University of Southern California Dissertations and Theses

Normative and network influences on electronic cigarette use among adolescents

(USC Thesis Other)

Normative and network influences on electronic cigarette use among adolescents

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content NORMATIVE AND NETWORK INFLUENCES ON
ELECTRONIC CIGARETTE USE AMONG ADOLESCENTS
by
Sarah Elizabeth Piombo, MPH
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(PREVENTIVE MEDICINE, HEALTH BEHAVIOR RESEARCH)
August 2024
Copyright 2024 Sarah E. Piombo

ii
Dedication
To my parents, John and Elizabeth,
who have always believed in me.

iii
Acknowledgements
First and foremost, I would like to acknowledge and thank my doctoral advisor and
dissertation chair, Dr. Thomas Valente, who has provided mentorship and wisdom throughout
my doctoral training and the writing of this dissertation. Without his knowledge, guidance, and
support, I wouldn’t be where I am today. I’m truly appreciative of the time and energy he has
invested in my education. He has taught me not only what it is to be a great academic, but a great
leader as well.
I would like to acknowledge my committee members, Dr. Kayla de la Haye, Dr.
Kimberly Miller, Dr. Lindsay Young, and Dr. Jessica Barrington-Trimis. I want to express my
gratitude and appreciation for your willingness to provide intellectual insights, mentorship, and
support. I am incredibly fortunate to have these brilliant, strong, and successful women as role
models. A special thank you to Dr. Miller, whom I’ve known and worked with since 2017. She
was one of my earliest supporters at USC and she inspired me to continue my education in
preventive medicine.
I would like to thank faculty members, colleagues, and collaborators at USC and
elsewhere that I’ve had the pleasure of working with, including Dr. Jennifer Unger, Dr. David
Freyer, Dr. Myles Cockburn, and Dr. George Vega Yon. I am also thankful for the ADVANCE
research team for their hard work and commitment to this project.
I’m incredibly grateful for the friends I’ve made in the doctoral program, Julia Stal and
Kelsey McAlister. These friendships have been a great source of support and joy. They have
helped me through the most difficult parts of this journey, and I honestly don’t know what I
would’ve done without them.

iv
I would like to thank my friends and family for their long-standing support, including my
best friend Ashley Waite, Dr. Nethika Ariyasinghe, Cordelia Leeder, and Olivia Leeder. A
tremendous thank you to Andy Kampfschulte for his partnership and steadfast support, which
I’m incredibly grateful for.
Finally, I must acknowledge and express my immense gratitude for my parents, John and
Elizabeth Piombo, for their love and everlasting support. They have always encouraged me to
pursue my education and made incredible sacrifices for me to do so. Thank you for being my
greatest supporters and always believing in me.
This dissertation was supported by the National Institutes of Health, National Institute on Drug
Abuse (F31DA05618).

v
TABLE OF CONTENTS
DEDICATION................................................................................................................................ ii
ACKNOWLEDGEMENTS........................................................................................................... iii
LIST OF TABLES....................................................................................................................... viii
LIST OF FIGURES .........................................................................................................................x
ABSTRACT................................................................................................................................... xi
CHAPTER 1: INTRODUCTION....................................................................................................1
Introduction
Problem significance................................................................................................2
Health risks of e-cigarettes.......................................................................................2
Social networks........................................................................................................5
Cigarette smoking in social networks..................................................................... 5
The influence of social norms in networks..............................................................6
Gaps in the literature..................................................................................10
Stochastic actor-oriented models and smoking......................................................10
Gaps in the literature..................................................................................13
Diffusion of behaviors in social networks.............................................................14
Network simulation studies of smoking diffusion .................................................15
Network-based interventions for reducing e-cigarette use ....................................15
Gaps in the literature..................................................................................16
Summary of gaps in the literature......................................................................................16
Introduction to the dissertation studies ..............................................................................17
Specific aims..........................................................................................................18
Research design and methods................................................................................20
Data sources and collection ...................................................................................20
Data processing......................................................................................................22
Tables.................................................................................................................................23
Figures................................................................................................................................25
CHAPTER 2: THE IMPACT OF SOCIAL NETWORKS AND NORMS ON
E-CIGARETTE USE AMONG ADOLESCENTS IN SOUTHERN CALIFORNIA:
A PROSPECTIVE COHORT STUDY .........................................................................................26
Abstract..............................................................................................................................27
Introduction........................................................................................................................28
E-cigarette use........................................................................................................28
Social networks......................................................................................................28
Social Norms..........................................................................................................28
Methods..............................................................................................................................31
Participant recruitment...........................................................................................31
Measures................................................................................................................32

vi
Statistical analysis..................................................................................................34
Confirmatory factor analysis......................................................................34
Logistic regression .....................................................................................34
Multiple imputation ...................................................................................35
Results................................................................................................................................36
Discussion..........................................................................................................................38
Strengths and limitations........................................................................................40
Public health implications......................................................................................41
Tables.................................................................................................................................43
Figures................................................................................................................................46
CHAPTER 3: NETWORK DYNAMICS OF SOCIAL INFLUENCE ON
E-CIGARETTE USE AMONG AN ETHNICALLY DIVERSE ADOLESCENT COHORT .....47
Abstract..............................................................................................................................48
Introduction........................................................................................................................49
Social network framework.....................................................................................50
Peer influence and perceived norms ......................................................................51
The current study ...................................................................................................52
Material and methods.........................................................................................................52
Participants and recruitment ..................................................................................52
Measures................................................................................................................53
Analysis..................................................................................................................54
Siena meta-analysis................................................................................................57
Results................................................................................................................................57
Friendship network dynamics................................................................................58
E-cigarette use behavior dynamics ........................................................................59
Discussion..........................................................................................................................59
Strengths ................................................................................................................60
Limitations.............................................................................................................61
Conclusions............................................................................................................61
Tables.................................................................................................................................63
Figures................................................................................................................................66
CHAPTER 4: THE IMPACT OF SOCIAL NORMS ON DIFFUSION DYNAMICS:
A SIMULATION OF E-CIGARETTE USE BEHAVIOR ...........................................................67
Abstract..............................................................................................................................68
Introduction........................................................................................................................70
Theoretical foundations .........................................................................................70
Simulating network interventions..........................................................................72
Methods..............................................................................................................................73
Measures................................................................................................................74
Analysis..................................................................................................................78
Results................................................................................................................................79
Discussion..........................................................................................................................83
Strengths ................................................................................................................84
Limitations ............................................................................................................85

vii
Conclusions............................................................................................................85
Tables.................................................................................................................................86
Figures................................................................................................................................89
CHAPTER 5: DISCUSSION.........................................................................................................92
Discussion..........................................................................................................................93
Study 1 ...................................................................................................................93
Study 2 ...................................................................................................................94
Study 3 ...................................................................................................................94
Summary of Findings.............................................................................................95
Implications........................................................................................................................96
Methodological implications .................................................................................96
Public health implications......................................................................................97
Limitations.............................................................................................................97
Future Research .................................................................................................................98
Conclusion .......................................................................................................................100
REFERENCES ............................................................................................................................101
APPENDICES
Appendix A......................................................................................................................110

viii
LIST OF TABLES
Table 1. Demographic and Behavioral Characteristics..................................................................23
Table 2. School Friendship Network Characteristics ....................................................................24
Table 3. Demographic Characteristics of the Analytic Sample .....................................................43
Table 4. Perceived Pro-vaping Norms Wave 1 and Wave 2 Confirmatory Factor Analysis.........44
Table 5. Past 6-month Vaping as a Function of Demographic Characteristics,
Network Exposure to Vaping, and Perceived Pro-Vaping Norms ................................................45
Table 6. Demographic and Behavior Characteristics ....................................................................63
Table 7. Social Network Characteristics........................................................................................64
Table 8. Meta-analysis of changes in friendship networks and e-cigarette use.............................65
Table 9. Diffusion prevalence and rate by intervention strategy
and social norms or seed condition................................................................................................86
Table 10. Diffusion prevalence and rate by intervention strategy
and social norms or seed condition...............................................................................................87
Table 11. Multivariate regressions of diffusion prevalence and rate on
network metrics, seed condition, intervention strategy and e-cigarette social norms ...................88
Supplemental Table 1. Imputed Model of Past 6-month Vaping
as a Function of Demographic Characteristics, Network Exposure to Vaping,
and Peer Vaping Norms..............................................................................................................110
Supplemental Table 2. Demographic Characteristics of Participants Lost to Follow Up ...........111
Supplemental Table 3. School Friendship Network Characteristics ...........................................112
Supplemental Table 4. Meta-analysis of changes in friendship networks
and e-cigarette use (extended results).........................................................................................113

ix
Supplemental Table 5. Logistic regression of individual e-cigarette use
on demographic covariates, pro-e-cigarette norms, friend selection,
previous network exposure and prior e-cigarette use ..................................................................114

x
LIST OF FIGURES
Figure 1. Data processing workflow diagram................................................................................25
Figure 2. School network plots at three time points ......................................................................46
Figure 3. A high school network at three time points (colorized) .................................................66
Figure 4. Simulated network intervention strategies .....................................................................89
Figure 5. Examples of simulated e-cigarette social norm distributions.........................................90
Figure 6. Diffusion of e-cigarette use over time ............................................................................91
Supplemental Figure 1. Pro-e-cigarette social norm scores from ADVANCE data....................115

xi
ABSTRACT
Electronic cigarettes (e-cigarettes/e-cigs) have rapidly increased in popularity and
become the most commonly used tobacco product among adolescents over the last ten years.
1-4
E-cigarettes are detrimental to health and are correlated with initiation of combustible cigarette
use among adolescents and young adults.5-7 Preventing e-cigarette use by focusing on changing
adolescent behaviors can be difficult due to the complex influences of inter- and intra-personal
factors. E-cigarette use is driven by a combination of psychological, social, environmental, and
systemic factors. Social influence can occur through our social networks, or the relationships and
connections to others that can shape thoughts, attitudes, and behaviors.8 During adolescence, the
importance of friendships, social connections, and the desire to fit in may leave adolescents more
susceptible to the social influence of friends.
The dissertation studies explore the relationship between perceived social norms, peer ecigarette use, and their joint influence on individual e-cigarette use through the analysis of social
network data from a diverse cohort of over 2,000 adolescents. The primary focus of these studies
is identifying and differentiating between the effects of social influence mechanisms driving ecigarette use in adolescent social networks. These studies take a rigorous approach using
different social network analytic methods to examine these relationships. The results of these
studies can be used to inform and design social network interventions to address normative
perceptions surrounding e-cigarettes with the goal of ultimately decreasing use among
adolescents.

1
CHAPTER 1
Introduction

2
INTRODUCTION
Problem significance
Adolescent e-cigarette use is a public health crisis. Electronic cigarette (e-cigarette/ecigs/vaping) use has rapidly gained popularity among youth, quickly ascending to the most
commonly used tobacco product among adolescents.1
In 2023, 10.0% of high school adolescents
in the U.S. reported current e-cigarette use in the past thirty days, with Hispanic/Latine and nonHispanic multi-racial students having the highest rates of e-cigarette use compared to any other
racial or ethnic group.4 Compared to combustible tobacco products, the increased popularity and
social acceptability of e-cigarettes indicates that different dynamics could be driving uptake,
particularly among youth. For many, adolescence is characterized by experimentation and social
pressure to conform, leaving this population susceptible to risky behaviors and peer influence.
Additionally, mounting evidence that e-cigarette use is associated with combustible cigarette use
and other negative health outcomes has led to increased concern about the increasing popularity
of e-cigarette use among adolescents and young adults.
9-12
Health risks of e-cigarettes
E-cigarettes pose multiple health threats and predispose one to continued substance use,
and nicotine addiction. E-cigarette liquid contains nicotine, solvents and flavoring compounds
which are vaporized and inhaled. Nicotine is a highly addictive substance which can lead to a
lifetime of tobacco product use and negative health effects.3
It can also be harmful to brain
development, impulse control, attention, learning and mood regulation.3
Multiple health risks
have been linked to e-cigarette/vaping use, including recent evidence of lung injury and
hospitalization associated with vaping. The CDC has received over 2,900 cases of lung injuries
from vaping of tobacco and marijuana products, and over 60 deaths occurred from an outbreak in

3
2019.13,14 Since the 2019 outbreak, e-cigarette use has continued to be linked to cases of
respiratory distress and nicotine toxicity among youth, sometimes leading to hospitalization or
ICU admission.15 There is emerging evidence that e-cigarette use may also have longer-term
negative pulmonary and immune effects, including inflammation of the airways and increased
risk of asthma and bronchitis.16 The National Institutes of Health (NIH) has called for research to
identify predictors of e-cigarette use among youth in an effort to better understand and address
this public health crisis. Curtailing e-cigarette use among youth is critical for preventing a
lifetime of nicotine addiction, tobacco-related morbidities, and mortalities.
The relationship between e-cigarettes and combustible cigarettes
E-cigarettes are more prevalent and popular among adolescents than combustible
cigarettes. Unfortunately, e-cigarettes can contain higher levels of nicotine, are more addictive,
and are often marketed towards a younger audience.17-19 Studies have shown that adolescents
have more positive attitudes towards e-cigarettes and perceive them as less harmful and more
socially acceptable than combustible cigarettes. In a study among high school students in
California, almost 20% of students believed that e-cigarettes contain water vapor and did not
know that e-cigs are tobacco products, while more than 40% considered them safer than
combustible cigarettes.20 Evidence shows that e-cigarettes are addictive, and that e-cigarette
dependence strongly predicts continuation of use.21 Initiation of e-cigarette use during
adolescence substantially increases the risk of developing a substance dependence problem and
becoming long-term or lifelong e-cigarette user.
Many arguments have been made for e-cigarettes as a tool for harm reduction since
vaping is often presented as a safer alternative to combustible cigarettes. However, e-cigarette
use can be a gateway to combustible cigarette use. Contrary to popular belief, instead of

4
curtailing smoking behavior, adolescents who vape are significantly more likely to use
traditional or combustible cigarettes compared to non-vapers.
22,23 One meta-analysis found that
the probability of combustible cigarette initiation was 30.4% for ever-users of e-cigarettes,
compared to 7.9% for never-users of e-cigarettes.24 Adolescents who vape are also more likely to
be recent or regular cigarette smokers, with e-cigarette users having 3.50 times the odds of past
30-day combustible cigarette use compared to non-e-cigarette users.24
Evidence has shown that vaping can result in both e-cigarette and combustible cigarette
dependence. In one study among adolescents who report past 30-day e-cigarette and combustible
cigarette use, 29.7% reported combustible cigarette dependence while 16.4% reported e-cigarette
dependence.21 Additionally, adolescents who report using any nicotine products have an
increased risk of experimenting with more harmful nicotine products and ultimately have an
increased risk of becoming poly-substance users.25 Therefore, e-cigarettes are not a safe
alternative to combustible cigarettes and are not necessarily a harm reduction tool.
E-cigarette use may function as a gateway to combustible cigarette use or dependence,
which is ultimately associated with an increased risk of health issues and premature mortality.
While prior e-cigarette use is associated with initiation of combustible cigarette use, there is also
growing evidence that being embedded in a social environment that is favorable to e-cigarette
use (having friends that use) is associated with initiation of combustible cigarette use among
adolescents and young adults as well.5-7 The social dynamics surrounding e-cigarette use require
examination and understanding these dynamics is critical to addressing this public health
problem.

5
Social networks
Social networks are the relationships or connections between people, organizations or
other entities.8 Social network analysis (SNA) is a defined theoretical perspective and set of
methods used to analyze and understand how these relationships and social interactions affect
individual attitudes, behaviors, and greater network characteristics and processes. SNA can
provide insight about communication patterns, social influence processes, and the spread of
behaviors or ideas throughout networks. Public health issues can be viewed from this theoretical
perspective because diseases, health knowledge, opinions, and behaviors are often transmitted
from person to person through networks. Network analysis has been widely applied in the social
sciences to explore a variety of health behaviors26-32 and the dissertation studies contribute to this
growing body of evidence by providing unique insight into the mechanisms driving e-cigarette
use in social networks.
Interactions with people in our social networks result in exposure to behaviors and the
exchange of information, opinions, and ideas. Inherent to being embedded in a social network,
we are subject to a constant, dynamic process of social influence which can predispose us to
changes in our own beliefs and behaviors. SNA can be used to understand dynamics at the
individual and network levels that contribute to the spread of e-cigarette use in high school social
networks. Each of the dissertation studies used a different SNA method to understand the social
influence dynamics surrounding e-cigarette use.
Cigarette smoking in social networks
Prior social network research has demonstrated that there is an established relationship
between having friends who smoke and traditional/combustible cigarette use.33-50
A breadth of
studies have explored social networks and combustible tobacco use using a variety of analytic

6
approaches. The association has been well demonstrated by the Add Health study51 in addition to
work by Valente,52-54 Ennett,34-37 Feinberg,
55-57 and others. To provide context and background
for the dissertation studies, this section provides a review of past research that focuses on
normative influence in social networks, stochastic actor-oriented models and smoking, and
agent-based network simulation models and smoking.
The influence of social norms in networks
Social norms have been extensively explored across the social sciences from a range of
theoretical perspectives. Social norms can be further defined as descriptive or injunctive
norms.58,59 Descriptive norms are the perceptions about the behaviors of individuals (what
people do), while injunctive norms are perceptions of approval from one’s social network (what
people think is acceptable behavior).60 Norms are generally disseminated across networks
through communication and interaction with others and are implicit to group membership.
58
Therefore, understanding network dynamics is essential to understanding the influence of social
norms.
There is limited research using social network data to empirically explore the relationship
between perceived norms in networks and individual behavior. However, several studies have
specifically investigated perceived norms in relation to health behaviors, such as family planning
and substance use. Valente et al. (1997)61 explored contraceptive use and social network norms
among Cameroonian women. This study found that perceptions of attitudes and behaviors
towards contraceptive use in one’s social network were associated with women’s own
contraception. Specifically, odds of using contraceptives were significantly higher for women
who perceived that other women in her network used contraception and for women who were
encouraged by network members to use contraception. Thus, both descriptive and injunctive

7
norms had a significant effect on individual behavior. However, in this study, only 31% of
women accurately reported contraceptive status of those in their networks, highlighting the fact
that people often have difficulty accurately assessing the behavior of others. Interestingly,
perception was the driving influential factor, as the perception of contraceptive use was
significantly associated with personal use.61 It did not matter whether the women were correct or
incorrect in their assumptions about the behavior of their network members, just the perception
of what others were doing was powerful enough to have an effect on a woman’s own individual
choice.
Iannotti (1992)62 led one of the first studies comparing the effects of perceived norms and
peer behavior on individual substance use. He found that for cigarettes, marijuana, and polysubstance use, perceived friend use was a stronger predictor than friends’ self-reported use. In a
logistic regression controlling for friends’ substance use and perceived peer pressure, perceived
substance use among friends significantly increased the odds of alcohol (OR = 1.36), cigarette
(OR = 1.63), marijuana (OR=2.12) and poly-substance use (Beta = 0.211).62 Surprisingly,
perceived peer pressure and friends’ use was non-significant across all substances. Iannotti
concluded that perceptions of substance use were more influential than actual peer substance use
or perceived peer pressure.
Rice et al. (2003)63 further explored network influences on substance use among a large
adolescent cohort. Estimates of peer substance use were compared to peer self-reported
substance use for cigarettes, alcohol, and marijuana. Overall, alcohol and cigarette use was
predicted by peer use at the current time period and increased susceptibility to peer pressure at
the previous time period. Estimates of peer substance use were only weakly correlated with peer
self-reports, although in this study students were only allowed to nominate up to three friends.

8
Rice concluded that it is necessary to include peer self-reports of behavior since peer estimates of
behavior appear to have low validity and that additional longitudinal studies comparing these
individual and network influences are needed.63
Recent literature has established stronger associations between perceived norms and
alcohol use in college social networks. DiGuiseppi et al. (2018)64 found that descriptive norms
about drinking frequency were positively associated with number of drinks per week and
individual binge drinking frequency. Further, there was an interaction effect between norms and
indegree/outdegree. When college students perceived more frequent binge drinking among their
peers, there was a significant association between indegree and outdegree with more alcohol
consumption, perhaps indicating a connection between drinking and popularity. Work by Cox et
al. (2019)65 examined misperceptions about drinking behavior among both general and close
peers. Among first year college students, 84.8% overestimated and 11.3% accurately estimated
heavy drinking among general peers. For close peers in their network, 42.0% accurately
estimated heavy drinking while 36.9% overestimated it. Even for a relatively visible and social
behavior among college students, people still had difficulty accurately assessing the behavior of
their close friends, and overestimation of heavy drinking was associated with more frequent
individual heavy drinking.
Fewer studies exist with a specific focus on tobacco use and peer norms in high school
social networks. Work conducted by Valente et al. (2012) compared the effects of different
social influence measures on smoking in a predominately Hispanic/Latine adolescent population
across 9th and 10th grade.54 In this study, both egocentric and sociometric network data were used
to explore the relationship between peer norms, peer influence, and individual smoking behavior.
Consistent with past research, at both waves the average number of perceived friends who smoke

9
was greater than the average number of friends who self-reported as smokers, indicating that
perceptions of smoking were greater than the behavior in personal networks. In cross-sectional
models, measures of perceived friend smoking were significantly associated with individual
smoking, once again reaffirming the strength of the relationship between norms and behavior.
Longitudinal models that included both 9th and 10th grade perceived friend smoking showed that
that an increase in perceived friend smoking (from 9th to 10th grade) increased the odds of
becoming a smoker. Agreement rates between perceived and actual friend smoking were low for
confirmed smokers, ranging from 7-10%.54 In summary, the results of this study showed that
perceived friend smoking was more strongly associated than friend self-reported smoking with
individual smoking behavior and that people had difficulty accurately estimating the smoking
status of others.
A recent systematic review and meta-analysis by East et al. (2021)60 on the longitudinal
effects of social norms and smoking found that descriptive norms were significant predictors of
smoking initiation among youth. Overall descriptive norms (perceived smoking behavior) of
close friends, parents, siblings, and adults were all significantly associated with smoking
initiation (OR = 1.88). In a random effects meta-regression descriptive norms for close friend
smoking (OR = 2.05) and sibling smoking (OR = 2.28) had the strongest positive associations
with smoking initiation These findings indicate that people’s perception of their friends’
behavior has a significant effect on individual behavior. While ample research exists on
descriptive and injunctive norms, few studies have directly and quantitatively explored the
relationship between perceived norms and individual behavior within a social network
framework.

10
Gaps in the literature
Several limitations or gaps in the literature currently exist surrounding peer/social norms,
social networks, and e-cigarette use. The literature comparing the influences of peer norms and
peer behavior is relatively scarce, as illustrated by the limited number of studies described above
that empirically test the relationships between these effects, and even fewer that do
longitudinally. While the difference between peer influence via norms or behavior exposure may
seem nuanced, these are distinct mechanisms that could impact intervention design. The
dissertation studies contribute to the literature by critically comparing these effects in the context
of adolescent e-cigarette use.
Stochastic actor-oriented models and smoking
Stochastic actor-oriented models (SAOM)66,67 are agent based models that estimate the
co-evolution of social networks and behavioral processes and discern between network influence
and selection effects. Factors that influence changes in s social network (i.e. friendship ties) are
called selection effects, while factors that influence changes in behavior (i.e. e-cigarette use) are
called influence effects (see Chapter 3 for details on SAOM). Longitudinal network studies have
found that influence and selection processes can occur simultaneously, sometimes making it
difficult to determine which mechanism is driving behavior change. SAOM are a useful analytic
technique for providing a more detailed understanding of these processes in social
networks.26,39,42,43,52,67-70

In adolescent social networks there is strong evidence of relationships between peer use,
social influence, and individual smoking. This association has been well demonstrated by the
Add Health study71 and work by Mercken,33,39,40,72 Osgood,55,57,73 Lakon,74,75 Schaefer,42,69
Steglich,67 Valente,52-54,76 and others.26,43,68,77,78 Past research has sought to disentangle the

11
effects of peer influence and peer selection on smoking behavior in adolescent social networks.
In studies using SAOM to examine peer influence and peer selection on combustible cigarette
use, there is consistent evidence supporting peer selection effects26,33,39,40,42,52,57,69,72,74,75,78 While
the relationship between peer influence and smoking is established, it is less frequently supported
compared to peer selection.33,39,40,57,69,72,74,75,78,79
The effects of peer influence and selection are complex, dependent on network
characteristics, and may also occur simultaneously. There is substantial evidence that even when
controlling for other predictors of friendship (e.g. gender) youth often tend to form friendships
with others who have similar smoking behaviors.26,40,42,68,69,75
Research by Green (2013)
demonstrated that current smokers are more likely to make friends and remain friends with other
smokers.26 Multiple studies by Mercken et al. have found stronger effects for friend selection on
smoking behavior compared to friend influence,39,40,72 with some evidence of these effects
decreasing over time.33,72 This suggests that influence and selection effects may be stronger at
the beginning of high school before friendships and networks have stabilized.
Schaefer et al. (2012)69 found that smoking behavior informed friend selection, with a
significant positive effect for smoking similarity (b = 0.683; p < .001) and a positive effect of
average similarity on smoking (b = 2.883; p < .001) indicating that individual smoking behavior
becomes more similar to friends over time. This was further supported by Osgood et al. (2015)
where findings revealed a reciprocal nature to these processes, where friends shape problem
behaviors and problem behaviors influence friendship choices, leading to friendship clusters that
are differentiated for that behavior.57
As demonstrated by the research summarized above, there is ample support for peer
selection effects, however, an argument can also be made for peer influence since adolescents

12
rarely initiate smoking without first being exposed to cigarette use among friends. Green et al.
(2013) found that adolescents are significantly more likely to initiate smoking if the rate of
smoking in their personal network is greater than the school smoking prevalence.26 Work by de
la Haye et al. (2019) did not find a significant association between friend selection and smoking
status in high school networks, but instead found that smoking initiation is predicted by peer
influence, or exposure to friends who have tried smoking.77 Alexander et al. (2001)52 found that
in adolescent networks, the likelihood of smoking significantly increased when at least half of
one’s peer network smoked, or if one or two best friends smoked, thus demonstrating peer
influence and network exposure effects on smoking initiation. Together, these findings suggest
that peer influence is a significant driver of smoking initiation, even when prevalence is low in
the overall network. Further, it’s possible that one’s personal network informs perceived social
norms, resulting in inaccurate perceptions about the prevalence of smoking behavior in the
overall network.
Work by Ragan (2016)78 also examined how smoking beliefs affected friendships and
smoking behavior. Ragan found that individual increased approval of smoking and positive
expectations for smoking were both associated with increases in cigarette use (b = 0.155, p <
0.001; b = 0.106, p < 0.001). Over time, expectations for smoking evolved to become more like
friends’ expectations (b = 1.774, p < 0.001). Interestingly, while friend smoking was not
associated with changes in individual smoking expectations, the association between friends’
approval and individual expectations was positive and significant (b = 0.128, p < 0.05). The key
finding from this study was that peer expectations and approval of smoking can shape individual
beliefs and expectations of smoking and these positive beliefs and expectations ultimately are
associated with increased individual cigarette use.

13
Gaps in the literature
The popularity and novelty of e-cigarettes highlights the need for research to understand
the roles of social selection and social influence on e-cigarette use. While the dynamics between
social influence and combustible cigarette use is well established, 7,45 less is known about the
social dynamics surrounding e-cigarette use. The rapid uptake of e-cigarette products among
youth not only parallels the early popularity of cigarettes in previous generations but is also
amplified by the influences of friends and social media. E-cigarettes are currently more prevalent
and popular than combustible cigarettes among adolescents, which suggests that there are
distinct social dynamics influencing use. There is a lack of knowledge about the effects of
perceived social norms on e-cigarette use, and how these norms compare to peer behavior effects
(i.e., the literature has yet to compare normative vs. network effects on e-cigarette use). Further,
the social dynamics driving e-cig use in adolescent networks has not been explored
longitudinally and as adolescent friendship networks evolve over time it is likely that perceived
norms and patterns of e-cigarette use change as well.
Diffusion of behaviors in social networks
Social influence occurs through interactions with people in our social networks when the
attitudes, beliefs, and behaviors of people we are close to impacts our own thoughts and
behaviors. Diffusion of innovations theory was developed to explain the spread of ideas,
practices and behaviors between and within communities or networks.80-82 When people have
more exposure to a behavior, this increases the likelihood of adopting the behavior due to
influence from those in their social network. As network exposure and adoption rates increase,
this results in the behavior spreading throughout the network. There are three classic data sets
that are considered foundational studies of diffusion theory: the medical innovation study83

14
which followed the diffusion of tetracycline prescribing practices among U.S. doctors in the
1950’s, the Brazilian farmers study84 which measured the diffusion of hybrid corn seed over
twenty years, and the Korean family planning study85 which examined family planning practices
over an eleven year period. These classic studies illustrate the diffusion paradigm and exemplify
how behaviors and practices spread throughout networks over time.
Network simulation studies of smoking diffusion
Prior simulation studies have used agent-based modeling with SAOMs to simulate the
effects of social influence and selection on behavior.79,86 A study by Schaefer et al. (2013)79
simulated how changes in peer influence and smoker popularity impact smoking prevalence.
After testing multiple combinations of influence and smoking popularity over thousands of
simulations, when peer influence was present in the model the smoking outcomes were
moderated by the strength of smoking popularity. When smokers were designated as the popular
network members, increases in peer influence led to more smoking initiation. However, when
smokers were unpopular, increases in peer influence had the opposite effect, thereby decreasing
smoking prevalence. Overall, the effects of peer influence and smoking popularity were
interdependent, demonstrating how the network position of initial adopters or users impacts
diffusion dynamics.
Building on this research, adams and Schaefer (2016)86 used simulations to evaluate how
initial smoking prevalence can also moderate changes in smoking throughout networks. This
study illustrated how interventions impact networks that are operating under different initial
conditions, emphasizing that interventions should be tailored to networks and that outcomes are
context dependent. Using data from 85 schools, an empirical distribution of smoking and the
relationship to friendship networks was generated, then SAOMs were fit to models with varying

15
prevalence distributions and the smoking outcomes of the simulations were recorded. In
summary, when peer influence was low, smoking rates did not increase. When peer influence
was moderate, increased smoking popularity resulted in stagnant smoking rates, while decreased
smoking popularity resulted in decreased smoking rates. Simulation models with the highest
levels of peer influence and smoking prevalence resulted in increased smoking rates, even in
networks with a low initial prevalence of smokers. This demonstrates that peer influence and
smoking popularity could override a network majority of non-smokers at baseline. The key point
of this study is that if an intervention was implemented and theoretically able to produce the
same effect (i.e. altering peer influence) then the results of the intervention would vary
depending on contextual network factors such as baseline smoking prevalence, popularity of
smoking, or other network characteristics.
Network-based interventions for reducing e-cigarette use
In health behavior change interventions, there is mounting evidence that including a
social network component is more effective than targeting individuals alone.87 Understanding
social network dynamics is key to implementing effective interventions and achieving desirable
outcomes. Multiple network-based intervention approaches exist, but selecting the optimal
strategy is dependent on the network dynamics and the target behavior.
Preliminary studies have begun to show potential for peer-led e-cigarette use
interventions.88 A recent study recruited peer leaders in schools to model healthy norms and
reduced acceptability of vaping. The leaders’ message had relatively good exposure throughout
the network, but some students were still difficult to reach.89 Recent vapers had increased
exposure to other high-risk students and to students who viewed vaping as minimally harmful
(potential network clustering effects), whereas students who had more peer leader friends were

16
less likely to report recent vaping.89 However, network dynamics surrounding vaping are still not
well understood. In this study, intervention messages that were promoted through the peer leader
networks were not associated with the perception that vaping is harmful. This intervention was
successful in reducing recent e-cigarette use behavior but not in shifting normative perceptions
of the behavior. Other intervention strategies may need to be implemented in addition to the peer
leader approach to successfully alter norms in social networks.
Gaps in the literature
Many of the preliminary studies on e-cigarette network-based interventions have been
limited in sample size or duration (i.e. cross-sectional or two-wave studies). The prevalence and
popularity of e-cigarettes suggests that there are distinct differences in the social dynamics
driving use compared to combustible cigarettes and different social norms surrounding use. In
the current literature, it is unclear what role perceived peer norms play or how altering norms
would aide e-cigarette prevention among adolescents. This knowledge gap highlights the need to
explore the impact of norms messaging as a prevention strategy to reduce the initiation and
spread of e-cigarette use in social networks.
SUMMARY OF GAPS IN THE LITERATURE
The popularity of e-cigarette use among youth highlights the urgent need to understand ecigarette use initiation and diffusion. The current gaps in the literature can be summarized by the
following points:
1. There are few empirical studies comparing the effects of perceived social norms vs.
network exposure on individual behavior.

17
2. E-cigarette use and the social dynamics surrounding e-cigarette initiation in adolescent
social networks has not been longitudinally examined.
3. There is a lack of knowledge about the impact of social norms messaging as an ecigarette use prevention strategy in adolescent networks.
INTRODUCTION TO THE DISSERTATION STUDIES
Electronic cigarettes have experienced a substantial increase in uptake and popularity,
becoming the most commonly used tobacco product among adolescents over the last ten years.
1-4
While there are many psychological, social, and environmental factors that contribute to tobacco
use, there is an ample body of evidence that social networks influence adolescent tobacco use. Ecigarettes have recently gained traction and favor among adolescents and young adults, but no
studies have examined the association between e-cigarette use and adolescent social networks
over time.
One possible mechanism driving e-cigarette use may be perceived social norms and the
desire to conform to these norms. However, these perceived norms are often not aligned with
reality, as people often overestimate risk behaviors among friends, and actual e-cigarette use
among friends may be much lower than adolescents believe. Currently, no studies have explored
the associations between perceived social norms and exposure to friend use on individual ecigarette use. This dissertation uses data from a high school cohort recruited in an NIH-funded
R01 study where participants completed surveys about tobacco/e-cigarette use, social network
(friendship) questions, other behavioral health attitudes and behaviors. Students were followed
throughout high school, resulting in longitudinal data on e-cigarette use and social networks. As
adolescent friendship networks evolved over time, perceived social norms and patterns of ecigarette use changed as well. The three dissertation studies are structured as stand-alone papers

18
that each relate to a specific aim. The main objective of the dissertation is to evaluate the
longitudinal effects of social influence on individual e-cigarette use. Additionally, these studies
aim to differentiate the effects of social influence mechanisms by quantitatively comparing the
network exposure to friend e-cigarette use and perceived social norms surrounding e-cigarettes.
Study 1: Impact of social networks and norms on e-cigarette use among adolescents in
Southern California: a prospective cohort study
Aim 1: Assess the mechanisms of social influence that drive e-cigarette use by evaluating the
differential effects of perceived social norms and friend use on individual e-cigarette use.
H1: Perceived rates of e-cigarette use among friends will be significantly greater than actual
levels of friends’ self-reported e-cigarette use.
H2: Individuals who perceive greater e-cigarette use and pro-e-cigarette norms among friends
will be significantly more likely to use e-cigarettes.
Study 2: Network dynamics of social influence on e-cigarette use among an ethnically
diverse adolescent cohort
Aim 2: Use stochastic actor-oriented models to evaluate the longitudinal relationship between
friend influence and friend selection on e-cigarette use (i.e., how e-cigarette use propagates
throughout networks over time) and clarify the role of perceived peer norms.
H3: Network exposure to peer e-cigarette use behavior and pro-e-cigarette social norms will both
be positively associated with e-cigarette initiation.
H4: E-cigarette use status will be associated with friend selection over time.
Study 3: The impact of social norms on diffusion dynamics: A simulation of e-cigarette use
behavior

19
Aim 3: Use parameters from Aims 1 and 2 in agent-based simulation models to evaluate the
potential impact of targeting perceived norms for e-cigarette use prevention and reduction in
adolescent social networks.
H6: In diffusion simulation models, individual e-cigarette use will decrease when pro-e-cigarette
norms decrease (i.e. when anti-e-cigarette norms increase).
The dissertation studies contribute to the current SNA literature by exploring the interaction
between perceived social norms, exposure to peer behaviors, and individual behaviors. When
exploring behavioral influences, the current literature rarely addresses that perceived norms vary
for adolescents depending on who is in their network (i.e., one person’s idea of normal
behavioral is likely different from someone else’s). The present studies have tested hypotheses
on the discrepancy between perceptions about peer norms, peer e-cigarette use, and friends’ selfreported behavior and their influences on individual behavior. Using complete social network
data allows us to compare the veracity of these perceptions against peer self-reported data in
these networks. The analyses in this dissertation seek to disentangle peer influence effects and
understand their individual and joint impacts on individual e-cigarette use.
Second, this project is part of the first large scale cohort study examining friendship
network data and e-cigarette use among adolescents. Given the sharp increase in popularity of ecigarettes/vaping among youth, there is an urgent need to understand the factors that have
contributed to rapid e-cigarette uptake, whereby they have surpassed all other tobacco products
in popularity. This dissertation critically examines peer influence effects in relation to patterns of
initiation and use over time, implemented through the application of mixed effects logistic
regression and stochastic actor-oriented models. These types of network models have not been
applied to e-cigarette use and friendship networks before.

20
Finally, in designing a behavioral intervention, targeting perceived norms and targeting
friend behavior are two different approaches, therefore understanding the network dynamics of
social norms vs. behavior is critical for designing effective interventions. Study3 focuses on how
social norms impact the effects of peer exposure on e-cigarette use and explore whether peer
norms messaging could be a viable intervention pathway in adolescent networks to reduce e-cig
use. The simulation models test varying levels of e-cigarette social norms under different
network conditions to understand how changes in norms affect e-cigarette diffusion throughout
networks. Study 3 also simulates several network intervention strategies with the aim of
identifying the most advantageous approach.
The dissertation studies take a rigorous methodological approach to examine the
relationships among network and normative influences on e-cigarette use. By clarifying the
effects of perceived peer norms on e-cigarette use, this dissertation seeks to inform future
network-based social norm interventions with the overarching objective of ultimately decreasing
e-cigarette use among adolescents.
Research design and methods
Using a prospective cohort design, the relationship between perceived friend/social
norms and individual e-cigarette use was assessed among a demographically diverse sample of
more than 2,000 high school students in Southern California over time. Details about the analytic
sample, data collection, and data processing are described below.
Data sources and collection
The dissertation studies use data collected from a racially and ethnically diverse
adolescent cohort. The ADVANCE (Assessing Developmental Patterns of Vaping, Alcohol,
Nicotine, and Cannabis Use and Emotional Wellbeing)90 study began data collection in

21
September 2020 with a pilot subset of five schools (class of 2024) and began data collection with
the full eleven school sample in Spring 2021. ADVANCE collected data on substance use,
mental health, social networks, and other health related information throughout the course of
high school. Students were recruited in 9
th grade (ages 13-15) at baseline and followed through
the end of high school (until ages 17-19). At each wave, electronic surveys are administered in
students’ classrooms either virtually or on-site depending on the individual schools’ COVID-19
preventive procedures during the first few waves. During Fall 2020/Spring 2021 enrollment,
4,183 students received parental consent, 3,968 eligible students were invited to participate in the
study, and 2,211 provided assent and completed the assessment in Spring 2021 (baseline for this
dissertation).
Each school is considered one network and as many 9th graders as possible were enrolled
into the cohort and followed throughout high school, with some students joining the study at
subsequent waves. There were eleven schools in the class of 2024 cohort, but one school
declined participation in the social network portion of the survey, therefore the dissertation
studies focus on ten school networks. ADVANCE follow up surveys were administered every 6
months (Fall/Spring semester) yielding 8 waves of data across 4 years of high school.
Dissertation studies 1 and 2 use data from the first three waves of complete data across the ten
schools, Spring 2021, Fall 2021, and Spring 2022.
Table 1 provides demographic and behavioral characteristics of the analytic sample at
baseline (10 schools). The sample is approximately 53.7% female, 69.4% heterosexual, 53%
Hispanic, and racially diverse. Network metrics and average degree scores by wave and school
are displayed in Table 2. The average network size is at baseline was 162 students, in general the
networks are sparse and decentralized. The average total degree at baseline was 6.5 ties, average

22
out-degree scores was 3.25 and approximately 50% of friendship nominations were reciprocated.
E-cigarette use prevalence was low at baseline but gradually increased with each subsequent
wave. At baseline, self-reported past 6-month e-cigarette was 2.0% and lifetime e-cigarette use
was 5.4%.
Data processing
All data cleaning, processing, and analysis were performed in the statistical system R
version 4.3.2.91 Initial processing of each wave followed the workflow in Figure 1. Friendship
networks were assessed by asking participants: Name up to seven (7) of your closest friends in
your grade at school in the spaces below. Enter your friends first and last real name, not their
nickname. School rosters were pre-loaded so that names would populate as students began
typing. Students also had the option to write in names if they did not automatically populate.
Write-in nominations were matched against school rosters using a string-matching algorithm and
nominations that could not be matched against the roster, were made to students outside of
school, or to students who did not consent to participate in the study were removed. Friendship
nominations were used to construct social networks for each school, at each study wave. Social
networks are represented as directed adjacency matrices, where each cell in the matrix represents
a friendship between a given pair of students in the cohort (xij) at that wave (a friendship
nomination from student i (ego) to student j (alter) = 1, and no friendship/tie = 0).
After cleaning nomination data and constructing networks, data sets were output for analysis.
Missing data was be handled according to the analytic plan for each model (please see chapters
2-3 for details).

23
Table 1. Demographic and Behavior Characteristics (N = 2,912 )
N(%)
Sex Assigned at Birth
Male 1,300 (44.6)
Female 1,563 (53.7)
Missing 49 (1.7)
Sexual Orientation^
Heterosexual 2,020 (69.4)
Sexual Minority 425 (14.6)
Questioning 145 (5.0)
Prefer not to disclose 118 (4.1)
Missing 204 (7.0)
Hispanic 1,572 (54.0)
Missing 132 (1.5)
Race
American Indian/Alaska Native 64 (2.2)
Asian 803 (27.6)
Black/African American 55 (1.9)
Native Hawaiian/Pacific Islander 13 (0.4)
White 467 (16.0)
Multi-racial 637 (21.9)
Other 682 (23.0)
Missing 191 (6.6)
Past 6-month E-cig Use Prevalence
Wave 1 58 (2.0)
Wave 2 146 (5.0)
Wave 3 208 (7.1)
Lifetime E-cig Use Prevalence
Wave 1 157 (5.4)
Wave 2 277 (9.5)
Wave 3 357 (12.3)
Mean (SD)
Baseline Age (years) 14.6 (0.6)
Perceived Pro-vaping Norms†
Wave 1 0.95 (0.5)
Wave 2 0.99 (0.5)
^Sexual minority, questioning, prefer not to disclose collapsed for analysis with
heterosexual as reference
† range 0-2

24
Table 2. School Friendship Network Characteristics
Wave 1
School Size Edges
Average
Degree
Average
Out-degree Density Reciprocity Transitivity APL Cohesion
101 169 429 5.08 2.54 0.02 0.38 0.59 6.42 0.16
102 58 80 2.76 1.38 0.02 0.23 0.48 3.91 0.26
103 131 228 3.48 1.74 0.01 0.25 0.53 7.63 0.13
104 222 597 5.38 2.69 0.01 0.29 0.54 7.39 0.14
105 80 134 3.35 1.68 0.02 0.27 0.54 3.84 0.26
106 93 256 5.51 2.75 0.03 0.32 0.50 2.91 0.34
107 109 304 5.58 2.79 0.03 0.24 0.42 2.50 0.40
112 255 1616 12.67 6.34 0.02 0.29 0.57 6.10 0.16
113 279 1666 11.94 5.97 0.02 0.32 0.54 6.38 0.16
114 223 1032 9.26 4.63 0.02 0.23 0.38 6.51 0.15
Mean 162 634.2 6.50 3.25 0.02 0.28 0.51 5.36 0.22
Wave 2
Size Edges
Average
Degree
Average
Out-degree Density Reciprocity Transitivity APL Cohesion
101 212 595 5.61 2.81 0.01 0.25 0.55 6.14 0.16
102 113 169 2.99 1.50 0.01 0.24 0.35 2.84 0.35
103 117 252 4.31 2.15 0.02 0.26 0.56 6.78 0.15
104 282 870 6.17 3.09 0.01 0.28 0.51 6.75 0.15
105 101 240 4.75 2.38 0.02 0.29 0.52 6.51 0.15
106 140 269 3.84 1.92 0.01 0.22 0.41 6.85 0.15
107 190 414 4.36 2.18 0.01 0.16 0.40 6.91 0.14
112 299 1045 6.99 3.49 0.01 0.30 0.53 5.95 0.17
113 331 1144 6.91 3.46 0.01 0.31 0.54 7.08 0.14
114 296 917 6.20 3.10 0.01 0.21 0.48 6.31 0.16
Mean 208 591.5 5.21 2.61 0.01 0.25 0.49 6.21 0.17
Wave 3
Size Edges
Average
Degree
Average
Out-degree Density Reciprocity Transitivity APL Cohesion
101 208 632 6.08 3.04 0.01 0.30 0.56 6.03 0.17
102 126 219 3.48 1.74 0.01 0.34 0.52 3.45 0.29
103 150 300 4.00 2.00 0.01 0.22 0.48 6.34 0.16
104 291 902 6.20 3.10 0.01 0.25 0.46 6.26 0.16
105 110 255 4.64 2.32 0.02 0.25 0.51 7.28 0.14
106 142 281 3.96 1.98 0.01 0.22 0.42 6.63 0.15
107 182 426 4.68 2.34 0.01 0.15 0.45 7.00 0.14
112 276 882 6.39 3.20 0.01 0.27 0.56 6.34 0.16
113 332 1186 7.14 3.57 0.01 0.25 0.54 5.94 0.17
114 292 931 6.38 3.19 0.01 0.24 0.50 5.89 0.17
Mean 211 601.40 5.29 2.65 0.01 0.25 0.50 6.12 0.17

25
Fig. 1. Data processing workflow diagram.

26
CHAPTER 2
The impact of social networks and norms on e-cigarette use among adolescents in
Southern California: A prospective cohort study

27
ABSTRACT
Objective: Using social network analysis, we assessed the mechanisms of social influence that
promote e-cigarette use in adolescent networks.
Methods: Data on health behaviors and friendship networks from a cohort of ten high schools in
Southern California (N = 2,912) were collected in grade 9 Spring 2021 (W1), grade 10 Fall 2021
(W2), and Spring 2022 (W3). Two mixed effects logistic regression models were estimated (full
sample and subsample non-vapers only) to evaluate the associations of W1 and W2 pro-vaping
norms, peer e-cigarette use exposure, and prior e-cigarette use (full sample) on past 6-month
vaping at W3, adjusting for demographic covariates and school clustering.
Results: Previous vaping was the strongest predictor of past 6-month vaping at W3 among the
full sample. Greater exposure to friend e-cigarette use at W2 (AOR=12.2, 95% CI=4.04, 36.5),
and greater pro-vaping norms at W2 (AOR = 2.63, 95% CI=1.24, 5.55) were significantly and
positively associated with increased odds of initiating e-cigarette use at W3 among students with
no lifetime e-cigarette use.
Conclusion: Peer network exposure and pro-vaping norms are significant predictors of vaping
initiation even when network vaping prevalence is low.

28
INTRODUCTION
E-cigarette Use
E-cigarette use (e.g., vaping) has rapidly increased and gained popularity among youth,
and is now the most commonly used nicotine product among adolescents.1
In the United States,
14.1% of high school students and 3.3% of middle school students reported using e-cigarettes in
the past 30 days in 2022.92 Mounting evidence that e-cigarette use is associated with combustible
cigarette use and other negative health outcomes has led to increased concern about the
popularity of e-cigarette use among youth.3,93 Trends suggest that youth have more positive
norms regarding e-cigarettes compared to combustible cigarettes, and report greater peer
approval of e-cigarettes.94 However, reducing adolescent vaping through behavior change
interventions can be challenging due to the complex influences of inter- and intra-personal
factors.
Social networks
Social networks are the relationships or connections between people, organizations or
other entities.8 Social influence occurs through social network interactions which shape
normative behavioral perceptions. For tobacco use, prior social network research has
demonstrated a positive association of having friends who smoke with individual combustible
cigarette use.33-40,42-46,48-50 Adolescence is a developmental period often characterized by
experimentation and the desire to fit in, making adolescents susceptible to social influence from
friends and peers. If an individual perceives vaping as normative in their network, then the desire
to conform with friends creates a unique social pressure, potentially driving e-cigarette initiation
and use.
Social norms

29
Social norms have been extensively explored across the social sciences from a range of
theoretical perspectives. Descriptive norms are the perceptions about the behaviors of individuals
in one’s network (what people do), whereas injunctive norms are perceptions of approval from
one’s social network or opinions on what one should or should not do (what people think).60
Norms are generally disseminated across networks through communication and interaction with
others and are implicit to group membership.58 Therefore, understanding social network
dynamics is essential to understanding the influence of social norms on behavior.
Past studies have demonstrated that perceptions of friends’ behavior compared to friends’
self-reported behavior are often not in alignment.38,63,95 In a nationwide survey of 6th
-12th grade
students in the U.S., 85% of students perceived that most students in their grade typically use
tobacco, while only 21% of students reported having ever used tobacco.96 These misperceptions
surrounding peer norms can lead to the belief that certain behaviors, such as tobacco or ecigarette use, are more common than they actually are, which leads to artificially high
perceptions about network prevalence. Compared to friends’ self-reported behavior, perceived
friend norms have significantly greater influence on individual behavior.38,54,63 This can be
problematic because adolescents may base decisions about e-cigarette use on inaccurate
perceptions about their peers in an effort to fit in.
Prior studies have laid the groundwork for the strength of association between perceived
norms and individual behavior,38,63,64,95 but few studies have examined tobacco use in the context
of norms and social networks. Valente et al. (2012)54 explored the effects of peer influence and
peer norms by examining multiple measures of peer influence on combustible tobacco use in a
predominantly Hispanic adolescent cohort. Perceptions of friend smoking were significantly and
consistently associated with individual smoking. Research has demonstrated that injunctive

30
norms are associated with adult smoking97 and a recent systematic review and meta-analysis on
the longitudinal effects of social norms and smoking found that descriptive norms were
significant predictors of smoking initiation among youth.60 Overall descriptive norms (perceived
smoking behavior) of close friends, parents, siblings, and adults were all significantly associated
with smoking initiation, with descriptive norms for close friend smoking and sibling smoking
having the strongest positive associations with smoking initiation.60 However, agreement
between perceived friend smoking and self-reported friend smoking was low among confirmed
smokers, further demonstrating the importance of using peer self-report data to accurately
measure exposure and to quantify the discrepancy between perceived and actual smoking
prevalence.
A breadth of studies have explored social networks and combustible tobacco use through
a variety of analytic approaches.33-37,39,40,42,43 However, many studies were limited either by
sample size because they focused on a single network69 or duration, often limited to one or two
waves of data,26,54,98 and none of these studies has specifically examined e-cigarette use in this
context. Given the rise in popularity of e-cigarettes, perceptions about prevalence and peer use
may be stronger, but exposure to friends’ vaping behavior may also be higher. Alternatively, provaping norms and social acceptability of vaping may persist even when exposure to vaping
among peers is low. Therefore, it is difficult to discern what social dynamics are more strongly
associated with individual use and what factors predict e-cigarette use initiation.
Another limitation in this area of research is that studies frequently operationalize peer
behavior through measures that only capture the participant’s perception of their friends’
behavior (i.e. asking people to estimate the number of friends that use a substance instead of
surveying those friends), which fails to capture the actual behavior of friends and tends to result

31
in an overestimation of peer use. Using network analysis allows us to compare perceived peer
norms to friends’ self-reported e-cigarette use, providing a better understanding of networkdependent factors that drive e-cigarette use.
This study uses social network analysis to assess the mechanisms of social influence that
drive e-cigarette use by evaluating the differential effects of perceived norms and friend use on
individual e-cigarette use. We will explore whether individuals who perceive greater e-cigarette
use, and pro-vaping norms are more likely to use e-cigarettes. Further, we will examine whether
pro-vaping norms or peer exposure to vaping has a stronger effect on individual vaping behavior.
METHODS
Participant Recruitment
ADVANCE (Assessing Developmental Patterns of Vaping, Alcohol, Nicotine, and
Cannabis Use and Emotional Wellbeing) is a prospective cohort study on the health behaviors of
high school students in the Southern California area.90 Among the ten high schools, three are in
low income, low resource settings, two are in middle to high income settings, and the remaining
five schools are in middle income urban settings. The schools all comprise racially and ethnically
diverse student populations with school size ranging from 1,700 to 2,700 students. Two schools
participate in the California TUPE (Tobacco Use Prevention Education) program, one school
which has the highest e-cigarette use prevalence in our sample. The TUPE program is a state
funded program aimed at reducing youth tobacco use through tobacco-specific, researchvalidated, education and reinforcement activities, including intervention and cessation programs
for students. Schools must undergo a competitive application process to participate. Data
collection began in Fall 2020 with a pilot subset, and then began in Spring 2021 with the full
sample. This analysis uses the first three full waves of data from Spring 2021 to Spring 2022

32
across ten high schools, each high school is considered one social network. Participants (N =
2,912) were recruited in 9th grade (ages 13-15) at baseline. Written parental consent and student
assent were obtained before data collection. STROBE (Strengthening the Reporting of
Observational Studies in Epidemiology) study cohort reporting guidelines were used99. This
study was approved by the University of Southern California Institutional Review Board (#HS19-00682-CR002).
Patient and Public Involvement
The public was not involved in developing, conducting, or reporting the results of this study due
to the sensitive nature of the data collected from minors.
Measures
Demographic covariates
Demographic covariates include sex assigned at birth (male, female), Hispanic ethnicity (nonHispanic, Hispanic), age at baseline, and sexual identity (heterosexual, sexual minority [gay,
lesbian, bisexual, asexual, queer], questioning, or prefer not to disclose). Due to unequal counts
in each category, respondents who identified as non-heterosexual were collapsed into one
category for analyses with heterosexual as the reference category.
Estimated E-cigarette Use among Friends
Participants were asked to estimate how many of their friends use e-cigarettes: How many of
your five closest friends use electronic cigarettes for vaping nicotine (E-cigs, vaporizer, JUUL,
Puff Bar)? (0 – 5; not sure).
Friendship networks
Friendship networks were assessed by asking participants: Name up to seven (7) of your closest
friends in your grade at school in the spaces below. Enter your friends first and last real name,

33
not their nickname. School rosters were pre-loaded so that names would populate as students
began typing. This nomination method has been successfully deployed in several studies,
including school-based adolescent studies and provides valid and reliable network data.8,54,70,100
Friendship nominations were used to construct social networks for each school, at each study
wave. Social networks are represented as directed adjacency matrices, where each cell in the
matrix represents a friendship between a given pair of students in the cohort (xij) at that wave (a
friendship nomination from student i to student j = 1, and no friendship/tie = 0).
Network Exposure
Friends’ e-cigarette use was operationalized as network exposure. Personal network exposure is
the proportion of people in one’s personal network who have adopted a behavior.8 Network
exposure is a measure of influence in one’s social network defined as:
𝑬𝒊 =
∑ 𝑾𝒊𝒋𝒚𝒋
𝑾𝒊+
(1)
Where W is the social network weight matrix and yj is a vector of adoption behavior. To
calculate network exposure, Wij is multiplied by yj to get a count of the number of friends who
have adopted the behavior. This value is then divided by the total number friends named to get a
proportion or percentage. Network exposure was calculated for each individual based on their
outgoing ties and the alter self-report data.
Perceived Pro-vaping Norms
Perceived pro-vaping norms measure an individual’s perception of pro-vaping/e-cigarette
attitudes and beliefs. The following questions were asked to measure perceptions about network
e-cigarette use and attitudes or approval of e-cigarette use: People who are important to me use

34
e-cigarettes; My friends don’t mind when other people use e-cigarettes around them; Ecigarettes are more socially acceptable than smoking cigarettes; A lot of people vape ecigarettes (do not agree = 0, don’t know = 1, agree = 2). A composite perceived pro-vaping
norms score variable was calculated as an average of the four items (see confirmatory factor
analysis section below).
E-cigarette Use
The outcome of this study was individual past 6-month vaping/e-cigarette use defined as: Any
electronic cigarette for vaping nicotine (E-cigs, vaporizer, JUUL, Puff Bar), dichotomized as
(Use in the past 6 months, no use in the past 6 months).
Statistical Analysis
Confirmatory Factor Analysis (Perceived Pro-Vaping Norms)
Confirmatory factor analysis for the four perceived pro-vaping norm indicators was
conducted using the lavaan package for latent variable modeling in R.101 The four ordinal
indicator items were loaded onto a single factor and measurement models were estimated
separately for each respective wave. As such, the four items were averaged to create a composite
measure of perceived pro-vaping norms for each wave (range 0-2), with higher scores indicating
stronger perceived pro-vaping norms. Cronbach’s alpha for the pro-vaping norms factor at each
wave were lower than desired (wave 1  = 0.58; wave 2  = 0.64).
Logistic Regression
Mixed effects logistic regression using complete cases was used to evaluate the
association of wave 1 and 2 perceived pro-vaping norms, peer e-cigarette use exposure, prior
wave e-cigarette past 6-month-use, and demographic covariates with wave 3 past 6-month ecigarette use (in one model), including random effects for school clustering. An initiation model

35
was run on a subsample of individuals with no lifetime vaping behavior as of wave 2 to test the
association of these factors with vaping initiation during the past 6 months at wave 3. An
additional model was also run separately using pooled data from the ten imputed data sets
(Appendix A; see details below). The model specification is as follows:
𝑙𝑜𝑔 (
𝑃𝑟(𝑦𝑡𝑖𝑗=1)
1−𝑃𝑟(𝑦𝑡𝑖𝑗=1)
) = 𝛼 + ∑ 𝛽𝑘𝑋𝑘𝑖𝑗 + 𝑦𝑡−1𝑖𝑗 + 𝜌(𝑝𝑒𝑒𝑟 𝑦𝑡−1𝑖𝑗) + 𝜌(𝑝𝑒𝑒𝑟 𝑦𝑡−2𝑖𝑗) + 𝑈𝑗 + 𝜀
𝑘
𝑘=1
(2)
Where i individuals are nested in j schools over t time periods, and t=3. Where yi is the
dichotomous individual past 6-month e-cigarette use outcome, α the intercept, k parameter
estimates for vectors of k covariates, which includes demographics and pro-vaping norms (Xki),
yt-1ij (prior e-cigarette use behavior), and ρ the parameter estimates for the network exposure to
peer e-cigarette use at waves 1 and 2. Additionally, Uj is the random intercept for each school
and 𝜺 is the residual error term.
Multiple Imputation
In addition to the observed model, an imputed model was run for comparison. Since
many individuals had data that was close to complete (i.e. missing a single covariate or outcome
at one wave), an imputed model was run to maximize use of the collected data and serve as a
robustness check. Out of the full sample (N=2,912), 1,599 had complete data for all three waves
and were included in the full model, 1,431 individuals had no lifetime e-cigarette use and were
included in the initiation model. After independent variable imputation, a sample of 2,154
individuals were retained in the imputed model. Missing data were imputed using multiple
imputation by chained equations with the MICE package in R.102 Multilevel imputation,
controlling for school clustering, was used to impute data for all independent variables for ten

36
imputed data sets (i.e. vaping behavior outcomes were not imputed for any wave) (Supplemental
Table 1 Appendix A). For pro-vaping norms, an average score was imputed instead of imputing
individual items. Network exposure to vaping was imputed as a continuous variable instead of
imputing the vaping behavior of friends. Imputed variable distributions were assessed against
observed distributions to evaluate the integrity of the imputations. All analyses were performed
using R (version 4.2.3).91
RESULTS
Analytic sample demographics are displayed in Table 3. The study sample was majority
female (58.3%), identified as heterosexual (73.7%) with a large proportion of Hispanic students
(44.9%). The average age at baseline was 14.4 years old (SD=0.5). Past 6-month vaping
prevalence was 2.4% at baseline and increased to 6.1% at wave 2, and 7.9% at wave 3. Among
the full sample, the estimated number of friends that currently use e-cigarettes at wave 3 was
0.53 friends on average, while the actual number of friends that used e-cigarettes based on their
self-reports was 0.36 friends on average (results not shown; t = -6.39, p < 0.001). Among ecigarette users at wave 3 (n = 126), the estimated number of friends that use e-cigarettes was 2.4
friends on average, while the actual average based on their self-reports was 0.83 friends using ecigarettes at wave 3 (t = -9.06, p < 0.001).
The factor loadings and measurement model fit for the confirmatory factor analysis are
shown in Table 4. Overall, factor loadings were moderate. The Comparative fit indices (CFI) for
both waves were > 0.95 and the Tucker Lewis Indices (TLI) for both waves were > 0.90. For
both waves, a perceived pro-vaping norms score was calculated as an average of the four items
(composite score range 0-2), with higher scores indicating greater agreement with the statements
and perceived approval of vaping (wave 1 mean = 0.95; wave 2 mean = 0.97).

37
The results for the mixed effects logistic regression for both the full sample and the
initiation models are displayed in Table 5. In the full sample model, previous 6-month vaping at
waves 1 and 2 were both the strongest predictors of past 6-month vaping at wave 3. On average,
students who were older at baseline were more likely to report past 6-month vaping. Greater
network exposure to friends that vape at the prior wave was significantly associated with past 6-
month e-cigarette use in adjusted models (AOR = 2.65; 95% CI: 1.02, 6.89). Perceived provaping norms at wave 2 were also significantly and positively associated with increased odds of
past 6-month vaping in adjusted models (AOR= 1.81; 95% CI: 1.08, 2.83). An indicator for
participation in the TUPE program was tested but its inclusion did not change the model results
so it was omitted.
To further test the strength of these relationships we ran an initiation model, constrained
to individuals who had no lifetime use of e-cigarettes at both waves 1 and 2 (Table 5). We found
that greater network exposure to friend vaping at wave 2 was the strongest predictor of reported
past 6-month vaping initiation at wave 3 (AOR=12.2; 95% CI: 4.04, 36.5). Positive perceived
pro-vaping norms from the previous wave were also significantly associated with vaping
initiation at wave 3 (AOR=2.63; 95% CI: 1.24, 5.55). The imputed model results were consistent
with the observed model results, but the strength of the association was greater for perceived provaping norms and weaker for network exposure (see Appendix A Supplemental Table 1). A
network plot of one school is shown in Figure 2 to illustrate differences in network structure and
vaping prevalence across the three time periods. In this study, trends show increased vaping
prevalence throughout the networks over time.

38
DISCUSSION
Network exposure to peer e-cigarette use and perceived pro-vaping norms among friends
were the strongest factors associated with e-cigarette initiation (among those who had no lifetime
use of e-cigarettes by wave 2); these factors were also associated with any past 6-month ecigarette use in the overall sample, though prior vaping was the strongest factor associated with
past 6-month vaping at wave 3. While students who were older at baseline were significantly
more likely to report past 6-month e-cigarette use, this finding was not observed in the initiation
model. This may indicate that students who are older than their peers are more likely to have
already initiated use before wave 3, which is why they were not eligible to be included in the
initiation model. Social networks not only influence individual behavior through direct contact
with network members but also through attitudes, norms, and ideas that reinforce behaviors. This
study also demonstrates that adolescents tend to significantly overestimate the number of friends
that use e-cigarettes and that this difference is especially pronounced among vapers.
Past research provides evidence that people have difficulty accurately assessing the
behavior of others, even if those individuals are close ties in their network. Discrepancies in
perceived peer norms and actual peer behavior could be attributed to several cognitive biases.
One phenomenon known as the majority illusion has been demonstrated in past social network
research.103 People often do not have enough information to accurately estimate the prevalence
of a behavior and instead tend to estimate the behavior of an entire network based on
observations in their personal network.103 When this occurs, people tend to overestimate
behaviors in the network based on the behavior of high degree or “popular” network members.
Since these popular members are more connected and more visible, this can lead to the belief

39
that certain behaviors, such as substance use, are more common which results in an artificially
high perception about the network prevalence.
Another contributor to inaccurately estimating peer behavior is the false consensus effect,
otherwise known as normative fallacy, in which people believe that others behave the same way
they do. This results in people projecting their own behavior onto others, leading substance users
to overestimate use among their friends. Work by Henry et al. has demonstrated this effect in
substance use estimates among adolescents cohorts.98 Our finding, that on average vapers
estimate more friends use e-cigarettes, is consistent with this viewpoint.
Perceptions of peer behavior is a significant, influential, predictor of individual behavior.
Evidence suggests that peer norm perceptions are often not aligned with reality and that actual ecigarette use among friends may be much lower than people believe. This discrepancy can be
problematic because compared to friends’ self-reported behavior, perceived norms have a
stronger association with individual behavior.38,54,63 Thus, one’s perception of friends’ behavior
can be more powerful and influential than the actual behavior of one’s friends. Adolescents may
be at further increased risk of vaping due to the high visibility and social acceptability of vaping
compared to smoking combustible cigarettes. A recent study by Valente et al. (2023) is one of
the first to show peer network exposure effects on e-cigarette use among youth.100 The findings
of the present study are consistent with the effects of network exposure but our study also
includes the effects of peer norms on behavior, both of which are associated with e-cigarette use
initiation.
Both network exposure and perceptions of peer pro-vaping norms increases the odds of
recent e-cigarette use among adolescents. This study demonstrates two important findings: Even
when prevalence of a behavior in a network is low, exposure effects can have a significant

40
impact on behavior, and second, inaccurate perceptions of other people’s behavior can also have
influential effects. In this study, even though network exposure to peer vaping was low, many
adolescents believe e-cigarette use is common. This illustrates that these perceptions may
continue to persist, even in the absence of peer network exposure.
This study included a sample of adolescents from diverse cultural and socioeconomic
backgrounds in Southern California. Emerging themes surrounding social network exposure to
peer vaping, norms, and individual use that could be applicable to adolescents in different
settings but may vary depending on laws surrounding e-cigarette sales and use. California
prohibits the sale of tobacco products to individuals under the age of 21 and recently enacted a
law prohibiting the sale of flavored tobacco products, including e-cigarettes or flavor
enhancers.104 States or countries with less stringent regulations may observe even greater rates of
use and peer exposure among adolescents.
Future research could focus on developing or using validated measures of perceived
norms on e-cigarette use. Interventions should focus on changing norms and perceptions about ecigarette use among adolescents and dispelling the notion that many of their peers use ecigarettes. Using norms messaging in combination with network intervention strategies could be
another approach to curtailing use. Early interventions using peer leaders have shown promising
results.88The efficacy of a norms based social network intervention should be evaluated in future
studies.
Strengths and Limitations
This study has several strengths and some limitations. Inherent to all longitudinal studies,
there may be some bias in which participants are lost to follow up. Hispanic individuals, and
those that identify as heterosexual were more likely to be lost to follow up between waves 2 and

41
3 (see Appendix A Supplemental Table 2). Some students nominated individuals who were not
students within their school or could not be identified from roster information yielding those
nominations unusable. We could not control for sibling or parent vaping because the survey did
not contain questions asking about family e-cigarette use. The peer norms factor is not based on
a validated scale, but moderate factor loadings indicate construct validity and the Cronbach’s
alpha indicates acceptable reliability. The relatively low prevalence of vaping in the networks
resulted in wide 95% confidence intervals for estimates of network exposure to vaping in all
models.
Strengths of this study include the prospective cohort design, which allows causal
inferences to be made about factors driving vaping initiation. The use of sociometric data allows
us to compare the veracity of perceived peer norms against the prevalence of vaping in the
network which has not been examined in e-cigarette use. Recent studies have demonstrated that
regression analyses on social networks have not been shown to overestimate social influence
compared to other methods, despite the non-independence of observations.105 This study is the
first to our knowledge to use social network analysis to examine both exposure and normative
influences on e-cigarette use among a large and diverse adolescent cohort.
Public Health Implications
Adolescents overestimate the acceptability of e-cigarette use among their peers which has
a significant impact on individual e-cigarette initiation and use. Findings from this study could
strengthen vaping prevention initiatives and be applied to youth interventions in different settings
and countries. Targeting network norms surrounding vaping could lead to changes in attitudes,
increased perceived behavioral control and may result in lower acceptance of vaping among
adolescents. Leveraging social norms and network dynamics to achieve a shift in attitudes could

42
ultimately lead to decreases in vaping behavior even in the presence of network exposure to
vaping.

43
Table 3. Demographic Characteristics of the Analytic Sample (N=1,599)
N(%)
Sex Assigned at Birth
Male 667 (41.7)
Female 932 (58.3)
Sexual Orientation^
Heterosexual 1,178 (73.7)
Sexual Minority 224 (15.3)
Questioning 104 (6.5)
Prefer not to disclose 73 (4.6)
Hispanic Ethnicity 718 (44.9)
Race
American Indian/Alaska Native 19 (1.2)
Asian 569 (35.6)
Black or African American 24 (1.5)
Native Hawaiian/Pacific Islander 6 (0.4)
White 254 (15.9)
Multi-racial 391 (24.5)
Other 324 (20.2)
Missing 12 (0.8)
Past 6-month E-cig Use Prevalence
Wave 1 38 (2.4)
Wave 2 98 (6.1)
Wave 3 126 (7.9)
Mean (SD)
Baseline Age (years) 14.4 (0.5)
Perceived Pro-vaping Norms†
Wave 1 0.95 (0.5)
Wave 2 0.97 (0.5)
Comparative Count of Friends that use E-cigs at
Wave 3 Mean (SD)
All participants (n=1,599)
Respondent Estimated Count 0.53 (1.1)
Friends’ Self-report Count 0.36 (0.7)
Among past 6-month e-cig users (n=126)
Respondent Estimated Count 2.4 (1.7)
Friends’ Self-report Count 0.83 (1.0)
^Sexual minority, questioning, prefer not to disclose collapsed for analysis
with heterosexual as reference
† range 0-2

44
Table 4. Perceived Pro-vaping Norms Wave 1 and Wave 2 Confirmatory Factor Analysis
Standardized
Factor Loadings^
Wave
1
Wave
2
Item
People who are important to me use e-cigarettes 0.57 0.66
My friends don’t mind when other people use e-cigarettes around them 0.48 0.55
E-cigarettes are more socially acceptable than smoking cigarettes 0.64 0.64
A lot of people vape e-cigarettes 0.67 0.73
Cronbach's alpha 0.58 0.64
Fit Indices
CFI 0.98 0.97
TLI 0.95 0.92
RMSEA 0.078 0.12
SRMR 0.040 0.057
^Standardized by the standard deviation of the factor and the item
CFI = Comparative fit index; TLI = Tucker-Lewis index; RMSEA = Root mean square error
of approximation; SRMR = Standardized root mean squared residual.

45
Table 5. Past 6-month Vaping as a Function of Demographic Characteristics, Network
Exposure to Vaping, and Perceived Pro-Vaping Norms
Full Model
(N = 1,599)
Initiation Model
(N = 1,431)
Fixed Effects AOR 95% CI AOR 95% CI
Female 1.39 (0.84, 2.31) 1.08 (0.55, 2.10)
Hispanic 1.35 (0.82, 2.23) 1.58 (0.82, 3.08)
Sexual minority^ 1.49 (0.91, 2.45) 1.63 (0.81, 3.31)
Age 1.57* (1.02, 2.39) 1.59 (0.88, 2.87)
Past 6-month Vaping
Wave 1 6.82*** (2.66, 17.5) - -
Wave 2 11.1*** (6.42, 19.3) - -
Network exposure to vaping
Wave 1 1.36 (0.48, 3.82) 0.74 (0.13, 4.3)
Wave 2 2.65* (1.02, 6.89) 12.2*** (4.04, 36.5)
Perceived Pro-vaping norms
Wave 1 1.61 (0.91, 2.83) 0.89 (0.41, 1.94)
Wave 2 1.81* (1.08, 3.01) 2.63* (1.24, 5.55)
Random Effects Variance Variance
School 0.15 0.25
^Sexual minority, questioning, prefer not to disclose collapsed for analysis with heterosexual as
reference
AOR = adjusted odds ratio for mixed logistic regression; CI = Confidence interval
* p<0.05; ** p<0.01; *** p<0.001

46
Figure 2. School network plots at three time points. Nodes are sized by total degree (number of
nominations sent and received). Circles are females and squares are males; dark nodes indicates
being an e-cigarette user.

47
CHAPTER 3
Network dynamics of social influence on e-cigarette use
among an ethnically diverse adolescent cohort

48
ABSTRACT
The objective of this study was to examine the mechanisms of social influence driving ecigarette use in adolescent social networks and differentiate between the effects of exposure to
friend behavior and social norms on individual use. Surveys on health behaviors and friendship
networks from nine high schools in Southern California (N = 2,245; 48% Hispanic) were
collected at three time points from Spring 2021 of grade 9, Fall 2021, and Spring 2022 of grade
10. Stochastic actor-oriented models for the co-evolution of social networks and behavior
dynamics tested for friendship network social influences on e-cigarette use. Two mechanisms of
social influence were estimated, exposure to friend behavior (e-cigarette use among friends) and
pro-e-cigarette social norms (perceived peer approval and use of e-cigarettes), while controlling
for social selection, individual covariates, and endogenous network effects. Results from the nine
schools were combined in a meta-analysis.
Findings revealed social influence effects through exposure to friend e-cigarette use and
pro-e-cigarette social norms, which both had significant positive influences on individual ecigarette initiation over time. Further, Hispanic/Latine individuals and females were more likely
to initiate e-cigarette use compared to males and non-Hispanic/Latine students. The importance
of these effects should be considered in tobacco prevention initiatives. Designing culturally
tailored interventions that target youth social networks and e-cigarette social norms could be
effective at curtailing adolescent use. Changing perceptions and social acceptability of ecigarettes could be one way to slow or prevent the spread of e-cigarette use in adolescent
networks.

49
INTRODUCTION
E-cigarettes have been the most popular tobacco product among adolescents for the last
ten years.3,4 In 2023, 10.0% of high school adolescents in the U.S. reported current e-cigarette
use in the past thirty days.4 Hispanic/Latine and non-Hispanic multi-racial students had the
highest rates of e-cigarette use compared to any other racial or ethnic group.4 The introduction of
e-cigarettes (e-cigs) has given the tobacco industry a new opportunity to target younger
audiences. Clever marketing and social media campaigns have distanced e-cigarettes from the
social and health stigmas of combustible cigarettes in the minds of many adolescents and
adults.106-109 However, the CDC and Surgeon General have declared that e-cigarettes are not safe
for youth and have issued a call for increased tobacco control strategies to reduce and prevent ecigarette use.3 The popularity of e-cigarettes now echoes the public health battle with
combustible cigarettes among youth in generations past: addictive, popular, and difficult to
regulate.
The popularity, acceptability, and heightened public salience of e-cigarette use leads to
distinct social dynamics that can accelerate e-cigarette uptake, particularly among adolescents.
Generally, when people are exposed to e-cigarettes in both real-life social networks and on social
media, they are more likely to adopt this behavior as a result of social influence.109 Adolescents
are in a development stage that is often characterized by the desire to fit in with friends and be
socially accepted by peers,110 which heightens their awareness of social norms and susceptibility
to peer influence. Since adolescent e-cigarette use is likely to be socially influenced, social
network analysis is a valuable tool to understand that influence and yield insights about the
social spread of e-cigarette use and potential intervention strategies to curtail use.

50
Social network framework
Social network analysis (SNA) is a defined theoretical perspective and set of methods
used to analyze and understand relationships between people and how they affect individual
beliefs and behaviors, and broader group outcomes.8 Social relationships and interactions expose
people to others’ behaviors and the exchange of information, opinions, role modeling, and
normative influences. SNA methods provide useful tools to study social influence in the context
of adolescent behavior, and to model network and behavior dynamics jointly and thus parse out
processes of social selection and social influence.67 Although many adolescents’ social networks
are comprised of relations to family, friends, and other contacts, SNA research has shown that
their dynamic relationships with same-age peers are a social context with a powerful influence
on their health and risk-taking behaviors.111
In studies of adolescent social networks and smoking, there is strong evidence that peer
smoking is positively associated with individual smoking.26,33,39,42,43,52-55,67-72,75,77,78 Longitudinal
studies of adolescent smoking and social networks, that model mechanisms driving these
relationships over time, have further established that social selection and influence are important.
There is consistent evidence of peer selection effects, where individuals are likely to form
friendships with others who have similar smoking behaviors.26,40,42,68,69,75 However, adolescents
are unlikely to begin smoking without being exposed to cigarette use among friends first. In
adolescent networks, the likelihood of being a current smoker increases when smoking is
prevalent in personal networks or among one’s best friends.26,52,76 Together, these findings
demonstrate the importance of peer influence and network exposure on adolescent smoking even
while accounting for endogenous factors that predict network change. While the relationship
between peer influence and smoking is established, it is not known if these same social

51
phenomena and mechanisms drive e-cigarette use among youth. The greater prevalence and
popularity of e-cigarettes compared to combustible cigarettes suggests that there are distinct
differences in their social acceptability and social dynamics driving use. The social visibility of
e-cigarette use dispels notions of harm and normalizes the behavior. Given these differences and
the popularity of e-cigarettes among adolescents over the last decade, it is important to
understand the mechanisms of peer influence driving uptake in order to effectively address it.
Peer influence and perceived norms
Understanding the mechanisms through which peer influence occurs is critical to
designing effective interventions that target the appropriate mechanism. Two key mechanisms
that play a role in peer influence among youth are social influence through direct behavior
exposure and social influence through behavioral norms. For e-cigarette use, network exposure
can be quantified as the proportion of friends in one’s social network that self-report using ecigarettes. For example, if 4 of 5 friend in one’s social network use e-cigarettes, then their
network exposure would be 0.80.
Social norms are people’s beliefs about what is typical. They can be defined as
descriptive norms, perceptions about others’ behavior, or injunctive norms, perceptions of
approval or acceptable behavior.58,59 For e-cigarette use, the role of perceived norms, such as
one’s perception of e-cigarette use prevalence among their peers, and perceptions of whether or
not peers approve of e-cigarettes, could be mechanisms of social influence. The positive
association between perceptions of peer smoking and individual smoking has been strongly
demonstrated among Hispanic/Latine adolescents.54 However, adolescents may inaccurately
overestimate how prevalent e-cigarette use is among friends due to the increased visibility of the
behavior, as modeled by their friends, popular peers, or the media, and thus have inaccurate

52
perceived norms. Social networks are instrumental in informing and communicating social
norms, and can be leveraged to correct incorrect views about behavior prevalence and
acceptability.58 One objective of this study is to differentiate between these mechanisms of social
influence—direct behavior exposure and social influence through behavioral norms-- to
determine if either or both mechanism influences e-cigarette use in adolescent social networks.
The Current Study
This study addresses a gap in our current understanding of how adolescent social
networks are linked to e-cigarette use. We focus on a cohort of 2,245 youth across nine high
schools from different socio-economic settings in Southern California. Our study population is
48% Hispanic/Latine, the ethnic group with the highest rate of e-cigarette use among
adolescents.4 The study models peer social networks and e-cigarette behavior dynamics in a
diverse adolescent cohort, to test for social influence on e-cigarette use. Because no studies, to
our knowledge, have explored specific mechanisms of social influence on e-cigarette use, a key
aim is to determine if youth e-cigarette use is influenced by direct network exposure, perceived
pro-e-cigarette norms, or both.
MATERIALS AND METHODS
Participants and Recruitment
ADVANCE (Assessing Developmental Patterns of Vaping, Alcohol, Nicotine, and
Cannabis Use and Emotional Wellbeing) is a prospective cohort study on the health behaviors of
high school students in the Southern California area.90 Data collection began in August 2020
with a pilot subset, and then continued in January 2021 with the full sample. Students were
surveyed each Fall and Spring semester. This analysis uses the first three full waves of data from
January 2021, August 2021, and January 2022 across nine high schools, with each high school

53
considered one social network. At wave 1 (Spring/January 2021), across eleven schools 3,968
eligible students were invited to participate. Of these, 2,211 students provided parental consent,
student assent, and completed the survey. One school did not participate in the SNA portion of
the survey, and one school’s model ultimately did not converge so these schools were removed
from the final analysis. The analytic sample was restricted to individuals who participated in at
least two waves of data collection without missing data on the dependent behavior variable (ecigarette use) (N = 1,510 at Spring 2021). Additional participants joined the study at subsequent
waves resulting in an analytic sample of 2,245 students across nine schools/networks and three
waves of data collection. Among the nine high schools, three are in low income, low resource
settings, two are in middle to high income settings, and the remaining four schools are in middle
income urban settings. The schools all comprise racially and ethnically diverse student
populations. Participants were recruited in 9th grade (ages 13-15) at baseline. Study approval was
granted by the University of Southern California IRB.
Measures
The outcome was ‘ever used e-cigarettes’, defined as a dichotomous behavioral outcome
(“never used” = 0 vs. “ever used” = 1). Perceived pro-e-cigarette norms were measured as
using items that reflected descriptive norms (perceptions about network e-cigarette use) and
injunctive norms (peer approval of e-cigarette use) within one’s social network. The following
questions were asked: “People who are important to me use e-cigarettes; My friends don’t mind
when other people use e-cigarettes around them; E-cigarettes are more socially acceptable than
smoking cigarettes; A lot of people vape e-cigarettes” (Response options 0 = disagree; 1= Don’t
know; 2 = Agree). Confirmatory factor analysis was used to assess construct validity (see
Chapter 2 Table 4 for details) and an average summary score of perceived pro-e-cigarette norms

54
was calculated, with higher scores indicating greater perceived pro-e-cigarette norms. We
controlled for demographic covariates sex at birth (female = 0, male= 1) and ethnicity (nonHispanic = 0, Hispanic = 1).
Analysis
Stochastic Actor Oriented Models (SAOM) are agent-based models for the co-evolution
of social networks and behavior change.66,112 SAOM model the changes in a social network from
the perspective of the individuals (i.e., ‘actors’). The model is a continuous-time Markov chain,
where changes occur through a series of unobserved mini-steps, in which actors have the option
of forming or dissolving friendships or changing their behavior. These longitudinal models
estimate the effect of factors that might influence these decisions, such as the network dynamics
and the characteristics of the actors, which are specified as parameters in the model.113 SAOM
have been used to examine the co-evolution of network changes and health behaviors or
outcomes such as physical activity,31 tobacco use,69,77 marijuana use,28 and sexually transmitted
infections.114
SAOM have two components or processes: (i) network evolution, where tie
changes/friendship changes among actors is the dependent variable, and (ii) behavior evolution,
where behavior changes among actors is the dependent variable. The co-evolution of the network
and behavior is estimated using method of moments implemented by computer simulation of the
change processes.66 Factors that influence changes in the friendship network are called selection
effects, while factors that influence changes in behavior are called influence effects. Since the
co-evolution of the network and behavior are measured by distinct processes, SAOM are a useful
analytic technique for differentiating effects of social selection from social influence, which may
both give rise to correlations between a behavior and social connections.26,39,42,43,67-70

55
SAOM were estimated for each of the nine high school networks separately using the RSiena
package.113. Modeling the e-cigarette behavior variable as initiation/ever-use allows us to
estimate the rate at which e-cig use spreads throughout the network and parameters that impact
behavior diffusion.115 Details of SAOM estimation and the RSiena package are described
elsewhere.66,67 We used these models to test effects of social influence on adolescent e-cigarette
use over time, controlling for social selection and other confounding effects. Social influence
was operationalized by two effects: the effect of friend e-cigarette use (direct peer exposure), and
perceived pro-e-cigarette norms, in predicting adolescent e-cigarette use. The results will provide
insight into how e-cigarette initiation and use spreads through adolescent peer networks over
time.
In these models, the network evolution part estimated selection effects that predict the
formation, maintenance, or dissolution of friendships. Selection effects were chosen based on
model requirements and previous research. We included three types of effects that represent how
individual attributes affect selection: (i) ego effects are the tendency for individuals with a given
characteristic to send friend nominations, (ii) alter effects are the tendency for individuals with a
given characteristic to receive friend nominations, and (iii) same (or similarity) effects are the
tendency for individuals to form friendships with others who are the same/similar to themselves
on a given characteristic. For example, using sex as an attribute, a positive female ego effect
would indicate that females are more likely to send friendship nominations, a positive female
alter effect would indicate that females are more likely to receive friendship nominations, and a
positive female same effect would indicate that individuals are more likely to form friendship
with others of the same sex. These three effects, or terms, represent network evolution processes
based on individual attributes and we included them for sex and ethnicity. Ego, alter, and same

56
effects for e-cigarette use were also initially included in order to estimate changes in the
friendship network based on the behavior of interest. As we detail in the results these e-cigarette
selection terms could not be estimated. The model also included effects that represent
endogenous network processes known to contribute to the formation of friendships, such as
degree related effects (friendship ties made and received), reciprocity, and the GWESP effect
which measures the tendency for individuals to form ties with friends of friends. The model
assumes that individuals have the opportunity to form or dissolve friendships at each mini-step in
the MCMC simulation, modeled by the rate of these changes for each period between observed
data points.
The behavior evolution part of the model includes effects that predict changes in ecigarette use over time based on individual and network characteristics. Our primary aim is to
test for social influence and determine if it is driven by direct exposure to peer e-cigarette use
among networks and/or perceived pro-e-cigarette norms. Direct exposure to peer e-cig use is
tested by the average exposure effect on the rate of e-cigarette use behavior. Perceived pro-e-cig
norms are tested by the pro-e-cig norms effect on the rate of e-cigarette use behavior. The model
also controls for other individual factors that could be predictive of behavior change, such as sex
and ethnicity.
A preliminary model was fit for two schools using forward model selection with scoretests (t-ratios, estimate divided by standard error) to test the significance of parameters to predict
co-evolution of friendship networks and e-cigarette use initiation. The same model was then fit
to all nine schools. To obtain a parsimonious model that could be fit across all networks we
chose effects that were theoretically relevant to our research question and/or important to control
for based on prior research. If a parameter could not be estimated for a specific network, the

57
effect was set to zero for that network to retain the network in the meta-analysis which requires
the same model specification across networks. Overall convergence ratios for each of the nine
networks were < 0.25 with convergence t-ratios for all parameters < 0.10.
Siena Meta-Analysis
A two-step meta-analytic approach was used to combine results from the nine
networks.113,116 Each high school grade network is considered an independent group sampled
from a population (of dynamic networks) and results are combined for each parameter. The
dependent variable is the population parameter estimate which is the result of combining the
parameter estimates from each individual network. Significance of the combined parameter
estimates are based on the IWLS modification of Snijders & Baerveldt (2003).117 All analyses
were implemented using RSiena (Simulation Investigation for Empirical Network Analysis)113
for the statistical system R version 4.3.91
RESULTS
Descriptive statistics for demographic and network characteristics of the analytic sample
are presented in Tables 6 and 7. Almost half of participants were Hispanic/Latine ethnicity which
is typical for Southern California. At baseline, the average network size was 167.8 students, with
a 5.4% prevalence of e-cigarette use (See Appendix A Supplemental Table 3 for separate school
network metrics). Regressions were run to determine if any characteristics were associated with
being lost to follow up (not shown). No covariates were associated with participants lost to
follow up from wave 1 to wave 2, however, Hispanic/Latine students were more likely to be lost
to follow up from wave 2 to wave 3. Figure 3 shows one of the high school friendship networks
at each time point with nodes colored to show e-cigarette use status. The results of the SAOM

58
meta-analysis are described below, focusing on predictors of friendship network dynamics first,
and then predictors of e-cigarette use, including the two social influence effects of interest.
Friendship Network Dynamics
The meta-analysis findings for changes in friendship networks and e-cigarette use are
shown in Table 8. Additional meta-analysis details can be found in Supplemental Table 4 in
Appendix A. If an effect was unable to be estimated, or if the standard error exceeded an
acceptable value, that network was dropped from the meta-analysis for that effect. Across
networks, friendship ties were significantly more likely to be reciprocated, transitive, and with
individuals of the same sex and ethnicity.
The reciprocity effect was significant and positive, indicating that on average students
tend to reciprocate friendship ties. Network transitivity was measured by the significant GWESP
effect, where there was a tendency to form ties with friends of friends. The positive in-degree
popularity effect suggests that individuals with more friendship nominations tend to continue
receiving more friendships nominations over time, increasing in popularity. The negative outdegree popularity effects can be interpreted as individuals who send more friendship nominations
receiving fewer nominations over time. The negative in-degree activity effect also suggests that
more popular individuals tend to send fewer nominations over time. Together, these can be
interpreted as more popular students receiving, but not sending, more friendship nominations.
Additionally, there was evidence of demographic homophily, where individuals were more likely
to form friendships with others of the same ethnicity and sex. E-cigarette use selection effects
were unable to be estimated and resulted in non-convergence. Please see Appendix A
Supplemental Table 5 for regression analyses examining friendship selection based on ecigarette use.

59
E-Cigarette Use Behavior Dynamics
On average, the rate of e-cigarette use initiation had a positive, but non-significant, trend
over each time period. Both of the hypothesized effects of social influence were supported. The
average exposure effect of peer e-cigarette use on individual e-cigarette use was able to be
estimated in six networks and was positive and significant in the meta-analysis (𝜇̂𝜃= 3.15, p <
0.01). Additionally, stronger pro-e-cigarette norms were significantly associated with an
increased rate of e-cigarette use (𝜇̂𝜃= 0.29, p < 0.01). Thus, both mechanisms had independent
and significant effects on adolescent e-cigarette use. In addition to social influence effects,
Hispanic individuals had a significantly greater rate of e-cigarette use initiation compared to nonHispanic students, while males had significantly lower e-cigarette use rates compared to females.
DISCUSSION
Peer influence plays a significant role in adolescent e-cigarette use. We explored peer
influence effects through network exposure to peer e-cigarette use and perceived e-cigarette
norms on individual e-cigarette use over time. To our knowledge this is the first study of its kind
to explore these associations longitudinally among a large, ethnically diverse, adolescent cohort.
Building on past social network and tobacco research,69
we used rigorous methodology to
examine social influence factors associated with e-cigarette use while controlling for individual
attributes and endogenous network effects. We found that peer influence operates through both
network exposure to friends who vape and through perceived positive norms surrounding ecigarette use.
These results provide evidence that peer influence is a persistent effect on adolescent ecigarette use across different school contexts, and that having a higher proportion of friends who
use e-cigs, and further, having pro-e-cig norms, increased the likelihood of e-cigarette use for

60
these adolescents. The importance of these two effects could vary across contexts. For example,
in schools with lower e-cigarette prevalence, students will have less opportunity for direct peer
exposure, but overall positive attitudes towards e-cigs may exist nonetheless and encourage
experimentation. The e-cigarette use prevalence in the schools in our study is relatively low at
baseline (approximately 5%), which demonstrates that peer influence can still operate in a low
prevalence setting, and that the role of perceived pro-e-cigarette norms may become even more
impactful in these networks.
While social network dynamics are nuanced and context dependent, these findings show
that addressing perceived norms and attitudes surrounding e-cigarette use could be a potential
intervention strategy. While having friends that use e-cigarettes in one’s network has a
significant effect on personal e-cigarette use, it can be difficult to create behavior change through
social network alteration. A norms messaging intervention aimed at reducing social acceptability
of e-cigarettes is potentially a more feasible and broadly applicable approach. In addressing
social network norms, it will also become important to understand where pro-e-cigarette norms
are originating, whether this is from social media, substance use among popular peers, or settings
outside of school. Our study found that Hispanic/Latine students were at greater risk of initiating
use, indicating that culturally tailored interventions and messaging strategies should also be
considered.
Strengths
This study has several strengths and some limitations. Estimating SAOM in RSiena
allows us to model the changes in friendship networks and e-cigarette use behavior while
controlling for non-independence of observations. Prior research has used these models with
regards to cigarette smoking, but none to our knowledge have examined e-cigarette use in a

61
large, ethnically diverse, adolescent cohort. Additionally, Hispanic/Latine adolescents have the
highest rate of e-cigarette use nationally, which highlights the need for increased understanding
of e-cigarette use among this population. Another strength of our study is the meta-analysis,
which allows us to make inferences about the underlying population that our networks are
sampled from, giving our findings more generalizability compared to studies focusing on a
singular network.
Limitations
Inherent to longitudinal survey-based research, limitations of this study include missing
data due to attrition or participants being lost to follow up. Given the unique dynamics of
networks, there was difficulty with fitting some of the effects across all networks. The low
prevalence of e-cigarette use resulted in the average exposure effect being inestimable in three of
the networks. Additionally, the low prevalence of e-cigarette use did not permit us to estimate
ego, alter, and similarity effects for e-cigarette use on friend selection, therefore we cannot draw
any conclusions on individuals selecting friends based on e-cigarette use status.
Conclusions
Our study is the first of its kind to explore friendship and e-cigarette use dynamics in a
large, diverse, adolescent cohort. We built on past research by parsing social influence into peer
exposure and normative effects and found that both factors exert influence on individual ecigarette use behavior. Targeting perceived pro-e-cigarette norms may be an important strategy
for creating behavior change, especially in networks with a low prevalence of e-cigarette use.
Norms messaging interventions aimed at reducing positive attitudes and social acceptability of ecigarettes may be instrumental in reducing the initiation and spread of e-cigarette use in

62
adolescent networks. Social factors have a strong influence on adolescent health behavior,
particularly risk behavior, and thus effective public health strategies require social responses.

63
Table 6. Demographic and Behavior Characteristics (N = 2,245 )
N (%)
Sex Assigned at Birth
Male 1,006 (44.8)
Female 1,225 (54.6)
Missing 14 (0.6)
Ethnicity
Hispanic 1,079 (48.1)
Missing 2 (0.1)
Race
American Indian/Alaska Native 46 (2.0)
Asian 734 (32.7)
Black or African American 44 (2.0)
Native Hawaiian/Pacific Islander 12 (0.5)
White 361 (16.0)
Multi-racial 511 (22.8)
Other 467 (20.8)
Missing 70 (3.1)
E-cigarette User
Wave 1 122 (5.4)
Wave 2 228 (10.2)
Wave 3 287 (12.8)
Mean (SD)
Age (years)
Wave 1 14.64 (0.6)
Wave 2 15.51 (0.4)
Wave 3 15.96 (0.4)
Pro-vaping norms^
Wave 1 3.81 (1.8)
Wave 2 3.91 (2.0)
Wave 3 -
^ range 0-8
Note: vaping norms not measured at wave 3

64
Table 7. Social Network Characteristics (N = 9 networks)
Average Network Metrics Wave 1 Wave 2 Wave 3
Size (number of individuals) 167.8 (81.1) 210.1 (93.5) 214.1 (84.9)
Edges (friendship ties) 670.9 (620.8) 611.2 (389.3) 620.9 (366.9)
Reciprocity (proportion) 0.52 (0.06) 0.50 (0.07) 0.51 (0.05)
Density 0.02 (0.01) 0.01 (0.004) 0.01 (0.003)
Transitivity 0.29 (0.05) 0.26 (0.04) 0.26 (0.04)
Average Path Length 5.68 (1.7) 6.14 (1.3) 6.02 (1.05)

65
Table 8. Meta-analysis of changes in friendship networks and e-cigarette use in nine high
schools (N=2,245)
Effect N Estimate (s.e.)
Network Dynamics
Friend rate (period 1) 7 16.71 (1.39)***
Friend rate (period 2) 9 6.7 (0.66)***
Out-degree (density) 9 -3.41 (0.13)***
Reciprocity 9 2.9 (0.05)***
GWESP 9 1.73 (0.04)***
Popularity (in-degree square root) 9 0.19 (0.04)**
Popularity (out-degree square root) 9 -0.35 (0.04)***
Activity (in-degree square root) 9 -0.13 (0.08)
Male
Alter 9 0.05 (0.02)*
Ego 9 0.06 (0.03)
Same 9 0.5 (0.03)***
Hispanic
Alter 9 0.03 (0.04)
Ego 9 0.07 (0.04)
Same 9 0.35 (0.06)***
Behavior Dynamics
Rate (period 1) 9 0.03 (0.01)**
Rate (period 2) 9 0.02 (0.004)**
Average network exposure 6 3.15 (0.50)**
Male 8 -0.66 (0.17)**
Hispanic 8 0.76 (0.24)*
Pro-vaping norms 9 0.29 (0.07)**
*p < 0.05; **p < 0.01; ***p < 0.001
N = number of schools on which the statistic for this effect are based;
Estimate (𝜇̂𝜃) = estimated average effect size; standard error (s.e.)
Note: E-cigarette ego, alter, and same effects were initially estimated but omitted from final models due to
non-convergence

66
Figure 3. A high school network at three time points: Spring 2021 (left panel), Fall 2021 (middle
panel), and Spring 2022 (right panel). Red nodes indicate e-cigarette users. Circular nodes are
female, square nodes are male. Nodes are sized by in-degree (number of friendship nominations
received).

67
CHAPTER 4
The impact of social norms on diffusion dynamics:
A simulation of e-cigarette use behavior

68
ABSTRACT
Background: Diffusion of innovations theory was developed to explain the spread of ideas,
practices, and behaviors between and within communities or networks. These principles can also
be used to understand how to prevent or slow the spread of a harmful behavior, such as substance
use in social networks. This study explores how different network intervention strategies could
impact diffusion dynamics through a series of network simulations based on observed social
norms and e-cigarette use data.
Methods: The baseline simulated network conditions were informed by baseline data collected
from the ADVANCE study. Simulations contained 300 nodes (average observed network size)
and eight time periods to mirror the ADVANCE study. Simulated networks varied by seed
condition (initial adopters/users), intervention condition, and norms distributions. Networks were
seeded with 5% e-cigarette use at baseline in three seed conditions: central nodes (high indegree), community based (one component of the network selected), and random seeding.
Social norms were assumed to be normally distributed about three different mean conditions of
pro-e-cigarette, anti-e-cigarette, or neutral norms. To test intervention strategies we assigned
greater pro- and anti-tobacco norms to 15% of the network based on four conditions: opinion
leadership (high in-degree nodes), betweenness (high betweenness centrality nodes),
segmentation (high in-degree nodes selected from each community), and random selection. This
creates twelve different seed x norm intervention conditions, each simulated 100 times.
Multivariate generalized linear models with a logit-link and robust standard errors were
estimated for e-cigarette diffusion prevalence and rate on network metrics, seed condition,
intervention strategy, and norms condition.

69
Results: Anti-e-cigarette norms significantly decreased average prevalence and rate across all
seed and intervention conditions (AOR = 0.66, p <0.001; AOR = 0.75, p <0.001). An interaction
between anti-e-cigarette norms and intervention strategy showed that assigning anti-tobacco
norms to high betweenness centrality nodes significantly decreased average prevalence and rate
(AOR = 0.79, p < 0.01; AOR = 0.83, p <0.001).
Conclusions: To buffer against the spread of harmful behaviors it is critical to understand how
personal and network factors affect diffusion dynamics. The results of this study show that
achieving a change in norms for 15% of a network can have substantial impact on e-cigarette use
prevalence. Targeting social norms through network-based interventions is one avenue for
slowing the spread of harmful behaviors.

70
INTRODUCTION
Social networks have a profound influence on health decisions, behaviors, and outcomes.
Our relationships and social interactions can lead to the exchange of ideas and information from
person to person. These interactions can influence attitudes, beliefs, and perceptions about what
is normative in one’s social network. In turn, individuals might change their behavior due to
social influence from others in the network, perceptions about social norms, or both.
While social networks can positively impact health behaviors, harmful behaviors are also
socially transmissible. For example, adolescent e-cigarette use is a prominent, sociallyinfluenced public health issue.4 Adolescents are at a developmental stage where they are acutely
aware of social norms and the opinions of friends, leaving them susceptible to social influence.
Peer influence can operate through social networks and can substantially impact adolescent
health behaviors such as substance use.26,75,86,100,118 Since social norms are communicated
through networks, understanding the relationship between social networks and social norms
presents an opportunity to use social norms to promote better health decisions. Social network
analysis (SNA) is a defined theoretical perspective and set of methods used to understand these
social relationships and provides opportunities to intervene on harmful behaviors.8 Social
networks can be leveraged in interventions to enhance social influence, communicate health
information, and promote positive behavior change. Network intervention strategies are rooted in
the diffusion of innovations theory, which is critical to understanding social influence processes
and the spread of behaviors.119
Theoretical foundations
Diffusion of innovations theory

71
Diffusion of innovations theory was developed to explain the spread of ideas, practices,
and behaviors between and within communities or networks.80-82,120,121 There are both individual
and network factors that influence diffusion. Diffusion may occur at a faster rate in some
network structures, such as small world networks, and network characteristics such as clustering
may increase diffusion within groups but slow diffusion between groups.122 Additionally,
individual characteristics affect behavior adoption. People may be more susceptible to social
influence from others who are similar to themselves or may be predisposed to adopting a new
behavior early in the diffusion process. Network position also plays a role, diffusion may occur
more rapidly if central or popular individuals have adopted the behavior because these
individuals have greater social influence.123
While classic diffusion models are based on network exposure to a behavior, or contagion
via cohesion, network exposure alone cannot always predict behavior adoption in real world
settings. The individual factor that most directly impacts behavior adoption is threshold, or the
level of network exposure a person must have before adopting the target behavior.80,81
Thresholds vary between individuals; some people adopt a behavior with low rates of exposure
and would be considered early adopters relative to others in the network, while others do not
adopt a behavior even with a high rate of network exposure and would be considered late
adopters.124 Additional factors that affect behavior diffusion include the position of the initial
adopters in the network, network structure, the threshold distribution, and the influence
mechanism.81 These parameters independently and jointly affect diffusion and also operate
differently under different network conditions, such as the initial prevalence of the target
behavior.

72
Previous work by Valente and Vega Yon (2020)125 tested simulation models that varied
by network structure, seed condition, threshold distribution and influence mechanism. After
examining diffusion rates and prevalence under various conditions there was evidence of faster
diffusion in networks with random and small world structures, high threshold variability, and
under conditions where cohesion was the influence mechanism. Across all networks, the rate of
diffusion and behavior prevalence were lowest when initial users were marginal nodes and
fastest when initial adopters were central (i.e. nodes with high degree or popularity).
Simulating network interventions
Previous studies have used agent-based models to simulate the spread of health behaviors
such as alcohol use, smoking, and physical activity.79,126,127 The principles of diffusion of
innovations theory can also be used to understand how to prevent or slow the spread of a
behavior.80 Since individuals are influenced by others who are proximal in social networks,
investigating the dynamics between social norms, network exposure, and health behaviors is
critical to network intervention design. Network simulation models can clarify the mechanisms
driving diffusion in different network settings, allowing us to test how changes in network
parameters or characteristics would theoretically impact behavior outcomes. This information
can be used to optimize network-based interventions and achieve behavior change.
Diffusion processes are typically measured as a function of threshold and network
exposure, but in real world settings decisions about behavior are often more complex. The effects
of personal attributes or social processes may significantly impact individual behavior adoption
and overall network diffusion trajectories. For example, social norms have demonstrated
influence on individual health behaviors ranging from contraceptive use to substance use,
including tobacco and alcohol.64,65,96,128,129 While network exposure to peer e-cigarette use still

73
drives diffusion, social norms may exacerbate or attenuate exposure effects by raising or
lowering one’s threshold for behavior adoption. Individuals who are surrounded by others with
anti-tobacco norms may feel less tempted to use e-cigarettes, even if they are exposed to the
behavior. By contrast, individuals who are exposed to pro-tobacco social norms among friends
might be more primed for experimentation if given the opportunity. Therefore, when examining
the spread of e-cigarette use in social networks both personal and network factors should be
considered. If social norms, individual thresholds, and behavior exposures shape greater network
diffusion trajectories, then modeling diffusion processes with the inclusion of personal factors
could reveal opportunities for interventions. Additionally, in designing a network intervention,
consideration must be given to what parameters can be altered. Changing normative perceptions
or attitudes surrounding substance use could be more feasible than altering the structure of a
social network, and targeting social norms may be one potential avenue for interventions.
The primary aim of this study is to determine how changes in social norms could be used
to prevent, slow, or halt e-cigarette use in adolescent networks through a series of network
diffusion simulation models. We expect that altering e-cigarette social norms for a portion of the
network (15% of nodes) will significantly impact diffusion dynamics. Another objective is to
compare how these changes in e-cigarette norms interact with different network intervention
strategies to determine the optimal intervention approach and inform future tobacco prevention
and other health promotion initiatives.
METHODS
Building on past research of simulation models to explore diffusion dynamics, the models
in this study explore how e-cigarette use changes or evolves under different network conditions
and with various intervention strategies using parameters based on observed data. In the current

74
study, the baseline simulated network conditions were informed by baseline data collected from
the ADVANCE study, a prospective cohort study of health behaviors and friendship networks of
adolescents in Southern California.90
In the ADVANCE study the average observed network
size was 297 nodes, with an average baseline e-cigarette use prevalence of approximately 5%.
Participants are surveyed biannually (every Fall and Spring semester) resulting in eight waves of
data. The simulated networks in this study are comprised of 300 nodes, with 15 e-cigarette users
at time 0 (5% prevalence), an average degree of 6, and a small-world network structure with
diffusion processes simulated across eight time points.
Measures
E-cigarette use prevalence
E-cigarette use prevalence is an outcome measure of behavior saturation, defined as the final
proportion of individuals in the network who have adopted e-cigarette use at the end of the
simulation (range 0-1).
E-cigarette use rate
Diffusion rate is an outcome measure of the rate of behavior adoption, or the average of the
proportion of adopters at each time period during the diffusion (range 0-1).
Seed condition (initial e-cigarette adopters)
1. Five percent of the nodes in each simulated network are designated as seeds, or nodes that act
as the initial users in the network. The seed condition can impact diffusion outcomes, with
centrally seeded networks often resulting in faster diffusion.130,131 Initial adopters were tested
under three different conditions:
Central seeds (high indegree): defined as the nodes with the highest number of incoming ties,
a measure of network centrality. Nodes with high indegree are often considered opinion

75
leaders in networks and may have more social influence.
Community/cluster seeds: A community detection algorithm based on random walks132 was
used to identify the number of communities or clusters in each simulated network. One
community was chosen at random as the initial seeds to mimic e-cigarette use originating
from one cluster or group of friends in the network.
2. Random seeds: Nodes selected completely at random without regard to degree or position.
Intervention strategies
There are multiple intervention strategies that can be tailored to network conditions and
the target behavior.119
One of the most common network intervention approaches is to
purposefully select individuals in a network to act as change agents.88,123 Often, these individuals
are selected because they occupy influential network positions or based on individual attributes.
For all simulations, intervention nodes were distinct from the seed nodes, such that no initial ecigarette users were selected as intervention nodes. Four network intervention conditions were
tested (Figure 4).
1. Opinion-leaders (high in-degree nodes): Having a high in-degree, or number of connections,
in a network is a common metric of popularity.133 These highly connected individuals are
considered opinion leaders because their status may make them more influential. The top
15% of nodes with the highest in-degree were selected in this condition.
2. Betweenness (high betweenness centrality nodes): Betweenness centrality is a measure of a
node’s strategic position in the network.133 Nodes that are high on betweenness centrality
occupy gate keeper positions, where they most frequently are on the shortest path connecting
other nodes. Nodes that are high on betweenness centrality may play an important role in

76
connecting subgroups or communities within the network. The top 15% of nodes with the
highest betweenness centrality were selected in this condition.
3. Segmentation + opinion leaders (OL): A community detection algorithm was used to
determine the number of communities in the network and the two nodes with the greatest indegree in each community/cluster were selected for the intervention. This intervention
approach ensures that the opinion leaders from each network community are included in the
intervention.119
4. Random: In this condition 15% of nodes were randomly selected without regard to degree or
network position for comparison.
E-cigarette social norms
In the ADVANCE study, e-cigarette social norms were measured with four items on a
likert-scale. A composite social norms score was calculated with higher values indicative of
greater pro-e-cigarette norms which were associated with increased individual e-cigarette use
over time.134 Social norms were approximately normally distributed in the observed networks
(see Appendix A supplemental figure 1). To adapt this information into the simulated networks,
e-cigarette norms were normally distributed (mean = 50, SD = 7.5). These values were chosen to
guarantee that 95% of the values would fall between 35-65, to reduce overlap with social norms
distribution of the intervention nodes. Changes in social norms were tested to represent
theoretical changes resulting from a tobacco-prevention intervention, where one might use peer
leaders or other network members to act as change agents.88,89 In each simulation run, 15% of the
nodes in the network were chosen according to the intervention condition and those nodes were
assigned different social norm distributions relative to the rest of the network. Three norms

77
conditions were tested: anti-e-cigarette norms (mean = 15, SD = 7.5), neutral e-cigarette norms
(mean = 50, SD = 7.5), and pro-e-cigarette norms (mean = 85, SD = 7.5) (Figure 5).
Threshold
Threshold for adoption is the theoretical level that would lead to the individual adopting the
behavior.124
𝑎𝑖 = {
1 𝑖𝑓 𝜏𝑖 < 𝐸𝑖
0 𝑖𝑓 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒 𝐸𝑖 =
∑𝑖 ≠𝑗𝑊𝑖𝑗𝑎𝑗
∑𝑖 ≠𝑗𝑊𝑖𝑗
(1)
Where adoption of the behavior 𝑎𝑖
, occurs if threshold 𝜏𝑖
, is less than exposure 𝐸𝑖
. And network
exposure to the behavior is defined as the proportion of people in one’s social network who have
adopted the behavior (i.e. initiated e-cigarette use) calculated as the adjacency matrix 𝑊𝑖𝑗 and the
alters who have adopted 𝑎𝑗
, divided by the number of alters in one’s network. Network
thresholds were normally distributed in the simulated networks (mean = 0.33, SD 0.16).
Intervention nodes were given higher or lower thresholds to correspond with anti-e-cigarette or
pro-cigarette norms (i.e. intervention leaders with anti-e-cig norms were given a higher threshold
to decrease the chance that they would become users, whereas the opposite is true for the pro-ecig norms condition). Thresholds were normally distributed for the intervention nodes in both the
pro-e-cig norms (mean = 0.25, SD = 0.10) and anti-e-cig norms conditions (mean = 0.75, SD =
0.10). Intervention nodes in the neutral e-cig norms condition were not assigned a change.
Social Influence
Many diffusion models measure behavior adoption as a function of network exposure, or direct
exposure to people in one’s network who have already adopted the behavior. Models were run

78
with diffusion as a function of network exposure weighted by the three social norm conditions
(anti-e-cig, neutral, and pro-e-cig).
𝐸𝑡 =
(𝑊𝑡 × [𝑥𝑡 ∘ 𝐴𝑡
])
( 𝑊𝑡 × 𝑥𝑡)
(2)
Where 𝐸𝑡
is network exposure at time t, 𝑊𝑡
is the graph at time t, 𝑥𝑡
is an attribute vector of size
n at time t (the attribute is perceived norms), and 𝐴𝑡
is the column of cumulative adopters at time
t in the cumulative adopter matrix. The symbol ′ × ′ indicates matrix product.
The classic model measures social influence as a function of network exposure through direct
neighboring nodes. Network exposure is calculated as the number of adopters in one’s personal
network at the previous time period divided by the total outdegree and expressed as a
proportion.125 Each individual node has an exposure score that ranges from zero to one for each
time period. When a node’s threshold is reached, they adopt the behavior. In the classic model,
exposure is only calculated on the behavior of interest with no attributes weighting the exposure.
In the three norms conditions tested, each node was assigned a value for e-cigarette social norms
from distributions described above. Network exposure (proportion of adopters in one’s network)
was weighted by the social norm values, providing a weighted exposure score for each
individual. This weighting represents the theoretical impact of alters’ e-cigarette norms and
attitudes on an individual’s adoption of the behavior (i.e. individuals consider not only the
behavior but also the attitude or approval of their alters).
Analysis

79
This study follows a similar analytic plan as a prior diffusion simulation study by Valente
and Vega Yon125 using the netdiffuseR package in R.135 Models were initialized with networks
of the same size (N = 300), structure (small world) and proportion of initial e-cigarette users
(5%) based on averages obtained from observed data. Seeding conditions (random, central,
community), and intervention condition (opinion leaders, betweenness, segmentation + OL,
random) were varied resulting in twelve seed x intervention conditions that were tested across
three e-cigarette social norm distributions (anti-e-cig, neutral, and pro-e-cig). Each seed x
intervention x norms condition was simulated 100 times. This resulted in a total of 36 conditions
x 100 diffusion simulations.
For each run of the model network metrics (average path length, modularity), threshold
distribution parameters (average, minimum, maximum), norm distribution parameters (average,
minimum, maximum), average infection rate, average susceptibility rate, average final
prevalence, and rate of e-cigarette use initiation were stored. Multivariate generalized linear
models with a logit-link (since outcomes are on the interval ratio scale) and robust standard
errors were estimated for e-cigarette diffusion prevalence and rate on network metrics, seed
condition, intervention strategy, and social norm conditions.
RESULTS
The main results from the simulations are displayed in Figure 6 and described in detail
below. The simulated changes in e-cigarette social norms had the hypothesized impact on
diffusion, with pro-e-cig social norms having the greatest diffusion prevalence and rate and antie-cig social norms having the lowest and slowest diffusion prevalence and rate. In the anti-e-cig
norms condition, the betweenness centrality approach had the greatest effect on diffusion
dynamics.

80
E-cigarette social norm distributions
Overall mean diffusion prevalence and rates are presented in Table 9. Across all seed and
intervention conditions, anti-e-cig norms had a lower average diffusion prevalence and slower
average rate compared to other conditions, while the pro-e-cig norms condition had a greater
average diffusion prevalence and accelerated diffusion rate compared to other conditions.
Seed Condition
Across all e-cigarette social norm distributions and intervention conditions, the
community-based seed condition had a noticeably lower average diffusion prevalence and slower
diffusion rate compared to other conditions. By contrast, the centrally seeded networks had a
greater average prevalence and accelerate rate compared to all other conditions (Table 9).
Intervention strategies
All intervention conditions had relatively similar average prevalence and diffusion rates
when pooling all seed conditions and norms distributions, with no single intervention strategy
emerging as superior overall. However, other patterns emerge after further stratifying the data.
Table 10 displays mean prevalence and rate for the varying e-cigarette social norm conditions by
intervention strategy (pooled across seed conditions) and seed conditions by intervention strategy
(pooled across social norms conditions). When stratifying intervention approaches by social
norms (still pooled across all seed conditions), the betweenness centrality intervention had the
lowest diffusion rate and prevalence in the anti-e-cig norms condition compared to all others
(mean prevalence = 0.49, mean hazard rate = 0.08). In the neutral e-cig norms condition, which
represents no change in norms, the opinion leader intervention condition had the lowest diffusion
prevalence and rate (mean prevalence = 0.62, mean rate = 0.12). Lastly, in the pro-e-cig

81
condition, the betweenness intervention had the highest average diffusion prevalence and rate
(mean prevalence = 0.79, mean rate = 0.20).
Multivariate analyses
Multivariate generalized linear models were estimated to demonstrate how simulated
changes in network conditions and intervention approaches would theoretically impact diffusion
dynamics (Table 11). Diffusion prevalence and rate were significantly impacted by network
metrics, seed conditions, changes in social norms, and interactions between social norms and
intervention conditions.
For network metrics, average path length was negatively associated with diffusion
prevalence and rate, such that increases in average path length resulted in slower diffusion. The
average threshold minimum was also associated with diffusion prevalence and rate, but the
average threshold maximum was only associated with rate. Networks seeded with central nodes
had a higher prevalence of e-cig adopters at the end of the simulation (AOR = 1.90) and
experienced a faster diffusion rate (AOR = 1.48) compared to networks seeded on random nodes.
By contrast, networks seeded on a single community had a lower prevalence of e-cig adopters
(AOR = 0.28) and slower diffusion rate (AOR = 0.37) compared to random seeds.
There was a significant main effect of changes in e-cigarette social norms for intervention
nodes on diffusion dynamics. When approximately 15% of nodes have anti-e-cig social norms
relative to the rest of the network there is significantly lower average prevalence and slower
average rate of diffusion (AOR = 0.66, AOR = 0.74) compared to networks with neutral norms
or no alteration in norms. Conversely, when the intervention nodes hold pro-e-cig norms, the
network experiences a significant increase in average prevalence and an accelerated average rate
of diffusion (AOR= 1.35, AOR = 1.24) compared to networks with neutral norms.

82
There was no significant main effect of intervention condition on diffusion dynamics,
however, when intervention strategy was interacted with e-cigarette social norms several
intervention x norm conditions emerged as significant. First, the betweenness centrality
intervention x anti-e-cig norms had the strongest effect on decreasing diffusion prevalence (AOR
= 0.79) and rate (AOR = 0.83), while the opinion leader x anti-e-cig condition resulted in
marginally significant slower diffusion rate (AOR = 0.91). Finally, all interactions between
intervention conditions and pro-e-cig norms resulted in significantly increased diffusion
prevalence and accelerated diffusion rates (Table 11).
Sensitivity Analysis
To determine if changes in diffusion could still be achieved with a lower proportion of
network members participating in the intervention, the simulation and regression analyses were
run again with 10% of network members delegated as intervention nodes. The results of the
multivariate models (results not shown) were consistent with the models presented in the current
study, with the betweenness centrality intervention strategy having the greatest impact on
diffusion outcomes. There was one additional significant condition, anti-e-cig norms x opinion
leaders, which resulted in decreased diffusion prevalence and rate. One reason for this could be
that in interventions where fewer network members are being recruited, opinion leaders'
influence is roughly equivalent to those chosen based on betweenness centrality, but when
recruiting a greater proportion of the network, having additional intervention nodes high on
betweenness centrality may amplify their effect. The main effect of both pro- and anti-e-cig
norms remained significant, and the betweenness centrality intervention strategy remained the
strongest of all intervention approaches.

83
DISCUSSION
This study builds on past diffusion simulation research in two ways: first, it incorporates
personal attributes into a measure of weighted exposure, and second it tests the effects of
different network-based intervention strategies on diffusion. This extension tests how simulated
changes in personal attributes, such as social norms, or different intervention conditions would
theoretically impact diffusion outcomes, both of which are important factors to consider in
behavior change applications.
While diffusion is governed by network structure, exposure, and the threshold
distribution, in real world settings personal attributes are likely to affect one’s predisposition to
behavior adoption. The social norms, attitudes, or opinions that are prevalent among one’s
network are likely to exert influence and affect one’s susceptibility to social pressures.58,97,136
Exploring network exposure weighted by personal attributes allows us to consider these effects
in simulations and offers a potential pathway for interventions. While network metrics,
thresholds, and exposures are measurable, altering these parameters through an intervention
presents challenges. First, thresholds cannot be established until one has adopted a behavior (i.e.
threshold is network exposure prior to adoption), so in tobacco prevention or other public health
applications, we couldn’t measure threshold until after individuals have adopted the behavior.
Social norms surrounding a behavior are measurable regardless of adoption status and could
provide insight on prevailing network attitudes.
Second, changing the network structure or an individual’s network exposure presents
challenges and essentially requires forming new ties or dissolving existing ones. This would
imply instructing individuals who they can/cannot be friends with. However, strategically

84
addressing social norms of key network members offers an approach for leveraging the existing
network structure to disseminate anti-tobacco messaging.
There is substantial evidence that social network interventions are effective at creating
behavior change.87 A key component of intervention effectiveness relies on which network
members are recruited to participate in the intervention. Research has demonstrated that selecting
individuals based on opinion leadership, or indegree centrality, is effective in achieving behavior
change.87-89,119,123,137 Findings from the present study are consistent with this approach, but
simulation results interestingly show that selecting individuals based on betweenness centrality
was more effective in impacting diffusion. Selecting individuals with high betweenness
centrality to lead interventions may offer several advantages over opinion leaders or popular
individuals. Those who have greater betweenness centrality occupy network positions where
they are often on the shortest path connecting others.8 Given this strategic position in the
network, they could be at an advantage for transmitting information between different social
groups or communities within the network. While an individual can have both high in-degree
centrality and high betweenness centrality, the individuals ranking highest in these metrics are
not always the same people.138 There may be scenarios where the popular individuals in the
network are e-cigarette users and would not be suitable to lead an intervention,53 or where the
most central individuals cannot reach less connected members of the network. Therefore, it is
important to consider and assess network structure, initial users, and susceptible individuals
when recruiting leaders to optimize intervention effectiveness.
Strengths
This study has several strengths, first of which is the use of simulation models to test the
effects of theoretical changes in network parameters, absent the constraints of empirical data.

85
Building on past diffusion simulation work, this study adds two important elements that are
relevant to public health applications. Examining changes in personal attributes allows us to
model how theoretical changes in attitudes could impact behavior adoption under certain
network conditions. Second, comparing the results of different intervention strategies provides
insight on which approaches could be optimal. Finally, the baseline network conditions are
informed by observed data, allowing us to design a network simulation with more fidelity to real
life social networks.
Limitations
The results of this study are contingent on the algorithms and distributions chosen to
represent these network features and will inevitably vary in real world settings which limits the
generalizability of findings. However, the results of this study provide insight on diffusion
processes as a function of network seeding, intervention approaches, and changes in attributes,
highlighting the importance of these factors and their interactions with one another.
Conclusions
The objective of this study was to evaluate the effects of changes in e-cigarette social
norms on diffusion prevalence and rate of e-cigarette use. These network simulation models
demonstrated that a theoretical change for approximately 10-15% of network members has a
significant impact on both prevalence and rate. Further, the comparison of different network
intervention strategies revealed that recruiting individuals with high betweenness centrality could
achieve the greatest change in diffusion outcomes under certain network conditions. The findings
from this study can be used to design network-based behavior change interventions and
demonstrate how network simulations could aide in optimizing intervention effectiveness.

86
Table 9. Diffusion prevalence and rate by intervention strategy and social
norms or seed condition
Mean Diffusion
Prevalence
Mean Diffusion
Rate
Intervention strategy
Opinion Leaders 0.626 0.133
Betweenness Centrality 0.635 0.136
Segmentation + Opinion Leaders 0.631 0.131
Random 0.621 0.124
E-cig Social Norm Distribution
Anti-E-cig 0.518 0.089
Neutral 0.628 0.125
Pro-E-cig 0.738 0.179
Seed Condition
Random 0.687 0.141
Central 0.802 0.194
Community 0.395 0.058

87
Table 10. Diffusion prevalence and rate by intervention strategy and social norms or seed condition
Intervention Strategy
Opinion Leaders Betweenness
Segmentation +
Opinion Leaders Random
E-cig Social Norm
Distribution
Mean
Prevalence
Mean
Rate
Mean
Prevalence
Mean
Rate
Mean
Prevalence
Mean
Rate
Mean
Prevalence
Mean
Rate
Anti-E-cig 0.498 0.084 0.485 0.078 0.553 0.100 0.537 0.094
Neutral 0.621 0.122 0.632 0.126 0.625 0.125 0.635 0.127
Pro-E-cig 0.759 0.194 0.788 0.204 0.716 0.168 0.69 0.152
Seed Condition
Random 0.684 0.143 0.687 0.143 0.690 0.139 0.686 0.138
Central 0.803 0.200 0.797 0.199 0.818 0.198 0.792 0.18
Community 0.391 0.057 0.421 0.066 0.386 0.055 0.384 0.055

88
Table 11. Multivariate regressions of diffusion prevalence and rate on network metrics, seed condition,
intervention strategy and e-cigarette social norms
Prevalence Rate
Parameter AOR CI AOR CI
Modularity 1.68 [0.05, 60.61] 0.73 [0.08, 6.57]
APL 0.26** [0.12, 0.59] 0.45** [0.27, 0.76]
Threshold Min 0.19*** [0.13, 0.28] 0.38*** [0.3, 0.48]
Threshold Max 0.79 [0.53, 1.18] 0.79^ [0.62, 1.02]
Intervention Strategy (ref. = random selection)
Opinion Leader (in-degree) 0.93 [0.83, 1.04] 0.95 [0.89, 1.02]
Betweenness Centrality 1.00 [0.89, 1.11] 0.99 [0.92, 1.06]
Segmentation + Opinion Leader 0.96 [0.86, 1.07] 0.98 [0.92, 1.05]
Seed Condition (ref. = random)
Central 1.90*** [1.81, 2.01] 1.48*** [1.43, 1.53]
Community 0.28*** [0.26, 0.29] 0.37*** [0.35, 0.39]
E-cigarette Social Norms (ref = neutral)
Anti-E-cig norms (negative) 0.66*** [0.58, 0.76] 0.74*** [0.69, 0.81]
Pro-E-cig norms (positive) 1.35*** [1.2, 1.51] 1.24*** [1.16, 1.33]
Intervention x Norms Condition
Opinion leader x Anti-E-cig 0.89 [0.77, 1.03] 0.91^ [0.83, 1.01]
Betweenness x Anti-E-cig 0.79** [0.68, 0.91] 0.83*** [0.75, 0.91]
Segmentation + OL x Anti-E-cig 1.11 [0.96, 1.29] 1.08 [0.98, 1.19]
Opinion leader x Pro-E-cig 1.60*** [1.35, 1.9] 1.43*** [1.3, 1.58]
Betweenness x Pro-E-cig 1.83*** [1.53, 2.19] 1.47*** [1.33, 1.62]
Segmentation OL x Pro-E-cig 1.20* [1.02, 1.41] 1.15** [1.04, 1.27]
Note: ref. = reference
^p < 0.10; *p < 0.05; **p < 0.01; ***p < 0.001

89
Figure 4. Simulated network intervention strategies. Red nodes indicate initial seeds/e-cigarette
users, yellow nodes indicate intervention nodes selected based on (a) opinion leaders/high indegree, (b) betweenness centrality, (c) segmentation + opinion leaders, and (d) random.

90
Figure 5. Examples of simulated e-cigarette social norm distributions where the intervention
nodes are assigned different norms relative to the rest of the network: (a) pro-e-cigarette norms,
(b) neutral e-cigarette norms, and (c) anti-e-cigarette norms.

91
Figure 6. Diffusion of e-cigarette use over time by seed condition, intervention strategy, and
social norms.

92
CHAPTER 5
Discussion

93
DISCUSSION
The overarching objective of this dissertation was to identify and differentiate between
mechanisms of social influence driving the propagation of e-cigarette use in adolescent social
networks with the aim of informing public health interventions to slow or curtail use. Using data
from the ADVANCE study, a cohort of racially and ethnically diverse of adolescents in Southern
California, each dissertation study took a different social network analytic approach to address
specific aims: (1) Assess the mechanisms of social influence that drive e-cigarette use by
evaluating the differential effects of perceived social norms and friend use on individual ecigarette use; (2) Use stochastic actor-oriented models to evaluate the longitudinal
relationship between friend influence and friend selection on e-cigarette use (i.e., how ecigarette use propagates throughout networks over time) and clarify the role of perceived
peer norms; and (3) Use parameters from Aims 1 and 2 in agent-based simulation models
to evaluate the potential impact of targeting perceived norms for e-cigarette use prevention
and reduction in adolescent social networks.
Study 1
The goal of study 1 was to compare the effects of perceived norms and peer exposure eon past 6-month e-cigarette use and to assess if individuals tend to over-estimate e-cigarette use
among friends. Two key findings emerged from study 1: First, on average all participants overestimate e-cig use among peers, and that current e-cigarette users significantly over-estimate use
among friends. This finding is consistent with past research on perceived norms surrounding
substance use where individuals tend to over-estimate risk behaviors among friends.38,65,96,98
Second, this study used several mixed effect regression models to examine factors associated
with past 6-month e-cig initiation and use. The full sample model determined that past vaping

94
behavior was the strongest predictor of current vaping, and that both peer network exposure and
perceived pro-vaping norms were significantly associated with use while controlling for past
behavior and other covariates. Among never users of e-cigarettes these associations still held,
with increases in both network exposure and pro-vaping norms being strongly associated with ecigarette use initiation at wave 3. This study laid the foundation for study 2, which took a more
nuanced look at network and behavior evolution.
Study 2
The primary objective of study 2 was to estimate SAOMs to evaluate peer influence and
selection effects, and further to differentiate between mechanisms of peer influence. SAOMs are
often considered the gold standard method for estimating changes in networks and behavior over
time and are a more rigorous approach than the regression models in study 1. The results of the
Siena meta-analysis showed that once again peer influence effects acting through peer exposure
and pro-e-cigarette social norms. Additionally, these models revealed that Hispanic/Latine and
female students were at increased risk of initiating e-cigarette use compared to their peers when
controlling for changes in individual and network factors. The hypothesized selection effects
were unable to be estimated in these models and instead were examined through separate
regressions (Appendix A Supplemental Table 5). Forming new friendships with e-cigarette users
(i.e. selecting e-cigarette users as friends) was associated with individual e-cigarette use at wave
3 but not at wave 2 in the regression analyses. Studies 1 and 2 provided baseline estimates and
insight on network influences that informed study 3.
Study 3
The third and final dissertation study utilized agent-based network simulations to
evaluate how changes in social norms affect diffusion dynamics while comparing several

95
network intervention approaches. Study 3 highlights the importance of social norms and
attitudes on behavior spread. In simulated network interventions, changing social norms for
approximately 15% of network members resulted in a marked decrease in diffusion prevalence
and rate. Selecting individuals with high betweenness centrality as intervention change agents
emerged as the optimal intervention strategy on the simulated network conditions. This study
helps us understand how additional factors (i.e. norms, attitudes) impact diffusion and illustrates
the utility of simulations in testing intervention approaches. The baseline networks in the
simulations were rooted in empirical measures from studies 1 and 2 to emulate a more realistic
simulation. The findings of this study could be used to inform future interventions and
demonstrate how addressing social norms and purposeful recruitment of intervention participants
can be used to optimize intervention effectiveness.
Summary of findings
The dissertation studies address several gaps in the literature on social networks, social
norms, and e-cigarette use. While the relationship between social networks and smoking is well
established, these are some of the first longitudinal analyses evaluating e-cigarette use in
adolescent social networks. The two hypothesized mechanisms of social influence were both
supported in studies 1 and 2. Social norms and peer exposure to e-cigarettes both have distinct
influences on individual use. Study 3 showed the utility of including personal attributes in
diffusion simulations to test how theoretical changes in these attributes could potentially impact
diffusion dynamics. Further, study 3 examined several different intervention strategies and found
that selecting network members based on betweenness centrality or opinion leadership (high indegree) are the most effectives approaches for behavior change interventions. These studies have
thoroughly examined mechanisms of social influence in adolescent networks.

96
IMPLICATIONS
The findings from the dissertation studies have several important methodological and public
health implications and can be summarized by two key points:
1. Social norms and behavioral perceptions influence individual e-cigarette use behavior,
even when controlling for peer network exposure.
2. There is great potential for social norms and social networks to be strategically leveraged
in behavior change interventions.
Methodological implications
The dissertation studies use various social network analytic methods in each chapter to
explore social influence effects. While study 1 uses mixed effects logistic regression and study 2
uses stochastic actor-oriented models, the findings from the two studies reflect similar findings.
While SAOM are often viewed as the most rigorous method for examining changes in networks
and behavior, recent work has demonstrated that estimates from SAOMs are not necessarily
more conservative than regression estimates.139 Both methods identified peer influence effects in
the dissertation studies, but the SAOM also identified associations between ethnicity, gender,
and e-cigarette use. Additionally, the regression models in study 1 are only looking at behavioral
outcomes not evaluating friendship selection effects, while study 2 provides insight on friendship
network evolution.
Several prior network simulation studies have used SAOMs and other agent-based
models to explore changes in networks and behavior.79,86,126 Building on past research, study 3
examines behavior change in a diffusion of innovations framework, which accounts for behavior
adoption thresholds in addition to network exposure effects in the simulations.

97
Public health implications
The key finding from the dissertation studies is that social influence operates through two
distinct mechanisms of social norms and behavior exposure in social networks. Knowing this, we
can begin to design social network interventions to discourage positive attitudes towards ecigarettes, clarify harms and risks, and address normative misperceptions about substance use
among friends. E-cigarette use is a social problem that requires a social response. Tobacco
prevention initiatives grounded in a social network framework offer great potential for shifting
norms and achieving behavior change.
Limitations
There are limitations to the dissertation studies, some inherent to study design and others
due to data demands. First, data collection began in Fall of 2020 during the COVID-19 pandemic
which presented several challenges. Schools were conducting classes via remote learning/online
teaching in Los Angeles County for the 2020-2021 academic year. This not only appears to have
resulted in lower enrollment in earlier waves, but also impacted friendship formation at the
beginning of high school, potentially resulting in the lower-than-expected degree scores in the
friendship networks. The ADVANCE research team went through great efforts to enroll as many
students as possible at each wave. The impact of the pandemic on data collection was felt by
many researchers across the country and was not under investigators’ control.
Second, incomplete network data created obstacles in data cleaning and limitations in
analyses. Not all students received parental consent/assented to participate in the study, which
resulted in their removal from the network data set. This meant that if participating students
made friendship nominations to non-participants, this data had to be removed. Additionally, if
students wrote-in nominations to students not attending their school or not listed on the school

98
roster, then these nominations also had to be removed. While a string-matching algorithm was
used to attempt to match the write-in nominations, there was still data loss at each wave. The
model in study 1 only included students that had complete data for all three waves, which limited
the sample size. However, supplemental analyses in study 1 were run with an imputed data set
and findings were in alignment.
Another limitation was that the social norms questions were not asked to all participants
at wave 3 (Spring 2022) due to changes in the health behavior/substance use portion of the
survey. Instead of imputing social norms scores for this wave, a decision was made the truncate
the dissertation analysis at wave 3.
While the dissertation studies thoroughly examine friend/peer influence effects on ecigarette use, no questions were included in the study about parent and/or sibling e-cigarette use,
therefore we were unable to control for these influences in the analyses. Research has shown
overall normative perceptions about tobacco use, as well as close friend and sibling smoking, are
associated with individual smoking.94 Future work should consider these effects.
The results of the dissertation studies are most applicable to adolescents in Southern
California and may have limited generalizability to other racial/ethnic and geographic
populations. Additionally, the results of study 3 are based on the chosen network parameters and
algorithms used in the simulation. While the baseline networks are grounded in empirical
network data, simulations cannot perfectly control for all conditions in real world social settings.
FUTURE RESEARCH
There are several opportunities and areas of future research to build on the current studies and
multiple avenues for public health applications.

99
1. Validated e-cigarette social norms scale
The perceived e-cigarette social norms scale in the dissertation studies comprised of four
items from the ADVANCE survey that had moderate factor loadings when analyzed
together as a scale. There are several e-cigarette scales pertaining to perceptions of ecigarette harm, e-cigarette advertising, flavor preference, and product types. The latter
three were developed in 2022 when the ADVANCE study was already underway. Given
the demonstrated association between e-cigarette use and social influences, a validated
perceived social norms scale would be a valuable instrument in future studies.
2. Culturally tailored studies
Study 2 findings agreed with national reports that Hispanic/Latine adolescents are at
increased risk of e-cigarette use compared to their non-Hispanic peers. This necessitates a
closer look at those factors driving use among this population, including social (peer,
sibling, parent), environmental, and personal factors. Further, this highlights the needs for
culturally tailored interventions and public health messaging designed to address ecigarette attitudes and norms in this population.
3. Selection effects
The relationship between e-cigarette use and popularity is still unclear. Future analyses
and/or analysis of later study waves could clarify if e-cigarette users are more likely to be
selected as friends.
4. Social network and e-cigarette use interventions
The findings from the dissertation studies are relevant to multiple public health
applications. This work lays the foundation for piloting social network interventions to

100
address social norms surrounding e-cigarette use. Strategically recruiting network
members with high betweenness or in-degree centrality to lead an intervention could be
an effective approach to slow the spread of e-cigarette use in adolescent networks.
CONCLUSION
The findings from this dissertation demonstrate social influences on e-cigarette use, but
in a broader context they exemplify how normative perceptions (and misperceptions) can have
significant impacts on individual behavior. Many public health issues would benefit from
examination within a social network framework and a systems level approach. As demonstrated
by these studies, social network analysis not only provides insight on dynamic interpersonal
influence processes but offers a path forward to address and remedy complex public health
problems.

101
REFERENCES
1. CDC/MMWR. Vital Signs: Tobacco Product Use Among Middle and High School
Students—United States, 2011–2018. Morbidity and Mortality Weekly Report. 2019;68
2. Barrington-Trimis JL, Berhane K, Unger JB, et al. Psychosocial Factors Associated With
Adolescent Electronic Cigarette and Cigarette Use. Pediatrics. Aug 2015;136(2):308-17.
doi:10.1542/peds.2015-0639
3. E-cigarette Use Among Youth and Young Adults: A Report of the Surgeon General
(2016).
4. Birdsey J, Cornelius M, A. J. Tobacco Product Use Among U.S. Middle and High School
Students — National Youth Tobacco Survey. MMWR Morb Mortal Wkly Rep.
2023;72(44):1173–1182. doi: http://dx.doi.org/10.15585/mmwr.mm7244a1
5. Barrington-Trimis J, Berhane K, Unger J, et al. E-cigarette use, the e-cigarette
psychosocial environment, and susceptibility to cigarette use. presented at: Society for Research
on Nicotine and Tobacco; 2015; Philadelphia, PA.
6. Barrington-Trimis JL, Urman R, Berhane K, et al. E-cigarettes and Future Cigarette Use.
Pediatrics. 2016;
7. Soneji S, Barrington-Trimis JL, Wills TA, et al. Association Between Initial Use of eCigarettes and Subsequent Cigarette Smoking Among Adolescents and Young Adults: A
Systematic Review and Meta-analysis. JAMA Pediatr. Jun
2017;doi:10.1001/jamapediatrics.2017.1488
8. Valente TW. Social Networks and Health: Models, Methods, and Applications. Oxford
University Press; 2010.
9. Gilbert PA, Kava CM, Afifi R. High-School Students Rarely Use E-Cigarettes Alone: A
Sociodemographic Analysis of Polysubstance Use Among Adolescents in the United States.
Nicotine & Tobacco Research. 2021;23(3):505-510. doi:10.1093/ntr/ntaa037
10. Barrington-Trimis JL, Kong G, Leventhal AM, et al. E-cigarette Use and Subsequent
Smoking Frequency Among Adolescents. Pediatrics. Dec 2018;142(6)doi:10.1542/peds.2018-
0486
11. Kong G, Chaffee BW, Wu R, et al. E-cigarette device type and combustible tobacco use:
Results from a pooled analysis of 10,482 youth. Drug and Alcohol Dependence. 2022/03/01/
2022;232:109279. doi:https://doi.org/10.1016/j.drugalcdep.2022.109279
12. Yang Z, Berhane K, Leventhal AM, Liu M, Barrington-Trimis JL, Thomas DC.
Modeling the longitudinal transitions of electronic cigarettes and conventional cigarettes with
time-dependent covariates among adolescents. Preventive Medicine. 2022/11/01/
2022;164:107294. doi:https://doi.org/10.1016/j.ypmed.2022.107294

102
13. CDC. Outbreak of lung injury associated with the use of e-cigarette, or vaping, products
2020. www.cdc.gov/tobacco/basic_information/e-cigarettes/severe-lung-disease.html
14. Krishnasamy VP, Hallowell BD, Ko JY, et al. Update: Characteristics of a Nationwide
Outbreak of E-cigarette, or Vaping, Product Use-Associated Lung Injury - United States, August
2019-January 2020. MMWR Morb Mortal Wkly Rep. Jan 24 2020;69(3):90-94.
doi:10.15585/mmwr.mm6903e2
15. Zutrauen S, Do MT, Ghandour L, et al. Acute injury or illness related to the inhalation of
vaping aerosols among children and adolescents across Canada: A cross-sectional survey of
Canadian paediatricians. Paediatr Child Health. Mar 2022;27(1):43-49.
doi:10.1093/pch/pxab062
16. Overbeek DL, Kass AP, Chiel LE, Boyer EW, Casey AMH. A review of toxic effects of
electronic cigarettes/vaping in adolescents and young adults. Critical reviews in toxicology.
2020;50(6):531-538. doi:10.1080/10408444.2020.1794443
17. Fielding-Singh P, Epperson AE, Prochaska JJ. Tobacco Product Promotions Remain
Ubiquitous and Are Associated with Use and Susceptibility to Use Among Adolescents. Nicotine
Tob Res. Jan 22 2021;23(2):397-401. doi:10.1093/ntr/ntaa136
18. Hung M, Spencer A, Goh C, et al. The association of adolescent e-cigarette harm
perception to advertising exposure and marketing type. Arch Public Health. Apr 8
2022;80(1):114. doi:10.1186/s13690-022-00867-6
19. Glantz S, Jeffers A, Winickoff JP. Nicotine Addiction and Intensity of e-Cigarette Use by
Adolescents in the US, 2014 to 2021. JAMA Network Open. 2022;5(11):e2240671-e2240671.
doi:10.1001/jamanetworkopen.2022.40671
20. Gorukanti A, Delucchi K, Ling P, Fisher-Travis R, Halpern-Felsher B. Adolescents'
attitudes towards e-cigarette ingredients, safety, addictive properties, social norms, and
regulation. Preventive medicine. 2017;94:65-71. doi:10.1016/j.ypmed.2016.10.019
21. Vogel EA, Cho J, Mcconnell RS, Barrington-Trimis JL, Leventhal AM. Prevalence of
Electronic Cigarette Dependence Among Youth and Its Association With Future Use. JAMA
Network Open. 2020-02-19 2020;3(2):e1921513. doi:10.1001/jamanetworkopen.2019.21513
22. Hair EC, Barton AA, Perks SN, et al. Association between e-cigarette use and future
combustible cigarette use: Evidence from a prospective cohort of youth and young adults, 2017-
2019. Addict Behav. Jan 2021;112:106593. doi:10.1016/j.addbeh.2020.106593
23. Leventhal AM, Strong DR, Kirkpatrick MG, et al. Association of Electronic Cigarette
Use With Initiation of Combustible Tobacco Product Smoking in Early Adolescence. JAMA.
2015;314:700-707.
24. Soneji S, Barrington-Trimis JL, Wills TA, et al. Association Between Initial Use of eCigarettes and Subsequent Cigarette Smoking Among Adolescents and Young Adults: A
Systematic Review and Meta-analysis. JAMA Pediatrics. 2017;171:788-797.

103
25. Kwon E, Adams Z, Seo D-C. Trajectories and determinants of adolescents’ nicotine
product use risk among U.S. adolescents in a nationally representative sample of longitudinal
cohort. Addictive behaviors. 2021;116:106812-106812. doi:10.1016/j.addbeh.2020.106812
26. Green HDJ, Horta M, de la Haye K, Tucker JS, Kennedy DR, Pollard M. Peer influence
and selection processes in adolescent smoking behavior: A comparative study. Nicotine &
Tobacco Research. 2013;15:534-541.
27. De la Haye K, Shin HS, Vega Yon G, Valente TW. Smoking Diffusion through Networks
of Diverse, Urban American Adolescents over the High School Period. Journal of Health and
Social Behavior. in press;
28. de la Haye K, Green. HD, Pollard M, Kennedy DP, Tucker JS. Befriending risky peers:
Factors driving adolescents’ selection of friends with similar marijuana use. Journal of Youth
and Adolescence. 2015;44:1914-1928.
29. de la Haye K, Green, H. D. Jr., Kennedy, D. R., Pollard, M., & Tucker, J. S. Selection
and influence mechanisms associated with marijuana initiation and use in adolescent friendship
networks. Journal of Research on Adolescence. 2013;23:474-486.
30. Tucker JS, de la Haye K, Pollard M, Green HDJ, Kennedy DR. .Peer influence on
marijuana use in different types of friendships. 54(1), 67-73. Journal of Adolescent Health.
2014;54:67-73.
31. de la Haye K, Robins G, Mohr P, Wilson C. How physical activity shapes, and is shaped
by, adolescent friendships. Social Science & Medicine. 2011;73:719-728.
32. De La Haye K, Robins, G., Mohr, P., Wilson, C. Homophily and Contagion as
Explanations for Weight Similarities Among Adolescent Friends. Journal of Adolescent Health.
2011;49:421-427.
33. Mercken L, Snijders T, Steglich C, Vertiainen E, de Vries H. Smoking-based selection
and influence in gender-segregated friendship networks: a social network analysis of adolescent
smoking. Addiction. 2010;105:1280-1289.
34. Ennett ST, Bauman KE. Peer group structure and adolescent cigarette smoking: A social
network analysis. Journal of Health and Social Behavior. Sep 1993;34(3):226-236.
35. Ennett ST, Bauman KE. The contribution of influence and selection to adolescent peer
group homogeneity: The case of adolescent cigarette smoking. Journal of Personality and Social
Psychology. Oct 1994;67(4):653-663.
36. Ennett ST, Bauman KE, Hussong A, et al. The peer context of adolescent substance use:
Findings from social network analysis. Journal of Research on Adolescence. 2006;16(2):159-
186.
37. Ennett ST, Faris R, Hipp J, et al. Peer smoking, other peer attributes, and adolescent
cigarette smoking: A social network analysis. Prevention Science. 2008;9:88-98.

104
38. Ianotti RJ, Bush PJ. Perceived vs. actual friends’ use of alcohol, cigarettes, marijuana,
and cocaine: Which has the most influence? . Journal of Youth and Adolescence. 1992;21:375-
389.
39. Mercken L, Snijders T, Steglich C, de Vries H. Dynamics of adolescent friendship
networks and smoking behavior: Social network analyses in six European countries. Social
Science & Medicine. 2009;69:1506-1514.
40. Mercken L, Candel M, Willems P, de Vries H. Social influence and selection effects in
the context of smoking behavior: Changes during early and mid adolescence. Health Psychology.
2009;28:73-82.
41. Pearson M, & West, P. . Drifting smoke rings: Social network analysis and Markov
processes in a longitudinal study of friendship groups and risk-taking. Connections.
2005;24(2):59-76.
42. Haas SA, Schaefer DR. With a little help from my friends? Asymmetrical social
influence on adolescent smoking initiation and cessation. Journal of Health and Social Behavior.
2014;55:126-143.
43. Hall J, Valente TW. Adolescent smoking networks: The effects of influence and selection
on future smoking. Addictive Behaviors. 2007;32:3954-3059.
44. Tucker JS, Green JHD, Zhou AJ, Miles JNV, Shih RA, D'Amico EJ. Substance Use
among Middle School Students: Associations with Self-Rated and Peer-Nominated Popularity.
Journal of Adolescence 2011;34:513–519.
45. Unger JB, Chen X. The role of social networks and media receptivity in predicting age of
smoking initiation: A proportional hazards model of risk and protective factors. Addictive
Behaviors. 1999;3:371-381.
46. Unger JB, Rohrbach LA, Cruz TB, et al. Ethnic variation in peer influences on adolescent
smoking. Nicotine & Tobacco Research. May 2001;3(2):167-176.
47. Urberg KA, Cheng CH, Shyu SJ. Grade changes in peer influence on adolescent cigarette
smoking: A comparison of two measures. Addictive Behaviors. 1991;16:21-28.
48. Urberg KA, Degirmencioglu SM, Pilgrim C. Close friend and group influence on
adolescent cigarette smoking and alcohol use. Developmental Psychology. Sep 1997;33(5):834-
844.
49. Kirke D. Chain reactions in adolescents' cigarett, alcohol and drug use: Similarity
through peer influence or the patterning of ties in peer networks. Social Networks. 2004;26:3-28.
50. Kobus K. Peers and adolescent smoking. Addiction. May 2003;98 Suppl 1:37-55.

105
51. Jeon KC, Goodson P. US adolescents' friendship networks and health risk behaviors: a
systematic review of studies using social network analysis and Add Health data. PeerJ (San
Francisco, CA). 2015;3:e1052-e1052. doi:10.7717/peerj.1052
52. Alexander C, Piazza, M., Mekos, D., & Valente, T. W. . Peer networks and adolescent
cigarette smoking: An analysis of the national longitudinal study of adolescent health. Journal of
Adolescent Health. Jul 2001;29(1):22-30.
53. Valente TW, Unger, J., & Johnson, A. C. . Do popular students smoke?: The association
between popularity and smoking among middle school students. Journal of Adolescent Health.
2005;37:323-329.
54. Valente T, Fujimoto K, Soto D, Ritt-Olson A, Unger J. A Comparison of Peer Influence
Measures as Predictors of Smoking among Predominately Hispanic/Latino High School
Adolescents. Journal of Adolescent Health. 2012;52:358–364.
55. Moody J, Brynildsen WD, Osgood DW, Feinberg ME, Gest S. Popularity Trajectories
and Substance Use in early Adolescence. Soc Networks. May 2011;33(2):101-112.
doi:10.1016/j.socnet.2010.10.001
56. Osgood DW, Feinberg ME, Gest SD, et al. Effects of PROSPER on the influence
potential of prosocial versus antisocial youth in adolescent friendship networks. J Adolesc
Health. Aug 2013;53(2):174-9. doi:10.1016/j.jadohealth.2013.02.013
57. Osgood DW, Feinberg ME, Ragan DT. Social Networks and the Diffusion of Adolescent
Problem Behavior: Reliable Estimates of Selection and Influence from Sixth Through Ninth
Grades. Prev Sci. Aug 2015;16(6):832-43. doi:10.1007/s11121-015-0558-7
58. Rimal RN, Real K. Understanding the influence of perceived norms on behaviors.
Communication Theory. 2003;13(2):184-203.
59. Rimal RN, Lapinski MK. A re-explication of social norms, ten years later.
Communication Theory. 2015;25(4):393-409.
60. East K, McNeill A, Thrasher JF, Hitchman SC. Social norms as a predictor of smoking
uptake among youth: a systematic review, meta-analysis and meta-regression of prospective
cohort studies. Addiction (Abingdon, England). 2021;doi:10.1111/add.15427
61. Valente TW, Watkins, S., Jato, M. N., Van der Straten, A., & Tsitsol, L. M. Social
network associations with contraceptive use among Cameroonian women in voluntary
associations. Social Science and Medicine. 1997;45:677-687.
62. Iannotti RJ, Bush PJ. Perceived vs. actual friends' use of alcohol, cigarettes, marijuana,
and cocaine : which has the most influence? Journal of youth and adolescence. 1992;21(3):375-
389. doi:10.1007/BF01537024
63. Rice RE, Donohew, L., & Clayton, R. Peer network, sensation seeking, and drug use
among junior and senior high school students. Connections. 2003;26(2):32-58.

106
64. DiGuiseppi GT, Meisel MK, Balestrieri SG, Ott MQ, Clark MA, Barnett NP.
Relationships between social network characteristics, alcohol use, and alcohol-related
consequences in a large network of first-year college students: How do peer drinking norms fit
in? Psychol Addict Behav. Dec 2018;32(8):914-921. doi:10.1037/adb0000402
65. Cox MJ, DiBello AM, Meisel MK, et al. Do misperceptions of peer drinking influence
personal drinking behavior? Results from a complete social network of first-year college
students. Psychol Addict Behav. May 2019;33(3):297-303. doi:10.1037/adb0000455
66. Snijders TAB, van de Bunt, G. G., & Steglich, C. . Introduction to stochastic actor-based
models for network dynamics. Social Networks. 2010;32:44-60.
67. Steglich C, Snijders TAB, Pearson M. Dynamic networks and behavior: Separating
selection from influence. Sociological Methodology. 2010;20:1-65.
68. Go M-H, Tucker JS, D. GJH, Pollard M, Kennedy D. Social distance and homophily in
adolescent smoking initiation. Drug and Alcohol Dependence. 2012;124:347-354.
69. Schaefer DR, Haas SA, Bishop NJ. A Dynamic Model of U.S. Adolescents’ Smoking and
Friendship Networks. American Journal of Public Health. 2012;102e:e12-e18.
70. Valente TW, Fujimoto K, Unger JB, Soto DW, Meeker D. Variations in network
boundary and type: A study of adolescent peer influences. Social Networks. Jul 2013;35(3):309-
316. doi:10.1016/j.socnet.2013.02.008
71. Harris KM, Udry JR. Data from: National Longitudinal Study of Adolescent to Adult
Health (Add Health), 1994-2018 [Public Use]. 2022. doi:10.3886/ICPSR21600.v24
72. Mercken L, Steglich C, Sinclair P, Holliday J, Moore L. A Longitudinal Social Network
Analysis of Peer Influence, Peer Selection, and Smoking Behavior Among Adolescents in British
Schools. Health psychology. 2012;31(4):450-459. doi:10.1037/a0026876
73. Osgood DW, Ragan DT, Wallace L, Gest SD, Feinberg ME, Moody J. Peers and the
Emergence of Alcohol Use: Influence and Selection Processes in Adolescent Friendship
Networks. J Res Adolesc. Sep 1 2013;23(3)doi:10.1111/jora.12059
74. Lakon CM, Hipp JR, Wang C, Butts CT, Jose R. Simulating Dynamic Network Models
and Adolescent Smoking: The Impact of Varying Peer Influence and Peer Selection. Am J Public
Health. Dec 2015;105(12):2438-48. doi:10.2105/ajph.2015.302789
75. Lakon CM, Wang C, Butts CT, Jose R, Timberlake DS, Hipp JR. A Dynamic Model of
Adolescent Friendship Networks, Parental Influences, and Smoking. J Youth Adolesc. Sep
2015;44(9):1767-86. doi:10.1007/s10964-014-0187-7
76. Lakon CM, Valente TW. Social integration in friendship networks: the synergy of
network structure and peer influence in relation to cigarette smoking among high risk
adolescents. Soc Sci Med. May 2012;74(9):1407-17. doi:10.1016/j.socscimed.2012.01.011

107
77. de la Haye K, Shin H, Yon GGV, Valente TW. Smoking Diffusion through Networks of
Diverse, Urban American Adolescents over the High School Period. Journal of Health and
Social Behavior. Sep 2019;60(3):362-376. doi:10.1177/0022146519870521
78. Ragan DT. Peer beliefs and smoking in adolescence: a longitudinal social network
analysis. Am J Drug Alcohol Abuse. Mar 2016;42(2):222-30.
doi:10.3109/00952990.2015.1119157
79. Schaefer DR, adams J, Haas SA. Social networks and smoking: exploring the effects of
peer influence and smoker popularity through simulations. Health Educ Behav. Oct 2013;40(1
Suppl):24s-32s. doi:10.1177/1090198113493091
80. Valente TW. Network Models and Methods for Studying the Diffusion of Innovations. In
P. Carrington, J. Scott and S. Wasserman (Eds.) Models and Methods in Social Network
Analsysis: Cambridge University Press; 2005. p. 98-116.
81. Valente T. Network models of the diffusion of innovations. Quantitative methods in
communication. Hampton Press; 1995.
82. Granovetter M, Soong R. Threshold models of diffusion and collective behavior. Journal
of Mathematical sociology. 1983;9(3):165-179.
83. Coleman JS, Katz E, Menzel H. Medical innovation: A diffusion study. Indianapolis:
Bobbs-Merrill Company; 1966.
84. Rogers EM, Ascroft JR, Röling NG. Diffusion of innovations in Brazil. Nigeria, and
India (Unpublished report) East Lansing, MI: Michigan State University. 1970;
85. Rogers EM, Kincaid DL. Communication networks: Toward a new paradigm for
research. Free Press; 1981.
86. adams J, Schaefer DR. How Initial Prevalence Moderates Network-based Smoking
Change: Estimating Contextual Effects with Stochastic Actor-based Models. J Health Soc
Behav. Mar 2016;57(1):22-38. doi:10.1177/0022146515627848
87. Hunter RF, de la Haye K, Murray JM, et al. Social network interventions for health
behaviours and outcomes: A systematic review and meta-analysis. . PLOS Medicine. 2019;
16(9): e1002890.https://doi. org/10.1371/journal.pmed.1002890
88. Chu K-H, Sidani J, Matheny S, et al. Implementation of a cluster randomized controlled
trial: Identifying student peer leaders to lead E-cigarette interventions. Addictive behaviors.
2021;114:106726-106726. doi:10.1016/j.addbeh.2020.106726
89. Wyman PA, Rulison K, Pisani AR, et al. Above the influence of vaping: Peer leader
influence and diffusion of a network-informed preventive intervention. Addict Behav. Feb
2021;113:106693. doi:10.1016/j.addbeh.2020.106693

108
90. Harlow AF, Vogel EA, Tackett AP, et al. Adolescent Use of Flavored Non-Tobacco Oral
Nicotine Products. Pediatrics. Sep 1 2022;150(3)doi:10.1542/peds.2022-056586
91. R: A Language and Environment for Statistical Computing. R Foundation for Statistical
Computing; 2023. https://www.R-project.org/
92. Cooper M, Park-Lee E, Ren C, Cornelius M, Jamal A, Cullen KA. Notes from the Field:
E-cigarette Use Among Middle and High School Students - United States, 2022. MMWR Morb
Mortal Wkly Rep. Oct 7 2022;71(40):1283-1285. doi:10.15585/mmwr.mm7140a3
93. National Academies of Sciences E, Medicine. Public Health Consequences of ECigarettes. The National Academies Press; 2018:774.
94. East KA, Hitchman SC, McNeill A, Thrasher JF, Hammond D. Social norms towards
smoking and vaping and associations with product use among youth in England, Canada, and the
US. Drug Alcohol Depend. Dec 1 2019;205:107635. doi:10.1016/j.drugalcdep.2019.107635
95. Valente TW, Watkins S, Jato MN, Van der Straten A, Tsitsol LM. Social network
associations with contraceptive use among Cameroonian women in voluntary associations.
Social Science and Medicine. 1997;45:677-687.
96. Perkins JM, Perkins HW, Jurinsky J, Craig DW. Adolescent Tobacco Use and
Misperceptions of Social Norms Across Schools in the United States. J Stud Alcohol Drugs. Nov
2019;80(6):659-668. doi:10.15288/jsad.2019.80.659
97. Yong H-H, Chow R, East K, et al. Do Social Norms for Cigarette Smoking and Nicotine
Vaping Product Use Predict Trying Nicotine Vaping Products and Attempts to Quit Cigarette
Smoking Amongst Adult Smokers? Findings From the 2016–2020 International Tobacco Control
Four Country Smoking and Vaping Surveys. Nicotine & Tobacco Research. 2022;25(3):505-
513. doi:10.1093/ntr/ntac212
98. Henry DB, Kobus K, Schoeny ME. Accuracy and bias in adolescents' perceptions of
friends' substance use. Psychol Addict Behav. Mar 2011;25(1):80-9. doi:10.1037/a0021874
99. Elm Ev, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP.
Strengthening the reporting of observational studies in epidemiology (STROBE) statement:
guidelines for reporting observational studies. BMJ. 2007-10-18 22:01:04 2007;335:806-808.
100. Valente TW, Piombo SE, Edwards KM, Waterman EA, Banyard VL. Social Network
Influences on Adolescent E-cigarette Use. Substance Use & Misuse. 2023:1-7.
doi:10.1080/10826084.2023.2188429
101. Rosseel Y. lavaan: An R Package for Structural Equation Modeling. Journal of Statistical
Software. 05/24 2012;48(2):1 - 36. doi:10.18637/jss.v048.i02
102. van Buuren S, Groothuis-Oudshoorn K. mice: Multivariate Imputation by Chained
Equations in R. Journal of Statistical Software. 12/12 2011;45(3):1 - 67.
doi:10.18637/jss.v045.i03

109
103. Lerman K, Yan X, Wu XZ. The "Majority Illusion" in Social Networks. PLoS One.
2016;11(2):e0147617. doi:10.1371/journal.pone.0147617
104. California Prohibits Retailers from Selling Flavored Tobacco Products. California
Department of Public Health. 2023.
https://www.cdph.ca.gov/Programs/CCDPHP/DCDIC/CTCB/Pages/CAFlavorTobaccoLaw.aspx
#
105. Ragan DT, Osgood DW, Ramirez NG, Moody J, Gest S. A Comparison of Peer Influence
Estimates from SIENA Stochastic Actor–based Models and from Conventional Regression
Approaches. Sociological Methods & Research. 2019;
106. Zheng X, Li W, Wong SW, Lin HC. Social media and E-cigarette use among US youth:
Longitudinal evidence on the role of online advertisement exposure and risk perception. Addict
Behav. Aug 2021;119:106916. doi:10.1016/j.addbeh.2021.106916
107. Zheng X, Lin HC. How Does Online e-cigarette Advertisement Promote Youth's ecigarettes Use? The Mediating Roles of Social Norm and Risk Perceptions. Health Commun. Jun
2023;38(7):1388-1394. doi:10.1080/10410236.2021.2010350
108. Vassey J, Valente T, Barker J, et al. E-cigarette brands and social media influencers on
Instagram: a social network analysis. Tob Control. Aug 2023;32(e2):e184-e191.
doi:10.1136/tobaccocontrol-2021-057053
109. Vassey J, Galimov A, Kennedy CJ, Vogel EA, Unger JB. Frequency of social media use
and exposure to tobacco or nicotine-related content in association with E-cigarette use among
youth: A cross-sectional and longitudinal survey analysis. Prev Med Rep. Dec 2022;30:102055.
doi:10.1016/j.pmedr.2022.102055
110. Berndt TJ. Developmental changes in conformity to peers and parents. Developmental
psychology. 1979;15(6):608.
111. Montgomery SC, Donnelly M, Bhatnagar P, Carlin A, Kee F, Hunter RF. Peer social
network processes and adolescent health behaviors: A systematic review. Prev Med. Jan
2020;130:105900. doi:10.1016/j.ypmed.2019.105900
112. Ripley RM, Snijders TAB, Boda Z, Vörös A, Preciado P. Manual for SIENA version 4.0.
University of Oxford, Department of Statistics; Nuffield College.: Oxford; 2023.
113. Ripley RM, Snijders TAB, Boda Z, Vörös A, Preciado P. Manual for SIENA version 4.0.
R Package version 1301. University of Oxford, Department of Statistics; Nuffield College.:
Oxford; 2021.
114. Young LE, Fujimoto K. The Co-evolution of online social networks and syphilis
incidence among young black men who have sex with men. Social Science & Medicine.
2021/03/01/ 2021;272:113764. doi:https://doi.org/10.1016/j.socscimed.2021.113764

110
115. Greenan CC. Diffusion of Innovations in Dynamic Networks. Journal of the Royal
Statistical Society Series A: Statistics in Society. 2014;178(1):147-166. doi:10.1111/rssa.12054
116. Snijders TAB. The Multiple Flavours of Multilevel Issues for Networks. In: Lazega E,
Snijders TAB, eds. Multilevel Network Analysis for the Social Sciences: Theory Methods and
Applications. Springer; 2016:chap 2.
117. Snijders TA, Baerveldt C. A multilevel network study of the effects of delinquent
behavior on friendship evolution. Journal of mathematical sociology. 2003;27(2-3):123-151.
118. Henneberger AK, Mushonga DR, Preston AM. Peer Influence and Adolescent Substance
Use: A Systematic Review of Dynamic Social Network Research. Adolescent Research Review.
2021/03/01 2021;6(1):57-73. doi:10.1007/s40894-019-00130-0
119. Valente TW. Network interventions. Science. Jul 6 2012;337(6090):49-53.
doi:10.1126/science.1217330
120. Rogers EM. Diffusion of innovations. 5th ed., Free Press trade pbk. ed. ed. Free Press;
2003.
121. Valente TW, Rogers EM. The origins and development of the diffusion of innovations
paradigm as an example of scientific growth. Science communication. 1995;16(3):242-273.
122. Watts DJ, Strogatz SH. Collective dynamics of ‘small-world’networks. nature.
1998;393(6684):440-442.
123. Valente TW, Davis RL. Accelerating the diffusion of innovations using opinion leaders.
The Annals of the American Academy of Political and Social Science. 1999;566(1):55-67.
124. Valente TW. Social network thresholds in the diffusion of innovations. Social networks.
1996;18(1):69-89.
125. Valente TW, Vega Yon GG. Diffusion/Contagion Processes on Social Networks. Health
Educ Behav. Apr 2020;47(2):235-248. doi:10.1177/1090198120901497
126. van Woudenberg TJ, Simoski B, Fernandes de Mello Araújo E, et al. Identifying
Influence Agents That Promote Physical Activity Through the Simulation of Social Network
Interventions: Agent-Based Modeling Study. Original Paper. J Med Internet Res.
2019;21(8):e12914. doi:10.2196/12914
127. McMillan C, Schaefer DR. Comparing targeting strategies for network-based adolescent
drinking interventions: A simulation approach. Soc Sci Med. Aug 2021;282:114136.
doi:10.1016/j.socscimed.2021.114136
128. Sedlander E, Rimal RN. Beyond Individual-Level Theorizing in Social Norms Research:
How Collective Norms and Media Access Affect Adolescents' Use of Contraception. J Adolesc
Health. Apr 2019;64(4s):S31-s36. doi:10.1016/j.jadohealth.2018.12.020

111
129. Padon AA, Rimal RN, Jernigan D, Siegel M, DeJong W. Tapping Into Motivations for
Drinking Among Youth: Normative Beliefs About Alcohol Use Among Underage Drinkers in
the United States. J Health Commun. Oct 2016;21(10):1079-87.
doi:10.1080/10810730.2016.1222030
130. Aral S, Muchnik LEV, Sundararajan A. Engineering social contagions: Optimal network
seeding in the presence of homophily. Network Science. 2013;1(2):125-153.
doi:10.1017/nws.2013.6
131. Badham J, Kee F, Hunter RF. Simulating network intervention strategies: Implications
for adoption of behaviour. Network Science. 2018;6(2):265-280. doi:10.1017/nws.2018.4
132. Pons P, Latapy M. Computing communities in large networks using random walks.
Springer; 2005:284-293.
133. Freeman LC. Centrality in social networks conceptual clarification. Social Networks.
1978/01/01/ 1978;1(3):215-239. doi:https://doi.org/10.1016/0378-8733(78)90021-7
134. Piombo SE, Barrington-Trimis J, Valente T, W. . Impact of social networks and norms on
e-cigarette use among adolescents in Southern California: a prospective cohort study. BMJ
Public Health. 2023;1(1):e000163. doi:10.1136/bmjph-2023-000163
135. netdiffuseR: Analysis of Diffusion and Contagion Processes on Networks. Version R
packge version 1.22.2. 2021. https://github.com/USCCANA/netdiffuseR
136. Rimal RN, Real K. How behaviors are influenced by perceived norms: A test of the
theory of normative social behavior. Communication research. 2005;32(3):389-414.
137. Iyengar R, Van den Bulte C, Eichert J, West B. How social network and opinion leaders
affect the adoption of new products. NIM Marketing Intelligence Review. 2011;3(1):16-25.
138. Valente TW, Coronges K, Lakon C, Costenbader E. How Correlated Are Network
Centrality Measures? Connect (Tor). Jan 1 2008;28(1):16-26.
139. Ragan DT, Osgood DW, Ramirez NG, Moody J, Gest SD. A comparison of peer
influence estimates from SIENA stochastic actor–based models and from conventional
regression approaches. Sociological Methods & Research. 2022;51(1):357-395.

112
APPENDIX A
Supplemental Table 1. Imputed Model of Past 6-month Vaping as a Function
of Demographic Characteristics, Network Exposure to Vaping, and Peer
Vaping Norms (N = 2,154)
Fixed Effects AOR 95% CI
Female 1.29 (0.79-2.12)
Hispanic 1.32 (0.8-2.17)
Sexual minority^ 1.52 (0.93-2.5)
Age 1.53* (1-2.33)
Past 6-month Vaping
Wave 1 7.65*** (3.03-19.3)
Wave 2 10.7*** (6.17-18.5)
Network exposure to vaping
Wave 1 1.34 (0.48-3.75)
Wave 2 2.73* (1.06-7.05)
Peer vaping norms
Wave 1 1.52 (0.87-2.66)
Wave 2 1.78* (1.07-2.96)
Random Effects Variance
School 0.15
^Sexual minority, questioning, prefer not to disclose collapsed for analysis with heterosexual as
reference
AOR = adjusted odds ratio for mixed logistic regression; CI = Confidence interval
* p<0.05; ** p<0.01; *** p<0.001

113
Supplemental Table 2. Demographic Characteristics of Participants Lost to Follow Up
Wave 2
(N = 140)
Wave 3
(N = 212)
N (%)
Sex Assigned at Birth
Male 58 (41.4) 90 (42.5)
Female 73 (52.1) 113 (53.3)
Missing 9 (6.0) 9 (4.3)
Sexual Orientation^
Heterosexual 103 (73.6) 173 (81.6)*
Missing 7 (5.0) 8 (3.8)
Hispanic 69 (49.3) 142 (67.0)*
Missing 9 (6.4) 19 (9.0)
Past 6-month E-cig Use Prevalence
Wave 1 6 (4.3) 4 (1.9)
Wave 2 NA 18 (8.5)
Mean (SD)
Baseline Age (years) 14.4 (0.53) 14.7 (0.66)
^Sexual minority, questioning, prefer not to disclose collapsed for analysis with heterosexual as reference
*Heterosexual sexual orientation and Hispanic identity were both associated with being lost to follow up from
wave 2 to wave 3

114
Supplemental Table 3. School Friendship Network Characteristics
Wave 1
School Size Edges Density Reciprocity Transitivity APL
101 169 429 0.02 0.59 0.38 6.42
102 58 80 0.02 0.48 0.23 3.91
103 131 228 0.01 0.53 0.25 7.63
104 222 597 0.01 0.54 0.29 7.39
105 80 134 0.02 0.54 0.27 3.84
106 93 256 0.03 0.50 0.32 2.91
112 255 1616 0.02 0.57 0.29 6.10
113 279 1666 0.02 0.54 0.32 6.38
114 223 1032 0.02 0.38 0.23 6.51
Wave 2
School Size Edges Density Reciprocity Transitivity APL
101 212 595 0.01 0.55 0.25 6.14
102 113 169 0.01 0.35 0.24 2.84
103 117 252 0.02 0.56 0.26 6.78
104 282 870 0.01 0.51 0.28 6.75
105 101 240 0.02 0.52 0.29 6.51
106 140 269 0.01 0.41 0.22 6.85
112 299 1045 0.01 0.53 0.30 5.95
113 331 1144 0.01 0.54 0.31 7.08
114 296 917 0.01 0.48 0.21 6.31
Wave 3
School Size Edges Density Reciprocity Transitivity APL
101 208 632 0.01 0.56 0.30 6.03
102 126 219 0.01 0.52 0.34 3.45
103 150 300 0.01 0.48 0.22 6.34
104 291 902 0.01 0.46 0.25 6.26
105 110 255 0.02 0.51 0.25 7.28
106 142 281 0.01 0.42 0.22 6.63
112 276 882 0.01 0.56 0.27 6.34
113 332 1186 0.01 0.54 0.25 5.94
114 292 931 0.01 0.50 0.24 5.89

115
Supplemental Table 4 Meta-analysis of changes in friendship networks and e-cigarette use in nine
Southern California high schools (N=2,245)
Effect N T^2 Mu-hat (s.e.) sigma Q
Q pvalue
Network Dynamics
Friend rate (period 1) 7 517.59*** 16.71 (1.39)*** 2.87 18.66 0.01
Friend rate (period 2) 9 1507.04*** 6.7 (0.66)*** 1.89 120.72 0.00
Out-degree (density) 9 3282.16*** -3.41 (0.13)*** 0.30 26.66 0.00
Reciprocity 9 6859.11*** 2.9 (0.05)*** 0.11 23.11 0.00
GWESP 9 3435.35*** 1.73 (0.04)*** 0.05 14.88 0.06
Popularity (in-degree square root) 9 81.80*** 0.19 (0.04)** 0.09 22.16 0.01
Popularity (Out-degree square root) 9 260.66*** -0.35 (0.04)*** 0.09 22.67 0.00
Activity (in-degree square root) 9 104.06*** -0.13 (0.08) 0.20 44.93 0.00
Male
Alter 9 12.24 0.05 (0.02)* 0.00 7.05 0.53
Ego 9 20.64** 0.06 (0.03) 0.06 15.34 0.05
Same 9 694.58*** 0.5 (0.03)*** 0.06 19.58 0.01
Hispanic
Alter 9 20.96** 0.03 (0.04) 0.06 18.93 0.02
Ego 9 22.35** 0.07 (0.04) 0.05 14.37 0.07
Same 9 235.80*** 0.35 (0.06)*** 0.14 47.59 0.00
Behavior Dynamics
Rate (period 1) 9 12.61 0.03 (0.01)** 0.00 2.31 0.97
Rate (period 2) 9 10.93 0.02 (0.004)** 0.00 3.88 0.87
Rate effects 3.15 (0.50)**
Average exposure 6 12.01* -0.66 (0.17)** 0.00 1.24 0.94
Male 8 11.77 0.76 (0.24)* 0.00 3.67 0.82
Hispanic 8 18.95** 0.29 (0.07)** 0.00 6.98 0.43
Pro-vaping norms 9 32.42*** 16.71 (1.39)*** 0.00 8.79 0.36
*p < 0.05; **p < 0.01; ***p < 0.001
N = number of schools on which the statistic for this effect were based;
Estimate (mu hat) = estimated average effect size; standard error (s.e.);
sigma = estimated true between-schools standard deviation of the effect size;
T ^2 = statistic for testing that total effect is zero;
Q = statistic for testing that true effect variance is zero, and p-value for associated test

116
Supplemental Table 5. Logistic regression of individual e-cigarette use on demographic
covariates, pro-e-cigarette norms, friend selection, previous network exposure and prior ecigarette use.
Wave 2
(N = 1595)
Wave 3
(N = 1909)
Effect Estimate (S.E.) AOR Estimate (S.E.) AOR
Male -0.73 (0.31)* 0.48 -0.44 (0.3) 0.64
Hispanic 1.3 (0.3)*** 3.68 0.31 (0.29) 1.37
E-cig use at previous wave 7.58 (1.03)*** 1954.41 8.71 (1.03)*** 6071.14
Network exposure at previous wave 1.88 (0.58)** 6.54 1.63 (0.56)** 5.12
Perceived Pro-e-cig norms 0.27 (0.08)*** 1.31 0.35 (0.08)*** 1.42
New e-cig user friends (selection) 0.16 (0.15) 1.18 0.48 (0.19)* 1.62
*p < 0.05; **p < 0.01; ***p < 0.001

117
Supplemental Figure 1. Pro-e-cigarette social norm scores from ADVANCE data wave 1 (top)
and wave 2 (bottom). Note: social norms not measured wave 3.

Abstract (if available)

Abstract Electronic cigarettes (e-cigarettes/e-cigs) have rapidly increased in popularity and become the most commonly used tobacco product among adolescents over the last ten years. E-cigarettes are detrimental to health and are correlated with initiation of combustible cigarette use among adolescents and young adults. Preventing e-cigarette use by focusing on changing adolescent behaviors can be difficult due to the complex influences of inter- and intra-personal factors. E-cigarette use is driven by a combination of psychological, social, environmental, and systemic factors. Social influence can occur through our social networks, or the relationships and connections to others that can shape thoughts, attitudes, and behaviors. During adolescence, the importance of friendships, social connections, and the desire to fit in may leave adolescents more susceptible to the social influence of friends.

The dissertation studies explore the relationship between perceived social norms, peer e-cigarette use, and their joint influence on individual e-cigarette use through the analysis of social network data from a diverse cohort of over 2,000 adolescents. The primary focus of these studies is identifying and differentiating between the effects of social influence mechanisms driving e-cigarette use in adolescent social networks. These studies take a rigorous approach using different social network analytic methods to examine these relationships. The results of these studies can be used to inform and design social network interventions to address normative perceptions surrounding e-cigarettes with the goal of ultimately decreasing use among adolescents.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Evaluating social networks and impact of micro-influencers who promote e-cigarettes on social media

PDF

A network analysis of online and offline social influence processes in relation to adolescent smoking and alcohol use

PDF

Social network influences on depressive symptoms among Chinese adolescents

PDF

Contextualizing social network influences on substance use among high risk adolescents

PDF

Adolescent social networks, smoking, and loneliness

PDF

Anxiety symptoms and nicotine use among adolescents and young adults

PDF

Effects of flavorings in electronic cigarettes on the use and appeal of e-cigarettes among youth and adults

PDF

Investigating factors that influence peer relationships and obesity during middle childhood

PDF

Exploring the role of peer influence, linguistic acculturation, and social networks in substance use

PDF

The role of depression symptoms on social information processing and tobacco use among adolescents

PDF

Social selection and influence of alcohol & marijuana implicit cognitions and behaviors: a longitudinal investigation of peer social network dynamics

PDF

Diffusion of a peer‐led suicide preventive intervention in secondary schools: strategies to increase effectiveness of peer‐led interventions

PDF

Friendship network position on adolescent behaviors: an examination of a broker position and the likelihood of alcohol and cigarette use

PDF

The role of social support in the relationship between adverse childhood experiences and addictive behaviors across adolescence and young adulthood

PDF

Examining the built and online environments to understand cannabis and e-cigarette availability, marketing, and product use in Southern California

PDF

The effects of mindfulness on adolescent cigarette smoking: Measurement, mechanisms, and theory

PDF

Opioid withdrawal symptoms, opioid use, and injection risk behaviors among people who inject drugs (PWID)

PDF

Social self-control and adolescent substance use

PDF

Distress tolerance and mindfulness disposition: associations with substance use during adolescence and emerging adulthood

PDF

Adolescent cigarette use and psychological distress: relation to adult obesity risk

Asset Metadata

Creator Piombo, Sarah Elizabeth (author)

Core Title Normative and network influences on electronic cigarette use among adolescents

School Keck School of Medicine

Degree Doctor of Philosophy

Degree Program Health Behavior Research

Degree Conferral Date 2024-08

Publication Date 01/15/2025

Defense Date 04/10/2024

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

Tag adolescent health,e-cigarettes,OAI-PMH Harvest,Public Health,social influence,social network analysis,tobacco

Format theses (aat)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Valente, Thomas (committee chair), Barrington-Trimis, Jessica (committee member), de la Haye, Kayla (committee member), Miller, Kimberly (committee member), Young, Lindsay (committee member)

Creator Email piombo@usc.edu,sarahpiombo13@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC113997P18

Unique identifier UC113997P18

Identifier etd-PiomboSara-13238.pdf (filename)

Legacy Identifier etd-PiomboSara-13238

Document Type Dissertation

Format theses (aat)

Rights Piombo, Sarah Elizabeth

Internet Media Type application/pdf

Type texts

Source 20240716-usctheses-batch-1183 (batch), University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright.

Repository Name University of Southern California Digital Library

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

Repository Email cisadmin@lib.usc.edu