer look
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
MOTIVATIONAL INTERVIEWING WITH ADOLESCENT SUBSTANCE
USERS: A CLOSER LOOK
BY
Elizabeth M. Barnett
___________________________________________________________
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
(PREVENTIVE MEDICINE)
May 2013
Copyright 2013 Elizabeth M. Barnett
i
DEDICATION
To my dad, Frank Barnett (1938 to 2007)
who always said,
“If you want respect, get a Ph.D.”
ii
ACKNOWLEDGEMENTS
I would like to thank my committee, Drs. Steve Sussman, Donna Spruijt-Metz,
Luanne Rohrbach, Ping Sun, Theresa Moyers, and Stan Huey, Jr. for making me a better
researcher and better writer each in their own way. To the team at USC, thank you for taking
a chance on me and tirelessly guiding me through. I would also like to thank Dr. Huey for
introducing me to Caitlin Smith, a kindred MI spirit, who I hope to conduct research with
throughout my researching days. Also special thanks to Dr. Moyers, agility trainer,
grandmother and MI master for taking on the extra work of mentoring me from afar.
I would also like to thank my professional network. First, the many MINT members
who chimed in on research related questions, especially, Mary Beth Abella, Melinda Hohman,
Denise Ernst and others who inspired me via the list serve. Second, thanks to Dale Parent,
Brad Bogue, Bill Woodward, and Brian Matteson for the unique stepping stone you provided
for my career, my specialty and my life. I appreciate you all more than you know.
In addition to the professors at USC, I want to thank my friends in the program, Drs.
Karla Wagner and Sue Schembre for being available to help me with analyses, tables and
various other aspects of the “art” of statistics. To Dr. Nadra Lisha and Doctor-in-training
Gillian O’Reilly for being in the boat with me and helping me keep my head up.
Finally, I must thank my wonderful, supportive husband, Kevin, for making it possible
for me to pursue this degree; also thanks to my in-laws, my mother, and my sister for helping
with childcare; and of course to Naomi, my 3-year old daughter, for warm send-offs and
welcome-homes. Special thanks to my mother-in-law, Susan, for proof-reading my work.
Finally, thanks to the local coffee shops that never charged me for office space. I couldn’t
have done this without any of you!
iii
TABLE OF CONTENTS
Dedication .......................................................................................................................................i
Acknowledgements........................................................................................................................ii
List of Tables ................................................................................................................................iv
List of Figures ...............................................................................................................................v
Abbreviations and Glossary ........................................................................................................vi
Abstract .......................................................................................................................................viii
Chapter 1: Introduction
Specific Aims & Hypotheses ....................................................................................................1
Background & Significance .......................................................................................................4
Randomized Controlled Trial Overall Methods ......................................................................16
Motivational Interviewing Booster
Booster Design and Implementation ............................................................................18
Training and Supervision of Interventionists ...............................................................21
Data Collection and Management .................................................................................22
Training and Supervision of Coders .............................................................................24
Coding of Recordings.....................................................................................................25
Chapter 2: Motivational Interviewing and Decision Making
Introduction...............................................................................................................................31
Methods ....................................................................................................................................33
Results ......................................................................................................................................36
Discussion.................................................................................................................................36
Chapter 3: Motivational Interviewing and Reflective Listening: A Closer Look
Introduction...............................................................................................................................42
Methods ....................................................................................................................................46
Results ......................................................................................................................................51
Discussion.................................................................................................................................51
Chapter 4: From Counselor Skill to Decreased Marijuana Use: Does Change Talk
Matter?
Introduction...............................................................................................................................57
Methods ....................................................................................................................................62
Results ......................................................................................................................................68
Discussion.................................................................................................................................69
Chapter 5: Summary and Conclusions ....................................................................................79
References ..................................................................................................................................86
iv
LIST OF TABLES
Table 1-1 MI Booster condition sample descriptive statistics .................................................28
Table 1-2 Target behaviors in MI contacts .............................................................................29
Table 1-3 Interventionist proficiency ratings ..........................................................................30
Table 2-1 Comparison of current sample to those lost to follow-up or incomplete data......39
Table 2-2 Bivariate correlations.................................................................................................40
Table 2-3 Results of logistic regression predicting drug use at 1-year follow-up..................41
Table 3-1 Counselor to client transitions ..................................................................................56
Table 4-1 Summary of studies including mediation analyses .................................................75
Table 4-2 Measurement details for all variables included in models ....................................76
Table 4-3 Univariate and bivariate statistics for all variables included in models ...............77
Table 4-4 Mediation results for MI indicators predicting reductions in marijuana use ......78
v
LIST OF FIGURES
Figure 1-1. Theoretical process model of Motivational Interviewing ....................................11
Figure 4-1 Proposed mediation model illustrating hypothesized causal mechanisms ..........73
Figure 4-2 Consort diagram. ......................................................................................................74
vi
ABBREVIATIONS AND GLOSSARY
FN: Follow/Neutral: Client language unrelated to change.
CACTI: CASAA Application for Coding Treatment Interactions
CASAA: Center on Alcoholism, Substance Abuse, and Addictions, University of New Mexico
CR: Complex Reflection: Counselor statement that adds direction, meaning or feeling
CT: Change Talk: Client language expressing reason, ability, or commitment to change.
CCT: Counter Change Talk: Client language expressing reason, ability, or commitment to
status quo.
MI: Motivational Interviewing
MICO: Motivational Interviewing Consistent: Behaviors such as affirmation, emphasizing
personal choice and control, asking permission before giving advice
MIIN: Motivational Interviewing Inconsistent: Behaviors such as confronting, warning and
giving advice
MINT: Motivational Interviewing Network of Trainers
MISC: Motivational Interviewing Skill Code
MI-SCOPE: Motivational Interviewing Sequential Coding of Process Exchanges
MITI: Motivational Interviewing Treatment Integrity
PCT: Percent Change Talk
PRCT: Percent Reflection of Change Talk
PRCCT: Percent Reflection of Counter Change Talk
RCT: Reflection of Change Talk
RCCT: Reflection of Counter Change Talk
RNEG: Reflection Negative: Reflection of client language away from change.
vii
RPOS: Reflection Positive: Reflection of client language toward change.
SR: Simple Reflection
TND: Project Towards No Drug Abuse
viii
ABSTRACT
This dissertation adds to the existing literature about how Motivational Interviewing
(MI) works and who responds to best to Motivational Interviewing interventions. It also
describes the development of an innovative MI booster intervention for a classroom-based
substance use prevention program that was able to accommodate participants regardless of
level of substance use or their substance of choice. This dissertation also describes the
supervision and training of interventionists and coders.
The most recent trial of Project Towards No Drug (TND) Abuse, a classroom-based
substance use prevention program, was designed to determine whether the addition of an MI
booster would enhance program effects for students attending continuation high schools in
Southern California. In Study 1 (Chapter 2), data from two conditions (TND only and
TND+MI) were used to investigate whether decision making attributes moderated the effects
of program condition on substance use outcomes. Results from this analysis found no effect
of decision making on program efficacy, leading to the conclusion that decision making does
not warrant further investigation as a moderator of MI efficacy.
Investigations regarding how MI works were conducted using data derived from
sequentially coding recorded treatment interactions from the MI condition. All sessions
pertaining to substance use were coded using a MI specific coding manual and MI specific
automated technology. Study 2 (Chapter 3) describes an investigation into the relationship
between a specific MI counselor behavior (reflective listening) and client language about
change (change talk). Counselor reflections were coded for complexity (simple or complex)
and valence (positive/toward or negative/away from change). Client language was coded as
toward change (change talk), away from change (counter change talk) and neutral about
ix
change (FN). A comparison of effect sizes shows that the valence of a reflection is more
strongly associated with subsequent change talk than is complexity. Study 3 (Chapter 4) used
summary scores derived from the sequentially coded data to characterize counselor’s fidelity
to MI. We investigated whether the percentage of client change talk mediated the relationship
between six unique counselor skills and marijuana use outcomes. Study findings showed that
the relationship between percentage open questions, percentage complex reflections,
counselor empathy and counselor MI spirit and outcomes were mediated by change talk.
Percentage reflection of change talk demonstrated a main effect on outcomes. Findings from
Paper 2 and Paper 3 support an existing call for a new measurement instrument that focuses
on how counselors respond to change talk. Such an instrument will be invaluable in the
development of MI skills and enhance our ability to improve program efficacy.
1
CHAPTER 1: INTRODUCTION
Specific Aims
A growing body of research demonstrates the efficacy of Motivational Interviewing
(MI) interventions with adolescent substance users (Jensen et al., 2011). Findings in studies of
adolescent substance users are consistent with those of young adult and adult users –
consistent but small effect sizes when compared against no treatment controls (Hettema,
Steele, & Miller, 2005; Jensen et al., 2011). Among adult users, some information about the
mechanisms by which MI works exists (Apodaca & Longabaugh, 2009). Current evidence
shows associations between counselor skills, client language, and substance use outcomes
(Moyers, Martin, Houck, Christopher, & Tonigan, 2009). However, to date, only one study
investigates these relationships in an adolescent sample (Baer, Beadnell, Garrett, Hartzler,
Wells, et al., 2008). Furthermore, little is known about client characteristics that moderate the
efficacy of MI. Although some evidence for such moderation exists within the adult and
young adult literature (Carey, Henson, Carey, & Maisto, 2007; Karno & Longabaugh, 2004),
findings do not appear to be theoretically driven and do not include adolescents. This
dissertation, a secondary data analysis of the most recent randomized controlled trial of
Project Towards No Drug Abuse (TND), a 12-session school-based curriculum targeting
youth at risk for drug abuse (Sussman, Sun, Rohrbach, & Spruijt-Metz, 2012), fills these
research gaps.
Study 1 explores the moderating role of individual level decision-making
characteristics on the relationship between treatment condition (TND+MI and TND only) and
substance use outcomes. Strengths or weaknesses in various aspects of decision-making may
interact with treatment condition, such that certain students may benefit more from the one-
2
on-one support provided in the MI condition. This study aims to characterize youth for whom
the addition of an MI booster enhances outcomes.
Study 2 uses sequentially coded client-counselor communication data to examine the
temporal relationship between different types of reflections and client change language.
Reflections have long been considered a critical skill in the practice of MI and represent a
major challenge for some practitioners to learn (Miller & Rollnick, 2002). Throughout the MI
literature, there is an implicit assumption that complex reflections are more "effective'" than
simple reflections, however this assumption has not been tested in relations to eliciting change
talk. Results from these analyses are examined for a) probabilistic differences between
associations of complex and simple reflections and “change talk,” i.e., client language toward
change, or “counter change talk,” i.e., client language away from change, and b) probabilistic
differences between associations positively and negative valenced reflection and change and
counter change talk.
Finally, Study 3 uses results from Study 2 to inform the development of a model to
test the mediating relationship of client language between counselor skill and substance use
outcomes. Counselor skill is measured using six unique measures of MI fidelity, including
percentage of reflections of change talk, the measure of valence reflections taken from Study
2. Consistent with the theoretical underpinnings of MI and empirical findings, we expect to
find that counselor behaviors consistent with MI will be associated with client language in the
direction of change, which will result in decreased substance use. This study will be the first
to examine the full mediational model among an adolescent substance-using population
In summary, the specific aims of this dissertation are:
3
Study 1
Aim 1-1: Explore the moderating influence of decision-making characteristics on the
association between treatment condition and substance use behavior.
Study 2
Aim 2-1: Explore the conditional relationships between simple and complex reflections and
client change language.
Aim 2-2: Explore the impact of positively and negatively valenced reflections on change talk
and counter change talk.
Aim 2-3: Compare effect sizes of associations between complexity and valence and client
language.
Study 3
Aim 3-1: Explore the mediating effect of percent of client change language on the relationship
between individual counselor skills and marijuana use.
4
Background & Significance
Public Health Significance of Adolescent Substance Use. Substance use continues
to be a major concern for the health and well-being of adolescents in the United States.
According to the 2012 Monitoring the Future Report (MTF), the trend of increasing marijuana
since 2007 has stopped among 8
th
, 10
th
and 12
th
graders. However, the perceived risk of
smoking marijuana continues to sharply decline, suggesting that this year’s data may not be
indicative of progress and that increases will resume in years to come. Daily marijuana use
remained flat with 1 in 15 seniors or 6.5% reporting daily use, still at its highest level since
1981. Other illicit drug use, such as inhalants, pain killers, and ecstasy, remains fairly stable
over the past three years with two new drugs coming on the scene “bath salts” and “synthetic
marijuana”. As for alcohol, after leveling off after years of decline to 40% of 12
th
graders
reporting drinking in the past month, alcohol use showed a 2% increase among 12
th
graders.
Binge drinking, defined as consuming five or more drinks in a row at least once in the last two
weeks, followed the same leveling-off pattern, settling at 5%,16%, and 24% overall for 8
th
,
10
th
, and 12
th
graders. Finally, cigarette smoking in the past month showed a statistically
significant drop from 11.7% in 2011 to 10.6% over all grades, with 10% of seniors reported
being current daily smokers. In sum, the rise in marijuana use has temporarily come to a stop,
while the gains made in reducing alcohol use have suffered a setback, and reductions in
tobacco use continue (Johnston, Bachman, & Schulenberg, 2012).
Alcohol use and abuse continues to present the greatest immediate threat to adolescent
health. Adolescents have different patterns of alcohol use than do adults. They have higher
levels of binge drinking, are less likely to meet dependency criteria, and are more likely to
meet abuse criteria (Erickson, Gerstle, & Feldstein, 2005). When it comes to drinking and
5
driving, the Center for Disease Control and Prevention’s (CDC) Youth Risk Behavior Survey
System (YRBSS) reports approximately 10% of adolescents operated a vehicle after drinking,
28% reported riding in a car with an impaired friend in the prior 30 days, and that motor
vehicle crashes represent the leading cause of death for this age group (Prevention, 2010).
Furthermore, the National Highway and Traffic Safety Administration (NHTSA) reports
drivers aged 15-20 made up 19% of all fatalities with drivers with Blood Alcohol
Concentration > 0.08 and 350,000 teens were treated in emergency rooms for motor vehicle
crashes. Meanwhile, a survey of emergency room visits in 13 metropolitan areas by the
Substance Abuse Mental Health Services Administration’s (SAMHSA) Drug Abuse Warning
Network (DAWN) reports approximately 900,000 substance-use-related emergency room
visits among those under age 21. These trends suggest that in order to make additional
progress in reducing the incidence of drug use, innovative and targeted efforts are needed.
Effective Strategies for Substance Use Prevention and Intervention. Substance use
prevention and intervention programs range from multi-session educational programs
delivered in schools or community-based organizations to individual, family, or group
counseling interventions to mass media and public service campaigns such as “Just Say No”
or “This is Your Brain on Drugs,” and brief opportunistic interventions of one or two
sessions. Interventions using the theory, skills, and spirit of Motivational Interviewing (MI)
make up one large category of brief intervention. A classic review of the more broadly
defined brief intervention literature by Bien, Miller, and Tonigan, (1993) identifies the
common features of brief intervention as 1) containing objective feedback, often of
assessment results, 2) promoting autonomy, choice and personal responsibility, 3) providing
6
advice, 4) providing a menu of strategy options to pursue, 5) providing empathy, and 6)
supporting client self-efficacy or ability to change. This review clearly shows that brief
intervention was as effective as more lengthy and costly interventions, and highlights the role
of brief intervention as paving the way for treatment enrollment and engagement. The authors
recommend that treatment programs avoid using waitlists, and rather provide single-session
brief intervention immediately for people seeking services, as research shows that one session
can be adequate to initiate a self-directed change process. They also confirm commonalities
between brief intervention and the principles of MI, with the primary difference being the use
of direct advice, which is not a standard feature of MI interventions.
What is Motivational Interviewing? MI is defined as a directive, non-judgmental
counseling style to explore and resolve ambivalence about behavior change (Miller &
Rollnick, 2002). MI originated as a backlash to the confrontational methods that were
pervasive throughout addiction treatments in the United States in the 1970s and 1980s (Miller
& White, 2007). Miller and Rose (2009) define the relational aspect of MI as expressing
empathy and demonstrating the “spirit of MI,” which they describe as “collaborative rather
than authoritarian, evoking the client’s own motivation rather than trying to install it, and
honoring client autonomy.” They further call MI a “way of being with clients” and a
necessary but not sufficient criteria for determining whether an intervention is truly MI. In
other words, an intervention cannot be considered MI if it does not adhere to the relational
aspect. The technical aspect of MI is the proficient use of MI techniques to increase change
talk (CT) and decrease counter-change talk (CCT). CT is defined as statements about the
desire, ability, reason, need and commitment to change the identified target behavior, while
7
CCT expresses the desire, ability, reason and need to retain the status quo. Reflective listening
and open questions are the primary technical skills used in MI. Reflective listening
demonstrates that the counselor has heard and understands the meaning of what has been said.
It can be simple, mere repeating, or complex, in which the counselor adds hypotheses about
client meaning. The addition of meaning and direction into reflections functions to steer a
conversation, in a similar but more skillful way than asking questions. Open questions allow
clients the freedom to explore a topic, while closed questions limit responses and may
reinforce the power differential that exists between counselor and client. Other techniques
used in MI include emphasizing the client’s personal choice and control, and asking
permission before sharing concerns, giving advice or providing information.
MI interventions assume that ambivalence about behavior change is normal and insist
that it not be met with judgment or confrontation. They differ from other interventions
because they do not assume a deficit model of functioning (Ingram & Snyder, 2006). Family-
based programs target family functioning deficits, while skill-building classes target
individual functioning deficits. MI approaches intervention from the premise that people have
the skills, insight and desire necessary to make healthier choices, and that these thoughts and
choices will manifest given an environment where clients do not face judgment and are
allowed a sense of respect and autonomy to make choices for their lives (Miller & Moyers,
2006). Further, MI assumes that these decisions are made based on thoughtful consideration,
not information exchange or the counselor’s psychological insights into client behavior
(Resnicow, Davis, & Rollnick, 2006). Resistance is considered to be a result of the
interaction between the counselor’s behavior and the client’s ambivalence about change and
need for autonomy. Rather than trying to convince someone to change, a counselor’s task is
8
to create a space where the client can consider the pros and cons of changing, and consider
how their current behavior fits with their own long-term goals and personal values (Resnicow
et al., 2006). Opponents and detractors of MI contend that MI is not a panacea. They argue
some people do better with direct advice and MI is just “good counseling.” Both of these
arguments have some support in the literature. Beutler, Mohr, Grawe, & Engle, (1991) found
that people low in psychological reactance perform better with more directive approaches.
Further, research suggests that interpersonal qualities of empathy, warmth, and genuineness
explain more of the treatment effects than do the technical skills that counselors use. (Miller,
Benefield, Tonigan 1993; Moyers & Miller, 2012)
Measuring Fidelity to Motivational Interviewing. In order to investigate the
assumptions made in MI much effort has been put into measuring it. Measuring fidelity to MI
interventions began with Project MATCH, the United States’ largest test of matching
alcohol interventions to client characteristics. In that study, Motivational Enhancement
Therapy was compared to a cognitive behavioral intervention, and a twelve step
facilitation. The MATCH Tape Rating Scale (MTRS) was developed in order to
discriminate between the three conditions and to ensure that counselor interpersonal
skill and therapeutic alliance did not significantly differ between conditions. Findings
showed that interventions were discernible and infrequently used techniques associated with
the other treatments (Carroll et al., 1998). The first study to go beyond interpersonal skill was
seen in Miller, Benefield, Tonigan (1993). They used a modified version of a resistance
coding system (Chamberlain, Patterson, Reid, Kavanagh, & Forgatch, 1984; Kavanagh,
Gabrielson, & Chamberlain, 1982) to perform behavioral coding of MI practice. Behavioral
9
coding refers to the classification of counselor statements into category such as confront,
query, listen, teach, or direct. Early experiences using existing instruments led to the 2000
public release of the Motivational Interviewing Skill Code (MISC 1.0; Miller, Moyers, Ernst,
& Amrhein, 2003). The MISC 1.0, developed by the University of New Mexico’s Center on
Alcoholism, Substance Abuse, and Addiction (CASAA), is an exclusive and exhaustive
coding scheme that 1) rates counselor global interpersonal skill on Likert scale measures of
constructs such as acceptance, empathy, collaboration, 2) counts instances of the use of
certain counselor behavioral skills such as open and closed questions, and simple and
complex reflection, and 3) counts instances of client language that indicate preparation and
commitment to change. Later Moyers, Martin, Manuel, Hendrickson, & Miller, (2005)
developed the Motivational Interviewing Treatment Integrity (MITI) through a factor analysis
of MISC data, as a simplified tool to measure counselor adherence to MI. The MITI was
intended to be used to provide coaching and supervision to practitioners, while the MISC is
the preferred instrument for investigating relationships between counselor skill and client
language/behavior. The next step in MI coding was to capture the sequential relationships
between counselor and client language. In 2005, Martin, Moyers, Houck, Christopher, &
Miller released the Sequential Coding of Process Exchanges (MI-SCOPE). Finally, in 2010,
an automated software program called the CASAA Application for Coding Treatment
Interactions (CACTI) was developed to more easily operationalize the collection and analysis
of sequential MISC data (Glynn, Hallgren, Houck, & Moyers, 2012). Both the MISC and the
MITI have been revised from their original versions; current and previous versions of the
coding manuals can be obtained via the www.casaa.unm.edu website.
10
Revisions to the MISC 1.0 were based on the work of linguist Paul Amrhein.
Amrhein, Miller, Yahne, Palmer, & Fulcher, (2003) proposed a two-factor structure of client
language grounded in linguist John Searle’s Theory of Indirect Speech Acts. Factor one
consists of client “change talk/preparatory language,” defined as an utterance that states
desire, ability, reason and need to change, and factor two consists of client “commitment
language” such as “I will, I must, I am going to.”. This two-factor structure was conceptually
supported by the two phases identified by Miller and Rollnick in their foundational text,
Motivational Interviewing: Preparing People to Change (1991, 2002). In that text, a two-
phase process (Phase I: Building Motivation and Phase II: Consolidating Commitment) is laid
out for using and training MI. Amrhein et al. (2003) further found commitment language to
mediate the relationship between change talk and behavior change, the strength of the
commitment language to be more important than its frequency, and an increasing amount, or
the slope, of change talk throughout an intervention to predict reduction in problem behavior.
However, more recently factor analysis on the data coded in Moyers et al. (2009)
found a more nuanced relationship between client language categories and outcomes than had
previously been proposed by Amrhein (Martin, Christopher, Houck, & Moyers, 2011). Martin
et al.’s (2011) findings suggest that there is a six-factor structure that includes factors for
counter-change talk, steps toward change, preparatory language, commitment language,
ability language, and following/neutral language.
Despite the availability of these coding instruments, our ability to understand the
mechanism underlying MI intervention is dependent upon fidelity measurement, which varies
greatly across MI studies. Fidelity measurement ranges from not being measured or reported
at all (Hodgins, Currie, & el-Guebaly, 2001; McCambridge & Strang, 2004), to receiving a
11
brief mention of the amount of MI training provided to interventionists (Emmons et al., 2001),
to information about the ongoing supervision of interventionists (Monti et al., 1999), to
counselor self-report of adherence to MI principles and intervention protocol (Strang &
McCambridge, 2004), to client report of counselor skill (McNally, Palfai, & Kahler, 2005),
and finally to the use of an objective, observational coding system which typically includes
the use of audio- or video-recorded interventions (Barnett et al., 2012a)(Moyers, Miller et al.
2005; Baer, Beadnell et al. 2008; Gaume, Bertholet et al. 2010; Sussman 2011). Observational
coding schemes range from homegrown systems, to adapted versions of existing scales, to
scales developed specifically to measure characteristics of MI such as the MISC, MITI, and
SCOPE (Hettema et al., 2005).
How Does Motivational Interviewing Work? Mediational studies in the MI
literature investigate either the theoretical process of MI, which posits counselor interpersonal
and behavior skill influences the content of client speech (arm 1) and client speech influences
client outcomes (arm 2), (See Figure 1; Miller & Rose, 2009) or (b) possible mediating
constructs, such as readiness to change, self-efficacy, resistance, engagement, or client
experience of discrepancy (Apodaca & Longabaugh, 2009).
Figure 1-1: Theoretical Process Model of Motivational Interviewing
12
A meta-analysis by Apodaca & Longabaugh (2009) on MI’s mechanisms of change
found 29 MI studies provided data on mediators of change. They found support for a
relationship between client change talk and client experience of discrepancy with better
substance use outcomes, as well as finding therapist use of MI Inconsistent (MIIN) behaviors,
such as directing, confronting, warning, giving advice without permission, to be related to
worse substance use outcomes. They report that as of 2009, there are only two full mediation
MI analyses (Karno & Longabaugh, 2004; McNally et al., 2005) in the MI literature and
neither of them used MI process-coding data. Karno’s studied the role of counselor
directiveness, while McNally tested the mediating role of the client’s experience of
discrepancy on reduced substance use outcomes.
Since Apodaca’s meta-analysis, mediation, as proposed in the theoretical process
model, has been supported by the first full mediational analysis of 118 sequentially coded
Project MATCH sessions (Moyers et al., 2009). This two-armed analysis revealed that
percentage of client change talk explains 30% of the relationship between MI Consistent
(MICO) counselor behaviors and alcohol use outcomes. A one-armed analysis by (Gaume,
Gmel, & Daeppen, 2008) found support for language about client ability to predict drinking
rates 12 months later, while (Baer, Beadnell, Garrett, Hartzler, & Wells, 2008), the only study
to report process data with an adolescent sample, finds that client language about reasons to
change predicts changes in substance-use rates at follow-up. Moyers et al. (2007) found a
single category of combined change talk and commitment language to predict alcohol use
outcomes, while Hodgins, Ching, & McEwen, (2009) found commitment language to predict
decreased gambling and decreased monetary losses.
13
For Whom Does Motivational Intervention Work Best? While understanding
grows about the active ingredients of MI, some have turned their attention to
understanding who responds best to MI interventions. Moderation findings in the MI
literature come from three sources: (1) results of Project MATCH, (2) meta-analyses, and
(3) individual RCTs that compare MI to other treatments. Results from the Project
MATCH matching hypotheses found the MI treatment condition to be better suited for
persons scoring high in anger (Karno & Longabaugh, 2004). Secondary analysis of
Project MATCH data found MI to have stronger effects for those reporting high
psychological reactance (Karno & Longabaugh, 2005) and for self-reported high worriers
(Westra, Arkowitz, & Dozois, 2009). Meta-analyses show persons of minority ethnicity
and persons with low motivation to change responded better to MI (Hettema, Steele, &
Miller, 2005). However, because studies do not consistently measure client-level
psychosocial variables, meta-analyses are unable to compare them. Readiness to change
appears as the only psychosocial variable consistently tested, though methods for
measurement and results are inconsistent, thus it is difficult to draw conclusions about
its importance. Findings from individual studies of college students show that MI
interventions work better for students with low future orientation, high self-regulation skills
(Carey, Henson, Carey, & Maisto, 2007) and “high need for cognition”, a tendency to engage
in and enjoy effortful cognitive processing (Capone & Wood, 2009)
Motivational Interviewing’s Fit with Adolescents. First applied to adult alcohol-
using populations, MI is now used for a wide range of health behavior change across the age
14
span. However, questions about adolescents’ ability to make decisions and envision the
future cast doubt on whether the technique is well-suited to this age group (D’Amico, Miles,
Stern, & Meredith, 2008; Erickson et al., 2005). Currently, MI is presumed appropriate for
adolescent substance users based on empirical evidence of its efficacy, and theoretical support
from Blos’ Theory of Development (Blos, 1966). Blos (1966) theorizes adolescence to be a
time of separation and individuation, when making choices for oneself is paramount.
Adolescence is a period when psychological reactance is at its height (Brehm, 1966; Hong,
Giannakopoulos, Laing, & Williams, 1994), and thus a non-confrontational approach is
imperative to achieve success with this population. More support for using MI's with
adolescents comes from studies showing MI to be more effective with the least motivated to
change, (Brown et al., 2003), as well as able to reach non-treatment-seeking participants
(Kealey et al., 2009; Miller et al., 1989). Since adolescents are often mandated to attend
treatment by school officials, juvenile justice professionals, or their families, they fall into
these categories – not motivated for change and non-treatment seeking. Adolescents’ lack of
motivation is often attributed to their having experienced few negative consequences from
their substance use (Tevyaw & Monti, 2004). Due to their age and relatively short duration of
their substance use, they often do not see themselves as having a problem or may believe that
they can stop using at any time (Kealey et al., 2009). Furthermore, among adolescents, some
substance use is considered normative, making it seem unnecessary and also undesirable to
quit (Erickson et al., 2005).
Five reviews of drug-use-related applications of MI with children and adolescents
currently exist (Barnett, Sussman, Smith, Rohrbach, & Spruijt-Metz, 2012b; Jensen et al.,
2011; Tevyaw & Monti, 2004; Erickson, Gerstly, and Feld, 2005; Grenard, Ames, Pentz, &
15
Sussman, 2006). Barnett et al. (2012b) found 67% of the reviewed studies had significant
improved outcomes. Jensen et al.’s (2011) meta-analysis revealed an overall small effect size
of d = .173, 95% CI ([.094, .252] n=21) (). Grenard et al., (2006) reported that interventions
are typically conducted in one-on-one sessions, and usually provided some type of
personalized feedback about the individual’s substance use. Tevyaw (2004) reported that MI
approaches show decreases in substance-use-related negative consequences, substance use,
and lead to increased treatment engagement. Their review also found stronger results for
heavier substance users and/or those less motivated to change. Furthermore, MI interventions
are adaptable to settings in which brief interventions are necessary (Barnett et al., 2012b) such
as emergency rooms frequented by adolescents for drug- use-related injury, or juvenile
detention settings where substance use may be related to their detention. The evidence
supports that MI interventions work with adolescents, but much less is known about the
characteristics of the individuals who respond best to this approach and the mechanisms
through which it works.
Rationale for the Proposed Study. Existing literature tells us that among adults,
persons reporting high in anger and reactance respond better to MI interventions. Among
college students studies show persons scoring low in future orientation, high in self-regulation
skills, and high in need for cognition appear to respond well to MI interventions. With respect
to mediation, there is evidence that counselor technical and interpersonal skill elicits certain
types of client language, and that increases in client change language produce decreases in
problem behavior. However, it is unclear whether these findings of moderation and mediation
hold among adolescent substance users.
16
In today’s environment of scarce resources, universal prevention efforts are
threatened, and more targeted approaches are needed. Although findings from Project
MATCH questioned the ability to predict or hypothesize about who responds best to
particular interventions, Karno & Longabough (2007) insists that “matching is not dead.”
They investigated the additive effect of mismatching clients with treatment modalities and
determined that being mismatched to treatment leads to increased substance use.
Understanding who responds best to specific intervention types promises to save time and
money and increase the effect sizes of interventions. Understanding which counselor
behaviors are most important for eliciting which client statements will also support
streamlined interventions and training that can increase effectiveness of MI interventions. The
studies in this dissertation provide additional insight into these relationships and increase our
ability to effect change among adolescent substance users. In the Methods section that
follows, I present details about the overarching study and the specifics about the MI booster
design and implementation.
Randomized Controlled Trial
Twenty-four continuation high schools in four counties in Southern California were
invited to participate in a three-arm randomized controlled trial of the Project Towards No
Drug (TND) Abuse program. Continuation schools accommodate students who due to lack of
credits will be unable to graduate from regular high schools. In order to be eligible for
inclusion schools met the following four criteria: 1) at least 5% Caucasian enrollment to be
consistent with previous TND studies, 2) situated within 75 miles of project headquarters (for
the ability to manage implementation of the program, 3) contain only grades 9 to 12, and 4)
17
have at least two classes with at least 60 students. Sixty-one schools met these criteria.
Thirty-seven declined to participate, were unreachable, or were on a wait list and were not
contacted after a sufficient number of schools were reached. Participating schools were
randomly assigned to the assessment-only control, the classroom program only (TND), or the
classroom program plus MI condition (TND+MI). At least two classrooms from each school,
with a minimum of 60 students, were identified to participate. In total, 2397 students were
enrolled in these classes. In order to be included in the study, those students under the age of
18 were required to return a signed parental consent form and a signed subject assent. Verbal
telephone consent was obtained for those not returning a written consent form. Parental
consent was not required for students over 18 years old. The University of Southern
California’s Institutional Review Board approved all study procedures. Of the enrolled
students, 1694 (70.7%) consented to participate in the study. . Reasons for non-participation
included 1) parental non-response (23.4%), 2) student decline (5.1%), and 3) parent decline
(0.8%; Lisha et al., 2012).
Self-report survey data was collected from students at pretest (T1), immediately
following the classroom portion of the program (T2), and approximately one year later (T3).
All pre- and post-test surveys were administered at school. At one-year follow-up, surveys
were administered at the schools or by telephone if the student was no longer enrolled at the
school. The classroom portion of the program was provided by four female health educators
(three Caucasian and one Hispanic), aged 25, 27, 45, and 45. All educators had previous
experience providing health education curricula. Health educators were counterbalanced
across conditions so that each instructor taught in each condition.
18
Overall intervention effects indicated that any Project TND programming (i.e.,
received TND-only or TND+MI) showed significant positive effects on lowering the
proportion of substance users and reducing the number of times respondents reported using
alcohol, tobacco, marijuana, or hard drugs in the past 30 days compared to the control group.
However, there were no statistically significant differences between the TND+MI and TND-
only groups (Sussman et al., 2012).
Motivational Interviewing Booster
Booster Design and Implementation. The booster intervention employed in the
TND+MI condition consisted of three 20-minute sessions between the youth and an MI
interventionist. The first session was conducted in person one to three days after the
completion of the classroom-based instruction and the immediate posttest administration. The
second and third sessions were conducted via the telephone at three- to four-month intervals.
In cases where the MI interventionist was unable to meet the youth in person for the first
session, participants were contacted by telephone and the second session was attempted in
person if they were still attending the same school. We attempted to audiotape all MI sessions
for subsequent content coding.
The MI booster intervention structure included seven components: an opening, finding
a target behavior, exploring ambivalence, summarizing, asking a key or transitional question,
action planning and closing. The structured opening informed the youth of the purpose of the
session (to gather their impressions about the TND program and discuss a behavior of interest
to them), discussed their rights and the limits of confidentiality, and provided an opportunity
to decline audio recording if he/she desired. For the second and third sessions, the opening
19
consisted of reestablishing an understanding of the purpose of the call and checking-in
regarding the topic or behavior discussed in the previous session.
Once interventionists perceived adequate rapport, they shifted their attention to
establishing a behavioral target for change. Interventionists prioritized finding a substance use
target behavior when possible. If students reported use of multiple substances, the
interventionist either asked the student which of the substances they wished to focus on, or
interventionists proceeded with the topic they judged most problematic or amenable to change
based on details provided by the student. In cases where students disclosed not using any
substances, they were either directly asked if there was a health behavior they would like to
work on or interventionists used an agenda-setting tool (Spruijt-Metz, Barnett, Resnicow,
2011) that allowed the youth to choose among a variety of target behaviors. According to
Rollnick et al. (1999), an agenda-setting tool facilitates client engagement as they select the
topic and can increase overall effectiveness. Based on input from the pilot study, we designed
an agenda-setting tool that covered a wide range of adolescent health and life concerns,
including getting a job, graduating from high school, practicing safe sex, smoking cigarettes,
drinking alcohol, smoking marijuana, using club drugs, becoming independent/moving out,
exercise, healthy eating, going to college, choosing friends, and having a baby. For the
second and third sessions, interventionists began by following up on the previously
established target Depending on reported progress or change, the interventionists either
decided to continue to pursue the topic or went through the agenda-setting process again to
find a new target behavior.
Once the target behavior had been established during the first session, interventionists
explored ambivalence by inquiring about the pros and cons of the behavior. If appropriate,
20
they also explored the pros and cons of changing the behavior. The second session employed
the use of a personal values exercise (Miller, de Baca, Matthews, & Wilbourne, 2001), where
students were asked over the phone to select three of 15 values that were read to them, such as
being a “good student,” “good son/daughter,” “good brother/sister,” and “belief in God.”
Once students had selected the values, interventionists inquired about why they chose those
particular values. Ultimately, they asked how the target behavior fit in with these values. For
the third session, interventionists asked students to choose three words from a list of positive
attributes, such as “reliable,” “strong,” “honest,” and “trustworthy” that the student felt
described them (Miller, Hedrick, & Orlofsky, 1991; Miller, Arciniega, 2004). These words
were then used by the interventionist to affirm client strengths, and support client self-efficacy
to change. The interventionist also explained that these words described people who were
successful in making changes in their behavior. Interventionists then asked students to
describe how they felt these attributes might be helpful to them in their efforts to change the
behavior they had been discussing.
After completing the exploration exercises, the protocol indicated the use of a
transitional summary and key question to reinforce the youth’s stated importance or
confidence to change, and a question that invited the youth to consider what his/her next steps
would be. If the student responded to the transitional question by indicating that some type of
action or change was needed, the interventionist proceeded to elicit action steps. If a student
indicated feeling stuck or ambivalent about making change, the interventionist acknowledged
this and proposed that the action be limited to checking in about the topic again in a few
months to see if his/her thoughts or feelings had changed. Finally, once all of the other steps
had been conducted, the closing consisted of thanking the student for his/her engagement,
21
openness and thoughtfulness, expressing optimism about his/her proposed change or
enthusiasm about talking to him/her again, confirming the best phone number to reach
him/her, and establishing an approximate time that the student would be called again.
Training and Supervision of Interventionists. We hired interventionists using a
two-stage process. First, applicants completed an adapted Helpful Response Questionnaire
(HRQ; Miller, et al., 1991), a tool used to assist in assessing counselor empathy. We used a
five-item measure asking them to write a response to a client statement that would indicate
they were listening (e.g., client statements included “Just because I use drugs doesn’t make it
a problem. Everybody uses drugs.”) Applicants responding with open-ended questions or
reflective listening received invitations to interview. In addition to a structured in-person
interview, applicants participated in a recorded mock-telephone interview. Two Motivational
Interviewing Network of Trainers (MINT) trained project staff reviewed the recordings for
both global skills and behavior counts as set forth in the Motivational Interviewing Treatment
Integrity (MITI 3.0; Moyers, et al., 2007) coding scheme. We selected interventionists based
on their ability to meet the global skill proficiency standard in the MITI.
We hired and trained a total of 15 interventionists, hoping to keep caseloads
manageable at 20-30 students per interventionist and to have the same interventionist for all
contacts with each student. All interventionists had at least a 4 year college degree. We
provided a minimum of 40 hours of MI training to interventionists and used the Video
Assessment of Simulated Encounters (VASE –R) as a post-test measure of skill (Rosengren,
Baer, Hartzler, Dunn, & Wells, 2005). The VASE-R presents three video scenarios that
trainees watch and then provide written responses to client statements. These statements are
22
then rated based on established criteria set forth in the manual. All interventionists met these
criteria by the end of training.
Data Collection and Management. We attempted to record all sessions between the
students and MI interventionists. In-person interviews were recorded with hand-held digital
recorders, while telephone sessions were recorded via a web-based client resource
management (CRM) system. The CRM provided an interface for interventionists to access
their caseload and keep notes about their conversations. The CRM tracked the date and time
of each attempt, providing information to supervisors for staff-management purposes.
Following each MI session, interventionists completed a written client engagement measure
that included identifying the target behavior and completing six items that assessed (1) how
comfortable the interventionist felt during the call, (2) how much rapport they felt, (3) how
engaged they believed the student to be, (4) how helpful they found the protocol to be, (5)
their beliefs about the helpfulness of the call, and (6) the likelihood that the participant would
follow through with the behavior change discussed. For each item, responses were measured
using a 5-point Likert scale ranging from “not at all” to “extremely.” Factor analysis revealed
that all items loaded on a single factor with a Cronbach’s alpha of 0.87 (Barnett, et al.,
2012a).
In total, interventionists conducted 1045 sessions, with 527 students, 92% of the
sample (see Table 1-1 for MI condition demographic and prevalence data). Four hundred
sixty two (80.6%) were reached for the first contact, 352 (61.4%) were reached for the second
contact, and 226 (39.4%) were reached for the third contact. Of those reached, 31% (n = 178)
of students were reached once, 36% (n = 207) were reached twice and 24% (n = 139) were
23
reached three times. Approximately 30% of sessions discussed substance use targets
(marijuana = 116, alcohol = 57, tobacco = 54, hard drugs = 8), while more than 50% talked
about graduating from high school or finding current or future employment (See Table 1-2 for
details). Of the entire sample, 856 (82%) of sessions were recorded; reasons for not being
recorded included participant declines, or technical problems recording in person or with the
telephone recording system. Of the substance use targets, 231 had available recordings.
Interventionists recorded target behaviors on a “counselor perception of client engagement”
measure completed after each session. All of these targets were later reviewed and confirmed
by coders (more information on coders provided below). In order to be considered a
substance-use target, substance use had to be addressed with the exploration exercise. For
example, if an interventionist asked about substance use and the student reported that they had
cut back, and the interventionist moved on to another topic, this would not be considered a
substance-use target even though the topic was mentioned. During this review, 15 recordings
were determined not to meet criteria as substance-use targets. The final sample of taped
substance-use target interviews was 223, representing 170 different students.
Fidelity to Intervention Protocol. In order to assess fidelity to the protocol, we
developed a dichotomous scale to assess whether interventionists adhered to each of its seven
components. That is, we coded calls for the presence of the opening, establishing a target,
exploring ambivalence, transitional summary, key question, action plan, and closing. Each
session could earn from 0 to 7 points. The majority of calls (69%) had five or more of the
components. The component most likely to be excluded was creating a change plan; only 40%
of the calls contained this element. The change plan was excluded when interventionists
determined it was inappropriate based on their perception of the student’s readiness to change.
24
Quality of Motivational Interviewing. Considered over all interventionists, the
intervention exceeded the proficiency standards for all global measures and behavior counts
set forth in the MITI 3.0 (Moyers, Martin, Manuel, Miller, & Ernst, 2007). Eighty-eight
percent of the sessions exceeded proficiency in percent of MI-Consistent behaviors, 68%
exceeded proficiency in percent of complex reflection, 62% exceeded proficiency in percent
of open questions, and 64% exceeded proficiency in reflection-to-question ratio. See table 1-3
for interventionists’ proficiency scores.
Training and Supervision of Coders. We provided 40 hours of initial training in the
coding instrument and software to five undergraduate and graduate students. Weekly coding
meetings were held throughout the project to improve or maintain reliability. Final coding
decisions were made by the supervisor. Coders practiced on a series of non-substance use
recordings until their inter-rater reliability was at criterion of 0.60 using established intraclass
correlation (ICC) guidelines (Cicchetti, 1994).
We coded the sample using the MISC 2.5 (Houck, Moyers, Miller, Glynn & Hallgren,
2013) from the Center on Alcoholism, Substance Abuse and Addictions http://casaa.unm.edu/
download/misc25.pdf). The MISC 2.5 is a hybrid of the MISC 2.1 and the Sequential Code
for Observing Process Exchanges (MI-SCOPE; Martin, Moyers, Houck, Christopher &
Miller, 2005) designed to optimize the features from each coding systems to allow sequential
coding of MI sessions. Specifically the MISC 2.5 allows for the capture of specific behaviors
from the MISC 2.1, as well as valenced reflections and temporal order from the SCOPE. Like
all versions of the MISC, it codes counselor and client language into mutually exclusive and
exhaustive categories. Coding was performed in two passes. In the first pass, coders parsed
the entire recording into utterances, or thought units, and then completed a set of six Likert
25
global ratings of counselor interpersonal skill and one Likert measure rating client self-
exploration. In the second pass, a different coder applied behavioral codes to each counselor
utterance and each client utterance.
Coding was conducted using the CASAA Application for Coding Treatment
Interactions (CACTI; Glynn, Hallgren, Houck, & Moyers, 2012). This software automates the
parsing of recordings and stores sequential coding of each utterance with no manual data
entry. Although CACTI software does not require or utilize transcripts, we transcribed our
entire sample of recordings to allow coders to refer to the transcription in difficult cases.
Coding of Recordings. According to the MISC 2.5, all client utterances are
assigned one of 15 client language codes. Client statements are categorized as either change
talk (CT), counterchange talk (CCT), or unrelated to change (FN; Follow/Neutral).
Determining change talk requires that coders know the target behavior before coding each
recording. As previously mentioned, interventionists documented targets at the time of the
session, and these classifications were confirmed by the parser. Disagreements were resolved
by a supervisor. As done in Moyers et al. (2009), we collapsed all change talk categories into
CT and CCT for increased reliability and ease of interpretation. Component categories of CT
included statements of commitment (“I will cut back on smoking”), taking steps ("I’ve
already slowed down"), desire (“I want to quit”), ability (“I think I can do it”), reason (“I have
to stop for my health”), need (“I need to cut back so I can keep a job”) and “other” statements
that do not fall into the previous categories. Components of CCT included statements counter
to commitment (“There’s no way I will stop”), taking steps (“I had a drink last night”), desire
(“I really don’t want to make a change”), ability (“There is no way I’d be able to give it up”),
26
reason (“It’s not affecting my health”), need (“I really don’t think I need to change”), or
“other” statements about change.
Counselor Skill. Each counselor utterance was assigned one of 17 counselor codes.
Coding of reflective listening, included a designation for complexity, either SR or CR, and a
designation for valence, either a positive (POS), negative (NEG), neutral (NEU), or both
positive and negative (BOTH) valence, resulting in 8 possible codes for reflection. Other
utterances were coded as either open (OQ) or closed questions (CQ); as MI-consistent
(MICO) including specific behaviors of affirming, supporting, and asking permission before
giving advice; MI-inconsistent behaviors (MIIN) including confronting, warning, and giving
advice without permission; or “Other” for behaviors such as providing information about the
session, filler, and comments designed to facilitate conversation. Summary scores included
POQ, PCR, RQR, and PRCT (see Table 1 for formulas to calculate variables). Percent MICO
was not included in analysis due to low base-rates of MIIN behaviors occurring in the dataset
In addition to coding language, The MISC 2.5 measures counselor interpersonal skills
on a Likert scale of 1 to 5, whereby a 1 represents a low level of the characteristic and 5 a
high level. We used 2 measures of interpersonal skill empathy and spirit, a mean composite
measure of the sum of autonomy, collaboration and evocation. These global ratings
characterize the entire interview and coders are instructed to consider each attribute as a
departure from the midpoint value. According to the MISC 2.5 Coding Manual, “empathy” is
defined as the counselor’s ability to convey an understanding of the client’s perspective;
“autonomy” represents the degree to which a counselor emphasizes the client’s ability to
choose their behavior; “evocation” refers to the counselor’s ability to draw out information
27
from the client; “collaboration” is defined as the sense that the relationship between counselor
and client is that of equal partners.
Coding Reliability. We randomly selected 20% of our final sample using a random
number generator for double coding. Cicchetti’s (1994) criterion identifies ICCs below .40 as
poor, .40-.59 as fair, .60-.74 as good, and above .75 as excellent. For our data, final ICCs
were .94 for open questions, .80 for closed questions, .94 for reflections overall, .48 for
simple reflections, .45 for complex reflections, .84 for RPOS and .82 for RNEG, .68 for
MICO behaviors and .29 for MIIN behaviors. Client codes were .92 for CT, .86 for CCT, and
.88 for FN responses. These results indicate that coders had difficulty differentiating SR from
CR, and difficulty reliably identifying MIIN, which occurred infrequently. Only seven
recordings, less than 1%, contained any MIIN behaviors in our dataset
28
Table 1-1: MI Booster condition sample demographics (N=573)
Characteristic
Gender
Percent
Female 40.3
Male 59.7
Age
15 years and under
8.2
16 years old
27.6
17 years old
44.9
18 years and over
19.3
Ethnicity
Latino/Hispanic 67.7
Caucasian 6.8
Mixed 12.6
African American
5.8
Other
4.4
Asian
2.0
Native American 0.7
Drug Use Prevalence in Past 30 days
Cigarettes
40.2
Alcohol
57.2
Drunk
41.8
Marijuana
46.2
Hard Drugs
23.5
29
Table 1-2: Target behaviors in MI contacts* (n=1040)
Behavior Number Percent
Graduation 432 41.08%
Employment 130 12.33%
Marijuana 116 11.01%
Interpersonal 91 8.63%
Drug related lifestyle 87 8.25%
Alcohol 57 5.41%
Tobacco 54 5.12%
Other 31 2.94%
Nutrition 24 2.28%
Physical Activity 20 1.90%
Hard Drugs 8 0.76%
Sex 3 0.28%
*total does not equal 1040 as some contacts were coded for multiple targets
30
Table 1-3: Interventionist proficiency ratings (n=231)
MITI
Proficiency
Standard
Average
Std
Dev
Global Skills
Evocation 3.5 3.63
0.77
Collaboration 3.5 4.00
0.82
Autonomy 3.5
3.78 0.70
Direction 3.5
4.18 0.90
Empathy 3.5 3.95
0.69
Behavior Count Summary Scores
Percent MI Consistent 90% 98.9%
Percent Open Question 50% 56.8%
Percent Complex 40% 56.2%
Reflection to Question Ratio
1:1 1.33
31
CHAPTER 2: MOTIVATIONAL INTERVIEWING AND DECISION MAKING
Introduction
Adolescent substance use is a major cause of adolescent morbidity and mortality
in the United States (Brannigan, Schackman, Falco, & Millman, 2004; Sussman & Ames,
2001) and school-based prevention programs are a common approach to addressing it
(Midford, 2010). These programs frequently produce significant but small effects. In
particular, Project Towards No Drug Abuse (Project TND), a 12-session classroom-based
program, has shown consistent effects on hard drug use and less consistent results on
other drug use (Sussman, Sun, Rohrbach, & Spruijt-Metz, 2012). One way to increase
effect size of classroom programs is thought to be the addition of booster programming
(Skara & Sussman, 2003). Meta-analysis has shown that programs that provide boosters
are better able to maintain effects (Rooney & Murray, 1996; White & Pitts, 1998), while
others have noted that few direct tests of boosters have been conducted (Cuijpers, 2002).
In the most recent trial of Project TND, a direct test of the effects of booster programming,
Sussman et al. (2012) failed to find an increased effect for the TND + Motivational
Interviewing (MI) booster condition, suggesting that overall the added MI booster did not
improve the efficacy of the classroom-based program. However, effect sizes for the TND +
MI condition were larger than those reported for the TND-Only condition. This finding
suggests that perhaps the intervention effects for some people were being masked by lack of
effects in others.
The search for moderators of program efficacy, specifically the search for whom
MI interventions work best, is not new. In the MI research literature moderators have
32
been identified from three primary sources of information: (1) results of Project MATCH,
the United States’ largest test of matching alcohol interventions to client characteristics
conducted in the early 1990s, (2) meta-analyses of MI efficacy, and (3) individual RCTs
that compare MI to other treatments. Findings from the Project MATCH show the MI
treatment condition to be better suited for persons scoring high in anger (Karno &
Longabaugh, 2004), persons scoring high psychological reactance (Karno & Longabaugh,
2005), and for self-reported high worriers (Westra, Arkowitz, & Dozois, 2009). Meta-analyses
have shown that persons of minority ethnicity status and persons with low motivation
to change respond better to MI (Hettema, Steele, & Miller, 2005; Lundahl, Kunz, Brownell,
Tollefson, & Burke, 2010). However, because studies do not consistently measure client-
level psychosocial variables, meta-analyses are unable to compare them. Readiness to
change (RTC) appears as the only psychosocial variable consistently tested, but because
methods for measurement and results are inconsistent, it is difficult to draw conclusions
about its importance. Findings from individual studies of college students show that MI
interventions work better for students with low future orientation, high self-regulation skills
(Carey, Henson, Carey, & Maisto, 2007) and high “need for cognition,” the tendency to
engage in and enjoy effortful cognitive processing (Capone & Wood, 2009).
Under investigation in this study is a 3-session MI booster that used a decisional
balance activity with all participants in the first session. Through meta-analytic approaches,
Apodaca and Longabaugh (2009) found that the use of decisional balance/exploration of pros
and cons has been associated with improved substance use outcomes (LaBrie, Pedersen,
Earleywine, & Olsen, 2006; Strang & McCambridge, 2004). The decisional balance exercise,
derived from Janis and Mann’s Theory of Conflict Decision Making (Janis & Mann, 1977), is
33
a structured activity designed to elicit both the pros and cons of behavior change. Janis and
Mann propose that weighing the pros and cons of options are central to the decision-making
process and that individuals differ with respect to their ability or inclination to do so. The
competing influence of pros and cons or ambivalence is common to persons considering
behavior change (Miller & Rollnick, 2002). Thus, we hypothesized that since MI
interventions rely heavily on the counselor’s ability to assist clients to explore ambivalence,
client differences in decision-making style or ability may interact with the intervention to
influence outcomes.
Understanding the influence of individuals' confidence in their decision-making
abilities, their decision-making style (i.e., active or avoidant), and the strength of their coping
style defined by decision making could greatly enhance our ability to target adolescents who
would benefit from the personalized assistance provided via an MI intervention. We proposed
that treatment effects would differ based on individual decision-making characteristics, thus
making the MI intervention a better "match" for some youth than others. We hypothesized
that MI would be more effective for youth who are active decision makers, have high
decision-making confidence, and score high in decision-making coping style.
Methods
Sample and Procedures
Data for this study were taken from 16 of the 24 continuation high schools that
participated in the 3-arm randomized controlled trial of the Project TND. This study included
only those schools assigned to the classroom program only (TND-Only), or the classroom
program plus Motivational Interviewing (TND+MI) condition. At least two classrooms from
34
each school, with a minimum of 60 students, were identified to participate. For more
information on school selection see Lisha et al. (2012). In order to be included in the study,
students under the age of 18 were required to return a signed parental consent form and a
signed subject assent. Parental consent was not required for students over 18 years old.
Baseline data was collected in the classroom, while one-year follow-up was collected either in
person at the school or via telephone if the student was no longer attending the same school.
The University of Southern California’s Institutional Review Board approved all study
procedures. For details about program efficacy and the MI intervention see Sussman et al.
(2012) and Barnett et al. (2012a), respectively. This sample included all students with
complete baseline and one-year follow-up data in either the TND only or TND + MI
conditions (N = 658).
Measures
Demographic information about age, gender, ethnicity, and highest level of parental
education was collected via survey. Ethnicity was collected as a seven-level categorical
indicator coded as White/Caucasian, Latino/Hispanic, African American/Black, Mixed
Ethnicity, Asian, and American Indian/Native American, or "Other." Due to the sample
distribution, 61% Hispanic with no other group over 10%, we collapsed ethnicity to
Latino/Non-Latino for analysis. Parental education was measured separately for each parent
and was recorded as the highest value from either parent. Response choices included six
options ranging from “did not complete 8th grade” to “attended or completed graduate
school.”
Respondents indicated how many times they used each of the following drugs during the
previous 30 days: alcohol, tobacco, marijuana, cocaine, hallucinogens, stimulants, inhalants,
35
ecstasy, pain killers, tranquilizers, or other drugs such as PCP, steroids, GHB, and K. For each
substance, there were 12 response categories (0, 1-10, 11-20, 21-30, 31-40, 41-50, 51-60, 61-70,
71-80, 81-90, and 91-100+ times). For data analyses, a dichotomous variable was created where
the outcome was defined as "true" if a specific substance was used one or more times in the past
30 days. A hard-drug composite measure was constructed that summed responses to six items
regarding use of cocaine, hallucinogens, inhalants, stimulants, ecstasy, and “other” drugs (i.e.,
depressants, PCP, steroids, heroin, or other drugs) in the last 30 days (α =0.91).
Three characteristics of decision making were measured. First, a decision-making style
of coping, taken from the Coping Assessment Battery (Bugen & Hawkins, 1981; Wills, 1986)
was measured using three 5-point Likert scale items ranging from “never” to “always.” Items
included “I think hard about what steps to take,” "I think about the choices before I do
anything,” and “I consider my actions very carefully” (mean 3.4, se l.0, α=.67). Decision-
making confidence was measured using three items taken from the Adolescent Decision Making
Questionnaire (ADMQ) (Friedman & Mann, 1993; Tuinstra, Van Sonderen, Groothoff, Van den
Heuvel, & Post, 2000). Items included “I like to make decisions myself,” “I think that I am a
good decision-maker,” and “The decisions I make turn out well.” These three items were reverse
coded (mean 2.7, se 0.59, α =.69). Active decision-making was measured using three items also
taken from the ADMQ. Items included “When faced with a decision, I go along with what
others suggest,” “I'd rather let someone else make a decision for me so that it won't be my
problem,” and “I prefer to leave decisions to others” (mean 3.4, se 0.54, α =.74).
Statistical Approach
Multi-level mixed logistic models (PROC GLIMMIX) were used to capture the
random effects of clustered data in SAS (SAS Institute, 2008). All models controlled for age,
36
gender, ethnicity, highest level of parental education, and baseline drug use. Moderator
variables were tested first as continuous variables, and where significance was found, further
tests by median and tertile splits were investigated. All continuous variables were
standardized to a mean of zero and a standard deviation of one for ease of interpretation.
Results
Frequencies, means, and significant differences between the sample and those lost to
follow-up or excluded from the sample due to missing data are reported in Table 1. The final
sample was younger than those lost to follow-up or missing data. The overall sample was
54% male, with a mean age of 16.7 years, and racial composition of 61% Latino, 11%
Caucasian, 5% African American, 3.5% Asian, 15% mixed race, 3% other, and less than one
percent Native American/Alaskan Islander.
Correlation results, presented in Table 2, showed two significant associations for hard
drug use, active decision making (r = -.12 , p = .001) and cognitive decision making (r = -.10
, p = .01). In this data being male was significantly correlated with having smoked in the past
30 days, and age was negatively associated with marijuana and hard drug use. Results of the
mixed logistic regression models, presented in Table 3, show no main or interaction effects
between treatment condition and decision making variables in the prediction of alcohol,
cigarette, marijuana, or hard drug use.
Discussion
Based on this analysis, decision making does not appear to influence MI efficacy and
should not be used to target participants for MI interventions. Rather, it appears that MI works
equally well regardless of a youth's decision-making style or confidence. Our data suggest
37
that MI may reinforce or complement decision-making strengths, as well as compensate for
decision-making weaknesses or maladaptive propensities.
Our ability to identify differential effects between the conditions may be limited due to
the length of time between students' MI sessions and the follow-up data collection. The
effects of brief interventions have been shown to decrease over time and may have needed to
be assessed earlier (McCambridge & Strang, 2005). Approximately one-third of the sample
received only one MI session, which occurred immediately after the classroom program
(Barnett et al., 2012a). For these students, a 12-month follow-up period may have been too
long to see results from one session of MI. Furthermore, we have no way to assess the effects
of the intervening attempts to contact participants. Our attempts may have signaled concern or
a reminder about the program message thereby decreasing substance use, or conversely, our
attempts may have signaled “nagging” and provoked a reactant response increasing their
participation in substance-using behaviors.
The ability of this study to find effects may also been limited by our measurement of
decision making characteristics and substance use outcomes. Although the decision making
measures all showed adequate internal reliability, ranging from α = .69 - .74, we used only
three items from each scale. Furthermore, although we looked at three conceptualizations of
decision making, competence, approach/avoidance, and coping mechanism, none included an
objective measure of student’s competence. Meanwhile the measurement of drug use in this
study relied on a 12-level ordinal scale that was not able to differentiate low levels of use. A
continuous measure or one capable of differentiating between occasional weekend use, using
every weekend, or using during the school week would have enhanced our ability to detect
effects.
38
Finally, findings from this study may lack generalizability due to the inclusion of only
high-risk adolescents, more than 60% of whom were Latino, in southern California. It is
unknown whether the decision-making characteristics of at-risk youth differ from regular high
school students or whether there exist cultural/ethnic differences in decision making that may
have influenced these findings. Though there is more to learn about decision making as a
moderator of MI effects, this study contributes to the MI literature by providing information
about one possible moderator. Researchers should continue to investigate the interactions
between psychosocial variables and MI to enhance our ability to target MI interventions
where they are likely to produce the greatest effect.
39
Table 2-1: Comparison of current sample and lost to follow-up (FU) or incomplete data`
Sample Lost to FU p value*
658 466
Male % (n) 54.0 (355) 59.2 (276) 0.08
Age (M ± SD) 16.7 ± .95 16.8 ± .96 0.006
Race/Ethnicity % (n) 0.35
Asian 3.5 (23) 3.2 (14)
Latino/Hispanic 61.1 (402) 63.2 (276)
African American 4.9 (32) 7.8 (34)
Caucasian 11.4 (75) 10.3 (45)
Native American 0.6 (4) 0.7 (3)
Mixed 15.2 (100) 11.7 (51)
Other 3.3 (22) 3.2 (14)
Drug Use Prevalence in the Past 30 days % (n)
Alcohol 55.8 (367) 59.7 (259) 0.2
Cigarettes 41.3 (272) 43.0 (188) 0.58
Marijuana 43.6 (287) 48.8 (214) 0.09
Hard Drugs 27.2 (179) 26.1 (122) 0.7
Decision Making Coping Style 3.4 ± 1.0 3.4 ± 1.1 0.95
Decision Making Approach 3.4 ± .54 3.4± .60 0.74
Decision Making Confidence 2.7 ± .59 2.7 ± .68 0.50
*T-tests and Chi Square tests conducted to determine statistically difference due to attrition.
40
Table 2-2: Bivariate correlations
Age Male Active Confidence Coping Alcohol Cigs Marijuana
Hard
Drugs
Age 1
Male 0.004 1
0.91
Active DM 0.005 -0.032 1
0.90 0.42
DM Confidence 0.016 0.121 0.185 1
0.68 0.0018 <.0001
Coping DM 0.199 0.032 0.125 0.279 1
<.0001 0.41 0.0013 <.0001
30 day alcohol use (YN) 0.018 -0.049 -0.004 -0.009 -0.060 1
0.64 0.21 0.91 0.82 0.13
30 day cigarette use (YN) -0.0479 0.094 -0.059 -0.005 -0.038 0.431 1
0.22 0.02 0.13 0.91 0.33 <.0001
30 day marijuana use (YN) -0.078 0.050 -0.005 0.016 -0.067 0.394 0.425 1
0.05 0.20 0.9 0.68 0.09 <.0001 <.0001
30 day hard drug use (YN) -0.111 -0.004 -0.125 0.005 -0.100 0.366 0.416 0.420 1
0.001 0.92 0.0013 0.91 0.01 <.0001 <.0001 <.0001
DM = Decision Making; Substance use variables measure the presence of use in the past 30 days.
41
Table 2-3: Results of logistic regression predicting drug use at 1 year follow-up*
Alcohol Cigarette Marijuana
Hard Drug
Use
β (se) β (se) β (se) β (se)
Model 1: Decision Making Coping Style
DM .05 (.05) -.04 (.05) -.05 (.05) -.01 (.05)
Condition -.04 (.09) -.06 (.07) .04 (.08) -.09 (.07)
DM* Condition .002 (.08) .05 (.07) .11 (.07) -.05 (.08)
Model 2: Active Decision Making
DM -.02 (.05) -.03 (.05) .01 (.05) -.03 (.05)
Condition -.04 (.09) -.06 (.07) .04 (.08) -.10 (.08)
DM* Condition .03 (.08) .06 (.07) -.01 (.07) -.04 (.08)
Model 3: Decision Making Confidence
DM .07 (.06) .03 (.05) -.02 (.05) .05 (.06)
Condition -.04 (.09) -.06 (.07) .04 (.08) -.10 (.07)
DM* Condition -.07 (.08) -.09 (.07) .04 (.08) -.06 (.08)
*All models control for age, gender, Latino (YN), highest parental education, and
baseline substance use; DM = Decision Making; Condition TND Only = 1, TND + MI
= 0; Substance use variables measure the presence of use in the past 30 days; no
significant findings.
42
CHAPTER 3: REFLECTIVE LISTENING AND CHANGE TALK: A CLOSER LOOK
AT MOTIVATIONAL INTERIVEIWNG
In the United States, reductions in adolescent use of alcohol, tobacco, and other hard
drugs have been leveling off while marijuana use has been climbing (Johnston, Bachman, &
Schulenberg, 2012). As social service systems try to address these trends, evidence-based
practices (EBP) like Motivational Interviewing (MI) are being mandated or encouraged by state
funding agencies. The costs associated with adopting EBPs can place substantial burdens on
state, county, and local agencies (Olmstead, Carroll, Canning-Ball, & Martino, 2011), making it
imperative that resources be used wisely.
MI is a client-centered counseling style directed at the exploration and resolution of
ambivalence about behavior change. Having its roots in Rogerian client-centered therapy
(Rogers, 1959), it emphasizes the importance of accurate empathy. It also prescribes the use of
MI-consistent (MICO) behaviors such as asking permission before providing advice, affirming
the client, being supportive and emphasizing personal choice and control, while proscribing the
use of MI-inconsistent (MIIN) behaviors such as confronting, arguing, directing and warning.
Furthermore, it encourages the use of open questions rather than closed questions, and more
reflective listening than questioning (Miller & Rollnick, 2002). In fact, quality reflective
listening is considered a hallmark of the approach and can be challenging to learn for clinicians,
para-professionals and lay persons alike (Miller & Moyers, 2006).
While not all MI efficacy trials have shown positive effects, meta-analyses have
determined that MI is more effective than no treatment and requires fewer sessions to achieve
comparable results when compared to alternative treatments (Hettema, Steele, & Miller, 2005;
43
Lundahl, Kunz, Brownell, Tollefson, & Burke, 2010). Further, a growing body of research has
identified mechanisms of change at work to produce improved client outcomes. Specifically, it
has been shown that the therapist’s MI skills are associated with the presence of client change
talk (CT), language that expresses the client’s desire, ability, reason, need or commitment to
change, and the amount of CT expressed is associated with improved outcomes (Catley et al.,
2006; Gaume, Bertholet, Faouzi, Gmel, & Daeppen, 2010; Gaume, Gmel, & Daeppen, 2008;
Gaume, Gmel, Faouzi, & Daeppen, 2008; Moyers, Martin, Houck, Christopher, & Tonigan,
2009; Moyers & Martin, 2006; Moyers, Martin, Christopher, et al., 2007). However, less is
known about the relative importance of the constituent skills within MI, providing little
guidance to counselors when choosing which skill to employ during treatment interactions. For
example, when facing an ambivalent statement from a client, should an interviewer choose a
reflection supporting change or an expression of empathy? MI theory does not provide a
hierarchy for selecting among the array of appropriate responses. Greater understanding of MI
skills at a micro level can enhance our ability to efficiently train practitioners and improve client
outcomes.
Specifically, the skill of reflective listening deserves a closer look. While there are many
ways to conceptualize reflective listening, this study will focus on complexity and valence, as
current MI coding schema include reliable measures for both. Complexity includes simple
reflections (SRs) that parrot or paraphrase client statements and complex reflections (CRs) that
add direction, meaning, emphasis and/or feeling to client statements. For example, in response to
a client statement “Quitting is too hard,” a counselor could respond with, “It’s really hard (SR).”
or “You feel like it is impossible (CR).” The importance of using CR is highlighted by the fact
that the standards used to establish fidelity to MI require more than 50% CR. This emphasis on
44
complex reflections is intended to offset the superficiality that occurs when simple reflections are
overused, as well as to deepen the client’s self exploration within the session (Miller & Rollnick,
2002).
Valence, on the other hand, refers to a clinician’s choice to reflect client statements
directed either toward or away from change. For instance, in response to hearing “Quitting is too
hard,” a negatively valenced reflection (RNEG) might be “It’s something you don’t think you
can do.” Conversely, in response to hearing “I really need to do this,” a positively valenced
reflection (RPOS) might be “You’re ready for change.” Frequently clients present language
toward and away from change in the same sentence, forcing the counselor to choose which
aspect to reflect. While the concept of simple and complex reflections are widely taught,
discussed and coded, valenced reflections are not. The discrepant treatment between complexity
and valence does not appear to have any empirical basis and is the primary impetus for this
study. Clarifying their respective importance would allow MI trainers and practitioners to hone
their efforts to improve practice and enhance treatment effectiveness.
Beyond establishing an association between client language and clinician behaviors, new
research is beginning to focus on the question of whether there is a causal relationship between
them. Can counselors intentionally manipulate client language, or is change talk simply a
reflection of some third variable, such as motivation to change? Glynn and Moyers (2010)
conducted an experimental manipulation using an ABAB design to investigate whether the use
of MI would produce significantly more CT than the comparison functional analysis (FA)
intervention. They found 13% more CT in the MI condition compared to the FA condition.
More commonly, studies of the temporal relationships between counselor and client language
have been conducted using sequentially coded data of interactions between therapists and clients
45
in MI sessions. In these studies, clinician speech is typically coded into five general categories,
MI consistent (MICO), MI inconsistent (MIIN), Reflections, Questions, and Other behaviors;
while client language is put into three categories, CT, CCT, and language unrelated to change
(FN). These studies confirm that MICO behaviors are more likely to be followed by CT, while
MIIN behaviors are more likely to be followed by CCT (Gaume et al., 2010; Gaume, Gmel,
Faouzi, et al., 2008; Moyers & Martin, 2006).
With respect to reflective listening, in a sample of young adult army recruits in Luasanne,
Switzerland, Gaume et al. (2010) found that CT followed SR 43% and CR 50% of the time;
while CCT followed SR 27% and CR 29% of the time, indicating that SR and CR do not produce
major differences in the type of client speech that follows. Meanwhile, findings about the
valence of reflections were reported by Moyers et al., (2009b) in their analysis of coded sessions
of adult alcohol users. They found that CT followed RPOS 44% and RNEG only 7% of the time,
and CCT followed RPOS 3% of the time and RNEG 38% of the time. Similar patterns were
seen in Moyers, Houck, Glynn, & Manuel (2011) whereby coded recordings showed CT
followed RPOS between 46-51% (across both conditions) and was not significantly associated
with RNEG, while CCT followed RNEG 37% of the time and was not significantly associated
with RPOS.
The association between reflective listening and outcomes has also been demonstrated in
studies using frequency count data. In a sample of African American smokers, Catley et al.
(2006), using the Motivational Interviewing Skill Code (MISC; Miller, Moyers, Ernst, &
Amrhein, 2003), conducted a series of regression models and found that an inclusive category of
reflection predicted CT (B = .57, SE .10, p <.001). While Gaume, Gmel, Faouzi, & Daeppen
(2009) found that % CR significantly predicted baseline to 12-month follow-up differences in
46
weekly drinking among a sample of hazardous drinkers treated in an emergency department in
Lausanne, Switzerland (B = .65, SE .15, p <.01).
Taken together, these findings support the hypothesis that counselor reflections may
cause client CT and thereby influence outcomes. But questions still remain as to whether
complexity or valence is a more important component of reflections. We are not aware of any
published studies that have compared both conceptualizations of reflective listening. In the
present study, we explore the temporal relationship between both types of reflections and CT.
We hypothesize that a) CR will more frequently precede CT than SR, b) RPOS will more
frequently precede CT than RNEG, and c) a comparison of estimates will show that valence is
more strongly associated with CT than complexity.
Methods
Sample & Procedures
In order to investigate these relationships, we used a subsample of MI sessions taken
from the MI condition of the most recent trial of Project Toward No Drug Abuse, a classroom-
based substance use prevention program, designed to investigate whether a three-session, MI-
booster enhanced program effects (Sussman, Sun, Rohrbach, & Spruijt-Metz, 2012). This
cluster-randomized controlled trial consisted of 24 continuation high schools (CHS) in three
conditions, classroom-only, classroom + MI, and assessment-only control. CHSs have notably
higher drug use prevalence rates than regular high schools and serve students unlikely to
graduate from traditional high school due to lack of credits and excessive absences. All study
procedures were approved by the University of Southern California’s Institutional Review
Board. For more details on school selection see Lisha et al., 2012.
47
The intervention consisted of three 20-minute contacts conducted at three- to four-month
intervals. Students received the first contact at school within three days of the classroom-based
program, while the second and third contacts were conducted over the telephone. Multiple
attempts were made to reach each student during each contact period. Each contact was
structured to focus on the use of an exercise (pros and cons, values, or character strengths)
designed to elicit change talk about a behavior the student was interested in changing. Of the
1040 contacts, approximately 30% focused on the use of substances. Mean number of contacts
per participant was 1.8 and mean length of conversations was 18.9 minutes.
Nineteen interventionists were provided 40 hours of training. All interventionists had at
least a four-year college degree. Initial training and ongoing supervision/coaching was conducted
by a member of the Motivational Interviewing Network of Trainers. Overall, the MI delivered
met the MI fidelity standards identified in the Motivational Interviewing Treatment Integrity
(MITI 3.0; Moyers, Martin, Manuel, Miller, & Ernst, 2007). Further detail on the development
and implementation of the MI booster intervention is published separately (Barnett et al., 2012).
For this study, only MI sessions where the behavioral target was substance use (n = 223,
representing 170 individual subjects and 17 interventionists) were included in the sample.
Substance use targets were identified by the interventionists and confirmed by the coders. In
order to be considered a substance use target, substance use had to be addressed with the
exploration exercise used during the session. For example, if a participant reported that he or she
had cut back on cigarette use, and the interventionist proceeded to explore a non-substance use
topic, this session would not be considered a substance use target.
48
Coding and parsing
We provided 40 hours of initial training in the coding instrument and software to five
undergraduate and graduate students. Weekly coding meetings were held throughout the project
to improve or maintain reliability. Final coding decisions were made by the supervisor. Coders
practiced on a series of non-substance use recordings until their inter-rater reliability was at
criterion of 0.60 using established intraclass correlation (ICC) guidelines (Cicchetti, 1994).
We coded the sample using the MISC 2.5 (Houck, Moyers, Miller, Glynn & Hallgren,
2013) from the Center on Alcoholism, Substance Abuse and Addictions (http://casaa.unm.edu/
download/misc25.pdf). The MISC 2.5 is a hybrid of the MISC 2.1 and the Sequential Code for
Observing Process Exchanges (MI-SCOPE; Martin, Moyers, Houck, Christopher & Miller,
2005) designed to optimize the features from each coding systems to allow sequential coding of
MI sessions. Specifically the MISC 2.5 allows for the capture of specific behaviors from the
MISC 2.1, as well as valenced reflections and temporal order from the SCOPE. Like all versions
of the MISC, it codes counselor and client language into mutually exclusive and exhaustive
categories. Coding was performed in two passes. In the first pass, coders parsed the entire
recording into utterances, or thought units, and then completed a set of six Likert global ratings
of counselor interpersonal skill and one Likert measure rating client self-exploration. In the
second pass, a different coder applied behavioral codes to each counselor utterance and each
client utterance.
Coding was conducted using the CASAA Application for Coding Treatment Interactions
(CACTI; Glynn, Hallgren, Houck, & Moyers, 2012). This software automates the parsing of
recordings and stores sequential coding of each utterance with no manual data entry. Although
49
CACTI software does not require or utilize transcripts, we transcribed our entire sample of
recordings to allow coders to refer to the transcription in difficult cases.
Client Language
According to the MISC 2.5, all client utterances are assigned one of 15 client language
codes. Client statements are categorized as either change talk (CT), counterchange talk (CCT), or
unrelated to change (FN; Follow/Neutral). Determining change talk requires that coders know
the target behavior before coding each recording. As previously mentioned, interventionists
documented targets at the time of the session, and these classifications were confirmed by the
parser. Disagreements were resolved by a supervisor. As done in Moyers et al. (2009), we
collapsed all change talk categories into CT and CCT for increased reliability and ease of
interpretation. Component categories of CT included statements of commitment (“I will cut back
on smoking”), taking steps ("I’ve already slowed down"), desire (“I want to quit”), ability (“I
think I can do it”), reason (“I have to stop for my health”), need (“I need to cut back so I can
keep a job”) and “other” statements that do not fall into the previous categories. Components of
CCT included statements counter to commitment (“There’s no way I will stop”), taking steps (“I
had a drink last night”), desire (“I really don’t want to make a change”), ability (“There is no
way I’d be able to give it up”), reason (“It’s not affecting my health”), need (“I really don’t think
I need to change”), or “other” statements about change.
Counselor Behavioral Skill Counts
Each counselor utterance was assigned one of 17 counselor codes, including a
designation for complexity, either SR or CR, and a designation for valence, either a positive
(POS), negative (NEG), neutral (NEU), or both positive and negative (BOTH) valence, resulting
50
in 8 possible codes for reflection. For the purposes of these analyses, four summary variables
were made: 1) all SR regardless of valence were collapsed into an SR category; 2) all CR
regardless of valence were collapsed into a CR category; 3) all SR and CR with positive valence
were collapsed into an RPOS category; and 4) all SR and CR with a negative valence were
collapsed into an RNEG category. In other published literature RPOS and RNEG have been
called a Reflection of Change Talk (RCT) and Reflection of Counter Change Talk (RCCT),
respectively (Moyers et al., 2009). We chose to refer to them as RPOS and RNEG to highlight
the concept of valence under investigation in this study.
Coding Reliability
We randomly selected 20% of our final sample using a random number generator for
double coding. Cicchetti’s (1994) criterion identifies ICCs below .40 as poor, .40-.59 as fair, .60-
.74 as good, and above .75 as excellent. For our data, final ICCs were .94 for open questions, .80
for closed questions, .94 for reflections overall, .48 for simple reflections, .45 for complex
reflections, .84 for RPOS and .82 for RNEG, .68 for MICO behaviors and .29 for MIIN
behaviors. Client codes were .92 for CT, .86 for CCT, and .88 for FN responses. As seen in other
studies, these results indicate that coders had difficulty differentiating SR from CR, and
difficulty reliably identifying MIIN, which occurred infrequently (Gaume et al. 2010; Moyers et
al. 2009). Only seven recordings, less than 1%, contained any MIIN behaviors in our dataset.
Analytical Approach
Conditional probabilities (CP) were calculated using GSEQ 5.1 software (Bakeman & Quera,
2002). CPs measure the likelihood that a specified counselor behavior will precede a certain
client behavior. We calculated the number of observed and expected transitions between client
51
and counselor categories, and the odds ratios and confidence limits for each transition, to
determine whether the behavior is more or less likely than chance to occur (Bakeman & Quera,
1995).
Results
The transition matrix included 14,505 transitions taken from 223 sessions. We analyzed only
those transitions where speech transitioned from counselor to client. Table 1 provides observed
and expected frequencies, and odds ratios with corresponding confidence limits. Confidence
limits that do not include zero indicate that the transition is significantly different than would be
expected by chance. For instance, confidence limits for all three transitions including SR include
zero, hence we conclude that SR was not significantly more likely than chance to be followed by
CT, CCT, or FN. All other transitions were significantly different than would be expected by
chance. Data showed that CR was 89% more likely than chance to be followed by CT, 62% more
likely to be followed by CCT and 57% less likely to be followed by FN., RPOS was almost 11
times more likely than chance to be followed by CT, 71% less likely to be followed CCT, and
85% less likely than chance to be followed by FN. While RNEG was 19 times more likely than
chance to be followed by CCT, and 65% less likely to be followed by CT and 84% less likely to
be followed by FN.
Discussion
Our data indicate that complex reflections on the part of interviewers are better suited for
encouraging client language in favor of change when compared to simple reflection. This
provides support for the current emphasis on training practitioners to go beyond parroting and
paraphrasing what clients say, and instead to infer meaning, feeling and direction to their words.
52
Our second hypothesis is also supported, as RPOS is much more likely than RNEG to be
followed by CT, a finding that is consistent with the current emphasis in MI training to
encourage change talk by reflecting it.
This research highlights the significant likelihood that a client will continue to talk about
not changing if a counselor reflects CCT. Counselors reflect CCT for both intentional and
unintentional reasons. First, MI practitioners may choose to reflect CCT to demonstrate empathy
and understanding for the sake of the client/counselor relationship. Second, counselors may be
using an amplified reflection, which involves reflecting CCT in an exaggerated fashion, as a
strategy for moving beyond a clinical impasse (Miller and Rollnick, 2002). Third, they may be
employing double-sided reflection where the counselor reflects both CT and CCT
simultaneously, to highlight the client’s ambivalence about change.
Another intentional response to CCT that should be mentioned is the use of reframing.
According to Miller and Rollnick (2002), reframing reinterprets or transforms the client’s words
“in a new light that is more likely to be helpful and support change" (p. 103). Training
practitioners to respond to client CCT by reflecting some aspect of CT could result in increased
utterances of CT while demonstrating understanding and empathy. For example, if a client states
that “Quitting is too hard" (CCT), the counselor might reframe it to “You’re really trying"
(RPOS). Although this is a sophisticated skill that may be difficult to inculcate through training,
enhanced ability to reframe is likely to result in increased CT.
Likely what is more problematic is the unintentional use of RNEG that subsequently may
reinforce CCT. RNEG may indicate a novice use of MI where a counselor engages in rote
repetition of client statements and lacks the directive element of MI. This phenomenon is
commonly expressed by trainees who feel they are reflecting what is being said, but not making
53
progress toward behavior change. RNEG may encourage CCT by simply attending to it, causing
the client to continue discussing CCT because the counselor has signaled that it is important.
This issue is central to the ongoing debate regarding the use of decisional balance exercises, as
these exercises intentionally elicit CCT as a way to help clients think through the consequences
of behavior change (Miller & Rollnick, 2009).
RNEG may not be the only way that CCT is reinforced. Any counselor behavior that
attends to CCT may encourage more CCT and reinforce the status quo. Research has shown that
MIIN responses, correcting, advising and arguing against CCT are more likely to be followed by
CCT (Moyers, Martin, Houck, Christopher, & Tonigan, 2009). Another counselor behavior that
may reinforce CCT is problem-solving in response to barriers to change. Problem-solving by
counselors may overemphasize the barriers, inadvertently reinforce the status quo, and impede
behavior change, instead of making the problem manageable and thereby facilitating change. A
focus on barriers to change may also prematurely orient the conversation toward action, limiting
the time devoted to developing the reasons and motivation for change. This research suggests
that counselors may need to restrict the impulse to focus their attention on the barriers in order to
reinforce a client’s intention to change.
Findings from this data should be considered in light of a unique sample, approximately
75% Latino and 70% male. Though MI has been shown to be effective across ethnicity and
gender, no published MI studies have investigated differences in counselor/client language
patterns based on ethnicity or gender. Future research should investigate whether differences
exist. Findings should also be considered in light of some important choices made in this study.
First, we chose only to investigate the transition between counselor statements and client
responses. It is also highly probable that similar relationships exist whereby client speech
54
influences counselor responses. Our investigation was designed to shed light on the role of
counselor behaviors, as MI training is intended to influence these choices. Future research could
investigate how the counselor responds to client statements, and test a more complicated model
including additional transitions. Second, we chose to collapse all types of CT into one over-
arching category for ease of interpretation. Future studies should consider investigating a unique
role for each category of change talk.
Findings should also be considered in light of the following limitations. First, despite
extensive training and supervision, coders achieved only “fair” ICCs when coding SR and CR,
indicating that conclusions about these high inference variables should be made cautiously.
However, because the patterns associated with these distinctions do exist, we believe that the
distinction remains important. Second, there was some self-selection bias in this sample, as
approximately 18% of the participants declined to be recorded, leaving open the possibility that
those who declined may have demonstrated different response patterns to clinician statements.
Third, though conditional probabilities do represent a time sequence between two variables, they
only measure the association between them, and not whether one causes the other. Only
experimental manipulations that investigate differential responses to the various types of
reflections can begin to establish a causal relationship between clinician behaviors and client
change talk within MI sessions. Finally, since these analyses do not control for any other
variable, the likelihood of responding with change talk may be attributable to some other factor,
namely client readiness to change, rather than counselor skills.
In conclusion, our hypotheses were supported. Results suggest that the valence of a
reflection may be more important in eliciting change talk than whether a reflection is simple or
complex. A greater emphasis on valence, and less attention to complexity, might be a welcome
55
relief to MI trainers and MI coders alike. First, influencing CR in training appears to be a major
challenge. Of the three studies reporting the ability of training to influence trainee’s percentage
of complex reflection, only one found significant results (Moyers, et. al., 2008). Meanwhile
training that targeted trainee response to change talk increased the amount of change talk elicited
by trainees compared to the training-as-usual control (Moyers et. al., 2011). Second, a review of
reported reliability estimates shows CR to be consistently the most difficult skill to achieve a
“good” coding reliability score, while RPOS and RNEG may be easier for coders to capture.
Reliability estimates from this study support this notion. These findings also support the
recommendation by Glynn and Moyers (2011) for a new rating system to evaluate counselors’
responses to change talk. At present, the MITI is the primary tool used to establish fidelity to
MI. However, the MITI does not require coders to know the targeted behavior change, so
valenced reflection cannot be evaluated. These findings also suggest that accurate empathy,
conveyed via complex reflection, may not be the only tool for creating an environment
conducive to change in an MI session. The clinician’s ability to influence the direction of the
client’s language may be as important as empathy in determining the success of a motivational
interview.
56
Table 3-1: Counselor to client transitions
Initial event → subsequent event
Observed
frequency
Expected
Frequency
Odds
Ratio
Confidence
Limits
Simple and Complex Reflections
SR → CT 762 803 0.91 .83-1.00
SR → FN 1544 1511 1.06 .98-1.16
SR → CCT 403 396 1.03 .91-1.16
CR → CT 1293 943 1.89 1.74-2.06
CR → FN 1261 1774 0.43 .40 - .47
CR → CCT 628 465 1.62 1.46-1.80
Positive and Negative Valenced Reflections
RPOS → CT 1768 706 10.93 9.87-12.10
RPOS→ FN 483 1328 0.15 .14-.17
RPOS → CCT 130 348 0.29 .25-.35
RNEG → CT 160 348 0.35 .30-.42
RNEG → FN 220 655 0.16 .14-.19
RNEG → CCT 795 172 18.99 16.59 - 21.73
Note: SR = Simple Reflection, CR = Complex Reflection, CT = Change Talk, CCT =
Counter Change Talk, FN = Follow/Neutral, language unrelated to change, RPOS =
Reflection Positive, RNEG = Reflection Negative.
57
CHAPTER 4: FROM COUNSELOR SKILL TO DECREASED MARIJUANA USE:
DOES CHANGE TALK MATTER?
Motivational Interviewing (MI), a client-centered counseling style used for the
exploration of ambivalence about behavior change (Miller & Rollnick, 2002), has been identified
as a promising intervention for adolescent substance use treatment (Macgowan & Engle, 2010),
appropriate for addressing a range of substances across a variety of settings (Barnett, et al.,
2012b). MI also has a well-specified theoretical model, whereby counselor behaviors are
expected to promote client language predictive of change (see figure 1; adapted from Miller &
Rose, 2009), and a growing body of evidence to support hypothesized causal mechanisms or
mediational paths within it. However, full mediation tests are rare and none have examined
these paths using unique MI skills or measures of MI gestalt as predictors.
The gestalt and skills of MI are commonly measured using the Motivational Interviewing
Treatment Integrity (MITI; Moyers, Martin, Manuel, Miller, & Ernst, 2007), a coding instrument
designed to assess counselor adherence to MI. The MITI measures gestalt by assessing how well
counselors convey empathy, collaborate with clients, respect client autonomy, and evoke rather
than provide information. These last three indicators are then combined into a measure of “MI
Spirit.” Meanwhile, MI behavioral skills are measured by categorizing counselor statements as
open or closed questions, complex or simple reflections, MI consistent (MICO) or MI
Inconsistent (MIIN) behaviors. Though the definition of MICO differs slightly between coding
instruments, it is a composite measure of counselor speech or behavioral indicators that
demonstrate adherence to the “way of being” prescribed in MI. MICO includes instances of
asking permission before giving advice or making suggestions, offering support, affirming,
58
emphasizing personal choice and control, and sometimes, depending upon the measurement
instrument used, includes open questions and reflections. To date much of the evidence for
mediation has been shown using MICO as the predictor. We contend that much can be learned
by modeling the gestalt and behavioral indicators separately.
The relationship between these counselor measures and outcomes (path c) has been
addressed in previous research. Gaume et al. (2009) modeled the unique MI counselor skills
separately to predict alcohol use at 12-month follow-up in a study of alcohol-using adults in an
emergency department. In so doing, they found significant relationships between MI Spirit,
percentage Complex Reflection, and the Reflection to Question Ratio on outcomes when
controlling for client ability language. While empathy was not a significant predictor of
outcome, it showed a trend in the hypothesized direction (p = .06). Similarly, McCambridge,
Day, Thomas, and Strang (2011) found significant relationships between MI Spirit and
percentage Complex Reflection on marijuana cessation at three months in a sample of youth ages
14-19 attending Further Education Colleges in London.
Various research designs have been used to investigate the relationship between
counselor skills and client language about change (i.e., change talk [CT] and counter change talk
[CCT]; path a). Sequential analyses have provided probabilistic support that MI-Consistent
(MICO) behaviors will be more strongly followed by CT, while MI-Inconsistent (MIIN)
behaviors are more strongly associated with CCT (Gaume, Gmel, Faouzi, & Daeppen, 2008;
Moyers & Martin, 2006; Moyers, Martin, Houck, Christopher, & Tonigan, 2009). Regression
analysis of non-sequential count data have shown associations between MI Spirit and MI-
Consistent (MICO) with increased CT (Catley et al., 2006). Experimental manipulations of
counselor approaches have shown a causal relationship between counselor behavior and change
59
talk. Glynn and Moyers (2010) showed in an ABAB design between MI and Functional
Analysis (FA) that MI resulted in increased amounts of change talk compared to FA, while
Morgenstern et al. (2012) found in a 3-condition randomized controlled trial that the directive
elements of MI are more instrumental in producing CT than the non-directive elements.
CT has further been shown to predict outcomes (path b) in several studies. Though CT
has been conceptualized differently across studies, support has been found for a single category
of combined CT and commitment language to predict alcohol use outcomes (Moyers et al.,
2007), as well as a combined category of client language about reasons, desires and ability to
change to predict improvements in substance use rates in a sample of homeless youth (Baer,
Beadnell, Garrett, Hartzler, & Wells, 2008). Measures of the strength of change talk have shown
that the strength of client ability language predicted drinking rates at 12-months (Gaume, Gmel,
Faouzi, et al., 2008), and that the strength of commitment language predicted decreased
gambling and decreased monetary losses (Hodgins, Ching, & McEwen, 2009).
In spite of the support for paths a, b, and c found in these studies, there have been only
four full mediation studies published, and only three of these investigate substance-using
populations. Mediation analyses are important for establishing the mechanisms by which MI
works (see figure 1). They aid in formulating a more complete understanding of what is
occurring during treatment. Conventional understanding of mediation holds that if there is no
main effect between the predictor and the outcome then there cannot be mediation. However,
many have argued that mediation may exist even in cases where a main effect does not (Hoyle &
Kenny, 1999; Kenny, Kashy, & Bolger, 1998; MacKinnon, Fairchild, & Fritz, 2007; Shrout &
Bolger, 2002; Zhao, Lynch, & Chen, 2010). In particular, it has been argued that when the
direction of effects differs between path a and path b, it is not uncommon for path c to appear
60
insignificant. In such cases, MacKinnon, Fairchild, and Fritz (2007) describe the mediator as a
suppressor variable. This lack of main effect may be a contributing factor to the dearth of
mediation studies present in the MI literature.
The existing mediation studies are presented below. Moyers et al. (2009) found
significant main effects of MICO on outcomes (path c) and significant indirect effects (path a*b)
for MICO, CT, and drinks per week at 5-week follow-up after the personalized feedback session.
Vader, Walters, Prabhu, Houck, and Field (2010), in a sample of college-age students, did not
find evidence for significant indirect effects (path a*b), though there were significant
relationships between the MICO and CT (path a) and CT and 3-month alcohol use (path b) in the
condition with personalized feedback. They did not report any information about a main effect
for MICO on alcohol use. Morgenstern et al. (2012) conducted a three-armed randomized
controlled trial comparing a standard care control, an MI condition that included personalized
feedback and directive MI skills, and a Spirit Only condition which relied on the non-directive
elements of MI. They found significant effects for condition on commitment language (path a)
and a trend toward significance for commitment language on alcohol use at 7-day follow-up
(path b), but no significant indirect effects and no main effects. Finally, Pirlott, Kisbu-Sakarya,
DeFrancesco, Elliot, and MacKinnon (2012), in a study using personalized feedback,
investigated the use of MI to encourage fruit and vegetable consumption. This study showed
significant effects for MICO and MI Spirit on Total CT (path a) and CT on 12-month fruit and
vegetable consumption (path b), significant indirect effects (path a*b) and no main effect (path
c).
Taken together, these mediation results remain inconclusive. While both the rigor and
design of these studies are solid, any comparison of their results should be made cautiously in
61
light of the fact that they often defined their predictors, mediators, and outcome variables
differently, used different versions of similar coding instruments, had widely varying lengths of
follow-ups, and used different statistical tests for mediation. See Table 1 for details about each
study.
Because these studies used MICO, a composite variable, as the predictor variable, none
of them provide guidance as to which of the MI component skills is the most effective at eliciting
change talk. Counselors must constantly decide which skill to employ during a session. For
instance, if client states, “My friends make quitting so hard!” a counselor could respond with an
open question (e.g., “What do they do?”) or a complex reflection that identifies feeling (e.g.,
“You’re really frustrated.”). Theoretically, this choice will influence the direction of the
subsequent interactions between the client and counselor. The question will likely evoke a list of
behaviors or scenarios that promote drug use (e.g., “They bring it to my house.”). Alternatively,
the reflection allows the client to expand on what is frustrating (e.g., “Yeah. I ask them not to
bring it around but they do.”). In this example, the reflection elicited CT (e.g., the client
expressed an action that had been taken to reduce drug use, indicating some desire to change the
status quo), whereas the question elicited CCT (e.g., the client expressed greater exploration of
barriers to change.) Empirical evidence to support choosing one behavior over the other is
presently lacking.
Using data from the MI condition of a three-armed randomized controlled trial of a
classroom-based substance abuse prevention program, we investigated whether the percentage of
change talk (PCT) present in an MI session mediates the relationships between indicators of MI
fidelity and marijuana outcomes. Though most previous analyses have used a frequency count
of CT as the outcome or mediator variable, we decided that percentage of change talk would be
62
more indicative of someone’s intention to change due to the highly variable length of sessions in
this sample. Indicators of MI included % Complex Reflections (PCR), % Open Questions
(POQ), % Reflection of Change Talk (PRCT), Reflection to Question Ratio (RQR), counselor
empathy (Empathy), and counselor MI Spirit (Spirit). This study is the first to conduct mediation
analyses on the individual MI skills and the first to do so with structural equation modeling
(SEM). In a series of 6 SEM models, we tested our hypotheses that PCT would mediate the
relationship between PCR, POQ, RQR, PRCT, empathy, MI Spirit and marijuana use outcomes.
Methods
Procedure and Sample
The sample used in this study is derived from the 7
th
randomized trial of Project Towards
No Drug Abuse, a classroom-based substance abuse prevention program. In this three-armed
trial, 24 continuation high schools in Southern California with a minimum of 60 students per
school were recruited to participate. More detail about school selection can be found in Lisha et
al. (2012). In total, 2397 students were enrolled in the selected classes and 1,704 (71.1%)
consented to participate in the study. Of these, 573 students at 8 schools were assigned to the 3-
session MI booster condition and completed the pre-test data collection. In order to be included
in the study, students under the age of 18 were required to return a signed parental consent form
and a signed subject assent. Parental consent was not required for students over 18 years old.
The University of Southern California’s Institutional Review Board approved all study
procedures.
In the MI booster condition, students were provided up to three MI sessions; the first
occurred at school within 1 to 3 days of the classroom program, and the following two sessions
63
were conducted via telephone at 3- to 4-month intervals. Of the 1040 MI sessions, 235 discussed
substance use; 12 of these did not have recordings, leaving 223 to be coded. In order to establish
independence between observations, only one session per student was included in the final
sample (N = 170). Of these, 122 completed the post-test (see Figure 2: Flow Chart). To be
included in this sample, the content of the MI session had to pertain to substance use. Recordings
were identified for coding using notes kept by the MI interventionists. Coders independently
assessed whether the sessions met the criteria as having a substance use target In order to be
considered a substance use target, substance use had to be addressed with the exploration
exercise used during the session. For example, if a participant reported that they had cut back on
their cigarette use, and the interventionist proceeded to explore a different topic, this session
would not be considered a substance use target
The sample includes data from 17 interventionists having from 1 to 49 sessions in the
sample. All interventionists participated in standardized training and regular supervision
conducted by a member of the Motivational Interviewing Network of Trainers (MINT). Details
on the training and supervision of the interventionists and the content of the booster are
published elsewhere (Barnett, et al., 2012a).
Coding and parsing
We coded the sample using the MISC 2.5 (Houck, Moyers, Miller, Glynn & Hallgren,
2013) from the Center on Alcoholism, Substance Abuse and Addictions. The MISC 2.5 is a
hybrid of the MISC 2.1 and the Sequential Code for Observing Process Exchanges (MI-SCOPE;
Martin, Moyers, Houck, Christopher & Miller, 2005) designed to optimize the features from
each coding system to allow sequential coding of MI sessions. Specifically the MISC 2.5 allows
for the capture of specific behaviors from the MISC 2.1, as well as valenced reflections and
64
temporal order from the SCOPE. Like all versions of the MISC, it codes counselor and client
language into mutually exclusive and exhaustive categories. Coding was performed in two
passes. In the first pass, coders parsed the entire recording into utterances, or thought units, and
then completed a set of six gestalt ratings measuring counselor interpersonal skill and one
measure rating client self-exploration. In the second pass, a different coder applied behavioral
codes to each counselor utterance and each client utterance. Coding was conducted using the
Center on Alcoholism Substance Abuse and Addictions (CASAA) Application for Coding
Treatment Interactions (CACTI; Glynn, Hallgren, Houck, & Moyers, 2012). This software
automates the parsing of recordings and stores sequential coding of each utterance with no
manual data entry. Using this process for parsing ensures that all coders code the same
utterances, thereby increasing reliability. Although CACTI software does not require or utilize
transcripts, we transcribed our entire sample of recordings for ease of parsing and coding.
Parsing. In order to perform MISC coding, an interview must first be broken into distinct
thoughts; these thoughts are identified by parsing dialogue into its most basic unit, the utterance.
MISC coding requires that any two consecutive counselor statements that merit different codes
(e.g., a reflection followed by a question), be identified as separate utterances. Furthermore,
utterances of client change language are always parsed into separate utterances, even if the client
emits consecutive utterances from the same change talk category. For example, if a client
reports that he or she needs to quit smoking “for my kids, my health, and my ability to participate
in activities,” each of these reasons would be parsed into separate utterances and given separate
codes, as they represent multiple reasons to quit.
Training and Supervision of Coders. We provided five undergraduate and graduate students 40
hours of initial training in the MISC 2.5 and the CACTI software. These coders were trained to
65
parse recordings when a new idea was spoken, and/or the speaker shifted from counselor to
client. Once a recording had been parsed and globally rated, it was assigned to a different coder
who then, using CACTI, assigned a code to each utterance. Weekly coding meetings were held
throughout coding to improve and maintain reliability. All coding disagreements were resolved
by a supervisor.
Coders practiced on a series of non-substance use recordings until their inter-rater
reliability was at criterion of 0.60 using established intraclass correlation (ICC) guidelines
(Cicchetti, 1994). We randomly selected 20% of our coded sample using a random number
generator for double coding. These 47 recordings were double coded in order to calculate final
ICCs. Cicchetti’s criterion identifies ICCs below .40 as poor, .40-.59 as fair, .60-.74 as good, and
above .75 as excellent. For our data, final ICCs for counselor codes were .94 for open questions,
.80 for closed questions, .94 for reflections overall, .48 for simple reflections, .45 for complex
reflections, .84 for reflections of change talk, .82 for reflections of counter change talk, .68 for
MI-consistent behaviors and .29 for MI-inconsistent behaviors. Client codes were .92 for change
talk, .86 for counter change talk, and .88 for neutral responses. These results indicate that coders
had some difficulty differentiating simple reflections from complex reflections, and difficulty
reliably identifying MI-inconsistent behaviors, which occurred infrequently. Only seven (.03%)
recordings contained any MI-inconsistent behaviors in our dataset
Measures
Predictors. Each counselor utterance is assigned one of 17 behavior skill codes. Utterances are
coded as either open (OQ) or closed questions (CQ); simple (SR) or complex reflections (CR)
with a positive (+), negative (-), neutral (0), or both positive and negative (±) valence; as MI-
66
consistent (MICO) including specific behaviors of affirming, supporting, and asking permission
before giving advice; MI-inconsistent behaviors (MIIN) including confronting, warning, and
giving advice without permission; or “Other” for behaviors such as providing information about
the session, filler, and comments designed to facilitate conversation. For this analysis, summary
scores for POQ, PCR, RQR, and PRCT were calculated (see Table 2 for formulas).
In addition to coding counselor language, the MISC 2.5 measures the gestalt of the
session on a Likert scale of 1 to 5, whereby a 1 represents a low level of the characteristic and 5 a
high level. We used two gestalt measures: 1) empathy and 2) MI Spirit (a mean composite
measure of the sum of autonomy, collaboration and evocation), in our analyses. These ratings
characterize the entire interview and coders are instructed to consider each attribute as a
departure from the midpoint value. According to the MISC 2.5 Coding Manual, “empathy” is
defined as the counselor’s ability to convey an understanding of the client’s perspective;
“autonomy” represents the degree to which a counselor emphasizes the client’s ability to choose
their behavior; “evocation” refers to the counselor’s ability to draw out information from the
client; “collaboration” is defined as the sense that the relationship between counselor and client
is that of equal partners.
Mediator. According to the MISC 2.5, all client utterances are assigned one of 15 client language
codes. Client statements are categorized as either change talk (CT), counterchange talk (CCT), or
unrelated to change (FN). Determining change talk requires coders know the target behavior at
the outset CT includes statements of commitment (“I will cut back on smoking”), taking steps
(“I’ve already slowed down”), desire (“I want to quit”), ability (“I think I can do it”), reason (“I
have to stop for my health”), need (“I need to cut back so I can keep a job”), and other statements
that do not fall into the previous categories. CCT includes statements counter to commitment
67
(“There’s no way I will stop”), taking steps (“I had a drink last night”), desire (“I really don’t
want to”), ability (“There is no way I’d be able to give it up”), reason (“It’s not affecting my
health”), need (“I really don’t think I need to change”). For this analysis, we calculated percent
change talk (PCT) as the total frequency of change talk divided by the total number of client
utterances, including CT, CCT, and FN.
Outcome Variable. An ordinal measure of marijuana use was collected by asking respondents
how many times they used marijuana during the past 30 days. Subjects were provided with
twelve response categories ranging from 0 to 100. The log of this variable was used to account
for non-normality.
Descriptives. Measures of client age, gender, and ethnicity were also collected in the surveys.
Ethnicity was collected as White/ Caucasian, Latino/Hispanic, African American/Black, Mixed
Ethnicity, Asian, and American Indian/Native American, or "other.”
Analytical Approach. Structural Equation Modeling (SEM) was conducted to test for mediation
using Mplus (v.6) (Muthén & Muthén, 1998). SEM allows for more precise estimates of direct
and indirect effects than traditional regression approaches (Bentler and Chou, 1987). Mplus
provides estimates for the relationship between indicator and mediator (path a), the relationship
between mediator and outcome (path b), the main effect, or relationship between indicator and
outcome (path c), the direct effect, or relationship between indicator and outcome when
controlling for the mediator (path c’) and indirect effects (path a*b) using the Delta method.
Mplus also provides R
2
estimates for the mediator and the outcome variable. Maximum
likelihood estimation was used to address missing data.
We tested six models, one for each of the following MI fidelity indicators: two gestalt
measures (counselor empathy and MI Spirit) and four counselor behavioral skill measures (POQ,
68
RQR, PCR, and PRCT). All models included percent change talk as the proposed mediator,
logged marijuana use as the outcome, and controlled for baseline marijuana use. For
interpretability, all mediation results are standardized.
Results
The sample investigated in this study included 170 youth (70% male, 71% Latino, with a
mean age of 16.7 years), with reported past 30 day drug use of 68% for alcohol use, 59% for
cigarette use, and 36% for other drugs. Forty percent (40%) reported not using marijuana in the
past 30 days, while 36% reported being daily or near-daily users of marijuana.
Bivariate correlations: Table 2 shows that although MI skill variables are significantly
correlated, most correlations were below .36 with the exception of MI Spirit and Empathy (r =
0.73) and PRCT and PCT (r = 0.87).
Main Effects (path c): Only one model showed a main effect between the MI indicator
and drug use outcomes. PRCT directly influenced marijuana use (β = -0.19, p < .05); all others
had coefficients under -0.06 and were non-significant. All models controlled for baseline drug
use.
Indirect Effects (path a*b): Significant indirect effects of MI skill on marijuana use were
found for POQ (β = -0.05, p < .05), PCR (β = -0.06, p < .05), Empathy (β = -0.05, p < .05), and
MI Spirit (β = -0.06, p < .05), and a trend toward significance was found for RQR (β = -0.04, p =
.07).
Percent Variance Explained for Marijuana Outcomes: The R
2
for each of the six models
explained 21% of the variance in marijuana use. When baseline marijuana use was dropped, the
R
2
for each model was approximately 6%. Hence, each MI indicator is capable of explaining
69
approximately 6% of the variance in outcomes. We further explored the R
2
with all MI indicators
included simultaneously in the model; even after including all MI indicators the percent variance
explained did not change.
Percent Variance Explained for the Mediator: The R
2
for Percent Change Talk showed
that Percent Reflection of Change Talk by the counselor explained 75% of the variance in PCT
by the client, while all other predictors explained from 3– 6% of the variance. All models were
significant except for RQR (p < .10).
Discussion
The goal of this study was to compare the relative strength of unique indicators of MI in a
test of the hypothesized mediation model. Overall, the behavioral and gestalt predictors showed
similar associations with the mediator and outcomes, and significant indirect effects were more
common than significant main effects of counselor skill on outcomes. We propose two
explanations for this lack of main effects. First, we propose that seeing no main effect, but a
significant indirect effect, suggests that there is no reason to believe that a high percentage of any
indicator, e.g., open questions, alone, would produce change (path c); our findings suggest that it
is only when these open questions result in change talk (path a) that one would presume change
to follow (path b). For example, we would not expect open questions such as, “What are the
reasons you drink?” to result in expression of change talk. Similar interpretations can be made
for percent complex reflection and the reflection-to-question ratio. Having a high percentage of
complex reflections or more reflections than questions without subsequent change talk would
likely not result in behavior change. Second, the findings may represent examples of
70
inconsistent mediation, where the effect of the mediator suppresses the relationship of the
predictor on the outcome (MacKinnon et al., 2007).
How then should we understand the existing literature that shows a main effect of
counselor skill on outcomes? Perhaps in those studies there was an unmeasured mediator
influencing findings. McCambridge et al. (2011) did not control for any measure of client
language when they found significant relationships between PCR and MI Spirit and outcomes at
3-month follow-up. In contrast, Gaume et al. (2009) controlled for one category of change talk,
client ability language, in their analyses and found an independent effect of the counselor
behavior on outcomes at 12-month follow-up.
In this data, the only indicator to demonstrate a main effect was PRCT, the percentage of
the session during which the counselor specifically reflected change talk. PRCT differs from the
other indicators because it is a discrete behavior tied directly to the explicit counseling goal, or
behavior change target Reflections of change talk capture the valence or direction of the
counselor’s response to a client’s statement about change; it measures whether the reflection is
toward, away from, or neutral about change. To a novice listener, counselor reflections may
appear to constitute a neutral mirroring of the content the client has offered. However, it is our
view that counselors commonly add meaning, feeling or direction, to their reflections or show a
preference in choosing which aspects of the client’s speech to reflect. For example, clients often
present change talk and counter change talk together (“I want to, but . . .”) and counselors then
choose how to respond. If, as we argue in another study (Barnett et al., n.d.), these counselor
choices result in differential amounts of client change talk then these choices are closely related
to the explicit manipulation of the causal mechanism and, consequently, outcomes. Because of
the strength of the main effect of PRCT on marijuana use outcomes in this data, we argue that
71
PRCT is an important indicator of competence in MI practice and should be considered one of
the core MI skills.
Limitations
Findings from this study should be considered in light of the variance of the measures. As
an efficacy trial, variance in counselor skills was restricted due to rigorous training and
supervision of interventionists. Although we did see significant relationships between counselor
skill and change talk, we were unable to see these relationships with outcomes measured 12-
months later. An effectiveness trial using existing staff persons may demonstrate different
results. Variation in our mediator was influenced by using recordings from the first session of the
MI booster, which included a decisional balance exercise. Requiring this exercise likely
increased the amount of counter change talk present and may have reduced variance in our
mediator. Finally, variance in the outcome variable may have been influenced by social-
desirability bias, as marijuana use was self-report only, and not biochemically verified. However,
as the sample in this study represented a community sample of at-risk youth, we contend that this
population may have provided greater variance on drug use and problems associated with drug
use than often seen in clinical samples.
Furthermore, there may be alternative explanations for our findings or alternative
untested mediational pathways that deserve consideration. First, while our analyses did not
control for nesting within counselors, it is conceivable that counselor characteristics beyond MI
skill may be associated with client change talk and outcomes. Second, though we did not adjust
for client readiness to change, it also may account for variation in the presence of change talk
(Hallgren & Moyers, 2011; Moyers, Martin, Houck, Christopher, & Tonigan, 2009). Finally,
72
our mediation model may be inaccurately specified. The indicator and mediator used in these
analyses represent correlational data, as they were collected at the same point in time. While we
know that path a and b preceded outcomes, we must consider that client language may influence
counselor language as much as counselors' language influences client language.
Despite limitations, this study contributes to the current search for causal mechanisms in
MI. It expands evidence for mediation to a study of MI on adolescent marijuana use without
personalized feedback; whereas the other mediation studies have all relied on hybrids of MI and
objective information-giving, this study relies on “relational” and “technical” aspects only
(Miller & Rose, 2009). Additionally, it is the only study to look at indicators of MI fidelity
separately and provide information to clinicians about the relative merit of different clinical
choices. Future research should attempt to replicate these findings in an effectiveness trial where
greater variance in counselor skill would enhance our understanding of the relationship between
indicators of MI fidelity and drug use outcomes. In conclusion, our findings support change talk
as an active ingredient of MI and provide new empirical support for individual MI skills and
their contribution to outcomes. Findings also support a call for a new coding scheme to measure
counselor responses to change talk, as the counselor’s ability or tendency to reflect change talk
appears to be an important predictor of client success.
73
Figure 4-1. Proposed mediation model illustrating the hypothesized causal mechanisms of MI.
74
Figure 4-2: Consort diagram
Students enrolled in
24 participating schools
N = 2397
Consented
N = 1704
Classroom + Motivational Interviewing
Condition
N = 573, Total Sessions = 1040
Classroom Only
Condition
N = 562
Control
Condition
N = 569
Session focused on Substance Use
N = 235
Recordings not available due
to participant refusal or
technical problems
N = 12
Recordings
N = 223
Unique students represented in recordings
N = 170
Students with 1 Year Follow-Up Data
N = 122
75
Table 4-1: Summary of studies including mediation analyses
Paths
Sample and Design Measures and Analysis
Feed
back C a b a*b c' Notes
Morgenstern
et al, 2012
Sample size: 89
Number of session: 4
Sample: Adult problem
drinkers
Design: RCT MI vs. SOMI
vs. SC
Length of FU: 2 weeks
Coding Manual: Amrhein, 2003
Analysis: Baron & Kenney
Predictor: Condition
Mediator: Mean Commitment
Strength (last 2 deciles)
Outcome: Sum of Standard Drinks
yes Ns sig trend ns ns
RCT results showed
change talk significantly
increased in the MI
condition (directive
elements) relative to the
SOMI (Spirit only)
condition.
Moyers et al,
2009
Sample size: 118
Number of session: 2
Sample: Adult problem
drinkers
Design: Single condition
from Project MATCH
Length of FU: 5 week
follow-up
Coding Manual: SCOPE
Analysis: PRODCLIN
Predictor: MICO
Mediator: Total Change Talk
Outcome: Drinks per week
yes Sig sig sig sig sig
Path c controlling for
readiness to change; Path
a investigated MIIN
separately; MICO
coefficient changed when
CT added, but difference
not significant. Recordings
taken from session #1.
Vader et al,
2010
Sample size: 73
Number of session: 1
Sample: College drinkers
Design: Single condition
follow-up from RCT
Length of FU: 3 months
Coding Manual: MISC 2.1
Analysis: PRODCLIN
Predictor: MICO
Mediator: Total Change Talk
Outcome: Composite Alcohol Use
yes NA sig sig ns NA
Findings from RCT, MIF
significantly reduced
drinking over assessment
only. Mediation analysis
controlled for FN and
analyzed ST separately.
Alcohol use composite
includes measures of
drinks per week, peak
BAC, and protective
behaviors.
Pirlott et al,
2012
Sample size: 43
Number of session: at least
4
Sample: Fire fighters
Design: Purposive sample
of changers and non-
changers
Length of FU: 12 months
Coding Manual: MISC 2.1
Analysis: Bayesian and
PRODCLIN
Predictor: MICO and MI Spirit
Mediator: Total Change Talk
Outcome: F & V Consumption
Change Score
yes Ns sig sig sig ns
Mediation results from
both approaches showed
significance. Mediation
results on both indicators
showed mediation.
Recordings taken from
session 2 after feedback.
Notes: NA indicates that no information was reported in the paper. MICO as defined in the MISC 2.1 includes reflection, open questions, reframe,
asking permission, affirming, and offering support. MICO as defined by the SCOPE does not include any type of reflections or open questions.
76
Table 4-2: Measurement details for all variables included in the models
Predictor
POQ: Percent Open Questions OQ/(OQ+CQ)
PCR: Percent Complex Reflection CR+ CR- + CR0 + CR± /(CR+ + CR- + CR0 + CR± + SR+ + SR- + SR0 + SR±)
RQR: Reflection to Question Ratio (CR+ + CR- + CR0 + CR± + SR+ + SR- + SR0 + SR±) /(OQ+CQ)
PRCT: Percent Reflection of Change Talk (CR+ + SR+)/ (CR+ + CR- + CR0 + CR± + SR+ + SR- + SR0 + SR±)
PMICO: Percent MI Consistent
a
MICO/(MICO+MIIN)
MI Spirit Mean of (Autonomy
b
+ Collaboration
b
+ Evocation
b
)
Empathy
b
Mediator
PCT: Percent Change Talk CT/ (CT+CCT+FN)
Outcome
Past 30-day Marijuana Use
c
0, 1–10, 11–20, 21–30, 31-40, 41-50, 51–60, 61–70, 71–80, 81–90, and 91–100+
a
not included in this analysis due to low base rate of MIIN in denominator;
b
Empathy, Autonomy, Collaboration, Evocation are
measured on a 5 point likert scale;
c
Past 30-day Marijuana Use at Baseline was included as a covariate.
77
Table 4-3: Univariate and bivariate statistics for all variables included in the models
MJ SP EMP OQ CR PRCT RQ PCT
One Year Follow-Up Marijuana (MJ)
a
1.00
MI Spirit (SP)
b
-0.08 1.00
Empathy (EMP)
b
-0.06 0.73 1.00
Percent Open Questions (OQ)
c
0.04 0.23 0.17 1.00
Percent Complex Reflection (CR)
c
0.04 0.07 0.12 0.22 1.00
Percent Reflection of Change Talk (PRCT)
c
-0.22 0.23 0.23 0.14 0.15 1.00
Reflection : Question Ratio (RQ)
c
0.04 0.23 0.23 0.36 0.29 0.07 1.00
Percent Change Talk (PCT)
c
-0.25 0.23 0.23 0.19 0.22 0.87 0.18 1.00
Number of Observations 121 170 170 170 168 170 170 170
Mean 1.15 3.82 3.96 0.56 0.57 0.42 1.29 0.33
Std Dev 0.63 0.64 0.7 0.18 0.26 0.21 0.61 0.17
Notes: Bold indicates significance p < .05;
a
Marijuana measured on a 12 point ordinal scale;
b
MI Spirit and
Empathy measured on 5 point likert scale;
c
MI Fidelity Indicators measured as percents.
78
Table 4-4: Mediation Results for MI Quality Indicators Predicting Change in Marijuana Use
X --> M M --> Y
Indirect
Effect
Direct
Effect
Total/
Main
Effect R
2
N a b a*b c' c MJ
Change
Talk
Percent Open Questions 160 0.22** -0.22** -0.05* 0.01 -0.04 0.21 0.05
Percent Complex Reflection 158 0.25*** -0.24** -0.06* 0.03 -0.03 0.21 0.06
Percent Reflection of Change Talk 160 0.86*** -0.2 -0.18 -0.02 -0.19* 0.21 0.75
Reflection to Question Ratio 160 0.18* -0.22** -0.04+ 0.03 -0.02 0.21 0.03
Empathy 160 0.24** -0.22** -0.05* 0.02 -0.04 0.21 0.06
Spirit 160 0.25** -0.21** -0.06* -0.01 -0.06 0.21 0.06
+ p < .10 *p < .05, **p < .01, ***p < .001; standardized results; a = percent change talk on quality indicator; b =
percent change talk on one year marijuana outcome; c' = effect of quality indicator on one year marijuana outcome
controlling for percent change talk; c = main effect of quality indicator on one year marijuana outcome; R2 = % variance
explained of the dependent variable
79
CHAPTER 5: SUMMARY AND CONCLUSIONS
The overall objective of this dissertation was to contribute to the field's understanding
of which clients benefit most from MI interventions and to enhance understanding of the
mechanisms through which MI works. While little was learned about which adolescents
respond best to MI interventions, much was learned about the mechanisms. Consistent with
the theoretical process model of MI, robust associations between counselor behaviors and
client language were demonstrated with sequential analysis of counseling interactions.
Mediation using structural equation modeling (SEM) with MI fidelity summary scores was
supported, as well. This research supports the percentage of client change talk (PRCT) as a
robust mediator of counselor skills and drug use outcomes.
In addition to adding support to existing findings, this research makes a major
contribution in that it provides empirical evidence to help counselors decide which skills to
use by comparing the relative merits of unique MI counseling skills. Of the skills measured in
the MITI, results from sequential analysis showed CR to be more strongly associated with CT
than are SR, a finding that supports current emphasis on training and coaching practitioners to
provide more complex reflections. While results from the SEM analysis showed the
behavioral measures of POQ and PCR and gestalt measures of MI spirit and empathy are
almost equivalent in their ability to explain the presence of change talk. The SEM also
showed RQR to be the weakest of all measures. The weakness of this measure suggests that
the overall amount of reflections compared to the overall amount of questions is not strongly
associated with change talk. Rather, as the other behavioral measures tell us, it is the nuance,
or greater detail, contained in the PCR and POQ that influence change talk most. In theory, it
80
would be more instructive to have a measure that calculated the ratio of PCR/POQ rather than
Total Reflections/Total Questions.
This research also provides new evidence about the importance of the valence, or the
direction, of reflections offered by counselors. Direction in MI is not a new concept; MI is,
after all, defined as “a directive client-centered counseling style” (Miller & Rollnick, 2002).
In the definition, the term directive refers to the intent of the intervention to promote behavior
change. While MI training has always emphasized "direction" by encouraging practitioners to
elicit client change talk, it has been less instructive on directive responses, or reflections of
change talk or counter change talk. In this data PRCT was the strongest predictor of improved
outcomes. Counselors and trainers should interpret this to mean that reflecting change talk is
preferred to reflecting counter change talk and that the ratio of these reflections should be
heavily weighted toward reflection of change talk. As a strategy this means that, first, when
clients express ambivalence about change, counselors should primarily focus on reflecting the
client’s statements about change, and second, counselors should develop their ability to
reframe counter change talk as containing some element of change talk. For example, if a
client says, “I’m too stressed to consider quitting right now. (CCT)” the counselor might
reply, "You’ve been really thinking about this (RCT).”
Furthermore, this concept of “direction” is typically absent when coaching
practitioners using feedback derived from the MITI. The MITI does not include any
behavioral measure of valence, thus it provides no feedback on the counselor’s ability to elicit
or respond to change talk. A new coding instrument to provide counselors feedback about
their use of direction is necessary. Counselors need to know how they respond to change and
counter change talk, when offered alone or sandwiched together; they need to know whether
81
their tendencies are to ask about motivation to change or to inquire about barriers. At present,
MITI coaching provides information on counselors' use of complex and simple reflections, as
well as open and closed questions, but does not address whether those reflections emphasize
change talk or counter change talk, or whether their questions elicit motivation to change
(change talk) or barriers to change (counter change talk). This type of feedback may
accelerate the development of a counselor’s MI skills. Similar to a recent trial that tested
whether a modified MI training more effectively influenced trainees’ ability to elicit change
talk (Moyers et al., 2011), a comparison of coaching strategies could provide important
information about optimizing the learning of MI.
In addition to its importance in addressing skill development, such an instrument could
help streamline some of the existing MI measures. At present there is the MITI for
establishing counselor fidelity and coaching, the MISC for looking at associations between
counselor and client language, and the SCOPE for investigating sequential relationships
between counselor and client language. An instrument that measures the most robust, known
counselor indicator for predicting outcomes could eliminate the need to code client change
talk without losing some of the value. In this data, PRCT and PCT were highly correlated at
.87, suggesting that PRCT is capable of serving as a proxy for associations of client language
and outcomes, as the measure simultaneously captures an element of client and counselor
speech. Furthermore, it explained 75% of the variance in the percentage of change talk, a vast
improvement when compared to the other indicators, which did not exceed 6% of explained
variance. PRCT has an additional advantage as it was not difficult to establish “excellent”
inter-rater reliability when coding, especially compared to the challenge of training coders to
reliably differentiate between complex and simple reflections.
82
Findings from these studies should be considered in light of some distinguishing
features of the parent study and the MI intervention. Generalizability of these findings may be
limited by the inclusion of only high-risk adolescents at continuation high schools in southern
California. Southern California schools have a unique demographic composition, typically
more than 60% Latino, and although these students are deemed high-risk because they attend
non-traditional high schools, they do not necessarily have substance use problems. This MI
intervention was not an indicated substance use intervention, as most are; rather, it served as a
motivational booster to a universal prevention program and was delivered to all students
regardless of the extent of their drug use. The skew of the data showed that the vast majority
of students reported using either 0 or 1 - 10 times in the past 30 days. Hence, floor effects,
limits on the possible reduction in drug use, were likely.
A more refined measure, one that could differentiate between students using
occasionally on a weekend (once or twice per month) from those using every weekend (four
to eight times per month) would have improved our ability to detect effects, particularly in
Study 1 where no effects were found. Under the current measurement system, both of these
drug-using constellations would have indicated using 1-10 times in the last 30 days despite
representing very different patterns of drug use. A refined measure would have increased the
variance among users at the low end of the scale. The Global Assessment Programme for
Drug Use (2003) sponsored by the United Nations Office of Drugs and Crime uses a 7-point
scale ranging from “0” to “40 or more” with categories of 0, 1-2, 3-5, 6-9, 10-19, 20-39, 40 or
more to address this. Furthermore, instrument design theory proposes that scale range
suggests to respondents that the mid-point may represent average use, making the scale a
frame of reference for estimating one’s own response (Schwarz, 1999). With the current
83
measurement system, the mid-point would be more than 40 times per month or more than
daily drug use.
Findings from these studies may have also been limited by length of time between
students' MI sessions and one-year follow-up. Approximately one-third of the sample
received only one MI session, occurring immediately after the classroom program. For these
students, a 12-month follow-up period may have been too long to see effects from one session
of MI. Furthermore, we have no way to assess the effects of the intervening attempts to
contact them. Our attempts may have signaled concern or a reminder about the program
message thereby decreasing substance use, or conversely, our attempts may have signaled
“nagging” and provoked a reactant response increasing their desire to partake in substance-
using behaviors.
Starting from Karno & Longabaugh’s (2007) premise that “matching is not dead,”
future research must continue to unravel the role of client and counselor characteristics on the
use of MI. There are three sets of relationships that need to be studied: first, that between
client characteristics and program efficacy, as was attempted in Study 1; second, the
relationship between counselor characteristics and aptitude for MI; and third, the matching of
counselor and client characteristics. Variables of interest include demographics such as race,
ethnicity, culture, and gender, as well as intrapersonal characteristics such as psychological
reactance (Beutler et al., 1991; Karno & Longabaugh, 2005), the “Big 5” personality traits
(Costa & McCrae, 1992) and Positive Psychology’s character strengths and values (Park &
Peterson, 2006; Peterson & Seligman, 2006).
From the perspective of mediation, the search for psychosocial mediators of change
must continue as well. Theories closely associated with MI provide some guidance as to
84
possible mediators of change. Such variables include intrinsic motivation, competence,
autonomy, and relatedness from Self-Determination Theory (Deci & Ryan, 1985; Markland,
Ryan, Tobin, & Rollnick, 2005; Vansteenkiste & Sheldon, 2006), self-efficacy from Social
Learning Theory (Bandura, 1977), readiness to change or stages of change from the
Transtheoretical Model (Prochaska, DiClemente, & Norcross, 1992; Prochaska &
DiClemente, 1982); and cognitive dissonance from the Theory of Cognitive Dissonance
(Festinger, 1957). Other common health promotion theories may provide additional
important mediators, such as behavioral intentions from the Theory of Planned Behavior
(Ajzen & Madden, 1986), and perceived descriptive norms from Social Norms Theory
(Perkins, 2003). In order for systematic investigation to occur a series of testable mediation
models are needed (Barnett et al., 2012b).
Another branch of MI research should focus on experimental manipulations of
counselor behaviors. Of particular interest, based on these findings, would be a randomized
controlled trial that investigates outcomes related to PRCT. For instance, in one condition
counselors would be instructed to respond to any discussion of barriers to change (CCT) with
exploration and troubleshooting, increasing counselors RCCT and thereby lower their PRCT.
The other condition would be instructed to acknowledge any expression of CCT or barriers,
but to continue to focus on developing motivation by eliciting and reflecting change talk. We
would expect the second condition to have higher PRCT and improved outcomes than the
comparison group.
Finally, another important frontier for MI research is the effectiveness trial.
Effectiveness trials, or “real world” applications of MI, where training and supervision are
provided in a business-as-usual setting, instead of under the scrutiny of rigorously controlled
85
trials, are needed to support the financial investments being made by state and local agencies.
One line of inquiry into “real world MI” would be to develop a greater understanding of
“natural” MI skills. “Naturals” would refer to those individuals who embody the spirit of MI
without effort, refrain from telling people what to do, use reflective listening as a primary
listening skill, and naturally attend to positive change talk. By simply coding existing staff for
these skills, we could determine whether these people tend to have better client outcomes than
their peers.
In conclusion, much is known about MI and much is left to discover. There is more to
learn about matching clients and counselors to treatment approaches that best suit them, more
to learn about antecedents of behavior change, and more to learn about this cadre of MI
naturals. With the diffusion of MI throughout a wide range of disciplines (e.g., medicine,
dentistry, public health, social work, nursing, education, and, of course, substance use and
mental health treatment), targeting every conceivable health behavior and issue (e.g.,
substance use, asthma, eating disorders, exercise, nutrition, cancer screenings, medication
adherence, dental checkups, classroom management, water conservation), the need to improve
our ability to measure, train, and coach MI is heightened.
86
REFERENCES
Amrhein, P. C., Miller, W. R., Yahne, C. E., Palmer, M., & Fulcher, L. (2003). Client
commitment language during motivational interviewing predicts drug use outcomes.
Journal of Consulting and Clinical Psychology, 71(5), 862.
Apodaca, T. R., & Longabaugh, R. (2009). Mechanisms of change in motivational
interviewing: a review and preliminary evaluation of the evidence. Addiction, 104(5),
705.
Ajzen, I., & Madden, T. J. (1986). Prediction of goal-directed behavior: Attitudes, intentions,
and perceived behavioral control. Journal of experimental social psychology, 22(5),
453–474.
Baer, J. S., Beadnell, B., Garrett, S. B., Hartzler, B., & Wells, E. A. (2008). Adolescent
change language within a brief motivational intervention and substance use outcomes.
Psychology of Addictive Behaviors, 22(4), 570–575. doi:10.1037/a0013022
Bakeman, R., & Quera, V. (1995). Analyzing interaction: Sequential analysis with SDIS &
GSEQ. Cambridge University Press.
Bakeman, R., & Quera, V. (2002). GSEQ 5.1: Software for the analysis of interaction
sequences. Retrieved from http://www2.gsu.edu/~psyrab/gseq/Download.html
Bandura, A. (1977). Self-efficacy: toward a unifying theory of behavioral change.
Psychological review, 84(2), 191.
Barnett, E., Spruijt-Metz, D., Unger, J. B., Sun, P., Rohrbach, L. A., & Sussman, S. (2012a).
Boosting a Teen Substance Use Prevention Program with Motivational Interviewing.
Substance Use & Misuse, 47(4), 418–428. doi:10.3109/10826084.2011.641057
87
Barnett, E., Spruijt-Metz, D., Moyers, T. B., Smith, C., Rohrbach, L. A., Sun, P., & Sussman,
S. (n.d.). Reflective Listening and Change Talk: A Closer Look at Motivational
Interviewing, Under review.
Barnett, E., Sussman, S., Smith, C., Rohrbach, L. A., & Spruijt-Metz, D. (2012b).
Motivational Interviewing for adolescent substance use: A review of the literature.
Addictive Behaviors, 37(12), 1325–1334. doi:10.1016/j.addbeh.2012.07.001
Beutler, L. E., Mohr, D. C., Grawe, K., & Engle, D. (1991). Looking for differential treatment
effects: Cross-cultural predictors of differential psychotherapy efficacy. Journal of
Psychotherapy Integration. 1, 121-142
Bien, T. H., Miller, W. R., & Tonigan, J. S. (1993). Brief interventions for alcohol problems:
a review. Addiction, 88(3), 315–336.
Blos, P. (1966). On adolescence. New York: Free Press.
Brannigan, R., Schackman, B. R., Falco, M., & Millman, R. B. (2004). The quality of highly
regarded adolescent substance abuse treatment programs: Results of an in-depth
national survey. Archives of pediatrics & adolescent medicine, 158(9), 904.
Brehm, J. W. (1966). A theory of psychological reactance. New York.
Brown, R., Ramsey, S., Strong, D., Myers, M., Kahler, C., Lejuez, C. (2003). Effects of
motivational interviewing on smoking cessation in adolescents with psychiatric
disorders. Tobacco Control, 12(suppl 4), iv3–iv10.
Bugen, L. A., & Hawkins, R. C. (1981). The coping assessment battery: Theoretical and
empirical foundations. Paper presented at the Meeting of the American Psychological
Association, Los Angeles, California.
Capone, C., & Wood, M. D. (2009). Thinking about drinking: Need for cognition and
readiness to change moderate the effects of brief alcohol interventions. Psychology of
Addictive Behaviors, 23(4), 684.
88
Carey, K. B., Henson, J. M., Carey, M. P., & Maisto, S. A. (2007). Which heavy drinking
college students benefit from a brief motivational intervention? Journal of consulting
and clinical psychology, 75(4), 663.
Carroll, K. M., Connors, G. J., Cooney, N. L., DiClemente, C. C., Donovan, D. M., Kadden,
R. R., Zweben, A. (1998). Internal validity of project MATCH treatments:
Discriminability and integrity. Journal of consulting and clinical psychology, 66(2),
290.
Catley, D., Harris, K. J., Mayo, M. S., Hall, S., Okuyemi, K. S., Boardman, T., & Ahluwalia,
J. S. (2006). Adherence to principles of motivational interviewing and client within-
session behavior. Behavioural and Cognitive Psychotherapy, 34(1), 43.
Centers for Disease Control and Prevention. (2010). Teen Driver Fact Sheet. Atlanta, GA:
U.S. Department of Health and Human Services.
Chamberlain, P., Patterson, G., Reid, J., Kavanagh, K., & Forgatch, M. (1984). Observation
of client resistance. Behavior Therapy, 15(2), 144–155.
Cicchetti, D. V. (1994). Guidelines, criteria, and rules of thumb for evaluating normed and
standardized assessment instruments in psychology. Psychological Assessment, 6(4),
284.
Costa, P. T., & McCrae, R. R. (1992). Four ways five factors are basic. Personality and
individual differences, 13(6), 653–665.
Cuijpers, P. (2002). Effective ingredients of school-based drug prevention programs: a
systematic review. Addictive behaviors, 27(6), 1009–1023.
D’Amico, E. J., Miles, J. N. V., Stern, S. A., & Meredith, L. S. (2008). Brief motivational
interviewing for teens at risk of substance use consequences: A randomized pilot study
in a primary care clinic. Journal of substance abuse treatment, 35(1), 53–61.
89
Deci, E. L., & Ryan, R. M. (1985). Intrinsic motivation and self-determination in human
behavior. New York: Plenum.
Emmons, K. M., Hammond, S. K., Fava, J. L., Velicer, W. F., Evans, J. L., & Monroe, A. D.
(2001). A randomized trial to reduce passive smoke exposure in low-income
households with young children. Pediatrics, 108(1), 18.
Erickson, S. J., Gerstle, M., & Feldstein, S. W. (2005). Brief interventions and motivational
interviewing with children, adolescents, and their parents in pediatric health care
settings: a review. Archives of Pediatrics and Adolescent Medicine, 159(12), 1173.
Festinger, L. (1957). A theory of cognitive dissonance. Stanford University Press.
Friedman, I. A., & Mann, L. (1993). Coping patterns in adolescent decision making: An
Israeli-Australian comparison. Journal of adolescence, 16(2), 187–199.
Gaume, J., Bertholet, N., Faouzi, M., Gmel, G., & Daeppen, J.-B. (2010). Counselor
motivational interviewing skills and young adult change talk articulation during brief
motivational interventions. Journal of Substance Abuse Treatment, 39(3), 272–281.
doi:10.1016/j.jsat.2010.06.010
Gaume, J., Gmel, G., & Daeppen, J. B. (2008). Brief alcohol interventions: Do counsellors’
and patients’ communication characteristics predict change? Alcohol and alcoholism,
43(1), 62–69.
Gaume, J., Gmel, G., Faouzi, M., & Daeppen, J. B. (2009). Counselor skill influences
outcomes of brief motivational interventions. Journal of substance abuse treatment,
37(2), 151–159.
Gaume, J., Gmel, G., Faouzi, M., & Daeppen, J.-B. (2008). Counsellor behaviours and patient
language during brief motivational interventions: a sequential analysis of speech.
Addiction, 103(11), 1793–1800. doi:10.1111/j.1360-0443.2008.02337.x
90
Glynn, L. H., Hallgren, K. A., Houck, J. M., & Moyers, T. B. (2012). CACTI: Free, Open-
Source Software for the Sequential Coding of Behavioral Interactions. PLoS ONE,
7(7), e39740. doi:10.1371/journal.pone.0039740
Glynn, L. H., & Moyers, T. B. (2010). Chasing change talk: The clinician’s role in evoking
client language about change. Journal of Substance Abuse Treatment, 39(1), 65–70.
doi:10.1016/j.jsat.2010.03.012
Grenard, J. L., Ames, S. L., Pentz, M. A., & Sussman, S. (2006). Motivational interviewing
with adolescents and young adults for drug-related problems. International journal of
adolescent medicine and health, 18(1), 53–68.
Hallgren, K. A., & Moyers, T. B. (2011). Does readiness to change predict in-session
motivational language? Correspondence between two conceptualizations of client
motivation. Addiction, 106(7), 1261–1269.
Hettema, J., Steele, J., & Miller, W. R. (2005). Motivational interviewing. Annual Review of
Clinical Psychology, 1, 91–111.
Hodgins, D. C., Ching, L. E., & McEwen, J. (2009). Strength of commitment language in
motivational interviewing and gambling outcomes. Psychology of Addictive
Behaviors, 23(1), 122.
Hodgins, D. C., Currie, S. R., & el-Guebaly, N. (2001). Motivational enhancement and self-
help treatments for problem gambling. Journal of consulting and clinical psychology,
69(1), 50.
Hong, S. M., Giannakopoulos, E., Laing, D., & Williams, N. A. (1994). Psychological
reactance: Effects of age and gender. The Journal of social psychology, 134(2), 223-
228.
Houck, Moyers, Miller, Glynn & Hallgren, (2013). Motivational Interviewing Skill Code 2.5.
Retrieved from the Center on Alcoholism, Substance Abuse and Addictions
http://casaa.unm.edu/ download/misc25.pdf)
91
Hoyle, R. H., & Kenny, D. A. (1999). Sample size, reliability, and tests of statistical
mediation. Statistical strategies for small sample research, 195–222.
Ingram, R. E., & Snyder, C. R. (2006). Blending the good with the bad: Integrating positive
psychology and cognitive psychotherapy. Journal of Cognitive Psychotherapy, 20(2),
117–122.
Janis, I. L., & Mann, L. (1977). Decision making: A psychological analysis of conflict,
choice, and commitment. New York: Free Press.
Jensen, C. D., Cushing, C. C., Aylward, B. S., Craig, J. T., Sorell, D. M., & Steele, R. G.
(2011). Effectiveness of motivational interviewing interventions for adolescent
substance use behavior change: A meta-analytic review. Journal of consulting and
clinical psychology, 79(4), 433–440.
Johnston, L. D., Bachman, J. G., & Schulenberg, J. E. (2012). Monitoring the Future National
Results on Adolescent Drug Use: Overview of Key Findings, 2011. Institute for Social
Research, 85.
Karno, M. P., & Longabaugh, R. (2004). What do we know? Process analysis and the search
for a better understanding of Project MATCH’s anger-by-treatment matching effect.
Journal of studies on alcohol, 65(4), 501–512.
Karno, M. P., & Longabaugh, R. (2005). Less directiveness by therapists improves drinking
outcomes of reactant clients in alcoholism treatment. Journal of Consulting and
Clinical Psychology, 73(2), 262.
Karno, M. P., & Longabaugh, R. (2007). Does matching matter? Examining matches and
mismatches between patient attributes and therapy techniques in alcoholism treatment.
Addiction, 102(4), 587–596.
Kavanagh, K., Gabrielson, P., & Chamberlain, P. (1982). Manual for coding client resistance.
Unpublished instrument and technical report, Oregon Social Learning Center.
92
Kealey, K. A., Ludman, E. J., Marek, P. M., Mann, S. L., Bricker, J. B., & Peterson, A. V.
(2009). Design and implementation of an effective telephone counseling intervention
for adolescent smoking cessation. Journal of the National Cancer Institute, 101(20),
1393.
Kenny, D. A., Kashy, D. A., & Bolger, N. (1998). Data analysis in social psychology. The
handbook of social psychology, 1(4), 233–265.
LaBrie, J. W., Pedersen, E. R., Earleywine, M., & Olsen, H. (2006). Reducing heavy drinking
in college males with the decisional balance: Analyzing an element of Motivational
Interviewing. Addictive behaviors, 31(2), 254–263.
Lisha, N. E., Sun, P., Rohrbach, L. A., Spruijt-Metz, D., Unger, J. B., & Sussman, S. (2012).
An evaluation of immediate outcomes and fidelity of a drug abuse prevention program
in continuation high schools: Project Towards No Drug Abuse (TND). Journal of
Drug Education, 42(1), 33–57.
Lundahl, B. W., Kunz, C., Brownell, C., Tollefson, D., & Burke, B. L. (2010). A meta-
analysis of motivational interviewing: Twenty-five years of empirical studies.
Research on Social Work Practice, 20(2), 137–160.
Macgowan, M. J., & Engle, B. (2010). Evidence for optimism: behavior therapies and
motivational interviewing in adolescent substance abuse treatment. Child and
adolescent psychiatric clinics of North America, 19(3), 527.
MacKinnon, D. P., Fairchild, A. J., & Fritz, M. S. (2007). Mediation Analysis. Annual Review
of Psychology, 58(1), 593–614. doi:10.1146/annurev.psych.58.110405.085542
Markland, D., Ryan, R. M., Tobin, V. J., & Rollnick, S. (2005). Motivational interviewing
and self–determination theory. Journal of Social and Clinical Psychology, 24(6), 811–
831.
93
Martin, T., Christopher, P. J., Houck, J. M. , & Moyers, T. B. (2011). The structure of client
language and drinking outcomes in project match. Psychology of Addictive Behaviors,
25(3), 439–445. doi: http://dx.doi.org.libproxy.usc.edu/10.1037/a0023129
Martin, T., Moyers, T. B., Houck, J., Christopher, P., & Miller, W. R. (2005). Motivational
Interviewing Sequential Code for Observing Process Exchanges (MI-SCOPE) coder’s
manual. Retrieved from http://casaa.unm.edu/download/scope.pdf
McCambridge, J., Day, M., Thomas, B. A., & Strang, J. (2011). Fidelity to Motivational
Interviewing and subsequent cannabis cessation among adolescents. Addictive
Behaviors, 36(7), 749–754. doi:10.1016/j.addbeh.2011.03.002
McCambridge, J., & Strang, J. (2004). The efficacy of single-session motivational
interviewing in reducing drug consumption and perceptions of drug-related risk and
harm among young people: results from a multi-site cluster randomized trial.
Addiction, 99(1), 39–52.
McCambridge, J., & Strang, J. (2005). Deterioration over time in effect of Motivational
Interviewing in reducing drug consumption and related risk among young people.
Addiction, 100(4), 470–478.
McNally, A. M., Palfai, T. P., & Kahler, C. W. (2005). Motivational interventions for heavy
drinking college students: examining the role of discrepancy-related psychological
processes. Psychology of Addictive Behaviors, 19(1), 79.
Midford, R. (2010). Drug prevention programmes for young people: where have we been and
where should we be going? Addiction, 105(10), 1688–1695.
Miller, W. R., & Arciniega, L. T. (2004). Combined behavioral intervention manual: A
clinical research guide for therapists treating people with alcohol abuse and
dependence. US Department of Health and Human Services, National Institutes of
Health, National Institute on Alcohol Abuse and Alcoholism.
94
Miller, W. R., Benefield, R. G., & Tonigan, J. S. (1993). Enhancing motivation for change in
problem drinking: A controlled comparison of two therapist styles. Journal of
consulting and clinical psychology, 61(3), 455.
Miller, W. R., C’de Baca, J., Matthews, D. B., & Wilbourne, P. L. (2001). Personal values
card sort. University of New Mexico, Albuquerque, NM. Retrieved from
http://www.guilford.com/etc/miller11/pers_val.pdf
Miller, W. R., Hedrick, K. E., & Orlofsky, D. R. (1991). The Helpful Responses
Questionnaire: A procedure for measuring therapeutic empathy. Journal of Clinical
Psychology; Journal of Clinical Psychology. 47(3), 444
Miller, W. R., & Moyers, T. B. (2006). Eight stages in learning motivational interviewing.
Journal of Teaching in the Addictions, 5(1), 3–17.
Miller, W. R., Moyers, T. B., Ernst, D., & Amrhein, P. (2003). Manual for the motivational
interviewing skill code (MISC). Version 1.0. Retrieved from
http://casaa.unm.edu/download/misc.pdf
Miller, W. R., & Rollnick, S. (2002). Motivational interviewing: Preparing people for
change. The Guilford Press.
Miller, W. R., & Rollnick, S. (2009). Ten things that motivational interviewing is not.
Behavioural and Cognitive Psychotherapy, 37(2), 129.
Miller, W. R., & Rose, G. S. (2009). Toward a theory of motivational interviewing. American
Psychologist, 64(6), 527.
Miller, W. R., Sovereign, R. G. (1989). T he Check-up: A model for early intervention in
addictive behaviors. In Addictive Behaviors: Prevention and Early Intervention. (pp.
219–231). Lisse, Netherlands: Swets & Zeitlinger Publishers.
95
Miller, W. R., & White, W. (2007). Confrontation in Addiction Treatment. Counselor: The
magazine for addictions professionals (October 4, 2007). Retrieved from
http://www.counselormagazine.com/content/view/608/63.
Monti, P. M., Colby, S. M., Barnett, N. P., Spirito, A., Rohsenow, D. J., Myers, M.,
Lewander, W. (1999). Brief intervention for harm reduction with alcohol-positive
older adolescents in a hospital emergency department. Journal of consulting and
clinical psychology, 67(6), 989.
Morgenstern, J., Kuerbis, A., Amrhein, P., Hail, L., Lynch, K., & McKay, J. R. (2012).
Motivational Interviewing: A Pilot Test of Active Ingredients and Mechanisms of
Change. Psychology of Addictive Behaviors. 26(4), 859-869. doi:10.1037/a0029674
Moyers, T. B., Houck,, J. M., Glynn, L. H., & Manuel, J. K. (2011). Can specialized training
teach clinicians to recognize, reinforce, and elicit client language in Motivational
Interviewing? Presented at the 34th annual meeting of the Research Society on
Alcoholism Nation Conference, Atlanta, GA.
Moyers, T. B., Manuel, J. K., Wilson, P. G., Hendrickson, S. M. L., Talcott, W., & Durand, P.
(2008). A randomized trial investigating training in motivational interviewing for
behavioral health providers. Behavioural and Cognitive Psychotherapy, 36(02), 149–
162.
Moyers, T. B., & Martin, T. (2006). Therapist influence on client language during
motivational interviewing sessions. Journal of Substance Abuse Treatment, 30(3),
245–251.
Moyers, T. B., Martin, T., Christopher, P. J., Houck, J. M., Tonigan, J. S., & Amrhein, P. C.
(2007). Client language as a mediator of motivational interviewing efficacy: Where is
the evidence? Alcoholism: Clinical and Experimental Research, 31, 40s–47s.
Moyers, T. B., Martin, T., Houck, J. M., Christopher, P. J., & Tonigan, J. S. (2009). From in-
session behaviors to drinking outcomes: A causal chain for motivational interviewing.
96
Journal of Consulting and Clinical Psychology, 77(6), 1113–1124. doi:
http://dx.doi.org.libproxy.usc.edu/10.1037/a0017189
Moyers, T.B., Martin, T., Manuel, J. K., Hendrickson, S. M. L., & Miller, W. R. (2005).
Assessing competence in the use of motivational interviewing. Journal of Substance
Abuse Treatment, 28(1), 19–26.
Moyers, T.B., Martin, T., Manuel, J., Miller, W., & Ernst, D. (2007). Revised global scales:
Motivational interviewing treatment integrity 3.0 (MITI 3.0). University of New
Mexico, Center on Alcoholism, Substance Abuse and Addictions (CASAA), 28.
Olmstead, T., Carroll, K. M., Canning-Ball, M., & Martino, S. (2011). Cost and cost-
effectiveness of three strategies for training clinicians in motivational interviewing.
Drug and Alcohol Dependence, 116(1–3), 195–202.
doi:10.1016/j.drugalcdep.2010.12.015
Park, N., & Peterson, C. (2006). Moral competence and character strengths among
adolescents: The development and validation of the Values in Action Inventory of
Strengths for Youth. Journal of Adolescence, 29(6), 891–909.
Perkins, H. W. (2003). The social norms approach to preventing school and college age
substance abuse: A handbook for educators, counselors, and clinicians. San
Francisco, CA: Jossey-Bass.
Peterson, C., & Seligman, M. E. P. (2006). The values in action (VIA) classification of
strengths. A life worth living: Contributions to positive psychology, 29–48.
Pirlott, A., Kisbu-Sakarya, Y., DeFrancesco, C., Elliot, D., & MacKinnon, D. P. (2012).
Mechanisms of motivational interviewing in health promotion: a Bayesian mediation
analysis. International Journal of Behavioral Nutrition and Physical Activity, 9:69.
doi:10.1186/1479-5868-9-69
97
Prochaska, J. O., & DiClemente, C. C. (1982). Transtheoretical therapy: Toward a more
integrative model of change. Psychotherapy: Theory, Research & Practice, 19(3),
276.
Prochaska, J. O., DiClemente, C. C., & Norcross, J. C. (1992). In search of how people
change: Applications to addictive behaviors. American Psychologist, 47(9), 1102.
Resnicow, K., Davis, R., & Rollnick, S. (2006). Motivational interviewing for pediatric
obesity: conceptual issues and evidence review. Journal of the American Dietetic
Association, 106(12), 2024–2033.
Rogers, C. R. (1959). A theory of therapy, personality, and interpersonal relationships, as
developed in the client-centered framework. Psychology: A study of a science, 3, 184–
256.
Rollnick, S., Mason, P., & Butler, C. (1999). Health behavior change: a guide for
practitioners. Edinburgh and New York.
Rosengren, D. B., Baer, J. S., Hartzler, B., Dunn, C. W., & Wells, E. A. (2005). The video
assessment of simulated encounters (VASE): Development and validation of a group-
administered method for evaluating clinician skills in motivational interviewing. Drug
and alcohol dependence, 79(3), 321–330.
Rooney, B. L., & Murray, D. M. (1996). A meta-analysis of smoking prevention programs
after adjustment for errors in the unit of analysis. Health Education & Behavior, 23(1),
48–64.
SAS Institute. (2008). SAS release 9.2. Cary, NC.
Schwarz, N. (1999). Self-reports: How the questions shape the answers. American
Psychologist, 54(2), 93.
Shrout, P. E., & Bolger, N. (2002). Mediation in experimental and nonexperimental studies:
new procedures and recommendations. Psychological Methods, 7(4), 422.
98
Skara, S., & Sussman, S. (2003). A review of 25 long-term adolescent tobacco and other drug
use prevention program evaluations. Preventive medicine, 37(5), 451–474.
Spruijt-Metz, D., Barnett, E., Resnicow, K., (2011). Obesity in Minorities. In S. Naar-King &
M. Suarez (Eds.), Motivational Interviewing with Adolescents and Young Adults (p. p.
135). New York: The Guilford Press.
Strang, J., & McCambridge, J. (2004). Can the practitioner correctly predict outcome in
motivational interviewing? Journal of substance abuse treatment, 27(1), 83–88.
Sussman, S., & Ames, S. (2001). The social psychology of drug abuse. Cambridge University
Press.
Sussman, S., Sun, P., Rohrbach, L. A., & Spruijt-Metz, D. (2012). One-year outcomes of a
drug abuse prevention program for older teens and emerging adults: Evaluating a
motivational interviewing booster component. Health Psychology, 31(4), 476.
Tevyaw, T. O. L., & Monti, P. M. (2004). Motivational enhancement and other brief
interventions for adolescent substance abuse: foundations, applications and
evaluations. Addiction, 99, 63–75.
Tuinstra, J., Van Sonderen, F. L. P., Groothoff, J. W., Van den Heuvel, W. J. A., & Post, D.
(2000). Reliability, validity and structure of the Adolescent Decision Making
Questionnaire among adolescents in The Netherlands. Personality and individual
differences, 28(2), 273–285.
Vansteenkiste, M., & Sheldon, K. M. (2006). There’s nothing more practical than a good
theory: Integrating motivational interviewing and self-determination theory. British
Journal of Clinical Psychology, 45(1), 63–82.
Vader, A. M., Walters, S. T., Prabhu, G. C., Houck, J. M., & Field, C. A. (2010). The
language of motivational interviewing and feedback: Counselor language, client
language, and client drinking outcomes. Psychology of Addictive Behaviors, 24(2),
190–197. doi:10.1037/a0018749
99
Westra, H. A., Arkowitz, H., & Dozois, D. J. A. (2009). Adding a motivational interviewing
pretreatment to cognitive behavioral therapy for generalized anxiety disorder: A
preliminary randomized controlled trial. Journal of anxiety disorders, 23(8), 1106–
1117.
White, D., & Pitts, M. (2002). Educating young people about drugs: a systematic review.
Addiction, 93(10), 1475–1487.
Wills, T. A. (1986). Stress and coping in early adolescence: Relationships to substance use in
urban school samples. Health psychology, 5(6), 503.
Zhao, X., Lynch, J. G., & Chen, Q. (2010). Reconsidering Baron and Kenny: Myths and
truths about mediation analysis. Journal of Consumer Research, 37(2), 197–206.
Abstract (if available)
Abstract
This dissertation adds to the existing literature about how Motivational Interviewing (MI) works and who responds to best to Motivational Interviewing interventions. It also describes the development of an innovative MI booster intervention for a classroom-based substance use prevention program that was able to accommodate participants regardless of level of substance use or their substance of choice. This dissertation also describes the supervision and training of interventionists and coders. ❧ The most recent trial of Project Towards No Drug (TND) Abuse, a classroom-based substance use prevention program, was designed to determine whether the addition of an MI booster would enhance program effects for students attending continuation high schools in Southern California. In Study 1 (Chapter 2), data from two conditions (TND only and TND+MI) were used to investigate whether decision making attributes moderated the effects of program condition on substance use outcomes. Results from this analysis found no effect of decision making on program efficacy, leading to the conclusion that decision making does not warrant further investigation as a moderator of MI efficacy. ❧ Investigations regarding how MI works were conducted using data derived from sequentially coding recorded treatment interactions from the MI condition. All sessions pertaining to substance use were coded using a MI specific coding manual and MI specific automated technology. Study 2 (Chapter 3) describes an investigation into the relationship between a specific MI counselor behavior (reflective listening) and client language about change (change talk). Counselor reflections were coded for complexity (simple or complex) and valence (positive/toward or negative/away from change). Client language was coded as toward change (change talk), away from change (counter change talk) and neutral about change (FN). A comparison of effect sizes shows that the valence of a reflection is more strongly associated with subsequent change talk than is complexity. Study 3 (Chapter 4) used summary scores derived from the sequentially coded data to characterize counselor’s fidelity to MI. We investigated whether the percentage of client change talk mediated the relationship between six unique counselor skills and marijuana use outcomes. Study findings showed that the relationship between percentage open questions, percentage complex reflections, counselor empathy and counselor MI spirit and outcomes were mediated by change talk. Percentage reflection of change talk demonstrated a main effect on outcomes. Findings from Paper 2 and Paper 3 support an existing call for a new measurement instrument that focuses on how counselors respond to change talk. Such an instrument will be invaluable in the development of MI skills and enhance our ability to improve program efficacy.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
The influence of contextual factors on the processes of adoption and implementation of evidence-based substance use prevention and tobacco cessation programs in schools
PDF
Sociocultural stress, coping and substance use among Hispanic/Latino adolescents
PDF
Social self-control and adolescent substance use
PDF
Motivation and the meanings of health behavior as factors associated with eating behavior in Latino youth
PDF
Mixed methods investigation of user engagement with a smoking cessation app
PDF
A "second chance" program for vulnerable young adults: a program evaluation and a randomized controlled trial of adjunctive motivational interviewing to improve retention
PDF
Energy drink consumption, substance use and attention-deficit/hyperactivity disorder among adolescents
PDF
A penta-dimensional longitudinal analysis of the predictors of compulsive internet use among adolescents using linear mixed model (LMM)
PDF
Role transitions, past life events, and their associations with multiple categories of substance use among emerging adults
PDF
The Internet activities, gratifications, and health consequences related to compulsive Internet use
PDF
Effects of sugar and fiber consumption in minority adolescents and self-tracking as a potential dietary intervention tool
PDF
A sociocultural and developmental approach to intimate partner violence among a sample of Hispanic emerging adults
PDF
Influences of specific environmental domains on childhood obesity and related behaviors
PDF
The effects of mindfulness on adolescent cigarette smoking: Measurement, mechanisms, and theory
PDF
Problematic alcohol use in Hispanic emerging adults: the role of perceived discrimination, cultural identity, and salient cultural values
PDF
Prospective associations of stress, compulsive internet use, and posttraumatic growth among emerging adults
PDF
A network analysis of online and offline social influence processes in relation to adolescent smoking and alcohol use
PDF
Tobacco and marijuana surveillance using Twitter data
PDF
The impacts of the COVID-19 pandemic on therapy utilization among racially/ethnically and socio-economically diverse children with autism spectrum disorder
PDF
Childhood cancer survivorship: parental factors associated with survivor's follow-up care behavior and mental health
Asset Metadata
Creator
Barnett, Elizabeth M.
(author)
Core Title
Motivational interviewing with adolescent substance users: a closer look
School
Keck School of Medicine
Degree
Doctor of Philosophy
Degree Program
Preventive Medicine (Health Behavior)
Publication Date
05/13/2013
Defense Date
05/13/2013
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
Decision making,mediation,motivational interviewing,OAI-PMH Harvest,reflective listening,substance use
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Spruijt-Metz, Donna (
committee chair
), Sussman, Steven (
committee chair
), Huey, Stanley J., Jr. (
committee member
), Moyers, Theresa B. (
committee member
), Rohrbach, Louise Ann (
committee member
), Sun, Ping (
committee member
)
Creator Email
embarnet@usc.edu,liz.barnett@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c3-255032
Unique identifier
UC11293370
Identifier
etd-BarnettEli-1685.pdf (filename),usctheses-c3-255032 (legacy record id)
Legacy Identifier
etd-BarnettEli-1685.pdf
Dmrecord
255032
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Barnett, Elizabeth M.
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
Tags
motivational interviewing
reflective listening
substance use