University of Southern California Dissertations and Theses

The role of social identity in adolescent smoking behavior

(USC Thesis Other)

The role of social identity in adolescent smoking behavior

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content
THE ROLE OF SOCIAL IDENTITY

IN ADOLESCENT SMOKING BEHAVIOR

by

Meghan Bridgid Moran

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

August, 2009

Copyright 2009 Meghan Bridgid Moran

ii

ACKNOWLEDGMENTS

Writing this dissertation has made me feel so lucky for the support of the
following people, without whom none of this would be possible.
Sheila Murphy has been a patient, thoughtful and thorough advisor for the past
five years. She has helped me flush out ideas, find resources, develop materials and
make the leap from a few rough sentences on paper to a full dissertation. She has been an
invaluable presence in my life, acting as needed as a teacher, mentor, defender and friend.
Sandra Ball-Rokeach has similarly been invaluable for her guidance, support and
mentorship. From the first class I took with her, Sandra has taught me things that truly
changed the way I thought and what I saw possible. More importantly, her belief in me
has shaped the way I think of myself as a researcher, teacher and person. Tom Valente
has somehow managed to find time in his schedule to advise me and for that I am
infinitely grateful. Tom’s expansive knowledge on tobacco research and sharp statistical
mind have made this dissertation literally possible – without his support, I am sure I
would have shifted topics long ago.
Many others at Annenberg have also been extremely important over the years.
Michael Cody, the first person I met at ASC, made me feel welcome immediately. Sarah
Banet-Weiser has been an amazing teacher, role model and friend. Anne Marie, Gail,
Christine, Imre, Justin, Abby, Patricia, Ray, Raymond and everyone else have helped me
navigate paperwork, given me opportunities to teach and helped me get into my
classroom whenever I was locked out.

iii

My fellow students and co-workers are truly how I maintained my sanity on a day
to day basis. Laura Portwood-Stacer, thank you for being a friend. I can’t imagine
having done this without you! Vikki Katz, Matthew Matsaganis, Holley Wilkin and
Yong-Chan Kim for their friendship and guidance; Chris Chavez for being my early
morning writing partner; Evelyn, Carmen, Charlotte, Yvonne and Renee for making
going into the office fun.
Finally, my friends and family who have encouraged and supported me from the
beginning. My best friends Kim, Lisa, Sarah, Suzanne, Stephanie, Ainsley and Lori ,
who have been constants for just about my entire life. Abby and Sara for helping L.A.
seem a little more like home. Ed, who I always knew would be here. My brother Kevin
for his jokes. Finally, my mother and father who have been endless cheerleaders and
encouragers, believing in me more than I believed in myself. I would like to officially
apologize for snapping at my mother when she asked how my writing was going, even
though I know she knows her concern meant the world to me. She always told me that
anything was possible and watching her strength and determination as she achieved her
dream made me believe it was true. I would also like to officially apologize to my father
for rolling my eyes during his ‘end of the year’ talks. At the end of every school year, I
received a talk about how the end was in sight and I just needed to ‘bear down and focus’
for finals. As I wrote my dissertation, he didn’t give me that talk but I heard his voice in
my head every day as I sat down to write and what a difference it made.

iv

TABLE OF CONTENTS
LIST OF TABLES .................................................................................................. vii
LIST OF FIGURES................................................................................................. xii
ABSTRACT ......................................................................................................... xiii
CHAPTER 1: THE PROBLEM OF ADOLESCENT SMOKING AND WHAT HAS
BEEN DONE TO UNDERSTAND IT .................................................................. 1
Introduction ............................................................................................................ 1
The Problem of Adolescent Smoking Behavior..................................................... 4
Predictors of Adolescent Smoking Behavior ......................................................... 5
Intrapersonal Influences ........................................................................................ 6
Sociodemographic traits ...................................................................................... 6
Sensation seeking ................................................................................................ 7
Emotional intelligence......................................................................................... 8
Problem behaviors ............................................................................................... 8
Depression/Depressive symptoms....................................................................... 9
Interpersonal Influences ...................................................................................... 10
Peer Influences .................................................................................................. 10
Familial Influences ............................................................................................ 13
Environmental Influences.................................................................................... 15
Ease of access .................................................................................................... 15
Other policies..................................................................................................... 16
Theories Used to Understand Adolescent Smoking Behavior .............................. 17
The Integrative Model of Behavioral Prediction ................................................ 18
Theoretical underpinnings .................................................................................. 18
Utility for adolescent smoking research ............................................................. 20
Social Norms Approach ..................................................................................... 21
Theoretical underpinnings .................................................................................. 21
Utility for adolescent smoking research ............................................................. 21
Sensation Seeking and the SENTAR Approach ................................................... 23
Theoretical underpinnings .................................................................................. 23
Utility for adolescent smoking research ............................................................. 24
Anti-Smoking Campaigns ..................................................................................... 26
Tobacco Company Run Campaigns ................................................................... 27
California Tobacco Education Media Campaign ............................................... 27
Massachusetts Tobacco Control Program Youth Prevention Campaign ........... 28
Florida truth™ Campaign ................................................................................... 29

v

Conclusion ............................................................................................................ 29

CHAPTER 2: SOCIAL IDENTITY THEORY AND PEER CROWD
AFFILIATION ..................................................................................................... 30
Social Identity Theory ........................................................................................... 30
Theoretical Foundations of Social Identity Theory............................................ 32
Why Do Individuals Develop Social Identities? ................................................ 35
Social Identity and Behavior .............................................................................. 36
Why Does Social Identity Affect Behavior? ...................................................... 37
Peer Crowd Affiliation and Adolescent Smoking ................................................. 39
Peer Crowd Affiliation Research........................................................................ 40
Peer Crowd Affiliation and Social Identity ........................................................ 41
Statement of Problem ............................................................................................ 42
Hypotheses ............................................................................................................ 44
Social Identity and Adolescent Smoking Behavior............................................ 44
Social Identity and the Sensation Seeking Approach......................................... 45
Social Identity and the Integrative Model of Behavioral Prediction.................. 46
Social Identity and the Social Norms Approach ................................................ 48
The Role of Social Identity in the truth™ Campaign......................................... 49

CHAPTER 3: SURVEY METHODOLOGY AND MEASURES........................ 51
Sampling Procedure............................................................................................... 51
Sample Characteristics .......................................................................................... 53
Age ..................................................................................................................... 53
Grade .................................................................................................................. 54
Gender ................................................................................................................ 54
Ethnicity ............................................................................................................. 54
Measures................................................................................................................ 55
Academic Achievement...................................................................................... 55
Religiosity........................................................................................................... 55
Smoking in the Home ........................................................................................ 56
Descriptive Norms.............................................................................................. 56
Injunctive Norms ................................................................................................ 56
Sensation Seeking............................................................................................... 56
Attitudes about not Smoking .............................................................................. 57
Attitudes about Cigarettes .................................................................................. 58
Attitudes about Cigarette Companies ................................................................ 59
Exposure to the truth™ Campaign ..................................................................... 60
Smoking Behavior .............................................................................................. 64
Smoking susceptibility ..................................................................................... 64
Smoking initiation ............................................................................................ 64
Established smoking......................................................................................... 65

vi

Smoking status.................................................................................................. 65
Social Identity..................................................................................................... 66
Conceptualizing social identity as a variable .................................................. 66
Measuring strength of identification with social category (SISC) .................. 69

CHAPTER 4: RESULTS ..................................................................................... 82
Interpreting Odds Ratios and Interaction Effects ................................................. 83
H1: Evaluating the Impact of Strength of Identification with each Social
Category on Smoking Behavior ............................................................................ 85
Preliminary Analysis .......................................................................................... 85
Hypothesis Testing .............................................................................................. 88
H2: Evaluating the Contribution of Strength of Identification with each
Social Category to the Sensation Seeking Approach ........................................... 94
Preliminary Analysis .......................................................................................... 94
Hypothesis Testing .............................................................................................. 95
H3: Evaluating the Contribution of Strength of Identification with each
Social Category to the Integrative Model of Behavioral Prediction .................. 102
Preliminary Analysis ........................................................................................ 102
Hypothesis Testing ............................................................................................ 107
H4: Evaluating the Contribution of Strength of Identification with each
Social Category on the Social Norms Approach ................................................ 110
Preliminary Analysis ........................................................................................ 111
Hypothesis Testing ............................................................................................ 111
H5: Evaluating the Impact of Strength of Identification with each
Social Category on the Effectiveness of the truth™ Campaign ......................... 114
Preliminary Analysis ........................................................................................ 115
Hypothesis Testing ............................................................................................ 115
Summary of Results ........................................................................................... 119
Evaluation of the Base Model .......................................................................... 119
Summary of Ability of SISC to Predict Smoking Behavior ............................. 120
Moderating Effect of SISC on Sensation Seeking ........................................... 122
Contribution of SISC to the Basic Integrative Model ...................................... 122
Moderating Effect of SISC on Social Norms ................................................... 123
Moderating Effect of SISC on Exposure to the truth™ Campaign .................. 124

CHAPTER 5: DISCUSSION AND CONCLUSION......................................... 126
Study Findings..................................................................................................... 127
Evaluation of the Base Model ........................................................................... 127
Does Identification with Certain Social Categories Have an Independent
Effect on Smoking Behavior? ............................................................................. 130
Does Identification with Certain Social Categories Contribute to the Sensation
Seeking Approach? ............................................................................................ 133

vii

Does Identification with Certain Social Categories Contribute to the
Integrative Model of Behavioral Prediction? ...................................................... 135
Does Identification with Certain Social Categories Contribute to the Social
Norms Approach?................................................................................................ 137
Does Identification with Certain Social Categories Increase the Effects of the
truth™ Campaign? .............................................................................................. 139
Summary of Results ............................................................................................ 142
Theoretical Contributions...................................................................................... 145
Practical Contributions .......................................................................................... 148
Methodological Contributions............................................................................... 156
Study Limitations ................................................................................................. 159
Conclusion............................................................................................................. 162

REFERENCES ................................................................................................... 164

APPENDIX A ..................................................................................................... 183
APPENDIX B...................................................................................................... 199

viii

List of Tables

Table 3.1. Sample Characteristics – Frequencies ……………………………………...61

Table 3.2. Sample Characteristics – Means ……………………………………………62

Table 3.3 Smoking Status Frequencies ………………………………………………...66

Table 3.4. Frequencies - Social categories ……………………………………………...71

Table 3.5. Social Identity Unstandardized and Standardized Scores …………………....75

Table 3.6. Bivariate Correlation Matrix of SISC items ………………………………...77

Table 3.7. SISC Pattern Matrix …………………………………………………………79

Table 3.8. Final SISC factors …………………………………………………………...80

Table 3.9. SISC Factor Descriptives …………………………………………………...80

Table 3.10. SISC Factor Bivariate Correlations ………………………………………..81

Table 4.1. One-Way ANOVA: SISC Item Means by Smoking Status ………………..86

Table 4.2. One-Way ANOVA: SISC Factor means by Smoking Status ………………87

Table 4.3 Multinomial Logistic Regression Analysis Parameter Estimates for
Base Model Predicting Smoking Status …………………………………………………89

Table 4.4 Multltinomial Logistic Regression Analysis Parameter Estimate for SISC
Items to Predict Smoking Status Controlling for Base Model…...………………………91

Table 4.5 Multinomial Logistic Regression Analysis Parameter Estimate for
SISC-Achiever Factor to Predict Smoking Status Controlling for Base Model…..……..93

Table 4.6 Multinomial Logistic Regression Analysis Parameter Estimates for
Interaction of SISC-Popular Item with Sensation Seeking to Predict Smoking
Status Controlling for Base Model ……………………………………………………...96

ix

Table 4.7 Multinomial Logistic Regression Analysis Parameter Estimate for
Interaction of SISC-Smart Item with Sensation Seeking to Predict Smoking
Status Controlling for Base Model ………………...……………………………………97

Table 4.8 Multinomial Logistic Regression Analysis Parameter Estimate for
Interaction of SISC-Musician Item with Sensation Seeking to Predict Smoking
Status Controlling for Base Model ……………………………...………………………97

Table 4.9 Multinomial Logistic Regression Analysis Parameter Estimate for
Interaction of SISC-Techie Item with Sensation Seeking to Predict Smoking
Status Controlling for Base Model ……………………………………………………...98

Table 4.10 Multinomial Logistic Regression Analysis Parameter Estimate for
Interaction of SISC-Non-conformist Item with Sensation Seeking to Predict
Smoking Status Controlling for Base Model …………………………...………………98
_
Table 4.11 Multinomial Logistic Regression Analysis Parameter Estimate for
Interaction of SISC-Elites Factor with Sensation Seeking to Predict Smoking
Status Controlling for Base Model ……………………………………………………...99

Table 4.12 Multinomial Logistic Regression Analysis Parameter Estimate for
Interaction of SISC-Outsiders Factor with Sensation Seeking to Predict Smoking
Status Controlling for Base Model …………………………………….......…………..100

Table 4.13 Multinomial Logistic Regression Analysis Parameter Estimate for
Interaction of SISC-Individualist Factor with Sensation Seeking to Predict
Smoking Status Controlling for Base Model ………………..…………………………101

Table 4.14 Multinomial Logistic Regression Analysis Parameter Estimate for
Interaction of SISC-Musical Arts with Sensation Seeking to Predict Smoking
Status Controlling for Base Model …………………………………………………….101

Table 4.15 Bivariate Correlations between Attitudes, Social Norm and Self
Efficacy………………………………………………………………………………....103

Table 4.16 Bivariate Correlations between SISC Items and Attitudes, Self Efficacy
and Social Norms ……………………………………………….…………….………..104

Table 4.17 Bivariate Correlations between SISC Factors and Attitudes, Self Efficacy
and Social Norms …………………………………………………….………………...106

x

Table 4.18 Bivariate Correlations between Attitudes, Social Norm and Self Efficacy
with Behavioral Intention to Smoke …………………………………………………...107

Table 4.19 Multiple Regression Analysis Parameter Estimate for Attitudes, Social
Norm and Self Efficacy to Predict Behavioral Intention to Smoke ………..…………..109

Table 4.20 Multiple Regression Analysis Parameter Estimate for SISC Items
with Attitudes, Social Norms and Self Efficacy to Predict Behavioral Intention to
Smoke ……………………………………………………………………………….…110

Table 4.21 Multiple Regression Analysis Parameter Estimate for SISC-Factors
with Attitudes, Social Norms and Self Efficacy to Predict Behavioral Intention to
Smoke …… ……………………………………………………………………..…….110

Table 4.22 One-way ANOVA: Average Percentage Friends Who Smoke by
Smoking Status………………………………………………………………...……….111

Table 4.23 Multinomial Logistic Regression Analysis Parameter Estimate for
Interaction of SISC-Theater Item with Social Norm to Predict Smoking
Status Controlling for Base Model ……………………………………...……………..113

Table 4.24 Multinomial Logistic Regression Analysis Parameter Estimate for
Interaction of SISC-Smart Item with Social Norm to Predict Smoking Status
Controlling for Base Model …………………………………………………..………..113

Table 4.25 Multinomial Logistic Regression Analysis Parameter Estimate for
Interaction of SISC-Achievers Factor with Social Norm to Predict Smoking
Status Controlling for Base Model ………………………………………...…………..114

Table 4.26 Multinomial Logistic Regression Analysis Parameter Estimate for
Interaction of SISC-Techie Item with Exposure to truth™ Campaign to Predict
Smoking Status Controlling for Base Model …………………………………..………116

Table 4.27 Multinomial Logistic Regression Analysis Parameter Estimate for
Interaction of SISC-Nerd Item with Exposure to truth™ Campaign to Predict
Smoking Status Controlling for Base Model …………………..………………………117

Table 4.28 Multinomial Logistic Regression Analysis Parameter Estimate for
Interaction of SISC-Goody-goody Item with Exposure to truth™ Campaign to
Predict Smoking Status Controlling for Base Model …………..………………………117

xi

Table 4.29 Multinomial Logistic Regression Analysis Parameter Estimate for
Interaction of SISC-Conservative Factor with Exposure to truth™ Campaign to
Predict Smoking Status Controlling for Base Model ………….……………………….118

Table 4.30 Multinomial Logistic Regression Analysis Parameter Estimate for
Interaction of SISC-Musical Arts Factor with Exposure to truth™ Campaign to
Predict Smoking Status Controlling for Base Model ………………….………….……118

Table 4.31 Summary of Base Model Parameter Estimates ………………………...….119

Table 4.32 Summary of SISC Items Ability to Predict Smoking Behavior ……..…….120

Table 4.33 Ability of SISC Factors to Predict Smoking Behavior ………………..…..121

Table 4.34 Summary of H2 Results: Interaction of SISC with Sensation
Seeking to Predict Smoking Status ……………………..……………………………...122

Table 4.35. Contribution of SISC to Basic Integrative Model Predicting
Behavioral Intention to Smoke …………………………………………………..…….123

Table 4.36 Summary of H4 Results: Interaction of SISC with Social Norms to
Predict Smoking Status ..………………………………………………………....…….124

Table 4.37 Summary of H5 Results: Interaction of SISC with Exposure to the
truth™ Campaign to Predict Smoking Status………………..…………………………125

Table 5.1 Summary of Effects of Identification with Each Social Category on
Smoking Behavior ……………………………………………………………………..144

xii

List of Figures

Figure 1.1. Integrative Model of Behavioral Prediction. ………………………………..20
Figure 3.1. Example of an individual’s SISC composition …………………………….74

xiii

Abstract

This dissertation introduces social identity (Tajfel, 1978; Tajfel and Turner, 1979) as a
useful construct for both understanding and preventing adolescent smoking. Social
identity, understood in this research as the strength of identification with a social
category, has been found to be useful in predicting other forms of behavior but has never
been used to understand adolescent smoking behavior. This dissertation builds on
research that has found peer crowd affiliation to affect smoking behavior, using Social
Identity Theory to understand the extent to which an adolescent’s strength of
identification with a social category (SISC) affects likelihood of being susceptible to
smoking, having tried smoking and being an established smoker. Specifically, this
dissertation hypothesizes (1) that social identity will impact smoking behavior, (2) that
social identity will interact with sensation seeking to impact smoking behavior, (3) that
social identity will lend predictive validity to the ability of the Integrative Model of
Behavioral Prediction to predict intentions to smoke, (4) that social identity will interact
with social norms to impact smoking behavior, and (5) that social identity will interact
with exposure to the truth™ anti-smoking campaign to impact smoking behavior. To test
these hypotheses, a cross-sectional survey of 224 14 and 15 year olds was conducted.
Results of this survey indicate that social identity is independently related to smoking
behavior after controlling for factors such as age, gender, ethnicity, school achievement,
religiosity, living with a smoker, sensation seeking and social norms. Certain identities,
such as artist, musician and nerd were associated with decreased smoking behavior while

xiv

other identities such as rebel and hipster were associated with increased smoking
behavior. Additionally, inclusion of social identity increased the predictive ability of a
basic Integrative Model containing social norms, attitudes and self efficacy. Finally,
social identity was also found to interact with sensation seeking, social norms and
exposure to the truth™ campaign to impact smoking behavior. These findings indicate
that social identity plays an important role in an adolescent’s smoking behavior. This
dissertation concludes with a discussion of the theoretical, practical and methodological
implications of these findings.

1

CHAPTER 1:
THE PROBLEM OF ADOLESCENT SMOKING AND WHAT HAS BEEN DONE TO
UNDERSTAND IT

Introduction
In the mid 1990’s, the attorneys general of 46 U.S. states brought suit against
corporate tobacco companies for Medicaid costs incurred due to smoking-related
behavior. This lawsuit represented the first public acknowledgment that by selling
cigarettes, tobacco companies were actively harming the American public. The
settlement borne out of this lawsuit was known as the Tobacco Master Settlement
Agreement. Among other restrictions and requirements, tobacco companies were
mandated to distribute funds to states for the creation of anti-smoking programs. With
this settlement came a renewed interest in adolescent smoking behavior.
Adolescent smoking behavior is an issue of interest to health practitioners and
researchers alike. Since the Master Settlement Agreement, rates of adolescent smoking
have declined (Johnston, O’Malley, Bachman & Schulenberg, 2008); however, with over
one-fifth of twelfth graders having reported smoking in the past 30 days, the problem has
clearly not been solved. Because over 80% of current smokers tried their first cigarette
before the age of 18 (Substance Abuse and Mental Health Service Administration:
SAMHSA, 2006), adolescence is a crucial time period on which to focus preventive
efforts.

2

This dissertation examines the problem of adolescent smoking behavior in the
context of social identity. Previous research has traditionally looked at the psychological,
interpersonal and environmental factors predicting adolescent smoking behavior. This
research introduces social identity (Tajfel, 1978; Tajfel and Turner, 1979), the idea that
an individual perceives him or herself as belonging to social categories with
corresponding prototypical behavior, to the study of adolescent smoking behavior. It is
argued that smoking is a communicative behavior that conveys something about the
smoker’s identity to others. As such, during adolescence, smoking may be motivated
primarily by identity. In other words, smoking is a way for adolescents to perform
various identities, rejecting some and maintaining others. The use of identity to predict
adolescent smoking behavior has been touched upon by researchers studying peer crowd
affiliation (Sussman, Pokhrel, Ashmore and Brown, 2007). This dissertation expands
upon this research (1) theoretically, by conceptually grounding this research in Social
Identity Theory, (2) methodologically, by proposing a new, valid and more detailed
measure of social identity; and (3) practically, by testing the effectiveness of anti-
smoking ads that use social identity appeals. Specifically, this project uses a sample of
224 14 and 15 year olds to examine whether the strength with which one identifies with a
social category (such as Nerd, Popular, Skater) influences their smoking behavior. The
concept of social identity is then applied to three models commonly used to understand
smoking behavior. Finally, the extent to which strength of identification with a social
category moderates the effects of an anti-smoking campaign is evaluated.

3

This dissertation consists of five chapters. The current chapter focuses on the
problem of adolescent smoking behavior and what researchers have done to understand
it. Intrapersonal, interpersonal and environmental predictors of smoking behavior are
examined. Then, theories used to understand smoking behavior are presented. Finally,
this chapter contains an overview of anti-smoking campaigns directed towards
adolescents.
Chapter Two introduces Social Identity Theory (SIT) as a framework through
which to understand adolescent smoking behavior. First, SIT is described and evidence
linking social identity to behavior is presented. Next, social identity is linked to
adolescent smoking behavior through an examination of the literature on the effects of
peer crowd affiliation on teen smoking. Chapter Two concludes with a presentation of
this dissertation’s hypotheses.
Chapter Three contains the methodology for the study used in this dissertation.
Sampling methodology, sample characteristics and survey measures are discussed. A
sample survey can be found in Appendix A. This chapter concludes with an in-depth
presentation pertaining to the development of a measure of social identity (Strength of
Identification with Social Category).
Chapter Four details the results of data analysis. Each hypothesis is presented in
turn, with findings summarized at the end of the chapter. In this chapter, key findings are
highlighted and are accompanied by tables presenting full parameter estimates. Non-
significant findings are presented in Appendix B.

4

Finally, Chapter Five contains an examination of the results presented in Chapter
Four. The methodological, theoretical and practical implications of the findings are
explored. This chapter – and this dissertation – concludes with recommendations for the
design of effective anti-smoking campaigns.
Adolescence is a time when individuals are particularly susceptible to risky
behaviors and is also a time when individuals are struggling to figure out who they are
and how to express their identity. The overarching goal of this dissertation is to
understand the extent to which smoking is the result of this identity formation. If
smoking is indeed an identity-based behavior, then researchers, health practitioners and
campaign designers alike can begin to take appropriate steps to create interventions that
effectively utilize this concept to reduce smoking rates.

The Problem of Adolescent Smoking Behavior
In 2007, 22.1% of 8th graders, 34.6% of 10
th
graders and 46.2% of 12
th
graders
reported ever trying a cigarette. Approximately seven percent of 8th graders, 14.0% of
10
th
graders and 21.6% of 12
th
graders reported smoking at least once in the past 30 days
and 3.0% of 8th graders, 7.2% of 10
th
graders and 12.3% of 12
th
graders reported being
daily smokers (Monitoring the Future, 2007). Among current smokers, 80% began before
the age of 18 (SAMHSA, 2006) and approximately one-third began before the age of 14
(Mowery, Brick & Farrelly, 2000). Consequently, many of these individuals who begin

5

smoking in adolescence can be expected to continue into adulthood and consequently
suffer any one of the myriad of smoking-related diseases.
The Centers for Disease Control (CDC: 2005) estimates that approximately
438,000 lives are lost every year due to smoking-related illnesses. The top three causes
of these deaths are lung cancer, chronic obstructive pulmonary disease and ischemic heart
disease. Other illnesses that have been linked to smoking include cardiovascular disease
such as arthrosclerosis and coronary heart disease (U.S. Surgeon General’s Office, 2004),
respiratory diseases such as chronic bronchitis and emphysema (American Cancer
Society, 2008; U.S. Surgeon General’s Office, 2004), as well as cancer of the bladder,
bronchus, cervix, esophagus, kidney, larynx, liver, nasal cavity, oral cavity, pancreas,
pharynx, stomach and trachea (Ontario Task Force on the Primary Prevention of Cancer,
1995; U.S. Surgeon General’s Office, 2004). All together, the cost of smoking-related
deaths and disease totals approximately $157 billion dollars per year (U.S. Surgeon
General’s Office, 2004). As these statistics underscore, addressing the issue of adolescent
smoking behavior is crucial to the health of the country.

Predictors of Adolescent Smoking Behavior
Because adolescent smoking behavior poses such a public health concern, public
health researchers and practitioners have devoted considerable resources to identifying its
predictors. Research examining the underlying factors associated with an adolescent’s
smoking behavior loosely falls into three areas: research which examines intrapersonal

6

factors, research which examines interpersonal factors and research which examines
factors that exist at the environmental level, mainly in the form of school or government
policy. The following section will present the main research findings in these areas.
Intrapersonal Influences
Research has found that certain demographic and personality traits are linked with
higher levels of adolescent smoking. These findings are useful in informing us both about
at-risk groups and also about psychological processes underlying smoking behavior. This
section details research that has examined intrapersonal influences on smoking behavior,
focusing specifically on sociodemographic traits, sensation seeking, emotional
intelligence, problem behaviors and depression/depressive symptoms.
Sociodemographic traits. A substantial body of research has consistently found
specific individual level factors that influence smoking behavior. Ethnicity is linked to
smoking in that African-American adolescents are less likely to smoke than Caucasians,
Native Americans and Latinos (Coogan et al., 1998; Ellickson, Perlman and Klein, 2003;
Johnson, Myers, Webber & Boris, 2004). Similarly, Asians are less likely to smoke than
Caucasians or Latinos (Ellickson, Perlman and Klein, 2003; Asbridge, Tanner &
Wortley, 2005). Gender is also a factor: female adolescents have been found to be more
likely than males to smoke (Coogan et al., 1998). Age is associated with smoking such
that as adolescents progress through middle and high school, they become more likely to
have ever tried a cigarette and to be an established smoker (Coogan et al., 1998).

7

While basic demographic information is useful in allowing practitioners to more
efficiently target at-risk groups, research evaluating why these factors are associated with
smoking is more useful in allowing us to understand the mechanisms through which the
acts of initiating and continuing smoking occur. For example, Ellickson, Perlman and
Klein (2003) found that Asians may be less likely to smoke because as a group they had
higher performance in school and were less likely to have a job. African Americans were
found to have higher levels of religiosity and lower levels of peer orientation, which may
explain the lower rate of smoking behavior (Ellickson, Perlman and Klein, 2003). The
gender difference in smoking behavior could be due to the fact that girls may start
smoking as a way to adopt and maintain a social image (Michell and Amos, 1997).
Individual personality characteristics also offer a way to not just identify at-risk groups
but to understand the process underlying an adolescent’s smoking behavior.
Sensation seeking. Sensation seeking is one such personality trait. Sensation
seeking is linked to a strong desire for novel, varied and extreme experiences and a high
willingness to take risks to obtain those experiences (Zuckerman, 1979, 1994).
Individuals who are high in sensation seeking are more likely to be involved in risky
behaviors such as substance use (Audrain-McGovern, et al., 2003; Palmgreen, Donohew,
Lorch, Hoyle and Stephenson, 2001; Stephenson and Palmgreen, 2001). Accordingly,
teen smokers have been consistently shown to rate high on sensation and novelty seeking
and to be more likely to participate in risk-taking behaviors (Audrain-McGovern et al.,
2004; Coogan et al., 1998; Dinn, Ayciegl & Harris, 2004; Frankenberger, 2004; Johnson

8

et al., 2004; Tercyak, 2003). Sargent, Beach, Dalton and colleagues (2001) found that
sensation seeking was related to increased likelihood of ever having tried a cigarette, an
association potentially due to the relationship between sensation seeking and exposure to
tobacco use in movies. Sensation seeking has also been shown to interact with Attention-
deficit hyperactivity disorder (ADHD) symptoms to produce an effect on smoking
initiation that was stronger than either by itself (Tercyak, 2003).
Emotional intelligence. Several other psychological factors have been found to
associate with smoking in adolescents. Lower levels of emotional intelligence — the
ability to understand, regulate and express emotions — have been found to increase the
likelihood a middle school student had engaged in smoking behavior (Trinidad &
Johnson, 2002; Trinidad, Unger, Chou, Azen & Johson, 2004). Further work by Trinidad
and colleagues showed that among those who had previously tried smoking, individuals
with high emotional intelligence were more likely than those with low emotional
intelligence to smoke again (Trinidad et al., 2004). On a similar note, Trudeau, Lillehoj,
Spoth and Redmond (2003) found that individual rights assertiveness and decision-
making skills – factors associated with high emotional intelligence – were related to
negative expectancies about smoking and higher intentions to refuse cigarettes. Both
negative expectancies about smoking and refusal intentions were negatively related to
smoking initiation.
Problem behaviors. Academic performance has also been associated with
adolescent smoking. Adolescents who perform better academically and who misbehave

9

less in school have consistently been found to be less likely to smoke. For example,
Bryant and colleagues found that poor academic achievement and misbehavior were
strongly associated with smoking behavior two years later (Bryant, Schulenberg,
Bachman, O’Malley and Johnston, 2000; Bryant, Schulenberg, O’Malley, Bachman &
Johnston, 2003). The researchers suggest that early negative experiences with school that
are due to earlier misbehavior and poor achievement may lead adolescents to skip school
and associate with delinquent students, indicating that ultimately the relationship between
school performance and smoking behavior is one of peer effects. The authors also
indicate that it may simply be the experience of high achievement that acts as a protective
factor against future smoking (Bryant et al., 2003). This research seemingly supports
Problem Behavior Theory (Jessor and Jessor, 1977) which hypothesizes that deviant
behaviors such as smoking, poor school performance and delinquency influence each
other reciprocally and are all part of a larger psychosocial ‘problem behavior’ syndrome.
Depression/Depressive symptoms. The relationship between
depression/depressive symptoms and smoking is complicated. Until recently, it was
believed that depression and depressive symptoms had a direct effect on adolescent
smoking behavior (Audrain-McGovern et al., 2004; Cornelius, Leech, Goldschmidt and
Day, 2005; Patton et al., 1998). In recent years, however, this relationship has been
questioned. Ritt-Olson, Unger, Valente and colleagues (2005) found that the relationship
between depression and smoking is mediated through peers, such that depression is
associated with having peers who have pro-smoking attitudes. Association with these

10

peers, in turn, leads to a greater likelihood of experimental smoking (Ritt Olson et al.,
2005).

Interpersonal Influences
While the above mentioned intrapersonal traits have received support for their
role in predicting adolescent smoking behavior, a substantial body of evidence has found
support for smoking as a social behavior, influenced by friends, peers and family
members. The following section presents this literature examining interpersonal
influences on adolescent smoking behavior.
Peer influences. An adolescent’s peers have well-documented effects on that
adolescent’s smoking behavior. This relationship remains strong whether the peer
reference group is close to the individual (e.g. best friends) or more distant (e.g. the
student body of a high school). At the most basic level, an adolescent who has a best
friend who smokes is more likely to smoke (Taylor, Conard, O’Byrne, Haddock &
Poston, 2004). Similarly, research has found that having a group of close friends who
smoke also increases the likelihood that an adolescent will smoke. For each additional
close friend who smokes, adolescents were found to be 1.48 times more likely to be a
regular smoker. Hoffman, Monge, Chou and Valente (2007) argue that the relationship
between an adolescent’s friends’ smoking behavior and that adolescent’s smoking
behavior can be broken into two components: peer selection and peer influence. Peer
selection involves an adolescent choosing friends who smoke or do not smoke; peer

11

influence involves an adolescent being coerced to smoke by friends who smoke
(Hoffman et al., 2007). These researchers found that peer selection was positively
associated with smoking behavior but that peer influence was negatively associated with
smoking behavior.
In a separate study, Hall and Valente (2007) examined the temporal influence of
peers. They asked sixth graders to nominate others as their friends and then measured the
extent to which selecting smokers as friends and being selected by smokers influenced
smoking behavior in 7
th
grade. Selecting smokers as friends in sixth grade increased the
likelihood that an adolescent would smoke in seventh grade, even after controlling for
sixth grade smoking behavior. On the other hand, being selected by smokers in sixth
grade and not reciprocating the nomination resulted in a decreased likelihood of smoking
in seventh grade. Even when an adolescent is not close friends with smoking peers, their
presence increases the likelihood that the adolescent will engage in smoking behavior.
However, regardless of reciprocation, being selected as a friend by smokers in sixth grade
increases the chances that one will nominate more smokers as friends in 7
th
grade. This
environment, where an adolescent has more reciprocal friendships with smokers, could
lead to increased smoking susceptibility and initiation. These results support those of
Hoffman and colleagues (2007).
Even the mere presence of large numbers of smoking students at school can make
an adolescent more likely to smoke. Powell, Tauras and Ross (2005) found that moving
an adolescent from a school where no one smokes to a school where 25% of the student

12

body smokes will increase that adolescent’s chances of smoking by 14.5%. Similarly, in
a study of high school students in Canada, Leatherdale, McDonald, Cameron and Brown
(2005) found that attending a school with a higher rate of 12
th
or 13
th
graders who smoke
increased an adolescent’s likelihood of smoking, such that a student attending a school
with a 60% senior-student smoking rate was 1.8 times more likely to be a regular smoker.
Leatherdale, Cameron, Brown, Jolin and Kroeker (2006) found similar evidence for the
effect of school smoking rate on smoking behavior. In their study, results showed that a
low-risk student (with no friends or family who smoked) was almost three times more
likely to smoke if he or she attended a school with a high prevalence of smokers
(Leatherdale, et al., 2006). In sum, simply being in an environment with peers who
smoke increases an adolescent’s likelihood of smoking. This influence becomes much
stronger when the adolescent chooses those smoking peers as friends.
Some researchers argue that peer influence is at heart, one of normative influence.
Brown, Teufel, Birch, Izenberg and Lyness (2006) found that perceived peer tobacco use
was positively associated with an adolescent’s likelihood of smoking and with his or her
reported past smoking behavior, regardless of actual peer tobacco use. This could be due
to the fact that adolescents are matching their smoking behavior to that which they
perceive to be the norm. Those who perceive higher rates of peer smoking may smoke in
order to match that norm. However, it has also been argued that those who already
smoke may misperceive the norm because they are more likely to spend time with others
who smoke (Brown et al., 2006).

13

Although these peer influences are often thought of as ‘peer pressure,’ it is crucial
to note that adolescents insist that peer pressure was not a factor in their decision to
smoke (Baillie, Lovato, Johnson and Kalaw, 2005). Rather, they see cigarettes as a way
to enter social situations and meet people, noting that smoking can be a common bond
between people (Baillie et al., 2005). Sterling and colleagues (2007) similarly found that
adolescents said smoking helped them fit in to social situations and made them look cool.
In any case, adolescents are adamant that they were not coerced or pressured into
smoking and that it was instead a choice. These remarks are consistent with quantitative
findings that peer influence is not associated with smoking behavior and that rather it is
peer selection (choosing friends who smoke) that correlates with one’s smoking behavior
(Hall & Valente, 2007; Hoffman et al., 2007). In other words, adolescents who choose
individuals who smoke as friends are more likely to also start smoking; however being
chosen by individuals who smoke as friends and not reciprocating that friendship was not
related to smoking behavior. These findings lend validity to adolescents’ claims that they
were not pressured or coerced to smoke by smoking friends and that rather they chose to
smoke because it felt appropriate for the social situation.
Familial influences. While peers have a significant impact on an adolescent’s
smoking behavior, family and home environment are also a factor that greatly affects the
likelihood of an adolescent smoking. There is significant evidence linking parental
smoking behavior to adolescent smoking behavior. When adolescents live with a parent
who smokes, they themselves are much more likely to model that behavior and also

14

smoke (Bricker, Peterson, Sarason, Andersen & Rajan, 2007; Johnson, Myers, Webber &
Boris, 2004; Leatherdale, et al., 2005; Taylor et al., 2004).
In addition to affecting adolescent smoking by acting as models, parents also have
an effect via parenting style. Chassin and colleagues (2005) measured two key
dimensions of parenting: behavioral control (monitoring, consistent discipline) and
acceptance (nurturance, warmth and attachment). The researchers found that lower levels
of behavioral control and acceptance were both positively related to levels of smoking. It
was also found that parental discussion about smoking had a protective influence on
adolescents. While higher levels of behavioral control and acceptance were positively
related to anti-smoking discussion, the variables had independent effects on adolescent
smoking behavior. This indicates that parenting style affects child smoking behavior in
ways other than through increased discussion devoted to smoking. Similarly, Castrucci
and Gerlach (2006) found that authoritative parenting (being caring and nurturing, yet
demanding responsible behavior and enforcing rules) was associated with reduced odds
of current smoking.
Ultimately, both peer and family models may combine to produce a particularly
potent effect on adolescent smoking behavior. Taylor and colleagues (2004) found that
individuals whose environments were saturated with smoking models such as friends,
parents, siblings and other family members, were significantly more likely to smoke than
individuals with no such smoking models. These researchers found that an individual

15

with four models who smoked were over 160 times more likely to smoke than their
counterparts with no smoking models (Taylor et al., 2004).
Environmental Influences
Expanding out even further, the environment in which an adolescent lives –
specifically, the extent to which the environment facilitates or obstructs access to
cigarettes – can also contribute significantly to his or her smoking behavior. Depending
on where an adolescent lives, there may be policies limiting his or her access to cigarette
purchase, limiting the places where an adolescent may smoke and the places where a
non-smoking adolescent may be exposed to smokers. This section examines the ways in
which various policies affect smoking behavior.
Ease of access. School and community policy have been touted as effective ways
to reduce adolescent smoking. For instance, across the United States, sales of cigarettes
to minors are restricted, meaning that no adolescent under the age of 18 should be able to
purchase cigarettes. However, data indicate otherwise. The U.S. Centers for Disease
Control and Prevention (CDC, 2005) found that 12.9% of current adolescents tried to buy
cigarettes in a store in the 30 days prior to taking the survey. Of those, almost half were
not asked to show identification. Attempts have been made to enforce laws restricting
tobacco sales to minors by penalizing merchants through fines. In communities where
illegal sales of tobacco to minors are lower, so are rates of adolescent smoking behavior
(Dent & Biglan, 2004; Pokorny, Jason & Shoeny, 2003). Unfortunately, there are many
alternate routes through which adolescents may obtain cigarettes: through the use of fake

16

IDs, having older friends purchase cigarettes, purchasing cigarettes at a store that does
not follow the law or obtaining cigarettes from a smoking parent or friend (Dent and
Biglan, 2004). Dent and Biglan (2004) found that rates of illegal merchant sales in
communities was positively associated with the rate of smoking among 11
th
graders.
However, lower rates of illegal merchant sale of tobacco was related to youths’ increased
reliance on social sources for cigarettes. Pokorny, Jason & Shoeny (2003) found similar
results, finding that the rate of illegal tobacco sales to minors was positively associated
with smoking initiation. A similar relationship between rate of illegal tobacco sales to
minors and continued cigarette use was not found. These researchers speculate that once
adolescents in these communities begin smoking regularly, they find other ways to obtain
cigarettes, such as using a fake ID or asking adults to make the purchase (Pokorny, Jason
& Shoeny, 2003). These findings indicate that while stricter enforcement of the law
restricting tobacco sales to minors can reduce the rate at which adolescents purchase
cigarettes in stores, smokers will tend to find alternate means through which to obtain
cigarettes.
Other policies. Other policies restricting smoking behavior include clean air acts,
which limit the places in which a smoker of any age may smoke. Many schools have
similar policies banning cigarette use on school grounds. Forster, Widome and Bernat
(2007) argue that clean air policies reduce the rates of adolescent smoking by reducing
the visibility of those who smoke. They speculate that because an adolescent sees fewer

17

people smoking, he or she perceives smoking as less normative and is consequently less
likely to smoke.

Theories Used to Understand Adolescent Smoking Behavior
As the prior literature review suggests, there are a myriad of factors affecting
adolescent smoking behavior. It is therefore not surprising that various theories have
been used to amalgamate these findings into workable models (Collins & Ellickson,
2004; Petraitis, Flay & Miller, 1995). These theories include the Theory of Reasoned
Action, Theory of Planned Behavior and Integrative Model of Behavioral Prediction
(Ajzen, 1988, 1991; Ajzen and Fishbein, 1977, 1980; Fishbein, 1980; Fishbein and
Ajzen, 1975; Fishbein and Cappella, 2006; Fishbein, Hennessy, Yzer and Douglas, 2003;
Fishbein and Yzer, 2003), Social Learning Theory/Social Cognitive Theory (Akers, 1977,
Bandura, 1977, 1986), the Social Norms Approach (Berkowitz, 2004a; Perkins, 2003;
Perkins & Berkowitz, 1986), Social Attachment Theory (Elliott et al., 1985; Hawkins and
Weis, 1985), Problem Behavior Theory (Jessor and Jessor, 1977) and the Sensation
Seeking Approach (Zuckerman, 1979). Petraitis, Flay and Miller (1995) organize
theories predicting adolescent substance use behavior into three categories:
intrapersonal, cultural/attitudinal and interpersonal.
The frameworks this dissertation will focus on are the Integrative Model of
Behavioral Prediction (Fishbein and Cappella, 2006; Fishbein, Hennessy, Yzer and
Douglas, 2003; Fishbein and Yzer, 2003), the Social Norms Approach (Berkowitz,

18

2004a; Perkins, 2003; Perkins & Berkowitz, 1986), and the Sensation Seeking Approach
(Zuckerman, 1979)
1
. These models are chosen (1) because they each represent one of
Petraitis, Flay and Miller’s (1995) three categories of theory, (2) because they are often
and currently used by communication scholars and (3) because they have been used to
design and evaluate anti-tobacco campaigns. The following section offers an evaluation
of these models and an examination of the ways in which they have been used to explain
adolescent smoking behavior and their utility for guiding campaigns and interventions.

The Integrative Model of Behavioral Prediction
Theoretical underpinnings. The Theory of Reasoned Action (Ajzen and Fishbein,
1977, 1980; Fishbein, 1980; Fishbein and Ajzen, 1975), the Theory of Planned Behavior
(Ajzen, 1988, 1991) and the Integrative Model of Behavioral Prediction (Fishbein, 2008;
Fishbein and Cappella, 2006; Fishbein, Hennessy, Yzer and Douglas, 2003; Fishbein and
Yzer, 2003) use cultural and attitudinal factors to predict behavioral intention, which
directly affects behavior. In the Theory of Reasoned Action (TRA), behavioral intention
is determined by attitudes (affective beliefs about the behavior and its outcomes) and
subjective norms (whether or not significant others approve of the behavior and
motivation to comply with those significant others). The Theory of Planned Behavior
(TPB) is similar to the TRA but adds the construct of perceived behavioral control

1
For a review of the models not addressed in this paper, please refer to Collins and Ellickson (2004) or
Petraitis, Flay and Miller (1995).

19

(Madden, Ellen and Ajzen, 1992). Perceived behavioral control consists of beliefs about
the resources and opportunities one has to perform a behavior and acts on behavior both
directly and indirectly through behavioral intentions; thus the TPB is able to predict
behaviors not thought to be under volitional control (Madden, Ellen and Ajzen, 1992).
The TRA and TPB subsequently were subsumed into what is currently known as
the Integrative Model of Behavioral Prediction (Fishbein and Cappella, 2006; Fishbein,
Hennessy, Yzer and Douglas, 2003; Fishbein and Yzer, 2003; see Figure 1.1). The
Integrative Model expanded upon the TRA and TPB, combining attitudinal and
normative factors with components of the health belief model (Janz and Becker, 1984;
Rosenstock, 1974) and social cognitive theory. In addition to recognizing that both
attitudes and subjective norms are important determinants of behavior, the Integrative
Model argues that self efficacy is a third determinant. In other words, individuals must
believe that they are able to perform the behavior in question. The Integrative Model also
takes background or distal variables such as sociodemographics and personality traits into
consideration, acknowledging that these factors will impact the relevant beliefs ultimately
predicting behavior (Fishbein and Cappella, 2006).

20

Figure 1.1. Integrative Model of Behavioral Prediction

Utility for adolescent smoking research. Ter Doest, Dijkstra, Gebhard and Vitale
(2007) found that attitudes toward smoking, subjective norms about smoking, and
perceived behavioral control over not smoking all predicted intentions to smoke. Hanson
(1997) found differences in how the theory worked in African-American, Puerto Rican
and non-Hispanic white female teenagers. Among African-American females, all three
components of the TPB – attitudes, subjective norms and perceived behavioral control –
predicted behavioral intention; however in Puerto Rican and non-Hispanic white females,
subjective norms was not a significant predictor of behavioral intention. McMillan and
Connor (2003) examined the effectiveness of the TPB in predicting smoking among
young adults aged 19-22. They found that both behavioral intentions and perceived
behavioral control had significant direct effects on smoking, but that only perceived

Distal Variables

•Demographics

•Attitudes Towards
Targets

•Personality Traits

•Other Individual
Difference Variables

Behavioral
Beliefs and
Outcome
Evaluations
Normative Beliefs
and Motivation to
Comply
Efficacy Beliefs
Attitud
e
Norm
s
Self
efficacy
Skills
Intention
Environment
al
Constraints
Behavior

21

behavioral control predicted behavioral intention to smoke over the next six months.
Harakeh and colleagues (2004) found that attitudes, self efficacy and social norms were
all related to behavioral intention to smoke, which, in turn, was related to current
smoking behavior and the onset of future smoking (Harakeh et al., 2004).

Social Norms Approach
Theoretical underpinnings. Social norms have been defined in the past as frames of
reference through which the world is understood (Sheriff, 1936) and were later broken
down into descriptive norms, or “what is commonly done,” and injunctive norms “what is
commonly approved and disapproved” (Kallgren, Reno and Cialdini, 2000, pp. 1002).
The Social Norms Approach focuses more in depth on the normative component of the
Integrative Model and primarily focuses on descriptive, as opposed to subjective, norms
as a predictor of behavior. According to the Social Norms Approach (Berkowitz, 2004a;
Perkins, 2003; Perkins & Berkowitz, 1986), individuals are motivated to behave in ways
similar to those around them. An individual’s perception of the behavior of those around
him or her will influence that individual’s behavior (Berkowitz, 2004b). However,
individuals often misperceive the behavior of those around them, either over- or under-
estimating the prevalence of a certain behavior. This leads to that individual behaving in
accordance with a non-existent social norm. Thus, individuals who over-perceive the
prevalence of a certain behavior will be more likely to partake in that behavior.
Utility for Adolescent Smoking Research. Because of these findings, the Social

22

Norms Approach is particularly applicable to adolescent substance use behavior. While
the Social Norms Approach has been most widely applied to the issue of over-
consumption of alcohol by college students (Perkins and Berkowitz, 1986; Campo,
Brossard, Frazer, Marchell, Lewis and Talbot, 2003; Russell, Clapp & DeJong, 2005;
Smith, Atkin, Martell, Allen and Hembroff, 2006; Thombs, Dotterer, Olds, Sharp and
Raub, 2004; Wechsler, Nelson, Lee, Seibring, Lewis and Keeling, 2003; Werch, Pappas,
Carlson, DiClemente, Chally and Sinder, 2000), it has also been shown to be useful to the
study of adolescent smoking behavior.
The smoking behavior of socially close friends and more distant peers has
consistently been found to have an impact on adolescent smoking behavior. The smoking
behavior of one’s friends has been identified as a main factor influencing teen smoking
(Unger, Rohrbach, Cruz, Baezconde-Garbanati, Howard, Palmer & Johnson, 2001).
Similarly, adolescents who had higher estimates of peer smoking behavior also had
higher levels of smoking and smoking susceptibility (Unger, et al., 2001). Finally,
Eisenberg and Forster (2003) found that adolescents did not even need to be socially
connected to peers for their smoking to have an effect. Simply noticing other teens
smoking was related to higher levels of smoking behavior. Researchers have begun to
recognize the importance of social norms on adolescent smoking behavior and have
designed campaigns accordingly. For instance, Linkenbach and Perkins (2003) found
that a campaign that told adolescents aged 10-17 that “MOST of us are tobacco free” was
successful in reducing rates of first time cigarette use. Martino-McAllister and Wessel

23

(2005) evaluated a similar media blitz campaign targeting adolescents aged 12-18,
containing information stating that most youth do not smoke. The campaign was found
to produce reduced perceptions of peer smoking. Along similar lines, Haines, Barker and
Rice (2003) found that a campaign informing high school students in DeKalb County, IL
that over 80 percent of their peers did not smoke was effective in reducing rates of 30-day
smoking by almost 10% two years later. A similar campaign conducted among high
school students in Evanston, IL was found to produce similar results (Christensen &
Haines, 2003).

Sensation Seeking and the SENTAR Approach
Theoretical underpinnings. The sensation seeking framework posits that certain
individuals have a greater need for psychological stimulation and consequently seek out
exciting and novel experiences to satisfy that need (Zuckerman, 1979). Thus, sensation
seeking is a personality trait linked to a strong desire for novel, varied and extreme
experiences and a high willingness to take risks to obtain those experiences (Zuckerman,
1979; 1994). Individuals who are high in sensation seeking are more likely to be
involved in risky behaviors such as drug use (Audrain-McGovern, Teriyaki, Shields,
Bush, Espinel and Lerman, 2003; Palmgreen, Donohew, Lorch, Hoyle and Stephenson,
2001; Stephenson and Palmgreen, 2001). This is because high sensation seekers may be
drawn to the thrill that comes from the risk involved with using illegal drugs, or they may
be drawn to the novel sensory experience that one may obtain from using drugs and other

24

substances (Yanovitzky, 2006). Additionally, high sensation seekers have a tendency to
overestimate their own invulnerability, while underestimating the risks involved with
substance use (Audrain-McGovern, et al., 2003; Yanovitzky, 2006). Petersen, Clausen
and Lavik (1989) found that while the disinhibition component of Zuckerman’s sensation
seeking scale was positively related to smoking behavior, the thrill and adventure-seeking
component was negatively related to smoking behavior. The authors argue that this
indicates cigarette smoking may be related to a specific type of sensation seeking
personalities who may be more restrained. That is, smoking may require a certain level
of disregard for its negative consequences, but is not a behavior associated with thrill or
adventure.
Utility for Adolescent Smoking Research. Because high sensation seekers are
especially at risk for substance use and abuse, researchers have identified this population
as a viable one for targeting by anti-substance campaigns. Palmgreen et al. (2001) have
developed an approach called SENTAR (for sensation seeking targeting). In this
approach, sensation seeking is used as a variable to segment the audience and campaign
ads are designed with this specific high sensation seeking audience in mind. Using
SENTAR, one should first conduct formative research with high sensation seekers to
facilitate the development of high sensation value messages. Once the messages are
developed, they should be placed in high sensation value contexts (for example, action-
filled television programs known to attract high sensation seeking audiences) because
high sensation seekers are drawn to media that can fulfill their desire for new and

25

stimulating experiences (Palmgreen et al., 2001; Stephenson, 2002; Stephenson, Morgan,
Lorch, Palmgreen, Donohew and Hoyle, 2002).
The value of this approach has received some empirical support. Stephenson et
al. (2001) found that high sensation seekers were significantly more likely than low
sensation seekers to have seen high sensation value public service announcements
specifically designed to target them. Palmgreen et al. (2001) tested the effects of anti-
marijuana public service announcements developed for high sensation seekers in
Kentucky. They found that there was no significant effect on low sensation seekers,
whose marijuana use was low to begin with, but there was a significant positive effect on
high sensation seekers.
Though Palmgreen et al. (2001) did not control for ad exposure, Stephenson and
Palmgreen (2001) propose that, independent of the likelihood of increased exposure, anti-
marijuana messages with a high sensation value may produce an effect in high sensation
seekers through various processing mechanisms. For example, in high sensation seekers,
the perceived message sensation value has been shown to be positively associated with
amounts of cognitive, narrative and sensory processing. Stephenson (2003) further found
that perceived message sensation value was associated with sympathetic distress (a
construct which incorporated levels of fear, sadness, sympathy, being afraid and being
upset). The level of sympathetic distress was found to mediate argument-based
processing among high sensation seekers, but not among low sensation seekers.

26

Consequently, higher levels of sympathetic distress in high sensation seekers elicited
stronger anti-marijuana attitudes.
Because sensation seeking is related to both marijuana use and cigarette use, these
findings can be applied to anti-smoking campaigns. It is likely that the same tactics that
worked for anti-marijuana ads will also work for anti-smoking ads. The following
section details the characteristics and effects of anti-smoking campaigns.

Anti-Smoking Campaigns
Anti-tobacco campaigns traditionally consist of school-based interventions, media
campaigns or some combination of the two. Reviews of school-based campaigns have
found that they produce moderate effects in the short-term (Bruvold, 1993; Bruvold &
Rundall, 1988; Rooney and Murray, 1996); however, support for long-term effects is
weak (Wiehe, Garrison, Christakis, Ebel and Rivara, 2005). Anti-smoking campaigns
carried out entirely through media outlets tend to be even less effective than school-
based, both in the short and long term. In a meta-analysis conducted by Snyder and
colleagues (2004), media-based anti-smoking campaigns had an effects size of = .06.
This indicates that across the board, media-based anti-smoking campaigns had relatively
small effects in reducing smoking behavior. Similar results were found by Rooney and
Murray (1996). In general, media campaigns are more effective at increasing knowledge
and changing attitudes than in changing behaviors; as such, many, though not all, media-
based campaigns had low levels of effectiveness for behavior change. Regardless, over

27

the past 15 to 20 years, many states, as well as tobacco companies, have instituted media-
only anti-smoking campaigns, most likely because of the cost and resources required for
school-based interventions.
Tobacco Company Run Campaigns
The tobacco industry has run anti-smoking advertising campaigns, as mandated
by the Tobacco Master Settlement Agreement of 1998. These campaigns include
“Tobacco is Whacko if You’re a Teen” from Lorillard and “Think. Don’t Smoke,” from
Phillip Morris. Phillip Morris claims that its ads were designed to show teens that they
don’t have to smoke to fit in (Biener, 2002; Wakefield, Terry-McElrath and Emery,
2006) and tobacco company websites insist that they are fully committed to the health of
the nation and its children (Warner, 2002). However, research has found the “Think.
Don’t Smoke” and “Tobacco is Whacko” campaigns to be wholly ineffective (Biener,
2002; Wakefield et al., 2006). In fact, exposure to tobacco company anti-smoking
campaigns was found to be positively related to intentions to smoke (Wakefield et al.,
2006). Lorillard has been accused of using its anti-smoking campaign as a way to
advertise to teens in outlets otherwise prohibited by law and using the campaign to gather
psychographic data about teens (Landman, Ling and Glantz 2002). As such, there have
been calls for tobacco companies to run their anti-smoking campaigns in more
responsible ways using channels and techniques proven to be effective, or to stop running
these campaigns altogether (Landman, Ling and Glantz, 2002; Warner, 2002).
California Tobacco Education Media Campaign

28

Statewide anti-tobacco campaigns have proven to be more successful than those
run by tobacco companies; however, levels of effectiveness vary from state to state. The
California Tobacco Education Media Campaign focused on the negative health effects of
second-hand smoke and on highlighting deceptive tactics tobacco companies use to sell
cigarettes (Balbach & Glantz, 1998; Honig & Kizer, 1990; Pechmann & Reibling, 2000).
Additional themes in the campaign included portraying smokers as poor role models and
teaching refusal skills (Pechmann & Reibling, 2000). Research conducted on ads aired in
1990 and 1991 found that the campaign had only limited success (r = .03) in preventing
adolescent smoking. However, in 1993 through 1995, while the national smoking
prevalence in smoking among 8
th
and 10
th
graders rose, this campaign is credited with
preventing a similar increase among California adolescents (Pechmann & Reibling,
2000).
Massachusetts Tobacco Control Program Youth Prevention Campaign
In 1999, the Massachusetts Tobacco Control Program began a youth prevention
campaign aimed to keep youth aged 9 to 17 from smoking. This campaign focused
heavily on the negative medical effects of smoking. It also included messages regarding
second-hand smoke and the negative cosmetic and social effects of smoking. The
campaign’s first round of public service announcements (PSAs) showed individuals
giving first-hand accounts of the negative outcomes of smoking. Later PSAs highlighted
the large number of Americans killed by smoking-related illness and also featured
industry attacks (Biener, 2002). This campaign was not uniformly effective in reducing

29

adolescent smoking rates. From 1993 to 1999, smoking prevalence among
Massachusetts adolescents varied. At times for certain groups smoking prevalence
decreased (10
th
graders; 1997-1999) while at other times it increased (10
th
graders; 1993-
1995) or remained flat while other age groups showed declines (12
th
graders, 1997-1999).
Florida truth™ Campaign
The Florida truth™ campaign is one state-wide media-based anti-smoking
intervention that has proven to be effective, so much so that it has developed into a
nationwide campaign. This campaign targets adolescents aged 12 to 17 and features
“edgy” and “cutting edge” youth, focusing primarily on exposing the deceptive marketing
approaches of tobacco companies (Farrelly, Healton, Davis, Messeri, Hersey and
Haviland, 2002, pp. 901). Research has found this campaign to have a marked level of
success in increasing anti-smoking attitudes and decreasing smoking behavior among
adolescents (Farrelly et al., 2002; Farrelly, Davis, Haviland, Messeri & Healton, 2005).
Because of the initial success of the campaign in Florida, the campaign began airing
nationally and continues to do so.

Conclusion
Despite all of the work devoted to understanding and preventing adolescent
smoking, there is still work to be done, particularly in the arena of translating theory to
campaign design. The case of the truth™ campaign deserves attention: this campaign has

30

had considerable success — more than most other media-based anti-smoking campaigns
— and yet there has been no research to determine the factors and theoretical
underpinnings contributing to its success. This dissertation proposes to fill this gap by
introducing Social Identity Theory (Tajfel, 1978; Tajfel and Turner, 1979) as a useful
construct for both understanding and preventing adolescent smoking. Substantial
amounts of research indicate social identity has potential to predict behavior, including
adolescent smoking behavior. This construct is being studied for its ability to predict
smoking behavior and also for its ability to translate into effective campaign design.
Ultimately, this dissertation argues that social identity may be a useful construct for teen
smoking research to consider because of the potential ability to (1) independently predict
teen smoking behavior; (2) contribute to existing models of teen smoking behavior
(Integrative Model of Behavioral Prediction; Social Norms Approach; Sensation
Seeking); and (3) contribute theoretically and practically to the design of anti-smoking
campaigns. The next chapter of this dissertation will explore Social Identity Theory and
research indicating that it is a viable way to understand adolescent smoking behavior.

31

CHAPTER 2:
SOCIAL IDENTITY THEORY AND PEER CROWD AFFILIATION
The social identity perspective is a broad framework encompassing Social Identity
Theory (SIT: Tajfel, 1978; Tajfel and Turner, 1979), Self-Categorization Theory (Turner
et al., 1987) and research based on the premise that individuals define at least part of who
they are from their membership in social groups and categories (Hogg, 2003; Hogg 2005,
Hogg, 2006; Hogg & Reid, 2006). The following chapter first examines the social
identity perspective, focusing specifically on aspects of Social Identity Theory that align
most closely with peer crowd affiliation research. Next, research on peer crowd
affiliation which has found that identifying with a certain group affects an adolescent’s
smoking behavior is presented. Finally, the concept of social identity is linked to these
findings in order to lay the foundation for this project’s hypotheses which are presented at
the end of this chapter.

Social Identity Theory
Social Identity Theory (SIT) posits that individuals understand and define
themselves in terms of the social groups or categories to which they belong. This theory
was initially developed to understand intergroup discrimination but over the years has
been used to explain how identification with social groups and categories affects a
myriad of behaviors. The way SIT is used in the present research is more in accordance

32

with these later interpretations. In this section, research detailing the theoretical
underpinnings of SIT and its effect on behavior are presented.

Theoretical Foundations of Social Identity Theory
According to Social Identity Theory (Tajfel, 1978; Tajfel and Turner, 1979),
individuals have numerous identities based on their self-categorization as members of
various groups. These are known as social identities, as opposed to personal identity
which is based on an individual’s idiosyncratic quirks or personality traits (Turner, 1999).
Specifically, social identity is defined as “a self definition in terms of social category
membership” (Turner, 1999, p. 10) or “the individual’s knowledge that he belongs to
certain social groups together with some emotional and value significance to him of the
group membership” (Tajfel, 1972, p. 31). Social groups occur when two or more people
identify themselves as belonging to the same social category (Turner, 1982b). For
example, ‘jocks,’ ‘techies,’ and ‘preppies’ would all be considered social categories.
Although social categories occur at the group level, social identity — the knowledge that
one belongs to a social category — is a property of the individual.
In order for one to develop a social identity, three basic processes must occur.
First, one must recognize that various social categories exist and are distinct from each
other. For example: Catholics are different from Protestants, Christians are different
from Muslims. These social categories may be broad (for example, female) or narrow
(member of the college basketball team) and can correspond to any kind of signifier that

33

people use to distinguish each other — for instance, race (African American), region
(Southern Californians), cultural (fan of a certain band), religion (Jewish), and so on.
The key in this first process is the recognition that people can be grouped according to
any number of properties. This process of distinction occurs when members of one
category (an in-group) are perceived to be similar to each other, and different from
members of another category (an out-group). Second, one must come to see him or
herself as belonging to certain social categories but not others. For example, one must
think of herself as a female but not male, a Los Angelino but not a San Franciscoan. That
one must have a perception of belonging to a group is crucial here. For example,
although others may place someone in a social category (academic, jock, punk), that
individual can only be said to possess that particular social identity if he or she actively
self-categorizes as such. In this way, self-categorization is a psychological phenomenon,
whereby an individual perceives him or herself as identifying with or belonging to a
social category with a corresponding prototype (Hogg and Abrams, 1999). Third, in
order for social identity to develop, the social categories into which one self-categorizes
must have prototypes that provide the individual with relevant ways with which to define
his or her self (Hogg and Abrams, 1988; Tajfel & Turner, 1979; Turner, 1982b). In other
words, prototypes define what a group or category means for the specific individual.
Prototypes are cognitive representations of the attributes that define a social
category (Hogg & Reid, 2006). They are, essentially, the quintessence of what it means
to be a group member – the collection of the group’s defining behaviors, values, attitudes

34

and beliefs. A group prototype is the perception of what is normative for that group. As
such, group prototypes define the group’s uniqueness, differentiating one group from
another (Hogg, Terry and White, 1995). Without prototypical behaviors, a social identity
is essentially meaningless. For example, one could be labeled a ‘preppy,’ but unless this
social category has corresponding prototypical behavior, such as wearing distinctive
clothing and espousing upper class ideals and aspirations, it fails to serve the function
(and act as) a social identity, which is to say it does nothing to distinguish ‘preppies’
from those who belong to other social groups and categories, such as ‘jocks’ or ‘nerds.’
Prototypes vary among groups, although some groups may share certain norms
and behavior. For example, ‘techies’ and ‘nerds’ may share the prototypical behavior of
performing well in school, but ultimately have prototypes that are very different. Thus it
is the collection of norms constituting a unique whole that make up the prototype, or
content, of a social identity.
Although prototypes are ultimately individual cognitive representations of what
defines a social group, there is often a consensus with respect to these defining features
among members because prototypes tend to be enacted and reinforced time and time
again (Hogg & Reid, 2006). When individuals self-categorize as group members, they
begin to think of themselves in terms of this prototype, a phenomenon known as
depersonalization. For example, individuals who use Mac computers may begin to define
themselves as “Mac people,” an identity which carries meanings about who they are as
creative, laid back and politically liberal individuals. Without a consensus about the key

35

characteristics of the prototypical Mac user, the Mac user identity would have little value.
Thus, without a group prototype, a social identity would hold no meaning in the mind of
an individual group member.
An individual can self-categorize as a member of any number of social categories
and consequently can possess numerous social identities. For example, an adolescent
may self-categorize as ‘punk,’ ‘nerd’ and ‘techie.’ However, an individual’s numerous
social identities can and often do vary in terms of fit and accessibility or salience (Hogg
& Reid, 2006). This individual may view him or herself as very similar to the ‘punk’
prototype yet only somewhat similar to the ‘techie’ prototype, and so the fit of the ‘punk’
prototype is better than that of the ‘techie’ prototype. Social identities can also be more
or less accessible, depending on the situation. Social identities tend to be most accessible
when they become the characteristic that defines an individual against a group of others
(Forehand, Deshpande, and Reed, 2002) – a person’s punk social identity may be
extremely accessible when the individual is in a room full of country music fans.
Why Do Individuals Develop Social Identities?
According to SIT, individuals break society into categories because it makes
dealing with the overwhelming amount of stimuli in life manageable by allowing the
individual to group similar items together, thus highlighting their similarities (Cantor,
Mischel and Schwartz, 1982; Hogg and Abrams, 1999; Tajfel, 1957; 1959). Second,
individuals compare these categories. When individuals are a member of a category, they
tend to display in-group bias and rate that category higher than others (Hogg and Abrams,

36

1999). This in-group bias serves the individual’s inherent need for self-esteem (Hogg
and Abrams, 1999; Tajfel and Turner, 1979; Turner, 1982a, 1982b). Thus, the need to
break an overwhelming society into smaller subsets and the need for self esteem are what
motivate individuals to categorize, compare and to adopt social identities. These social
identities in turn provide individuals with a stable and manageable framework for society
and with a positively enhanced sense of self.
Social Identity and Behavior
SIT was originally developed as a way to explain intergroup behavior.
Specifically, Tajfel and Turner developed the theory to explain intergroup bias and the
minimal group paradigm – that is, why individuals display positive bias towards those to
whom they are in a group with, even when the group is randomly assigned (Hogg, 2006;
Tajfel, Flament, Billing & Bundy, 1971; Tajfel, 1972, Turner, 1978). In the minimal
group paradigm, the social identity of being randomly assigned as a ‘Group A’ or ‘Group
B’ member was sufficiently strong and salient as to impact individual’s behavior towards
the other group. Over the years, however, SIT has been used to explain other forms of
behavior such as recycling or sunscreen use (Sparks & Shephard, 1992; Terry & Hogg,
1996).
The extent to which one self-categorizes as a member of a specific group has been
shown to have an impact on other forms of behavior. When an individual subscribes to
an identity for which a particular behavior is prototypical or normative, he or she is more
likely to perform that behavior (Hogg and Reid, 2006; Sparks, 2000, Terry & Hogg,

37

1996). For example, Greene (2004) found that for individuals who had partisan social
identities (as Republicans or Democrats), the social identity affected partisan behavior,
such as voting and participating in partisan political activities. Berger and Rand (2008)
found that associating a negative health behavior with an undesirable out-group reduced
the likelihood an individual would perform that behavior. Undergraduates were less
likely to choose junk food options form a menu when they were told that eating junk food
was typical of graduate students (the out-group). Invernizzi, Falomir-Pichastor, Muntoz-
Rojas and Mugny (2003) examined the role of social identity in continuing or quitting
smoking and proposed that the salience of the ‘smoker’ identity influences one’s
motivation to quit. They found smokers viewed their ‘smoker’ identities positively and
tended to resist anti-smoking messages because they were perceived as a threat to their
identities.
It is important to note that individual prototypes of a social identity may vary;
because of this the strength of a social identity will impact only behavior that one sees as
prototypical of that identity. So, two individuals could possess equally strong “green
consumer” identities. If one person sees recycling as a prototypical behavior of the green
consumer identity, SIT predicts he would be more likely to recycle. If the second person
does not see recycling as a prototypical behavior of that identity, the strength of his
“green consumer” identity would have no bearing on his likelihood to recycle.
Why Does Social Identity Affect Behavior?

38

The mechanisms through which social identity affects behavior are still contested
(Dickson & Scheve, 2006). The theoretical explanation of why social identity impacts
behavior hinges upon the premise that behavioral motivations vary on a continuum from
interpersonal to intergroup (Tajfel, 1974; 1978). At the interpersonal extreme, behavioral
motivation is located at the person-to-person level, determined by individual personality
and idiosyncratic quirks. At the intergroup extreme, behavioral motivation is located at
the group level, such that an individual is motivated to act in accordance with a group
prototype. In the case of the minimal group experiments (Tajfel, 1970), the ‘Group A’
and ‘Group B’ prototypes were simple: the groups were defined solely by their label
distinction from the other. That is, Group A’s defining characteristic was that it was not
Group B. Thus, behavior towards ‘Group B’ was motivated by being a ‘Group A’
member and vice versa. According to SIT, when an individual acts according to the
prototype for one of their social identities, they are doing so because they are acting out
who they are. So, because Group A’s prototype consisted solely of being other than
Group B, in displaying bias against Group B, Group A members were acting out who
they were as Group A members. In other words, social identity served a “self definitional
function” (Hogg & Reid, 2006, pp. 12) and the prototype acted as a framework for
organizing behavior according to who one is as a member of a social category.
Hogg and Abrams (1999) argue that there are three processes that must occur in
order for social identity to exert influence on behavior. First, individuals must identify
themselves as belonging to a specific social category. Second, individuals must possess a

39

prototype of that category. Third, individuals must self-stereotype and assign that
prototype to themselves. It is crucial to note here that, since prototypes are cognitive
representations of group norms, the actual group behavior need not matter for social
identity to influence behavior.
In certain situations and for certain actions, behavior is more likely to be
governed by a specific social identity than in others. Hogg and Turner (1987) identify
three factors that impact the influence of social identity on behavior. First, an individual
is more likely to act in accordance to a prototype when that prototype is clear and the
social category is relevant. Reed and Forehand (2003) argue that a social identity will
only influence behavior to which it is relevant. For example, self-categorizing as a ‘jock’
might influence sneaker purchase (Nikes and not Sketchers) but may have no relevance
to movie choice. That is to say movies may not be included in the prototype for ‘jock.’
Second, social identity is more likely to influence behavior when the social category is
desirable to the individual. Thus, an individual who believes that the ‘preppy’ social
category carries some sort of status would be much more likely to act according to his or
her ‘preppy’ social identity than to their ‘nerd’ social identity. Finally, situational factors
affect the extent to which social identity influences behavior. When a situation is
uncertain, individuals are more likely to look to relevant social identities for guidance as
to how to behave.

Peer Crowd Affiliation and Adolescent Smoking

40

Adolescents are a group for whom SIT is particularly relevant, given their
tendency to self-categorize into different peer group types (Ashmore, Del Boca, & Beebe,
2002; Palmonari, Pombeni, and Kirchler, 1990; Kirchler, Palmonari, and Pombeni,
1993). Sussman, Pokhrel, Ashmore and Brown (2007) argue that adolescents identify
with social groups that provide them with a sense of identity. Often, teens simply come to
identify with “reputation-based collective[s]” (Sussman et al., 2007). In other words,
adolescents tend to classify themselves into social categories with established and fairly
stable prototypes. While SIT has been used mainly to understand intergroup
discrimination among adolescents (e.g. Tarrant, North, Edridge, Kirk, Smith & Turner,
2001; Tarrant, 2002), peer crowd affiliation research takes a very similar perspective and
has been useful in predicting how identification with peer groups affects adolescent
smoking behavior. In this section, I present research examining the extent to which peer
crowd affiliation impacts adolescent smoking behavior and then link this research to
Social Identity Theory.
Peer Crowd Affiliation Research
Peer crowd identification research has explored the social crowds or categories
into which adolescents group themselves and the ways in which this self-categorization
predicts behavior such as smoking. In general, this research has identified five broad
groups into which adolescents tend to self-categorize: “elites,” “athletes,” “deviants,”
“academics” and “others” or as a similar variation thereof (Sussman et al., 2007). Elites
tend to have a high level of status among their peers and do well academically and

41

socially. Athletes also tend to be popular and participate in school sports. Academics
tend to excel in the classroom, more so than the elites, but tend to have less social status.
In contrast, deviants do not usually do well academically or participate in extracurricular
activities. Others include a variety of groups that did not fall into the first four categories.
Some groups that fall into others include normals, outsiders and floaters who identify
with a variety of peer groups.
Across the board, research has found some peer groups to be more likely to
engage in problem behavior such as smoking and some peer groups to be less likely to
engage in this behavior. Academics are considerably less likely to engage in alcohol use
(Ashmore, Del Boca & Beebe, 2002; Downs & Rose, 1991), drug use (Downs & Rose,
1991; Urberg et al., 2000). Jocks were more likely to use alcohol (Barber et al., 2001).
Criminals were more likely to smoke cigarettes (Cohen, 1979; Eckert, 1983; La Greca et
al, 2001; Mosbach & Leventhal, 1988; Sussman et al., 1990; Sussman et al., 1993;
Sussman et al., 1994; Sussman et al., 1999; Sussman et al., 2000; Urberg, 1992), use
drugs (Barber et al., 2001; Downs & Rose, 1991; La Greca et al., 2001; Sussman et al.,
1990; Sussman et al., 1999; Sussman et al., 2000; Sussman, Unger & Dent, 2004) and use
alcohol (Barber et al., 2001; Cohen, 1979; Downs & Rose, 1991; La Greca et al., 2001;
Mosbach & Leventhal, 1988; Sussman et al., 1990; Sussman et al., 1999; Sussman et al.,
2000). Elites were more likely to smoke cigarettes (Dolcini & Adler, 1994; Mosbach &
Leventhal, 1988), use alcohol (Dolcini and Adler, 1994; Mosbach & Leventhal, 1988;

42

Tolone & Tieman, 1990) and use drugs (Dolcini and Adler, 1994; Tolone and Tieman,
1990).
Peer Crowd Affiliation and Social Identity
Social identity theory offers a way to theoretically explain peer crowd affiliation
findings. The crucial aspect of this research is that it was not necessarily the behavior of
individual peers adolescents socialized with (i.e. the number of their friends who
smoked), but rather the peer crowds with which they identified, that predicted smoking
behavior. These findings seem to indicate that smoking is a behavior that is potentially
motivated by group membership. According to SIT, each of these peer crowds should
have a corresponding prototype of which smoking, drug or alcohol use are a part. These
prototypes provide adolescents with a frame of reference for behavior. Behaving in
accordance with the group prototype (i.e. smoking) allows adolescents to perform their
social identities and to distinguish themselves from others.

Statement of Problem
Research on adolescent smoking behavior has offered an extraordinary amount of
theoretically sound and empirically supported insights. However, there is still work to be
done, particularly in the arena of campaign design. This dissertation uses peer crowd
affiliation research as a starting point and incorporates Social Identity Theory to further
interrogate the role of social categories in adolescent smoking behavior.
Tajfel argued that behavior varies along a continuum from intergroup to

43

interpersonal. At the intergroup extreme, all behavior is determined by one’s
membership in a social group or category; at the interpersonal extreme, behavior is
determined by idiosyncratic personality traits and quirks (Turner, 1999). This
dissertation argues that smoking has been largely (though not entirely) treated as a
behavior that falls on the interpersonal end of the continuum, determined largely by
individual factors such as knowledge, attitudes, family situation and personality traits.
This dissertation takes the position that, instead, smoking should be treated is a behavior
that falls on the intergroup end of the spectrum, determined by membership in a group.
This, I argue, is why research has found peer crowd affiliation to have an independent
effect on smoking behavior.
This dissertation contributes to the extant literature on adolescent smoking
behavior in a number of ways. First, this project offers a methodological contribution.
Currently, the way in which peer crowds are classified is broad and non-specific to the
behavior in question. Peer crowds are generally broken down into five categories:
“elites,” “athletes,” “deviants,” “academics” and “others” (Sussman et al., 2007). While
these categories are well-understood and tend to apply to adolescents in a variety of
regions, they are still relatively unspecific. Additionally, the content of these peer crowds
with regard to the behavior in question (in this case, smoking) has not been thoroughly
interrogated. In general, group members tend to be asked whether they partake in a
certain behavior and then this information is aggregated by the researcher. There is no
assessment on the part of the adolescent as to whether this behavior is considered

44

prototypical of the group, or how central that identity is to that person’s self-concept.
This project offers a more nuanced way of measuring social identity that is guided by
what adolescents believe to be the prototypes of certain categories.
Second, this dissertation makes a theoretical contribution. To begin with, this
project conceptually grounds peer crowd affiliation research using Social Identity
Theory. Perhaps the main value Social Identity Theory lends to peer crowd affiliation
research is the concept of prototypes. Prototypes, or cognitive representations of what a
social category means, offer a way to theoretically link one’s affiliation with a peer group
to his or her smoking behavior. Additionally, this project examines how social identity as
a member of a peer group or social category can fit into the Integrative Model of
Behavioral Prediction, the Social Norms Approach and the Sensation Seeking Approach.
This will further our understanding of how social identity relates to other variables
commonly used to predict adolescent smoking behavior.
Finally, this dissertation offers a practical contribution. This project evaluates the
effectiveness of the truth™ anti-smoking campaign and the role social identity plays in
its effectiveness. This project will determine of which groups both smoking and not
smoking are prototypical. It will also elucidate which of these groups do and do not exert
influence on the smoking behavior of their members. In doing so, this project will
provide useful information to health and media practitioners and researchers as to how
anti-smoking campaigns can use the social identity approach to target adolescents and
increase the effectiveness of their campaigns.

45

Hypotheses
Social Identity and Adolescent Smoking Behavior
The research conducted in this project consisted of a survey of 250 adolescents
aged 14-15. The purpose of this survey is to evaluate the extent to which social identity
plays a role in smoking behavior; to evaluate the extent to which social identity
contributes to the Integrative Model of Behavioral Prediction, the Social Norms
Approach and the Sensation Seeking Approach; and to evaluate the extent to which social
identity contributes to the effectiveness of the truth™ anti-smoking campaign.
As noted above, a significant amount of prior research has found an association
between membership in certain peer crowds and smoking behavior. This project seeks
to understand whether social identity is the mechanism underlying this finding:

H1a: Identification with social identities for which smoking is prototypical will
be positively related to adolescent (a) smoking susceptibility, (b) trying smoking
and (c) established smoking.
H1b: Identification with social identities for which smoking is not prototypical
will be negatively related to adolescent (a) smoking susceptibility (b) trying
smoking and (c) established smoking.

46

Social Identity and the Sensation Seeking Approach
Research has shown that the sensation seeking personality trait is positively
related to smoking behavior. Social Identity Theory can be used to interrogate this
finding by hypothesizing that certain social identities will interact with sensation seeking
to produce an increased or decreased likelihood of smoking. The image of the smoker is
still very much perceived as risk taking and rebellious (e.g. Lloyd & Lucas, 1998;
Pederson, Koval, & O’Connor, 1997). It is possible, then, that social identities for which
smoking is prototypical are also associated with rebellious and risk taking and will
magnify the effect of sensation seeking on smoking behavior, and that social identities
that are associated with more conservative behaviors will suppress the effect of sensation
seeking on behavior. In other words, while an individual may psychologically possess the
trait of high sensation seeking, if he or she does not possess a corresponding rebellious
identity, then sensation seeking behavior such as smoking may be inhibited. Thus, this
project hypothesizes that:
H2: Social identity will interact with sensation seeking to affect (a) smoking
susceptibility, (b) trying smoking and (c) established smoking, such that the
relationship between sensation seeking and smoking behavior will be stronger
among adolescents who identify with social identities for which smoking is
prototypical.

47

Social Identity and the Integrative Model of Behavioral Prediction
The Integrative Model of Behavioral Prediction has also found support for its
utility in predicting behavior. This project argues that the predictive validity of the
Integrative Model could be strengthened by incorporating the construct of social identity.
Relatively few researchers have examined the link between social identity and variables
such as attitudes; however the few studies that have done so showed evidence supporting
the inclusion of social identity into the theory of planned behavior, a forerunner of the
Integrative Model. Sparks and Shephard (1992) found both that social identity was
significantly and positively correlated with attitudes, subjective norms and perceived
behavioral control regarding a prototypical behavior for that identity. The researchers
also found that social identity had an independent effect on that behavior, indicating
identity could be a useful addition to the Integrative Model of Behavioral Prediction. The
Integrative Model proposes that the intermediate variables of attitudes, norms and self
efficacy are predicted by background variables such as demographics, personality traits
and other factors unique to the individual (Yzer, 2004); in its inclusion of individual
difference variables, this model readily allows for this conceptual incorporation of social
identity (Fishbein et al., 2003; Yzer, 2004). According to Social Identity Theory, a
behavior should be prototypical for a corresponding social identity (Sparks and Shepard,
1992). This project argues that specific attitudes, norms and self efficacy beliefs are all
prototypical for a social identity as well and as such, the inclusion of social identity as a
background variable will add power to the predictive ability of these variables. Thus, in

48

addition to having an independent effect on behavior, social identity should also impact
behavior via its effect on the intermediate variables of attitudes, norms and self efficacy.
This project hypothesizes the following:
H3: Social identity will lend predictive power to the ability of the Integrative
Model of Behavioral Prediction to predict adolescent (a) smoking susceptibility,
(b) trying smoking and (c) maintenance, such that the relationship between
attitudes, normative beliefs and self efficacy beliefs and smoking behavior will be
stronger among adolescents who identify with social identities for which smoking
is prototypical.

Social Identity and the Social Norms Approach
According to social norms research, reference groups are an extremely important
component in determining the extent to which norms predict behavior. Social Identity
Theory has been identified as a potentially useful way to explore the role of reference
groups as a factor mediating social norms and behavior (Terry and Hogg, 1996;
Yanovitzky et al., 2006). This project examines whether Social Identity Theory offers
any power to the predictive validity of social norms. The Social Norms Approach posits
that the norms of a relevant reference group will affect an individual’s behavior. This can
be linked to Social Identity Theory by thinking of group norms as group prototypes and
the reference group as a social category. When an individual strongly identifies with a

49

reference group or a social category, he or she is more likely to adhere to the requisite
norms or prototype.
Although research in this area is limited, there have been a few attempts to link
Social Identity Theory to the Social Norms Approach. Terry, Hogg and White (1999)
found that the inclusion of social identity increased the ability of social norms to predict
behavior. In addition to finding an independent effect of social identity on behavior, the
researchers also found an interaction effect for social identity with social norms, whereby
the recycling norm of a relevant group was related to behavior, but only for those who
strongly identified with the group. This project argues that social identity functions in a
way similar to that found by Terry, Hogg and White (1999) to predict adolescent
smoking behavior. Each social identity has prototypical, or normative, beliefs, attitudes
and behaviors. An individual who has a social identity (such as the rebel) for which
smoking is seen as prototypical and normative may be more likely to adhere to that norm
and consequently engage in tobacco use. Alternately, an individual who has a social
identity for which tobacco abstinence is the norm may be more likely to abstain from
smoking. So, among those who have a strong social identity as part of a reference group,
that group’s social norms should be effective in predicting behavior, whereas we would
not expect to see that same relationship between norms and behavior in those who have a
weak social identity as part of a particular group. Thus, the proposed research project
hypothesizes:

50

H4: Social identity will interact with social norms to affect adolescent smoking
(a) susceptibility, (b) uptake and (c) maintenance such that the social norms of a
relevant group will only predict behavior among those adolescents who strongly
identify with the group.

The Role of Social Identity in the truth™ Campaign
Evaluations of televised anti-smoking campaigns indicate that these campaigns
have significant yet small effects (Goldman & Glantz, 1998; Snyder et al., 2004). Those
PSAs that tend to be most effective tend to focus on the effects of secondhand smoke or
on industry manipulation — that is, ads that inform viewers that tobacco companies are
‘tricking’ them into smoking (Farrelly, Healton, Davis, Messeri, Hersey and Haviland,
2002; Farrelly, Davis, Haviland, Messeri & Healton, 2005; Goldman & Glantz, 1998).
One campaign whose success seems to have exceeded the others is the truth™ campaign
which features ‘edgy’ adolescents staging public displays in which they expose tobacco
company lies (Farrelly et al., 2002). This campaign has been attributed with reducing
smoking behavior and increasing anti-tobacco attitudes in adolescents (Farrelly, Healton,
Davis, Messeri, Hersey and Haviland, 2002; Farrelly, Davis, Haviland, Messeri &
Healton, 2005).
Kleine, Kleine and Kernan (1993) argue that media teach individuals how to
perform identities. According to Social Identity Theory, it can be argued that the truth™
campaign is effective in part because it teaches individuals that not smoking is

51

prototypical of desirable social identities (‘edgy’ and ‘cutting-edge’, in the case of truth™
(Farrelly et al., 2002)). The truth™ campaign can be said to ascribe not smoking and
rebelling against smoking as prototypical for these identities. It is hypothesized that anti-
smoking ads, such as truth™, which target specific social identities, will see increased
effects among those who identify as such.
H5: Social identity will interact with exposure to the truth™ campaign to
affect adolescent smoking behavior, such that those who have seen the truth™
campaign and who strongly identify with the social categories presented in the
truth™ ads will be less likely to engage in smoking behavior.

This dissertation uses peer crowd affiliation research as a point of entry to
determine the validity and usefulness of social identity as a construct to understand
adolescent smoking behavior. The following chapters detail the methods, findings and
implications of this research.

52

CHAPTER 3:
SURVEY METHODOLOGY AND MEASURES

In order to determine whether social identity affected smoking behavior, an online
survey was conducted. This survey sampled 248 adolescents and took place during the
Fall of 2008. The current section details the sampling procedure, sample characteristics
and survey measures.
Sampling Procedure
The study sample was obtained using a recruitment firm (Authentic Response)
who specialized in recruitment for online surveys. The online survey modality was
determined to be acceptable given the fact that 93% of teenagers have internet access
(Lenhart, Arafeh, Smith & Macgil, 2008); however the researcher is cognizant of the fact
that by not incorporating the 7% of teenagers who are not online, the sample may not be
fully representative.
The sample contained 248 14-15 year olds, randomly selected from a panel of
153,759 adolescents aged 13-17. This panel was demographically similar to the U.S.
population of 14 and 15 year olds in terms of ethnicity, family income, parents’ education
level. Parental consent to participate in online surveys was already obtained by the
recruitment firm. Two hundred and fifty panelists were randomly selected and e-mailed
an invitation to participate in the survey, which elicited a final sample of 248
respondents. Panelists who were interested in participating could click a link directing

53

them to the survey homepage, where they read an information sheet and had to check a
box agreeing to consent before being taken to the actual survey. Those who chose to
participate were compensated two dollars, regardless of whether they completed the
survey. Because the first wave of recruitment did not elicit the desired number of
responses, an additional wave of emails was sent out to a second set of randomly selected
panelists. The response rate was calculated using the method proposed by the American
Association of Public Opinion Research for use in telephone surveys. Using this method,
the response rate was 8.7%. However, it must be noted that the survey closed once 250
responses were collected and this calculation of response rate did not account for those
who attempted to take the survey after it closed.
Ultimately, this procedure elicited 252 responses. Twenty-five of these
participants indicated they were 12, 13, 16, 17 or 18 years old. Because this study was
evaluating only 14 and 15 year olds, these individuals were not included in the analysis.
Additionally, three participants were not included in the analysis because they
discontinued the survey before answering the items on smoking behavior or gave
responses that brought the validity of their response set into question (for example, by
writing “I’m a dog” in for their ethnicity). Thus, the final sample size was 224.

Sample Characteristics
Participants were asked about demographic factors previously found to have an
impact on smoking behavior. These variables were grade in school, age (in order to

54

ensure participants met the sampling criterion of being 14 or 15 years old), gender
(Coogan et al., 1998) and ethnicity (Asbridge, Tanner & Wortley, 2005; Coogan et al.,
1998; Ellickson, Perlman and Klein, 2003; Johnson, Myers, Webber & Boris, 2004). For
a complete breakdown of sample characteristics on these measures, please refer to Table
3.1 (for frequencies) and Table 3.2 (for means and standard deviations).
Age
The sampling procedure was designed to recruit 14 and 15 year olds. This age
group was chosen for study because it is an age where adolescents are trying cigarettes
for the first time (Johnston, O’Malley, Bachman & Schulenberg, 2008) and an age where
adolescents begin to try on different social identities (Erikson, 1968; Marcia, 1966).
Several participants responded that they were younger than 14 or older than 15.
Specifically, of 248 respondents, 47.6% indicated they were 14, 42.7% indicated they
were 15. .4% (N = 1) indicated they were 12, 7.3% (N =18) indicated they were 13, .8%
(N=2) indicated they were 16, .4% (N=1) indicated they were 17 and .4% (N=1) indicated
they were 18 and .4% (N=1) declined to state their age. Because of this, only 14 and 15
year olds were included in the analysis. This left a final sample of N=224. Of this
sample, 52.7% (N=118) were 14 year olds and 47.3% (N=106) were 15 year olds.
Grade
Most of the sample was in 9
th
(50.9%, N=114) or 10
th
(33.5%, N=75) grade.
12.1% (N=27) were in 8
th
grade, 2.2% (N=5) were in 11
th
grade and 1.3% (N=3) were in
12
th
grade.

55

Gender
Seventy-one percent (N=159) of the sample was female and 29% (N=65) was
male.
Ethnicity
Respondents were asked to indicate which ethnicity(ies) they described
themselves as. Respondents could check multiple categories. Eight percent (N=18)
identified as American Indian or Alaskan Native, 2.2% (N=5) identified as Asian, 12.9%
(N=29) identified as Black or African American, 16.5% (N=37) identified as Hispanic or
Latino, 1.8% (N=4) identified as Middle Eastern, .9% (N=2) identified as Native
Hawaiian or Other Pacific Islander, 72% (N=162) identified as white and 5.8% (N=13)
identified as other.

Measures
In addition to measuring age, grade, gender and ethnicity, the following were also
measured as control variables. All of the measures used as control variables were
selected based on previous research suggesting a relationship to smoking behavior.
Academic Achievement
Academic achievement has been found to affect adolescent smoking behavior
such that individuals who perform better in school are less likely to smoke (Bryant,
Schulenberg, Bachman, O’Malley and Johnston, 2000; Bryant, Schulenberg, O’Malley,
Bachman & Johnston, 2003; Ellickson, Perlman and Klein, 2003). To measure academic

56

achievement, respondents were asked to report the grades they usually got in school,
ranging from A+, A, A-, B+, B, B- down to F or below. There were thus 13 levels of
grades one could report receiving, with A+ scored highest at 13 and F or below scored
lowest at 1.
Religiosity
Increased levels of religiosity have been found to be related to lower levels of
smoking (Ellickson, Perlman and Klein, 2003). To measure religiosity, respondents were
asked how often they attended religious services. Respondents were given seven options
from which to choose, with 7 being most frequent attendance and 1 being least frequent.
These options were “Every day” (scored as 7), “More than once a week,” “Once a week,”
“A few times a month,” “About once a month,” “A few times a year” and “Never”
(scored as 1).
Smoking in the Home
Research has shown that living with someone who smokes has a profound effect
on the likelihood an adolescent will also smoke (Bricker, Peterson, Sarason, Andersen &
Rajan, 2007; Johnson, Myers, Webber & Boris, 2004; Leatherdale, et al., 2005; Taylor,
Conard, O’Byrne, Haddock & Poston, 2004). To measure this, participants were asked
whether they lived with anyone who smokes. Response options were “Yes,” “No” and “I
don’t know.” Individuals who answered “I don’t know” were collapsed into the “No”
category.

57

Descriptive Norms
Participants were given four items asking them to indicate what percentage of
their friends and their peers have “ever tried smoking” and “have smoked every day for
the last 30 days.” Response options were every fifth percent from 0%, 5%, 10% through
100%.
Injunctive Norms
Participants were given two items asking them to indicate on a seven-point scale
the extent to which they agreed or disagreed with statements that their friends would
approve of them smoking cigarettes and that their family would approve of their smoking
cigarettes. The response option of “strongly disagree” was coded as 1 and the response
option of “strongly agree” was coded as 7.
Sensation Seeking
Sensation seeking was measured using the four-item Brief Sensation Seeking
Scale (BSSS-4) validated by Stephenson, Hoyle, Palmgreen and Slater (2003). The scale
consisted of the following items: (a) I would like to explore strange places; (b) I like to
do frightening things; (c) I like new and exciting experiences, even if I have to break the
rules; and (d) I prefer friends who are exciting and unpredictable. Response options
ranged on a 7-point scale from 1 for “strongly disagree” to 7 for “strongly agree.” Higher
numbers indicate greater levels of sensation seeking. Factor analysis indicated that all
items loaded on one factor. Responses were then added and divided by 4. Scale
reliability was sufficient (alpha = .882).

58

Attitudes about Not Smoking
Attitude items were obtained from the Legacy Media Tracking Survey, a survey
developed by the American Legacy Foundation to evaluate the effectiveness of the
truth™ campaign (see Farrelly, Healton, Davis, Messeri, Hersey & Haviland, 2002;
Niederdeppe, Farrelly & Haviland, 2004). The items measuring attitudes about not
smoking were asked to individuals who indicated they had never tried smoking.
Participants were asked to indicate how much they disagreed or agreed (on a 7-point
scale) with four items about their decision not to smoke. These items were “I don’t want
to smoke because it would mean cigarette companies are using me”; “I don’t want to
smoke because it doesn’t fit my image”; “Smoke from other people’s cigarettes bothers
me”; and “I don’t want to smoke because it would make me less attractive.” These items
were then summed and divided by 4 to create one measure. Reliability for this scale was
alpha = .735. Higher scores on this scale indicate more positive feelings about not
smoking.
Attitudes About Cigarettes
Attitudes about cigarettes were measured by asking individuals to indicate how
much they disagreed or agreed (on a 7-point scale) with ten statements about cigarettes.
First, these items were summed and divided by 10; however, scale reliability was
unsatisfactory using the standards recommended by Nunnally and Bernstein (1994)
(alpha = .663). A factor analysis (principal component analysis, with a Varimax rotation)
revealed that the 10 items were primarily loading on two factors. The first factor

59

consisted of six items that were more related to the social aspects of smoking. The items
on the first factor were “Not smoking is a way to express your independence,” “Smoking
is a way to express your independence,” “Smoking cigarettes makes people your age look
cool,” “Smoking cigarettes can help keep your weight down,” “Nonsmokers don't like to
date someone who smokes” and “Smoking cigarettes makes people your age fit in.” The
second factor consisted of four items that tended to relate to health aspects of smoking.
These items were “The smoke from other people's cigarettes is harmful to you,” “People
your age who smoke cigarettes have more friends than people who don't smoke,” “It is
safe to smoke for only a year or two, as long as you quit after that” and “People who
smoke regularly have a much harder time keeping up in sports and athletic activities.”
New scales were then created by summing the items from the first factor and dividing by
six and summing the items from the second factor and dividing by four. The two
resulting scales had satisfactory reliability (alpha = .855 and alpha = .718, respectively).
For each of these scales, higher numbers indicate more positive attitudes about smoking.
Attitudes About Cigarette Companies
Attitudes about cigarette companies were measured by asking individuals to
indicate how much they disagreed or agreed (on a 7-point scale) with 13 statements about
cigarette companies and their practices. Factor analysis found these items loaded on two
separate factors. The first factor was more related to cigarette company practices and
contained the items “Cigarette companies lie,” “Cigarette companies deny that cigarettes
cause cancer,” “Cigarette companies deny that cigarettes are addictive,” “I would like to

60

see cigarette companies go out of business,” “The people who run cigarette companies
know what they are doing is wrong,” “Cigarette companies target teens to replace
smokers who die,” “Cigarette companies try to get young people to start smoking,” and
“Cigarette companies target minority groups.” The second factor was more related to
cigarette company rights and contained the items “Cigarette companies should have the
same right to sell cigarettes as other companies have to sell their product,” “Cigarette
companies get too much blame for young people smoking,” “Anti-smoking
advertisements are no more honest than cigarette advertisements.” “Cigarette companies
should have the same right to make money as other companies” and “The government
should let companies sell whatever they want.” Reliability for both factors was sufficient
(Cronbach’s alpha = .895 and Cronbach’s alpha = .835, respectively). The items on each
factor were then averaged to create two measures of attitudes toward cigarette companies.
Exposure to the truth™ Campaign
Participants were first asked to indicate whether they had ever seen any
campaigns about not starting to smoke, about quitting smoking, messages against
smoking, messages about or against cigarette companies, about not using alcohol, about
not drinking and driving, about not using chewing tobacco or about not using drugs.
Participants could give multiple responses if they had seen more than one campaign.
Participants who indicated they had seen a campaign about “not starting to smoke,”
“messages against smoking,” “messages about or against cigarette companies” and
“quitting smoking” were then probed as to whether they had seen these campaigns in the

61

past 30 days. Participants who indicated they had seen an anti-smoking campaign in the
past 30 days were then asked to select the theme or slogan of this campaign. The choices
given were “Think. Don’t Smoke,” “Truth,” “Tobacco is whacko,” “Become an ex,”
Tobacco vs. Kids: Where America Draws the Line” or they could write in the slogan if it
was not included in this list of options.
To further probe exposure to the truth™ campaign, participants were then asked
to indicate whether they were aware of any other campaign against smoking that was
currently taking place. Those who answered yes were asked to choose or write in the
slogan of the particular campaign. The response options for this question were the same
as in the first probe.
There were a total of 85 (37.9%) individuals who had been exposed to the truth
campaign. Of those, 91.8% had seen the truth™ campaign in the past 30 days. Because
of the relatively small number of respondents who had seen the truth™ campaign but not
in the past 30 days, exposure to the truth™ campaign was treated dichotomously. All
those who indicated in either probe that they had seen the truth™ campaign were counted
as having been exposed to the campaign; all those who had not were treated as not having
been exposed to the campaign.

Table 3.1. Sample Characteristics – Frequencies .
% (N) .
Age
14 years 52.7% (118)
15 years 47.3% (106)

62

Table 3.1. Sample Characteristics – Frequencies (continued) .
% (N) .
Grade
8
th
grade 12.1% (27)
9
th
grade 50.9% (114)
10
th
grade 33.5% (75)
11
th
grade 2.2% (5)
12
th
grade 1.3% (3)

Ethnicity
American Indian/Alaskan Native 8% (18)
Asian 2.2% (5)
Black/African American 12.9% (29)
Hispanic/Latino 16.5% (37)
Middle Eastern 1.8% (4)
Native Hawaiian/Pacific Islander 0.9% (2)
White 72% (162)
Other 5.8% (13)

Academic achievement
A- through A+ 41.9% (93)
B- through B+ 37.8% (84)
C- through C+ 16.2% (36)
D- through D+ 1.8% (4)
F or below 2.3% (5)

Church attendance
Never 29.9% (67)
Few times a year 23.2% (52)
About once a month 8.9% (20)
A few times a month 9.8% (22)
Once a week 18.3% (41)
More than once a week 9.4% (21)
Every day 0.4% (1)

Live with smoker
Yes 43% (97)
No 50.9% (114)
Don’t know 5.8% (13)

Exposure to truth™ campaign

63

Table 3.1. Sample Characteristics – Frequencies (continued) .
% (N) .
Yes 37.9% (85)
No 62.1% (139)

Table 3.2. Sample Characteristics – Means .
Mean (SD)
Media Use
TV use in hours 4.40 (3.21)
Radio use in hours 4.81 (4.18)
Internet use in hours 5.46 (2.86)

Social Norms
% Friends – tried smoking 29.38 (33.9)
% Friends – every day for 30 days 13.34 (23.47)
% People your age – tried smoking 56.79 (32.16)
% People your age – every day for 30 days 39.08 (31.13)

Sensation seeking scale 4.59 (1.68)
I would like to explore strange places. 4.76 (2.00)
I like new and exciting things, even if I have to break the rules. 4.51 (1.96)
I like to do frightening things. 4.26 (1.95)
I prefer friends who are exciting and unpredictable. 4.85 (1.873)

Attitude toward smoking – social 2.45 (1.258)
Not smoking is a way to express your independence. 2.08 (1.487)
Smoking is a way to express your independence. 5.06 (1.722)
Smoking cigarettes makes people your age look cool. 5.72 (1.644)
Smoking cigarettes can help keep your weight down. 5.43 (1.652)
Nonsmokers don’t like to date someone who smokes. 2.58 (1.815)
Smoking cigarettes makes people your age fit in. 5.75 (1.567)

Attitude toward smoking – health 2.852 (1.385)
People your age who smoke have more friends than people 3.09 (1.93)
who don’t smoke.
The smoke from other people’s cigarettes is harmful to 5.51 (1.707)
your health.

64

Table 3.2. Sample Characteristics – Means (continued) .
Mean (SD)
It is safe to smoke for only a year or two, as long as 2.31 (1.819)
you quit after that.
People who smoke regularly have a much harder time 4.49 (1.865)
keeping up in sports and athletic activities.

Attitude toward cigarette companies – rights 3.454 (.909)
Cigarette companies should have the same right to sell 3.13 (1.888)
cigarettes as other companies.
Cigarette companies get too much blame for young 3.09 (1.662)
people smoking.
Anti-smoking advertisements are no more honest than 3.59 (1.657)
cigarette ads.
Cigarette companies should have the same right to make money 4.84 (1.704)
as any other type of company.
The government should let companies sell whatever they want. 4.57 (1.758)

Attitude toward cigarette companies – practices 3.388 (1.399)
The people who run cigarette companies know what they 3.42 (1.96)
are doing is wrong.
Cigarette companies lie. 5.24 (1.758)
I would like to see cigarette companies go out of business. 4.54 (1.922)
Cigarette companies target teens to replace smokers who died. 4.73 (1.866)
Cigarette companies deny that cigarettes are addictive. 5.02 (1.846)
Cigarette companies deny that cigarettes cause cancer and 5.29 (1.888)
other harmful diseases.
Cigarette companies try to get young people to smoke. 3.49 (1.779)
Cigarette companies target minority groups. 3.45 (1.806)

Smoking Behavior
Smoking behavior was measured using adolescent self-report. Biochemical
testing, such as measuring saliva content of the chemicals cotinine or thiocyanate was not
used to verify self-report due to cost restraints and the nature of the online survey.
Although biochemical validation of self-report tends to be the most thorough measure of

65

smoking behavior, meta-analysis has found adolescent self-report of their own smoking
behavior to be valid (Brener, Billy & Grady, 2003).
Smoking susceptibility. Smoking susceptibility is defined as the perceived
likelihood of trying smoking. Participants were asked if they would smoke a cigarette if
their best friend offered it to them and if they thought they will smoke a cigarette at any
time in the next year. As per the work of Pierce and colleagues (Pierce, Choi, Gilpin,
Farkas and Berry, 1998; Pierce, Choi, Gilpin, Farkas & Merritt, 1996), respondents were
classified as non-susceptible if they answered ‘probably not’ or ‘definitely not’ to both
these questions. All other responses were classified as susceptible.
Smoking initiation. Participants were asked if they had ever tried a cigarette, even
just one of two puffs (Pierce, Choi, Gilpin, Farkas and Berry, 1998; Pierce, Choi, Gilpin,
Farkas & Merritt, 1996). This variable was treated dichotomously: those who answered
‘no’ were considered never smokers and those who answered ‘yes’ were treated as
having tried smoking and were asked one more question about their smoking behavior.
Established smoking. To determine the extent to which participants were
established smokers, they were asked if they have smoked at least 100 cigarettes in their
life (Pierce, Choi, Gilpin, Farkas and Berry, 1998; Pierce, Choi, Gilpin, Farkas & Merritt,
1996). Participants who answered no, but indicated they had tried smoking were
categorized as having tried smoking. Participants who answered ‘yes’ were categorized
as established smokers.

66

Smoking status. These measures detailed above were collapsed to form one four-
level variable measuring smoking status. This variable consists of four categories: (1)
never tried, not susceptible; (2) never tried, susceptible; (3) ever tried, not established and
(4) ever tried, established. To reiterate, respondents who indicated they had never
smoked and that they did not plan on smoking in the next year were categorized as “Not
susceptible.” Respondents who indicated they had never smoked but might try smoking
in the next year were categorized as “Susceptible.” Those who indicated they had tried
smoking, but had not smoked more than 100 cigarettes in their life were categorized as
“Experimental Smokers.” Those who indicated they had tried smoking and had smoked
more than 100 cigarettes in their life were categorized as “Established Smokers.” Table
3.3 provides frequencies for each category.

Table 3.3. Smoking Status - Frequencies .
Status Percentage N
Not susceptible 57.5% 143
Susceptible 11.3% 28
Experimental smoker 24.6% 61
Established smoker 6.5% 16

Social Identity
Conceptualizing social identity as a variable. Currently, there is no standardized
measure of social identity or of peer crowd affiliation. Bagozzi and Lee (2002) asked
participants to think of the group of friends with whom they ate lunch. Participants were
then asked to indicate how much their self-image overlapped with that of these friends

67

using both a visual scale of eight circles overlapping) as well as a 7-point Likert scale
anchored with the points ‘very much’ and ‘not at all.’ Participants were also asked how
attached they were to the group and how strong their feelings of belonging to the group
were. Along similar lines, Terry and Hogg (1996) used a 6-point scale that measured the
extent to which individuals identified with and had feelings of belonging to their “friends
and peers at university” (pp. 782). Bat-Chava (1994) used two items measuring the extent
to which individuals were involved with activities unique to a social identity and the
extent to which individuals preferred most of their friends to be from/part of that social
identity. Bond and Hewstone (1988) used three items measuring how important a social
identity was to the way participants viewed themselves, how much they preferred their
group to others and how satisfied they were with the group status quo.
As evident from the brief review above, social identity has been conceptualized
and operationalized in a variety of ways, many of which seem to be incompatible, make
comparisons between studies difficult (Jackson & Smith, 1999). In addition to the
conceptual ambiguity of these operationalizations, the existing measures of social identity
are potentially inadequate due to their neglect of group prototype. For instance, Terry
and Hogg (1996) hypothesized that identification with “friends and peers” would
correlate with use of sunscreen without assessing the extent to which using sunscreen was
prototypical of these groups. The researchers simply used injunctive norms (e.g. “My
friends and peers would approve of me using sunscreen”) as a proxy for prototypicality

68

and did not truly assess the extent to which the group “friends and peers” was diagnostic
for using sunscreen.
Attempts to operationalize peer crowd affiliation are similarly varied. Sussman,
Unger and Dent (2004), for instance, presented adolescents with a list of peer crowds
such as “burn-outs,” “popular,” “surfers,” “don’t have a best friend” and “goofies.” and
asked them to circle the one they feel they are most a part. Downs and Rose (1991) used
a similar method, but first asked adolescents to identify the groups in their school and
then circle the one they fit in with. La Greca, Prinstein and Fetter (2002) developed the
Peer Crowd Questionnaire using a multi-step methodology. First, the researchers looked
to previous research as well as to focus groups of local high school graduates for a list of
names and descriptions of crowds. Then, study participants were given this list and asked
to verify that they existed in their school and to name any other crowds. Participants
were then asked to indicate which crowd they most identified with and to indicate how
long they had been a member of that particular crowd. Finally, to assess friends' peer
crowd affiliations, adolescents were asked to name up to three of their very best friends;
for each friend named, adolescents were asked to identify that friend's peer crowd
affiliation. This procedure elicited five peer crowds and each respondent was categorized
as a member of only one crowd.
In general, these measures fall short in three ways. First, measures of peer crowd
affiliation do not account for adolescents who may identify with more than one group.
Social identity theory posits that individuals possess multiple social identities and it is

69

quite possible for an adolescent to identify equally strongly with two or more peer
crowds. Additionally, these measures do not account for the wide range of social
categories that exist, opting instead to focus only on “elites,” “athletes,” “deviants,”
“academics” and “others”
2
. For example, while Sussman and colleagues asked
respondents to choose from a list of 16 peer crowds, these groups were ultimately
collapsed into five groups. While these five groups may encompass the majority of
adolescents and explain some of the between-group variance in smoking behavior, they
do not capture the more specific social categories into which adolescent identify. Finally,
existing measures of social identity and peer crowd affiliation do not account for group
prototypes and group diagnosticity. According to Social Identity Theory, a social identity
should predict only those behaviors that are prototypical for that specific social category.
If one wants to understand how social identity affects smoking behavior, it is essential to
first know for which social identities smoking and not smoking are prototypical.
Measuring Strength of Identification with Social Category (SISC) Given the
variation and ambiguity surrounding operationalizations of social identity, it is crucial to
be explicit about what, exactly, the measure used in this project is actually measuring.
The proposed measure uses as a starting point Turner’s (1999) definition of social
identity: “a self definition in terms of social category membership” (p. 10). To possess a
social identity, an individual must first recognize that a social category exists, actively

2
It should be noted that La Greca, Prinstein and Fetter (2002) used adolescents to help generate peer crowd
names, they ultimately only used six categories, including a non-descript other category.

70

identify as belonging to that social category and possess a prototype for that social
category. In this framework, social identity is a property of the individual – an
individual’s perception that he or she belongs to a social category with a corresponding
prototype. Group prototypes are similarly properties of the individual. Group prototypes
are defined as individual cognitive representations of the social category’s defining
values, behaviors, norms and so on. Group prototypes are also properties of the
individual. However, although prototypes are ultimately individual perceptions of what
is normative for a social category, there is typically general consensus among members
about what constitutes a social category’s prototype. Thus, one individual’s group
prototype tends to be shared by others.
The measure developed for this study integrates measures used both by social
identity researchers and by peer crowd affiliation researchers. Although social identity
varies on several dimensions, such as fit and accessibility (Hogg, 2006; Hogg & Reid,
2006), this measure looks only at strength of identification. Following the work of
Terry and Hogg (1996) this measure operationalizes strength of identification with a
social category (SISC) as the key factor by which social identities vary within the
individual. This concept is integrated with work by Sussman and colleagues (Sussman et
al., 1999; Sussman et al., 2000; Sussman, Unger and Dent, 2004) who use peer crowds at
school as a frame, contextualizing the social categories within the school environment by
referencing “groups at school.” As such, this study measures one dimension of social

71

identity: the extent to which an individual identifies with social categories at school for
which smoking or not smoking are prototypical.
The first step in operationalizing this variable was to conduct a pre-test designed
to elicit the social categories for which smoking and not smoking were prototypical. This
pre-test was conducted on 61 college students at a large southwestern university.
Participants were asked to think about the typical smoker and the typical non-smoker.
They were then asked to indicate the social categories to which they felt these people
belonged. Finally, participants were asked to watch three truth™ campaign PSAs and
indicate the social categories to which they thought the people in the ads belonged. All
responses were open-ended. The ads that respondents watched were titled “Body Bags,”
“Lie Detector” and “The Magical Amount.” The first ad was a public display of body
bags representing people who had died from tobacco. The second ad featured young
adults at a tobacco company trying to get the marketing department to take a lie detector
test. The third ad featured two young men singing satirically about the nicotine in
cigarettes.
A total of 143 responses were given for ‘not smoking.’ A total of 166 responses
were given for ‘smoking.’ A total of 372 responses were given for the truth™ campaign
ads. These responses were coded and collapsed into categories. For example, nerds,
geeks and dorks were collapsed into ‘Nerd.’ Responses that were deemed to represent
individual characteristics (e.g. “beautiful”) or moods (e.g. “angry”) were not included.
This process resulted in a list of 25 social categories (see Table 3.4). Although not

72

specifically mentioned in the pre-test, Techie was added to this list as a variable of
interest.
Table 3.4. Frequencies - Social categories .
truth™
Name of social category Not smoking Smoking campaign
Artist 0 8 2
Athlete 30 7 11
Average 4 0 3
Band (Marching band) 3 0 4
Class clown 0 0 3
Emo 0 9 1
Gamer 0 0 1
Goody-goody 5 0 1
Goth 0 5 0
Hippie 1 2 1
Hipster 0 1 3
Involved 7 0 7
Misfit 0 9 1
Musician 0 2 1
Nerd 24 0 10
Non-conformist 5 1 9
Partier 0 15 0
Popular 3 11 6
Preppy 4 0 5
Rebel 1 10 7
Religious 3 0 0
Skater 0 3 2
Smart 17 1 17
Straightedge 3 0 0
Techie -- -- --
Theater/Drama 0 8 9

This list of social categories was then used in the survey to measure strength of
identification with each social category (SISC). The procedure was based on those used
by Sussman and colleagues (Sussman, Unger and Dent, 2004) who asked adolescents to

73

circle the group they best fit in with and La Greca, Prinstein and Fetter’s (2002) Peer
Crowd Questionnaire, which asked adolescents to circle the group with which they most
identified. Participants were asked to indicate on a sliding scale from 0 – 100 how much
they identified with each of the 26 groups identified in pre-testing. Participants dragged a
bar across the screen for each social category, which allowed them to adjust and compare
across categories, and make changes accordingly. Specifically, individuals were asked
the following: “People often hang out in different groups at school. For example, a lot of
schools have a group of “jocks.” Some students gave the following list of groups. Please
indicate how much you identify with each group by dragging the bar across the screen.
Dragging the bar to 100 means you identify with this group very much and dragging the
bar to 0 means you do not identify with this group at all.” This measure allowed
individuals to identify with more than one social category and to indicate varying levels
of identification. See Figure 3.1 for an example of one individual’s social identity make-
up.

74

Figure 3.1. Example of an individual’s SISC composition

Next, scores were standardized within individual. This was done to obtain an
accurate indication of the extent to which an individual identified as one social category
in relation to all others. For example, one individual’s scores may cluster high while

75

another’s may cluster low. Standardizing scores within individual accounts for the fact
that, to the individual who clusters high, a score of 60 may mean relatively little fit,
whereas to the individual who clusters low, a score of 60 may mean relatively high fit.
To illustrate the relation between the unstandardized and standardized SISC scores, take
the case of an individual whose unstandardized average of SISC items was 50, with a
standard deviation of 25. An item with an unstandardized score of 75 would be translated
into a standardized score of 1, indicating that item was rated one standard deviation
above the mean for that individual. An item with an unstandardized score of 25 would be
translated into a standardized score of -1, indicating that item was rated one standard
deviation below the mean for that individual. Table 3.5 provides the unstandardized and
standardized means of each of the 26 SISC items. The unstandardized mean is the
original average score, which ranged from 0-100. The standardized mean is the mean for
each item after standardizing scores within the individual.

Table 3.5. Social Identity Unstandardized and Standardized Scores
Unstandardized Standardized
Social Category Mean Median Mean Median
Artist 38.04 34.50 -.08 -.24
Athlete 48.58 50.00 .27 .32
Emo 32.89 16.00 -.24 -.57
Involved 47.24 49.50 .23 .20
Musician 47.52 47.00 .24 .23
Popular 54.64 56.00 .47 .64
Preppy 43.10 40.50 .10 .01
Band 30.86 20.00 -.28 -.56
Goth 25.95 10.00 -.49 -.76
Techie 23.70 12.50 -.52 -.72
Nerd 38.77 36.00 -.46 -.68

76

Table 3.5 (cont.). Social Identity Unstandardized and Standardized Scores
Straightedge 35.31 31.00 -.06 -.27
Smart 56.89 57.00 .55 .59
Class clown 55.20 59.00 .48 .63
Misfit 40.06 35.50 .00 -.20
Gamer 35.44 30.50 -.12 -.28
Skater 46.10 48.50 .21 .19
Nonconformist 29.84 19.00 -.34 -.57
Religious 31.26 20.00 -.29 -.49
Goody-goody 33.59 27.50 -.23 -.40
Hippie 23.90 10.00 -.54 -.73
Rebel 47.42 46.00 .23 .16
Partier 54.38 58.50 .44 .59
Hipster 28.96 20.50 -.35 .-.55
Average 65.87 71.00 .93 1.04

In addition to considering each of the 26 SISC variables separately, factor
analysis was used to examine the extent to which certain social identities clustered
together. Exploratory factor analysis was run on the 26 SISC variables using maximum
likelihood extraction for its ability to compute a variety of goodness of fit indexes (as
recommended by Fabrigar, Wegener, MacCallum & Strahn, 1999) and promax rotation,
given the high likelihood of inter-factor correlation (Fabrigar et al., 1999; Abdi, 2003).
The sample-to-item ratio was 8.96:1 (224 subjects to 25 items), exceeding the 5:1 ratio
recommendation of Gorsuch (1983) and Hatcher (1994), but just below the 10:1 ratio
recommended by Nunnally (1978). It is important to note that because of this, there may
be an increased chance of inflated error rates which could potentially lead to items being
statistically miscategorized (Osborne & Costello, 2004). To counter this, the researcher
took previous theory and research into consideration when finalizing which items
belonged on which factors.

77

Initial maximum likelihood estimation (MLE) direct oblique rotation exploratory
factor analysis (EFA) was conducted and revealed that the matrix was not positive
definitive. Examination of a bivariate correlation matrix of the 26 items indicated that
the items SISC-Emo and SISC-Goth
3
were highly correlated (r = .699) (see Table 3.6).

3
SISC-Emo is a social category of individuals who listen to music with extreme emotional content (usually
about heartbreak). Its members tend to dress in black, with dyed black hair. SISC-Goth is a social
category that is also defined by dark dress and a particular style of music; however Goth music tends to be
less emotional. There is a considerable amount of overlap in the musical and fashion preferences of the
two groups.

78

Table 3.6. Bivariate Correlation Matrix of SISC items

79

A second factor analysis was run using the same specifications, but with SISC-
Emo removed. This produced eight factors. The decision to retain eight factors was
informed by each factor’s Eigenvalue, as well as a scree test, as recommended by
Costello & Osborne (2005). Table 3.7 provides factor loadings and names.

Table 3.7. SISC Pattern Matrix
.
Factor
1 2 3 4 5 6 7 8
SISC-Artist -.248 .024 .241 -.253 -.048 -.125 .206 -.137
SISC-Athlete .626 -.012 -.040 .059 -.042 .016 -.197 -.265
SISC-Average -.041 .056 -.047 .341 .042 -.099 -.204 -.187
SISC-Rebel -.214 -.539 -.238 -.066 -.060 -.029 .036 .075
SISC-Band .041 .317 -.231 .171 -.077 -.158 .413 -.085
SISC-Class clown .295 -.494 .029 -.155 -.021 .056 .095 -.116
SISC-Gamer -.082 .072 -.035 .015 -.049 .975 -.075 .032
SISC-Goody-goody .010 .112 .077 .545 .034 -.030 -.068 .193
SISC-Goth -.319 .199 -.119 -.512 -.054 -.104 -.107 .033
SISC-Hippie -.108 .109 -.083 .004 -.056 .049 -.073 .506
SISC-Hipster .219 -.197 -.108 .026 -.006 -.067 -.081 .362
SISC-Involved .264 .253 .324 .055 -.214 -.064 .042 .113
SISC-Misfit -.719 .081 -.168 -.139 -.096 -.067 -.187 -.068
SISC-Musician .070 .233 -.193 -.049 .051 .092 .552 -.039
SISC-Nerd -.130 .349 .342 -.077 .019 -.087 -.062 .108
SISC-Nonconformist -.474 .098 .037 -.082 .015 .106 -.170 -.107
SISC-Partier .146 -.663 -.011 -.047 -.021 -.093 -.044 .060
SISC-Popular .742 -.114 -.061 -.177 -.072 -.035 -.086 -.122
SISC-Preppy .555 .091 -.081 .076 .003 -.106 -.305 .084
SISC-Religious .009 .096 .058 .473 -.109 .030 .050 -.136
SISC-Skater -.246 -.181 -.494 .122 -.087 .012 .036 .053
SISC-Smart -.212 -.085 .805 .229 -.036 .010 -.038 -.123
SISC-Straightedge -.017 .062 -.009 -.004 .996 -.050 .013 -.055
SISC-Techie .035 .358 .089 -.209 .026 .111 .037 .116
SISC-Theater -.078 -.155 .094 -.021 .003 -.103 .477 -.035
Extraction Method: Maximum Likelihood.
Rotation Method: Promax with Kaiser Normalization.
a Rotation converged in 13 iterations.

80

Eight final factors were decided upon: Elites, Outsiders, Achievers,
Conservatives, Substance-free, Individualists, Musical Arts and Cultural Rebels (see
Table 3.8). Although SISC-Skaters had the highest loading on the ‘Goody-goodies’
factor, it was ultimately included in the ‘Individualists’ factor as previous research has
shown individuals who identify with Skaters to be more similar to independents or non-
conformists (Sussman, Unger & Dent, 2004). Due to the high correlation with Goth, Emo
was not included in the factor analysis but was ultimately included in the ‘Outsiders’
factor.

Table 3.8. Final SISC factors .
Elites Outsiders Achievers Conservatives
Athlete Goth Artist Average
Partier Misfit Involved Goody-goody
Popular Emo Smart Religious
Preppy Techie
Class clown Nerd .
Substance Free Individualists Musical Arts Cultural Rebels
Straightedge Gamer Band Hipster
Nonconformist Musician Hippie
Skater Theater Rebel

Individual scores for each of the eight final factors were calculated by summing
the standardized SISC items on each factor and dividing by the number of items. See
Table 3.9 for descriptive statistics of each factor and Table 3.10 for bivariate correlations.
Table 3.9. SISC Factor Descriptives .
Factor Mean (SD) .
Popular .328 (.660)
Misfit -.3261 (.561)
Achievers .2158 (.607)
Goody-goodies .1224 (.622)
Straightedge -.1202 (.891)

81

Non-conformists -.0820 (.557)
Musical Arts -.0250 (.593)
Cultural Rebels -.2000 (.554)

Table 3.10. SISC Factor Bivariate Correlations .
1 2 3 4 5 6 7 8
1 Involved 1
2 Cult. Rebels -.384
**
1
3 Goody-goodies .113
*
-.320
**
1
4 Misfit -.045 -.009 -.524
**
1
5 Musical Arts .139
*
-.291
**
-.067 .000 1
6 Non-conformists -.325
**
.159
**
-.333
**
.245
**
-.104 1
7 Popular -.264
**
-.002 .182
**
-.644 -.377
**
-.352
**
1
8 Straightedge -.103 -.102 -.005 -.015 -.001 -.095 -.097 1

The next chapter reports the results for Hypotheses 1 through 5. These
hypotheses were designed to test the independent effect of strength of identification with
each social category on smoking behavior; the added validity of strength of identification
with each social category to the Sensation Seeking Approach, the Integrative Model of
Behavioral Prediction and the Social Norms Approach; and the moderating effect of
strength of identification with each social category on the relationship between exposure
to the truth™ campaign and smoking outcomes.

82

CHAPTER 4: RESULTS

This chapter presents the results of the data analysis designed to test the
hypotheses that strength of identification with a social category (SISC) would predict
adolescent smoking behavior (H1a and H1b); that SISC would moderate the effects of
sensation seeking on smoking behavior (H2); that SISC would add predictive validity to a
model containing attitudes, social norms and self efficacy (H3); that SISC would
moderate the effects of social norms on smoking behavior (H4); and that SISC would
moderate the effects of exposure to the truth™ campaign on smoking behavior (H5).
These hypotheses were tested by first examining the effects size and significance
of the standardized coefficient of the variables of interest in the models. This allows for
interpretation of the individual effects of the variables of interest. Then, to determine the
extent to which each variable significantly improved the overall model fit, the fit of each
model containing that specific variable of interest was compared to the fit of the model
without it. This was done by conducting a likelihood ratio test using the likelihood
statistics of the multinomial logistic regression models with and without the variable of
interest, or by looking at the size and significance of the R
2
change value in the multiple
regression models. Results of each hypothesis are presented in turn. Any relevant
preliminary analyses, such as bivariate correlation tables, are presented first. Then,
results of hypothesis testing using each of the 26 SISC items are presented. Finally,
hypotheses are then further tested using each of the eight SISC factors. Due to the large
volume of results, only the most important aspects of the findings are detailed in the text.

83

Tables in the chapter present details of significant findings, while non-significant findings
are presented in tables located in Appendix B.

Interpreting Odds Ratios and Interaction Effects
Before delving into explanation of the results, this section will briefly discuss the
interpretation of odds ratios. Odds ratios are the statistic of interest when measuring
effects size in multinomial regression analysis, yet often are misinterpreted (Davies,
Crombie and Tavakoli, 1998). An odds ratio should be interpreted exactly as its name
implies – as a ratio of odds. Odds are calculated by dividing the probability of
experiencing an event (such as trying smoking) by the probability of not experiencing the
event. Odds ratios are then calculated by dividing the odds of the event occurring by the
odds of the event not occurring mathematically,

OR = p / (1 - p)
q / (1 - q)

where p = the probability of experiencing an event and q = the probability of not
experiencing an event. The odds ratio then, is the ratio of the odds of experiencing an
event to the odds of not experiencing the event per one unit change in an independent
variable. For example, assume the parameter estimate for the odds ratio of sensation
seeking on trying smoking was 2.0. This can be interpreted as such: a one unit change in
sensation seeking increases one’s odds of smoking by a factor of two. Additionally, it is

84

crucial to keep in mind that odds ratios increase exponentially for every unit change in
the independent variable. So a two unit change in sensation seeking would correspond to
an odds ratio of four (2
2
). A three unit change in sensation seeking would correspond to
an odds ratio of eight (2
3
). Thus, the odds ratios discussed in this dissertation are to be
interpreted as the ratio of the odds of a type of smoking behavior occurring to the odds of
being an unsusceptible non-smoker.
Interpreting interaction effects in logistic regression models can similarly be
tricky. The interaction terms used for this analysis are multiplicative, whereby one
variable was multiplied by another. Significant odds ratios for interaction terms imply
that the effectiveness of one term varies across levels of the other term. However, the
designation of a focus variable and a moderator variable is arbitrary and at the discretion
of the researcher (Jaccard, 2001). For the purposes of this dissertation, strength of
identification with the various social categories is the moderator variable while sensation
seeking, social norms and exposure to the truth™ campaign are taken as the focus
variable. In other words, these results will be interpreted as the extent to which the
effects of the focus variables (sensation seeking, social norms and exposure to the truth™
campaign) vary across levels of the moderator variable (strength of identification with
each social category). This designation is based solely on theory: the focus variables
have been previously found to be directly associated with smoking behavior.

85

H1: Evaluating the impact of social identity on smoking behavior

The purpose of this section is to test Hypothesis 1a and 1b which propose that
strength of identification with social category (SISC) will be related to (a) smoking
susceptibility (b) trying smoking and (c) maintenance. This hypothesis was tested using
multinomial logistic regression analysis. The size and significance of the coefficients for
the SISC variables were examined to evaluate the extent to which the SISC variables
predicted an increased or decreased likelihood of smoking behavior. Additionally, the
extent to which the SISC variables increased the fit of the base model was evaluated
using a likelihood ratio test.
Preliminary Analysis
A one-way analysis of variance (ANOVA) was conducted to examine the mean
differences in the 26 social identity categories by smoking status (never smoker,
susceptible smoker, experimental smoker, established smoker). The analysis found that
never smokers, susceptible smokers, experimental smokers and established smokers
varied significantly on the following social identities: Artists, Involved, Musicians,
Band, Goth, Nerds, Smart, Misfit, Skaters, Hippie, Rebels and Partiers. Susceptible
smokers had a lower average strength of identification with the Musician category.
Experimental smokers had lower average strength of identification on the Artist,
Involved, Musician, Band, Nerd and Rebel categories. However, experimental smokers
had higher average strength of identification on the Smart, Skater and Partier categories.
Established smokers had lower average strength of identification on the Musician

86

category and higher averages on the Goth, Rebel and Hippie category. Table 4.1
provides means for the 26 SISC items by smoking status. A one-way ANOVA was also
used to examine mean differences in the eight SISC factors. The analysis found that
Achievers, Cultural Rebels, Conservatives and Musical Arts varied significantly between
levels of smoking status. Adolescents who were susceptible to smoking had higher
identification with the Cultural Rebel factor and lower identification with the Musical
Arts factor. Both experimental smokers and established smokers had lower levels of
identification with the Achiever, Cultural Rebel and Musical Arts categories. This
indicates that adolescents who highly identify as Achievers and Musical Arts are less
likely to be current smokers, while adolescents who identify as Achievers, Musical Arts
and Goody-goodies are less likely to be current smokers. On the other hand, adolescents
who identify as Cultural Rebels are more likely to be current smokers. Table 4.2
provides means for the eight SISC factor items by smoking status.

Table 4.1. One-Way ANOVA: SISC Item Means by Smoking Status .
Never tried Susceptible Tried Established
Artist .0706 -.3276 -.3317
*
-.0788
Athletes .3545 .0688 .3014 -.2214
Emo -.3472 -.1753 .0234 -.2483
Involved .4584 .0527 -.1519
*
-.1587
Musicians .4776 -.1781
*
-.0279
*
-.1946
*

Popular .4401 .4976 .5116 .4963
Preppy .0505 .2235 .1401 .1162
Band -.1292 -.5067 -.4925
*
-.5246
Goth -.5613 -.5071 -.4681 .0960
*

Techies -.4863 -.5411 -.6751 -.2141
Nerds -.3460 -.3918 -.7284
*
-.6042
Theater -.0695 -.3374 -.1847 -.3385
Straightedge -.2055 .0614 -.0982 -.1439

87

Table 4.1. One-Way ANOVA: SISC Item Means by Smoking Status (continued) .
Never tried Susceptible Tried Established
Smart .7157 .5115 .2530
*
.2555
Class clown .4164 .4406 .6466 .5432
Misfit -.1241 -.0239 .2341
*
.2748
Gamers -.1328 -.1246 -.1225 -.1775
Skaters .0121 .3042 .6652
*
.1036
Nonconformist -.3670 -.2832 -.2962 -.3712
Religious -.1631 -.5588 -.3874 -.5736
Goody-goody -.1444 -.1507 -.4089 -.4664
Hippie -.6702 -.3652 -.4822 -.1282
*

Rebels -.0896 -.3372 -.7214
*
1.0451
*

Partiers .2314 .7340 .7206
*
.7787
Hipsters -.4707 -.0829 -.2497 -.0827
Average .9297 1.278 .9086 .4587 .
*
Significant at p < .05
Reference groups in italics

Table 4.2. One-Way ANOVA: Eight SISC Factor Means by Smoking Status .
Never tried Susceptible Tried Established
Achievers .3996 .0859 -.0927
*
-.0610
*

Cultural Rebels -.3921 .0169
*
-.0138
*
.4066
*

Goody-goodies .1929 .2021 .0016 -.1878
*

Misfit -.3402 -.3594 -.3294 -.1289
Musical Arts .1442 -.3899
*
-.1757
*
-.3575
*

Nonconformists -.1534 -.0439 .0820 -.1167
Popular .2627 .3971 .4527 .3278
Straightedge -.1912 .0938 -.0433 -.1340 .
*
Significant at p < .05
Reference groups in italics

Hypothesis Testing
In order to test the hypothesis that SISC predicts smoking status, a base model
not including any of the SISC variables was established. Multinomial logistic regression
analysis was used to test this base model not including any of the SISC variables. More
specifically, ethnicity (being white and being African American), achievement,

88

religiosity, living with a smoker, social norm (perceived number of friends who had
smoked everyday for 30 days), sensation seeking and exposure to the truth™ campaign
were entered into the equation). These variables were chosen based on previous research
that found them to be predictive of smoking behavior. The method of entry was forced
entry likelihood ratio as recommended by Studenmund and Cassidy (1987), Field (2005)
and Hauck and Donner (1977). Smoking status was the dependent variable. Because the
preliminary one-way ANOVA indicated that “unsusceptible” (never smoker) was the
category from which the others differed, this was used as the reference category for the
dependent variable. Power to detect an effect of .10 was .938.
Results of the analysis found that none of the variables included in the base model
predicted likelihood of being a susceptible smoker as compared to an unsusceptible never
smoker. However, being white (e
β
=.398, p < .05), achievement in school (e
β
= .804, p <
.01) and sensation seeking (e
β
=1.461, p < .01) predicted having ever tried smoking while
achievement in school (e
β
= .743, p < .01) and social norm (e
β
= 1.062, p < .001)
predicted being a current smoker. Table 4.3 presents parameter estimates for the base
model.

Table 4.3. Multinomial Logistic Regression Analysis Parameter Estimates for Base
Model Predicting Smoking Status .
Smoking Status(a) β SE β Sig. e
Β
.
Susceptible Intercept -1.289 1.494 0.388
African American = Yes -1.280 0.869 0.141 0.278
White = Yes -0.446 0.543 0.411 0.640
Female = Yes -0.156 0.487 0.749 0.855
Lives with Smoker = Yes 0.097 0.296 0.744 1.102

89

Table 4.3. Multinomial Logistic Regression Analysis Parameter Estimates for Base
Model Predicting Smoking Status (continued) .
Smoking Status(a) β SE β Sig. e
Β
.
Achievement -0.086 0.094 0.363 0.918
Religiosity 0.081 0.128 0.527 1.085
Social Norm 0.016 0.011 0.152 1.016
Sensation Seeking 0.153 0.145 0.289 1.166
truth™ Exposure 0.350 0.462 0.449 1.419
Tried Intercept -0.539 1.244 0.665
African American = Yes -1.172 0.645 0.069 0.310
White = Yes -0.921 0.432 0.033 0.398
*

Female = Yes 0.393 0.433 0.364 1.481
Lives with Smoker = Yes 0.111 0.248 0.655 1.117
Achievement -0.219 0.073 0.003 0.804
**

Religiosity -0.084 0.107 0.434 0.919
Social Norm 0.013 0.009 0.147 1.014
Sensation Seeking 0.379 0.127 0.003 1.461
**

truth™ Exposure 0.168 0.380 0.659 1.183
Current Intercept 0.552 2.579 0.830
African American = Yes -0.579 1.629 0.722 0.561
White = Yes 2.114 1.552 0.173 8.285
Female = Yes -1.120 0.776 0.149 0.326
Lives with Smoker = Yes 0.137 0.454 0.762 1.147
Achievement -0.297 0.142 0.037 0.743
*

Religiosity -0.440 0.262 0.094 0.644
Social Norm 0.060 0.014 0.000 1.062
***

Sensation Seeking -0.168 0.235 0.476 0.846
truth™ Exposure 0.914 0.751 0.224 2.494
a The reference category is: Never, not susceptible.
χ2(27) = 85.656, p < .001

Next, each of the 26 SISC items were added to the above model and tested
individually. Again, unsusceptible was the reference category for the dependent variable.
Power to detect an effect of .1 was .938. Results of these analyses indicated that strength
of identification with several social categories was significantly associated with smoking
behavior. Table 4.4 provides parameter estimates for the SISC items. Parameter

90

estimates for the full models containing the SISC items can be found in Appendix B.
Strength of identification with the social categories of Artists (e
β
= .565, p<.05),
Musicians (e
β
= .334, p< .01), Theater (e
β
= .349, p < .01) and Religious (e
β
= .499, p <
.05) were associated with a decreased likelihood of being susceptible to smoking.
Strength of identification with the categories of Average (e
β
= 1.528, p < .05), Hipsters
(e
β
= 1.969, p < .05), Partiers (e
β
= 1.825, p < .05) and Rebels (e
β
= 1.93, p < .05) were
associated with an increased likelihood of being susceptible to smoking. Strength of
identification with the categories of Artists (e
β
= .460, p < .01), Involved (e
β
= .603, p <
.05), Musicians (e
β
= .537, p < .010) and Nerds (e
β
= .533, p < .05) were associated with a
decreased likelihood of ever having tried smoking. Alternately, strength of identification
with the categories Skaters (e
β
= 1.821, p < .01), Rebels (e
β
= 2.099, p < .01) and Partiers
(e
β
= 1.511, p < .05) were associated with an increased likelihood of ever having tried
smoking. None of the 26 social categories were associated with a decreased likelihood of
being a current smoker. Finally, the categories of Hippie (e
β
= 3.372, p < .01) and Rebels
(e
β
= 4.003, p < .01) were associated with an increased likelihood of being a current
smoker.
In order to evaluate the extent to which the SISC items significantly added
predictive value to the base model, a likelihood ratio test was used, whereby a chi-square
statistic was calculated by subtracting the likelihood statistic of the model with the SISC
item from the likelihood statistic of the model without the SISC item. Results of this test
found that the inclusion of the items for Artist (χ2(1) = 26.981, p < .001), Musician (χ2(1)

91

= 31.387, p < .001), Theater (χ2(1) = 22.842, p < .001), Partier (χ2(1) = 7.098, p < .01),
Rebel (χ2(1)=15.461, p < .001), Involved (χ2(1)=20.159, p < .01), Nerd (χ2(1)=20.071, p
< .001), Skater (χ2(1) = 23.490, p < .001) and Hippie (χ2(1) = 7.671, p < .01) all
significantly improved the fit of the base model.
Strength of identification with the categories of Athletes, Emo, Popular, Preppy,
Band, Goth, Techies, Smart, Misfit, Gamers, Nonconformists, Goody-goodies, Class
clowns and Straightedge were not associated with either an increased or decreased
likelihood of being susceptible to smoking, having ever tried smoking or being a current
smoker. Tables presenting detailed parameter estimates for these models can be found in
Appendix B.
Table 4.4. Multinomial Logistic Regression Analysis Parameter Estimate for SISC Items
to Predict Smoking Status Controlling for Base Model .
Smoking Status(a) β SE β Sig. e
β
.
Susceptible Artist -0.572 0.289 0.048 0.565
*

Musician -1.097 0.331 0.001 0.334
**

Theater -1.054 0.380 0.006 0.349
**

Religious -0.695 0.332 0.036 0.499
*

Average 0.424 0.210 0.044 1.528
Hipster 0.677 0.281 0.016 1.969
*

Partier 0.601 0.252 0.017 1.825
*

Rebel 0.658 0.281 0.019 1.930
*

Involved -0.352 0.253 0.164 0.703
Nerd 0.126 0.282 0.656 1.134
Skater 0.312 0.268 0.245 1.366
Hippie 0.430 0.273 0.115 1.538
Tried Artist -0.777 0.240 0.001 0.460
**

Musician -0.622 0.228 0.006 0.537
**

Theater -0.303 0.245 0.216 0.738
Religious -0.090 0.243 0.711 0.914
Average 0.099 0.176 0.575 1.104

92

Table 4.4. Multinomial Logistic Regression Analysis Parameter Estimate for SISC Items
to Predict Smoking Status Controlling for Base Model (continued) .
Hipster 0.200 0.237 0.400 1.221
Partier 0.413 0.194 0.034 1.511
*

Rebel 0.741 0.235 0.002 2.099
**

Involved -0.507 0.205 0.014 0.603
*

Nerd -0.628 0.267 0.019 0.533
*

Skater 0.599 0.218 0.006 1.821
**

Hippie 0.268 0.234 0.252 1.307
Current Artist 0.120 0.438 0.784 1.128
Musician -0.874 0.492 0.076 0.417
Theater -0.150 0.508 0.768 0.861
Religious -0.524 0.554 0.344 0.592
Average -0.404 0.369 0.274 0.668
Hipster 0.501 0.403 0.214 1.651
Partier 0.645 0.412 0.118 1.905
Rebel 1.387 0.519 0.008 4.003
**

Involved -0.412 0.380 0.278 0.662
Nerd -0.305 0.514 0.554 0.737
Skater -0.511 0.468 0.275 0.600
Hippie 1.215 0.415 0.003 3.372
**

a The reference category is: Never, not susceptible.
*
p < .05
**
p < .01
***
p < .001

In addition to testing the SISC items, the hypothesis that strength of identification
with social categories would be related to smoking behavior was evaluated in terms of the
SISC factors. Each of the eight SISC factors was added to the base model and tested
individually. Forward stepwise likelihood ratio was the method of entry and never, not
susceptible was the reference category for the dependent variable.
These analyses found that several SISC factors were associated with smoking
behavior. Identifying as an Achiever was associated with a decreased likelihood of being
susceptible to smoking (e
β
= .426, p < .05) and of ever having tried smoking (e
β
= .240, p

93

< .001) (see Table 4.5). Identifying as a Conservative was associated with a decreased
likelihood of being a current smoker (e
β
= .181, p < .049) (see Table 4.6). Identifying
with the category of Musical Arts was associated with a decreased likelihood of ever
having tried smoking (e
β
= .392, p < .05) (see Table 4.7). Identification as a Cultural
Rebel was associated with an increased likelihood of being susceptible to smoking (e
β
=
4.766, p < .01), of ever having tried smoking (e
β
= 3.123, p < .01) and being a current
smoker (e
β
= 15.275, p < .010) (see Table 4.8). Identification with Elite, Individualist,
Substance free and Outsider were not significantly associated with smoking behavior.
Likelihood ratio testing showed that the inclusion of identification with Achievers
(χ2(1) = 30.039, p < .001), Musical Arts (χ2(1) = 34.553, p < .01) and Cultural Rebel
(χ2(1) = 287.769, p < .001) significantly improved the fit of the base model.

Table 4.5. Multinomial Logistic Regression Analysis Parameter Estimate for SISC-
Achiever Factor to Predict Smoking Status Controlling for Base Model _
Smoking Status(a) β SE β Sig. e
β

Susceptible Achievers -0.853 0.432 0.048 0.426
*

Conservative 0.174 0.402 0.666 1.190
Musical Arts -2.016 0.541 0.000 0.133
***

Cultural Rebel 1.562 0.465 0.001 4.766
**

Tried Achievers -1.426 0.380 0.000 0.240
***

Conservative -0.088 0.334 0.792 0.916
Musical Arts -0.937 0.366 0.010 0.392
*

Cultural Rebel 1.139 0.392 0.004 3.123
**

Current Achievers -0.785 0.719 0.275 0.456
Conservative -1.707 0.866 0.049 0.181
*

Musical Arts -1.238 0.761 0.103 0.290
Cultural Rebel 2.726 0.761 0.000 15.275
***

a The reference category is: Never, not susceptible.
*
p < .05
**
p < .01
***
p < .001

94

H2: Evaluating the Contribution of Strength of Identification with Each Social Category
to the Sensation Seeking Approach

The second hypothesis of this project proposes that identifying as belonging to a
social category will moderate the relationship between sensation seeking and likelihood
of being susceptible to smoking, of having ever tried smoking and of being a current
smoker. Multinomial logistic regression was used to test the extent to which the
interaction between SISC and sensation seeking predicted smoking behavior. The
following section presents the results of the data analysis.
Preliminary Analysis
To test this hypothesis, multinomial logistic regression analysis was used to
evaluate the extent to which the interaction of sensation seeking with SISC was
associated with smoking behavior. Interaction terms were created by first mean-
centering the sensation seeking variable and each of the SISC variables. Although
bivariate correlation showed no significant multicollinearity issues for sensation seeking
and any of the SISC variables, centering the variables was still done as a precaution to
ensure no issues of multicollinearity (Aiken and West, 1991). Variables were centered
by subtracting the variable’s mean from each score. Next, each centered SISC variable
was multiplied by the centered sensation seeking variable to create interaction terms.
Hypothesis Testing
To test the hypothesis that SISC will moderate the relationship between sensation
seeking and smoking behavior, a series of analyses to evaluate the impact of the sensation
seeking variable, the SISC item and the interaction of the sensation seeking with SISC

95

were run. The base model (ethnicity, gender, living with a smoker, achievement,
religiosity and social norm) was included as a control. Power to detect an effect of .1 was
.928. The effect size and significance of the sensation seeking variable, the SISC
variable and the interaction term were assessed. Additionally, the -2 log likelihood test
was used to determine the extent to which the interaction term significantly contributed to
the overall predictive utility of the model.
Results of the analysis found that several SISC items interacted with sensation
seeking to significantly predict smoking behavior. SISC-Popular and sensation seeking
interacted to produce an increased likelihood of being susceptible to smoking (e
β
= 1.382,
p < .05) (see Table 4.6)
4
. SISC-Smart also interacted with sensation seeking to produce
an increased likelihood of being a susceptible smoker (eβ = 1.477) (see Table 4.7).
SISC-Musician and sensation seeking interacted to produce a decreased likelihood of
smoking (e
β
= .584, p < .01) (see Table 4.8). SISC-Techie also interacted with sensation
seeking to produce a decreased likelihood of being susceptible to smoking (e
β
= .629, p <
.05) (see Table 4.9). SISC-Nonconformists interacted with sensation seeking to produce
a decreased likelihood of having ever tried smoking (e
β
= .627, p < .05) (see Table 4.10).
Sensation seeking did not interact with any of the SISC variables to produce an increased
likelihood of having ever tried smoking or of being a current smoker.
The interactions of sensation seeking with SISC-Popular (χ2(1)=9.587, p < .01),
SISC-Smart (χ2(1)=4.17, p < .05), SISC-Musician (χ2(1) = 10.297, p < .01), SISC-

4
Tables provide parameter estimates for each SISC item, the sensation seeking item and the interaction
term. Tables providing parameter estimates for all variables in the model can be found in Appendix B.

96

Techies (χ2(1) = 6.651, p < .05) and SISC-Nonconformist (χ2(1) = 6.933, p < .01) all
significantly improved the overall fit of the sensation seeking model (containing the
ethnicity, gender, living with a smoker, achievement, religiosity, social norm, sensation
seeking and the SISC item). This indicates that models containing these interactions fit
the data significantly better than models without these interactions.

Table 4.6. Multinomial Logistic Regression Analysis Parameter Estimates for Interaction
of SISC-Popular Item with Sensation Seeking to Predict Smoking Status Controlling for
Base Model .
Smoking Status(a) β SE β Sig. e
Β
.
Susceptible SISC-Popular 0.080 0.234 0.733 1.083
Sensation Seeking 0.185 0.158 0.243 1.203
SISC-Popular
*
SS(b) 0.323 0.154 0.036 1.382
*

Tried SISC-Popular 0.212 0.186 0.255 1.236
Sensation Seeking 0.366 0.135 0.007 1.443
**

SISC-Popular
*
SS(b) -0.016 0.122 0.893 0.984
Current SISC-Popular 0.390 0.413 0.345 1.477
Sensation Seeking -0.223 0.248 0.369 0.800
SISC-Popular
*
SS(b) 0.197 0.232 0.394 1.218
a The reference category is: Never, not susceptible.
b SS = Sensation Seeking
χ2 = 93.698(30) p < .001
-2LL χ2 = 374.053
*
p < .05
**
p < .01

Table 4.7. Multinomial Logistic Regression Analysis Parameter Estimate for Interaction
of SISC-Smart Item with Sensation Seeking to Predict Smoking Status Controlling for
Base Model .
Smoking Status(a) β SE β Sig. e
Β
.
Susceptible SISC-Smart -0.060 0.297 0.841 0.942
Sensation Seeking 0.151 0.157 0.339 1.162
SISC-Smart
*
SS(b) 0.390 0.172 0.023 1.477
*

Tried SISC-Smart -0.379 0.248 0.126 0.685
Sensation Seeking 0.350 0.136 0.010 1.419
*

SISC-Smart
*
SS(b) 0.156 0.153 0.308 1.169

97

Table 4.7. Multinomial Logistic Regression Analysis Parameter Estimate for Interaction
of SISC-Smart Item with Sensation Seeking to Predict Smoking Status Controlling for
Base Model (continued) .
Smoking Status(a) β SE β Sig. e
Β
.

Current SISC-Smart -0.569 0.503 0.258 0.566
Sensation Seeking -0.187 0.316 0.554 0.830
SISC-Smart
*
SS(b) 0.142 0.285 0.617 1.153
a The reference category is: Never, not susceptible.
b SS = Sensation Seeking
χ2 = 90.023(30) p < .001
-2LL χ2 = -377.728
*
p < .05

Table 4.8. Multinomial Logistic Regression Analysis Parameter Estimate for Interaction
of SISC-Musician Item with Sensation Seeking to Predict Smoking Status Controlling for
Base Model .
Smoking Status(a) β SE β Sig. e
β

Susceptible SISC-Musical Arts Factor -1.171 0.358 0.001 0.310
**

Sensation Seeking 0.031 0.162 0.848 1.032
SISC-Musical Arts Factor
*
SS(b) -0.537 0.185 0.004 0.584
**

Tried SISC-Musical Arts Factor -0.781 0.260 0.003 0.458
**

Sensation Seeking 0.429 0.146 0.003 1.535
**

SISC-Musical Arts Factor
*
SS(b) 0.098 0.179 0.584 1.103
Current SISC-Musical Arts Factor -0.880 0.516 0.088 0.415
Sensation Seeking -0.230 0.243 0.344 0.795
SISC-Musical Arts Factor
*
SS(b) -0.428 0.255 0.093 0.652
a The reference category is: Never, not susceptible.
b SS = Sensation Seeking
χ2 = 110.971(30) p < .001
-2LL χ2 = 356.780
**
p < .01

Table 4.9. Multinomial Logistic Regression Analysis Parameter Estimate for Interaction
of SISC-Techie Item with Sensation Seeking to Predict Smoking Status Controlling for
Base Model .
Smoking Status(a) β SE β Sig. e
β

Susceptible SISC-Techies Item -0.189 0.383 0.622 0.828
Sensation Seeking 0.125 0.159 0.431 1.134
SISC-Techies Item
*
SS(b) -0.463 0.198 0.020 0.629
*

98

Table 4.9. Multinomial Logistic Regression Analysis Parameter Estimate for Interaction
of SISC-Techie Item with Sensation Seeking to Predict Smoking Status Controlling for
Base Model (continued) .
Smoking Status(a) β SE β Sig. e
β

Tried SISC-Techies Item -0.325 0.329 0.324 0.723
Sensation Seeking 0.366 0.135 0.007 1.442
**

SISC-Techies Item
*
SS(b) 0.079 0.209 0.705 1.082
Current SISC-Techies Item 0.107 0.483 0.824 1.113
Sensation Seeking -0.154 0.264 0.560 0.857
SISC-Techies Item
*
SS(b) -0.334 0.241 0.166 0.716
a The reference category is: Never, not susceptible.
b SS = Sensation Seeking
χ2 = 89.690(30) p < .001
-2LL χ2 = 378.061
*
p < .05
**
p < .01

Table 4.10. Multinomial Logistic Regression Analysis Parameter Estimate for _
Interaction of SISC-Non-conformist Item with Sensation Seeking to Predict Smoking
Status Controlling for Base Model _
Smoking Status(a) β SE β Sig. e
β

Susceptible SISC-Non-conformist Item 0.332 0.336 0.324 1.393
Sensation Seeking 0.064 0.159 0.685 1.067
SISC-Non-conformist Item
*
SS(b) -0.356 0.224 0.112 0.700
Tried SISC-Non-conformist Item 0.512 0.298 0.086 1.669
Sensation Seeking 0.309 0.139 0.026 1.363
*

SISC-Non-conformist Item
*
SS(b) -0.467 0.192 0.015 0.627
*

Current SISC-Non-conformist Item 0.378 0.519 0.466 1.460
Sensation Seeking -0.223 0.257 0.385 0.800
SISC-Non-conformist Item
*
SS(b) -0.543 0.326 0.096 0.581
a The reference category is: Never, not susceptible.
b SS = Sensation Seeking
χ2 = 92.788(30) p < .001
-2LL χ2 = -394.104
*
p < .05

Several of the eight SISC factors also interacted with sensation seeking to
significantly predict smoking behavior. SISC-Elite and sensation seeking interacted to
produce an increased likelihood of being susceptible to smoking (E
β
= 2.012, p <.05; see

99

Table 4.11). SISC-Outsiders and sensation seeking interacted to produce a decreased
likelihood of being susceptible to smoking (e
β
=.552, p < .05; see Table 4.11). SISC-
Individualists interacted with sensation seeking to produce a decreased likelihood of
having ever tried smoking (e
β
= .573, p < .05; see Table 4.13). SISC-Musical Arts
interacted with sensation seeking to produce a decreased likelihood of being a current
smoker (e
β
= .428, p < .038; see Table 4.14).
The interactions of sensation seeking with SISC-Elites (χ2(1)=7.501, p < .01),
SISC-Individualist (χ2(1)=5.141, p < .05) and SISC-Musical Arts (χ2(1)=5.841, p < .05)
significantly improved the overall fit of the base sensation seeking model.

Table 4.11. Multinomial Logistic Regression Analysis Parameter Estimate for Interaction
of SISC-Elites Factor with Sensation Seeking to Predict Smoking Status Controlling for
Base Model .
Smoking Status(a) β SE β Sig. e
β

Susceptible Sensation Seeking 0.197 0.165 0.234 1.217
SISC-Elite Factor 0.301 0.395 0.446 1.351
SISC-Elite Factor
*
SS(b) 0.699 0.278 0.012 2.012
*

Tried Sensation Seeking 0.407 0.140 0.004 1.502
**

SISC-Elite Factor 0.691 0.324 0.033 1.996
*

SISC-Elite Factor
*
SS(b) 0.003 0.207 0.989 1.003
Current Sensation Seeking -0.182 0.270 0.499 0.833
SISC-Elite Factor 0.066 0.706 0.926 1.068
SISC-Elite Factor
*
SS(b) 0.654 0.457 0.152 1.924
a The reference category is: Never, not susceptible.
b SS = Sensation Seeking
χ2 = 93.698(30) p < .001
-2LL χ2 = 374.053
*
p < .05
**
p < .01

100

Table 4.12. Multinomial Logistic Regression Analysis Parameter Estimate for Interaction
of SISC-Outsiders Factor with Sensation Seeking to Predict Smoking Status Controlling
for Base Model .
Smoking Status(a) β SE β Sig. e
β

Susceptible Sensation Seeking 0.168 0.160 0.292 1.183
SISC-Outsider Factor -0.097 0.445 0.827 0.907
SISC-Outsider Factor
*
SS(b) -0.594 0.294 0.044 0.552
*

Tried Sensation Seeking 0.365 0.136 0.007 1.440
**

SISC-Outsider Factor -0.319 0.382 0.404 0.727
SISC-Outsider Factor
*
SS(b) -0.032 0.247 0.896 0.968
Current Sensation Seeking -0.280 0.263 0.287 0.756
SISC-Outsider Factor 0.392 0.679 0.564 1.479
SISC-Outsider Factor
*
SS(b) -0.107 0.373 0.774 0.898
a The reference category is: Never, not susceptible.
b SS = Sensation Seeking
χ2 = 85.908(30) p <. 001
-2LL χ2 = 381.843
**
p < .01

Table 4.13. Multinomial Logistic Regression Analysis Parameter Estimate for Interaction
of SISC-Individualist Factor with Sensation Seeking to Predict Smoking Status
Controlling for Base Model .
Smoking Status(a) β SE β Sig. e
β

Susceptible Sensation Seeking 0.093 0.156 0.550 1.098
Table 4.13. Multinomial Logistic Regression Analysis Parameter Estimate for Interaction
of SISC-Individualist Factor with Sensation Seeking to Predict Smoking Status
Controlling for Base Model (continued) .
SISC-Individualist Factor 0.393 0.449 0.381 1.482
SISC-Individualist Factor
*
SS(b) -0.471 0.288 0.102 0.625
Tried Sensation Seeking 0.347 0.142 0.015 1.414
*

SISC-Individualist Factor 0.957 0.394 0.015 2.604
*

SISC-Individualist Factor
*
SS(b) -0.557 0.260 0.032 0.573
*

Current Sensation Seeking -0.200 0.259 0.438 0.818
SISC-Individualist Factor -0.482 0.731 0.509 0.618
SISC-Individualist Factor
*
SS(b) -0.407 0.403 0.312 0.665
a The reference category is: Never, not susceptible.
b SS = Sensation Seeking
χ2 = 92.211(30) p < .001
-2LL χ2 = -375.540
*
p < .05

101

Table 4.14. Multinomial Logistic Regression Analysis Parameter Estimate for Interaction
of SISC-Musical Arts with Sensation Seeking to Predict Smoking Status Controlling for
Base Model .
Smoking Status(a) β SE β Sig. e
β

Susceptible Sensation Seeking 0.059 0.174 0.735 1.061
SISC-Musical Arts Factor -2.099 0.560 0.000 0.123
***

SISC-Musical Arts Factor
*
SS -0.467 0.296 0.114 0.627
Tried Sensation Seeking 0.399 0.139 0.004 1.490
**

SISC-Musical Arts Factor -1.074 0.397 0.007 0.342
**

SISC-Musical Arts Factor
*
SS 0.068 0.243 0.780 1.070
Current Sensation Seeking -0.346 0.262 0.187 0.707
SISC-Musical Arts Factor -1.694 0.890 0.057 0.184
SISC-Musical Arts Factor
*
SS -0.849 0.409 0.038 0.428
*

a The reference category is: Never, not susceptible.
b SS = Sensation Seeking
109.681(30) p < .001
-2LL χ2 = 358.070
*
p < .05
**
p < .01
***
p < .001

H3: Evaluating the Contribution of Strength of Identification with Each Social Category
to a Basic Integrative Model of Behavioral Prediction

The third hypothesis of this project is that SISC will contribute predictive validity
to a basic Integrative Model of Behavioral Prediction (i.e. a model using attitudes, social
norms and self efficacy to predict behavioral intention). Multiple regression analysis was
used to test this hypothesis. The size and significance of standardized coefficients and R
2
change were evaluated to determine the extent to which SISC contributed to the model.

102

Preliminary Analysis
The Integrative Model of Behavioral Prediction hypothesizes that behavioral
intention is the result of attitudes, social norms and self efficacy. The corresponding
attitude variables for this analysis were attitude towards health aspects of smoking,
attitude towards social aspects of smoking and attitude towards cigarette companies. The
corresponding social norm variables were percentage of friends who smoke and friend
and family approval of smoking. The corresponding self efficacy variables were ability
to easily not smoke and ability to refuse a cigarette offered by a friend. Because the
attitude variables and the two self efficacy variables were somewhat similar, they posed a
potential issue with multicollinearity. To determine the extent to which multicollinearity
existed between variables in the model, bivariate correlations were run. These
correlations examined associations between the predictors in the model (attitudes, social
norm, self efficacy), the SISC items and the predictors in the model and the SISC factors
and the predictors in the model. No problems with multicollinearity were indicated;
additional diagnostics (tolerance and variance inflation factor) also did not indicate an
issue with multicollinearity. Consequently all variables were included in the model.
Bivariate correlations were also run to examine the association of the attitudes, social
norm and self efficacy variables with intention to smoke.
These preliminary analyses found that attitudes toward smoking and attitudes
toward cigarette companies were positively associated. Percentage of friends who
smoked was positively associated with attitude toward smoking (emotional), attitude

103

toward cigarette companies and friends and family approval of smoking. Friends and
family approval of smoking was also positively correlated with the three attitude
variables. Finally, the self efficacy variables were negatively associated with attitudes
toward smoking and toward cigarette companies. Self efficacy to refuse a cigarette
offered by a friend was negatively associated with percentage of friends who smoked.
Table 15 provides details on these bivariate correlations.

Table 4.15. Bivariate Correlations between Attitudes, Social Norm and Self Efficacy
.
1 2 3 4 5 6 7 8

1 Attitude toward smoking –
Emotional
1
2 Attitude toward smoking –
Practical
.183
**
1
3 Attitude toward cigarette
companies - Practices
.132
*
.494
**
1
4 Attitude toward cigarette
companies - Rights
.437
**
.212
**
.131
**
1
5 Percentage friends who
smoke
.184
**
.060 .057 .199
**
1
6 Friends and family approval
of smoking
.233
**
.264
**
.237
**
.225
**
.285
*
1
7 Self efficacy to not smoke

-.246
**
-.227
**
-.185
*
-.120 .046 -.029 1
8 Self efficacy to refuse a
cigarette
-.308
**
-.187
*
-.203
*
-.124 -.195
*
-.106 .272
*
1
*
p < .05
**
p < .01

Table 4.16 provides the bivariate correlation matrix between the SISC items and
the attitude, self efficacy and social norm variables. Overall, identification as a member
of the categories Artist, Band, Involved, Musician and Theater were negatively associated
with attitudes about smoking, social norms and were positively associated with self

104

efficacy to not smoke. Identification as a Rebel was associated with more positive
attitudes about smoking, social norms and lower self efficacy.
Table 4.16. Bivariate Correlations between SISC Items and Attitudes, Self efficacy and
Social Norms .
1 2 3 4 5 6 7 8 .
Artist
-.240
**

-0.089

-.137
*
-0.128
-0.115

-0.046

-0.006

0.131

Athlete
0.014

-0.126

-0.018 -0.044
0.025

0

0.079

0.162
*

Average
-0.011

-0.056

-0.020 -0.041
0.001

-0.064

-0.021

-0.065

Band
-0.101

-0.068

-0.001 -.158
*

-.168
*

-0.120

-0.030

0.345
**

Class
Clown
0.065

0.078

0.060 0.049
0.015

0.065

0.100

-0.003

Emo
-0.005

0.044

-0.043 0.157
*

0.018

0.084

-0.002

-0.080

Gamer
0.096

0.061

0.027 -0.003
-0.051

-0.043

-0.023

-0.017

Goody-
goody
-0.024

-0.058

-0.009 -0.127
0.005

-0.166
*

0.063

-0.012

Goth
0.032

0.032

-0.026 0.200
**

0.061

0.097

-0.078

-0.094

Hippie
0.031

.214
**

0.109 0.016
0.092

0.127

-0.016

-0.268
**

Hipster
0.109

0.106

0.030 0.045 0.124
0.011

-0.152

-0.302
**

Involved
-.204
**

.177
**

-0.039 -.237
**

-0.137
*

-0.103

0.008

0.162
*

Misfit
0.034

0.013

-0.019 .192
**

0.155
*

0.105

0.094

-0.058

Musician
-0.129

-0.010

-0.010 -.163
*

-0.201
**

-0.122

-0.015

0.285
**

Nerd
-0.080

-0.074

0.051 -0.090
-0.143
*

-0.110

-0.049

-0.024

Nonconf-
ormist
-0.006

-0.017

-0.028 0.070
0.021

0.06

0.143

0.058

Partier
0.010

0.116

0.061 0.142
*

0.150
*

0.094

0.079

-0.232
**

Popular
0.108

0

-0.014 -0.011
0.003

0.041

-0.095

0.012

Preppy
0.073

0.006

0.023 0.002
0.148
*

0.004

-0.051
-0.038

Rebel
0.170
*

0.218
**

0.051 0.220
**

0.207
**

0.260
**

0.136
-0.100

Religious
0.011

-0.107

0.027 -0.031
-0.033

-0.129

-0.080

-0.008

105

Skater
0.231
**

0.131

0.015 0.088 0.117 0.087 -0.005 -0.179
*

Smart -0.205
**
0.209
**
-0.142
*
-0.123 -0.109 -0.122 0.087 0.123
Straightedg
e
0.056 0.038 0.042 0.048 -0.043 0.053 -0.019 0.012
Techie 0.074 0.027 0.124 0.030 -0.061 -0.047 -0.193
*
0.035
Theater -0.149
*
-0.073 -0.083 -0.138
*
-0.222
**
-0.089 -0.009 0.09
*
p < .05
**
p < .01
***
p < .001
1 Attitude toward smoking -Emotional
2 Attitude toward smoking - Practical
3 Attitude toward cigarette companies – Practices
4 Attitude toward cigarette companies - Rights
5 Percentage friends who smoke
6 Friends and family approval of smoking
7 Self efficacy to not smoke
8 Self efficacy to refuse a cigarette

Table 4.17 provides the bivariate correlation matrix showing the relationships
between the eight SISC factors and attitudes, self efficacy and social norms. In general,
identification with the categories of Achiever and Musical Arts was negatively associated
with attitudes and social norms and positively associated with self efficacy. Strength of
identification with the social category of Cultural Rebels was positively associated with
attitudes and social norms and negatively associated with self efficacy.

106

Table 4.17 Bivariate Correlations between SISC Factors and Attitudes, Self Efficacy and
Social Norms .
1 2 3 4 5 6 7 8 .
Elite 0.08 0.017 0.03 0.038 0.101 0.054 0.025 -0.031
Outsider 0.015 0.016 0.016 .167
*
0.02 0.053 -0.062 -0.077
Achiever -.326
**
-.241
**
-.157
*
-.250
**
-.183
**
-.138
*
0.046 .213
**

Conservative -0.012 -0.109 -0.003 -0.097 -0.012 -.173
**
-0.02 -0.044
Substance Free 0.056 0.038 0.042 0.048 -0.043 0.053 -0.019 0.012
Individualist .171
*
0.09 0.008 0.082 0.062 0.061 0.052 -0.081
Musical Arts -.184
**
-0.074 -0.044 -.225
**
-.288
**
-.163
*
-0.027 .356
**

Cultural Rebels .165
*
.282
**
0.098 .156
*
.224
**
.218
**
-0.007 -.352
**

*
p < .05
**
p < .01
***
p < .001
1 Attitude toward smoking -Emotional
2 Attitude toward smoking - Practical
3 Attitude toward cigarette companies – Practices
4 Attitude toward cigarette companies - Rights
5 Percentage friends who smoke
6 Friends and family approval of smoking
7 Self efficacy to not smoke
8 Self efficacy to refuse a cigarette

Finally, Table 4.18 provides bivariate correlations illustrating the relationship
between attitudes, social norms and self efficacy and behavioral intention to smoke.
More positive attitudes about smoking and about cigarette companies were associated
with stronger behavioral intention to smoke. Friends and family approval of smoking
was also associated positively with behavioral intention to smoke. Self efficacy to refuse
a cigarette offered by a friend was negatively associated with behavioral intention to
smoke.

107

Table 4.18. Bivariate Correlations between Attitudes, Social Norm and Self Efficacy with
Behavioral Intention to Smoke .
Behavioral Intention to Smoke
Attitude toward smoking - Emotional .252
**

Attitude toward smoking - Practical .198
*

Attitude toward cigarette companies - Practices .131
*

Attitude toward cigarette companies – Rights .203
*

Percentage friends who smoke .135
Friends and Family approval of smoking .275
**

Self efficacy to not smoke -.043
Self efficacy to refuse a cigarette -.360
**

*
p < .05
**
p < .01
***
p < .001

Hypothesis testing
To test the hypothesis that SISC would contribute to a basic Integrative Model a
series of forward stepwise multiple regression analyses were run. Behavioral intention to
smoke in the next year was used as the dependent variable. Because this item was asked
only to those who had never smoked, those who had tried smoking or who were current
smokers were not included in the analysis. A base model (ethnicity, gender, living with a
smoker, achievement, religiosity, sensation seeking and exposure to the truth™
campaign) and the attitude items (attitudes toward smoking and attitudes toward cigarette
companies), two social norm items (percentage of friends who had ever tried smoking
and friend and family approval of smoking) and two self efficacy items (self efficacy to
easily not smoke in the next month and ability to turn down a cigarette offered by a

108

friend) were included in the first step. Each of the SISC variables was entered in the
second step. Power to detect an effect of .10 was .914.
Results of the analysis found that of the variables included in the base model, only
sensation seeking (β = .218, p < .01), and self efficacy to turn down a cigarette (β = -.271,
p < .001) predicted behavioral intention to smoke. Overall, the base model accounted for
approximately 32% of the variance in intention to smoke (R
2
= .317, F(16,136) = 3.940,
p < .001). The addition of two SISC variables increased the predictive validity of the
base model. Inclusion of SISC-Musician (β = -.205, p < .05) increased the R
2
by .033 (R
2

= .340, F(17,131) = 3.970, p < .001). Inclusion of SISC-Rebel (β = .191, p < .05)
increased the R
2
by .029 (R
2
= .346,F(17,135) = 4.192, p < .001).
Of the eight SISC factors, only identification with the categories Musical Arts and
Cultural Rebel were associated with behavioral intention to smoke. Strength of
identification with Musical Arts was negatively related to intention to smoke (β = -.185, p
< .05) and increased the R
2
of the base model by .025 (R
2
= .332, F(17,131) = 3.833, p <
.001). Strength of identification with the category Cultural Rebel was positively related
to intention to smoke (β = .220, p < .01) and increased the R
2
of the base model by .034
(R
2
= .351, F(17,135) = 4.297, p < .001).
Table 4.19 presents the full parameter estimates of the base model. Table 4.20
presents parameter estimates of the SISC items that contributed significantly to the base

109

model, while Table 4.21 presents parameter estimates of the SISC factors that contributed
significantly to the model
5
.

Table 4.19 Multiple Regression Analysis Parameter Estimate for Attitudes, Social Norm
and Self Efficacy to Predict Behavioral Intention to Smoke .
B SE(B) β Sig. .
(Constant) 1.291 1.015 0.206
African American = Yes -0.315 0.342 -0.081 0.359
White = Yes -0.29 0.269 -0.093 0.282
Female = Yes -0.025 0.219 -0.008 0.908
Lives With Smoker = Yes -0.128 0.137 -0.073 0.350
Achievement -0.042 0.046 -0.070 0.364
Religiosity 0.110 0.059 0.145 0.065
Sensation Seeking 0.175 0.064 0.218 0.007
Identified truth 0.302 0.215 0.105 0.162
Attitude Toward Smoking – Emotional 0.034 0.102 0.029 0.737
Attitude Toward Smoking – Practical 0.185 0.098 0.178 0.062
Attitude Toward Cig. Companies – Practices -0.045 0.139 -0.029 0.745
Attitude Toward Cig. Companies – Rights 0.089 0.084 0.085 0.291
Percentage Friends who Tried Smoking 0.007 0.005 0.129 0.140
Friends and Family Approval of Smoking 0.150 0.102 0.120 0.143
Self Efficacy to Not Smoke 0.045 0.054 0.066 0.405
Self Efficacy to Turn Down Cigarette -0.219 0.063 -0.271 0.001

Dependent Variable: Behavioral Intention to Smoke
R
2
=.317, F(16,136)=3.940, p < .001
*
p < .05
**
p < .01
***
p < .001

Table 4.20. Multiple Regression Analysis Parameter Estimate for SISC Items with
Attitudes, Social Norms and Self Efficacy to Predict Behavioral Intention to Smoke .

B SE(B) β Sig.
Musician -.306 .120 -.205 .012
Rebel .296 .121 .191 .016

5
For detailed models of each analysis please refer to Appendix B.

110

Dependent Variable: Behavioral Intention to Smoke
*
p < .05
**
p < .01
***
p < .001

Table 4.21. Multiple Regression Analysis Parameter Estimate for SISC-Factors with
Attitudes, Social Norms and Self Efficacy to Predict Behavioral Intention to Smoke
.
B SE(B) β Sig
Musical Arts -.396 .179 -.185 .029
Cultural Rebel .589 .220 .220 .001
Dependent Variable: Behavioral Intention to Smoke
*
p < .05
**
p < .01
***
p < .001

H4: Evaluating the Contribution of Strength of Identification with Each Social Category
on the Social Norms Approach

Hypothesis Four proposes that SISC will moderate the effect of social norms on
adolescent smoking behavior. This hypothesis was tested by creating an interaction term
for each of the 26 SISC variables and the social norms item (percentage of friends who
are current smokers) and then evaluating the impact of the interaction effect on a standard
social norms model.
Preliminary Analysis
Although previous analysis did not indicate any multicollinearity issues between
social norms and the SISC variables, both were centered as a precaution. A one-way
ANOVA was used to examine basic mean differences in percentage of friends who are
current smokers by smoking status (see Table 4.22). Results of this analysis showed that

111

individuals who were not susceptible to smoking had the lowest percentage of friends
who smoked, reporting an average of 7.75%. Individuals who were susceptible to
smoking reported that an average of 14.62% of their friends currently smoked, and
individuals who had tried smoking reported an average of 16.2% of their friends smoked.
Individuals who were current smokers had the highest percentage of friends who smoked,
reporting an average of 49.0%. This was significantly higher than the percentage
reported by unsusceptible smokers, but not significantly higher than those reported by
susceptible smokers or individuals who had tried smoking.

Table 4.22. One-way ANOVA: Average Percentage Friends Who Smoke by Smoking
Status .
Unsusceptible Susceptible Tried Current
Percentage
Friends who 7.75 14.615 16.203 49.000
*

Smoke _
Note: Reference group in italics
*
p < .05

Hypothesis Testing
Multinomial logistic regression analysis was used to test the hypothesis that SISC
would moderate the effect of social norms on smoking behavior. Again, forced entry
likelihood ratio was used as the method of entry. To create the interaction term, the
centered social norms item was multiplied by each of the centered SISC variables. Power
to detect an effect of .10 was .928. To determine the extent to which the interaction term
was associated with smoking behavior, standardized effects size and significance were
examined. To determine the extent to which the interaction term contributed

112

significantly to the predictive validity of a base social norms model, a -2 log likelihood
test was run comparing the full model to a base model containing the control variables
(ethnicity, gender, lives with smoker, achievement, religiosity, sensation seeking and
exposure to truth™ campaign) and the social norms variable and SISC variable.
Results of this analysis did not find an interaction effect between any of the SISC
variables and social norm on likelihood of being susceptible to smoking, nor on
likelihood of being a current smoker. However, SISC-Theater interacted with social
norm to produce a slightly increased likelihood of having ever tried smoking (e
β
= 1.008,
p < .05). SISC-Smart also interacted with social norm to produce an increased likelihood
of having ever tried smoking (e
β
= 1.006, p < .05). Both the interaction of social norm
with SISC-Theater (χ2(1) = 8.437, p < .01) and SISC=Smart (χ2(1) = 7.318, p < .01)
significantly increased model fit. Tables 5.23 and 5.24 provide details for these models
6
.
Of the eight SISC factors, only SISC-Achievers significantly interacted with
social norms to affect smoking behavior. The interaction of SISC-Achievers with social
norms was associated with an increased likelihood of having ever tried smoking (e
β
=
1.044, p < .028). This interaction term also significantly improved the fit of the base
social norms model (χ2(1) = 6.494, p < .05).

Table 4.25 provides details for these models.

Table 4.23. Multinomial Logistic Regression Analysis Parameter Estimate for Interaction
of SISC-Theater Item with Social Norm to Predict Smoking Status Controlling for Base
Model .
β SE β Sig. e
β
___.

6
Tables provide parameter estimates for the SISC items, the social norms variable and the interaction term.
For parameter estimates of the other variables included in the model, please refer to Appendix B.

113

Susceptible Social Norm 0.002 0.021 0.935 1.002
SISC-Theater -0.991 0.391 0.011 0.371
*

SISC-Theater
*
Social Norm 0.007 0.020 0.742 1.007
Tried Social Norm 0.029 0.012 0.015 1.029
*

SISC-Theater -0.296 0.260 0.254 0.744
SISC-Theater
*
Social Norm 0.033 0.013 0.010 1.034
*

Current Social Norm 0.064 0.019 0.001 1.066
**

SISC-Theater 0.330 0.559 0.555 1.391
SISC-Theater
*
Social Norm -0.006 0.018 0.747 0.994
a The reference category is: Never, not susceptible.
χ2(30)=100.185, p < .001
-2LL χ2 = 367.566
*
p < .05
**
p < .01
***
p < .001

Table 4.24. Multinomial Logistic Regression Analysis Parameter Estimate for Interaction
of SISC-Smart Item with Social Norm to Predict Smoking Status Controlling for Base
Model .
Smoking Status(a) β SE β Sig. e
β
___

Susceptible Social Norm 0.008 0.014 0.575 1.008
SISC-Smart -0.122 0.308 0.694 0.886
SISC-Smart
*
Social Norm -0.008 0.020 0.701 0.993
Tried Social Norm 0.022 0.010 0.033 1.022
*

SISC-Smart -0.314 0.253 0.215 0.731
SISC-Smart
*
Social Norm 0.036 0.015 0.019 1.037
*

Current Social Norm 0.062 0.017 0.000 1.064
***

SISC-Smart -0.375 0.523 0.473 0.687
SISC-Smart
*
Social Norm -0.007 0.024 0.773 0.993
a The reference category is: Never, not susceptible.
χ2(30)=93.171, p < .001
-2LL χ2 – 374.580
*
p < .05
**
p < .01
***
p < .001

114

Table 4.25. Multinomial Logistic Regression Analysis Parameter Estimate for Interaction
of SISC-Achievers Factor with Social Norm to Predict Smoking Status Controlling for
Base Model .
Smoking status(a) β SE β Sig. e
β
.
Susceptible Social Norm -0.004 0.019 0.846 0.996
SISC-Achievers Factor -0.847 0.451 0.061 0.429
SISC-Achievers Factor
*
SN(b) -0.017 0.026 0.509 0.983
Tried Social Norm 0.024 0.011 0.026 1.025
*

SISC-Achievers Factor -1.409 0.402 0.000 0.244
***

SISC-Achievers Factor
*
SN(b) 0.043 0.020 0.028 1.044
*

Current Social Norm 0.070 0.016 0.000 1.073
***

SISC-Achievers Factor -0.895 0.787 0.256 0.409
SISC-Achievers Factor
*
SN(b) 0.039 0.028 0.156 1.040
a The reference category is: Never, not susceptible.
b SN = Social Norm
χ2(30)=105.820, p < .001
-2LL χ2 =361.931
*
p < .05
**
p < .01
***
p < .001

H5: Evaluating the Impact of Strength of Identification with Each Social Category on the
Effectiveness of the truth™ Campaign

The final hypothesis of this project proposes that SISC will moderate the effects
of exposure to the truth™ campaign on smoking behavior. To test this hypothesis,
multinomial logistic regression analysis was used to evaluate the impact of the interaction
term on the same base model. Size and significance of the standardized coefficients were
evaluated, as was the significance and size of the R
2
change and likelihood ratio test.
Preliminary Analysis

115

Chi-square analysis was run to determine the extent to which individuals varied in
their smoking status as a function of exposure to the truth™ campaign. The analysis
found no significant differences in smoking status based on whether or not someone had
seen the truth™ campaign in the past 30 days.
Hypothesis Testing
For each of the hypotheses, an interaction term was created for exposure to the
truth™ campaign by SISC. All SISC variables were mean-centered. To test H5,
multinomial logistic regression was used. Forced entry likelihood ratio was the method
of entry, with all variables entered on one step. Power to detect an effect of .10 was .928.
Results of this analysis showed that several of the SISC items interacted with exposure to
the truth™ campaign to significantly predict smoking behavior. SISC-Techie and
exposure to the truth™ campaign interacted to produce a decreased likelihood of being
susceptible to smoking (e
β
= .153, p < .05). SISC-Nerd also interacted with exposure to
the truth™ campaign to produce a decreased likelihood of being susceptible to smoking
(e
β
= .194, p < .05). Finally, SISC-Goody-goody interacted with exposure to the truth™
campaign to produce an increased likelihood of being susceptible to smoking (e
β
= 6.008,
p < .01). 4.26 provides parameter estimates for these models
7
.
Several of the eight SISC factors also interacted with exposure to the truth™
campaign to predict smoking behavior. The interaction of SISC-Conservatives with
exposure to the truth™ campaign was significantly associated with an increased

7
Tables provide parameter estimates for the SISC items, the variable measuring exposure to the truth™
campaign and the interaction term. For details on the full models, please refer to Appendix B.

116

likelihood of being susceptible to smoking (e
β
= 5.124, p < .05). SISC-Musical Arts
interacted with exposure to the truth™ campaign to produce a decreased likelihood of
being susceptible to smoking (e
β
= .061, p < .034).
Of these interaction terms mentioned above, SISC-Techie (χ2(1) = 6.729, p <
.01), SISC-Nerd (χ2(1) = 9.877, p < .01), SISC-Conservative (χ2(1) = 5.624, p < .01) and
SISC-Musical Arts (χ2(1)=5.824, p < .01) significantly improved the fit of the base
model. Tables 4.27 through 5.30 provide parameter estimates for these models
7
.

Table 4.26. Multinomial Logistic Regression Analysis Parameter Estimate for Interaction
of SISC-Techie Item with Exposure to truth™ Campaign to Predict Smoking Status
Controlling for Base Model .
Smoking Status(a) β SE β Sig. e
β
.
Susceptible SISC-Techie Item 0.562 0.392 0.152 1.754
Exposure to truth campaign 0.297 0.524 0.571 1.346
Exposure
*
SISC-Techie -1.874 0.871 0.031 0.153
*

Tried SISC-Techie Item 0.140 0.347 0.686 1.150
Exposure to truth campaign 0.021 0.402 0.958 1.021
Exposure
*
SISC-Techie -1.036 0.671 0.122 0.355
Current SISC-Techie Item 0.573 0.585 0.327 1.774
Exposure to truth campaign 0.728 0.797 0.361 2.071
Exposure
*
SISC-Techie -0.726 0.923 0.432 0.484
χ2(33)=89.768, p < .001
-2LL χ2 = 377.983
*
p < .05
**
p < .01
***
p < .001
Table 4.27 Multinomial Logistic Regression Analysis Parameter Estimate for Interaction
of SISC-Nerd Item with Exposure to truth™ Campaign to Predict Smoking Status
Controlling for Base Model .
Smoking Status(a) β SE β Sig. e
β
.
Susceptible SISC-Nerd Item 0.649 0.349 0.063 1.914
Exposure to truth campaign 0.538 0.509 0.291 1.712
Exposure
*
SISC-Nerd -1.641 0.676 0.015 0.194
*

Tried SISC-Nerd Item -0.340 0.316 0.281 0.712
Exposure to truth campaign -0.034 0.434 0.938 0.967

117

Exposure
*
SISC-Nerd -1.051 0.654 0.108 0.349
Current Sensation Seeking -0.226 0.254 0.374 0.798
SISC-Nerd Item 0.293 0.612 0.633 1.340
Exposure
*
SISC-Nerd -2.176 1.512 0.150 0.113
a The reference category is: Never, not susceptible.
χ2(33) = 99.235, p < .001
-2LL χ2 = 368.516
*
p < .05
**
p < .01
***
p < .001

Table 4.28. Multinomial Logistic Regression Analysis Parameter Estimate for Interaction
of SISC-Goody-goody Item with Exposure to truth™ Campaign to Predict Smoking
Status Controlling for Base Model .
Smoking Status(a) β SE β Sig. e
β
.
Susceptible SISC-Goody-goody Item -0.803 0.442 0.069 0.448
Exposure to truth campaign 0.120 0.570 0.833 1.128
Exposure
*
SISC-Goody-goody 1.793 0.615 0.004 6.008
**

Tried SISC-Goody-goody Item -0.284 0.304 0.350 0.753
Exposure to truth campaign 0.179 0.385 0.641 1.196
Exposure
*
SISC-Goody-goody 0.200 0.451 0.657 1.222
Current SISC-Goody-goody Item -0.625 0.721 0.386 0.535
Exposure to truth campaign 0.907 0.817 0.267 2.476
Exposure
*
SISC-Goody-goody -0.023 0.970 0.981 0.978
a The reference category is: Never, not susceptible.
χ2(33) = 98.840, p < .001
-2LL χ2 = 388.052
*
p< .05
**
p < .01
***
p < .001

Table 4.29 Multinomial Logistic Regression Analysis Parameter Estimate for Interaction
of SISC-Conservative Factor with Exposure to truth™ Campaign to Predict Smoking
Status Controlling for Base Model .
Smoking Status(a) β SE β Sig. e
β
. .
Susceptible Exposure to truth campaign 0.206 0.498 0.679 1.229
SISC-Conservative Factor -0.402 0.491 0.413 0.669

118

Exposure
*
SISC 21.634 0.795 0.040 5.124
*

Tried Exposure to truth campaign 0.154 0.384 0.688 1.167
SISC-Conservative Factor -0.205 0.405 0.613 0.815
Exposure
*
SISC-Conservative 20.281 0.623 0.652 1.324
Current Exposure to truth campaign 1.322 0.929 0.155 3.752
SISC-Conservative Factor -2.499 1.113 0.025 0.082
*

Exposure
*
SISC-Conservative 21.80 21.449 0.214 6.064
a The reference category is: Never, not susceptible.
χ2(33)=96.237, p < .001
-2LL χ2 = 390.655
*
p < .05
**
p < .01
***
p < .001

Table 4.30 Multinomial Logistic Regression Analysis Parameter Estimate for Interaction
of SISC-Musical Arts Factor with Exposure to truth™ Campaign to Predict Smoking
Status Controlling for Base Model .
Smoking Status(a) β SE β Sig. e
β
.
Susceptible Exposure to truth campaign -0.613 0.746 0.411 0.542
SISC-Musical Arts Factor -1.274 0.580 0.028 0.280
SISC-Musical Arts Factor
*
Exposure -2.8051.324 0.034 0.061
*

Tried Exposure to truth campaign -0.034 0.408 0.934 0.967
SISC-Musical Arts Factor -0.747 0.421 0.076 0.474
SISC-Musical Arts Factor
*
Exposure -0.7910.831 0.341 0.453
Current Exposure to truth campaign 0.861 0.898 0.337 2.367
SISC-Musical Arts Factor -1.525 1.034 0.140 0.218
SISC-Musical Arts Factor
*
Exposure 0.342 1.536 0.824 1.408
a The reference category is: Never, not susceptible.
χ2(33)=109.664, p < .001
-2LL χ2 = 358.087,
*
p < .05
**
p < .01
***
p < .001

119

Summary of Results

Evaluation of the Base Model
Results of this data analysis indicated that none of the established predictors of
smoking behavior predicted susceptibility to smoking for this sample. Being white and
having better grades in school were associated with a decreased likelihood of having tried
smoking, while higher levels of sensation seeking were associated with increased
likelihood of having tried smoking. Having better grades in school was also associated
with a decreased likelihood of being a current smoker, while having more friends who
smoked was associated with an increased likelihood of being a current smoker. Table
4.31 summarizes results of the base model.

Table 4.31 Summary of Base Model Parameter Estimates.
Susceptible Tried Current
African American = Yes NS NS NS
White = Yes NS
-
NS
Female = Yes NS NS NS
Lives with Smoker = Yes NS NS NS
Academic Achievement NS
- -
Religiosity NS NS NS
Social Norm NS NS
+
Sensation Seeking NS
+
NS
truth™ Exposure NS NS NS
Χ2(27) = 85.656, p < .001

Summary of H1: Ability of SISC to Predict Smoking Behavior
Several of the 26 SISC items and resultant eight factors were found to be
associated with either a decreased or increased likelihood of being susceptible to

120

smoking, having tried smoking and being a current smoker. Identification with the social
categories of Artist, Involved, Musician, Nerd, Theater, Religious, Achiever and
Conservative is associated with decreased smoking behavior. Identification with the
social categories of Hippie, Rebel, Partier, Hipster, Average and Cultural Rebel is
associated with increased smoking behavior. Table 4.32 summarizes which SISC items
are associated with which levels of smoking status. Table 4.33 summarizes which SISC
factors are associated with which levels of smoking status.

Table 4.32 Summary of SISC Items Ability to Predict Smoking Behavior

Susceptible Tried Current Significantly
improves base
model fit?
Artist
- -
NS

Athletes NS NS NS
Emo NS NS NS
Involved NS
-
NS

Musicians
- -
NS

Popular NS NS NS
Preppy NS NS NS
Band NS NS NS
Goth NS NS NS
Techies NS NS NS
Nerds NS
-
NS

Theater
-
NS NS

Straightedge NS NS NS
Smart NS NS NS

121

Class clown NS NS NS
Misfit NS NS NS
Gamers NS NS NS
Skaters NS
+
NS

Nonconformist NS NS NS
Religious
-
NS NS
Goody-goody NS NS NS
Hippie NS NS
+

Rebels
+ + +

Partiers
+ +
NS

Hipsters
+
NS NS
Average
+
NS NS

Table 4.33 Ability of SISC Factors to Predict Smoking Behavior

Susceptible Tried Current Significantly
improves
model fit?
Achievers
- -
NS

Cultural Rebels
+ + +

Conservatives NS NS
-
Outsiders NS NS NS
Musical Arts
-
NS NS

Individualists NS NS NS
Elites NS NS NS
Substance Free NS NS NS
Truth™ Exposure NS NS NS

122

Summary of H2: Moderating Effect of SISC on Sensation Seeking
Several SISC items and factors were found to interact with sensation seeking to
produce either an increased or decreased likelihood of being susceptible to smoking and
having tried smoking. Sensation seeking and identification with the social categories of
Popular, Smart and Elite interacted to produce increased smoking behavior. In contrast,
sensation seeking and identification with the social categories of Musician, Techie,
Nonconformist, Outsider, Musical Arts and Individualist interacted to produce decreased
smoking behavior. Table 4.34 summarizes these results.

Table 4.34 Summary of H2 Results: Interaction of SISC with Sensation Seeking to
Predict Smoking Status .

Susceptible Tried Current Significantly
improves
base model
fit?
SISC-Items

Popular
*
SS
+
NS NS

Smart
*
SS
+
NS NS

Musician
*
SS
-
NS NS

Techie
*
SS
-
NS NS

Nonconformist
*
SS NS
-


SISC-Factors

Elite
*
SS
+
NS NS


123

Outsiders
*
SS
-
NS NS
Musical Arts
*
SS
-
NS NS

Individualists
*
SS NS
-
NS


Contribution of SISC to a Basic Integrative Model of Behavioral Prediction
Strength of identification with the categories Artist, Musician, Band, Theater,
Rebel, Partier, Hipster, Musical Arts and Cultural Rebel all contributed significantly to a
basic Integrative Model containing attitudes, social norms and self efficacy.
Identification with the social categories Artist, Musician, Band and Theater were
negatively associated with behavioral intention to smoke, while identification with the
social categories Rebel, Partier, Hipster and Cultural Rebel were positively associated
with behavioral intention to smoke. Table 4.35 summarizes the results for this
hypothesis.

Table 4.35. Contribution of SISC to Basic Integrative Model Predicting Behavioral
Intention to Smoke .
.
Β Sig. R
2
change Sig
SISC-Items

Musician -.205 p < .05 .085 p < .05
Rebel .191 p < .05 .029 p < .05
SISC-Factors
Musical Arts -.185 p < .05 .025 p < .05
Cultural Rebel .220 P < .01 .034 p < .01

124

Summary of H4: Moderating Effect of SISC on Social Norms
There were several SISC variables that interacted with social norms to affect
smoking behavior. Strength of identification with the categories Theater, Smart and
Achiever all interacted with social norms to produce an increased likelihood of ever
having tried smoking. There was no interaction effect that produced a decreased
likelihood of smoking behavior. Table 4.36 summarizes the results related to
Hypothesis 4.

Table 4.36 Summary of H4 Results: Interaction of SISC with Social Norms to Predict
Smoking Status .
.
Susceptible Tried Current Significantly
improves
base model
fit?
SISC-Items

Theater NS
+
NS

Smart NS
+
NS

SISC-Factors
Achiever NS
+
NS


H5: Moderating Effect of SISC on Exposure to the truth™ campaign
Strength of identification with the categories Techie, Nerd, Goody-goody,
Musical Arts and Conservative all interacted with exposure to the truth™ campaign to

125

affect smoking behavior. SISC-Techie, SISC-Nerd and SISC-Musical Arts interacted
with exposure to the truth™ campaign to produce a decreased likelihood of being
susceptible to smoking, while SISC-Goody-goody and SISC-Conservative interacted
with exposure to the truth™ campaign to produce an increased likelihood of being
susceptible to smoking. Table 4.37 summarizes the findings relating to this hypothesis.

Table 4.37 Summary of H5 Results: Interaction of SISC with Exposure to the truth™
Campaign to Predict Smoking Status .

Susceptible Tried Current Significantly
improves
base model
fit?
SISC-Items

Techie
*
Exposure
-
NS NS

Nerd
*
Exposure
-
NS NS

Goody-goody
*

Exposure
+
NS NS
SISC-Factors

Conservative
*

Exposure
+
NS NS

Musical Arts
*

Exposure
-
NS NS


Overall, these results indicate that the strength of an adolescent’s identification
with a variety of social categories impact his or her smoking behavior in a variety of
ways. The next and final chapter of this dissertation includes a discussion of the
implications, limits and practical applications of these results.

126

CHAPTER 5: DISCUSSION AND CONCLUSION

This dissertation hypothesized that social identity plays a role in adolescent
smoking behavior. To test this hypothesis, a pre-test survey was conducted, whereby
adolescents were asked to indicate which social categories they associated with smoking
and not smoking. These responses were then condensed into a list of social categories
associated with smoking and a list of categories associated with not smoking. Next, a
separate sample of adolescents was surveyed to understand how strength of identification
with these social categories influenced their smoking behavior. Results of the data
analysis indicate that there is indeed a relationship between the strength with which an
adolescent identifies as a member of a certain social category and their smoking behavior.
More specifically, certain social categories are associated with increased smoking
behavior while identification with other social categories is associated with decreased
smoking behavior. This final chapter discusses these results and evaluates the
methodological, practical and theoretical implications of these findings.

Study Findings
Evaluation of the Base Model
Previous research has identified ethnicity, gender, living with a smoker, academic
achievement, religiosity, social norms, sensation seeking and exposure to the truth™ anti-
smoking campaign to be strong predictors of smoking behavior. These variables were

127

entered into a multinomial logistic regression model which served as a base model to
which to compare models containing the social identity items. Likelihood of being a
susceptible smoker, of having ever tried smoking and of being a current smoker in
relation to being unsusceptible to smoking were examined as the dependent variables in
the present research. The current analysis found that none of the items entered into the
base model were associated with being susceptible to smoking. Being white and having
higher levels of school achievement were associated with a decreased likelihood of
having tried smoking, while sensation seeking was associated with an increased
likelihood of having tried smoking. School achievement was also associated with a
decreased likelihood of being a current smoker, while the social norm of having friends
who smoked was positively associated with being a current smoker.
The results using the base model are consistent with previous research. School
achievement was positively associated with a lowered likelihood of ever having tried
smoking or of being a current smoker, which supports the results found by Bryant and
colleagues (Bryant, Schulenberg, Bachman, O’Malley and Johnston, 2000; Bryant,
Schulenberg, O’Malley, Bachman and Johnston, 2003). It has been suggested that the
relationship between academic achievement and smoking behavior is one of peer effects,
such that those who perform poorly in school are more likely to associate with smoking
peers. However, this study found that academic achievement was associated with
smoking behavior even after controlling for social norms, indicating that the effect is not
completely due to friend or perceived peer behavior.

128

Sensation seeking was positively associated with increased likelihood of having
tried smoking, but not with smoking susceptibility or with being a current smoker. These
findings indicate that the transition from being a non-smoker to trying smoking is
motivated primarily by sensation seeking. However, the transition from having tried
smoking to being an established smoker is motivated by another factor or factors. While
trying smoking may be a thrilling or novel behavior that high sensation seekers look for,
over time this behavior may lose its appeal. After having tried a cigarette once or twice,
smoking no longer remains a new experience to high sensation seekers. Thus, while
trying smoking may satisfy a high sensation seeker’s desire for new and exciting
experiences, continuing to smoke does not serve this function and as such, is not
associated with sensation seeking.
Continued smoking behavior was found to be predicted by social norms. The
higher percentage of one’s friends who smoke, the more likely that individual was to be a
current smoker. This indicates that the leap from trying smoking to continuing to smoke
may be motivated by one’s friends. In this way, the influence of friends on the transition
to cigarette addiction becomes apparent. An adolescent may transition from trying
smoking once or twice to smoking regularly due to implicit encouragement from his or
her friends. In other words, simply being around friends who smoke can motivate an
adolescent to continue smoking (National Institute on Drug Abuse, 2006). Additionally,
as the adolescent becomes addicted to smoking, seeing others smoke can act as a cue that

129

increases his or her craving, and makes quitting more difficult, furthering the cycle of
addiction.
Although the variables mentioned above were found to be statistically significant
predictors of smoking in this dataset, the variables of gender, being African American,
living with a smoker, religiosity and exposure to the truth™ campaign were not
associated with smoking behavior. These findings diverge from previous research that
has identified these factors as important predictors of smoking behavior. This difference
could be due to the relatively small sample size – for instance, while the percentage of
African Americans in the sample (12.9%) was comparable to national demographic data,
this only equated to 29 cases. Although multicolinearity was not identified as a
problematic issue, one alternate explanation for the failure to find significant
relationships between these variables and smoking behavior could be that certain items
had already accounted for the variance in smoking behavior. For instance, the effect of
religiosity may have already been accounted for by the school achievement variable, as
the two were related.

Does Identification with Certain Social Categories Have an Independent Effect on
Smoking Behavior?
This research tested the effects of the 26 social category items and the eight social
category factors on smoking behavior. The 26 social category items measured
adolescents’ strength of identification with single social categories while the eight social

130

category factors measured strength of identification with identity clusters (for instance,
the category Musical Arts consisted of the items Musician, Band and Theater). Data
analysis found that identification with several social categories was associated with
smoking behavior. Identifying as an Artist, as Involved, as a Musician, as a Nerd, as a
Theater member and as Religious were associated with decreased smoking behavior. The
social identity category factors of Musical Arts, Conservatives and Achievers factors
were also associated with decreased smoking behavior. In contrast, identifying as a
Skater, Hippie, Rebel, Partier, Hipster and Average was positively associated with
increased smoking behavior. Among the eight broad social category factors, only
Cultural Rebel – comprised of Hippie, Rebels and Partiers – was associated with smoking
behavior.
These findings indicate that identification with these specific social categories has
an independent and pronounced effect on smoking behavior, even after controlling for
ethnicity, gender, living with a smoker, religiosity, achievement, sensation seeking and
social norms. Of these, identification as a Rebel was most strongly associated with
increased smoking behavior. With a one-unit increase in identification as a Rebel, odds
of an adolescent’s having tried smoking went up over two-fold and their odds of being a
current smoker increased over four-fold. Similarly, a one-unit increase in identification
as a Hippie nearly doubled odds of being a current smoker. Alternately, identification as
a Musician was the strongest predictor of decreased smoking behavior. With every one
unit increase in identification as a Musician, the odds of being a susceptible smoker

131

decreased by two-thirds and the odds of having ever tried smoking went down by almost
half.
The significance of the social category factors indicates that there are certain
identity clusters that predict smoking behavior. While the categories of Elite, Outsiders,
Individualists and Substance-free did not predict smoking behavior, the categories of
Achiever, Conservative, Musical Arts and Cultural Rebel did. In the case of Cultural
Rebel, for instance, the combination of strongly identifying with the Hippie, Rebel and
Hipster categories increases the likelihood of being a current smoker, though not of being
an experimental smoker. This indicates that adolescents who identify as Cultural Rebels
may be more likely to continue smoking after trying once. While individual social
categories were found to predict smoking behavior, the identity clusters one identifies
with – as represented by the eight social category factors – were found to be important
predictors as well.
Social identity theory makes a behavioral prediction that as strength of
identification as a group or category member increases, so will compliance to the group
prototype. Pre-testing identified the categories of Musician, Nerd, Involved (in school),
Theater and Religious to be associated with not smoking as a prototypical behavior.
Conversely, pre-testing identified the categories of Skater, Hippie, Rebel, Partier and
Hipster as associated with smoking as a prototypical behavior. The behavioral prediction
of Social Identity Theory is borne out in these results – strength of identification with
categories for which not smoking is prototypical is associated with decreased likelihood

132

of smoking, whereas strength of identification with categories for which smoking is
prototypical is associated with increased likelihood of smoking. These findings show
that identification with these social categories increases the extent to which adolescents
comply with the categories’ prototypical behaviors of either smoking or not smoking.
Finally, preliminary analysis showed that certain of these social categories were
associated with increased social norms around smoking and increased sensation seeking.
Individuals who identified with the non-smoking categories tended to have lower
percentages of friends who smoked and to be lower in sensation seeking, while the
opposite was true of individuals who identified with the smoking categories. However,
results of this analysis showed that strength of identification with these social categories
was associated with smoking behavior independently of social norms or sensation
seeking. Although certain social categories may interact with sensation seeking or social
norms to affect smoking behavior (as will be discussed in the following sections), these
findings indicate that identification with the above social categories has an effect on
smoking behavior above and beyond that which can be accounted for by one’s sensation
seeking or one’s friends’ behavior.

Does Identification with Certain Social Categories Contribute to the Sensation Seeking
Approach?
Results of the data analysis also suggest that strength of identification with certain
social categories can significantly contribute to the Sensation Seeking Approach. The

133

items for Musician, Techie, Nonconformist and the factors for Musical Arts, Individualist
and Outsiders all interacted with sensation seeking to produce a decreased likelihood of
smoking behavior. In contrast, the items for Popular and Smart and the factor for Elite
interacted with sensation seeking to produce an increased likelihood of smoking
behavior.
These findings indicate that the effects of sensation seeking on smoking behavior
are weakened when one identifies with the categories Musician, Techie, Nonconformist,
Musical Arts, Individualist and Outsiders. When these categories have not-smoking as a
prototypical behavior, this effect may be caused by identification suppressing the effects
of sensation seeking. In other words, the desire to behave according to a category’s
prototype may be stronger than the desire to satisfy one’s need for exciting and novel
experiences like smoking. Alternately, these social categories may offer prototypical
alternatives that satisfy one’s desire for sensation. For instance, performing music in
front of an audience is a behavior which is likely prototypical for Musicians and which is
also likely to be exciting. Or, Nonconformists may participate in a variety of behaviors
that make them feel as they are rebelling which can satisfy a desire for stimulation and
excitement. As such, smoking is no longer needed as a way to satisfy a sensation
seeker’s desire for these kinds of experiences.
On the other hand, identification with the categories Popular, Smart and Elite all
strengthened the effect of sensation seeking on smoking behavior. In other words, when
individuals strongly identified as either Popular, Smart or Elite, and when they were also

134

high in sensation seeking, the likelihood of engaging in smoking behavior increased.
None of these categories were independently associated with smoking behavior. This
indicates that there is something unique about the interaction between these categories
and sensation seeking that influences smoking behavior. This effect may be due to the
fact that these social categories do not offer alternate prototypical behaviors that act as
healthy outlets for high sensation seekers, particularly in the case of the category Smart.
Thus, high sensation seekers in this category may look elsewhere – such as to smoking –
to satisfy their need for new and stimulating experiences.
Another explanation is that these identities may prime smoking as a valid means
of stimulation and excitement, making it likely that high sensation seekers in these
categories turn to cigarettes. Cigarettes have long been portrayed in the media and
advertisements as cool, sexy and glamorous. In fact, tobacco companies have identified
and targeted several young adult market segments, including Enlightened Go-Getters,
Uptown Girls and Mavericks (National Cancer Institute, 2009). This kind of marketing
strengthens the association between these identities and cigarette use, positioning
smoking as prototypical of a Go-Getter or Uptown Girl. It is quite possible that there is
considerable overlap between the individuals who fall into these market segments and the
individuals who identify with the social categories of Smart, Popular and Elite. This
advertising could then prime smoking as an acceptable risky behavior for a sensation
seeker who is already pre-disposed to take risks.

135

Overall, the construct of social identity is a useful addition to the sensation
seeking model. Social identity has an effect on smoking independent of sensation
seeking, but also interacts with sensation seeking to produce an effect. In this case, social
identity is particularly valuable for helping researchers understand when they should and
should not expect a positive association between sensation seeking and increased
smoking behavior.
Does Identification with Certain Social Categories Contribute to the Integrative Model of
Behavioral Prediction?
Data analysis found that strength of identification with several social categories
significantly improved the ability of attitudes, norms and self efficacy to predict intention
to smoke. In this model, identification with the categories of Artist, Musician, Band,
Theater and the factor of Musical Arts were negatively associated with intention to
smoke. Identification with the categories Rebel, Partier and Hipster and the SISC factor
of Cultural Rebel were associated with increased intentions to smoke.
Preliminary analysis indicated that some of these identity categories were
correlated with attitudes, norms and self efficacy beliefs about smoking. For instance,
strength of identification with the Musician category was positively related to self
efficacy to refuse a cigarette. However, the findings indicate that while certain social
categories may be related to attitudes, norms and self efficacy beliefs about smoking,
there is still an independent effect of social identity on intention to smoke. This indicates
that while identification with a social category may be related to attitudes, norms and self

136

efficacy beliefs (insofar as they are prototypical of that social category), this
identification impacts smoking behavior above and beyond these relationships.
Interestingly, of the attitudes, social norms and self efficacy variables, only percentage of
friends who smoked and self efficacy to refuse a cigarette were associated with intention
to smoke. This finding diverges from previous research that has found all three factors
significantly predict intention to smoke (Ter Doest, et al., 2007; Hanson, 1997; Harakeh
et al., 2004; Macmillin & Connor, 2003). The attitudes about smoking and cigarette
companies that were measured in this survey were not associated with intention to smoke.
While the attitudes measured in this study do not predict intention to smoke, there could
still potentially be different attitudes that do.
Regardless, identification with the social categories of Artist, Musician, Band,
Theater, Musical Arts, Rebel, Partier, Hipster and Cultural Rebel remained a strong
predictor of intention to smoke even after controlling for attitudes, behaviors and norms.
This suggests that identification with the above social categories affects behavioral
intentions to smoke independently of any effect on social norms, attitudes or self efficacy
and that inclusion of these variables significantly improves the predictive ability of the
Integrative Model.

137

Does Identification with Certain Social Categories Contribute to the Social Norms
Approach?
Social norms are a factor that has been found to strongly correlate with smoking
behavior (Brown et al., 2006; Canada, et al., 2005; Eisenberg and Forster, 2003; Hall and
Valente, 2007; Hoffman, et al., 2007; Leatherdale, et al., 2006; Powell, Tauras and Ross,
2005; Taylor, et al., 2004; Unger et al., 2001). This data analysis found that while social
norms were independently associated with increased smoking behavior, they also
interacted with the social categories of Theater, Smart and Achiever to produce an
increased likelihood of smoking behavior (specifically of having tried smoking). This
indicates that the effect of social norms on smoking behavior increases as individuals
more strongly identify as members of these categories. On their own, however,
identification with the categories of Theater and Achiever were associated with decreased
smoking behavior. In other words, when not looking at the interactions of Theater and
Achiever with social norms, it would appear that identification with these groups is
linked to a lower likelihood of engaging in smoking behavior. But, when individuals
strongly identify with these particular groups and perceive a high percentage of their
friends to engage in smoking behavior, the likelihood of the individual smoking
increases.
These findings highlight the variability of group prototypes. According to Social
Identity Theory, group prototypes tend to be shared across group members but are
ultimately individual perceptions of standard norms and behavior (Hogg and Reid, 2006).

138

In other words, the concept of group prototype is similar to that of perceived norm: not
what is the actual standard of behavior for a group, but what the individual perceives that
standard to be. The significance of these interaction effects between identification with
the above mentioned social categories and social norms suggests that for some
individuals who identify with the categories of Theater, Smart and Achiever, smoking
may be a prototypical behavior, as indicated by the increasing effect of social norms.
While identification with these categories may typically be associated with decreased
smoking behavior, there may be some individuals who identify with these categories and
who find smoking behavior to be prototypical. For these individuals, as identification
with the category increases, so does the influence of the perceived norm. These results
support prior research that found that individuals are more likely to perform a behavior
when that behavior is normative or prototypical for an identity to which they subscribe
(Hogg and Reid, 2006; Sparks, 2000; Terry & Hogg, 1996). Ultimately, the
incorporation of social identity into the Social Norms Approach could be useful, both for
social identity’s independent effect on smoking behavior, and also for its interaction with
social norms, allowing researchers to better understand when the norms of a group will
influence smoking behavior.

139

Does Identification with Certain Social Categories Increase the Effects of the truth™
Campaign?
Exposure to the truth™ campaign has been statistically associated with more
negative attitudes toward smoking and decreased smoking behavior (Farrelly et al., 2002,
2005). As previously described, the truth™ campaign uses various techniques to expose
the lies of tobacco companies. Although the designers of the campaign attempted to
portray edgy youth, viewers of the ad felt that the people in the ads predominantly
belonged to the social categories of Smart, Nerd and Athlete. This indicates that the ads
were perhaps not as cutting edge as the designers had hoped. Regardless, there were
several interaction effects between exposure and identification with certain social
categories. Data analysis found that exposure to the truth™ campaign interacted with
identification as a Techie, Nerd and Musical Artist to produce a decreased likelihood of
being susceptible to smoking. Exposure to the truth™ campaign interacted with
identification as a Goody-goody and as a Conservative to produce an increased likelihood
of being susceptible to smoking.
The interaction of exposure with identification as a Techie, Nerd and Musical
Arts indicates that while the truth™ campaign may have been ineffective across the board
at reducing smoking behavior, it did result in decreased smoking behavior for individuals
who strongly identified with these social categories. This indicates that there was
something unique to the truth™ campaign that appealed to Techies, Nerds and Musical
Arts. Pre-testing found that individuals who watched the ads predominantly categorized

140

the actors as Nerds, Smart and Athletes. Moreover, the truth™ ads currently airing have
similar actors and features that might make them appeal to Nerds, Techies and Musical
Arts. For example, one series of ads currently on air is called “The Sunny Side of
truth™.” These ads feature young adults singing and dancing in a way that mocks
tobacco companies. The young adults in these ads seem to fit into the Musical Arts
category and as such, this campaign positions rebelling against tobacco companies and
not smoking as prototypical for this category. Similarly, the truth™ website has many
features that would appeal to someone interested in technology. For instance, the website
features games, online messaging, a blog and videos. One of the most prominent videos
has a theme that resembles Star Wars. The website also addresses a technologically
savvy user (for instance a note that reads “…If you don’t know what YouTube is, you
probably live under a rock”). By treating its users as Techies, it simultaneously ascribes
not smoking as prototypical for that group.
On the other hand, one series of truth™ ads features individuals engaging in
rebellious behavior, the mildest of which resembles public art and captures the attention
of people on the street and the most outrageous of which resembles a public protest. It is
possible that people who identify as Goody-goody and Conservative understood the
young adults in the truth™ campaign ads as wild rebels or lawbreakers – identities quite
dissimilar to those of Goody-goody or conservative. Thus, for these individuals, the
truth™ campaign may have positioned the act of not smoking as something prototypical
for rebels or outlaws. According to Social Identity Theory, social identities are

141

understood in contrast to one another. It is likely that the Goody-goody and conservative
identities are understood in contrast to the rebel or outlaw identities present in this subset
of truth™ campaign ads. This may explain why exposure to the truth™ campaign was
associated with increased susceptibility to smoke for Goody-goody and Conservatives.
Individuals who identified as with these groups and who were also exposed to the truth™
campaign became more likely to contrast their identity with those featured in the ads. If
the ads depicted not smoking as prototypical of a rebellious identity, and Goody-goodies
or conservatives felt different, or opposite to that identity then it is easy to see how the
truth™ campaign failed to decrease smoking behavior in these two groups.
Overall, though, this study did not find exposure to the truth™ campaign to have
the same results documented by Farrelly and colleagues (2002; 2005; 2008). There are
two potential explanations for the failure to find a wide range of significant effects. First,
the exposure measure was relatively simplistic and may not have captured respondents
actual exposure to the campaign Because of the nature of this measure, which only asked
respondents to recall whether they had seen the campaign in the past 30 days, a dose-
response analysis was impossible. Second, the truth™ campaign has been airing heavily
for quite a few years and may have reached saturation. It is quite possible that those who
indicated they had not seen the ad in the past 30 days had previously been exposed to the
ad. This issue will be further discussed in the Limitations section. One additional caveat
that may account for these different results is that the researchers (e.g. Farrelly et al,
2002; 2005; 2008) who found the truth™ campaign to be effective measured only beliefs,

142

attitudes and behavioral intention to smoke whereas the current research measured actual
smoking behavior and found no direct relationship between exposure to the truth™
campaign and smoking behavior. However, results of the current study did find that the
ad was effective in reducing smoking susceptibility for members of the Techie, Nerd and
Musical Arts social category. This indicates that the campaign had features that appealed
specifically to adolescents with those social identities and highlights the utility of identity
in ad design and targeting (which will be discussed further in the Practical Contributions
section).

Summary of Results
Overall, there were several social categories associated with smoking behavior;
however these associations manifested themselves in different ways. For example, while
identification with the categories of Artist, Involved, Skaters, Religious, Partier, Hipster
and Average all contributed significantly to the base model, their effects were no longer
significant when controlling for the attitude, social norms and self efficacy variables in
the integrative model. This suggests that the effect identifying with these social
categories had on smoking behavior may be due instead to attitudes, social norms or self
efficacy. Alternately, other social categories were only related to smoking behavior when
another factor was present. For instance, identifying with the social category of Popular
was only related to smoking when an individual was high in sensation seeking.
Similarly, identifying with the social category of Goody-goody was only related to

143

smoking behavior when the individual had been exposed to the truth™ campaign. These
findings indicate that strongly identifying with these social categories makes one
predisposed to also be strongly affected by another variable. For example, identifying as
a Goody-goody may not affect smoking behavior on its own, but when people who
strongly identify as Goody-goodies are exposed to the truth™ campaign, they become
much more likely to engage in smoking behavior. Finally, some social categories such as
Musicians and Theater had independent effects on smoking behavior and also interacted
with other variables to produce magnified effects. Identification as a Musician was
negatively related to smoking behavior and also interacted with sensation seeking to
impact smoking behavior, such that those who strongly identified as Musicians and who
were also low sensation seekers were much less likely to engage in smoking behavior.
Table 5.1 summarizes the specific mechanisms through which the social categories
affected smoking behavior (only social categories with significant relationships reported,
refer to Appendix B for non-significant findings).

5.1. Summary of Effects of Identification with Each Social Category on Smoking
Behavior .

Contributed to
Base Model
Interacted
with
Sensation
Seeking
Contributed to
Integrative
Model
Interacted
with Social
Norms
Interacted
with
Truth™
Campaign
Exposure
Artist
√
-- -- -- --
Involved
√
-- -- -- --
Musicians
√ √ √
-- --

144

Nerds
√
-- -- --
√
Theater
√
-- --
√
--
Skaters
√
-- -- -- --
Religious
√
-- -- -- --
Rebels
√
--
√
-- --
Hippie
√
-- -- -- --
Partiers
√
-- -- -- --
Hipsters
√
-- -- -- --
Average
√
-- -- -- --
Popular
--
√
-- -- --
Smart
--
√
--
√
--
Techie
--
√
-- --
√
Nonconformist
--
√
-- -- --
Goody-goody
-- -- -- --
√

Theoretical Contributions
This study contributes to the literature on peer crowd affiliation and smoking
behavior by incorporating Social Identity Theory. The results detailed above indicate
that there is a social identity component involved in smoking behavior. Social Identity
Theory offers a framework through which to interpret these findings. According to
Social Identity Theory, membership in a social category is more likely to influence
behavior (1) when it is diagnostic for that behavior, (2) when the social category is

145

desirable to the individual and (3) when the individual is in an uncertain situation (Hogg
and Turner, 1987; Reed and Forehand, 2003). These three notions can be used to explain
why identification with certain social categories did or did not predict smoking behavior.
The diagnosticity of the social categories in this study was evaluated by asking
individuals to list social categories for which smoking was prototypical and social
categories for which not smoking was prototypical. It was found that several, though not
all, of the social categories associated with smoking and not smoking were predictive of
those respective behaviors. The results of this study support the notion that diagnosticity
is an important factor in social identity’s value in behavioral prediction. At the same
time, there were several social categories that were not statistically related to smoking
behavior, even though pre-testing determined they were diagnostic identities. This
illustrates the variability of social identity. As noted previously, although group
prototypes are generally shared by the members of a group, they are ultimately individual
perceptions of that group’s standard behavior. Additionally, group prototypes can vary
by region, gender, age, etc. – for example, the prototype for the Popular category in
California may include playing volleyball and wearing expensive flip flops while the
prototype for the Popular category in New York may include playing lacrosse and
wearing an expensive coat. The data collected for this study suggest that a similar
situation may be at play. One group of individuals was given a pre-test to identity social
categories that they thought were diagnostic of smoking and not smoking. However,
survey results revealed that identification with only about half of those social categories

146

actually affected smoking behavior. It is possible that this occurred because the pre-test
sample was a group of older adolescents in Los Angeles while the survey sample was a
group of younger adolescents from a wide variety of regions. The diagnostic social
categories given by the pre-test respondents, then, may have only been diagnostic for
those living in Los Angeles, while the more geographically diverse survey sample may
have felt there were other social categories that were diagnostic for smoking or not
smoking.
The role of desirability can also be used to explain why those identities predicted
smoking behavior. Insofar as an identity is desirable to an individual, he or she will be
more likely to enact that identity by performing prototypical behavior. Thus, when an
identity is diagnostic for smoking or not smoking and is a desirable identity to the
individual, he or she is more likely to exhibit either decreased or increased smoking
behavior in accordance with the identity’s prototype. Although this study did not
measure the desirability of each social identity, the fact that so many individuals behaved
as prescribed by the group prototype indicates that these identities were also desirable for
these individuals.
Finally, social identity is more strongly associated with behavior when the
individual is in an uncertain social situation. This notion can be used to describe
temporary social settings (such as a party) where an individual may feel uncertain of
social norms, but can also be used to describe broader time periods in an individual’s life
when he or she lacks a sense of security or certainty about his or her place in society.

147

Adolescence is one such time where individuals struggle to determine who they are apart
from their families and in the context of their peers (Blos, 1975; Seltzer, 1982).
Adolescents often try on a series of identities before finally committing to one or more
that can be used as a framework to understand the self (Erikson, 1959; Marcia, 1966).
Because of this uncertainty, the various identities that an adolescent tests and finally
commits to can be particularly effective at influencing behavior.
More broadly, Social Identity Theory is hypothesized to influence behavior
because it serves a “self-definitional function” (Hogg & Reid, 2006, pp. 12). Individuals
act according to group prototypes because it allows them to perform who they are and in
doing so, solidify their identity. When an individual identifies as a member of a social
category, understands that category’s prototype and ascribes features of that prototype to
oneself (self-stereotyping), the social category serves an organizing function whereby the
actions of the individual are understood in the context of this identity. By acting in
accordance with this identity, the individual is able to re-establish this identity (Tajfel &
Turner, 1986). Additionally, acting out prototypical behavior serves to reaffirm and
communicate one’s social identity to others (Emler & Reicher, 1995). When an
individual enacts a prototypical behavior, he or she is using that behavior to communicate
who he is to others. Thus, when an adolescent who strongly identifies as a Rebel lights a
cigarette, he or she is not only reaffirming the Rebel identity to him or herself, but also
using the cigarette as a cue to signal his or her identity to others.

148

Practical Contributions
It is here, in contextualizing smoking as a performance and a communicative
behavior, that social identity contributes to understanding the effect of identification with
peer crowds on smoking behavior. The results of this study corroborate those researching
peer crowd affiliation and smoking (e.g. Downs & Rose, 1991; La Greca, Prinstein &
Fetter, 2002; Sussman, et al., 2004, 2007) and social identity provides a useful theoretical
framework through which to interpret and apply them. Specifically, Social Identity
Theory allows for an understanding of smoking and not smoking as identity-based
behaviors. This is not to marginalize the roles of other smoking predictors, some of
which were found to have significant effects in this study. Rather, by looking at smoking
as an identity-based and communicative behavior, campaign designers can use identity
appeals in the creation of anti-smoking ads.
According to SIT, individuals adhere to behavior that is prototypical of social
categories with which they identify (Hogg & Abrams, 1999; Turner & Hogg, 1987). In
doing so, these individuals are performing their identity as members of this social
category (Hogg and Abrams, 1999). The current study has found considerable support
for the notion that smoking is a performance of identity (specifically, Partier, Rebel,
Skater and Hippie). Similarly, the act of not smoking is also a performance of identity
(specifically, Artist, Musician, Theater, Involved and Nerd). Insofar as the behaviors of
smoking and not smoking are motivated by identity, health practitioners need to take this
into account when designing anti-smoking ads. Using the work of Basu and Wang

149

(2009) and contextualizing the argument within consumer culture theory (see Arnould &
Thompson, 2005), this dissertation argues that branding anti-smoking campaigns and the
act of not smoking is the key way to leverage the findings that identity is directly related
to smoking behavior. This section first addresses consumer culture theory that positions
goods as resources for identity construction and performance and discusses branding as
the way through which this positioning occurs. Next, cigarettes as branded consumer
goods are discussed and it is argued that the act of not smoking must be viewed in a
similar vein. Finally, an evaluation of how this framework relates to the current study’s
findings and how they can be applied to anti-smoking campaigns is presented.
Experts in consumer culture and advertising have long understood behavior –
specifically the purchase and use of products – as motivated by and based in identity
(Leiss, Kline, Jhally & Botterill, 2005). SIT posits that we act in accordance with the
prototypes for the groups to which we belong and consumer culture theorists understand
that the purchase of various products is an important prototypical behavior, viewing the
sphere of consumption as a marketplace of tools (i.e. products) for the establishment and
communication of identity (Arnould & Thompson, 2005). The use of products to
perform identity has been described as follows: As traditional sources of identity, such as
work, gender, family and ethnicity become more unstable, U.S. citizens are increasingly
using products as substitute. In this way, products help consumers to understand who
they are and to communicate their identity and its corresponding values to others (Cross,
2000; Holt, 2002; Muniz & O’Guinn, 2001). Consumer products, in short, have become

150

the physical manifestations of the cultural principles and identity categories through
which we organize society (McCracken, 1986; Sherry, 1987).
The way in which products become associated with certain social categories is
through advertising and specifically, branding (McCracken, 1986). By definition, a brand
is “a name, term, sign, symbol, or design, or combination of them which is intended to
identify … goods and services” (Kotler, 1991, p. 442). Not since the 1940’s has reason-
why advertising been favorable; instead advertisers have increasingly used branding to
sell products by targeting specific market segments or social categories by using identity
cues to align the product with the consumers’ interest and lifestyle (Sivulka, 1998).
Cigarettes in particular have been branded extensively. As early as the 1920’s, tobacco
companies concentrated on creating a brand image that would distinguish their product
from the others. For example, Lucky Strike branded their cigarette as fashionable,
respectable and classy – ideal for a woman who could smoke with style (Sivulka, 1998).
On the other hand, the creation of the Marlboro man by Phillip Morris effectively
associated Marlboro cigarettes with masculinity. This brand image was created in the
1950’s, with the goal of rebranding Marlboro (a cigarette previously associated with
women) as manly, tough and rugged. The identity of the “Marlboro Man” – a handsome
yet tough cowboy figure – was the personification of this identity and Marlboro cigarettes
were his brand of choice (Kellner, 1991).
Due to the way cigarettes have been branded, understanding the relationship
between identity and consumption is particularly important for anti-tobacco researchers.

151

While smoking may be a health behavior, it is also undeniably a consumer behavior. For
decades, tobacco companies have spent billions of dollars branding their product,
strategically constructing smoking as the ideal way to perform a variety of identities. To
illustrate, in 2005, the five largest cigarette companies in the U.S. spent a combined sum
of $13.11 billion dollars on advertising and promotion (Federal Trade Commission,
2007) and Americans spent approximately $82 billion dollars purchasing cigarettes
(Capehart, 2008).
Because of this, anti-smoking campaigns face a particularly uphill battle. Anti-
smoking advocates are forced to battle tobacco companies who have spent decades and
billions of dollars creating cigarette advertisements that have strategically constructed
desirable brand images for smoking. However, it has recently been argued that the very
same branding techniques that cigarette companies use to sell their product may be useful
to anti-smoking campaigns. In recent years, there have been an increasing number of
arguments that health needs to be understood in the context of the consumer sphere (Basu
& Wang, 2009). Consumers, particularly adolescents, are bombarded with an astounding
amount of persuasive messages. Some estimate that the average teenager is exposed to
over 3,000 advertisements in one day (Schwartz, 2004). These advertisements use
sophisticated techniques to sell products to consumers, leveraging values, identities and
lifestyles in order to align a product with the interest of the consumer (Reynolds &
Gutman, 2001). Health campaigns – particularly anti-smoking campaigns – need to
adopt these same strategies if they are to encourage the consumer to adopt a health

152

behavior (Howgill, 1998). Basu and Wang (2009) argue that branding is one such
technique that health campaign designers need to adopt if they are to compete in the
saturated advertising world. Branding, in the case of health marketing, is the process of
establishing an identity or image for a behavior (Aaker, 1997) so that behavior becomes
meaningful to consumers. Just as cigarettes have become identity cues, that is, more than
just objects to smoke, the behavior of not smoking must similarly be branded so that
adolescents feel their decision to not smoke represents something crucial about their
identity.
Traditionally, anti-tobacco campaigns directed towards adolescents have focused
more on rational tactics. These campaigns focused on the long-term health effects of
smoking or more immediate negative consequences such as bad breath. Campaigns also
focused on teaching adolescents about peer pressure and strategies to refuse a cigarette
(Farrelley, Niederdeppe & Yarsevich, 2003). More recently, however, campaigns have
focused on denormalizing smoking and on exposing tobacco companies’ manipulative
tactics (Farrelley, Niederdeppe & Yarsevich, 2003). These newer tactics are more in line
with strategies of branding and attempt to create an “edgy” and “cutting edge” identity
that is associated with not smoking (Farrelly et al., 2002, pp. 901). Indeed, Basu and
Wang (2009) cite the truth™ campaign as an exemplar of a successful health behavior
branding campaign.
The current study found considerable evidence for smoking and not smoking as
behaviors motivated by identity. Because of the impact social identity has on smoking

153

behavior, it can be used as a unique tool to create anti-smoking campaigns and target at-
risk audiences. Future anti-smoking campaigns can use social identity to target and
persuade adolescents in several ways. To begin with, social identity can be used to target
at risk groups. Results of the current study indicate that smoking is prototypical for
several social identities, such as Rebels, Partiers and Hipsters and that the more an
individual identifies as such, the more likely he or she is to behave according to the
prototype and smoke. These individuals are at higher risk than those who identify as
Musicians or Artists, for instance, and should be more heavily targeted. Campaign
designers can use media research data containing information about audience
psychographics and media use to determine which media sources these at-risk youth are
using and then advertise in those sources more heavily.
Social identity can also be extremely valuable for the design of campaign content.
As stated, adolescents who identify with certain social categories are more likely to
engage in smoking behavior. Particularly in the case of those who identify as Rebels,
Hipsters and Partiers, this use of cigarettes to communicate identity is likely due to a long
history of cigarette advertising that has branded smoking as cool, glamorous and
rebellious. In targeting these youth, the task of anti-smoking advertisements is to de-
brand cigarettes and re-brand the act of not smoking. Anti-smoking campaigns need to
devalue cigarettes of their ‘cool’ connotations while positioning the act of not smoking as
something desirable to at-risk groups. The truth™ campaign is one example of
advertisements that have tried to align the act of not smoking with cool and rebellious

154

identities. In other words, anti-smoking campaigns need to remove the act of smoking
from these groups’ prototypes and replace it with the act of not smoking.
Another identity-based tactic anti-smoking campaigns can utilize is to decrease
the desirability of identities associated with smoking while increasing the desirability of
identities associated with not smoking. This is a considerably harder task, as the Rebel,
the Hipster and so on have long been culturally desirable icons (Holt, 2004). However,
the advertising industry has shown a considerable amount of influence over what is and is
not fashionable, so this type of campaign is feasible, particularly if a variety of media
outlets participate. The intended effect of this campaign would be to reduce the number
of individuals who adopt these smoking-related identities which, in turn, would reduce
the number of smokers. Additionally, because individuals are less likely to perform a
group prototype when the group is undesirable, there would be reduced smoking among
those who still adopted these identities.
In sum, campaign practitioners need to create a positive brand for the act of not
smoking, similar to the way Nike, Apple, American Apparel or any company
successfully marketing to adolescents has done. If goods are how individuals understand
themselves and construct and enact identities, then campaigns (1) must portray cigarettes
in such a way that they lose their cache for identity construction and (2) position not
smoking as a branded behavior that communicates information about the actor. Granted,
because there is no associated product – merely the absence of one – branding the act of
not smoking becomes more difficult. But, by positioning not smoking as its own

155

behavior (and not just the lack of one), with corresponding desirable identities, anti-
smoking practitioners can encourage adolescents to not smoke.
One caveat to this approach is that it requires a significant amount of coordination
on the part of campaign designers. On any given day, an adolescent may be subjected to
several incongruent messages from different sources, persuading him or her not to smoke
(Basu & Wang, 2009). The strategies leveraged by these campaigns differ and while the
intentions of the campaign practitioners is good, it simply leads to confusion and
produces a fragmented image for the behavior of not smoking. It is crucial that a unified
brand be presented so that adolescents nationwide know what smoking and not smoking
means. Additionally, this kind of campaign must be heavily pre-tested. As discussed,
although the truth™ campaign attempted to present a cool and rebellious identity, most
individuals in the current study identified the people in the ads as Smart or Nerds. Thus,
campaign designers must devote considerable resources to testing ads to ensure the
intended message is the received message. Working together with teenagers to design
campaigns will also lead to more effective ads. Although labor intensive, this sort of
testing and collaboration will produce stronger and more effective anti-smoking
campaigns.

Methodological Contributions
As noted previously, social identity and peer affiliation have been measured in a
myriad of ways. Generally, these measures neglect the fact that adolescents may identify

156

with multiple groups in varying levels. Additionally, traditional measures typically do
not account for the many nuanced social categories with which adolescents identify.
Finally, these measures do not account for group prototypes – that is, the extent to which
a social category is diagnostic of smoking behavior. The current project developed a
more detailed procedure to measure strength of an adolescent’s identification with a
social category.
The first step of this process involved a pre-test, whereby a group of 61
individuals were asked to identify social categories or groups for which the behaviors of
smoking and not smoking were prototypical. These responses were then grouped into 26
social categories, some of which were associated with smoking and some of which were
associated with not smoking. Next, a separate sample of 224 individuals were surveyed
and asked to indicate, on a scale of 1 to 100 how much they identified with each of these
26 social categories. These responses were standardized within each individual to
provide an accurate representation of how much he or she identified with each social
category in comparison to all others. This measure is also versatile in that factor analysis
was used to reduce the 26 social categories into eight factors, several of which also
predict smoking behavior.
This process contributes to the literature on social identity and peer crowd
affiliation by providing a nuanced measure of how much an individual identifies with a
social category. First, the current measure accounts for a wide range of social categories
generated by adolescents themselves. Most importantly, this measure accounts for the

157

fact that individuals may identify with more than one social category. Previous research
(e.g. Down & Rose, 1991; LaGreca, Prinstein & Fetter, 2002; Sussman, Unger & Dent,
2004) asked adolescents to indicate the one group they most identified with. The current
research advances that methodology by allowing individuals to indicate the extent to
which they identify with each of 26 social categories. While providing a measure of
which category an individual identifies with most, this method has the additional
advantage of providing data that shows how much an individual identifies with one social
category compared to all others. This measure also provides data that allows researchers
to understand which social categories tend to cluster together. In the current research,
exploratory factor analysis determined eight factors into which the 26 social categories
could be collapsed. These factors provide a more parsimonious conceptualization of
social identity, and several factors were found to predict smoking behavior. While an
understanding of the 26 social categories and how they individually affect smoking
behavior offers the most detailed understanding of the relationship between social
identity and smoking behavior, the eight factors can be useful to researchers and
practitioners who are looking for a more economical or efficient approach.
Finally, this measure of social identity contributes to previous methodology by
accounting for group prototypes. Reed and Forehand (2003) argue that a social identity
will only influence behavior for which it is diagnostic. In other words, if a behavior is
not a part of a group’s prototype, then researchers should have no reason to expect that
identity to influence that behavior. Because of this, when studying the relationship

158

between social identity and smoking behavior, it is crucial to focus on identities for
which the act of smoking or its obverse, not smoking, is prototypical. In the case of the
current measure, each of the 26 social categories was selected because pre-testing
identified it as an identity for which smoking or not smoking was prototypical.
Accounting for identity diagnosticity in pre-testing increases the likelihood that the social
categories in question will actually predict smoking behavior.

Study Limitations
Although this study found several significant results that could lead to potentially
successful interventions, the reader should keep in mind several limitations when
interpreting the findings. To begin with, the sample posed several issues. Potential
respondents received an e-mail which allowed them to link to the survey. The first 250
individuals to complete the survey were included in the sample. Thus, the sample size of
the study was relatively small (N = 224). While power analyses revealed adequate power
to detect effects of .10, the sample was not large enough to adequately detect smaller
effects. Additionally, the small sample resulted in less than ideal size for certain
subgroups (for instance, only 29 African Americans). The sample drew from a nation-
wide sampling frame that was demographically similar to the population to which results
are to be generalized. While this may increase the generalizability of the results, an
analysis of the findings revealed that group prototypes of certain social categories may
vary. Geographic data was not collected, so it is not certain the extent to which this

159

variability was due to geographic difference. Several geographically concentrated
samples would be useful for probing location as a source of variation. Additionally, data
on parents’ income was not collected, and so the notion that variation in-group prototypes
was due to class differences cannot be ruled out. Thus, two potential recommendations
are made. First, this study should be replicated with a larger sample size that would
ensure detection of significant effects. Second, this study should be replicated in several
geographically concentrated areas, with income data collected. While it is possible that
there will still be considerable variation within concentrated locations and socioeconomic
classes, this process would allow the researcher to determine the extent to which group
prototypes vary as a function of location and class.
Additionally, the sample was not truly random. The survey for this study was
conducted online using a random sample from a large demographically representative
sampling frame. This method was chosen for its cost-efficiency and likelihood to
increase response rate. The online survey modality was determined to be acceptable
given the fact that 93% of teenagers have Internet access (Lenhart et al., 2008); however
the researcher is cognizant of the fact that by not incorporating the 7% of teenagers who
are not online, the sample may not be fully representative.
Next, due to the large number of analyses conducted, there is a possibility that
certain statistically significant results were due to chance. In defense, however, the study
was largely explorative in nature and evaluated many variables that had not been looked
at before. Moreover, each regression was necessary to determine the effect of the social

160

identity variable. Future research in this area can use the findings of this study to conduct
a more focused series of analyses using far fewer variables and statistical tests.
This study employed a pre-test designed to identify which social categories were
associated with smoking and not smoking. However, it was found that some categories
identified in the pre-test as being associated with either smoking or not smoking did not
have this same association in the actual survey. This speaks to the notion of a group
prototype as an ultimately individual perception of what is normative for a social
category. Future research should ask the survey respondents themselves to indicate
whether they thought each social category was associated with smoking, not smoking or
neither.
The current study was also limited by its measure of exposure to the truth™
campaign. The measure employed was developed by the American Legacy Foundation
for use in their ongoing tracking survey to measure the effects of the truth™ campaign.
Farrelly and colleagues (e.g. Farrelly, Davis, Duke & Messeri, 2008) used this measure
and found exposure to the truth™ campaign was related to negative beliefs and attitudes
toward smoking and decreased intention to smoke. However, the majority of this data
was collected between 2000 and 2003, when the truth™ campaign was still relatively
new on a nationwide scale. These researchers were likely not faced with issues of
message saturation. In the current study, data was collected in 2008 and adolescents were
much more likely to have lifetime exposure to the truth™ campaign, even if not in the
past 30 days. Because of this, the employed measure of exposure did not capture

161

variation between those who had never seen the campaign, who had been mildly exposed
to the campaign and who had been heavily exposed to the campaign. Future research
should develop a more nuanced measure of exposure so that a dose-response relationship
can be evaluated. Additionally, use of an experiment where certain individuals are
shown campaign ads while others aren’t could lend support to survey data examining the
relationship between exposure to the truth™ campaign and smoking behavior.
Finally, there are several issues that, while interesting, were not within the scope
of this project. For instance, there are many other ways in which social identity might
affect smoking behavior. The scope of this dissertation was to evaluate its contribution to
the Sensation Seeking Approach, the Social Norms Approach, the Integrative Model of
Behavioral Prediction and its interaction with exposure to the truth™ campaign. Other
interactions were not tested for parsimony’s sake but should be considered for future
research (Bagley, White and Golomb, 2001).
Additionally, while the finding that social identity did predict smoking behavior
has significant implications for campaign design, it raises the question of how social
identities are adopted. Why do some individuals identify with the category of Rebel or
Skater, while others identify with the category of Nerd or Techie? Although this issue
has gone largely unaddressed by social identity theorists, its answer is one that can bring
new insight to the study of adolescent smoking behavior. By understanding the process
through which identities are adopted, anti-smoking practitioners can better design
campaigns that persuade individuals to adopt identities for which not smoking is

162

prototypical. Longitudinal studies examining how adolescents try and eventually settle
on an identity will be useful in understanding this issue.

Conclusion
This study found that social identity had significant effects on adolescent smoking
behavior. Analyses revealed that social identity works to independently affect smoking
behavior, but also interacts with factors such as sensation seeking and social norms to
produce increased effects. In other words, while social identity direct impacts smoking
behavior it also works with other variables to produce increased effects. These findings
underscore the importance of understanding smoking as not just a health behavior, but
also a consumer behavior that is based in identity. Goods and products, including
cigarettes, are constantly marketed towards adolescents for the identities they represent.
This study offers support for the argument that, with the help of marketers, the use of
goods is one important way that adolescents make sense of their worlds. As such, we
must look at smoking as a meaningful and communicative behavior that speaks volumes
about how an adolescent sees him or herself and how he or she wants to be seen. Health
practitioners and campaign designers need to come together to produce a unified brand
image for the act of not smoking and work to devalue the identity associations of
smoking. Doing so will increase the success of anti-smoking media campaigns by
providing adolescents with new and positive ways to construct their identities.

163

References

Aaker, D.A. (1996), Building Strong Brands, The Free Press, New York, NY, .

Abdi, H. (2003). Factor rotations in factor analyses. In M. Lewis-Beck, A. Bryman & T.
Futing (Eds.) Encyclopedia of social science research methods (pp. 978-982). Thousand
Oaks, CA: Sage.

Abrams, D. & Hogg, M.A. (1999). Social identity and social cognition. Malden, MA,
US: Blackwell Publishing.

Aiken, L.S., & West, S.G. (1991). Multiple regression: Testing and interpreting
interactions. Thousand Oaks, CA: Sage.

Ajzen, I. (1988). Attitudes, personality, and behavior. Milton Keynes: Open University
Press.

---- (1991). The theory of planned behavior. Organizational behavior and human
decision processes, 50, 179-211.

Ajzen, I. & Fishbein, M. (1977). Attitude-behavior relations: A theoretical analysis and
review of empirical research. Psychology Bulletin, 84(5), 888-918.

---- (1980). Understanding attitudes and predicting social behavior. Englewood Cliffs,
NJ: Prentice Hall.

Akers, R. L. (1977). Deviant Behavior: A social learning approach. 2nd edition.,
Belmont, CA: Wadsworth.

American Cancer Society (2008). Cigarette smoking. Available at:
http://www.cancer.org/docroot/PED/content/PED_10_2X_Cigarette_Smoking.asp

Arnould, E.J. & Thompson, C.J. (2005). Consumer culture theory (CCT): Twenty years
of research. Journal of Consumer Research, 31(4), 868-882.

Asbridge, M., Tanner, J., & Wortley, S. (2005). Ethno-specific patterns of adolescent
tobacco use and the mediating role of acculturation, peer smoking, and sibling smoking.
Addiction, 100, 1340-1351.

Ashmore, R.D., Del Boca, F.K., & Beebe, M. (2002). “Alkie,” “Frat Brother,” and
“Jock”: Perceived types of college students and stereotypes about drinking. Journal of
Applied Social Psychology, 32, 885-907.

164

Audrain-McGovern, J., Tercyak, K.,, Shields, A., Bush, A., Espinel, C.F., & Lerman,
C. (2003). Which adolescents are most receptive to tobacco industry marketing?
Implications for counter-advertising campaigns. Health Communication, 15(4), 499-513.

Audrain-McGovern, J., Rodriguez, D., Tercyak, K.P., Cuevas, J., Rodgers, K., &
Patterson, F. (2005). Identifying and characterizing adolescent smoking trajectories.
Cancer Epidemiology, Biomarkers & Prevention, 13(12), 2023-34.

Bagley, S.C., White, H., & Golomb, B.A. (2001). Logistic regression in the medical
literature: Standards for use and reporting, with particular attention to one medical
domain. Journal of Clinical Epidemiology, 54(10), 979-985.

Bagozzi, R.P., & Lee, K. (2002). Multiple routes for social influence: The role of
compliance, internalization, and social identity. Social Psychology Quarterly, 65(3),
226-247.

Balbach, E.D., & Glantz, S.A. (1998). Tobacco control advocates must demand high-
quality media campaigns: The California experience. Tobacco Control, 7, 397-408.

Baillie, L., Lovato, C.Y., Johnson, J.L, & Kalaw, C. (2005). Smoking decisions from a
teen perspective: A narrative study. American Journal of Health Behavior, 29(2), 99-
106.

Bandura, A. (1977). Social learning theory. Englewood Cliffs. New Jersey: Prentice-Hall.

---- (1986). Social foundations of thought and action: A social
cognitive theory. Engiewood Cliffs. NJ: Prentice Hall.

Barber, B.L., Eccles, J.S., & Stone, M.R. (2001). Whatever happened to the Jock, the
Brain, and the Princess? Young adult pathways linked to adolescent activity involvement
and social identity. Journal of Adolescent Research, 16(5), 429-455.

Basu, A. & Wang, J. (2009). The role of branding in public health campaigns. Journal of
Communication Management, 13(1), 77-91.

Bat-Chava, Y. (1994). Group identification and self-esteem of deaf adults. Personality
and Social Psychology Bulletin, 20(5), 494-502.

Berger, J. & Rand, L. (2008). Shifting signals to help health: Using identity signaling to
reduce risky health behaviors. Journal of Consumer Research, 35(3), 509-518.

165

Berkowitz, A.D. (2004a). An overview of the social norms approach. In L. Lederman, L.
Stewart, F. Goodhart & L. Laitman (Eds.), Changing the culture of college drinking: A
socially situated prevention campaign. Cresskill, NJ: Hampton Press.

Berkowitz, A. (2004b). The social norms approach: Theory, research and annotated
bibliography. Available at www.alanberkowitz.com.

Biener, L. (2002). Recall of anti-tobacco ads/Impact of warnings/Advertising. Tobacco
Control, 11(Supplement2), ii43-ii46.

Blos, P. (1975). The second individuation process of adolescence. In A. Esman (Ed.),The
psychology of adolescence: Essential readings. New York: International Universities
Press.

Bond, M.H. & Hewstone, M. (1988). Social identity theory and the perception of
intergroup relations in Hong Kong. International Journal of Intercultural Relations,
12(2), 153-170.

Brener, N.D., Billy, J.O.G., & Grady, W.R. (2003). Assessment of factors affecting the
validity of self-reported health-risk behavior among adolescents: evidence from the
scientific literature. Journal of Adolescent Health, 33(6), 436-457.

Bricker, J.B., Peterson Jr., A.V., Sarason, I.G., Anderson, M.R., & Rajan, K.B. (2007).
Changes in the influence of parents’ and close friends’ smoking on adolescent smoking
transitions. Addictive Behaviors, 32, 740-747.

Brown, S.L., Teufel, J.L, Birch, D.A., Izenberg, N., & Lyness, D. (2006). Perceptions of
tobacco use in early adolescents. The Journal of Primary Prevention, 27(5), 515-526.

Bruvold, W. H. (1993) A meta-analysis of adolescent smoking prevention programs.
American Journal of Public Health, 83, 872–880.

Rundall, T.G., Bruvold, W.H., 1988. A meta-analysis of school-based smoking
and alcohol use prevention programs. Health Education Quarterly, 15(3), 317–334.

Bryant, A.L., Schulenberg, J.E., Bachman, J.G., O’Malley, P.M., & Johnston, L.D.
(2000). Understanding the links among school misbehavior, academic achievement, and
cigarette use: A national panel study of adolescents. Prevention Science, 1, 71-87.

Bryant, A.L., Schulenberg, J.E., O’Malley, P.M., Bachman, J.G. & Johnston, L.D (2003).
How academic achievement, attitudes, and behaviors relate to the course of substance use

166

during adolescence: A 6-year, multiwave national longitudinal study. Journal of
Research on Adolescence, 13(3), 361-397.

Campo, S, Brossard, D, Frazer, M.S., Marchell, T., Lewis, D. & Talbot, J. et al. (2003).
Are social norms campaigns really magic bullets? Assessing the effects of students
misperceptions on drinking behavior. Health Communication, 15(4), 481-497.

Cantor, N., Mischel, W., & Schwartz, J. (1982). Social knowledge: Structure, content,
use and abuse, in A.H. Hastorf & A.M. Isen (Eds). Cognitive social psychology. New
York: Elsevier.

Capehart, Tom. (2008). Expenditures for tobacco products and disposable personal
income, 1989–2005. Compiled from reports of the Department of Commerce, Bureau of
Economic Analysis. Available at
http://www.ers.usda.gov/Briefing/Tobacco/Data/table21.pdf.

Castrucci, B.C., & Gerlach, K.K. (2006). Understanding the association between
authoritative parenting and adolescent smoking. Maternal and Child Health Journal,
10(2), 217-224.

Centers for Disease Control. (2005). Targeting tobacco use: The nation’slLeading
cause of death. Available at:
www.cdc.gov/tobacco/basic_information/00_pdfs/AAGTobacco2007.pdf

Chassin, L., Presson, C.C., Rose, J., Sherman, S.J. Davis, M.J., & Gonzalez, J.L. (2005).
Parenting style and smoking-specific parenting practices as predictors of adolescent
smoking onset. Journal of Pediatric Psychology, 30(4), 333-344.

Christensen, S. & Haines, M. (2003). Communities use a social norms approach to
reduce teen alcohol and tobacco use: Two case studies. In Selected Abstracts from the
2004 National Conference on the Social Norms Model. The Report on Social Norms:
Working Paper #14, Little Falls, NJ: PaperClip Communications.

Cohen, J. (1979). High school subcultures and the adult world. Adolescence, 14(55), 491-
502.

Collins, R.L., & Ellickson, P.L. (2004). Integrating four theories of adolescent smoking.
Substance Use & Misuse, 39(2), 179-209.

Coogan, P.F., Adams, M., Geller, A.C., Brooks, D., Miller, D.R., Lew, R.A., et al.
(1998). Factors associated with smoking among children and adolescents in Connecticut.
American Journal of Preventive Medicine, 15(1), 17-24.

167

Cornelius, M.D., Leech, S.L., Goldschmidt, L., & Day, N.L. (2005). Is prenatal tobacco
exposure a risk factor for early adolescent smoking? A follow-up study. Neurotoxicology
and Teratology, 27(667-676).

Costello, A.B. & Osborne, J.W. (2005). Best practices in exploratory factor analysis:
Four recommendations for getting the most from your analysis. Practical Assessment,
Research & Evaluation, 10(7), 1-9.

Cross, G.S. (2000). An all-consuming century: Why commercialism won in modern
America. New York: Columbia University Press.

Davies, H.T.O., Crombie, I.K., & Tavakoli, M. (1998). When can odds ratios mislead?
British Medical Journal, 316(March), 989-991.

Dent, C. & Biglan, A. (2004). Relation between access to tobacco and adolescent
smoking. Tobacco Control, 13, 334-338.

Dickson, E.S. & Scheve, K. (2006). Social identity, political speech, and electoral
competition. Journal of Theoretical Politics, 18(1), 5-39.

Dinn, W.M., Aycicegi & Harris, C.L. (2004). Cigarette smoking in a student sample:
Neurocognitive and clinical correlates. Addictive Behaviors, 29(1), 107-126.

Dolcini, M.M. & Adler, N.E. (1994). Perceived competencies, peer group affiliation, and
risk behavior among early adolescents. Health Psychology, 13(6), 496-506.

Downs, R.D., & Rose, S.R. (1991). The relationship of adolescent peer groups to the
incidence of psychosocial problems. Adolescence, 26(102), 473-492.

Eckert, P. (1983). Beyond the statistics of adolescent smoking. American Journal of
Public Health, 73(4), 439-441.

Eisenberg, M., & Forster, J. (2003). Adolescent smoking behavior: Measures of social
norms. American Journal of Preventive Medicine, 25(2), 122-128.

Ellickson, P. L., Perlman, M. & Klein, D. J. (2003) Explaining racial/ethnic differences in
smoking during the transition to adulthood. Addictive Behaviors, 28, 913–931.

Elliott. D. S.. Huizinga. D., & Ageton. S. S. (1985). Explaining deiinquent behavior and
drug use. Beverly Hills. CA: Sage.

168

Emler, N., & Reicher, S. D. (1995). The social psychology of adolescent delinquency.
Oxford, UK: Basil Blackwell.

Erikson, E.H. (1968). Identity: Youth and Crisis. New York: Norton.

Fabrigar, L. R., Wegener, D. T., MacCallum, R. C., & Strahan, E. J. (1999). Evaluating
the use of exploratory factor analysis in psychological research. Psychological Methods,
4(3), 272-299.

Farrelly, M.C., Davis, K.C., Duke, J., & Messeri, P. (2009). Sustaining 'truth': Changes in
youth tobacco attitudes and smoking intentions after 3 years of a national antismoking
campaign. Health Education Research, 24(1), 42-48.

Farrelly, M.C., Davis, K.C., Haviland M.L., Messeri, P. & Healton, C.G. (2005).
Evidence of a dose-response relationship between “truth” antismoking ads and youth
smoking prevalence. American Journal of Public health, 95(3), 425-431.

Farrelly, M.C., Healton, C.G., Davis, K.C., Messeri, P., Hersey, J.C., & Haviland, C.G.
(2002). Getting to the truth: Evaluating national tobacco countermarketing campaigns.
American Journal of Public health, 92(6), 901-907.

Farrelley, M.C., Niederdeppe, J., & Yarsevich, J. (2003). Youth tobacco prevention mass
media campaigns: past, present, and future directions. Tobacco Control, 12(Supplement
1), i35-i47.

Federal Trade Commission (2007). Federal Trade Commission Cigarette Report.
Available at: www.ftc.gov/reports/tobacco/2007cigarette2004-2005.pdf

Field, A. (2005). Discovering Statistics Using SPSS. London: Sage Publications, Ltd.

Fishbein, M. (1980). A theory of reasoned action: some applications and implications.
Nebraska Symposium on Motivation, 27, 65-116.

Fishbein, M. & Azjen, I. (1975). Belief, attitude, intention, and behavior: An
introduction to theory and research. Reading, MA: Addison-Wesley.

Fishbein, M. & Cappella, J. (2006). The role of theory in developing effective health
communications. Journal of Communication, 56(s1), S1-S17.

Fishbein, M., Hennessy, M., Yzer, M., & Douglas, J. (2003). Can we explain why some
people do and some people do not act on their intentions? Psychology, Health &
Medicine, 8(1), 3-18.

169

Fishbein, M., & Yzer, M. (2003). Using theory to design effective health behavior
interventions. Communication Theory, 13(2), 164-183.

Forehand, M.R., Deshpande, R., & Reed, A. (2002). Identity salience and the influence
of differential activation of the social self-schema on advertising response. Journal of
Applied Psychology, 87(6), 1086-1099.

Forster, J.L., Widome, R., & Bernat, D.H. (2007). Policy interventions and surveillance
as strategies to prevent tobacco use in adolescents and young adults. American Journal
of Preventive Medicine, 33(6 Suppl), S335-S339.

Frankenberger, K. D. (2004). Adolescent egocentrism, risk perceptions, and sensation
seeking among smoking and nonsmoking youth. Journal of Adolescent Research, 19,
576–590.

Goldman, L.K. & Glantz, S.A. (1998). Evaluation of antismoking advertising campaigns.
Journal of the American Medical Association, 279, 772-777.

Gorusch, R. L. (1983). Factor Analysis (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum
Associates.

Greene, S. (2004). Social identity theory and party identification. Social Science
Quarterly, 85(1), 136-153.

Haines, M.P., Barker, G.P., & Rice, R. (2003). Using social norms to reduce alcohol and
tobacco use in two Midwestern high schools. In H.W. Perkins (Ed.), Preventing school
and college age substance abuse: A handbook for educators, counselors, clinicians. San
Francisco: Jossey-Bass.

Hall, J.A., & Valente, T. (2007). Adolescent smoking networks: The effects of influence
and selection on future smoking. Addictive Behaviors, 32(12), 3054-3059.
Hanson, M.J. (1997). The theory of planned behavior applied to cigarette smoking in
African-American, Puerto Rican, and non-Hispanic White teenage females. Nursing
Research, 46(3), 155-162.

Harakeh, Z., Scholte, R.H., Vermulst, A.A., de Vries, H., & Engels, R. C. M.E. (2004).
Parental factors and adolescents' smoking behavior: An extension of The theory of
planned behavior. Preventive Medicine: An International Journal Devoted to Practice
and Theory, 39(5), 951-961.

170

Hatcher, L. (1994). A Step-by-Step Approach to Using the SAS® System for Factor
Analysis and Structural Equation Modeling. Cary, N.C.: SAS Institute, Inc.

Hauck Jr., W. W. & Donner, A. (1977). Wald’s test as applied to hypotheses in logit
analysis. Journal of the American Statistical Association, 72(360), 851-853.

Hawkins. J. D.. Br Weis. J. G. (1985). The social development model: An
integrated approach to delinquency prevention. Journal qf Primary,
Prevention. 6. 73-97.

Hoffman, B.R., Monge, P.R., Chou, C., & Valente, T.W. (2007). Perceived peer
influence and peer selection on adolescent smoking. Addictive Behaviors, 32, 1546-
1554.

Hogg, M. A. (2003). Social identity. In M. R. Leary & J. P. Tangney (Eds.), Handbook of
self and identity (pp. 462–479). New York: Guilford.

---- (2005). The social identity perspective. In S. A. Wheelan (Ed.), The handbook of
group research and practice (pp. 133–157). Thousand Oaks, CA: Sage.

---- (2006). Social identity theory. In P. J. Burke (Ed.), Contemporary social
psychological theories (pp. 111–136). Palo Alto, CA: Stanford University Press.

Hogg , M.A. & Abrams, D. (1988). Social identifications: A social psychology of
intergroup relations and group processes. London: Routledge.

Hogg, M.A. & Reid, S.A. (2006). Social identity, self-categorization, and the
communication of group norms. Communication Theory, 16, 7-30.

Hogg, M.A. & Turner, J.C. (1987). Social identity and conformity: A theory of referent
informational influence. In W. Doise & S. Moscovici, (eds). Current Issues in European
Social Psychology, (vol. 2). Cambridge, UK: Cambridge University Press.

Hogg, M., Terry, D., & White (1995). A tale of two theories: A critical comparison of
identity theory with social identity theory. Social Psychology Quarterly, 58(4), 255-269.

Holt, D.B. (2002). Why do brands cause trouble? A dialectical theory of consumer
culture and branding. Journal of Consumer Research, 32, 70-89.

---- (2004). How Brands Become Icons: The Principles of Cultural Branding. Cambridge,
MA: Harvard Business School Press.

171

Honig, B., & Kizer, K.(1990). Toward a Tobacco-Free California. Sacramento:
California Department of Health Services..

Howgill, M.W.C. (1998). Health care consumerism, the information explosion, and
branding: why 'tis better to be the cowboy than the cow. Managed Care Quarterly,6(4),
33-43.

Invernizzi, F., Falomir-Pichastor, J.M., Munntoz-Rojas, D., & Mugny, G. (2003). Social
influence in personally relevant contexts: the respect attributed to the source as a factor
increasing smokers’ intention to quit smoking. Journal of Applied Social Psychology,
33(9), 1818-1836.

Kallgren, C. A., Reno, R. R., & Cialdini, R. B. (2000). Focus theory of normative
conduct: When norms do and do not affect behavior. Personality and Social Psychology
Bulletin, 26, 1002–1012.

Kellner, D. (1991). Reading images critically: Toward a postmodern pedagogy. In H.A.
Giroux (Ed.), Postmodernism, feminism, and cultural politics: Redrawing educational
boundaries (pp. 60-82). Albany, NY: SUNY Press.

Kirchler, E., Palmonari, A. and Pombeni, M. L. (199e). Developmental tasks and
adolescents’relationships with their peers and their family. In S. Jackson and T. H.
Rodriguez (Eds.), Adolescence and its social worlds,. Hove, U.K.: Lawrence Erlbaum
Associates.

Kleine III, R., Kleine, S., & Kernan, J. (1993). Mundane consumption and the self: A
social-identity perspective. Journal of Consumer Psychology, 2(3), 209-235.

Kotler, P. (1991) Marketing Management: Analysis, Planning and Control, 8th Edition,
Englewood Cliffs, New Jersey, Prentice Hall, Inc.

Jaccard, J. (2001). Interaction Effects in Logistic Regression. Thousand Oaks, CA:
SAGE.

Jackson, W. & Smith, E.R. (1999). Conceptualizing social identity: A new framework
and evidence for the impact of different dimensions.
Personality and Social Psychology Bulletin, 25(1),120-135.

Janz, N.K., & Becker, M.H. (1984). The health belief model: A decade later. Health
Education Quarterly, 11, 1-47.

Jessor, R. & Jessor, S.L. (1977). Problem behavior and psychosocial development. New

172

York: Academic Press.

Johnson, C.C., Myers, L., Webber, L.S., Boris, N.W. (2004). Profiles of the adolescent
smoker: Models of tobacco use among 9
th
grade high school students Acadiniana
Coalition of Teens against Tobacco. Preventive Medicine, 39, 551-558.

Johnston, L.D., O’Malley, P.M., Bachman, J.G., & Schulenberg, J.E. (2008). Monitoring
the Future national results on adolescent drug use: Overview of key findings, 2007 (NIH
Publication No. 08-6418). Bethesda, MD: National Institute on Drug Abuse.

La Greca, A.M., Prinstein, M.J., & Fetter, M.D. (2001). Adolescent peer crowd
affiliation: Linkages with health-risk behaviors and close friendships. Journal of
Pediatric Psychology, 26(3), 131-143.

Landman A., Ling P.M., Glantz S.A. (2002). Tobacco industry youth smoking prevention
programs: protecting the industry and hurting tobacco control. American Journal of
Public Health, 92, 917–30.

Leatherdale, S.T., McDonald, P.W., Cameron, R., & Brown, K.S. (2005). A multilevel
analysis examining the relationship between social influences for smoking and smoking
onset. American Journal of Health Behavior, 29(6), 520-530.

Leatherdale, S.T., Cameron, R., Brown, K.S., Jolin, M., & Kroeker, C. (2006). The
influence of friends, family and older peers on smoking among elementary school
students: low risk students in high risk schools. Preventive Medicine, 42(3), 218-22.

Leiss, W., Kline, S., Jhally, S. & Botterill, J. (2005). Social Communication in
Advertising: Consumption in the Mediated Marketplace, 3
rd
ed. New York: Taylor and
Francis.

Lenhart, A., Arafeh, S., Smith, A., & Macgil, A. (2008). Writing, Technology and Teens.
Pew Internet and American Life Project. Available at pewinternet.org.

Linkenbach, J.W., & Perkins, H.W. MOST of us are tobacco free: An eight-month
social norms campaign reducing youth initiation of smoking in Montana. In H.W. Perkins
(Ed.) The social norms approach to preventing school and college age substance abuse:
A handbook for educators, counselors, and clinicians. (pp. 224-234). San Francisco, CA:
Jossey-Bass.

Lloyd, B. & Lucas, K. (1998). Smoking in Adolescence: Images and Identities. London
and New York: Routledge.

173

Macmillan, B. & Conner, M. (2003). Using the theory of planned behaviour to
understand alcohol and tobacco use in students. Psychology, Health & Medicine, 8(3),
317-328.

Madden, T.J., Ellen, P.S., & Ajzen , I. (1992). A comparison of the theory of planned
behavior nd the theory of reasoned action. Personality and Social Psychology Bulletin,
19(1), 3-9.

Marcia,J. (1966). Development of validation of ego-identity status. Journal of
Personality and Social Psychology, 3, 551–558.

Martino-McAllister, J., & Wessel, M.T. (2005). An evaluation of a social norms
marketing project for tobaccl prevention with middle, high and college students; Use of
funds from the Tobacco Master Settlement (Virginia). Journal of Drug Education, 35(3),
185-200.

McCracken, G. (1986). Culture and consumption: A theoretical account of the structure
and movement of the cultural meaning of consumer goods. The Journal of Consumer
Research, 13(1), 71-84.

Michell, L., & Amos, A. (1997). Girls, pecking order and smoking. Science & Medicine,
44(12), 1861-1869.

Monitoring the Future. (2007). Trends in Prevalence of Use of Cigarettes for Eighth,
Tenth, and TwelfthGraders Ann Arbor, MI: The University of Michigan

Mosbach, P., & Leventhal, H. (1988). Peer group identification and smoking:
Implications for intervention. Journal of Abnormal Psychology, 97(2), 238−245.

Mowery P.D., Brick P.D., & Farrelly M.C. (2000). Legacy First Look Report 3.
Pathways to Established smoking: Results from the 1999 National Youth Tobacco Survey.
Washington DC: American Legacy Foundation.

Muniz Jr., A.M. & O’Guinn, T.C. (2001). Brand community. Journal of Consumer
Research, 27, 412-432.

National Cancer Institute. (2009). The role of the media in promoting and reducing
tobacco use: Tobacco control monographs. Monograph 19. Available at:
http://cancercontrol.cancer.gov/tcrb/monographs/19/index.html.

National Institute on Drug Abuse. (2006). Research Report Series: Tobacco Addiction.
NIH Publication No. 06-4342. Rockville, MD: U.S. Department of Health and Human

174

Services National Institutes of Health. Available at:
http://www.nida.nih.gov/ResearchReports/Nicotine/Nicotine.html.

Niederdeppe, J., Farrelly, M.C., & Haviland, M.L. (2004). Confirming "truth": more
evidence of a successful tobacco countermarketing campaign in Florida.
American Journal of Public Health, 94(2), 255-7.

Nunnally, J.C. (1978). Psychometric Theory. New York: MacGraw-Hill.

Nunnally, J.C., & Bernstein, I.H. (1994). Psychometric Theory (3
rd
ed.).
New York: McGraw-Hill.

Ontario Task Force. (1995). Report of the Ontario Task Force on the Primary Prevention
of Cancer, Recommendations for the Primary Prevention of Cancer, March.

Osborne, J.W. & Costello, A.B. (2004). Sample size and subject to item ratio in principal
components analysis. Practical Assessment, Research & Evaluation, 9(11).

Palmgreen, P., Donohew, L., Lorch, E.P., Hoyle, R.H., & Stephenson, M.T. (2001).
Television campaigns and adolescent marijuana use: tests of sensation seeking targeting.
American Journal of Public Health, 91(2), 292-6.

Palmonari, A., Pombeni, M.S., & Kirchler, E. (1990). Adolescents and their peer groups:
A study on the significance of peers, social categorization processes and coping with
developmental tasks. Social Behaviour, 5(1), 33-48

Patton, G.C., Carlin, J.B., Coffey, C., Wolfe, R., Hibbert, M. & Bowes, G. (1998).
Depression, anxiety, and smoking initiation: A prospective study over 3 years.
American Journal of Public Health, 88(10), 1518-1522.

Pechmann, C., & Riebling, E.T. (2000). Anti-smoking advertising campaigns targeting
youth: Case studies from USA and Canada. Tobacco Control, 9(Supplement 2), ii18-
ii31.

Pedersen, W., Clausen, S.E., & Lavik, N.J. (1989). Patterns of drug use and sensation-
seeking among adolescents in Norway. Acta Psychiatrica Scandinavica, 79(4), 386-390.

Pederson, L.L., Koval, J.K., & O’Connor, K. (1997). Are psychosocial factors related to
smoking in grade-6 students? Addictive Behaviors, 22(2), 169-181.

175

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

Perkins, H.W., & Berkowitz, A.D. (1986). Perceiving the community norms of alcohol
use among students: Some research implications for campus alcohol education
programming. International Journal of the Addictions, 21(9-10), 961-976.

Petraitis, J., Flay, B.R. & Miller, T.Q. (1995). Reviewing theories of adolescent
substance use: Organizing pieces in the puzzle. Psychological Bulletin, 117(1), 67-86.

Pierce, J.P., Choi, W.S., Gilpin, E.A., Farkas, A.J. (1996). Validation of susceptibility as
a predictor of which adolescents take up smoking in the United States. Health
Psychology, 15, 511–515.

Pierce, J.P., Choi, W.S., Gilpin, E.A., Farkas, A.J. & Berry, C.C. (1998). Tobacco
industry promotion of cigarettes and adolescent smoking. Journal of the American
Medical Association, 279(7), 511-515.

Pokorny, S.B., Jason, L.A., & Schoeny, M.E. (2003). The relation of retail tobacco
availability to initiation and continued smoking. Clinical Child and Adolescent
Psychology, 32(2), 193-204.

Powell, L.M., Tauras, J.A., & Ross, H. (2005). The importance of peer effects, cigarette
prices and tobacco control policies for youth smoking behavior. Journal of Health
Economics, 24(5), 950-968.

Reed, A. & Forehand, M. (2003). Social identity in marketing research: An integrative
framework. Under review at Journal of Marketing.

Reynolds, T.J., & Gutman, J. (2001). Advertising is image management. In T.J. Reynolds
and J.C. Olson (Eds.), The means-end approach to marketing and advertising strategy,
(pp.145-162). New York: Routledge.

Ritt-Olson, A., Unger, J., Valente, T., Nezami, E., Chou, C., Trinidad, D., et al. (2005).
Exploring peers as a mediator of the association between depression and smoking in
young adolescents. Substance Use & Misuse, 40(1), 77-98.

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 Quarterly, 23(1), 48-
64.

176

Rosenstock, I.M. (1974). Historical origins of the health belief model. Health Education
Monographs, 2, 1-8.

Russell, C.A., Clapp, J.D., & DeJong, W. (2005). Done 4: Analysis of a failed social
norms marketing campaign. Health Communication, 17(1), 57-65.

Sargent, J.D., Beach, M.L., Dalton, M.A., Mott, L.A., Tickle, J.J, Ahrens, M.B., et al.
(2001). Effect of seeing tobacco use in films on trying smoking among adolescents: Cross
sectional study. British Medical Journal, 323(7326), 1394,

Schwartz, B. (2004). The paradox of choice: Why more is less. New York: Harper
Collins.

Seltzer, V.C. (1982) Adolescent social development: dynamic functional interaction.
Lexington: Lexington Books.

Sherif, M. (1936). The psychology of social norms. New York: Harper.

Sherry, J. F. Jr. (1987), Advertising as a cultural system. In J. Umiker-Sebeok (Ed.)
Marketing and Semiotics: New Directions in the Study of Signs for Sale (pp. 441-62).
Berlin: Mouton de Gruyter.

Sivulka, J. (1988). Soap, sex and cigarettes: A cultural history of American advertising.
Belmot, CA: Wadsworth Publishing Company.

Smith, S., Atkin, C., Martell, D., Hembroff R., & Allen, L. (2006). A social judgment
theory approach to conducting formative research in a social norms campaign.
Communication Theory, 16(1), 141-152.

Snyder, L.B., Hamilton, M.A., Mitchell, E.W., Kiwanuka-Tondo, J., Fleming-Milici, F.
& Proctor, D. (2004). A meta-analysis of the effect of mediated health communication
campaigns on behavior change in the United States. Journal of Health Communication,
9(6), 71-96.

Sparks, P. (2000). Subjective expected utility -based attitude-behavior models: the utility
of self- identity. In D. J. Terry and M.A. Hogg (Eds.), Attitudes, behavior and social
context: the role of norms and group Mmembership. Mahwah, NJ: Erlbaum.

Sparks, P., & Shephard, R. (1992). Self-identity and the theory of planned behavior:
Assessing the role of identification with “green consumerism”. Social Psychology
Quarterly, 55(4), 388-399.

177

Stephenson, M. (2002). Sensation seeking as a moderator of the processing of anti-
heroin public service announcements. Communication Studies, 53(4), 358-81.

Stephenson, M.T., Hoyle, R.H., Palmgreen, P. and Slater, M.D. (2003). Brief measures
of sensation seeking for screening and large-scale surveys. Drug and Alcohol
Dependence, 72(3), 279-286.

Stephenson, M., Morgan, S., Lorch, E., Palmgreen, P., Donohew, L., & Hoyle, R.H.
(2002). Predictors of exposure from an antimarijuana media campaign: Outcome
research assessing sensation seeking targeting. Health Communication, 14(1), 23-43.

Stephenson, M. & Palmgreen, P. (2001). Sensation seeking, perceived message sensation
value, personal involvement, and processing of anti-marijuana PSAs. Communication
Monographs, 68(1), 49-71.

Sterlin, K.L., Diamond, P.M., Mullen, P.D., Pallonen, U., Ford, K.T., & McAlister, A.L.
(2007). Smoking-related self-efficacy, beliefs, and intention: Assessing factorial validity
and structural relationships in 9th-12th grade current smokers. Addictive Behaviors,
32(9), 1863-1876.

Studenmund, A. H. & Cassidy, H.J. (1987). Using Econometrics: A Practical Guide.
Boston: Little Brown and Co.

Substance Abuse and Mental Health Services Administration. (2006). Calculated based
on data in 2006 National Household Survey on Drug Use and Health. Available at:
http://www.oas.samhsa.gov/nsduh.htm,

Sussman, S., Dent, C.W., McAdams, L.A., Stacy, A.W, Burton, D., & Flay, B.R. (1994).
Group self-identification and adolescent cigarette smoking: A 1-year prospective study.
Journal of Abnormal Psychology, 10(3), 576-580.

Sussman, S., Dent, C.W., & McCullar, W.J. (2000). Group self-identification as a
prospective predictor of drug use and violence in high-risk youth. Psychology of
Addictive Behaviors, 14(2), 192-196.

Sussman, S., Dent, C. W., Simon, T. R., Stacy, A. W., Burton, D., & Flay, B. R. (1993).
Identification of which high-risk youth smoke cigarettes regularly. Health Values, 17(1),
42−53.

Sussman, S., Dent, C. W., Stacy, A. W., Burciaga, C., Raynor, A., Turner, G. E., et al.
(1990). Peer group association and adolescent tobacco use. Journal of Abnormal
Psychology, 99(4), 349−352.

178

Sussman, S., Pokhrel, P., Ashmore, R.D., & Brown, B.B. (2007). Adolescent peer group
identification and characteristics: A review of the literature. Addictive Behaviors,
32(8), 1602-1627.

Sussman, S., Simon, T. R., Stacy, A.W.,Dent,C. W.,Ritt, A., Kipke, M. D., et
al.(1999).The Association ofgroup self-identification and adolescent drug use in three
samples varying in risk. Journal of Applied Social Psychology, 29(8), 1555−1581.

Sussman, S., Unger, J., & Dent, W. (2004). Peer group self-identification among
alternative high school youth: A predictor of their psychosocial functioning five years
later. International Journal of Clinical and Health Psychology, 4(1), 9-25.

Tajfel, H. (1957). Value and the perceptual judgement of magnitude. Psychological
Review, 64, 192-204.

---- (1959). Quantitative judgment in social perception. British Journal of Psychology,
50, 16-29.

Tajfel (1970). Experiments in intergroup discrimination. Scientific American, 223, 96-
102.

---- (1978). Social categorization, social identity, and social comparisons. In
Tajfel, H. (Ed.), Differentiation between social groups. London: Academic Press.

---- (1972). Social categorization. English manuscript of ‘‘La cate´ gorisation sociale.’’ In
S. Moscovici (Ed.), Introduction a` la psychologie sociale (Vol. 1, pp. 272–302). Paris:
Larousse.

---- (1974). Social identity and intergroup behaviour.
Social Science Information/sur les sciences sociales, 13(2), 65-93

Tajfel, H., Flament, C. Billig, M., & Bundy, R.P. (1971). Social categorization and
intergroup behaviour. European Journal of Social Psychology, 1, 149-178.

Tajfel & Turner (1979). An integrative theory of intergroup conflict. In W.G. Austin &
S. Worchel (Eds.), the social psychology of intergroup relations (pp. 33-147). Pacific
Grove, CA: Brooks/Cole.

----- (1986). The social identity theory of intergroup behavior. In S. Worchel & W.
Austen (Eds.) Psychology of intergroup relations (7-24), Chicago: Nelson Hall.

179

Tarrant, M. (2002). Adolescent peer groups and social identity. Social Developmetn,
11(1), 110-123.

Tarrant, M., North, A.C., Edridge, M.D., Kirk, L.E., Smith, E.A. & Turner, R.E. (2001).
Social identity in adolescence. Journal of Adolescence, 24(5), 597-609.
Taylor, Conard, O’Byrne, Haddock & Poston, 2004Tolone & Tieman, 1990

Taylor, J., Conard, M.W., O’Byrne, .K., Haddock, C.K., & Poston, W.S.C. (2004).
Saturation of tobacco smoking models and risk of alcohol and tobacco use among
adolescents. Journal of Adolescent Health, 35(3), 190-196.

Ter Doest, L., Dijkstra, A., Gebhardt, W.A. & Vitale, S. (2007). Cognitions about
smoking and not smoking in adolescence. Health Education and Behavior, July, 1-9.

Tercyak, K.P. (2003). Psychosocial risk factors for tobacco use among adolescents with
asthma. Journal of Pediatric Psychology, 28(7), 495-504.

Terry, D. & Hogg, M. (1996). Group norms and the attitude-behavior relationship: A
role for group identification. Personality and Social Psychology Bulletin, 22(8), 776-
793.

Terry, D., Hogg, M., & White, K. (1999). The theory of planned behaviour: Self-identity,
social identity and group norms. British Journal of Social Psychology, 38(3), 225-244.

Thombs, D.L., Dotterer, S., Olds, R.S., Sharp, K.E. & Raub, C.G. (2004). A close look at
why one social norms campaign did not reduce student drinking. Journal of American
College Health, 53(2), pp. 61-68.

Tolone, W.L., & Tierman, C.R. (1990). Drugs, delinquency and “nerds”: Are loners
deviant? Journal of Drug Education, 20(2), 153-162.

Trinidad, D.R., & Johnson, C.A. (2002). The association between emotional intelligence
and early adolescent tobacco and alcohol use. Personality and Individual Differences,
32(1), 95-105.

Trinidad, D.R., Unger, J.B., Chou, C., Azen, S., & Johnson, C.A. (2004). Emotional
intelligence and smoking risk factors in adolescents: Interacitons on smoking intentions.
Journal of Adolescent Health, 34(1), 46-55.

180

Trudeau, L., Lillehoj, C., Spoth, R., & Redmond, C. (2003). The role of assertiveness and
decision making in early adolescent substance initiation: Mediating processes. Journal
of Research on Adolescence, 13(3), 301-328.

Turner, J.C. (1978). Social categorization and social discrimination in the minimal group
paradigm. In H. Tajfel (Ed.), Differentiation between social groups: Studies in the social
psychology of intergroup relations. London: Academic Press.

----. (1982a). Towards a cognitive redefinition of the social group. In H. Tajfel (Ed.),
Social identity and intergroup relations (pp. 15–40). Cambridge, UK: Cambridge
University Press.

---- (1982b). Social identity and intergroup relations. Cambridge, UK: Cambridge
University Press.

---- (1985). Social categorization and the self-concept: A social cognitive theory of
group behavior. In E.J. Lawler, (ed.). Advances in Group Processes: Theory and
Research, vol. 2. Grennwich, CT: JAI Press.

---- (1999). Turner, J. (1999). Some current issues in research on social identity and self-
categorization theories. In Ellemers, N., Spears, R., & Doosje, B. (Eds.), Social
identity: Context, commitment, content, 6-34. Oxford: Blackwell Publishers.

Turner, J.C., Hogg, M.A., Oakes, P.J., Reicher, S.D., & Wetherell, M.S. (1987).
Rediscovering the social group: A self-categorization theory. Oxford: Blackwell.

Urberg, K. A. (1992). Locus of peer influence: Social crowd and best friend. Journal of
Youth and Adolescence, 21(4), 439−450.
Unger, J.B., Rohrbach, L.A., Cruz, T.B., Baezconde-Garbanati, L., Howard, K.A., &
Palmer, P.H., et al. (2001). Ethnic variation in peer influences on adolescent smoking.
Nicotine & Tobacco Research, 3(2), 167-176.

Urberg, K. A. (1992). Locus of peer influence: Social crowd and best friend. Journal of
Youth and Adolescence, 21(4), 439−450

Urberg, K.A., Degirmencioglu, S.M., Tolson, J.M., & Halliday-Scher, K. (2000).
Adolescent social crowds: Measurement and relationship to friendships. Journal of
Adolescence Research, 15(4), 427-445.

U.S. Surgeon General’s Office, (2004). The Health Consequences of Smoking: A Report
of the Surgeon General. Available at:
http://www.surgeongeneral.gov/library/smokingconsequences/.

181

Wakefield, M., Terry-McElrath, Y., Emery, S., Saffer, H., Chaloupka, F., Szczypka, G.,
et al. (2006). Effect of televised, tobacco company-funded smoking prevention
advertising on youth smoking-related beliefs, intentions, and behavior. American
Journal of Public Health, 96(12), 2154-2160.

Warner, K.E. (2002). Tobacco harm reduction: Promise and perils.
Nicotine & Tobacco Research, 4(Suppl2), S61-S71

Wechsler, H., Nelson, T.F., Lee, J.E., Seibring, M., Lewis, C., & Keeling, R. (2003).
Perception and reality: A national evaluation of social norms marketing interventions to
reduce college students' heavy alcohol use. Quarterly Journal of Studies on Alcohol,
64(4), 484-494.

Werch, C.E., Pappas, D.M., Carlson, J.M., DiClemente, C.C., Chally, P.S., & Sinder, J.A.
(2000). Results of a social norm intervention to prevent binge drinking among first year
residential college students. Journal of American College Health, 49(2), 85-92.

Wiehe, S.E., Garrison, M.M., Christakis, D.A., Ebel, B.E., & Rivara, F. “School-based
smoking prevention research": Reply. Journal of Adolescent Health, 37(1), 6-7

Yanovitzky, I. (2006). Sensation seeking and alcohol use by college students:
Examining multiple pathways of effects. Journal of Health Communication, 11(3), 269-
280.

Yanovitzky, I., Stewart, L., & Lederman, L. (2006). Social distance, perceived drinking
by peers, and alcohol use by college students. Health Communication, 19(1), 1-10.

Yzer, M. (2004, September). The integrative model of behavioral prediction as a tool for
designing STD prevention messages. EPIDASA Workshop.

Zuckerman, M. (1979). Sensation seeking: Beyond the optimal level of arousal.
Hillsdale, NJ: Lawrence Earlbaum Associates.

Zuckerman, M. (1994). Behavioral expressions and biological bases of sensation
seeking. Cambridge, UK: Cambridge University Press.

182

Appendix A: Complete Survey Measures

People often hang out in different groups at school. For example, a lot of schools have a
group of “jocks.” Some students gave the following list of groups. Please indicate how
much you identify with each group by dragging the bar across the screen. Dragging the
bar to 100 means you identify with this group very much and dragging the bar to 0 means
you do not identify with this group at all.

0----------------------------------------------------------------------------
100
Artists/Artsy kids
Athletes/Jocks
Emo kids
Involved in school
Musicians
Popular/cool kids
Preppy kids
Band kids (marching band)
Goth kids
Techies/AV kids
Nerds
Theater/Drama kids
Straightedge
Smart kids
Class clowns
Misfits/Outsiders
Gamers
Skaters
Nonconformists
Religious kids
Goody-goodies
Hippies
Bad kids/Rebels
Partiers
Hipsters
Average/Regulars

On average, how many days do you watch television?

0 hours
Between 0 and 1 hour

183

1 hour
2 hours
3 hours
4 hours
5 hours
6 hours
7 hours
8 hours
9 hours
10 hours
11 or more hours

On average, how many days do you listen to the radio?

0 hours
Between 0 and 1 hour
1 hour
2 hours
3 hours
4 hours
5 hours
6 hours
7 hours
8 hours
9 hours
10 hours
11 or more hours

On average, how many days do you use the Internet?

0 hours
Between 0 and 1 hour
1 hour
2 hours
3 hours
4 hours
5 hours
6 hours
7 hours
8 hours
9 hours
10 hours

184

11 or more hours

Have you ever tried cigarette smoking, even 1 or 2 puffs?

Yes No

How old were you when you smoked a whole cigarette for the first time?

I have never smoked a whole cigarette
8 years old or younger
9 years old
10 years old
11 years old
12 years old
13 years old
14 years old
15 years old
16 years old
17 years old

About how many cigarettes have you smoked in your entire life?

1 or more puffs but never a whole cigarette
1 cigarette
2 to 5 cigarettes (about ½ a pack total)
16-25 (about 1 pack total)
26-99 (more than 1 pack but less than 5 packs)
100 or more cigarettes (5 or more packs)

During the last 30 days, on how many days did you smoke cigarettes, even just 1 or 2
puffs?

During the past 30 days, on the days you smoked, how many cigarettes did you smoke
per day?

Less than 1 cigarette per day
2 to 5 cigarettes per day
6 to 10 cigarettes per day
11 to 20 cigarettes per day
More than 20 cigarettes per day

Have you ever smoked at least one cigarette every day for 30 days?

185

Yes No

Do you think you would be able to quit smoking cigarettes if you wanted to?

1 Definitely not
2
3
4 I’m not sure
5
6
7 Definitely yes

Do you think you will smoke a cigarette, even 1 or 2 puffs, at any time during the next
year?

1 Definitely not
2 Probably not
3 Maybe not
4 I’m not sure
5 Maybe yes
6 Probably yes
7 Definitely yes

If one of your best friends offered you a cigarette in the next 30 days, would you smoke
it?

1 Definitely not
2 Probably not
3 Maybe not
4 I’m not sure
5 Maybe yes
6 Probably yes
7 Definitely yes

If you wanted to, you could easily not smoke cigarettes during the next month.

1 Definitely not
2
3
4 I’m not sure
5

186

6
7 Definitely yes

If a friend offered you a cigarette, you could easily say no.

1 Strongly disagree
2
3
4 I’m not sure
5
6
7 Strongly agree

Please indicate how much you agree or disagree with the following statements.

I don’t want to smoke because it would mean cigarette companies are using me.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

I don’t want to smoke because it doesn’t fit my image.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

Smoke from other people’s cigarettes bothers me.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

187

I don’t want to smoke because it would make me less attractive.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

Do you think you will be smoking cigarettes 5 years from now?
1 Definitely not
2 Probably not
3 Maybe not
4 I’m not sure
5 Maybe yes
6 Probably yes
7 Definitely yes

Do you think you will be smoking cigarettes 10 years from now?
1 Definitely not
2 Probably not
3 Maybe not
4 I’m not sure
5 Maybe yes
6 Probably yes
7 Definitely yes
Please indicate how much you agree or disagree with the following statements.

Smoking cigarettes makes people your age look cool.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

Smoking cigarettes can help keep your weight down.
1 Strongly disagree
2 Disagree
3 Slightly disagree

188

4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

People who smoke regularly have a much harder time keeping up in sports and athletic
activities.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

Smoking cigarettes makes people your age fit in.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

Smoking is a way to express your independence.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

The smoke from other people's cigarettes is harmful to you.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

189

It is safe to smoke for only a year or two, as long as you quit after that.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

Not smoking is a way to express your independence.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

Nonsmokers don't like to date someone who smokes.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

People your age who smoke cigarettes have more friends than people who don't smoke.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

Please indicate how much you agree or disagree with each of the following statements.

Cigarette compnies deny that cigarettes are addictive.
1 Strongly disagree
2 Disagree

190

3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

I would like to see cigarette companies go out of business.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

Cigarette companies lie.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

The people who run cigarette companies know what they are doing is wrong.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

Cigarette companies should have the same right to sell cigarettes as other companies have
to sell their products.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

191

Cigarette companies deny that cigarettes cause cancer and other harmful diseases.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

Cigarette companies target teens to replace smokers who die.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

People have different views about the issue of smoking and cigarette companies.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

Cigarette companies target minority groups.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

Anti-smoking advertisements are no more honest than cigarette ads.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree

192

5 Slightly agree
6 Agree
7 Strongly agree

The government should let companies sell whatever they want.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

Cigarette companies try to get young people to start smoking.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

Cigarette companies get too much blame for young people smoking.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

Cigarette companies should have the same right to make money as any other type of
company.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

193

Have you ever seen or heard any ads or messages about any of the following issues?
(Please check all that apply.)
Not using drugs
Not starting to smoke
Quitting smoking
Messages against smoking
Messages about or against cigarette companies
Not using alcohol
Not drinking and driving while intoxicated
Not using chewing tobacco, snuff or dip
None of the above

During the past 30 days, have you seen or heard any ads or messages about not starting to
smoke, quitting smoking, messages about or against cigarette companies or messages
against smoking? Yes No

During the past 30 days, which of the following issues have you seen or heard about?
(Please check all that apply.)
Not starting to smoke
Quitting smoking
Messages against smoking
Messages about or against cigarette companies

In the past 30 days, where did you see these messages about smoking? (Check all that
apply.)
On TV
On the radio
In newspapers or magazines
On a billboard
On the internet
Somewhere else

What is the theme/slogan of this advertising campaign?
Think. Don't Smoke.
Truth.
Tobacco is whacko.
Become an ex.
Tobacco vs. Kids. Where America Draws the Line.
Other (please write in):

Who (or what group of people) is this advertising or campaign supposed to help?

194

Who is running this advertising or campaign?

Are you aware of any other advertising or campaign against tobacco or smoking that is
now taking place?
Yes No

What is the theme/slogan of this advertising or campaign?
Truth.
Think. Don't smoke.
Tobacco is wacko.
Become an ex.
Tobacco vs. Kids. Where America Draws the Line.
Other (please write in):

Who (or what group of people) is this advertising or campaign supposd to help?

Who is running this advertising or campaign?

What percentage of your friends have smoked every day for the last 30 days?

What percentage of your friends have ever tried smoking, even just 1 or 2 puffs?

What percentage of people your age do you think have smoked every day for the last 30
days?

What percentage of people your age do you think have ever tried smoking, even just 1 or
2 puffs?

Do you live with anyone who smokes?
Yes No I don't know.

Who in your house smokes? (Check all that apply.)
Mother
Father
Sibling
Grandmother
Grandfather
Other (please indicate)

My friends would approve of me smoking cigarettes.
1 Strongly disagree
2 Disagree

195

3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

My family members would approve of me smoking cigarettes.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

Please indicate the extent to which you agree or disagree with the following statements.

I would like to explore strange places.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

I like new and exciting experiences, even if I have to break the rules.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

I like to do frightening things.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree

196

7 Strongly agree

I prefer friends who are exciting and unpredictable.
1 Strongly disagree
2 Disagree
3 Slightly disagree
4 Neither agree nor disagree
5 Slightly agree
6 Agree
7 Strongly agree

How old are you?

What grade are you in?

In school, the grades I usually get are...
A+ A A- …………… D+ D D- F or below
-

How often do you attend religious services (church, temple, etc.) with your family?
Never
A few times a year
About once a month
A few times a month
Once a week
More than once a week
Every day

Are you a.....?
Male Female

How do you describe yourself? You can check more than one category.
American Indian or Alaskan Native
Asian
Black or African American
Hispanic or Latino
Middle Eastern
Native Hawaiian or Other Pacific Islander
White
Other (please indicate)

197

198

Appendix B: Unreported Statistical Tests

Multinomial Logistic Regression – SISC-artists
Smoking Status(a) B SE B Sig. Exp(B)
Never, susceptible Intercept -1.464 1.625 0.368
African American = Yes -1.033 0.885 0.243 0.356
White = Yes -0.456 0.556 0.412 0.634
Female = Yes 0.008 0.515 0.988 0.992
Lives with Smoker = Yes 0.207 0.307 0.501 1.230
Achievement -0.126 0.099 0.203 0.882
Religiousitysity 0.096 0.136 0.479 1.101
Social Norm 0.008 0.013 0.511 1.008
Sensation Seeking 0.174 0.155 0.263 1.190
truth™ Exposure 0.493 0.478 0.302 1.637
SISC-artists -0.572 0.289 0.048 0.565
*

a The reference category is: Never, not susceptible.

Multinomial Logistic Regression – SISC-musicians
Smoking Status(a) B SE B Sig. Exp(B)
Never, susceptible Intercept -0.793 1.627 0.626
African American = Yes -1.305 0.912 0.152 0.271
White = Yes -0.436 0.562 0.438 0.646
Female = Yes -0.347 0.539 0.519 0.707
Lives with Smoker = Yes 0.272 0.316 0.388 1.313
Achievement -0.123 0.101 0.224 0.885
Religiosity 0.136 0.142 0.340 1.145
Social Norm 0.006 0.013 0.666 1.006
Sensation Seeking 0.158 0.159 0.318 1.172
truth™ Exposure 0.418 0.489 0.392 1.519
SISC-musicians -1.097 0.331 0.001 0.334
a The reference category is: Never, not susceptible.

Multinomial Logistic Regression – SISC-theater
Smoking Status(a) B SE B Sig. Exp(B)
Never, susceptible Intercept -2.535 1.666 0.128
African American = Yes -1.188 0.911 0.192 0.305
White = Yes -0.558 0.567 0.325 0.572
Female = Yes 0.148 0.529 0.780 1.159
Lives with Smoker = Yes 0.424 0.324 0.190 1.529
Achievement -0.086 0.098 0.383 0.918
Religiosity 0.169 0.138 0.222 1.184

199

Social Norm 0.002 0.013 0.878 1.002
Sensation Seeking 0.184 0.160 0.248 1.203
truth™ Exposure 0.496 0.486 0.307 1.643
SISC-theater -1.054 0.380 0.006 0.349

Table 5.6 Multinomial Logistic Regression – SISC-religious
Smoking Status(a) B SE B Sig. Exp(B)
Never, susceptible Intercept -1.978 1.555 0.203
African American = Yes -1.257 0.873 0.150 0.284
White = Yes -0.507 0.554 0.360 0.602
Female = Yes -0.064 0.496 0.898 0.938
Lives with Smoker = Yes 0.150 0.295 0.611 1.162
Achievement -0.065 0.095 0.495 0.937
Religiosity 0.216 0.144 0.134 1.241
Social Norm 0.016 0.011 0.144 1.017
Sensation Seeking 0.079 0.147 0.589 1.082
truth™ Exposure 0.267 0.469 0.570 1.305
SISC-religious -0.695 0.332 0.036 0.499

Multinomial Logistic Regression – SISC-Average
Smoking Status(a) B SE B Sig. Exp(B)
Never, susceptible Intercept -1.978 1.578 0.210
African American = Yes -1.291 0.883 0.144 0.275
White = Yes -0.527 0.554 0.342 0.591
Female = Yes -0.155 0.496 0.754 0.856
Lives with Smoker = Yes 0.181 0.308 0.556 1.199
Achievement -0.075 0.098 0.448 0.928
Religiosity 0.043 0.133 0.749 1.044
Social Norm 0.014 0.011 0.208 1.014
Sensation Seeking 0.208 0.151 0.167 1.231
truth™ Exposure 0.330 0.468 0.481 1.391
SISC-average 0.424 0.210 0.044 1.528

Multinomial Logistic Regression – SISC-hipster
Smoking Status(a) B SE B Sig. Exp(B)
Never, susceptible Intercept -1.798 1.556 0.248
African American = Yes -1.192 0.881 0.176 0.304
White = Yes -0.139 0.563 0.804 0.870
Female = Yes -0.119 0.497 0.811 0.888
Lives with Smoker = Yes 0.121 0.300 0.687 1.128
Achievement -0.066 0.096 0.491 0.936

200

Religiosity 0.115 0.133 0.386 1.122
Social Norm 0.013 0.011 0.248 1.013
Sensation Seeking 0.169 0.147 0.250 1.184
truth™ Exposure 0.391 0.472 0.407 1.479
SISC-hipster 0.677 0.281 0.016 1.969

Multinomial Logistic Regression – SISC-partier
Smoking Status(a) B SE B Sig. Exp(B)
Never, susceptible Intercept -1.890 1.527 0.216
African American = Yes -1.417 0.893 0.112 0.242
White = Yes -0.374 0.555 0.501 0.688
Female = Yes -0.227 0.497 0.648 0.797
Lives with Smoker = Yes 0.129 0.302 0.670 1.138
Achievement -0.051 0.098 0.601 0.950
Religiosity 0.093 0.132 0.479 1.098
Social Norm 0.013 0.011 0.260 1.013
Sensation Seeking 0.158 0.143 0.267 1.171
truth(tm) Exposure 0.303 0.470 0.519 1.354
SISC-partier 0.601 0.252 0.017 1.825

Multinomial Logistic Regression – SISC-rebel
Smoking Status(a) B SE B Sig. Exp(B)
Never, susceptible Intercept -1.865 1.570 0.235
African American = Yes -1.290 0.875 0.141 0.275
White = Yes -0.414 0.549 0.451 0.661
Female = Yes -0.132 0.493 0.789 0.877
Lives with Smoker = Yes 0.055 0.298 0.855 1.056
Achievement -0.034 0.097 0.726 0.967
Religiosity 0.120 0.132 0.362 1.128
Social Norm 0.015 0.011 0.175 1.015
Sensation Seeking 0.096 0.148 0.517 1.100
truth™ Exposure 0.458 0.473 0.333 1.582
SISC-rebel 0.658 0.281 0.019 1.930

Multinomial Logistic Regression – SISC-artists
Smoking Status(a) B SE B Sig. Exp(B)

Tried Intercept -0.414 1.347 0.758
African American = Yes -1.096 0.713 0.124 0.334

201

White = Yes -0.740 0.448 0.098 0.477
Female = Yes 0.405 0.444 0.362 1.499
Lives with Smoker = Yes 0.080 0.256 0.756 1.083
Achievement -0.266 0.078 0.001 0.766
Religiosity -0.099 0.113 0.385 0.906
Social Norm 0.011 0.010 0.267 1.011
Sensation Seeking 0.405 0.137 0.003 1.499
truth™ Exposure 0.214 0.397 0.589 1.239
SISC-artists -0.777 0.240 0.001 0.460

Multinomial Logistic Regression – SISC-musicians
Smoking Status(a) B SE B Sig. Exp(B)

Tried Intercept 0.267 1.313 0.839
African American = Yes -1.387 0.711 0.051 0.250
White = Yes -0.779 0.443 0.078 0.459
Female = Yes 0.173 0.446 0.697 1.189
Lives with Smoker = Yes 0.107 0.256 0.675 1.113
Achievement -0.244 0.077 0.002 0.783
Religiosity -0.068 0.112 0.544 0.934
Social Norm 0.012 0.010 0.230 1.012
Sensation Seeking 0.351 0.132 0.008 1.421
truth(tm) Exposure 0.101 0.390 0.795 1.107
SISC-musician -0.622 0.228 0.006 0.537

Multinomial Logistic Regression – SISC-nerds
Smoking Status(a) B SE B Sig. Exp(B)
Tried Intercept -0.900 1.369 0.511
African American = Yes -1.362 0.695 0.050 0.256
White = Yes -0.825 0.435 0.058 0.438
Female = Yes 0.497 0.438 0.257 1.644
Lives with Smoker = Yes 0.141 0.256 0.583 1.151
Achievement -0.204 0.077 0.008 0.816
Religiosity -0.107 0.112 0.337 0.898
Social Norm 0.014 0.010 0.158 1.014
Sensation Seeking 0.316 0.134 0.018 1.371
truth(tm) Exposure 0.080 0.391 0.838 1.083
SISC-nerds -0.628 0.267 0.019 0.533

Multinomial Logistic Regression – SISC-skaters

202

Smoking Status(a) B SE B Sig. Exp(B)
Tried Intercept -0.894 1.348 0.507
African American = Yes -1.307 0.685 0.056 0.271
White = Yes -0.960 0.437 0.028 0.383
Female = Yes 0.406 0.440 0.356 1.500
Lives with Smoker = Yes 0.128 0.258 0.619 1.137
Achievement -0.189 0.077 0.014 0.828
Religiosity -0.041 0.113 0.717 0.960
Social Norm 0.013 0.010 0.193 1.013
Sensation Seeking 0.328 0.131 0.013 1.388
truth(tm) Exposure 0.089 0.391 0.821 1.093
SISC-skaters 0.599 0.218 0.006 1.821

Multinomial Logistic Regression – SISC-rebels
Smoking Status(a) B SE B Sig. Exp(B)
Tried Intercept -1.147 1.304 0.379
African American = Yes -1.227 0.663 0.064 0.293
White = Yes -0.869 0.442 0.049 0.420
Female = Yes 0.362 0.438 0.409 1.436
Lives with Smoker = Yes 0.075 0.249 0.762 1.078
Achievement -0.156 0.076 0.041 0.856
Religiosity -0.030 0.111 0.790 0.971
Social Norm 0.014 0.009 0.135 1.014
Sensation Seeking 0.302 0.131 0.021 1.353
truth(tm) Exposure 0.281 0.394 0.476 1.324
SISC-rebels 0.741 0.235 0.002 2.099

Multinomial Logistic Regression – SISC-partiers
Smoking Status(a) B SE B Sig. Exp(B)
Tried Intercept -0.832 1.239 0.502
African American = Yes -1.278 0.655 0.051 0.279
White = Yes -0.857 0.438 0.050 0.424
Female = Yes 0.328 0.437 0.452 1.389
Lives with Smoker = Yes 0.136 0.252 0.589 1.146
Achievement -0.201 0.073 0.006 0.818
Religiosity -0.078 0.109 0.475 0.925
Social Norm 0.012 0.009 0.208 1.012
Sensation Seeking 0.378 0.127 0.003 1.459
truth(tm) Exposure 0.149 0.386 0.698 1.161
SISC-partiers 0.413 0.194 0.034 1.511

203

Multinomial Logistic Regression – SISC-hippies
Smoking Status(a) B SE B Sig. Exp(B)
Current Intercept 1.432 2.746 0.602
African American = Yes -0.028 1.698 0.987 0.972
White = Yes 2.111 1.612 0.190 8.258
Female = Yes -0.976 0.834 0.242 0.377
Lives with Smoker = Yes -0.132 0.489 0.787 0.876
Achievement -0.357 0.148 0.016 0.700
Religiosity -0.448 0.276 0.105 0.639
Social Norm 0.059 0.014 0.000 1.061
Sensation Seeking -0.155 0.251 0.539 0.857
truth(tm) Exposure 0.421 0.804 0.600 1.524
SISC-hippies 1.215 0.415 0.003 3.372

Multinomial Logistic Regression – SISC-rebels
Smoking Status(a) B SE B Sig. Exp(B)
Current Intercept -0.406 2.658 0.879
African American = Yes -0.329 1.526 0.829 0.720
White = Yes 1.938 1.442 0.179 6.944
Female = Yes -1.139 0.818 0.164 0.320
Lives with Smoker = Yes 0.082 0.453 0.856 1.085
Achievement -0.222 0.145 0.125 0.801
Religiosity -0.359 0.264 0.173 0.698
Social Norm 0.058 0.014 0.000 1.060
Sensation Seeking -0.261 0.237 0.269 0.770
truth(tm) Exposure 0.929 0.773 0.229 2.533
SISC-rebels 1.387 0.519 0.008 4.003

Multinomial Logistic Regression – SISC-athletes
Smoking Status(a) B SE B Sig. Exp(B)
Never, susceptible Intercept -1.282 1.606 0.425
African American = Yes -1.191 0.882 0.177 0.304
White = Yes -0.572 0.546 0.295 0.564
Female = Yes -0.108 0.517 0.835 0.898
Lives With Smoker = Yes 0.233 0.305 0.445 1.262
Achievement -0.092 0.096 0.339 0.912
Religiosity 0.122 0.135 0.367 1.130
Social Norm 0.012 0.013 0.331 1.012
Sensation Seeking 0.115 0.151 0.447 1.122

204

truth Exposure 0.425 0.475 0.372 1.529
SISC-Athlete -0.153 0.239 0.521 0.858
Tried Intercept -0.554 1.331 0.677
African American = Yes -1.393 0.698 0.046 0.248
White = Yes -0.873 0.434 0.044 0.418
Female = Yes 0.428 0.443 0.333 1.535
Lives With Smoker = Yes 0.149 0.260 0.568 1.160
Achievement -0.228 0.075 0.002 0.796
Religiosity -0.078 0.110 0.475 0.925
Social Norm 0.014 0.010 0.160 1.014
Sensation Seeking 0.363 0.135 0.007 1.437
truth Exposure 0.185 0.392 0.638 1.203
SISC-Athlete 0.151 0.189 0.423 1.163
Current Intercept 1.623 3.262 0.619
African American = Yes -0.630 1.945 0.746 0.533
White = Yes 2.848 1.960 0.146 17.245
Female = Yes -1.616 0.847 0.057 0.199
Lives With Smoker = Yes -0.254 0.541 0.638 0.776
Achievement -0.310 0.173 0.074 0.733
Religiosity -0.502 0.289 0.082 0.605
Social Norm 0.073 0.017 0.000 1.075
Sensation Seeking -0.243 0.242 0.315 0.785
truth Exposure 0.197 0.806 0.807 1.217
SISC-Athlete -0.820 0.437 0.061 0.441
a The reference category is: Never, not susceptible.

Multinomial Logistic Regression – SISC-emo
Smoking Status(a) B SE B Sig. Exp(B)
Never, susceptible Intercept -1.482 1.581 0.349
African American = Yes -1.229 0.878 0.162 0.293
White = Yes -0.548 0.547 0.317 0.578
Female = Yes -0.047 0.510 0.927 0.954
Lives With Smoker = Yes 0.266 0.304 0.381 1.305
Achievement -0.095 0.099 0.335 0.909
Religiosity 0.112 0.134 0.403 1.119
Social Norm 0.011 0.013 0.372 1.011
Sensation Seeking 0.134 0.154 0.384 1.143
truth Exposure 0.453 0.472 0.338 1.573
z_siemo -0.036 0.235 0.878 0.965
Tried Intercept -0.225 1.282 0.861
African American = Yes -1.374 0.698 0.049 0.253
White = Yes -0.868 0.435 0.046 0.420

205

Female = Yes 0.365 0.434 0.401 1.440
Lives With Smoker = Yes 0.113 0.256 0.660 1.119
Achievement -0.231 0.077 0.003 0.794
Religiosity -0.075 0.110 0.491 0.927
Social Norm 0.015 0.010 0.117 1.016
Sensation Seeking 0.337 0.132 0.010 1.401
truth Exposure 0.106 0.385 0.782 1.112
z_siemo -0.026 0.183 0.888 0.975
Current Intercept 0.982 2.749 0.721
African American = Yes -0.239 1.721 0.889 0.787
White = Yes 2.485 1.694 0.142 12.004
Female = Yes -1.332 0.802 0.097 0.264
Lives With Smoker = Yes 0.053 0.482 0.913 1.054
Achievement -0.268 0.154 0.081 0.765
Religiosity -0.533 0.287 0.063 0.587
Social Norm 0.067 0.017 0.000 1.069
Sensation Seeking -0.254 0.252 0.314 0.775
truth Exposure 0.596 0.781 0.445 1.815
z_siemo -0.006 0.353 0.986 0.994
a The reference category is: Never, not susceptible.

Multinomial Logistic Regression – SISC-popular
Smoking Status(a) B SE B Sig. Exp(B)
Never, susceptible Intercept -1.786 1.654 0.280
African American = Yes -1.202 0.877 0.171 0.301
White = Yes -0.488 0.554 0.378 0.614
Female = Yes -0.018 0.510 0.972 0.983
Lives With Smoker = Yes 0.306 0.312 0.326 1.358
Achievement -0.087 0.097 0.372 0.917
Religiosity 0.110 0.134 0.415 1.116
Social Norm 0.011 0.013 0.390 1.011
Sensation Seeking 0.142 0.153 0.353 1.152
truth Exposure 0.514 0.480 0.285 1.672
z_sipopu 0.156 0.227 0.490 1.169
Tried Intercept -0.602 1.357 0.657
African American = Yes -1.350 0.698 0.053 0.259
White = Yes -0.807 0.439 0.066 0.446
Female = Yes 0.390 0.434 0.368 1.477
Lives With Smoker = Yes 0.155 0.261 0.552 1.168
Achievement -0.220 0.076 0.004 0.803
Religiosity -0.078 0.110 0.477 0.925
Social Norm 0.015 0.010 0.129 1.015

206

Sensation Seeking 0.352 0.133 0.008 1.422
truth Exposure 0.188 0.395 0.634 1.207
z_sipopu 0.172 0.180 0.339 1.188
Current Intercept 0.148 2.799 0.958
African American = Yes 0.027 1.745 0.987 1.028
White = Yes 2.496 1.659 0.132 12.137
Female = Yes -1.332 0.794 0.093 0.264
Lives With Smoker = Yes 0.208 0.519 0.689 1.231
Achievement -0.243 0.153 0.112 0.784
Religiosity -0.534 0.293 0.068 0.586
Social Norm 0.066 0.016 0.000 1.068
Sensation Seeking -0.220 0.248 0.375 0.802
truth Exposure 0.777 0.804 0.334 2.174
z_sipopu 0.448 0.421 0.287 1.565
a The reference category is: Never, not susceptible.

Multinomial Logistic Regression – SISC-preppy
Smoking Status(a) B SE B Sig. Exp(B)
Never, susceptible Intercept -1.530 1.582 0.333
African American = Yes -1.135 0.885 0.200 0.321
White = Yes -0.521 0.552 0.345 0.594
Female = Yes -0.065 0.511 0.899 0.937
Lives With Smoker = Yes 0.289 0.306 0.344 1.335
Achievement -0.100 0.098 0.305 0.905
Religiosity 0.110 0.135 0.412 1.117
Social Norm 0.009 0.013 0.467 1.009
Sensation Seeking 0.155 0.154 0.316 1.167
truth Exposure 0.486 0.473 0.304 1.625
z_siprep 0.203 0.228 0.373 1.225
Tried Intercept -0.286 1.285 0.824
African American = Yes -1.316 0.705 0.062 0.268
White = Yes -0.846 0.436 0.052 0.429
Female = Yes 0.347 0.436 0.427 1.414
Lives With Smoker = Yes 0.138 0.258 0.592 1.148
Achievement -0.240 0.076 0.002 0.787
Religiosity -0.076 0.110 0.489 0.927
Social Norm 0.013 0.010 0.198 1.013
Sensation Seeking 0.366 0.133 0.006 1.442
truth Exposure 0.163 0.388 0.674 1.177
z_siprep 0.225 0.182 0.218 1.252
Current Intercept 1.535 2.903 0.597
African American = Yes -0.547 1.814 0.763 0.579

207

White = Yes 2.449 1.723 0.155 11.578
Female = Yes -1.346 0.818 0.100 0.260
Lives With Smoker = Yes -0.008 0.472 0.987 0.992
Achievement -0.230 0.160 0.152 0.795
Religiosity -0.666 0.312 0.033 0.514
Social Norm 0.081 0.020 0.000 1.085
Sensation Seeking -0.426 0.280 0.129 0.653
truth Exposure 0.246 0.821 0.765 1.279
z_siprep -0.771 0.510 0.131 0.463
a The reference category is: Never, not susceptible.

Multinomial Logistic Regression – SISC-band
Smoking Status(a) B SE B Sig. Exp(B)
Never, susceptible Intercept -1.764 1.609 0.273
African American = Yes -1.340 0.883 0.129 0.262
White = Yes -0.446 0.552 0.419 0.640
Female = Yes -0.002 0.516 0.997 0.998
Lives With Smoker = Yes 0.256 0.302 0.397 1.292
Achievement -0.083 0.097 0.396 0.921
Religiosity 0.132 0.136 0.334 1.141
Social Norm 0.008 0.013 0.525 1.008
Sensation Seeking 0.115 0.151 0.445 1.122
truth Exposure 0.313 0.478 0.513 1.367
z_siband -0.430 0.285 0.132 0.651
Tried Intercept -0.347 1.280 0.786
African American = Yes -1.432 0.699 0.041 0.239
White = Yes -0.802 0.435 0.066 0.449
Female = Yes 0.381 0.436 0.382 1.464
Lives With Smoker = Yes 0.103 0.255 0.686 1.109
Achievement -0.224 0.076 0.003 0.800
Religiosity -0.067 0.111 0.547 0.936
Social Norm 0.014 0.010 0.166 1.014
Sensation Seeking 0.325 0.129 0.012 1.384
truth Exposure 0.015 0.390 0.969 1.016
z_siband -0.250 0.213 0.240 0.779
Current Intercept 0.885 2.824 0.754
African American = Yes -0.548 1.750 0.754 0.578
White = Yes 2.686 1.735 0.122 14.675
Female = Yes -1.384 0.806 0.086 0.251
Lives With Smoker = Yes 0.079 0.470 0.866 1.083
Achievement -0.282 0.158 0.074 0.754
Religiosity -0.521 0.286 0.069 0.594

208

Social Norm 0.065 0.016 0.000 1.068
Sensation Seeking -0.291 0.247 0.239 0.747
truth Exposure 0.490 0.783 0.531 1.633
z_siband -0.654 0.554 0.238 0.520
a The reference category is: Never, not susceptible.

Multinomial Logistic Regression – SISC-goth
Smoking Status(a) B SE B Sig. Exp(B)
Never, susceptible Intercept -1.667 1.611 0.301
African American = Yes -1.274 0.883 0.149 0.280
White = Yes -0.558 0.545 0.306 0.573
Female = Yes -0.053 0.509 0.917 0.948
Lives With Smoker = Yes 0.270 0.306 0.376 1.310
Achievement -0.094 0.098 0.340 0.911
Religiosity 0.109 0.135 0.420 1.115
Social Norm 0.011 0.013 0.384 1.011
Sensation Seeking 0.155 0.156 0.319 1.168
truth Exposure 0.458 0.471 0.331 1.581
z_sigoth -0.221 0.278 0.428 0.802
Tried Intercept -0.293 1.294 0.821
African American = Yes -1.388 0.700 0.047 0.250
White = Yes -0.868 0.434 0.046 0.420
Female = Yes 0.360 0.433 0.406 1.433
Lives With Smoker = Yes 0.112 0.257 0.662 1.119
Achievement -0.232 0.076 0.002 0.793
Religiosity -0.076 0.109 0.487 0.927
Social Norm 0.015 0.010 0.127 1.015
Sensation Seeking 0.348 0.134 0.009 1.417
truth Exposure 0.114 0.384 0.766 1.121
z_sigoth -0.081 0.203 0.689 0.922
Current Intercept 0.584 3.072 0.849
African American = Yes -0.089 1.863 0.962 0.915
White = Yes 3.173 2.078 0.127 23.872
Female = Yes -1.353 0.836 0.106 0.258
Lives With Smoker = Yes -0.034 0.480 0.943 0.966
Achievement -0.270 0.163 0.097 0.764
Religiosity -0.497 0.290 0.086 0.608
Social Norm 0.072 0.017 0.000 1.075
Sensation Seeking -0.302 0.250 0.227 0.739
truth Exposure 0.572 0.803 0.476 1.773
z_sigoth 0.555 0.365 0.128 1.742
a The reference category is: Never, not susceptible.

209

Multinomial Logistic Regression – SISC-techies
Smoking Status(a) B SE B Sig. Exp(B)
Never, susceptible Intercept -1.444 1.587 0.363
African American = Yes -1.230 0.883 0.164 0.292
White = Yes -0.555 0.547 0.311 0.574
Female = Yes -0.039 0.519 0.940 0.962
Lives With Smoker = Yes 0.263 0.303 0.386 1.301
Achievement -0.094 0.099 0.340 0.910
Religiosity 0.113 0.135 0.404 1.119
Social Norm 0.011 0.013 0.366 1.011
Sensation Seeking 0.128 0.151 0.395 1.137
truth Exposure 0.448 0.472 0.342 1.566
z_sitech 0.040 0.357 0.910 1.041
Tried Intercept -0.254 1.288 0.844
African American = Yes -1.328 0.699 0.057 0.265
White = Yes -0.847 0.434 0.051 0.429
Female = Yes 0.293 0.443 0.508 1.340
Lives With Smoker = Yes 0.113 0.256 0.659 1.120
Achievement -0.223 0.077 0.004 0.800
Religiosity -0.081 0.110 0.463 0.922
Social Norm 0.015 0.010 0.130 1.015
Sensation Seeking 0.334 0.129 0.010 1.396
truth Exposure 0.094 0.384 0.806 1.099
z_sitech -0.169 0.297 0.570 0.844
Current Intercept 0.649 2.804 0.817
African American = Yes -0.465 1.847 0.801 0.628
White = Yes 2.467 1.788 0.168 11.787
Female = Yes -1.075 0.844 0.203 0.341
Lives With Smoker = Yes 0.059 0.485 0.903 1.061
Achievement -0.268 0.151 0.077 0.765
Religiosity -0.479 0.291 0.100 0.619
Social Norm 0.066 0.016 0.000 1.068
Sensation Seeking -0.266 0.244 0.274 0.766
truth Exposure 0.626 0.787 0.426 1.871
z_sitech 0.297 0.464 0.523 1.345
a The reference category is: Never, not susceptible.

Multinomial Logistic Regression – SISC-smart
Smoking Status(a) B SE B Sig. Exp(B)
Never, susceptible Intercept -1.720 1.636 0.293
African American = Yes -1.152 0.887 0.194 0.316

210

White = Yes -0.513 0.548 0.349 0.599
Female = Yes 0.007 0.517 0.989 1.007
Lives With Smoker = Yes 0.274 0.308 0.374 1.315
Achievement -0.072 0.103 0.484 0.931
Religiosity 0.113 0.134 0.400 1.120
Social Norm 0.012 0.013 0.359 1.012
Sensation Seeking 0.125 0.150 0.406 1.133
truth Exposure 0.455 0.474 0.337 1.576
z_sismar -0.126 0.290 0.664 0.882
Tried Intercept -0.874 1.344 0.516
African American = Yes -1.179 0.706 0.095 0.307
White = Yes -0.794 0.434 0.068 0.452
Female = Yes 0.488 0.442 0.270 1.628
Lives With Smoker = Yes 0.194 0.257 0.450 1.214
Achievement -0.176 0.081 0.031 0.839
Religiosity -0.076 0.111 0.492 0.927
Social Norm 0.015 0.010 0.117 1.016
Sensation Seeking 0.328 0.130 0.011 1.388
truth Exposure 0.154 0.388 0.691 1.167
z_sismar -0.389 0.241 0.106 0.678
Current Intercept -0.143 2.928 0.961
African American = Yes -0.050 1.736 0.977 0.951
White = Yes 2.588 1.727 0.134 13.309
Female = Yes -1.106 0.823 0.179 0.331
Lives With Smoker = Yes 0.123 0.490 0.802 1.131
Achievement -0.198 0.162 0.224 0.821
Religiosity -0.500 0.286 0.080 0.606
Social Norm 0.066 0.016 0.000 1.069
Sensation Seeking -0.237 0.237 0.316 0.789
truth Exposure 0.625 0.788 0.428 1.868
z_sismar -0.613 0.487 0.208 0.542
a The reference category is: Never, not susceptible.

Multinomial Logistic Regression – SISC-misfits
Smoking Status(a) B SE B Sig. Exp(B)
Never, susceptible Intercept -1.515 1.582 0.338
African American = Yes -1.312 0.890 0.140 0.269
White = Yes -0.579 0.552 0.295 0.561
Female = Yes -0.056 0.510 0.913 0.946
Lives With Smoker = Yes 0.279 0.305 0.360 1.322
Achievement -0.101 0.098 0.305 0.904

211

Religiosity 0.098 0.135 0.470 1.102
Social Norm 0.012 0.013 0.353 1.012
Sensation Seeking 0.165 0.157 0.295 1.179
truth Exposure 0.481 0.473 0.310 1.617
z_simisf -0.257 0.286 0.368 0.773
Tried Intercept -0.166 1.283 0.897
African American = Yes -1.310 0.702 0.062 0.270
White = Yes -0.845 0.435 0.052 0.429
Female = Yes 0.371 0.435 0.394 1.449
Lives With Smoker = Yes 0.097 0.257 0.705 1.102
Achievement -0.226 0.076 0.003 0.798
Religiosity -0.070 0.110 0.525 0.932
Social Norm 0.016 0.010 0.107 1.016
Sensation Seeking 0.306 0.134 0.022 1.358
truth Exposure 0.069 0.388 0.859 1.071
z_simisf 0.141 0.216 0.514 1.151
Current Intercept 0.945 2.747 0.731
African American = Yes -0.217 1.716 0.899 0.805
White = Yes 2.518 1.686 0.135 12.405
Female = Yes -1.342 0.789 0.089 0.261
Lives With Smoker = Yes 0.051 0.483 0.915 1.053
Achievement -0.266 0.151 0.079 0.766
Religiosity -0.535 0.287 0.062 0.586
Social Norm 0.067 0.016 0.000 1.069
Sensation Seeking -0.254 0.249 0.307 0.775
truth Exposure 0.601 0.787 0.445 1.824
z_simisf -0.013 0.417 0.975 0.987
a The reference category is: Never, not susceptible.

Multinomial Logistic Regression – SISC-gamers
Smoking Status(a) B SE B Sig. Exp(B)
Never, susceptible Intercept -1.421 1.589 0.371
African American = Yes -1.219 0.879 0.166 0.296
White = Yes -0.547 0.547 0.318 0.579
Female = Yes -0.083 0.536 0.877 0.920
Lives With Smoker = Yes 0.267 0.305 0.381 1.306
Achievement -0.092 0.097 0.344 0.912
Religiosity 0.109 0.135 0.417 1.116
Social Norm 0.011 0.013 0.365 1.011
Sensation Seeking 0.128 0.150 0.392 1.137
truth Exposure 0.454 0.472 0.336 1.575
z_sigame -0.059 0.289 0.837 0.942

212

Tried Intercept -0.295 1.300 0.821
African American = Yes -1.373 0.695 0.048 0.253
White = Yes -0.883 0.434 0.042 0.414
Female = Yes 0.412 0.453 0.362 1.510
Lives With Smoker = Yes 0.107 0.254 0.673 1.113
Achievement -0.229 0.075 0.002 0.796
Religiosity -0.073 0.110 0.503 0.929
Social Norm 0.015 0.010 0.117 1.016
Sensation Seeking 0.337 0.130 0.009 1.400
truth Exposure 0.086 0.386 0.823 1.090
z_sigame 0.105 0.228 0.646 1.111
Current Intercept 0.934 2.715 0.731
African American = Yes -0.201 1.718 0.907 0.818
White = Yes 2.467 1.689 0.144 11.789
Female = Yes -1.360 0.808 0.092 0.257
Lives With Smoker = Yes 0.081 0.487 0.868 1.084
Achievement -0.256 0.153 0.094 0.774
Religiosity -0.545 0.290 0.060 0.580
Social Norm 0.066 0.016 0.000 1.068
Sensation Seeking -0.246 0.242 0.310 0.782
truth Exposure 0.567 0.781 0.468 1.762
z_sigame -0.143 0.454 0.752 0.867
a The reference category is: Never, not susceptible.

Multinomial Logistic Regression – SISC-nonconformists
Smoking Status(a) B SE B Sig. Exp(B)
Never, susceptible Intercept -1.245 1.529 0.416
African American = Yes -1.292 0.873 0.139 0.275
White = Yes -0.465 0.556 0.403 0.628
Female = Yes -0.146 0.491 0.767 0.865
Lives With Smoker = Yes 0.098 0.296 0.741 1.103
Achievement -0.087 0.095 0.359 0.917
Religiosity 0.083 0.129 0.519 1.087
Social Norm 0.016 0.011 0.152 1.016
Sensation Seeking 0.148 0.150 0.324 1.159
truth Exposure 0.337 0.469 0.472 1.401
z_sinonc 0.043 0.291 0.882 1.044
Tried Intercept -0.461 1.271 0.717
African American = Yes -1.185 0.647 0.067 0.306
White = Yes -0.949 0.444 0.032 0.387
Female = Yes 0.405 0.435 0.351 1.500
Lives With Smoker = Yes 0.114 0.248 0.646 1.120

213

Achievement -0.221 0.074 0.003 0.802
Religiosity -0.081 0.108 0.452 0.922
Social Norm 0.013 0.009 0.147 1.014
Sensation Seeking 0.371 0.130 0.004 1.450
truth Exposure 0.143 0.389 0.713 1.154
z_sinonc 0.067 0.235 0.775 1.069
Current Intercept 0.395 2.626 0.881
African American = Yes -0.520 1.647 0.752 0.595
White = Yes 2.140 1.558 0.170 8.496
Female = Yes -1.147 0.784 0.143 0.318
Lives With Smoker = Yes 0.156 0.459 0.734 1.169
Achievement -0.291 0.142 0.041 0.748
Religiosity -0.444 0.264 0.093 0.642
Social Norm 0.060 0.014 0.000 1.061
Sensation Seeking -0.150 0.244 0.537 0.860
truth Exposure 0.961 0.768 0.211 2.615
z_sinonc -0.116 0.457 0.800 0.891
a The reference category is: Never, not susceptible.

Multinomial Logistic Regression – SISC-goody-goodies
Smoking Status(a) B SE B Sig. Exp(B)
Never, susceptible Intercept -1.268 1.513 0.402
African American = Yes -1.274 0.868 0.142 0.280
White = Yes -0.437 0.542 0.420 0.646
Female = Yes -0.178 0.499 0.721 0.837
Lives With Smoker = Yes 0.100 0.298 0.736 1.105
Achievement -0.087 0.095 0.355 0.916
Religiosity 0.080 0.130 0.539 1.083
Social Norm 0.015 0.011 0.159 1.015
Sensation Seeking 0.163 0.149 0.274 1.177
truth Exposure 0.348 0.465 0.454 1.417
z_sigood 0.052 0.270 0.846 1.054
Tried Intercept -0.737 1.261 0.559
African American = Yes -1.194 0.648 0.065 0.303
White = Yes -0.919 0.433 0.034 0.399
Female = Yes 0.451 0.436 0.301 1.570
Lives With Smoker = Yes 0.130 0.249 0.600 1.139
Achievement -0.212 0.074 0.004 0.809
Religiosity -0.067 0.109 0.536 0.935
Social Norm 0.014 0.009 0.133 1.014
Sensation Seeking 0.356 0.130 0.006 1.428
truth Exposure 0.207 0.383 0.589 1.230

214

z_sigood -0.208 0.230 0.366 0.812
Current Intercept -0.122 2.643 0.963
African American = Yes -0.543 1.692 0.748 0.581
White = Yes 2.124 1.624 0.191 8.361
Female = Yes -0.856 0.799 0.284 0.425
Lives With Smoker = Yes 0.176 0.452 0.696 1.193
Achievement -0.294 0.142 0.038 0.745
Religiosity -0.406 0.267 0.128 0.666
Social Norm 0.062 0.014 0.000 1.064
Sensation Seeking -0.233 0.237 0.324 0.792
truth Exposure 1.022 0.760 0.179 2.778
z_sigood -0.675 0.480 0.160 0.509
a The reference category is: Never, not susceptible.

Multinomial Logistic Regression – SISC-straightedge
Smoking Status(a) B SE B Sig. Exp(B)
Never, susceptible Intercept -1.478 1.585 0.351
African American = Yes -1.246 0.890 0.162 0.288
White = Yes -0.514 0.549 0.349 0.598
Female = Yes -0.093 0.515 0.857 0.911
Lives With Smoker = Yes 0.200 0.307 0.515 1.221
Achievement -0.081 0.098 0.405 0.922
Religiosity 0.130 0.136 0.339 1.138
Social Norm 0.013 0.013 0.289 1.013
Sensation Seeking 0.120 0.150 0.422 1.128
truth Exposure 0.379 0.477 0.427 1.461
z_sisxe 0.364 0.262 0.165 1.439
Tried Intercept -0.209 1.279 0.870
African American = Yes -1.368 0.696 0.050 0.255
White = Yes -0.876 0.436 0.045 0.416
Female = Yes 0.359 0.434 0.408 1.432
Lives With Smoker = Yes 0.095 0.258 0.712 1.100
Achievement -0.227 0.075 0.003 0.797
Religiosity -0.074 0.110 0.501 0.929
Social Norm 0.016 0.010 0.103 1.016
Sensation Seeking 0.333 0.130 0.010 1.395
truth Exposure 0.085 0.386 0.826 1.088
z_sisxe 0.089 0.210 0.671 1.093
Current Intercept 1.045 2.763 0.705
African American = Yes -0.380 1.811 0.834 0.684
White = Yes 2.589 1.738 0.136 13.317
Female = Yes -1.354 0.792 0.087 0.258

215

Lives With Smoker = Yes -0.033 0.506 0.949 0.968
Achievement -0.272 0.155 0.079 0.762
Religiosity -0.530 0.288 0.066 0.589
Social Norm 0.068 0.016 0.000 1.071
Sensation Seeking -0.259 0.243 0.285 0.772
truth Exposure 0.573 0.783 0.464 1.773
z_sisxe 0.279 0.413 0.499 1.322
a The reference category is: Never, not susceptible.

Multinomial Logistic Regression – SISC-class clowns
Smoking Status(a) B SE B Sig. Exp(B)
Never, susceptible Intercept -1.538 1.578 0.330
African American = Yes -1.229 0.879 0.162 0.293
White = Yes -0.531 0.556 0.339 0.588
Female = Yes -0.052 0.511 0.919 0.950
Lives With Smoker = Yes 0.263 0.303 0.385 1.301
Achievement -0.090 0.096 0.349 0.914
Religiosity 0.118 0.135 0.380 1.126
Social Norm 0.012 0.012 0.353 1.012
Sensation Seeking 0.127 0.150 0.399 1.135
truth Exposure 0.436 0.473 0.357 1.546
z_siclas 0.066 0.288 0.818 1.068
Tried Intercept -0.342 1.288 0.791
African American = Yes -1.360 0.697 0.051 0.257
White = Yes -0.792 0.438 0.071 0.453
Female = Yes 0.290 0.435 0.505 1.336
Lives With Smoker = Yes 0.127 0.254 0.618 1.135
Achievement -0.227 0.075 0.002 0.797
Religiosity -0.069 0.110 0.532 0.933
Social Norm 0.016 0.010 0.100 1.016
Sensation Seeking 0.330 0.130 0.011 1.390
truth Exposure 0.102 0.385 0.791 1.107
z_siclas 0.289 0.238 0.226 1.335
Current Intercept 1.029 2.716 0.705
African American = Yes -0.245 1.589 0.877 0.783
White = Yes 2.118 1.502 0.158 8.312
Female = Yes -1.623 0.835 0.052 0.197
Lives With Smoker = Yes 0.190 0.497 0.703 1.209
Achievement -0.271 0.157 0.085 0.763
Religiosity -0.554 0.283 0.051 0.575
Social Norm 0.068 0.016 0.000 1.071
Sensation Seeking -0.224 0.243 0.357 0.799

216

truth Exposure 0.717 0.784 0.360 2.049
z_siclas 0.769 0.518 0.138 2.157
a The reference category is: Never, not susceptible.

Table 5.33 Multinomial Logistic Regression – Achievers _
Smoking Status(a) B SE B Sig. Exp(B)
Never, susceptible Intercept -2.156 1.659 0.194
African American = Yes -0.962 0.884 0.276 0.382
White = Yes -0.457 0.549 0.405 0.633
Female = Yes 0.159 0.526 0.763 1.172
Lives With Smoker = Yes 0.230 0.308 0.454 1.259
Achievement -0.037 0.100 0.712 0.964
Religiosity 0.096 0.136 0.480 1.100
Social Norm 0.009 0.013 0.466 1.009
Sensation Seeking 0.122 0.151 0.421 1.129
truth(tm) Exposure 0.486 0.478 0.309 1.626
sisc_achievers -0.853 0.432 0.048 0.426
Tried Intercept -1.321 1.382 0.339
African American = Yes -0.942 0.711 0.185 0.390
White = Yes -0.733 0.444 0.099 0.481
Female = Yes 0.648 0.454 0.154 1.911
Lives With Smoker = Yes 0.113 0.257 0.661 1.119
Achievement -0.152 0.078 0.052 0.859
Religiosity -0.100 0.114 0.382 0.905
Social Norm 0.013 0.010 0.213 1.013
Sensation Seeking 0.329 0.133 0.014 1.390
truth(tm) Exposure 0.241 0.401 0.548 1.272
sisc_achievers -1.426 0.380 0.000 0.240
Current Intercept 0.462 2.741 0.866
African American = Yes -0.079 1.676 0.962 0.924
White = Yes 2.422 1.648 0.142 11.264
Female = Yes -1.188 0.803 0.139 0.305
Lives With Smoker = Yes 0.009 0.491 0.986 1.009
Achievement -0.226 0.151 0.133 0.797
Religiosity -0.519 0.283 0.066 0.595
Social Norm 0.064 0.016 0.000 1.066
Sensation Seeking -0.226 0.241 0.349 0.798
truth(tm) Exposure 0.614 0.786 0.434 1.849
sisc_achievers -0.785 0.719 0.275 0.456
a The reference category is: Never, not susceptible.

217

Multinomial Logistic Regression – Interaction of SISC-Popular with Sensation Seeking
Smoking Status(a) Variable B SE B Sig. e
Β

Susceptible Intercept -1.044 1.451 0.472
African American = Yes -1.233 0.906 0.174 0.291
White = Yes -0.627 0.568 0.269 0.534
Female = Yes 0.078 0.518 0.881 1.081
Live with smoker = Yes 0.261 0.309 0.399 1.298
Achievement -0.061 0.097 0.532 0.941
Social Norm 0.010 0.013 0.446 1.010
Religiosity 0.070 0.136 0.605 1.073
SISC-popular 0.080 0.234 0.733 1.083
Sensation Seeking 0.185 0.158 0.243 1.203
SISC-popular
*
SS 0.323 0.154 0.036 1.382
*

Tried Intercept 1.136 1.125 0.313
African American = Yes -1.303 0.694 0.060 0.272
White = Yes -0.817 0.441 0.064 0.442
Female = Yes 0.422 0.435 0.332 1.525
Live with smoker = Yes 0.128 0.260 0.624 1.136
Achievement -0.219 0.077 0.005 0.804
**

Social Norm 0.015 0.010 0.125 1.015
Religiosity -0.094 0.110 0.393 0.911
SISC-popular 0.212 0.186 0.255 1.236
Sensation Seeking 0.366 0.135 0.007 1.443
**

SISC-popular
*
SS -0.016 0.122 0.893 0.984
Current Intercept -0.696 2.701 0.797
African American = Yes 0.321 1.684 0.849 1.378
White = Yes 2.659 1.666 0.110 14.288
Female = Yes -1.270 0.787 0.106 0.281
Live with smoker = Yes 0.143 0.506 0.777 1.154
Achievement -0.205 0.146 0.161 0.814
Social Norm 0.065 0.016 0.000 1.067
***

Religiosity -0.599 0.293 0.041 0.549
*

SISC-popular 0.390 0.413 0.345 1.477
Sensation Seeking -0.223 0.248 0.369 0.800
SISC-popular
*
SS 0.197 0.232 0.394 1.218
a The reference category is: Never, not susceptible.
*
p < .05
**
p < .01
***
p < .001

218

Multinomial Logistic Regression – Interaction of SISC-Smart with Sensation
Seeking_______________________________________________________________
Smoking Status(a) B SE B Sig. e
Β

Susceptible Intercept -0.812 1.547 0.599
African American = Yes -1.079 0.901 0.231 0.340
White = Yes -0.654 0.566 0.248 0.520
Female = Yes 0.070 0.531 0.895 1.073
Live with smoker = Yes 0.181 0.315 0.565 1.199
Achievement -0.093 0.105 0.374 0.911
Social Norm 0.012 0.013 0.326 1.013
Religiosity 0.102 0.136 0.453 1.107
SISC-smart -0.060 0.297 0.841 0.942
Sensation Seeking 0.151 0.157 0.339 1.162
SISC-smart
*
SS 0.390 0.172 0.023 1.477
*

Tried Intercept 0.660 1.229 0.591
African American = Yes -1.145 0.707 0.105 0.318
White = Yes -0.842 0.439 0.055 0.431
Female = Yes 0.487 0.441 0.269 1.628
Live with smoker = Yes 0.163 0.258 0.528 1.177
Achievement -0.186 0.082 0.023 0.830
*

Social Norm 0.016 0.010 0.110 1.016
Religiosity -0.083 0.110 0.452 0.921
SISC-smart -0.379 0.248 0.126 0.685
Sensation Seeking 0.350 0.136 0.010 1.419
*

SISC-smart
*
SS 0.156 0.153 0.308 1.169
Current Intercept -1.529 3.005 0.611
African American = Yes 0.327 1.627 0.841 1.387
White = Yes 2.645 1.717 0.123 14.089
Female = Yes -1.027 0.817 0.209 0.358
Live with smoker = Yes 0.090 0.490 0.853 1.095
Achievement -0.181 0.158 0.252 0.834
Social Norm 0.065 0.016 0.000 1.068
***

Religiosity -0.528 0.278 0.058 0.590
SISC-smart -0.569 0.503 0.258 0.566
Sensation Seeking -0.187 0.316 0.554 0.830
SISC-smart
*
SS 0.142 0.285 0.617 1.153
a The reference category is: Never, not susceptible.

219

Multinomial Logistic Regression – Interaction of SISC-Techie with Sensation
Seeking_________________________________________________________________
Smoking Status(a) B SE B Sig. Exp(B)
Susceptible Intercept -0.902 1.425 0.527
African American = Yes -1.323 0.893 0.139 0.266
White = Yes -0.645 0.560 0.249 0.525
Female = Yes -0.144 0.526 0.784 0.866
Live with smoker = Yes 0.298 0.305 0.329 1.347
Achievement -0.064 0.101 0.524 0.938
Social Norm 0.010 0.013 0.434 1.010
Religiosity 0.127 0.137 0.353 1.136
SISC-techies -0.189 0.383 0.622 0.828
Sensation Seeking 0.125 0.159 0.431 1.134
SISC-techies
*
SS -0.463 0.198 0.020 0.629
*

Tried Intercept 1.346 1.117 0.228
African American = Yes -1.306 0.700 0.062 0.271
White = Yes -0.866 0.434 0.046 0.421
*

Female = Yes 0.293 0.447 0.512 1.341
Live with smoker = Yes 0.120 0.256 0.640 1.127
Achievement -0.218 0.077 0.005 0.804
**

Social Norm 0.015 0.010 0.123 1.015
Religiosity -0.082 0.110 0.456 0.921
SISC-techies -0.325 0.329 0.324 0.723
Sensation Seeking 0.366 0.135 0.007 1.442
**

SISC-techies
*
SS 0.079 0.209 0.705 1.082
Current Intercept -0.464 2.773 0.867
African American = Yes -0.087 1.797 0.961 0.916
White = Yes 2.791 1.805 0.122 16.305
Female = Yes -1.404 0.910 0.123 0.246
Live with smoker = Yes 0.055 0.510 0.915 1.056
Achievement -0.240 0.143 0.094 0.787
Social Norm 0.065 0.016 0.000 1.067
***

Religiosity -0.513 0.291 0.078 0.599
SISC-techies 0.107 0.483 0.824 1.113
Sensation Seeking -0.154 0.264 0.560 0.857
SISC-techies
*
SS -0.334 0.241 0.166 0.716

Multinomial Logistic Regression – Interaction of SISC-Non-conformists with Sensation
Seeking_________________________________________________________
Smoking Status(a) B SE B Sig. Exp(B)
Susceptible Intercept -0.339 1.342 0.800

220

African American = Yes -1.222 0.880 0.165 0.295
White = Yes -0.523 0.564 0.354 0.593
Female = Yes -0.076 0.495 0.877 0.926
Live with smoker = Yes 0.071 0.295 0.810 1.073
Achievement -0.087 0.094 0.353 0.916
Social Norm 0.016 0.011 0.147 1.016
Religiosity 0.061 0.128 0.633 1.063
SISC-nonconformists 0.332 0.336 0.324 1.393
Sensation Seeking 0.064 0.159 0.685 1.067
SISC-nonconformists
*
SS -0.356 0.224 0.112 0.700
Tried Intercept 1.463 1.106 0.186
African American = Yes -1.152 0.661 0.081 0.316
White = Yes -1.009 0.453 0.026 0.364
*

Female = Yes 0.434 0.437 0.321 1.543
Live with smoker = Yes 0.093 0.249 0.708 1.098
Achievement -0.225 0.075 0.003 0.799
**

Social Norm 0.014 0.009 0.126 1.014
Religiosity -0.099 0.109 0.360 0.905
SISC-nonconformists 0.512 0.298 0.086 1.669
Sensation Seeking 0.309 0.139 0.026 1.363
*

SISC-nonconformists
*
SS -0.467 0.192 0.015 0.627
*

Current Intercept -0.045 2.515 0.986
African American = Yes -0.276 1.580 0.861 0.759
White = Yes 1.984 1.517 0.191 7.274
Female = Yes -0.948 0.779 0.224 0.387
Live with smoker = Yes 0.074 0.465 0.873 1.077
Achievement -0.250 0.140 0.073 0.779
Social Norm 0.058 0.014 0.000 1.059
***

Religiosity -0.503 0.257 0.051 0.605
SISC-nonconformists 0.378 0.519 0.466 1.460
Sensation Seeking -0.223 0.257 0.385 0.800
SISC-nonconformists
*
SS -0.543 0.326 0.096 0.581
a The reference category is: Never, not susceptible.

Multinomial Regression – SISC-theater by Social Norms .
Smoking Status(a) B SE B Sig. Exp(B)
Never, susceptible Intercept -1.431 1.424 0.315
White = Yes -1.158 0.919 0.207 0.314
African-American = Yes -0.620 0.572 0.279 0.538
Female = Yes 0.216 0.536 0.687 1.241

221

Lives with Smoker = Yes 0.394 0.329 0.231 1.482
Achievement -0.083 0.097 0.395 0.921
Social Norm 0.002 0.021 0.935 1.002
Religiousitysity 0.152 0.136 0.264 1.164
SISC-Theater -0.991 0.391 0.011 0.371
*

Sensation Seeking 0.193 0.158 0.223 1.213
Social Norm
*
SISC-theater 0.007 0.020 0.742 1.007
Tried Intercept 1.318 1.161 0.256
White = Yes -1.659 0.753 0.028 0.190
*

African-American = Yes -1.089 0.451 0.016 0.336
*

Female = Yes 0.663 0.457 0.146 1.941
Lives with Smoker = Yes 0.049 0.271 0.857 1.050
Achievement -0.229 0.076 0.003 0.795
**

Social Norm 0.029 0.012 0.015 1.029
*

Religiousitysity -0.068 0.114 0.550 0.934
SISC-theater -0.296 0.260 0.254 0.744
Sensation Seeking 0.384 0.138 0.005 1.468
**

Social Norm
*
SISC-theater 0.033 0.013 0.010 1.034
*

Current Intercept -0.007 2.723 0.998
White = Yes 0.524 1.645 0.750 1.688
African-American = Yes 3.200 1.758 0.069 24.539
Female = Yes -1.317 0.807 0.103 0.268
Lives with Smoker = Yes -0.024 0.505 0.962 0.976
Achievement -0.217 0.153 0.155 0.805
Social Norm 0.064 0.019 0.001 1.066
**

Religiousitysity -0.607 0.292 0.038 0.545
SISC-theater 0.330 0.559 0.555 1.391
Sensation Seeking -0.254 0.248 0.304 0.775
Social Norm
*
SISC-theater -0.006 0.018 0.747 0.994
a The reference category is: Never, not susceptible.

Table 5.18 Multinomial Regression – SISC-Smart by Social Norm
Smoking Status(a) B SE B Sig. Exp(B)
Never, susceptible Intercept -0.821 1.497 0.583
White = Yes -1.098 0.889 0.216 0.333
African American = Yes -0.545 0.556 0.327 0.580
Female = Yes -0.003 0.526 0.995 0.997
Lives with Smoker = Yes 0.250 0.311 0.423 1.283
Achievement -0.070 0.103 0.497 0.933
Social Norm 0.008 0.014 0.575 1.008
Religiousitysity 0.090 0.133 0.500 1.094

222

SISC-smart -0.122 0.308 0.694 0.886
Sensation Seeking 0.130 0.150 0.386 1.139
Social Norm
*
SISC-smart -0.008 0.020 0.701 0.993
Tried Intercept 0.210 1.257 0.868
White = Yes -0.984 0.719 0.171 0.374
African American = Yes -0.700 0.438 0.110 0.497
Female = Yes 0.584 0.448 0.192 1.793
Lives with Smoker = Yes 0.273 0.260 0.294 1.313
Achievement -0.166 0.082 0.042 0.847
*

Social Norm 0.022 0.010 0.033 1.022
*

Religiousitysity -0.056 0.113 0.620 0.946
SISC-smart -0.314 0.253 0.215 0.731
Sensation Seeking 0.347 0.134 0.010 1.415
*

Social Norm
*
SISC-smart 0.036 0.015 0.019 1.037
*

Current Intercept -0.886 2.928 0.762
White = Yes 0.381 1.602 0.812 1.464
African American = Yes 2.910 1.734 0.093 18.352
Female = Yes -1.124 0.837 0.179 0.325
Lives with Smoker = Yes 0.119 0.489 0.808 1.127
Achievement -0.158 0.159 0.319 0.854
Social Norm 0.062 0.017 0.000 1.064
***

Religiousitysity -0.546 0.281 0.052 0.579
SISC-smart -0.375 0.523 0.473 0.687
Sensation Seeking -0.269 0.244 0.271 0.764
Social Norm
*
SISC-smart -0.007 0.024 0.773 0.993
a The reference category is: Never, not susceptible.

Abstract (if available)

Abstract This dissertation introduces social identity (Tajfel, 1978

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The social groups approach to quitting smoking: An examination of smoking cessation online and offline through the influence of social norms, social identification, social capital and social support

PDF

The role of depression symptoms on social information processing and tobacco use among adolescents

PDF

Relationship formation and information sharing to promote risky health behavior on social media

PDF

Acculturation and Latino immigrants ' health attitudes and behaviors: a bi-dimensional acculturation model

PDF

Contagious: social norms about health in work group networks

PDF

Rediscovering the communication engine of neighborhood effects: how the interaction of residents and community institutions impacts health literacy and how it can be leveraged to improve health c...

PDF

Genetic variants and smoking progression in Chinese adolescents

PDF

Constructing collective memories of movements for African peoples' equality: Containing the legitimation crisis for white domination

PDF

A network analysis of online and offline social influence processes in relation to adolescent smoking and alcohol use

PDF

Neuropsychological mechanisms predictive of adolescent addictive behaviors

PDF

Contextualizing social network influences on substance use among high risk adolescents

PDF

Adolescent social networks, smoking, and loneliness

PDF

Parents' attunement: relation to adolescent problem behavior and the moderating role of marital conflict

PDF

The link between maternal depression and adolescent daughters' risk behavior: the mediating and moderating role of family

PDF

A multilevel communication approach to understanding human trafficking prevention behaviors in Indonesia

PDF

Social network influences on depressive symptoms among Chinese adolescents

PDF

Hispanic advertising as social orientation: an examination of the advertising industry as a field of cultural production

PDF

The role of social control in promoting healthy eating behavior among Chinese immigrants: an ecological approach

PDF

The effects of mindfulness on adolescent cigarette smoking: Measurement, mechanisms, and theory

PDF

Apparel and attribute: The social construction of status in New England colonies and the United States

Asset Metadata

Creator Moran, Meghan Bridgid (author)

Core Title The role of social identity in adolescent smoking behavior

Contributor Electronically uploaded by the author (provenance)

School Annenberg School for Communication

Degree Doctor of Philosophy

Degree Program Communication

Publication Date 08/06/2009

Defense Date 05/11/2009

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

Tag adolescent smoking,adolescent tobacco use,integrative model of behavioral prediction,OAI-PMH Harvest,sensation seeking,social identity,social norms

Language English

Advisor Murphy, Sheila T. (committee chair), Ball-Rokeach, Sandra (committee member), Valente, Thomas W. (committee member)

Creator Email meghanmo@gmail.com,meghanmo@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m2504

Unique identifier UC1219663

Identifier etd-Moran-3098 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-186620 (legacy record id),usctheses-m2504 (legacy record id)

Legacy Identifier etd-Moran-3098.pdf

Dmrecord 186620

Document Type Dissertation

Rights Moran, Meghan Bridgid

Type texts

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

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu