University of Southern California Dissertations and Theses

The power of social media narratives in raising mental health awareness for anti-stigma campaigns

(USC Thesis Other)

The power of social media narratives in raising mental health awareness for anti-stigma campaigns

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

THE POWER OF SOCIAL MEDIA NARRATIVES IN RAISING MENTAL HEALTH
AWARENESS FOR ANTI-STIGMA CAMPAIGNS

by

Hye Min Kim

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
(COMMUNICATION)

August 2022

Copyright 2022 Hye Min Kim

ii
DEDICATION
In loving memory to my grandmother, Doo-Eui Park (1925-2006).

iii
ACKNOWLEDGEMENTS

I would like to express my heartfelt gratitude to a number of people who supported me in
countless ways throughout my academic life. It certainly took a village and I owe many thanks to
them who nurtured me, encouraged me, and trusted my potential. First, I am deeply indebted to
my advisor and mentor Dr. Margaret McLaughlin for her unwavering support and guidance from
the day one. I vividly remember the first time I met her as my first-year mentor when I started
my PhD study at USC Annenberg. She carefully listened to my interest and goal and invited me
to join her weekly meetings, which was the very beginning of my scholarly journey in the field
of health communication and social media. Peggy taught me the importance of rigorousness in
research and perseverance in pursuing academic career through the numerous meetings we had at
her office and over Skype/Zoom. I am grateful for her allowing me to explore my topical and
methodological interests from in and outside of the Annenberg over the years. Such flexibility
and autonomy have been the source of my academic growth and independence, which will
undoubtedly serve me well in my next chapter as a faculty member.
It is hard to imagine completing my PhD program without my two other committee
members, Dr. Lynn Miller and Dr. Sheila Murphy. I was very fortunate to have received close
mentorship from Dr. Lynn Miller while we collaborated on multiple projects since the early days
of my PhD program. I learned from her what constitutes a good question as well as a good
academic writing, and such thoroughness will continue to guide me in my scholarly pursuits. I
can never forget her kindness and generosity to spend time with me going over my conference
slides at hotel lobby in Prague to help me better prepare for the ICA presentation. I will treasure
that memory forever. I would also like to express my sincere thanks to Dr. Sheila Murphy whose
iv
work greatly inspired my research interest in using social media narratives for health campaigns.
Her valuable comments and suggestions on the theoretical framework and research design
tremendously helped me consolidate and turn my half-baked ideas into research papers as well as
this dissertation project.
My special appreciation goes to other mentors from and beyond the USC Annenberg who
have encouraged me to pursue what I truly love and care. I am deeply grateful to Dr. David
Jeong for his enormous help and support in navigating my PhD journey as well as my job search
process during the pandemic. I would also like to thank Dr. Aimei Yang for her patience and
guidance. I was lucky to have her as my qualifying exam committee and develop my passion for
computational methods while working closely with her. Having regular meetings with her in the
year of my job market while completing my dissertation when things were all transferred to
online due to the global pandemic greatly helped me keep sane and cross the finish line. I truly
appreciate fond memories of that virtual happy birthday song by the team Aimei on my birthday
and her innumerable crossing lucky fingers for me. I am also indefinitely grateful to Dr. Kyu Ho
Youm for his continuous support and mentorship along the way since our fortuitous encounter at
his talk in 2014. Finally, I would like to thank Dr. Eunjin Kim, Dr. Su Jung Kim, Dr. Wenlin
Liu, Dr. Daphna Oyserman, Dr. Jillian Pierson, Dr. Jieun Shin, and Dr. Robin Stevens for their
time and invaluable advice whenever I needed help in my personal or professional endeavors.
This dissertation would not have been possible without the help and support from many
other individuals. I want to express my gratitude to the staff at USC Annenberg, especially Sarah
Holterman and Anne Marie, for their patience and kindness in always willing to help me with so
many questions. Knowing that they are on the lookout for students always reassured me that I am
part of such a supportive community. I would also like to thank my brilliant colleagues and
v
friends at Annenberg, Jeeyun Sophia Baik, Sukyoung Choi, Brianna Ellerbe, Joowha Hong,
Eugene Jang, Do Own Kim, Hyun Tae Kim, Steffie Kim, Eugene Lee, Yiqi Li, Lauren Sowa,
Jingyi Sun, Nathan Walter, Larry Zhimming Xu, and Yu Xu and my academic sisters, Jillian
Kwong, Mina Park, Yao Sun, Yunwen Wang, and Aveva Yusi Xu for their wonderful
companionship. My thanks extend to my dear friends, Youjin Choi, Hansol Choi, Jiwoo Han,
Soyeong Kim, Taehyung Kim, Jiwon Koo, Heejo Lee, Yoonkyung Kim, and Amber and Teddy
Zeng, for their priceless friendship and emotional support over the years despite the physical
distance.
Words cannot express enough my gratitude toward my parents for such unconditional
love and endless faith in which I could learn how to grow and explore my own ways in freedom.
I thank my mom and dad for the best gift they gave me, my younger sister and brother, and all
the good memories we cherish together since our childhood, which has been and will be the
greatest source of my strength in my life. And finally, I am immensely grateful to my husband,
Jihoon Sohn, for his love, understanding, patience, and humor. I certainly could not have made it
through this journey without you. Thank you for feeding me, taking me out for a walk, and
making my days full of happiness and laughter. The best thing I have ever done in my life is
marrying you.

vi
TABLE OF CONTENTS
DEDICATION ii
ACKNOWLEDGEMENTS iii
LIST OF TABLES viii
LIST OF FIGURES ix
ABSTRACT x
CHAPTER 1: INTRODUCTION 1
Communication Efforts for Mental Health Promotion 3
Mental health literacy 4
Stigma 5
Proposed Studies 6

CHAPTER 2: POWER OF PERSONAL STORIES 7
Theoretical Framework 8

CHAPTER 3: SOCIAL MEDIA AS A PUBLIC HEALTH CAMPAIGN TOOL 11
Story Sharing as Social Media Health Campaign Outcome 12

CHAPTER 4: AN EXPERIMENTAL STUDY OF SOCIAL MEDIA HEALTH STORY 15
Introduction 15
Explanatory Mechanism: Social Presence and Transportation 17
Parasocial Intimacy with Public Figures 19
Interpersonal Outcome of Social Media Personal Stories Self-Referencing Engagement 20
Method 22
Pilot Study 22
Procedure 23
Participants 25
Measures 25
Results 28
Preliminary Analysis 28
Hypotheses Testing 29
Discussion 38

CHAPTER 5: COMPUTATIONAL APPROACH TO SOCIAL MEDIA CAMPAIGNS 43
Social Media Health Campaign Effectiveness 43
Power of Personal Stories in Communication Network 44
Power of Affective Stories in Communication Network 47
Homophily Effects of Message Engagement in Communication Network 48
Method 49
Study Background 49
vii
Data Collection 50
Network Construction 51
Analytical Procedures 52
Measures 53
Results 54
Discussion 57

CHAPTER 6: CONCLUSION 59

REFERENCES 62

APPENDIX 80

viii
LIST OF TABLES
Table 4.1. Bivariate Correlations among Variables 29

ix
LIST OF FIGURES
Figure 4.1. Attitudes toward help-seeking by different media conditions for personal stories 30
Figure 4.2. Behavioral intentions of seeking professional help by different media conditions 31
Figure 4.3. Behavioral intentions of seeking informal help by different media conditions 31
Figure 4.4. Serial multiple mediation model with attitudes toward help-seeking as an outcome 33
Figure 4.5. Serial multiple mediation model with seeking professional help as an outcome 34
Figure 4.6. Serial multiple mediation model with seeking informal help as an outcome 35
Figure 4.7. Interaction effects on help-seeking attitudes 36
Figure 4.8. Interaction effects on self-referencing engagement 38
Figure 5.1. Tweet counts in the mental health week of 2018 51
Figure 5.2. Quote network 52
Figure 5.3. MCMC-based degeneracy diagnostics 56
Figure 5.4. Goodness-of-fit plots for ERGM 56

x
ABSTRACT

This dissertation project studies the effects of personal health stories on social media in
raising mental health awareness and reducing the stigma. Mental health problems or mental
disorders are one of the increasingly imminent public health issues in recent years. Promoting
mental health care with anti-stigmatizing messages is therefore an important health
communication goal to achieve health equity in the access to needed information, care, and
services (CDC, 2021). Drawing from the communication theories of narrative and media
affordances while taking a combination of experimental and computational methodologies, this
project embarks on unpacking what makes the use of personal stories, especially through social
media as a communication channel, an effective health communication strategy in normalizing
the issue of mental health and seeking professional help for mental health care. In doing so, 2
studies were proposed: 1) an experimental study that tests the effectiveness of personal stories in
different media formats on both persuasion and engagement outcomes, more specifically on
reducing the stigma, increasing help-seeking behavior, and eliciting relevant conversation, and 2)
a computational study that takes a naturalistic setting to further examine the engagement effects
of social media personal stories. Findings from the proposed studies provide actionable insights
for health organizations and agencies to use storytelling through social media for mental health
campaigns that are designed to increase public awareness as well as to promote favorable
attitudes and behavior about mental health care and services.
1
CHAPTER 1: INTRODUCTION

“The most personal is the most creative”
– Bong Joon Ho, Oscars, 2020 quoting Martin Scorsese

Storytelling and experiential messages have been widely used as effective
communication strategies for delivering health messages (Barbour, Doshi, & Herná ndez, 2016;
Kim, Bigman, Leader, Lerman, & Cappella, 2012; Dunlop, Wakefield, & Kashima, 2010; Kreps,
2017). A growing literature has shown that individuals’ stories of their direct experience with
health struggles and demonstrated success in coping with them facilitate readers’ message
processing and changes in beliefs, attitudes, and behavioral intentions (Quintero Johnson,
Yilmaz, & Najarian, 2017). Moreover, digital communication technologies have transformed the
way people access and share information and messages (Kreps, 2017), expanding the
opportunities for users to narrate their own personal experience including their health issues
(McLaughlin et al., 2012; Merolli, Gray, & Martin-Sanchez, 2013; Pereira, Quinn, & Morales,
2016) as well as to interact and engage with such stories.
What is unique about using social media narratives is that they involve both mass and
interpersonal communication (Walther, 1996), meaning that the messages are disseminated to
mass audience, while users can engage with the messages as well as interact with other users
(Neiger et al., 2012). In light of Rogers (2002)’s emphasis that effective message diffusion
process comes from both mass and interpersonal communication channels, the combination of
social media’s mass and interpersonal features may make health campaigns more effective by
facilitating the message delivery to broad audience members and their interpersonal engagement
2
with the message or among themselves. Moreover, such a media convergence of mass and
interpersonal communication further adds communication intimacy (Walther, 1996) to the health
campaign context. Indeed, previous research suggested a great potential of using social media,
especially for mental health promotion, is that it offers an inexpensive way for users to have
conversation about mental health and stigma (Link & Phelan, 2001) while exchanging relevant
information or sharing stories (e.g., Betton et al., 2015).
A major goal of mental health promotion is to normalize open conversation about mental
health – that is, it is just another type of health problem that is treatable where professional help
and resources are available and accessible. Therefore, many of mental health campaigns aim to
raise the awareness of mental health care and services and foster positive attitudes and beliefs
about them. Against this backdrop, personal stories about mental health on social media and their
interpersonal effects of instigating related conversation may be strategically used for mental
health promotion. Yet, little is known about how the effects of such stories with lived experience
through social media work. Despite empirical findings and advances in the area of mediated
communication and of narrative, more work is needed to fully understand how the two aspects of
communication, message and media, combine, not in isolation, to operate for effective social
media health campaigns.
The present research has two broad aims: first (1) to advance our understanding of the
mechanisms underpinning the narrative’s effects on social media for health campaigns (e.g.,
enhanced communication intimacy with the speaker of personal health story), and second, (2) to
advance the conceptual and methodological approach to the evaluation of social media campaign
effectiveness (e.g., engagement and persuasion outcomes). In the following chapters, I first
elaborate what the problem is: that is, the challenge of mental health stigma that underlies the
3
reluctance to seek help. Then I introduce and discuss the promise of using social media as an
effective public health communication channel in delivering personal health stories to promote
mental health literacy and reduce the stigma. And finally, I empirically test the explanatory
mechanism of social media personal stories’ effectiveness in terms of persuasion and
engagement outcomes.
Communication Efforts for Mental Health Promotion
Mental health is an integral and essential component of one’s health. It is more than just
the absence of mental disorders but a state of well-being where one can cope with daily stress of
life and work productively (World Health Organization, 2022). There are many different types of
mental disorders such as depression, anxiety disorder, and various eating disorders, which would
have a considerable impact on one’s mood, cognitions, and daily activities such as sleeping, food
intake, and working (National Institute of Mental Health, 2016). What is concerning, however, is
the lack of mental health care and services. It was further exacerbated by the recent COVID-19
pandemic that has disrupted the worldwide use of mental health care and services (World Health
Organization, 2020). For instance, more than 75% of people with depression especially from the
socio-economically disadvantaged countries received no treatment at all (World Health
Organization, 2021).
Prior studies have identified that primary reasons for the paucity of help-seeking behavior
are negative stereotypes or social stigma about seeking professional help for mental health
(Boysen & Vogel, 2008; Corrigan, 2004; Corrigan, Morris, Michaels, Rafacz, & Rüsch, 2012)
and the low level of mental health literacy (Jorm, 2012; Swami, 2012). Mental health literacy is
defined as “knowledge and beliefs about mental disorders which aid their recognition,
4
management, or prevention” (Jorm et al., 1997, p. 182). Increasing the public’s mental health
literacy is particularly difficult given the prevailing fear of being judged and the social stigma
associated with discussing the topic of mental health in general (e.g., Ritsher & Phelan, 2004).
As a result, the barrier to accessing information and services/care needed for mental health is
even greater than the barriers to physical health care (Henderson, Evans-Lacko, & Thornicroft,
2013). Against this backdrop, an increasing number of scholars and experts are calling for more
theory-driven research on strategic anti-stigma mental health message design to promote timely
help-seeking behaviors (Ritsher & Phelan, 2004).
Mental Health Literacy
Symptoms of mental health problems may seem neither immediate nor observable
compared to physical health symptoms, and thus people may tend to attribute such symptoms to
something else such as chronic fatigue and a stressful day. Therefore, recognizing one’s
struggles as indicative of a mental health problem and perceiving the need for care are the first
steps towards help-seeking behavior (Villatoro, Mays, Ponce, & Aneshensel, 2018). Such
knowledge and beliefs about mental health problems and prevention is referred to as mental
health literacy (MHL), which includes the ability to recognize the symptoms and to know the
risk factors, causes, and how to seek appropriate information as well as professional help and
treatments (Jorm et al., 1997; Jorm, 2012; Swami, 2012). That is, it is not simply a matter of
having knowledge, but specifically requires the ability to link the knowledge to the possibility of
action to benefit one’s own mental health or that of others. Therefore, MHL involves both
knowledge and efficacy.
5
Increasing MHL matters both at the individual level and at the public level. At the
individual level, failure to identify what one is going through as symptoms of mental health
problems prohibits the subsequent steps towards help-seeking behaviors (Jorm, 2012; Jorm et al.,
1997; Swami, 2012). This process of recognizing the problem and needs for help is referred to as
self-labeling, representing a pivotal antecedent that drives help-seeking (Thoits, 1985). Self-
labeling processes entail how one makes sense of what he or she is going through, how it can be
labeled as a problem that needs treatment, and then what needs to be done. On the other hand, at
the broader public level, limited understanding of mental health care further exacerbates the
tendency to stigmatize those with mental health problems (Pan, Liu, & Kreps, 2018).
Stigma
Attitudes toward people with mental health problems are associated with attitudes toward
seeking professional help for mental health care and treatment, thus such stigmatizing attitudes
impede help-seeking behavior. In many cases, people with mental health problems do not report
or even recognize their symptoms (i.e., self-labeling), partially due to the negative stereotypes
and social stigma toward people with mental health problems (Parcesepe & Cabassa, 2013) – for
instance, being weak, inferior, incompetent, vulnerable, shameful, or socially unacceptable
(Swami, 2012). Such stereotyped attitudes and prejudices are the very obstacles that keep people
from seeking mental health treatment (Corrigan et al., 2012). Indeed, besides the lack of
knowledge about available service and treatments, concerns about stigma have been found to be
the key barrier to perceived need for help and treatment-seeking behavior (Hunt & Eisenberg,
2010). Therefore, reducing stigma constitutes an important agenda for health communication
research to normalize mental health issue and promote help-seeking behavior.
6
Proposed Studies
This dissertation aims to investigate what makes social media an effective public health
communication tool for raising mental health awareness as well as promoting help-seeking
behavior and to explain the anti-stigmatizing effects of social media personal stories. To do so,
two studies were proposed here: 1) an experimental study that tests the effectiveness of personal
stories in different media formats on both persuasion and engagement outcomes, more
specifically on reducing the stigma, increasing help-seeking behavior, and eliciting relevant
conversation, and 2) a computational study that takes a naturalistic setting to further examine the
engagement effects of social media personal stories.

7
CHAPTER 2: POWER OF PERSONAL STORIES

A growing body of research points to the effectiveness of using narrative as a health
communication tool. One of the commonly used methods of narrative in health communication is
personal narrative or testimonial, anecdotal evidence that uses storytelling to depict individual
experience, events, and consequences (Slater & Rouner, 2002). In a typical personal narrative, a
main character tells a story about his or her successful experience that directly or indirectly
encourages the audience to follow his or her example (Braverman, 2008). This is different from
an alternative technique that presents factual information with expository reports of events,
professional opinions, or statistical evidence. To compare these two communication techniques
for their relative effectiveness, McQueen, Kreuter, Kalesan, and Alcaraz (2011) examined and
showed that the longitudinal effects of personal narratives about mammography and breast
cancer were greater than content-equivalent informational messages. Similarly, Kreuter and
colleagues demonstrated the cancer survivors’ personal narratives increased mammography
uptake among African-American women (Kreuter et al., 2007; Kreuter et al., 2008). Specifically,
readers of the survivors’ stories showed greater social and emotional support, modeled their
coping skills, and were more likely to share information and resources with others (Kreuter et al.,
2008). To this end, narrative communication, especially personal stories are seen as offering
unique advantages over traditional expository communication in the context of promoting
desirable health behaviors.

8
Theoretical Framework
The effectiveness of using personal narrative has long been recognized by scholars not
only in the field of health communication but also in news framing studies that feature thematic
versus episodic frames (Scheufele, 1999). Thematic frames emphasize broader trends or social
conditions of public issues, which foster a sense of shared responsibility and promote actions. By
contrast, episodic frames depict such issues with specific instances, providing little insight into
the larger social circumstances, but instead focusing on a single event or a case. Oftentimes
conceptualized as exemplification (Zillmann, 2002), the use of personal narrative or exemplar in
a news article (defined as “personal descriptions by people who are concerned or interested in an
issue” (Brosius, 1999, p. 214) has been shown to influence news audience perception and
judgement about the issue described (Kim et al., 2012).
Specifically, compared to a news article that addresses a broader social issue, featuring an
individual story in a news article was found to be more easily comprehended and recalled, evoke
greater emotional responses such as empathic feelings and compassion, and even more effective
in increasing news engagement (e.g., Zillmann, 2006). On the other hand, studies also purported
that given its focus on individual anecdote, personal narrative or episodic frame of news reports
may highlight individual causes and solutions of social problems, and thus make the public
blame the person in the story (e.g., Barry, Brescoll, & Gollust, 2013). That is, without context,
readers focus on the individuals in the stories and may find the people portrayed in the story
accountable for the problem as well as the solution (Barry et al., 2013). From this perspective,
using personal narrative to reduce stigma about mental health problems may backfire if it only
instigates the attribution of the problem to the individual featured in the story. Therefore, care
should be taken in using a personal narrative, especially in addressing a stigmatized issue.
9
Given this, emphasis has been placed on enhancing efficacy and positive aspects of the
story character (Major, 2018; Niederdeppe et al., 2011). For instance, personal narratives that
include individual efforts to lose weight were successful in increasing beliefs that society is
responsible for causing obesity in the United States (Niederdeppe et al., 2011). Another study
compared how readers of episodic and thematic frames of news coverage attribute the causes and
solutions for depression (Major, 2018). Interestingly, participants who read news article that used
an episodic frame attributed more responsibility to the society for depression. Major (2018)
explained that the participants may have seen depression as something beyond an individual’s
control because the story character of news article was described as they did everything that
could be done at an individual level – seeking counseling and taking medication. This highlights
the importance of personal agency and efficacy in using personal stories to avoid the potential of
unjust blame or trivializing the issue (e.g., Chen, McGlone, & Bell, 2015).
Narrative invites audience members into the world of its character, enabling them to
experience cognitive and emotional empathy for the character (Slater & Rouner, 2002).
Cognitive empathy transpires by taking the perspective of the character whereas emotional
empathy takes place by sharing the emotional experience of the character (Igartua & Frutos,
2017). When the story character is from a stigmatized group, audience may come to understand
what it is like to experience the described events that the character goes through by reading their
personal stories (Mar & Oatley, 2008). Such vicarious experience is what induces empathic
responses and emotional engagement (Cialdini, Brown, Lewis, Luce, & Neuberg, 1997), major
components of narrative engagement (Busselle & Bilandzic, 2009). This could be especially
useful in facilitating audiences’ perspective-taking and favorable attitudes toward stigmatized
groups, reducing their inclination to distance themselves from stigmatized characters (Chung &
10
Slater, 2013). For instance, narrative formats produced more empathetic attitudes toward
immigrants or elderly people compared to non-narrative formats through greater compassion
with the groups (Oliver, Dillard, Bae, & Tamul, 2012).
Taken together, even from the vantage point of framing theory that characterizes personal
stories as episodic frames, there is ample research evidence that recognizes them as promising
tools for reducing stigma against mental health problems via enhancing compassion and
emotional engagement as far as story characters’ efforts and agency are demonstrated and
highlighted. Based on the foregoing, now that we know that perspective-taking is key to using
personal stories for anti-stigmatizing messages, the question is what specific features of personal
stories may make it more or less effective. To address this question, in the next chapter I propose
to incorporate media-specific affordances into the study of narrative. I first explain how
perspective-taking experience can be enhanced through social media as a communication tool.
Instead of movies or a television series, this dissertation zooms in on messages that can feasibly
be encountered online for health communication campaigns, namely short stories delivered by an
affected individual (e.g., Braverman, 2008; de Wit, Das, & Vet, 2008).

11
CHAPTER 3: SOCIAL MEDIA AS A PUBLIC HEALTH
CAMPAIGN TOOL

An increasing number of health organizations are leveraging social media to disseminate
campaign messages and promote public engagement with such messages. Social media afford a
new synthesis of mass and interpersonal communication, referred to as “masspersonal
communication” (O’Sullivan & Carr, 2018). Nevertheless, not much theoretical attention has
been paid to social media’s mass personal communication feature for narrative. Social media
provide users with the opportunity to narrate their personal experience including their health
issues (Merolli et al., 2013; McLaughlin et al., 2012). Against this backdrop, this dissertation
sheds light on social media as a public health campaign tool, especially using narrative or
personal stories for its content.
Stemming from its unique potential to reach a large number of diverse audiences, social
media enable fast and broad-reaching campaign message dissemination. Many health
organizations already use social media like Twitter because of the broad reach (Park, Reber, &
Chon, 2016; Park, Rodgers, & Stemmle, 2013). Most importantly, unlike traditional mass media-
based health campaign where audiences are mere recipients of campaign messages, social media
users take a more active role by sharing, sending, commenting upon, or reposting campaign
messages (Kim, Hou, Han, & Himelboim, 2016). In other words, social media have expanded the
availability of and access to public health information and campaign messages as well as
changed the way such information is disseminated, ultimately maximizing the reach of health
campaigns (Chou, Hunt, Beckjord, Moser, & Hesse, 2009; Villagran, 2011). An example of
successful social media health campaign would be the “ice bucket challenge” that caught on a
12
few years ago. In an attempt to promote awareness of Lou Gehrig’s disease and encourage
donations for research, social media users who participated in this campaign all over the world
sequentially and publicly designated another person to participate using the hashtags of
#IceBucketChallenge. The campaign was disseminated through the participants’ networks over
the same hashtag. Social media amplified individual-led conversations and even channeled them
to mainstream media (Betton et al., 2015).
Story Sharing as Social Media Health Campaign Outcome
Facilitating interpersonal conversation in response to a campaign message constitutes a
valuable role in effective message dissemination and successful health campaigns (Rogers, 2002;
Southwell & Yzer, 2009). Social media afford users the opportunity to connect with one another
and to enhance collaboration through information sharing and engagement (Merolli et al., 2013).
Thus, on social media, an effective health campaign may initiate conversation among members
of the audience (Betton et al., 2015). For instance, people share their ideas, stories, or relevant
information in response to campaign messages that target increasing mental health awareness
and challenging stigma.
Notably, a study showed that viewers transported by a personal narrative about
lymphoma subsequently sought out more information about lymphoma as well as cancer in
general and discussed these issues with friends and family (Murphy, Frank, Moran, & Patnoe-
Woodley, 2011). More directly pertain to social media, a study examined the effects of narrative
messages (vs. non-narrative messages) about global health issues on individuals’ intentions to
engage in interpersonal communication and information sharing (Barbour et al., 2016). They
found that the narrative message encouraged message sharing interpersonally and through social
13
media because compared to non-narrative message, it was more immersive and evocative
(McQueen et al., 2011; Murphy, Frank, Chatterjee, & Baezconde-Garbanati, 2013).
Highlighting the utility of social media that enable people to share experiences and
engage in unsolicited communication, Naslund and colleagues (2016) noted their potential
opportunities for people with mental health problems to challenge stigma and to access relevant
interventions. As much as public interest and awareness of mental health problems are
encouraged through open discussion, it is essential that such public discourse ultimately leads to
help-seeking behavior. For example, people with mental health problems were motivated to seek
formal mental health care after communicating with similar others online (Powell, McCarthy, &
Eysenbach, 2003). More specifically, social connectedness was forged by interacting with peers,
sharing personal stories, and learning strategies for coping with day-to-day challenges (Naslund
et al., 2016). Such benefits of peer support include feeling more confident and empowered in
making health decisions after learning about others’ personal experiences (Entwistle et al.,
2011).
Having a mental health problem could be deemed private and personal, which requires
certain levels of intimacy to communicate about it in interpersonal context. According to the
theory of communication privacy management (CPM, Petronio & Venetis, 2017; Beck,
Aubuchon, McKenna, Ruhl, & Simmons, 2014), people think they are the owners of their
personal information, but it easily becomes no longer personal once it is disclosed and shared.
Moreover, this shared private information could have further impact on participating
communicators. For example, a study showed that sharing one’s story on social media about
personal struggles with eating disorder problem which is likely to be considered as private health
information invited other users’ supportive and positive reactions, some of which contained their
14
own struggles as well (Pereira et al., 2016). This suggests social media may make a safe and
creative way to openly talk about mental health problems, which people might otherwise keep to
themselves as personal issues. And this process happens through extending their ownership of
personal information about mental health problems to other users on social media through
engagement and participation (Smith & Brunner, 2016).
Explicating the engagement and participation aspects of social media health campaigns,
this dissertation focuses on the sharing of one’s story and experience in response to a campaign
message and its interpersonal implications for effective health campaigns using social media.
This is conceptualized as self-referencing engagement, meaning that it is an engagement not just
with the message alone passively but involving self-referencing content, which goes beyond the
sharing of a campaign message itself. I argue this concept explains the process of how a personal
health story elicits another personal health story on social media, which constitutes an important
step for successful health campaigns. In the following chapters, I present two studies. The first
included self-referencing engagement, in addition to persuasion, as primary outcome variables of
social media health campaign effectiveness. In the second study, self-referencing engagement
was further examined in a naturalistic setting of social media health campaigns through the
methodological lens of communication network and tie formation. More specifically, Study 1
investigated what makes social media an effective public health campaign tool in terms of
persuasion and self-referencing engagement outcomes respectively and the underlying
mechanisms. Study 2 took a computational approach to the process of self-referencing
engagement for social media health campaign effectiveness and illustrated that the power of
personal stories comes from its power of eliciting other personal stories.

15
CHAPTER 4: AN EXPERIMENTAL STUDY OF SOCIAL
MEDIA HEALTH STORY

Introduction
Previous studies suggest that media narratives have not only direct effects but also
indirect effects on attitudes and behavior by getting people to talk about the issues (Bandura,
2009). In particular, media narratives that address public figures’ health events have considerable
impact on increasing the public awareness about the health issues implicated in the events (e.g.,
Myrick, Noar, Willoughby, & Brown, 2014) as well as information seeking (e.g., Brown &
Basil, 1995; Dean, 2016; Metcalfe, Price, & Powell, 2011). For example, a study showed that
media coverage on public figures’ confession about their experience of panic attack motivates
public to seek more information about mental illness and disorders (Lee, 2018).
Research has demonstrated that the actions and statements of public figures can influence
public knowledge, interest, attitudes, behaviors, and willingness to discuss health issues (Dean,
2016; Noar, Willoughby, Myrick, & Brown, 2014). Through this increased public discussion, it
may even reduce stigma associated with the health issue (e.g., Corrigan et al., 2012). According
to a review that examined the publicly shared private health issues of 157 public figures such as
athletes, actors, musicians, and politicians, their disclosures of personal health issues indeed
triggered public activism and learning (Beck et al., 2014). For instance, in the wake of a
basketball player Earvin “Magic” Johnson’s announcement of HIV, more than 40,000 people
made calls to the National AIDS hotline in a single day (Brown & Basil, 1995). Moreover, it
instigated public discussion about Johnson’s disclosure as well as HIV in general with greater
16
interest in getting additional information (Kalichman & Hunter, 1992). More recently, Myrick
and colleagues (2014) found that sadness over the death of Steve Jobs, a prior chief executive
officer (CEO) at Apple, had an impact on public information-seeking behavior concerning
pancreatic cancer. A handful of research has corroborated the public figures’ influential power in
generating public interest in health issues and behavioral change, including the “The Angelina
Effect” (i.e., Angelina Jolie’s cancer announcement) on the public’s cancer-related beliefs and
behaviors as well as information seeking (Dean, 2016; Metcalfe et al., 2011; Kosenko, Binder, &
Hurley, 2016). These effects have been attributed to the availability of news information about
public figures (Dean, 2016) and individuals’ inherent motivations to discuss them (including
their health issues) with others (e.g., Myrick et al., 2014).
These days, communication channel for public figures’ health issues is not limited to
mass media coverage or news report but extends to social media postings, which are becoming
increasingly popular places that people discuss as well as learn about it. Receiving extensive
media coverage and social media discussion have potential to shape people’s attitudes, beliefs,
and interactions with others (Noar et al., 2014). However, little is known about the role of public
figure’s direct testimony on social media. It was also pointed out by Knoll and Matthes (2017) in
their meta-analysis that the effectiveness of public figure endorsements in health-related
communication has been understudied. In addition, most studies on the effects of such
endorsements are cross-sectional and they tend to follow a news event. Therefore, this chapter
proposes an experimental study that aims to investigate the effects of public figures’ personal
health story on social media, especially with regards to mental health problems. Building upon
the previous research that suggests media narrative about public figures’ disclosure of mental
health problems stimulates interpersonal communication about the topic, and thereby reduces the
17
stigma (Myrick, 2018), this study explicates how social media as communication channel makes
it even more effective.
Explanatory Mechanism: Social Presence and Transportation
Given that a story character featuring a stigmatized group member may make it difficult
for audience members to identify with them and thus to interfere with the rest of message process
(Kilk, Williams, & Reynolds, 2019), the story should not only address the struggles and
problems, but also should highlight what actions were taken with personal agency and efficacy
with a reference to getting help or treatment (Major, 2018). This also raises an important point
that the distance between the story character and audience member should remain close when
using personal stories for anti-stigma health campaigns.
Social presence, described as being aware of the other person in online environments
(Biocca, Harms, & Burgoon, 2003), may provide a useful mechanism to explain the relative
effectiveness of social media for personal stories. Prior studies have established that the
perceptual experience of being together with a mediated other is a critical concept that facilitates
individuals’ psychological responses in a virtual environment (Jin, 2012; Jin & Park, 2009; Kim
& Song, 2016; Lee & Nass, 2005). For instance, Lee (2012) found that a politician’s social
media posting about their private life induced greater social presence than posting about their
public life. More specifically, social media users felt a sense of being together with the politician,
which in turn enhanced the likability of that politician. Indeed, increased social presence was
found to be associated with positive evaluation of message sources including greater
trustworthiness (Bente, Rüggenberg, Krämer, & Eschenburg, 2008) and more favorable
impressions regardless of the sources’ physical attractiveness (Skalski & Tamborini, 2007).
18
Conceptually, social presence bears a similarity to another concept, identification, one of
the key constructs that explains narrative persuasion. Identification transpires when audience
understands and shares the narrative characters’ emotions and perspective, internalizing their
goals or even losing self-awareness (Cohen, 2001; Moyer-Gusé, 2008). Both social presence and
identification are related to story characters, not story itself. However, social presence concerns
more about media-induced perception toward the mediated person or character whereas
identification concerns more about message-induced (i.e., narrative format) effects. Given the
current study focuses on the effectiveness of using social media to deliver public figure’s
personal stories (not a different narrative characteristic that may determine the message-induced
effects), social presence was more pertinent than identification to this study context thus chosen.
Will personal stories in a social media format induce greater story involvement?
Enhanced social presence makes the mediated experience seem more real (Lombard & Ditton,
1997), fostering a stronger story involvement. Studies have shown that when people perceive a
greater social presence of a mediated person, they feel as if they are engaging in real, non-
mediated interaction with her/him. That feeling of vividly being part of a conversation suggests
its relevance to the concept of transportation. Transportation refers to the extent to which the
audience is engrossed (transported) into a story (Green & Brock, 2000). According to Green et
al. (2008), while reading a book or viewing a film, people experience transportation into the
story world, “a state of cognitive and emotional immersion in a text” (p. 513). The result of being
transported into a story leaves the audience more susceptible to beliefs and attitude change, as
the transported state weakens their ability to counterargue or challenge the message content
(Green & Brock, 2000; Slater & Rouner, 2002). As a form of experiential response to narratives,
transportation makes one’s story experience seem more like a real experience (Green & Brock,
19
2000; Green et al., 2008). Taken together, social presence and transportation will serially
mediate the effects of social media personal stories on the persuasion outcomes. Based on this,
the following hypotheses are proposed:
H1: Participants who read a public figure’s mental health stories in a social media format
will have (a) more positive attitudes toward professional help for mental health care and
(b) greater help-seeking behavioral intentions than their counterparts who read the same
stories in a non-social-media online article format.
H2: Participants who read a public figure’s mental health stories in a social media format
will experience a stronger sense of (a) social presence with the public figure in a story
and (b) transportation into the story than their counterparts who read the same stories in a
non-social-media online article format.
H3: The relationship in H1 will be mediated by social presence and transportation: That
is, participants who read a public figure’s mental health stories in social media format
will experience a stronger sense of social presence that is (a) positively associated with
the level of transportation, which in turn leads to (b) more positive attitudes toward
professional help for mental health care and (c) greater help-seeking behavioral
intentions.
Parasocial Intimacy with Public Figures
Previous research suggests that the development of intimate relationships with media
personae influences the way audience members interpret the actions and the message (Horton &
Wohl, 1956). Such relationships based on the audience members’ pre-existing intimacy with
public figures are conceptualized as parasocial intimacy (Dai & Walther, 2018). For instance,
20
McCracken (1989) demonstrated that the persuasive effects of public figures’ endorsement of a
product depend on how strong audience feels close with them. Given the similarities between the
interpersonal and mediated context where the story’s effects may differ by whose story it is –
whether the audience feels intimate with the story character or not – their pre-existing intimacy
with a public figure may moderate the hypothesized mechanisms of persuasion. To explore the
potential interaction effects, the following research question was proposed:
RQ1: How will the participants’ pre-existing parasocial intimacy with a public figure
influence the persuasive route of (a) social presence and (b) transportation, respectively
on the outcome variables?
Interpersonal Outcome of Social Media Personal Stories: Self-
Referencing Engagement
Optimizing the content of health campaigns and effectively disseminating the campaign
messages are essential to successful public health campaigns. The use of social media in public
health has been largely examined for its ability to reach broad audience and foster public
engagement (Neiger et al., 2012). Given the interactive communication features available such as
comments and replies, audience members now actively engage with health campaign messages
on social media (e.g., Bou-Franch, Lorenzo-Dus, & Blitvich, 2012). Such engagement and
participation not only facilitate the delivery of campaign messaging but also enhance the
interaction among the audience members (Kreps, 2017; Neuhauser & Kreps, 2014). However,
relative to the amount of discussion on the dissemination aspect of health messages, not much
empirical research has been done on how we can optimize campaign messages to invite greater
engagement with such messages on social media.
21
In addition to persuasiveness, a traditional outcome of health campaign effectiveness, I
argue that the engagement aspect should be considered as an important outcome for health
campaigns especially that use social media as a primary communication channel. As outlined in
the previous chapter, I emphasize the concept of self-referencing engagement. Self-referencing
occurs when a message reminds audience members of their own experiences and they connect it
to their own lives (Burnkrant & Unnava, 1995; Kreuter & Wray, 2003). Self-referencing may
happen based on one’s personal experience or issue involvement often times regardless of the
similarity with or relevance to the story character (Chen, Bell, & Taylor, 2016). Accordingly,
self-referencing engagement as campaign outcome could be defined as engagement with
campaign message in response to personal relevance that they find in connection with the
message.
Why is it important to consider self-referencing engagement for social media campaign
outcomes? What implications does self-referencing engagement have for anti-stigmatizing
campaign? A study that extracted major topics of mental health on Twitter may provide relevant
insights (Berry et al., 2017). In this study, they examined tweet messages with a hashtag,
#WhyWeTweetMH and identified four themes: (1) sense of community, (2) raising awareness
and combating stigma, (3) safe space for expression, (4) coping and empowerment. Similarly,
Naslund and his colleagues (2014) found that YouTube served as an environment where
individuals with severe mental disorders openly shared their personal experience with symptoms
and strategies for coping with day-to-day challenges, which received positive, encouraging, and
insightful reactions.
Indeed, knowing that there are others facing similar concerns, experiences, and
challenges can be highly reassuring as well as promote a sense of social bonding (Naslund et al.
22
2014). Especially for marginalized individuals, benefits include feelings of empowerment,
stronger personal identity, and pride by connecting with similar others online (van Uden-Kraan,
Drossaert, Taal, Seydel, & van de Laar, 2009). Given that disclosing one’s health problems is an
important component of addressing stigma (Corrigan, Kosyluk, & Rüsch, 2013), one of the
indications of successful campaign that aims to raise mental health awareness would be inviting
audience members’ sharing of their personal stories (i.e., self-referencing engagement). Thus, I
propose to test if using social media for delivering personal stories will be effective in inviting
self-referencing engagement. Additionally, RQ2 further explores whether the self-referencing
engagement varies by the level of parasocial intimacy:
H4: Participants who read a public figure’s mental health stories in a social media format
will report greater self-referencing engagement than their counterparts who read the same
stories in a non-social-media online article format.
RQ2: How will the participants’ pre-existing parasocial intimacy with a public figure
influences the self-referencing engagement effects of social media personal stories?
Method
Pilot Study
In developing experimental stimuli for this study, it was important to decide which public
figures’ stories would be used and ensure they are not particularly positive or negative in general
public perception. Therefore, a pilot study was designed with a list of eight different public
figures (Ariana Grande, Mariah Carey, Kristen Bell, Selena Gomez; Dwayne Johnson, Kevin
Love, Michael Phelps, Ryan Reynolds) who disclosed their mental health problem publicly in
23
2018 (Families for Depression Awareness, 2018; Today, 2018). In total, 221 participants living
in the United States took part in the pilot study through Amazon Mechanical Turk. Participants’
ages ranged from 20 to 65 (M = 37.9, SD = 10.12). Males constituted 55.2% of the sample. The
largest ethnic groups in the sample were White/Caucasian (78.2%), Hispanic/Latino (7.2%),
African American/Black (6.3%), and Asian American (5.4%). Participants were asked to indicate
if they knew the public figure and to rate the likability of each person on 5-point scale (1 = Very
unfavorable; 5 = Very favorable). Based on the results from one-sample t-tests, two female
(Kristen Bell, M = 3.50, SD = 1.52, t(220) = .02, p = .982; Mariah Carey, M = 3.63, SD = 1.37,
t(220) = 1.39, p = .164) and two male public figures (Dwayne Johnson, M = 3.59, SD = 1.37,
t(220) = 1.01, p = .315; Kevin Love, M = 3.39, SD = 1.96, t(220) = -.84, p = .400) whose scores
were not statistically significant from the mid-point scale were chosen to be used in the main
study (for the others: Ariana Grande, M = 3.69, SD = 1.51, t(220) = 1.85, p = .065; Selena
Gomez, M = 3.88, SD = 1.57, t(220) = 3.62, p < .001; Michael Phelps M = 3.77, SD = 1.35,
t(220) = 3.02, p = .003; Ryan Reynolds, M = 4.00, SD = 1.54, t(220) = 4.86, p < .001).
Procedure
Based on the results of the pilot study, eight different messages were created by adjusting
the real stories from news articles, interviews, and if any, each public figure’s own personal
social media mental health disclosure story. For the social media condition, Twitter was chosen
to enhance the realism of the stimuli for representing a social media condition that is often used
by public figures and celebrities for mass communication with broader audience
(Tanupabrungsun & Hemsley, 2018). For the comparison group, the online article condition was
created to feature a story from The New York Times. Corresponding messages from the Twitter
conditions were presented in quotations so that the perspectives of stories remained as a first-
24
person point of view across the conditions. To ensure the validity of the manipulation while
decreasing the likelihood that the effects found were unique to particular messages (i.e.,
idiosyncrasies in experimental manipulation), I used heterogeneity in the manipulations (Slater,
Peter, & Valkenburg, 2015; or “variability” in O’Keefe, 1999) to reflect real-world stories,
which allows for generalization across multiple messages. The stories followed the same
structure with balanced information across different conditions (i.e., their struggles, their first
reaction, dealing with stigma, how they overcame) and word counts were kept consistent in
length (328 words for Kristen Bell’s story; 324 words for Mariah Carey’s story, 321 for Dwayne
John’s story; 328 for Kevin Love’s story) with the same message used between two different
media conditions for each public figure. Message stimuli can be found in the Appendix page.
Upon their consent to participate, participants were first asked to check any social media
account they were currently using. They were also asked to indicate whether they recognize the
public figures and rate their pre-existing attitudes toward them. Participants then proceeded to be
randomly assigned to either no-message control condition or experimental condition (i.e., media:
Twitter vs. online article, public figures: Kristen Bell or Mariah Carey for female participants
and Dwayne John or Kevin Love for male participants). That is, within the experimental
condition, participants were randomly assigned to one of four distinct messages (i.e., 2 different
media formats x 2 different public figures who appeared to share the same gender as the
participant) that contained a personal story about their mental health struggle and help-seeking
experience. Participants who were randomly assigned to an experimental condition first rated
their pre-existing attitudes toward and parasocial intimacy with the public figures, then read a
single story of a public figure (randomly assigned between Kristen Bell or Mariah Carey for
female participants and between Dwayne Johnson or Kevin Love for male participants) either in
25
a Twitter or online article format. Then they proceeded with completing the remaining questions.
Participants who were randomly assigned to the no-message control group did not go through
any story-related questions, but only answered the questions that measured persuasion outcome
variables.
Participants
For the main study, 389 participants over age 18 in the United States were recruited using
Qualtrics. To ensure participants’ familiarity with the message stimuli, they were screened for
those who currently had a Twitter account at the time of participation out of multiple options in
the questionnaire that included Facebook, YouTube, Instagram, Snapchat, Pinterest, LinkedIn
(Pew Research Center, 2021). Participants’ ages ranged from 18 to 65 (M = 38.7, SD = 13.8).
Females constituted 51.1% of the sample (n = 190) and the predominant ethnic groups in the
sample were White/Caucasian (61.2%), Hispanic/Latino (13.4%), African American/Black
(13.1%), and Asian American (5.7%).
Measures
Pre-existing attitude. Participants’ pre-existing attitudes toward public figures were
measured with other non-relevant public figures acting as fillers to avoid sensitizing participants
to the purpose of the study. Items were from Dai and Walther (2018) that included “foolish-
wise,” unacceptable-acceptable,” “unfavorable-favorable,” “wrong-right,” “bad-good,” and
“negative-positive” on 7-point scale (Cronbach’s alpha = .94, M = 4.59, SD = 2.59).
Parasocial intimacy. Participants’ parasocial intimacy with public figure was measured
using 7 items from Chung and Cho (2017) that concern parasocial intimacy with celebrity in the
26
social media context. Items include “I think I can understand what kind of person s/he is,” “S/he
makes me feel comfortable as if I am with a friend,” “I see her/him as a natural, down-to-earth
person,” “I would like to have a friendly chat with her,” “If she were not a famous person, we
would have been good friends,” “If there were a story about her/him, I would read it,” “I feel as
if I have known her/him for a long time.” Participants rated on 7-point scale (1 = Strongly
disagree, 7 = Strongly agree; Cronbach’s alpha = .92, M = 3.75, SD = 2.18).
Transportation. Participants rated how much they were transported into the story on 7-
point scale. The items used were from the Transportation Scale-Short From (TS-SF, Appel,
Gnambs, Richter, Green, & 2015; Green & Brock, 2000): “I could picture myself in the events
described in the story,” “I was mentally involved in the story while reading it,” “I wanted to
learn how the story ended,” “The story affected me emotionally,” “While reading I had a vivid
image of the story.” (Cronbach’s alpha = .90, M = 5.27, SD = 1.30).
Social presence. The extent to which participants felt as if they were being together in a
conversation with the story character was measured with 4 items on 7-point scale (Lee & Shin,
2014): “I felt as if I were engaging in an actual conversation with her/him,” “I felt like I was in
the same room with her/him,” “I felt as if s/he was speaking directly to me,” “I could imagine
her/him vividly” (Cronbach’s alpha = .92, M = 5.04, SD = 1.34).
Help-seeking behavior intention. Items for the behavioral intent to seek help comprised
of two sub-dimensions (Sheffield, Fiorenza, & Sofronoff, 2004; Hoffner & Cohen, 2015):
seeking professional help (Psychologist, Psychiatrist, Doctor, and Telephone crisis hotline) and
informal help (Family, Friends, Coworkers, Religious group). Participants rated how likely they
would seek help from each if they were to experience mental health problem in the next 12
months on 7-point scale (1 = Not at all likely, 7 = Very likely). The items were then indexed for
27
intentions to seek professional help (Cronbach’s alpha = .82, M = 4.81, SD = 1.42) and informal
help (Cronbach’s alpha = .74, M = 4.19, SD = 1.44), respectively.
Attitudes toward help-seeking behavior. Three items on 7-point semantic differential
scales (unimportant/important, worthless/valuable, undesirable/desirable) were taken from Shi
and Dai (2020)’s study to measure the attitudes toward getting professional help for mental
health care. The scores were averaged to form an index for attitudes toward professional help-
seeking behavior (Cronbach’s alpha = .91, M = 6.17, SD = 1.13).
Self-referencing engagement. To measure the participants’ self-referencing engagement
with message, an open-ended question was used. The participants were asked to take 2 minutes
to list the thoughts that came across their mind while reading the message. Then, the responses
were analyzed using a computational software, LIWC 2022 (Linguistic Inquiry and Word Count;
Boyd, Ashokkumar, Seraj, & Pennebaker, 2022) that calculates a variety of different linguistic
styles. Here I focused on the use of first-person singular pronouns (e.g., I’d, I’m, me, my,
myself) to operationalize self-referencing thoughts. Individual scores were extracted from the
proportions of the use of first-person singular pronouns in relation to the total number of words
in the response. This method has been widely used and validated by previous works that
especially considered message style (i.e., how it is communicated) rather than content (i.e., what
is being communicated) as a primary interest of study since the early development of LIWC
(e.g., Gil-Lopez et al., 2018). For instance, using this approach, studies have shown that style
words such as pronouns and articles are the most common and consistent in our daily
conversation that reflect psychological processes that are independent of the content or topic of
communication (Tausczik & Pennebaker, 2010). Also, the machine-coded scores were shown to
be reliable in comparison with human raters’ estimates (e.g., Bantum & Owen 2009; Cohn,
28
Mehl, & Pennebaker 2004). The scores ranged from 0 to 28.57 and higher score indicated a
greater self-referencing engagement (M = 6.22, SD = 5.03).
Results
Preliminary Analysis
To ensure successful randomization, participants’ attitudes toward the public figures of
each condition were checked. It showed there was no significant differences in the pre-existing
attitudes (MKristen Bell = 5.57, SDKristen Bell = 1.06; MMariah Carey = 5.44, SDMariah Carey = 1.39; MDwayne
Johnson = 5.73, SDDwayne Johnson = 1.16; MKevin Love = 5.45, SDKevin Love = 1.13) across different
conditions, F(3, 321) = 1.087, p = .371. In other words, no public figure was particularly more or
less favored than others. To further assess if there were differences derived from different public
figures within the same media format (i.e., twitter vs. article), a series of ANOVAs was
conducted to compare four messages within each media condition on the outcome variables. No
significant differences were found in the same media condition regardless of whom the message
was about (all p-values > .05). Therefore, all subsequent comparisons between conditions were
made by collapsing results across individual public figures within each condition of different
media format. Finally, participants in the conditions (Twitter vs. Article vs. No-message-control)
did not vary by age, gender, or race (p-values > .1).

29
Table 4.1. Bivariate Correlations among Variables
Variables 1 2 3 4 5 6 7 8 9 10
1. Age –
2. Gender -.27*** –
3. Education .07 -.13* –
4. Presence .11* -.04 -.06 –
5. Transportation .11 -.01 -.09 .74*** –
6. Parasocial -.17*** -.12* .01 -.17** .02 –
7. Attitudes .07 -.02 -.10 .22*** .33*** .01 –
8. Professional -.01 .02 -.11* .27*** .44*** .07 .46*** –
9. Informal .01 -.07 -.02 .26*** .38 .01 .16** .53*** –
10. Self-reference -.08 .06 -.05 -.01 -.01 .21*** -.01 .02 -.11 –
Note. Gender: male = 1, female = 2. *p < .05. **p < .01. ***p < .001.

Hypotheses Testing
What makes social media effective for persuasion?
To test the hypotheses, a series of analyses of variance (ANOVAs) were conducted.
Planned contrasts were also performed to test for differences between specific conditions. I first
examined the effects of different media conditions on outcome variables of persuasion. Results
showed significant differences in the attitudes toward professional help for mental health care,
F(2, 369) = 5.02, p = .007, η
2
= .03, and help-seeking behavioral intentions, F(2, 369) = 6.958, p
= .001, η
2
= .04. Specifically, personal stories in a social media format triggered more positive
attitudes toward professional help for mental health care (M = 6.32, SD = .97) than baseline
control group (M = 5.83, SD = 1.58) at p = .005. However, there was no evidence that it differed
from online article format (M = 6.16, SD = .94, p = .463), partially supporting H1a.

30

Figure 4.1. Attitudes toward help-seeking by different media conditions.
Note. Means with no letters in common were significantly different to each other.

In terms of intentions for help-seeking behavior, public figure’s mental health story in a
social media format (M = 5.09, SD = 1.40) led to the greatest intention of seeking professional
help, F(2, 369) = 6.958, p = .001, η
2
= .04 compared to the same story in an online article format
(M = 4.54, SD = 1.43, p = .004) as well as no-message baseline control (M = 4.50, SD = 1.52, p
= .008). A similar pattern was observed for the intentions of seeking informal help such that
participants who read a public figure’s mental health story in a social media format (M = 4.28,
SD = 1.45) showed the greatest intentions of seeking informal help, F(2, 369) = 4.108, p = .017.
However, such difference was statistically significant from the no-message control only (M =
3.85, SD = 1.45, p = .020), not the online article format (M = 4.04, SD = 1.4, p = .113), partially
supporting H1b.

31

Figure 4.2. Behavioral intentions of seeking professional help by different media conditions.
Note. Means with no letters in common were significantly different from each other.

Figure 4.3. Behavioral intentions of seeking informal help by different media conditions.
Note. Means with no letters in common significantly different from each other.
32

To assess the mechanisms underlying the observed effects, how personal stories in
different media formats influence social presence and transportation was examined. First,
personal stories in a social media format were associated with greater social presence of the story
character (M = 5.35, SD = 1.27) than the online article format (M = 4.71, SD = 1.35), t(286) = -
4.11, p < .001, Cohen’s d = .48, supporting H2a. Additionally, participants who read personal
stories in a social media format showed greater transportation (M = 5.50, SD = 1.18) than those
who read it in an online article format (M = 5.03, SD = 1.27), t(286) = -3.21, p = .379, Cohen’s d
= . 38, supporting H2b.
H3 predicated a serial mediation model where transportation mediates the effects of
social presence in response to different media formats on the outcome variables. This hypothesis
was tested with Model 6 in Hayes’ PROCESS Macro (Hayes, 2013) with each outcome variable.
PROCESS generated bias-corrected bootstrap confidence intervals for all indirect effects using
5,000 bootstrap samples. First, the indirect effects of different media conditions on the attitudes
toward help-seeking was significant, b = .11, SE = .04, 95% CI [.039, .196] such that presenting
personal stories in a social media format (vs. in an online article) elicited greater social presence,
b = .63, SE = .15, 95% CI [.330, .937] which in turn was related to greater transportation into a
story (H3a supported), b = .68, SE = .04, 95% CI [.123, .374], leading to a more positive
attitudes toward help-seeking behavior (H3b supported), b = .25, SE = .06, 95% CI [.123, .374].
33

Figure 4.4. Serial multiple mediation model with attitudes toward help-seeking as an outcome.
Note. Coefficients and standard error in parentheses. Total indirect effect: b = .12, SE = .05, 95%
CI [.036, .213]. *p < .05, **p < .01, ***p < .001.

In the next model, behavioral intention to seek professional help was entered as an
outcome variable. As expected, social presence and transportation serially mediated the effects of
different media conditions on seeking professional help (condition → presence →
transportation→ seeking professional help), b = .13, SE = .05, 95% CI [.043, .239]. That is,
social media condition elicited greater social presence, b = .63, SE = .15, 95% CI [.330, .937],
which was then translated into a higher level of transportation, b = .68, SE = .04, 95% CI
[.604, .754], leading to a greater intention to seek professional help, b = .30, SE = .09, 95% CI
[.125, .477]. Results also revealed another significant indirect path from media conditions to
seeking professional help via presence while controlling for transportation (condition →
presence → seeking professional help), b = .15, SE = .09, 95% CI [.141, .476].

34

Figure 4.5. Serial multiple mediation model with seeking professional help as an outcome.
Note. Coefficients and standard error in parentheses. Total indirect effect: b = .29, SE = .09, 95%
CI [.141, .476]. *p < .05, **p < .01, ***p < .001.

Lastly, a similar pattern was found with seeking informal help as an outcome variable.
That is, the social media condition indirectly influenced greater intention of informal help-
seeking behavior via greater social presence and transportation in serial (condition → presence
→ transportation→ seeking informal help), b = .09, SE = .04, 95% CI [.019, .184], supporting
H3b. Likewise, an indirect effect of media conditions on seeking informal help was significant
via presence alone (condition → presence → seeking informal help), b = .27, SE = .07, 95% CI
[.139, .420] while transportation alone did not mediate the effects. Notably, the direct effects of
different media conditions (X) on Ys were not significant (dashed lines in Figures), highlighting
the indirect mechanism (M1 and M2) that explains the observed effects in H1.

35

Figure 4.6. Serial multiple mediation model with seeking informal help as an outcome.
Note. Coefficients and standard error in parentheses. Total indirect effect: b = .27, SE = .07, 95%
CI [.139, .420]. *p < .05, **p < .01, ***p < .001.

To address RQ1, whether these effects were moderated by the level of parasocial
intimacy was tested. Results showed that there was no evidence that parasocial intimacy
interacted with social presence on persuasive outcome variables FAttitudes(1, 284) = 1.96, p = .163;
FProfessionalHelp(1, 284) = .44, p = .509; FInformalHelp(1, 284) = .02, p = .965. However, parasocial
intimacy interacted with transportation on the attitudes toward professional help for mental
health care, F(1, 284) = 7.76, p = .006. That is, the effects of increased transportation on positive
attitudes toward help-seeking were more prominent among those who have greater parasocial
intimacy with the public figure. However, such interaction effects were not observed with
respect to behavioral outcomes, FProfessionalHelp(1, 284) = 1.76, p = .186; FInformalHelp(1, 284) = .81,
p = .368.
36

Figure 4.7. Interaction effects on help-seeking attitudes.

What makes social media effective for engagement?
For the illustrative purpose, sample responses that received low self-referencing
engagement scores were as follow: “Very true those words, sometimes persons with mental
health need to be heard about how they feel”; “Always respected Love since he was at UCLA,
Glad he is open about himself, respected”; “She was open and honest about her struggles. Mariah
pointed out it is okay to have issues. She took time to focus on her health and received medicine.
It is important to share mental health because most people deal with it.”; “He was conveying his
own experience on mental health and how he thought help was not needed for him. He then
realized how it could help him and how help could do well for others who had the same thoughts
as he did.”; “In describing the incident with his mother I felt that this was a heavy burden for any
young teenager. Caretaking of a parent with mental health issues is an increasingly common
burden and can hold children back from achieving their goals. His rise to stardom has given him
37
a platform to share his experiences and advocate on mental health issues in a way that would not
be available to many who face similar struggles. He has shown a light on not merely the problem
of youngsters, particularly boys, struggling with mental health issues but also lighted a pathway
for them to reach out for help.”
On the other hand, examples of high self-referencing engagement scores were as follow:
“I related to the message very much. These similar things happened to me also around the time I
started college.”; “I remember thinking that is exactly how my anxiety started. The feelings she
described were very similar to mine. My anxiety started in college as well.”; “It took me back to
times where I've felt like Mariah felt. Feeling overwhelmed and like I wasn't doing enough for
everyone around me.” In sum, the responses that received lower scores in self-referencing
engagement were mostly on the story itself whereas high scores in self-referencing engagement
contained references to themselves, demonstrating the frequent use of I-words.
Analyzing such responses, different media effects on self-referencing engagement were
examined. Between the two different media conditions, personal stories in a social media format
triggered greater self-referencing engagement (M = 6.86, SD = 5.42) than those in an online
article format (M = 5.54, SD = 4.47), t(282) = -2.23, p = .027, Cohen’s d = .26, supporting H4. In
answering RQ2, results showed significant interaction effects of media conditions and parasocial
intimacy, F(1, 280) = 5.136, p = .024. As can be seen from Figure 8, the effects of media
conditions were more prominent with a lower level of parasocial intimacy. In other words, even
if the story character was not somebody whom participants felt greatly intimate with, as long as
the message was in a social media format, they found it more self-referencing and engaged more
with the message. In contrast, there was little difference in the media-dependent effects of
38
personal stories when the participants had already parasocial intimacy with the public figure in
the story.

Figure 4.8. Interaction effects on self-referencing engagement.

Discussion
This study examined the power of public figures’ social media personal stories in terms
of two different approaches to the effectiveness of social media health campaigns: persuasion and
engagement. First, this study used multiple messages for experimental stimuli to minimize
variation due to idiosyncrasies in the manipulation and to increase the generalizability of the
observed findings (e.g., Jensen, 2008; Kim et al., 2012). Second, by including two comparison
conditions – a non-social media format (online article) and a baseline control condition, it
offered a more nuanced approach to testing the relative effectiveness of social media for public
health campaign and communication channel. A public figure’s mental health stories in a social
39
media format showed significantly persuasive effects on different outcome variables including
attitudes toward help-seeking, as well as behavioral intentions to seek professional and informal
help. In particular, such effects were significantly different from the same story in an online
article format when it came to seeking professional help for mental health care. This corroborates
the media-specific effects on outcome variables, highlighting that the observed effects are not
merely from reading a story itself but that what media the story is delivered in matters. Indeed,
non-significant differences in the persuasion outcomes between the participants who read
personal stories in an online article format and the participants who did not read any message at
all (i.e., baseline control group) suggests there was no evidence of relative effectiveness of
narrative story alone in this study. However, the same stories in a social media format were
consistently better at achieving statistically significant effects than the baseline control group.
These findings shed light on the power of using social media for effective mental health story
format choice in promoting professional help-seeking behavior.
In addition, this study further explicated the underlying mechanisms of how the effects of
social media format worked. Findings showed that it was through enhanced social presence
elicited from social media story, which was translated into a greater level of story involvement
(i.e., transportation) that influenced the persuasion outcomes. Notably, the observed direct effects
of media conditions on transportation in H2 no longer existed in the serial mediation models
(H3), indicating that media format does not influence the level of story involvement when
controlling for the effects of social presence. This again highlights the importance of social
presence in the model explaining the persuasive power of social media personal stories for anti-
stigmatizing campaigns.
40
What does the positive relationship between social presence and transportation tell us?
One speculation could be related to the distance from the public figure story character. Increased
social presence essentially means participants felt reduced distance between themselves and the
story character (i.e., public figure). According to the construal level theory (Trope & Liberman,
2010), people tend to construe distant entities with a broad and abstract mindset (i.e., high-level
construal) whereas they construe close entities with narrow and concrete mindset (i.e., low-level
construal). Therefore, such reduced distance may activate the readers’ low-level construal,
making them more focused on the vivid details of the story. In other words, the findings from H2
(i.e., significant effects of media conditions on transportation) imply that the social media format
makes a story more engaging and transportable; but according to the findings from H3 (i.e., non-
significant effects of media conditions on transportation when considering social presence), it
may be rather that the social media format makes readers, not the story itself, more ready to
concretely process the story, and thus more transported into the story. Therefore, social presence
serves a crucial theoretical mechanism that underlie the media-dependent effects of personal
stories for subsequent persuasion route, that is, transportation.
Notably, the relationship between transportation and the attitudes toward help-seeking
behavior was significantly moderated by the level of paraoscial intimacy that participants had
with the public figure featured in a story. The positive relationship between transportation and
attitudes was prominent among those who had a stronger parasocial intimacy with the character.
This is consistent with the findings in previous literature that the persuasive effects of interacting
with mediated characters or their story are dependent on the relationship or intimacy with the
media personae (Horton & Wohl, 1956). Interestingly, however, paraoscial intimacy did not
influence the effects of social presence on persuasion. In other words, regardless of the extent to
41
which an individual feels close to the storyteller, the social media-induced effects of reducing
distance with the person are strong enough to have persuasive power. Instead, how much an
individual feels close to the storyteller interacted with the following story involvement such that
when engrossed with personal stories of a public figure with whom he or she has already
developed an intimate parasocial relationships, the persuasive effects of story become stronger.
It is not news that social media facilitates broader and faster dissemination of health
campaigns while enabling people to share the message, connect with others, and support one
another, but how it really works has long been warranting further investigation. This study
empirically unpacked how social media users engage with public figure’s health stories by
sharing their own story. The findings provide empirical support for the benefits of using social
media stories to promote mental health campaigns coming from its power of generating relevant
conversation that is also composed of personal stories (i.e., self-referencing engagement). Self-
referencing engagement was proposed in an attempt to distinguish the effectiveness of social
media as a health campaign tool from other media formats given that audience participation
constitutes an important part of social media campaigns. It is not just engaging with the
campaign message itself (i.e., liking or forwarding) but more about sharing one’s story motivated
by self-referencing thoughts in response to the campaign message. Furthermore, such effects of
social media personal stories were also salient among those who had little parasocial intimacy
with the storyteller. In other words, even if a story is featuring a public figure who one does not
regard as close or intimate, the social media format of story enables people to find it relevant and
be willing to share their own story.
Although the format was limited to a Twitter in this study, future studies may explore
other social media formats and see if the findings are replicated. In addition to the media-specific
42
variations, message modality may be of interest to consider. For instance, prior research
demonstrated that it is more effective to use audio or visual modality for testimonials while text
modality for factual information (i.e., non-narrative) (Braverman, 2008). Given the expanded
diversity in terms of message modality used for health campaigns nowadays, social media
personal stories may take video format that contains public figures’ own voice and visual. With
such increased modality richness (Burgoon et al., 2002; Chaiken & Eagly, 1983; Ramirez &
Burgoon, 2004), social presence will likely be even more enhanced, which has an important
insight to anti-stigmatizing campaigns where reducing the distance with the stigmatized group is
one of the key goals. Lastly, this study only addressed personal stories that are directly delivered
by public figures from their own social media account, but whether the same stories made by
organizations or agencies, not individuals, have similar effects may be further explored. As noted
earlier, the main explanatory mechanism was the enhanced feeling of social presence that
participants felt as if they were in a real conversation in a less mediated environment with the
storyteller. Therefore, making social presence salient even with such health organizations or
agencies will serve an important direction for future research on health campaigns using social
media.
43
CHAPTER 5: COMPUTATIONAL APPROACH TO SOCIAL
MEDIA CAMPAIGNS

This chapter aims to further investigate how self-referencing engagement works with a
case study of social media health campaign celebrating Mental Health Awareness Week. In the
previous chapters, I have emphasized in conceptualizing self-referencing engagement that it
involves 1) social media users’ engagement with the campaign message 2) contains own
personal story reflecting their self-referencing thoughts and experiences, demonstrated by the
frequent use of first-person pronouns. In this chapter, I took a naturalistic setting of a social
media environment to empirically examine how self-referencing engagement unfolds and what it
means to social media health communication and campaign research. The main idea that guides
this project is taking a computational approach to the investigation of social media campaign
effectiveness with the operationalization of engagement as tie formation in the social media
communication network.
Social Media Health Campaign Effectiveness
Broadly, there are three different ways of evaluating health campaigns: formative,
process, and summative evaluation (Atkin, Rice, & Valdivia, 2012). Formative evaluation is
conducted during the development and planning stage to revise message and strategies, which
requires collaborative work through interpersonal feedback. Process evaluation, on the other
hand, is done during the implementation to see how well the campaign is working. Summative
evaluation is conducted after the campaign, assessing behavioral and attitudinal changes in the
target population. On social media, user interaction and traces of communication are usually
44
publicly available, providing informative data for campaign evaluation (Neiger et al., 2012). For
instance, system-generated metrics that appear in the form of number of views, likes, and shares
may provide quantitative information regarding the degree to which the campaign message has
reached (i.e., popularity or virality of certain campaign messages) as well as what the audience
response to that has been like (e.g., McNeill & Briggs, 2014; Park et al., 2016). Accordingly,
scholars have developed assessment metrics to manage and analyze social media-based
campaigns (Duke, Hansen, Kim, Curry, & Allen, 2014; Neiger et al., 2012; Peters, Chen,
Kaplan, Ognibeni, & Pauwels, 2013).
However, metrics alone may not be sufficient to capture how the campaign influences the
audience. Given the communication environment on social media connects people and message,
essentially forming networks, campaign effectiveness may be evaluated within such network
framework. In particular, self-referencing engagement involves interpersonal effects – that a
personal story brings about others’ personal stories. Notably, in this context, what is being
connected is not limited to users but also their messages. That is, the effects of health campaign
messages on other messages can be examined by the process of how such messages are
connected to each other. Therefore, in the following sections, I first demonstrate this network
approach to examining the effects of personal stories (i.e., self-referencing engagement) and
propose the hypotheses that this study aims to test.
Power of Personal Stories in Communication Network
On social media, all the communicative actions result in networks. Network ties form
when an individual or organization mentions, follows, likes, or shares the other actor or their
messages. In this case, the node of such network could be an individual user or an individual
45
message. For instance, consider a hyperlink network on Facebook. Postings (i.e., messages) may
be connected to one another based on the shared external hyperlink (e.g., news article, event
calendar, etc.) – that is, if two messages share the same link, they form a hyperlink tie and if two
messages do not share any hyperlink together, they are separate, not connected. From the
engagement point of view, two messages form a tie when a message quotes the other message in
response, which is done by users’ engagement behavior. In other words, whether a message is
likely to be engaged by others can be conceptualized as how much the message is likely to form
a tie with other messages.
Building upon the discussion of the power of personal stories in the previous chapters, I
propose to extend it to empirically testing the effectiveness of personal stories on social media in
a naturalistic setting of social media communication network in terms of tie formation as an
indication of message engagement. First, it raises an important question to address: what makes a
story personal in the context of social media health campaigns? As narrative framework
suggests, it refers to testimonial that describe personal experience or anecdotal evidence
(Braverman, 2008; Slater & Rouner, 2002). For example, in the Study 1 of this dissertation
project, the experimental stimuli messages for mental health personal stories were made
following the suggestion by the existing literature (e.g., Major, 2018) such that they contained
individuals’ struggle with mental health problems, how they first found out and reacted to it, and
finally how they dealt with stigma and successfully overcame it. Recall that besides the overall
structure and the content, the perspective of the storytelling was consistent in the first-person
point of views. That is, the story was presented in the voice of the speaker.
Narrative stories in the first-person point of views (Chen et al., 2016, or voice of
narrative, Kaufman & Libby, 2012) are often used interchangeably with personal testimony. In
46
the first-person narrative, central character narrates the story from his or her point of view, using
pronouns such as “I,” making audience members have access to the perception and thoughts of
that narrator. On the other hand, in the third-person narrative, an unspecified entity or uninvolved
person serves the narrator to convey the story (Banerjee & Greene, 2012). Research has shown
that personal stories in the first-person perspective are perceived to be more interesting and
enjoyable (Slater & Rouner, 2002), easier to read (Brosius & Bathelt, 1994), and more vivid
(Taylor & Thompson, 1982) in comparison with the third-person perspective (Winterbottom,
Bekker, Conner, & Mooney, 2008). More importantly, it was suggested that readers may
perceive first-person narrative as testimonial while third-person narrative more as an exemplar,
limiting the effects of personalized experience from third-person narrative (Banerjee & Greene,
2012) because the first-person narrative enables readers to see the thought process of characters,
making the stories feel more personal (Cohen, 2001; de Graaf, Hoeken, Sanders, & Beentjes,
2012; Oatley, 1999).
Whether a narrative message is in the first-person perspective can be determined by the
linguistic features of text. According to Pennebaker (2011), analyzing what people write or say
may reflect the ways people think or connect with others. Distinguishing the content of what
people are saying from how they are saying it, namely language style, his work emphasizes
analyzing different style of language or word use in predicting the personalities or psychological
state. Scholars have established a variety of different categories of language style into
computerized programs, such as Linguistic Inquiry and Word Count (LIWC; Pennebaker, Booth,
& Francis, 2007). A number of empirical studies have used this approach to analyze the language
style in text from a wide range of fields such as in marketing (Berger et al., 2020), clinical
psychology (Blackburn, Wang, Pedler, Thompson, & Gonzales, 2021), advertising (Aleti,
47
Pallant, Tuan, & van Laer, 2019), and communication (Gonzales, Hancock, & Pennebaker,
2010).
One of the language styles that pertains to the current study is the use of personal
pronouns – more specifically, the use of first-person pronouns. Findings from Pennebaker
(2011)’s seminal work suggest that the frequent use of words such as I and me indicate being
self-reflective and is associated with being honest and personal. In fact, how much personal a
text is has long been measured by the use of pronouns: for example, pronoun use was measured
to distinguish genres of literature such that romance novels were more personal than mysteries
(Biber, 1988) or to identify self-disclosures in online discussion board (Chang & Bazarova,
2016) and self-referencing messages in health blogs (Rains, 2014). Based on this reasoning of
greater use of first-person pronouns as an indication of more personal stories, I propose to test if
the effectiveness of such personalization is observed as tie formation in a communication
network:
H1: The more personalized a social media message is, the more likely it is to form a tie
with other messages.
Power of Affective Stories in Communication Network
Scholarly attention has been paid to identifying what factors influence engagement with
social media messages. One of the major findings is that messages containing a demonstrable
affective tone tend to be shared more widely and more quickly than neutral ones (Stieglitz &
Dang-Xuan, 2013). Relatedly, emotional broadcaster theory (Harber & Cohen, 2005) posits that
stories that carry greater emotional impact are more likely to be shared, resulting in both emotion
and information traveling across social networks. Similarly, the secondary social sharing
48
approach (Christophe & Rimé , 1997) argues that emotionally evocative events shared with
others tend to be further shared to other sets of people. For instance, a content analysis of the
most e-mailed The New York Times stories revealed that stories with hopeful tones were most
likely to go viral (Berger & Milkman, 2012). In a study on obesity-related messages on Twitter
(So et al., 2016), tweets that evoked strong emotion were found to be more frequently retweeted
than their counterparts that did not.
However, findings have been mixed in terms of whether it is negative or positive tone
that makes a difference in sharing. For instance, Meng and colleagues (2018) suggest
practitioners use loss frames or words that convey negative sentiments to boost the spread of
information on Twitter. Notably, it was rather a mixed emotion such as feelings of compassion
that was associated with information sharing (Myrick & Oliver, 2015). In the context of mental
health related stories, messages addressing symptoms, troubling experiences, struggles, and
hardships are likely to be considered as having negative tone but also likely to contain a positive
tone when discussing empowerment and efficacy. Therefore, instead of distinguishing these two,
the focus of this research is on whether the effects of affective tone on engagement are realized
as tie formation in a communication network. The following hypothesis was proposed:
H2: The greater the affective tone a social media message exhibits, the more likely it is to
form a tie with other messages.
Homophily Effects on Message Engagement in Communication
Network
Research has shown that homophily is an important force that drives tie formation in a
network between the actors that share the same identity or characteristics (Xu & Zhou, 2020).
49
Homophily effects, when it comes to the messages as focal nodes, suggest that the messages are
connected based on similar content. Extending this to the campaign context, such homophily
effects may indicate a successful interpersonal outcome that campaign messages promoted
relevant conversation among the audience members. Therefore, I propose two research questions
to explore if such homophily effects of personalization and affective tone of a message have an
impact on its tie formation. In particular, homophily effects of personalization will demonstrate
the self-referencing engagement, that is, a personal story brings about another personal story:
RQ1: Will the homophily effects of personalization be observed in the tie formation?
RQ2: Will the homophily effects of affective tone be observed in the tie formation?
Method
Study Background
To examine the social media communication network on the topic of mental health
awareness, this project took the case study of Twitter communication during the Mental Health
Awareness Week, which is the first week of October. The Mental Health Awareness Week was
first established in 1990 by the National Alliance on Mental Illness to raise awareness of the
importance of mental health and wellness in the United States. Since then, every year during this
week, organizations and mental health advocates across the country including healthcare
professionals and individuals distribute related programs and campaigns to promote education
and activities that are designed to initiate conversation and combat stigma.

50
Data Collection
On Twitter, there are several different ways a communication network could be formed:
by following or mentioning other users (or accounts), retweeting or quoting others’ tweet
messages. Given that the aim of this study is to examine how linguistic features of messages
contribute to forming a tie, a node (or vertex) should be a message, not a user or an account.
Compared to retweets, which is simply relaying what others have said, people can “quote” the
other person’s tweet by adding their own remarks or comments, making it possible to study the
message characteristics of both nodes that constitute a communication tie, that is, both quoting
and quoted tweets. Therefore, a decision has been made to focus on the tweets that are connected
by quotes.
To extract the quote network, I first collected the full data of tweets made in the Mental
Health Awareness Week of 2018, using Twitter API v2. In 2021, Twitter launched the academic
research product track which allows academic researchers to have access to the entire archive of
tweets dating back to 2006, upon approval of each project application. Using the search query of
‘mental health OR #MentalHealthWeek OR #MentalHealthAwarenessWeek OR
#MentalHealthAwareness’ as search terms and English as language used while excluding the
retweets (i.e., mere duplicates of original tweet message), full tweet archive data in the first week
of October 2018 (October 1 to October 7) were accessed (Ntotal tweets = 334,988) and only those
tweets connected by quote features were included in the analysis (nquote tweets = 9747). The Twitter
data included user- and tweet-level data. The user-level data included the information related to
the author of tweets such as number of following, followers, and total tweets made until the
specified end date of the data. The tweet-level data included the body of text as well as the
engagement metrics such as retweets, likes, and quotes.
51

Figure 5.1. Tweet counts in the mental health week of 2018

Network Construction
The quote network was extracted with the following procedure. In this network, a tie (i to
j) is made when a tweet (i) is quoting the other tweet (j), making it a directed graph where a node
is each tweet and an edge represents a quote relation. Accordingly, an edge list with source (i)
and target (j) was created where the source was the tweet message quoting the other tweet
message and the original quoted tweet became the target. A separate dataset was made for a node
list which contain unique values of the attributes for each tweet. The quote network consisted of
9747 nodes and was connected by 5546 edges.
52

Figure 5.2. Quote network

Analytical Procedures
For analysis, I took a few steps before testing the hypotheses. First, to obtain the
measures of personalization and affective tone, LIWC 2022 (Linguistic Inquiry and Word Count;
Boyd et al., 2022) was used. LIWC is a computerized, dictionary-based linguistic analytic
software program to detect a wide range of linguistic categories by calculating the percentage of
specific category in relation to the total number of words within a text. For any given text, LIWC
generates variables that represent the percentage of words associated with each dictionary (a
detailed explanation can be found in the Method section of the previous study). Then I used an
ERGM (Exponential Random Graph Model) to test the hypotheses. ERGM provides a way to
predict the presence and absence of ties as a function of individual covariates or network
structure (Hunter, Handcock, Butts, Goodreau, & Morris, 2008). That is, ERGM estimates and
53
evaluates the effects of exogenous covariates that are associated with the nodes and edges (i.e.,
local factors) or endogenous network structure (i.e., structural factors) and whether they do so in
a way that is significantly different from random chance. All the analyses were performed using
R.
Measures
Personalization. Following the previous studies that used the word count approach to
personal pronouns for self-reference or self-disclosure measures (Chang & Bazarova, 2016;
Rains, 2014), the extent to which a tweet message is personalized was operationalized as the use
of first-person singular pronouns. The aggregated proportion of words corresponding to the
category of “first-person singular” (e.g., I, I’d, me, my, myself) were machine-coded using
LIWC. Higher score indicated a greater personalization (M = 1.36, SD = 2.84).
Affective tone. The affective tone of a given tweet message was evaluated using LIWC.
Words that were classified as exhibiting affective tone by the LIWC category include good, well,
love, hope, happy, hurt, afraid, fear, frustr*, disappoint*, etc. A higher score indicated greater
affective tone present in a given tweet (M = 27.64, SD = 34.67).
Control variables. A set of measures capturing the account (user-level) characteristics
that may exert confounding effects on the relationship between tweet-level characteristics (i.e.,
message style and tone) and tie formation were included as control variables. To rule out the
potential effects of tendency that the author of tweet (i) is active user who follows many other
users and frequently makes tweets in quoting other tweets (j), the author of each tweet’s number
of following (M = 1967.41, SD = 6788.94) and the total number of tweets (M = 34674.06, SD =
80954.2) during the time period were measured to be controlled.
54
Results
Predictive models were built and assessed using ERGM. In the first model, edges and
other control variables were included. First, the baseline model with the edge term estimated the
baseline tendency of tie occurrence, indicating the marginal probability of two nodes being
connected by an edge (i.e., quote). This is also kwon as density (i.e., probability of edge
formation) and is often used in a similar way as using an intercept term in linear regression
models. For example, the negative coefficient of edges in the model (b = -9.768e+00, p < .001)
indicates that a tie is more likely not to form than form in the network. Then the other control
variables were entered. Results showed that the author’s following number and the author’s total
tweet number were positively associated with the likelihood of forming ties by quoting others.
Notably, coefficients of ERGM are the change in the (log-odds) likelihood of a tie for a
unit change in a predictor. To make the interpretation of the logged-odds coefficients more
intuitive, I calculated the inverse-logit to convert them into the probabilities of two nodes being
connected by an edge. The log-odds of the effects of number of following (b = 3.263e-06, p =
0.023) and total number of tweets (b = 4.271e-07, p = 0.003) were translated to the probabilities
of .50s for the presence of ties in the network. In other words, the greater the number of
following and the total tweets were made by the author of a tweet, the more likely their tweet is
to be quoted at a probability of 50%.
In the second model, main variables of interest were added. Results showed that tweets’
personalization scores were positively associated with the likelihood of quote tie formation (b =
2.351e-02, p < .001). That is, the greater personalization a tweet exhibited, the more likely it was
to be quoted with a probability of 50.6%, supporting H1. Before combined into the affective
55
tone, positive affect and negative affect were examined with a separate model for comparison
purposes. As expected, both positive (b = 9.794e-03, p < .001) and negative affect (b = 2.179e-
02, p < .001) were associated with a tweet’s greater likelihood of being quoted. Then the model
for hypotheses testing with affective tone revealed that tweets with greater affective tone were
more likely to be quoted (b = 1.315e-02, p < .001) with a probability of 50.3%, supporting H2.
To test the homophily effects of personalization and affective tone, absdiff() ERGM terms were
added for both personalization and affective tone scores, respectively. Results revealed
significant negative coefficients, meaning that the greater likelihood of forming ties can be
accounted by smaller differences in two tweets’ personalization (b = -1.241e-02, p = .020,
49.7%) and affective tone (b = -1.173e-02, p < .001, 49.7%; absdiff(positive tone) b = -9.058e-
03, p < .001; absdiff(negative tone) b = -1.594e-02, p < .001). Given this network is directed
with quoted tweets (j) being preceded by quoting tweets (i), the significant results indicate that
personalized tweets were likely to bring about a similar level of personalization through
communication tie formation (i.e., quote) (RQ1) and tweets that contained affective tone were
also likely to be quoted by bringing about similarly affective tone (RQ2).
To ensure the model was well-specified to be a good representation of the observed
network thus the estimates can be trusted, the model degeneracy was assessed (i.e., whether
model estimation fails converging; Goodreau, 2007). As can be seen from the MCMC trace plots
(Figure 11), values from the sampling procedure follow a bell-curve, indicating a good
convergence. It was also evaluated by the goodness-of-fit model statistics, which basically
simulate networks from the ERGM estimates and compare the distribution to the observed
values. The results are demonstrated in Figure 12 where the black bold lines represent the values
from the observed network while the box plots reflect the distribution of statistics for networks
56
simulated from the model. It shows a good fit with the observed values only overlapping with the
grey area but also being close to the mean of simulated networks within 95% confidence
intervals (pedges = .84, pfollowing = 1.00, ptweets = .92, p1st.per = .92, pabsdiff.1st.per = .98, paffect = .90,
pabsdiff.affect = .84, where the p-value closer to 1 is better, meaning that simulated networks are
similar to the observed one).

Figure 5.3. MCMC-based degeneracy diagnostics

Figure 5.4. Goodness-of-fit plots for ERGM

57
Discussion
This study set out to complement the Study 1 to further investigate the use of
personalization and affective tone for communication tie formation as an indicator of social
media campaign effectiveness. It was first operationalized with two different measures: 1) tie
formation (i.e., being quoted) and 2) homophily effects (i.e., self-referencing engagement).
Taking individual messages from a quote network as focal nodes, this study empirically and
computationally tested the effects of message characteristics on the likelihood of being engaged
while inviting personal stories.
First, this study supported the findings from Study 1 that the use of personal stories is
effective in promoting engagement on social media. Despite the word limit in a single unit of
message on Twitter, the extent to which each message was personalized was obtained through
the computerized program by calculating the percentage of first-person pronouns. As expected,
messages with greater personalization were likely to invite a greater engagement (i.e., quoted by
others) compared to less personalized ones. Additionally, the tone of message was also a
significant predictor of the likelihood of tie formation: the more affective tone a message
exhibits, the more likely it invites engagement (i.e., quoted by others).
In addition to the benefits of using personalization and affective tone in a message, the
findings of homophily effects further demonstrated the similarity in the level of personalization
and affective tone used between the connected messages (i.e., quoted tweet and quoting tweet).
Given the directed network, the observed homophily effects corroborate the causality that the
greater use of personalization and affective tone of a message invited other messages with a
similar level of personalization and affective tone. The personalization of the original quoted
58
message in this study represents the personal story in a social media format that Study 1
conceptualized while the personalization of the quoting message in this study matches with the
self-referencing engagement that Study 1 conceptualized. Therefore, the finding supports the
overarching argument of this dissertation project that the power of personal stories comes from
its power in eliciting other personal stories, while adding that tone also matters.
Extending the findings, future research may explore what other factors in addition to or in
combination with the first-person pronouns and affective tone may differently influence the
effects of social media personal stories on tie formation. For instance, given the focus of this
study on the message-specific features, whose story it is may worth consideration. Social media
communication provides information about who the speaker is – whether it is an individual,
organization, news outlet, or public figure – thus whose personal stories are more or less
effective in tie formation or self-referencing engagement could be further examined. From the
health campaign perspective, identifying influential storytellers and their positions in the
communication network will add insightful contributions to effective health campaign design
and evaluation. Additionally, the current study only considered the presence and absence of tie in
the tie formation. Future studies may focus on weighted ties so that the number of quotes sending
or receiving are taken into consideration.

59
CHAPTER 6: CONCLUSION
The goal of this project was theoretically to extend the study of narrative into the social
media environment and methodologically to advance the approach to the study of social media
health campaign effectiveness. While there is ample research on the power of social media as a
primary communication channel and of narrative for public health campaigns, little work has
done to combine them to explicate the use of social media for narrative effects. In light of the
“masspersoanl communication” (O’Sullivan & Carr, 2018) that social media afford, this project
investigated what makes social media an effective anti-stigma campaign tool in terms of
achieving both persuasion and engagement outcomes.
In doing so, I proposed and tested an explanatory mechanism underlying the effects of
public figure’s social media mental health story. In addition, I proposed a new concept of self-
referencing engagement which refers to the engagement with the message while expressing one’s
self-referencing thoughts by sharing own story, as an important outcome of social media health
campaigns. To empirically test how it works, I used computational approach to both measuring
the self-referencing engagement (i.e., quantitatively analyzed the open-ended responses to
personal stories as well as the real-world social media text data) and effects of it (i.e., social
network analysis of communication tie formation). It is my hope that this methodological
approach and the theoretical construct of self-referencing engagement would contribute to
expanding the understanding of narrative’s effects on persuasion and engagement in the social
media context.
In Study 1, the findings highlight the benefits of using public figure’s personal stories in a
social media format to promote positive attitudes toward help-seeking for mental health and
60
behavioral intentions to seek professional and informal help for mental health care. These effects
were explained by enhanced social presence elicited from the social media story, which was
translated into a greater level of story involvement (i.e., transportation) that influenced the
persuasion outcomes. Parasocial intimacy further explained the relationship such that the effects
of transportation on the attitudes became more salient when the story was from a public figure
who was perceived to be close. Moreover, personal stories in a social media format invited
greater self-referencing engagement, constituting the power of social media personal stories. In
Study 2, such effects of social media personal stories were further supported. That is, the greater
personalization a message exhibited, the more likely it was to receive engagement on Twitter.
Moreover, a similar level of personalization in response to personalized tweets were observed in
the form of communication network (i.e., quote). Same pattern was observed in the use of
affective tone and homophily effects on communication tie formation.
Given that an increasing number of health campaign efforts are made to create and
communicate in the mediated environment especially through social media, findings from Study
1 highlight the utility of social media and a public figure’s personal stories as an effective health
campaign tool, while shedding some light on the expanded understanding of the campaign
effectiveness into engagement beyond persuasion. In sum, it is personal stories that are not only
persuasive but also inviting other personal stories that matters in making successful social media
health campaigns. Future health communication researchers and practitioners may pay more
attention to using public figure or celebrity advocacy into anti-stigmatizing health campaigns by
focusing on the potential of public figures’ direct use of social media to narrate their personal
health story.
61
Findings from Study 2 have important implications with regards to anti-stigma health
campaigns. In raising mental health awareness and normalizing the conversation about the
stigmatized issue of mental health, audience members’ participatory engagement with campaign
messages would be the first goal to be achieved. Moreover, if such engagement involves the
participants’ sharing own personal stories, it is not mere a passive reaction or response to the
campaign message, but at the same time, it has its own power to further instigate other personal
stories, contributing to an important step towards normalizing the open conversation about the
stigmatized issue. After all, such collective sharing of personal stories essentially represents the
self-referencing engagement that has been emphasized throughout this dissertation project. These
findings highlight that the effectiveness of using social media for public health campaigns not
only lies in reaching a broader mass audience and facilitating information dissemination (i.e.,
mass communication) but also in promoting self-referencing engagement (i.e., interpersonal
communication).

62
REFERENCES
Aleti, T., Pallant, J. I., Tuan, A., & Van Laer, T. (2019). Tweeting with the stars: Automated text
analysis of the effect of celebrity social media communications on consumer word of
mouth. Journal of Interactive Marketing, 48(1), 17-32.
Appel, M., Gnambs, T., Richter, T., & Green, M. C. (2015). The transportation scale–short form
(TS–SF). Media psychology, 18(2), 243-266.
Atkin, C. K., Rice, R. E., & Valdivia, A. N. (2012). Advances in public communication
campaigns. In C. K. Atkin & R. E. Rice (Eds.), The international encyclopedia of media
studies (pp. 526–551). London, UK: Wiley-Blackwell.
Bandura, A. (2009). Social cognitive theory of mass communication. In Media effects (pp. 110-
140). Routledge.
Banerjee, S. C., & Greene, K. (2012). Role of transportation in the persuasion process: Cognitive
and affective responses to antidrug narratives. Journal of Health Communication, 17(5),
564-581.
Bantum, E. O. C., & Owen, J. E. (2009). Evaluating the validity of computerized content
analysis programs for identification of emotional expression in cancer
narratives. Psychological Assessment, 21(1), 79.
Barbour, J. B., Doshi, M. J., & Hernández, L. H. (2016). Telling global public health stories:
Narrative message design for issues management. Communication Research, 43(6), 810-
843.
Barry, C. L., Brescoll, V. L., & Gollust, S. E. (2013). Framing childhood obesity: How
individualizing the problem affects public support for prevention. Political
Psychology, 34(3), 327-349.
63
Beck, C. S., Aubuchon, S. M., McKenna, T. P., Ruhl, S., & Simmons, N. (2014). Blurring
personal health and public priorities: An analysis of celebrity health narratives in the
public sphere. Health Communication, 29(3), 244-256.
Bente, G., Rüggenberg, S., Krämer, N. C., & Eschenburg, F. (2008). Avatar-mediated
networking: Increasing social presence and interpersonal trust in net-based
collaborations. Human Communication Research, 34(2), 287-318.
Berger, J., & Milkman, K. L. (2012). What makes online content viral? Journal of Marketing
Research, 49(2), 192-205.
Berger, J., Humphreys, A., Ludwig, S., Moe, W. W., Netzer, O., & Schweidel, D. A. (2020).
Uniting the tribes: Using text for marketing insight. Journal of Marketing, 84(1), 1-25.
Berry, N., Lobban, F., Belousov, M., Emsley, R., Nenadic, G., & Bucci, S. (2017).
#WhyWeTweetMH: Understanding why people use Twitter to discuss mental health
problems. Journal of Medical Internet Research, 19(4).
Betton, V., Borschmann, R., Docherty, M., Coleman, S., Brown, M., & Henderson, C. (2015).
The role of social media in reducing stigma and discrimination. The British Journal of
Psychiatry, 206(6), 443-444.
Biber, D. (1988). Variation across speech and writing. Cambridge University Press.
Biocca, F., Harms, C., & Burgoon, J. K. (2003). Toward a more robust theory and measure of
social presence: Review and suggested criteria. Presence: Teleoperators & virtual
environments, 12(5), 456-480.
Blackburn, K. G., Wang, W., Pedler, R., Thompson, R., & Gonzales, D. (2021). Linguistic
markers in women’s discussions on miscarriage and abortion illustrate psychological
64
responses to their experiences. Journal of Language and Social Psychology, 40(3), 398-
411.
Bou-Franch, P., Lorenzo-Dus, N., & Blitvich, P. G. C. (2012). Social interaction in YouTube
text-based polylogues: A study of coherence. Journal of Computer-Mediated
Communication, 17(4), 501-521.
Boyd, R. L., Ashokkumar, A., Seraj, S., & Pennebaker, J. W. (2022). The development and
psychometric properties of LIWC-22. Austin, TX: University of Texas at Austin.
Boysen, G. A., & Vogel, D. L. (2008). Education and mental health stigma: The effects of
attribution, biased assimilation, and attitude polarization. Journal of Social and Clinical
Psychology, 27(5), 447-470.
Braverman, J. (2008). Testimonials versus informational persuasive messages: The moderating
effect of delivery mode and personal involvement. Communication Research, 35(5), 666-
694.
Brosius, H. B., & Bathelt, A. (1994). The utility of exemplars in persuasive
communications. Communication Research, 21(1), 48-78.
Brosius, H. B. (1999). Research note: The influence of exemplars on recipients' judgements: the
part played by similarity between exemplar and recipient. European Journal of
Communication, 14(2), 213-224.
Brown, W. J., & Basil, M. D. (1995). Media celebrities and public health: Responses
to'Magic'Johnson's HIV disclosure and its impact on AIDS risk and high-risk
behaviors. Health Communication, 7(4), 345-370.
Burgoon, J. K., Bonito, J., Ramirez, Jr., A., Dunbar, N. E., Kam, K., & Fischer, J. (2002).
Testing the interactivity principle: Effects of mediation, propinquity, and verbal and
65
nonverbal modalities in interpersonal interaction. Journal of Communication, 52, 657–
677.
Burnkrant, R. E., & Unnava, H. R. (1995). Effects of self-referencing on persuasion. Journal of
Consumer Research, 22(1), 17-26.
Busselle, R., & Bilandzic, H. (2009). Measuring narrative engagement. Media
Psychology, 12(4), 321-347.
CDC (2021). Health Equity Considerations for Developing Public Health Communications.
https://www.cdc.gov/healthcommunication/Comm_Dev.html
Chaiken, S., & Eagly, A. H. (1983). Communication modality as a determinant of persuasion:
The role of communicator salience. Journal of Personality and Social Psychology, 45(2),
241.
Chang, P. F., & Bazarova, N. N. (2016). Managing stigma: Disclosure-response communication
patterns in pro-anorexic websites. Health Communication, 31(2), 217-229.
Chen, M., Bell, R. A., & Taylor, L. D. (2016). Narrator point of view and persuasion in health
narratives: The role of protagonist–reader similarity, identification, and self-
referencing. Journal of Health Communication, 21(8), 908-918.
Chen, M., McGlone, M. S., & Bell, R. A. (2015). Persuasive effects of linguistic agency
assignments and point of view in narrative health messages about colon cancer. Journal of
Health Communication, 20(8), 977-988.
Chou, W. Y. S., Hunt, Y. M., Beckjord, E. B., Moser, R. P., & Hesse, B. W. (2009). Social
media use in the United States: implications for health communication. Journal of Medical
Internet Research, 11(4).
66
Christophe, V., & Rimé, B. (1997). Exposure to the social sharing of emotion: Emotional impact,
listener responses and secondary social sharing. European Journal of Social
Psychology, 27(1), 37-54.
Chung, S., & Cho, H. (2017). Fostering parasocial relationships with celebrities on social media:
Implications for celebrity endorsement. Psychology & Marketing, 34(4), 481-495.
Chung, A. H., & Slater, M. D. (2013). Reducing stigma and out-group distinctions through
perspective-taking in narratives. Journal of Communication, 63(5), 894-911.
Cialdini, R. B., Brown, S. L., Lewis, B. P., Luce, C., & Neuberg, S. L. (1997). Reinterpreting the
empathy–altruism relationship: When one into one equals oneness. Journal of Personality
and Social Psychology, 73(3), 481.
Cohen, J. (2001). Defining identification: A theoretical look at the identification of audiences
with media characters. Mass Communication and Society, 4, 245–264.
Cohn, M. A., Mehl, M. R., & Pennebaker, J. W. (2004). Linguistic markers of psychological
change surrounding September 11, 2001. Psychological science, 15(10), 687-693.
Corrigan, P. (2004). How stigma interferes with mental health care. American
Psychologist, 59(7), 614.
Corrigan, P. W., & Shapiro, J. R. (2010). Measuring the impact of programs that challenge the
public stigma of mental illness. Clinical Psychology Review, 30(8), 907-922.
Corrigan, P. W., Morris, S. B., Michaels, P. J., Rafacz, J. D., & Rüsch, N. (2012). Challenging
the public stigma of mental illness: a meta-analysis of outcome studies. Psychiatric
Services, 63(10), 963-973.
Corrigan, P. W., Kosyluk, K. A., & Rüsch, N. (2013). Reducing self-stigma by coming out
proud. American Journal of Public Health, 103(5), 794-800.
67
Dai, Y., & Walther, J. B. (2018). Vicariously experiencing parasocial intimacy with public
figures through observations of interactions on social media. Human Communication
Research, 44(3), 322-342.
Dean, M. (2016). Celebrity health announcements and online health information seeking: an
analysis of Angelina Jolie’s preventative health decision. Health Communication, 31(6),
752-761.
de Graaf, A., Hoeken, H., Sanders, J., & Beentjes, J. W. J. (2012). Identification as a mechanism
of narrative persuasion. Communication Research, 39, 802–823.
De Wit, J. B., Das, E., & Vet, R. (2008). What works best: objective statistics or a personal
testimonial? An assessment of the persuasive effects of different types of message
evidence on risk perception. Health Psychology, 27(1), 110.
Duke, J. C., Hansen, H., Kim, A. E., Curry, L., & Allen, J. (2014). The use of social media by
state tobacco control programs to promote smoking cessation: a cross-sectional
study. Journal of Medical Internet Research, 16(7).
Dunlop, S. M., Wakefield, M., & Kashima, Y. (2010). Pathways to persuasion: Cognitive and
experiential responses to health-promoting mass media messages. Communication
Research, 37(1), 133-164.
Entwistle, V. A., France, E. F., Wyke, S., Jepson, R., Hunt, K., Ziebland, S., Thompson, A.
(2011). How information about other people’s personal experiences can help with
healthcare decision-making: A qualitative study. Patient Education and Counseling 85,
e291–e298.
68
Families for Depression Awareness (2018). Dwayne “The Rock” Johnson, Michael Phelps, and
Mariah Carey: Celebrities Speaking Out about Depression and Bipolar Disorder.
http://familyaware.org/celebrities-speak-out/
Gil-Lopez, T., Shen, C., Benefield, G. A., Palomares, N. A., Kosinski, M., & Stillwell, D.
(2018). One size fits all: Context collapse, self-presentation strategies and language styles
on Facebook. Journal of Computer-Mediated Communication, 23(3), 127-145.
Gonzales, A. L., Hancock, J. T., & Pennebaker, J. W. (2010). Language style matching as a
predictor of social dynamics in small groups. Communication Research, 37(1), 3-19.
Goodreau, S. M. (2007). Advances in exponential random graph (p*) models applied to a large
social network. Social Networks, 29(2), 231-248.
Green, M. C., & Brock, T. C. (2000). The role of transportation in the persuasiveness of public
narratives. Journal of Personality and Social Psychology, 79, 701–721.
Green, M. C., Kass, S., Carrey, J., Herzig, B., Feeney, R., & Sabini, J. (2008). Transportation
across media: Repeated exposure to print and film. Media Psychology, 11(4), 512-539.
Harber, K. D., & Cohen, D. J. (2005). The emotional broadcaster theory of social
sharing. Journal of Language and Social Psychology, 24(4), 382-400.
Hayes, A. F. (2013). Introduction to mediation, moderation, and conditional process analysis: A
regression-based approach. Guilford Press.
Henderson, C., Evans-Lacko, S., & Thornicroft, G. (2013). Mental illness stigma, help seeking,
and public health programs. American journal of public health, 103(5), 777-780.
Hoffner, C. A., & Cohen, E. L. (2015). Portrayal of mental illness on the TV series Monk:
Presumed influence and consequences of exposure. Health Communication, 30(10), 1046-
1054.
69
Horton, D., & Wohl, R. (1956). Mass communication and para-social interaction: Observations
on intimacy at a distance. Psychiatry, 19(3), 215-229.
Hunt, J., & Eisenberg, D. (2010). Mental health problems and help-seeking behavior among
college students. Journal of Adolescent Health, 46(1), 3-10.
Hunter, D. R., Handcock, M. S., Butts, C. T., Goodreau, S. M., & Morris, M. (2008). ergm: A
package to fit, simulate and diagnose exponential-family models for networks. Journal of
Statistical Software, 24(3), nihpa54860.
Igartua, J. J., & Frutos, F. J. (2017). Enhancing attitudes toward stigmatized groups with movies:
Mediating and moderating processes of narrative persuasion. International Journal of
Communication, 11, 20.
Jensen, J. D. (2008). Scientific uncertainty in news coverage of cancer research: Effects of
hedging on scientists’ and journalists’ credibility. Human Communication
Research, 34(3), 347-369.
Jin, S. A. A., & Park, N. (2009). Parasocial interaction with my avatar: Effects of interdependent
self-construal and the mediating role of self-presence in an avatar-based console game,
Wii. CyberPsychology & Behavior, 12(6), 723-727.
Jin, S. A. A. (2012). “Toward integrative models of flow”: Effects of performance, skill,
challenge, playfulness, and presence on flow in video games. Journal of Broadcasting &
Electronic Media, 56(2), 169-186.
Jorm, A. F., Korten, A. E., Jacomb, P. A., Christensen, H., Rodgers, B., & Pollitt, P. (1997).
“Mental health literacy”: a survey of the public's ability to recognise mental disorders and
their beliefs about the effectiveness of treatment. Medical Journal of Australia, 166(4),
182-186.
70
Jorm, A. F. (2012). Mental health literacy: Empowering the community to take action for better
mental health. American Psychologist, 67(3), 231.
Kalichman, S. C., & Hunter, T. L. (1992). The disclosure of celebrity HIV infection: its effects
on public attitudes. American Journal of Public Health, 82(10), 1374-1376.
Kaufman, G. F., & Libby, L. K. (2012). Changing beliefs and behavior through experience-
taking. Journal of Personality and Social Psychology, 103(1), 1.
Klik, K. A., Williams, S. L., & Reynolds, K. J. (2019). Toward understanding mental illness
stigma and help-seeking: A social identity perspective. Social Science & Medicine, 222,
35-43.
Kim, E., Hou, J., Han, J. Y., & Himelboim, I. (2016). Predicting retweeting behavior on breast
cancer social networks: Network and content characteristics. Journal of Health
Communication, 21(4), 479-486.
Kim, H. S., Bigman, C. A., Leader, A. E., Lerman, C., & Cappella, J. N. (2012). Narrative health
communication and behavior change: The influence of exemplars in the news on intention
to quit smoking. Journal of Communication, 62(3), 473-492.
Kim, J., & Song, H. (2016). Celebrity's self-disclosure on Twitter and parasocial relationships: A
mediating role of social presence. Computers in Human Behavior, 62, 570-577.
Knoll, J., & Matthes, J. (2017). The effectiveness of celebrity endorsements: a meta-
analysis. Journal of the Academy of Marketing Science, 45(1), 55-75.
Kosenko, K. A., Binder, A. R., & Hurley, R. (2016). Celebrity influence and identification: a test
of the Angelina effect. Journal of Health Communication, 21(3), 318-326.
71
Kreuter, M. W., Green, M. C., Cappella, J. N., Slater, M. D., Wise, M. E., Storey, D., & Hinyard,
L. J. (2007). Narrative communication in cancer prevention and control: a framework to
guide research and application. Annals of Behavioral Medicine, 33(3), 221-235.
Kreuter, M. W., Buskirk, T. D., Holmes, K., Clark, E. M., Robinson, L., Si, X., ... & Mathews,
K. (2008). What makes cancer survivor stories work? An empirical study among African
American women. Journal of Cancer Survivorship, 2(1), 33-44.
Kreuter, M. W., & Wray, R. J. (2003). Tailored and targeted health communication: strategies
for enhancing information relevance. American Journal of Health Behavior, 27(1), S227-
S232.
Kreps, G. L. (2017). Online information and communication systems to enhance health outcomes
through communication convergence. Human Communication Research, 43(4), 518-530.
Lee, K. M., & Nass, C. (2005). Social-psychological origins of feelings of presence: Creating
social presence with machine-generated voices. Media Psychology, 7(1), 31-45.
Lee, E. J. (2012). Effectiveness of politicians' soft campaign on Twitter versus TV: Cognitive
and experiential routes. Journal of Communication, 63(5), 953-974.
Lee, E. J., & Shin, S. Y. (2014). When the medium is the message: How transportability
moderates the effects of politicians’ Twitter communication. Communication
Research, 41(8), 1088-1110.
Lee, S. Y. (2018). The effect of media coverage of celebrities with panic disorder on the health
behaviors of the public. Health Communication, 1-11.
Link, B. G., & Phelan, J. C. (2001). Conceptualizing stigma. Annual review of Sociology, 27(1),
363-385.
72
Lombard, M., & Ditton, T. (1997). At the heart of it all: The concept of presence. Journal of
Computer-Mediated Communication, 3(2).
Major, L. H. (2018). Mental health news: How frames influence support for policy and civic
engagement intentions. Journal of Health Communication, 23(1), 52-60.
Mar, R. A., & Oatley, K. (2008). The function of fiction is the abstraction and simulation of
social experience. Perspectives on Psychological Science, 3(3), 173-192.
McCracken, G. (1989). Who is the celebrity endorser? Cultural foundations of the endorsement
process. Journal of Consumer Research, 16(3), 310-321.
McLaughlin, M., Nam, Y., Gould, J., Pade, C., Meeske, K. A., Ruccione, K. S., & Fulk, J.
(2012). A videosharing social networking intervention for young adult cancer
survivors. Computers in Human Behavior, 28(2), 631-641.
McNeill, A. R., & Briggs, P. (2014, April). Understanding Twitter influence in the health
domain: a social-psychological contribution. In Proceedings of the 23rd International
Conference on World Wide Web (pp. 673-678). ACM.
McQueen, A., Kreuter, M. W., Kalesan, B., & Alcaraz, K. I. (2011). Understanding narrative
effects: the impact of breast cancer survivor stories on message processing, attitudes, and
beliefs among African American women. Health Psychology, 30(6), 674.
Meng, J., Peng, W., Tan, P. N., Liu, W., Cheng, Y., & Bae, A. (2018). Diffusion size and
structural virality: The effects of message and network features on spreading health
information on twitter. Computers in Human Behavior, 89, 111-120.
Merolli, M., Gray, K., & Martin-Sanchez, F. (2013). Health outcomes and related effects of
using social media in chronic disease management: a literature review and analysis of
affordances. Journal of Biomedical Informatics, 46(6), 957-969.
73
Metcalfe, D., Price, C., & Powell, J. (2011). Media coverage and public reaction to a celebrity
cancer diagnosis. Journal of Public Health, 33(1), 80-85.
Moyer‐Gusé, E. (2008). Toward a theory of entertainment persuasion: Explaining the persuasive
effects of entertainment‐education messages. Communication Theory, 18(3), 407-425.
Murphy, S. T., Frank, L. B., Moran, M. B., & Patnoe-Woodley, P. (2011). Involved, transported,
or emotional? Exploring the determinants of change in knowledge, attitudes, and behavior
in entertainment-education. Journal of Communication, 61(3), 407-431.
Murphy, S. T., Frank, L. B., Chatterjee, J. S., & Baezconde-Garbanati, L. (2013). Narrative
versus nonnarrative: The role of identification, transportation, and emotion in reducing
health disparities. Journal of Communication, 63(1), 116-137.
Myrick, J. G. (2018). An experimental test of the roles of audience involvement and message
frame in shaping public reactions to celebrity illness disclosures. Health Communication,
1-9.
Myrick, J. G., & Oliver, M. B. (2015). Laughing and crying: Mixed emotions, compassion, and
the effectiveness of a YouTube PSA about skin cancer. Health Communication, 30(8),
820-829.
Myrick, J. G., Noar, S. M., Willoughby, J. F., & Brown, J. (2014). Public reaction to the death of
Steve Jobs: Implications for cancer communication. Journal of Health
Communication, 19(11), 1278-1295.
Naslund, J. A., Grande, S. W., Aschbrenner, K. A., & Elwyn, G. (2014). Naturally occurring
peer support through social media: the experiences of individuals with severe mental
illness using YouTube. PLOS one, 9(10), e110171.
74
Naslund, J. A., Aschbrenner, K. A., Marsch, L. A., & Bartels, S. J. (2016). The future of mental
health care: peer-to-peer support and social media. Epidemiology and Psychiatric
Sciences, 25(2), 113-122.
National Institute of Mental Health. (2016). Depression. https://www.nimh.nih.
gov/health/topics/depression/index.shtml.
Neiger, B. L., Thackeray, R., Van Wagenen, S. A., Hanson, C. L., West, J. H., Barnes, M. D., &
Fagen, M. C. (2012). Use of social media in health promotion: Purposes, key performance
indicators, and evaluation metrics. Health Promotion Practice, 13(2), 159-164.
Neuhauser, L., & Kreps, G. L. (2014). Integrating design science theory and methods to improve
the development and evaluation of health communication programs. Journal of Health
Communication, 19(12), 1460-1471.
Niederdeppe, J., Shapiro, M. A., & Porticella, N. (2011). Attributions of responsibility for
obesity: Narrative communication reduces reactive counterarguing among liberals. Human
Communication Research, 37(3), 295-323.
Noar, S. M., Willoughby, J. F., Myrick, J. G., & Brown, J. (2014). Public figure announcements
about cancer and opportunities for cancer communication: a review and research
agenda. Health Communication, 29(5), 445-461.
Oatley, K. (1999). Why fiction may be twice as true as fact: Fiction as cognitive and emotional
simulation. Review of General Psychology, 3(2), 101-117.
O'Keefe, D. J. (1999). Variability of persuasive message effects: Meta-analytic evidence and
implications. Document Design, 1(2), 87-97.
75
Oliver, M. B., Dillard, J. P., Bae, K., & Tamul, D. J. (2012). The effect of narrative news format
on empathy for stigmatized groups. Journalism & Mass Communication Quarterly, 89(2),
205-224.
O’Sullivan, P. B., & Carr, C. T. (2018). Masspersonal communication: A model bridging the
mass-interpersonal divide. New Media & Society, 20(3), 1161-1180.
Pan, J., Liu, B., & Kreps, G. L. (2018). A content analysis of depression-related discourses on
Sina Weibo: attribution, efficacy, and information sources. BMC Public Health, 18(1),
772.
Parcesepe, A. M., & Cabassa, L. J. (2013). Public stigma of mental illness in the United States:
A systematic literature review. Administration and Policy in Mental Health and Mental
Health Services Research, 40(5), 384-399.
Park, H., Reber, B. H., & Chon, M. G. (2016). Tweeting as health communication: health
organizations’ use of Twitter for health promotion and public engagement. Journal of
Health Communication, 21(2), 188-198.
Park, H., Rodgers, S., & Stemmle, J. (2013). Analyzing health organizations' use of Twitter for
promoting health literacy. Journal of Health Communication, 18(4), 410-425.
Pennebaker, J. W., Booth, R. J., & Francis, M. E. (2007). Linguistic inquiry and word count:
LIWC [Computer software]. Austin, TX: liwc. net, 135.
Pennebaker, J. W. (2011). The secret life of pronouns: What our words say about
us. Bloomsbury Press/Bloomsbury Publishing.
Pereira, L. M., Quinn, N., & Morales, E. (2016). Breaking news: “I have an eating disorder.”
Video testimonials on YouTube. Computers in Human Behavior, 63, 938-942.
76
Peters, K., Chen, Y., Kaplan, A. M., Ognibeni, B., & Pauwels, K. (2013). Social media
metrics—A framework and guidelines for managing social media. Journal of Interactive
Marketing, 27(4), 281-298.
Petronio, S., & Venetis, M. K. (2017). Communication privacy management theory and health
and risk messaging. In Oxford Research Encyclopedia of Communication.
Pew Research Center (2021). Social Media Use in 2021.
https://www.pewresearch.org/internet/2021/04/07/social-media-use-in-2021/
Powell, J., McCarthy, N., & Eysenbach, G. (2003). Cross-sectional survey of users of Internet
depression communities. BMC Psychiatry, 3(1), 19.
Quintero Johnson, J. M., Yilmaz, G., & Najarian, K. (2017). Optimizing the presentation of
mental health information in social media: the effects of health testimonials and platform
on source perceptions, message processing, and health outcomes. Health
Communication, 32(9), 1121-1132.
Rains, S. A. (2014). The implications of stigma and anonymity for self-disclosure in health
blogs. Health Communication, 29(1), 23-31.
Ramirez, Jr, A., & Burgoon, J. K. (2004). The effect of interactivity on initial interactions: the
influence of information valence and modality and information richness on computer‐
mediated interaction. Communication Monographs, 71(4), 422-447.
Ritsher, J. B., & Phelan, J. C. (2004). Internalized stigma predicts erosion of morale among
psychiatric outpatients. Psychiatry Research, 129(3), 257-265.
Rogers, E. M. (2002). Diffusion of preventive innovations. Addictive Behaviors, 27, 989–993.
doi:10.1016/S0306-4603(02)00300-3
77
Scheufele, D. A. (1999). Framing as a theory of media effects. Journal of Communication, 49(1),
103-122.
Sheffield, J. K., Fiorenza, E., & Sofronoff, K. (2004). Adolescents' willingness to seek
psychological help: Promoting and preventing factors. Journal of Youth and
Adolescence, 33(6), 495-507.
Skalski, P., & Tamborini, R. (2007). The role of social presence in interactive agent-based
persuasion. Media Psychology, 10(3), 385-413.
Slater, M. D., & Rouner, D. (2002). Entertainment—education and elaboration likelihood:
Understanding the processing of narrative persuasion. Communication Theory, 12(2), 173-
191.
Slater, D. M., Peter, J., & Valkenburg, P. M. (2015). Message variability and heterogeneity: A
core challenge for communication research. Annals of the International Communication
Association, 39(1), 3-31.
Smith, S. A., & Brunner, S. R. (2016). The great whoosh: Connecting an online personal health
narrative and communication privacy management. Health Communication, 31(1), 12-21
So, J., Prestin, A., Lee, L., Wang, Y., Yen, J., & Chou, W. Y. S. (2016). What do people like to
“share” about obesity? A content analysis of frequent retweets about obesity on
Twitter. Health Communication, 31(2), 193-206.
Southwell, B. G., & Yzer, M. C. (2009). When (and why) interpersonal talk matters for
campaigns. Communication Theory, 19(1), 1-8.
Stieglitz, S., & Dang-Xuan, L. (2013). Emotions and information diffusion in social media—
sentiment of microblogs and sharing behavior. Journal of Management Information
Systems, 29(4), 217-248.
78
Swami, V. (2012). Mental health literacy of depression: gender differences and attitudinal
antecedents in a representative British sample. PloS one, 7(11), e49779.
Tanupabrungsun, S., & Hemsley, J. (2018). Studying celebrity practices on Twitter using a
framework for measuring media richness. Social Media+ Society, 4(1),
2056305118763365.
Tausczik, Y. R., & Pennebaker, J. W. (2010). The psychological meaning of words: LIWC and
computerized text analysis methods. Journal of Language and Social Psychology, 29(1),
24-54.
Taylor, S. E., & Thompson, S. C. (1982). Stalking the elusive" vividness" effect. Psychological
Review, 89(2), 155.
Thoits, P. A. (1985). Self-labeling processes in mental illness: The role of emotional
deviance. American Journal of Sociology, 91(2), 221-249.
Today (2018). 5 celebrities who have opened up about their mental health.
https://www.today.com/video/5-celebrities-who-have-opened-up-about-their-mental-
health-1239283779685
Trope, Y., & Liberman, N. (2010). Construal-level theory of psychological
distance. Psychological Review, 117(2), 440.
van Uden-Kraan, C. F., Drossaert, C. H., Taal, E., Seydel, E. R., & van de Laar, M. A. (2009).
Participation in online patient support groups endorses patients’ empowerment. Patient
Education and Counseling, 74(1), 61-69.
Villagran, M. (2011). Methodological diversity to reach patients along the margins, in the
shadows, and on the cutting edge. Patient Education and Counseling, 82(3), 292-297.
79
Villatoro, A. P., Mays, V. M., Ponce, N. A., & Aneshensel, C. S. (2018). Perceived need for
mental health care: The intersection of race, ethnicity, gender, and socioeconomic
status. Society and Mental Health, 8(1), 1-24.
Walther, J. B. (1996). Computer-mediated communication: Impersonal, interpersonal, and
hyperpersonal interaction. Communication Research, 23(1), 3-43.
Winterbottom, A., Bekker, H. L., Conner, M., & Mooney, A. (2008). Does narrative information
bias individual's decision making? A systematic review. Social Science &
Medicine, 67(12), 2079-2088.
World Health Organization. (2020). COVID-19 disrupting mental health services in most
countries, WHO survey. https://www.who.int/news/item/05-10-2020-covid-19-disrupting-
mental-health-services-in-most-countries-who-survey.
World Health Organization. (2021). Depression. https://www.who.int/news-room/fact-
sheets/detail/depression.
World Health Organization. (2022). Mental health: Strengthening our response.
https://www.who.int/news-room/fact-sheets/detail/mental-health-strengthening-our-
response.
Xu, S., & Zhou, A. (2020). Hashtag homophily in twitter network: Examining a controversial
cause-related marketing campaign. Computers in Human Behavior, 102, 87-96.
Zillmann, D. (2002). Exemplification theory of media influence. In Media effects (pp. 29-52).
Routledge.
Zillmann, D. (2006). Exemplification effects in the promotion of safety and health. Journal of
Communication, 56, S221-S237.

80
APPENDIX

81
82
83

84

85
86
87

Abstract (if available)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Gamification + HCI + CMC: effects of persuasive video games on consumers’ mental and physical health

PDF

Relationship formation and information sharing to promote risky health behavior on social media

PDF

School-based Health Centers 2.0: Using social media to increase utilization of UMMA Wellness Center Services

PDF

Social media and health: social support and social capital on pregnancy-related social networking sites

PDF

The experience of Eritrean refugee women in addressing their mental health needs

PDF

Social norms intervention to promote help-seeking with depressed Asian and European Americans: A pilot study

PDF

How media multitasking during narrative viewing affects persuasion: the mediating roles of transportation, engagement, identification, enjoyment, and emotion

PDF

The formation and influence of online health social networks on social support, self-tracking behavior and weight loss outcomes

PDF

Three essays on behavioral economics approaches to understanding the implications of mental health stigma

PDF

Exploring mental health care utilization and academic persistence among college students: evaluating racial and stigma-related disparities

PDF

Exploring the social determinants of health in a population with similar access to healthcare: experiences from United States active-duty army wives

PDF

Mental health advocacy and navigation partnerships: the case for a community collaborative approach

PDF

Contagious: social norms about health in work group networks

PDF

The effect of Twitter social support on health outcomes and its mediators: a randomized controlled trial of a social support intervention on Twitter for patients affected by cancer

PDF

Emotion culture through the lens of PUA (pickup artist): emotions and the social media narratives of structual gendered violence

PDF

Benefits of social networking in online social support groups

PDF

Mobile self-tracking for health: validating predictors, effects, mediator, moderator, and social influence

PDF

Translating data protection into practice: exploring gaps in data privacy and cybersecurity within organizations

PDF

Cultural influences on mental health stigma in Asian and European American college students

PDF

Pregnancy in the time of COVID-19: effects on perinatal mental health, birth, and infant development

Asset Metadata

Creator Kim, Hye Min (author)

Core Title The power of social media narratives in raising mental health awareness for anti-stigma campaigns

School Annenberg School for Communication

Degree Doctor of Philosophy

Degree Program Communication

Degree Conferral Date 2022-08

Publication Date 07/25/2022

Defense Date 04/22/2022

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

Tag engagement,health campaigns,health promotion,mental health,narrative,OAI-PMH Harvest,persuasion,social media,stigma

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor McLaughlin, Margaret (committee chair), Miller, Lynn Carol (committee member), Murphy, Sheila Teresa (committee member)

Creator Email hyemink@usc.edu,hyemink1029@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC111375346

Unique identifier UC111375346

Legacy Identifier etd-KimHyeMin-10979

Document Type Dissertation

Format application/pdf (imt)

Rights Kim, Hye Min

Type texts

Source 20220728-usctheses-batch-962 (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. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

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