University of Southern California Dissertations and Theses

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

(USC Thesis Other)

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

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Running head: A SOCIAL SUPPORT INTERVENTION ON TWITTER 1

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

Yujung Nam
University of Southern California

A SOCIAL SUPPORT INTERVENTION ON TWITTER 2
Table of Contents
Abstract 6
Introduction 9
Literature Review 18
Bridging and Bonding Social Capital 18
Mediational Analysis 22
Social Support 25
Self-Efficacy 29
Method 32
Recruitment 32
Participant 33
Research Design 34
Pre-intervention Procedure 34
Random Allocation to Intervention and Control
Condition
34
Intervention, Twitter Support Group 35
Facilitation by Periodic Tweet Assignments 38
Post-Intervention Procedure 39
Participants 40
Measures 45
Bridging Social Capital 45
Bonding Social Capital 45
Social Support 46

A SOCIAL SUPPORT INTERVENTION ON TWITTER 3
Self-efficacy 46
Health 46
Participation on Twitter, Relational and Self-
Contained Tweets
47
Perceived benefits and usability of Twitter 48
Control Variables 48
Analyses 50
Missing Data 51
Results 53
Descriptive Results 53
Tweet Content Results 53
Mediators of the Effect of Twitter Intervention on Health Outcomes 57
Bridging and Bonding Social Capital 57
Self-Efficacy 66
Social Support 67
Discussion 71
Bridging and Bonding Social Capital 72
Direct, Mediated, and Total Effect of SNS Participation on Health
Outcomes
77
Self-Efficacy 81
Social Support 85
Limitations 87
Implications 90

A SOCIAL SUPPORT INTERVENTION ON TWITTER 4
Conclusion 91
References 99

Tables, Figures, and Appendices

Tables
Table 1 Demographic Characteristics of Sample 42
Table 2 Means and Standard Deviations of Key Variables and Intercorrelations
at Baseline (N =206)
55
Table 3

ANCOVA Results From Testing Effects of Variables on Mediators
With Adjusting for Baseline Scores.
63
Table 4

ANCOVA Results From Testing Mediators of the Effect of Twitter
Intervention on Health With Adjusting for Baseline Scores and Initial
Level of Dependent Variable.
64
Table 5

Mediation Estimates of the Effect of Twitter Intervention on Health
Outcome.
65

Figures
Figure 1 Flowchart of the study procedure 37
Figure 2 Image of Twitter Social Support Group 40
Figure 3 Summary of findings related to (a) H1 and H3; (b) H2 and H4; (c) H5
and H6 and; (d) H7 and H8
61
Figure 4 Changes in mean levels of health and the four mediators from baseline 70

A SOCIAL SUPPORT INTERVENTION ON TWITTER 5
to post-treatment separately for those who were allocated in the Twitter
group and those in the control group

Appendices

Appendix A Average number of monthly searches on cancer-related search
keywords and suggested bidding price for each click on Google in
September 2012
93
Appendix B List of the facilitating tweet assignments and comments 94
Appendix C Wrap-Up Handout on Coping with Caner 97

A SOCIAL SUPPORT INTERVENTION ON TWITTER 6
Abstract

This study investigates the impact of participation in patient social media on one SNS platform,
Twitter, on health outcomes among cancer patients. By conducting a randomized controlled
intervention study using a Twitter-based SNS social support program, this study empirically
evaluates whether health status differs between the cancer patients who participated in the SNS
intervention on Twitter and those who were randomly assigned to the control group. The study
also theorizes the ways in which key theoretical constructs and potential mediators, drawn from
extant literature, affect the processes and consequences of support sharing among cancer patients
on Twitter and their health outcomes.
Several traditional health and persuasive communication theories are integrated into the
conceptual design of the study, including bridging and bonding social capital (Putnam, 2000;
Williams, 2006), social support (Zimet, Dahlem, Zimet, & Farley, 1988), and self-efficacy
(Bandura, 1977). Based on the findings from literature, the study proposes that participation in
the active support seeking and providing activities on SNSs would be likely to lead to higher
bridging social capital, bonding social capital, self-efficacy, and the sense of being supported.
Heightened levels of bridging and bonding social capital, social support, and self-efficacy would
be significantly associated with better assessments of health among cancer patients who
participate in SNS activities. Furthermore, literature on the social networks, social capital,
online patient social support, and self-efficacy suggests that the effect of the SNS intervention on
the health outcomes of cancer patients would be mediated by the amount of accumulated
bridging and bonding social capital, social support, and self-efficacy.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 7
The study employed a pre- and post-intervention within-subject design with 206
participants. Each individual assigned to the social media intervention condition completed a
one month long Twitter-based social media support and networking program. The study also
examined how individuals’ preexisting levels of key variables before intervention affect the
effect of social media use on Twitter and health afterward.
A series of full-factorial univariate Analysis of Covariance (ANCOVA) tests investigated
the influences of factors related to the effectiveness of the online health intervention (i.e.,
bridging and bonding social capital, social support and self-efficacy) and baseline differences on
the results of the social media intervention and health outcomes. ANOVA tests allowed for the
comparison of the dependent variable (health outcomes) between two groups (i.e., Twitter
intervention group vs. control group) while taking the influence of continuous covariates into
account (i.e., bridging social capital, bonding social capital, social support and self-efficacy).
Participation in the Twitter-based SNS cancer social support intervention exerted a
significant impact by increasing levels of bridging social capital, providing a forum for seeking
and receiving social support, and fostering self-efficacy. Except self-efficacy, these factors, in
turn, significantly predicted improved health outcomes among cancer patients. Bonding social
capital, despite its lack of association with participation in SNS, also had a strong, positive
influence on acquiring positive health outcomes.
Participation in the Twitter-based SNS intervention also had a significant direct effect on
achieving positive health outcomes, even after controlling for the preexisting levels as well as the
changed amounts of all four predictors and mediators; bridging social support, bonding social
support, social support, and self-efficacy. The support for the proposed mediational model

A SOCIAL SUPPORT INTERVENTION ON TWITTER 8
reaffirms that regular use of SNSs as a community platform is a significant predictor for better
health results among cancer patients, with the preexisting levels and changed values of the key
variables being equal. This strong association between activities in SNSs and receiving health
outcomes was also mediated by the changed scores of bridging social capital and social support.
At the core of this structure, SNSs provide a forum for exchanging emotional and
logistical support, collectively building and expanding relational networks, and reinforcing self-
efficacy as related to disease management among patients affected by cancer. Collective
network building, emotional and informational support sharing, and the cognitive empowering of
members to take control of their disease management facilitated the achievement of greater
physical health and well-being among cancer patients.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 9
Introduction

Cancer is a significant medical condition that requires intensive treatment and management, and
that creates immense physical, emotional, psychological, and financial ramifications for the
patient. Approximately 13.7 million Americans have a history of cancer. In 2014 alone, there
will be an estimated 1,665,540 new cancer cases diagnosed in the U.S., not including
noninvasive cancer or skin cancer, and 585,720 cancer deaths, averaging almost 1,600 people per
day (American Cancer Society, 2014). Cancer remains the second most common cause of death
in the US, accounting for nearly 1 of every 4 deaths (American Cancer Society, 2014).
Online cancer-related activities are common experiences for many patients, caregivers,
and survivors (Eysenbach, Powell, Englesakis, Rizo, & Stern, 2004). Seeking cancer-related
information and social support online has wide-ranging implications and has been studied
extensively (Bender et al., 2012; Chou, Liu, Post, & Hesse, 2011; Eysenbach et al., 2004;
McLaughlin et al., 2012; Portier et al., 2013).
Almost half of all patients with cancer use the Internet to search for medical information
on diagnosis, treatment options, and the side-effects of their treatment, and nearly as many
search for emotional support (Duggan & Smith, 2013; Elkin, 2008; Fogel, Albert, Schnabel,
Ditkoff, & Neugut, 2002; Jordan-Marsh, 2010; Namkoong et al., 2013; Rimer et al., 2005).
These patients also heavily utilize the Internet for the purpose of accessing up-to-date
publications that relate to their conditions (Koch-Weser, Bradshaw, Gualtieri, & Gallagher, 2010;
Leydon et al., 2000; Tustin, 2010), as well as general healthcare delivery and administration
(Murray, Burns, See, Lai, & Nazareth, 2005; J Sarasohn-Kahn, 2012).

A SOCIAL SUPPORT INTERVENTION ON TWITTER 10
Overall, this is partly attributed to the lengthy duration of illness, the pervasive and the
serious nature of conditions, and the unique social relationship and bonding experience among
patients affected by cancer (Balka, Krueger, Holmes, & Stephen, 2010; Jordan-Marsh, 2010;
Meier, Lyons, Rimer, Frydman, & Forlenza, 2007; Novotny et al., 2010). The Internet is an
indispensable source for cancer-related health information as anonymous social interactions
provide a safe space with reduced social stigma suitable for cancer patients who are concerned
about privacy. Numerous studies have investigated the patterns of information and support
sharing on the Internet among cancer patients, and these have drawn various general conclusions
regarding health information sharing among patients affected by chronic conditions (Chou et al.,
2011; Freimuth & Quinn, 2004; S. C. Kim, Shah, Namkoong, McTavish, & Gustafson, 2013;
Portnoy, Scott-Sheldon, Johnson, & Carey, 2008).

Research has documented the positive effects of online social support and virtual communities
on cancer patients’ health and wellbeing (Gotsis, Wang, Spruijt-Metz, Jordan-Marsh, & Valente,
2013; S. C. Kim et al., 2013; Klemm, 2012; Pinar, Okdem, Buyukgonenc, & Ayhan, 2012).
Previous randomized studies have evaluated a variety of widely available features in online
communities, such as discussion forums (Bender, Jimenez-Marroquin, & Jadad, 2011; Gustafson
et al., 2008; Owen, Klapow, Roth, & Tucker, 2004; Winzelberg et al., 2003), cancer decision-
assistance services (Gustafson et al., 2008; Radina, Ginter, Brandt, Swaney, & Longo, 2011),
structured training exercises for coping skills (Owen et al., 2004), the keeping and sharing of
personal journals (Greene, Choudhry, Kilabuk, & Shrank, 2011; Gustafson et al., 2008; Owen et
al., 2004; Winzelberg et al., 2003), and other information services (Gustafson et al., 2008; Owen
et al., 2004). Randomized studies suggest that introducing Internet-based support resources to

A SOCIAL SUPPORT INTERVENTION ON TWITTER 11
cancer survivors and seeking or accepting information offered on the Internet result in
significantly positive outcomes regarding overall quality of life, social support, competence in
finding information (Chung, 2013; Gustafson et al., 2008; Sayers, Riegel, Pawlowski, Coyne, &
Samaha, 2008), depression (Linden & Vodermaier, 2012; Winzelberg et al., 2003), and self-
perceived health status (Owen et al., 2004; Sayers et al., 2008). Explorative studies have also
indicated that Internet groups empower cancer patients and facilitate new social networks (M. T.
Høybye, Johansen, & Tjørnhøj‐ Thomsen, 2005; Lieberman & Goldstein, 2006).
Despite the vast amounts of research on Internet usage among patients faced with chronic
conditions like cancer, the increasing adoption by cancer patients of online social networking
services (SNS) and social media outlets, such as Facebook and Twitter, has gained renewed
scholarly attention. The Pew Internet and American Life Project (2013) has reported that nearly
two-thirds of all American adults who are online (115 million) have adopted and frequently use
SNS, with Facebook, Twitter, Pinterest, and Instagram being some of the most dominant services.

Online social media services, such as Twitter and Facebook, enable real-time dissemination of
news, information, personal information, and other details via a highly interactive form of social
media, and they have become important online tools for patients. Ellison and Boyd (2013)
defined SNS as “a networked communication platform in which participants 1) have uniquely
identifiable profiles that consist of user supplied content, content provided by other users, and/or
system-provided data; 2) can publicly articulate connections that can be viewed and traversed by
others; and 3) can consume, produce, and/or interact with streams of user-generated content
provided by their connections on the site” (p. 158). By creating real-life identities in their online

A SOCIAL SUPPORT INTERVENTION ON TWITTER 12
profiles and postings, users voluntarily engage in friendship building with others who share
similar interests and concerns within a bound system that they determine.
In contrast to earlier static health-related websites, this type of dynamic online
communication among patients, commonly referred to by the term, “Health 2.0” (Eysenbach et
al., 2004; Farmer, Holt, Cook, & Hearing, 2009), now offers patients an opportunity to build and
benefit from a social network, and to learn about their illness and gain support from others with
similar experiences. The SNS medium, thus, plays an important role in the modern social
community of cancer patients.
Within SNSs, people establish social relationships and share personal information, stories,
images, and news updates with other members who are connected as social network members or
as “friends.” SNSs are often perceived by patients as more trustworthy, familiar, and more
conducive to exchanging information and support than traditional online communities (Kontos,
Emmons, Puleo, & Viswanath, 2010; McLaughlin et al., 2012; Schneider, Jackson, & Baum,
2009; Sugawara et al., 2012). On the Internet, patients have virtually unlimited possibilities for
finding health information and support, but within SNS platforms, they engage in interpersonal
conversations with their familiar social networks and get information from “someone just like
me” (Edelman, 2012). SNSs are often considered the most reliable source of health information
among consumers and have surpassed health professionals and academic experts with respect to
the amount of trust and credibility they inspire (Edelman, 2012).

The study of social networks and persuasive health communication has demonstrated that
relational patterns in individuals and intangible assets accrued in their social networks exert a
strong influence on health outcomes through their interdependent nature (Fowler & Christakis, ,

A SOCIAL SUPPORT INTERVENTION ON TWITTER 13
Christakis, 2008; Pan & Jordan-Marsh, 2010; Perkins & Berkowitz, 1986; Sugawara et al., 2012;
Valente, Unger, & Johnson, 2005; Wasserman & Faust; Wellman & Berkowitz, 1988). Sources
of health knowledge and emotional support are created, consumed, and disseminated within the
web of social relations in SNSs, where the concepts of trustworthiness, friendship, and
boundaries have been dramatically transformed (Arnst, 2008; Breckons, Jones, Morris, &
Richardson, 2008; Eysenbach, 2008; Hawn, 2009; Jimison et al., 2008; Kordzadeh & Warren,
2013; Jane Sarasohn-Kahn, 2008). Healthcare consumers’ ubiquitous usage of Health 2.0 tools,
including SNSs like Twitter and Facebook, and the prevalence of online health-related
collaborations on web 2.0 outlets or peer-to-peer software (Jordan-Marsh, 2010, p. 115) are
compelling evidence of shifts away from consumption of static Internet pages and are reflective
of how health information and supportive influence are diffused in the context of social networks.
Recent studies of SNSs, including Facebook and various online health portals, have also
linked social observational and learning behavior to positive health outcomes (D. Kim & Chang,
2007; Webb, Joseph, Yardley, & Michie, 2010). Within social networks, people form
interpersonal relationships and engage in observational behavior, which, in turn, facilitates the
transmission of information and experience among members, accelerating social learning and
helping individuals develop idiosyncratic characteristics (Bandura, 2001). Outside of the Internet,
social networks have been shown to improve disease management skills and the general health
outcomes for patients (Christakis & Allison, 2006; Christakis & Fowler, 2008; Sayers et al.,
2008).
There are a wide variety of SNSs specific to the needs of cancer patients on the Internet,
including those catering to specific purposes and conditions, like “I had cancer,” “Know Cancer,”
“Planet Cancer,” “Stupid Cancer,” and “Association of Cancer Online Resources” (Digitome,

A SOCIAL SUPPORT INTERVENTION ON TWITTER 14
2011). General health issue-oriented SNSs include: “Med Help,” one of the largest health
databases with about 12 million monthly visitors “Patients Like Me,” with over 116,080 global
users and over 500 conditions; “Vitals,” with medical information on 720,000 physicians across
the U.S.; and “Face to Face Health,” an SNS that automatically matches patients with similar
diagnoses (Digitome, 2011; Elkin, 2008; Fox, 2011).

In the broader framework of the SNS culture, Twitter is one of the most widely adopted social-
networking and micro-blogging services. With over 500 million registered users worldwide,
Twitter is an important online meeting place for social networking (Duggan & Smith, 2013).
Many disease-specific groups and organizations have arisen on Twitter, representing important
sources of information, support, and engagement for patients with chronic disease (Hawn, 2009;
J Sarasohn-Kahn, 2012; Shapiro, 2009).
Launched in 2006, Twitter is a leading method of disseminating brief online messages to
a potentially global audience (Duggan & Smith, 2013). Twitter enables its millions of users to
send and read each other’s “tweets,” short posts limited to 140 characters. Followers of a
specific Twitter user can view or respond to tweets online or via smart phones and other mobile
devices, allowing for a nearly instantaneous dialogue between the user and his or her followers.
The service has more than 240 million active users worldwide and processes about 500 million
tweets per day (Twitter Inc., 2013).
In 2013, among adult Internet users, which is about 80% of the US population, 15% used
Twitter on a regular basis (Pew Internet & American Life Project, 2013). Twitter use is also
shown to be well distributed across gender, race, income, and education levels compared to the

A SOCIAL SUPPORT INTERVENTION ON TWITTER 15
overall population of Internet users (Duggan & Brenner, 2013; Duggan & Smith, 2013;
Schneider et al., 2009).

Recent health-focused analyses of the American Twitter stream have revealed that Twitter may
be a useful medium for healthcare providers, medical researchers, health-related news media
outlets and patients. Healthcare professionals and researchers are increasingly adopting Twitter
for a wide range of reasons, including better patient care and treatment, patient education,
seamless communication with other healthcare providers, and a constant stream of medical news
and updates (Chretien, Azar, & Kind, 2011; De la Torre-Díez, Díaz-Pernas, & Antón-Rodríguez,
2012; Desai et al., 2012; Fortinsky, Fournier, & Benchimol, 2012; Rajani, Berman, & Rozanski,
2011). Doctors, in particular, are shown to heavily utilize Twitter to share medical updates and
case studies with fellow physicians, with nearly half of their tweet content being attributed to
health topics (Hofer & Aubert, 2013; Signorini, Segre, & Polgreen, 2011; Thackeray, Neiger,
Smith, & Van Wagenen, 2012). Studies also focused on the substantial influence of Twitter on
public health through its rapid and timely dissemination of useful health information (Heaivilin,
Gerbert, Page, & Gibbs, 2011; Kubben, 2011). Medical news events and new scientific
discoveries are often shared and discussed in real- time via Twitter directly among patients.
Studies have documented that patients with breast cancer, colorectal cancer, testicular cancer,
diabetes and other chronic disease have used Twitter to share medical information and seek
support about their conditions (Armstrong & Powell, 2009; Bender et al., 2012; FDA, 2009;
Hawn, 2009; Jain, 2009; Thompson et al., 2008).

A SOCIAL SUPPORT INTERVENTION ON TWITTER 16
In sum, Twitter is an interactive, real-time SNS medium, which has been widely adopted
recently for the exchange of medical information and support. Healthcare providers, public
health organizations, and news media outlets frequently utilize Twitter for the purposes of
disseminating health topics and influencing public opinions. Patients also exchange personal
stories and medical information via Twitter at no or relatively low cost in terms of the time,
effort, and expertise required for use. As a result, Twitter has the potential to play a critical role
in the cancer patient’s information seeking and social support exchange. However, studies
regarding the role of Twitter and other SNSs’ roles in influencing cancer patients’ health remain
limited. Relatively little research to date has explored the nature of SNS usage among cancer
patients on Twitter. Detailed information about cancer patients’ use of Twitter for such purposes
has yet to be fully studied.
This study investigates the impact of participation in patient social media on one SNS
platform, Twitter, on health outcomes among cancer patients. By conducting a randomized
controlled intervention study using a Twitter-based SNS social support program, this study
empirically evaluates whether health status differs between the cancer patients who participated
in the SNS intervention on Twitter and those who were randomly assigned to the control group.
The study also theorizes the ways in which key theoretical constructs and potential mediators,
drawn from extant literature, affect the processes and consequences of support sharing among
cancer patients on Twitter and their health outcomes.
Several traditional health and persuasive communication theories are integrated into the
conceptual design of the study, including bridging and bonding social capital (Putnam, 2000;
Williams, 2006), social support (Zimet, Dahlem, Zimet, & Farley, 1988), and self-efficacy
(Bandura, 1977). Based on the findings from the literature, the study proposes that participation

A SOCIAL SUPPORT INTERVENTION ON TWITTER 17
in the active support seeking and providing activities on SNSs would likely lead to higher
bridging social capital, bonding social capital, self-efficacy, and the sense of being supported.
Heightened levels of bridging and bonding social capital, social support, and self-efficacy would
be significantly associated with better assessments of health among cancer patients who
participate in SNS activities. Furthermore, literature on the social networks, social capital,
online patient social support, and self-efficacy suggests that the effect of the SNS intervention on
the health outcomes of cancer patients would be mediated by the amount of accumulated
bridging and bonding social capital, social support, and self-efficacy.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 18
Literature Review

Bridging and Bonding Social Capital

Actions by patients and caregivers in social networking communities and online health portals
are investigated from the perspective of social capital, which is analogous to economic or human
capital (Cacioppo, Berntson, Larsen, Poehlmann, & Ito, 2000; Coleman, 1989; Jordan-Marsh,
2010; Putnam, 2000). Though social capital is a broad concept and does not have one clear
definition, it is based on the idea that social relationships have value, which can be deployed for
productive and health-oriented ends (Putnam, 2000; Resnick, 2001). According to Putnam
(2001), the term social capital describes the “intangible substances” that account for a
community working together towards a common goal (Hannifan, 1920, as cited inPutnam, 2001).
The benefits of social relationships, such as trust, reciprocity, and norms, are called capital
because individuals or communities can do more work with less cost when they have such
relationships. Just as money provides financial capital, relationships among people have value
that can be translated into benefits to those individuals who are involved in a network of
relationships.
Within social networks, different subtypes of social capital have been theorized to affect
the mental and physical well-being of individuals. A few important dimensions of social
networks include differing tie strengths, mechanisms of social connections, and cohesion or
dispersion levels in social networks (Walther & Boyd, 2002; K. B. Wright, 2005). The most
significant differentiating factor of many subtypes of social capital is the difference between
bonding and bridging social capital (Putnam, 2000; Williams, 2006).

A SOCIAL SUPPORT INTERVENTION ON TWITTER 19
Bonding social capital is cultivated through strong, tight, and deep relational ties, such as those
between close friends and families, whereas bridging capital often accrues from weak, loose, and
shallow social relationships. Bonding social capital refers to strong, reinforcing interactions that
reaffirm exclusionary relationships (Putnam, 2000; Wuthnow, 2002). Bonding social capital,
which is accumulated through strong tie connections, is often an important source for emotional
well-being, and the term “bonding” describes how tightly people are connected within these
networks (Williams, 2006). In social network theory, bonding social capital is related to social
cohesion as described by Coleman (1989), who noted that this type of relationship can be found
in “fraternal organizations, church-based women’s reading groups, and fashionable country clubs”
(Putnam, 2000, p. 22).
Bridging social capital, on the other hand, refers to inclusive interactions that draw
people together across social boundaries. It is an important component of social interactions
because it is through bridging interactions with people outside one’s immediate circle, so-called
“weak ties,” that people are able to tap into heterogeneous groups and access unique resources
(Leonard & Onyx, 2003). Bridging social capital is found in the “civil rights movement, many
youth service groups, and ecumenical religious organizations” (Putnam, 2000, p. 22).

Theoretical and empirical studies have examined whether the two subtypes of social capital
translate into the online environment and the SNS outlets, and whether they can lead to health
benefits. Previous debates have focused on whether online interaction yields an accumulation of
bonding or bridging social capital and whether it spurs civic participation and stronger
community ties (Bender et al., 2011; Norris, 2004; Van Alstyne & Brynjolfsson, 2005). Studies
have shown that at both the individual and community levels clear links exist between online

A SOCIAL SUPPORT INTERVENTION ON TWITTER 20
activities and increased social capital in general (Drentea & Moren‐ Cross, 2005; N. B. Ellison,
Steinfield, & Lampe, 2007; Walther & Boyd, 2002; K. B. Wright, Rains, & Banas, 2010; Zhou,
2011). Online communities can facilitate social capital, with weak-ties and extended networks
on SNSs being most beneficial in community building and indicative of the effectiveness of
participation in times of physical or mental health crisis (Haythornthwaite, 2005; Norris, 2004;
Richardson, 2012). Active SNS participation leads to the accumulation of health information
and emotional resources, such as trust (Papacharissi, 2010), is manifested at both the individual
and the community level (N. B. Ellison et al., 2007; Valenzuela, Park, & Kee, 2009), encourages
civic engagement (Skoric, Ying, & Ng, 2009), and increases social capital overall (N. B. Ellison
et al., 2007). In the present study, participants in a cancer social media intervention program on
Twitter are expected to report higher bridging (H1a) and bonding (H2a) social capital compared
to those in the control group.

H1a: Participants in the social media intervention will have a significantly increased level
of bridging social capital at the end of the intervention compared to those who were
randomly assigned to the control condition.
H2a: Participants in the social media intervention will have a significantly increased level
of bonding social capital at the end of the intervention compared to those who were
randomly assigned to the control condition.

Social capital is not just an outcome of online social interaction and networking. Social capital
also further facilitates social engagement among people by building trust and coordinating social
exchanges, which, in turn, affects their health. The idea that levels of social capital can influence

A SOCIAL SUPPORT INTERVENTION ON TWITTER 21
individuals’ health was introduced in one study of suicide and social cohesion (Durkheim, 1951).
Studies by Kawachi et al. (Kawachi, Kennedy, & Glass, 1999; Kawachi, Kennedy, Lochner, &
Prothrow-Stith, 1997) have established an association between social capital, mortality, and
morbidity. Despite vast amounts of research over the past decade, the literature surrounding
social capital and its association with individual health outcomes has often suggested different
findings (Giordano & Lindstrom, 2010; Hawe & Shiell, 2000; Muntaner, 2004; Pearce & Davey
Smith, 2003; Szreter & Woolcock, 2004).

Recent studies on social capital, relational assets, and networking patterns among online
community participants have revealed new associations between social capital and health
outcomes through the way patients and caregivers emotionally assist one another, share
experiences, and evaluate or distribute health knowledge (Cheng et al., 2013; Giordano &
Lindstrom, 2010; Hofer & Aubert, 2013; M. Høybye et al., 2010; Namkoong et al., 2013;
Walther & Boyd, 2002; K. B. Wright et al., 2010). The individual levels of bridging and
bonding social capital of online participants have also been linked to their health outcomes
(Giordano & Lindstrom, 2010; Phua, 2013). To expand and clarify, associations between
subtypes of social capital, bridging (H1b) and bonding (H2b) social capital, and health outcomes
in SNSs were proposed in this study. Higher levels of bridging and bonding social capital in
SNSs were theorized to be associated with overall better health outcomes.

H1b: Increased level of bridging social capital among participants will be significantly
associated with an increased level of health outcomes at the end of the social media
intervention.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 22
H2b: Increased level of bonding social capital among participants will be significantly
associated with an increased level of health outcomes at the end of the social media
intervention.

A number of studies have tested mediational hypotheses to explain the effects of virtual
communities and online interactions on individual health outcomes. Berkman and Glass (2000)
show that social interaction in online communities directly or indirectly facilitates the flow of
information and acts as a supportive influence in online health communities. This is thought to
be achieved via a variety of mediating mechanisms, including the provision of social support,
perceived or actual, and changes in social context, such as increased levels of social influence
and social engagement. Kim et al. (2010) have prospectively tested the mediator effects of
online activities on health and found that social cohesion and organizational participation
significantly influence individual health outcomes. Increased participation in social activities
and civic engagement, which is a subset of social capital, significantly facilitates the positive
impact of SNSs on health outcomes. Mediators of effects of bridging and bonding social capital
on health among cancer patients who regularly participate in SNS activities have not been
extensively explored yet.

Mediational Analysis

This study investigates associations between participation in a SNS-based intervention for cancer
patients and individual health outcomes, while adjusting for the same health determinants and
preexisting scores as mediators. Simultaneously adjusting all considered variables, including the

A SOCIAL SUPPORT INTERVENTION ON TWITTER 23
predictor variables’ baseline scores and changed levels during the intervention, is an effective
way to investigate the consequences of supportive sharing in SNSs and its expected outcomes on
participants’ overall health and well-being.
The aim was to demonstrate whether subscales of social capital, as well as two additional
key theoretical concepts, social support and self-efficacy, are valid predictors of improvement in
self-rated health among cancer patients after participating in a SNS-based intervention. By
controlling for each construct’s baseline and changed levels, the mediated, indirect impact of the
SNS intervention on individual health outcomes could be accurately estimated, as well as the
SNS’s pure, direct impact on positive health change. The total impact of the Twitter social
support intervention on health outcomes was also calculated by combining the mediated and
direct impacts, given mediators.
The mediators included subscales of social capital and social support in order to replicate
and extend the previously discussed findings by Kim et al. (2010) and Berkman and Glass (2000).
Self-efficacy was also tested as a predictor and mediator. The theoretical grounds for including
social support and self-efficacy as predictors and mediators will be discussed in a later section.
Understandably, the manner in which bridging and bonding social capital, social support, and
self-efficacy influence or mediate health outcomes among cancer-focused SNS participants may
help resolve the debate surrounding the role of social capital and other constructs within the field
of health communication and public health.
The following hypotheses were proposed with respect to the indirect, mediated effect of
the SNS intervention on health outcomes through the role of bridging and bonding social capital.
First, it was speculated that the Twitter intervention would have mediated effects on the health
outcomes.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 24

H3a: Participants in the social media intervention will have a significantly increased level
of health outcomes at the end of the intervention compared to those who were
randomly assigned to the control condition through the mediated role of bridging
social capital.
H4a: Participants in the social media intervention will have a significantly increased level
of health outcomes at the end of the intervention compared to those who were
randomly assigned to the control condition through the mediated role of bonding
social capital.

Second, the Twitter intervention was expected to have a direct effect on the health outcomes
independent of the role of mediators.

H3b: Participants in the social media intervention will have a significantly increased level
of health outcomes at the end of the intervention compared to those who were
randomly assigned to the control condition while controlling for the mediated role
of bridging social capital.
H4b: Participants in the social media intervention will have a significantly increased level
of health outcomes at the end of the intervention compared to those who were
randomly assigned to the control condition while controlling for the mediated role
of bonding social capital.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 25
Third, the research addressed the overall and total effects of the social media intervention on
health outcomes given the direct and indirect role of key variables as mediators.

H3c: Participants in the social media intervention will have a significantly increased level
of health outcomes at the end of the intervention compared to those who were
randomly assigned to the control condition through the direct and indirect role of
bridging social capital.
H4c: Participants in the social media intervention will have a significantly increased level
of health outcomes at the end of the intervention compared to those who were
randomly assigned to the control condition through the direct and indirect role of
bonding social capital.

Social Support

One scholar in the communication scholarship discipline has defined social support as, “verbal
and nonverbal communication between recipients and providers that reduces uncertainty about
the situation, the self, the other, or the relationship, and functions to enhance a perception of
personal control in one’s experience” (Albrecht & Adelman, 1987, p. 19). A sense of social
support enhances the individual’s self-concept, and it generates relational and emotional aids that
mitigate stressful circumstances caused by medical conditions and health-related events (J. Kim
et al., 2010).

A SOCIAL SUPPORT INTERVENTION ON TWITTER 26
The literature provides an empirical and theoretical understanding of the reasons why
patients turn to broad social networks and communities to seek social support, including
informational, emotional, and instrumental support, rather than relying on immediate circles of
friends and family (Barak, Boniel-Nissim, & Suler, 2008; Bender et al., 2012; Hofer & Aubert,
2013; Stefanone, Kwon, & Lackaff, 2012; K. Wright, 2000). Health-oriented online social
networks and social tools have been reported to provide perceptions of being supported among
patients and caregivers, and, in turn, this facilitates health benefits by connecting people afflicted
with similar difficulties, enabling them to share, understand, empathize, and empower one
another (Barak et al., 2008; Bender et al., 2012; Haythornthwaite, 2005; Walther, 1996; Walther
& Boyd, 2002; K. Wright, 2000; K. B. Wright & Bell, 2003). Active participation in online
communities leads to a greater level of intimacy, self-disclosure, and emotional and
psychological well-being, which has a positive influence on perceived social support (Han et al.,
2012; Rodgers & Chen, 2005; Shaw, Hawkins, McTavish, Pingree, & Gustafson, 2006). This
study hypothesizes that participation in a SNS social support intervention on Twitter would
indicate a significantly positive association with social support on the site (H7a).

H5a: Participants in the social media intervention will have a significantly increased level
of social support at the end of the intervention compared to those who were
randomly assigned to the control condition.

Several scholarly studies have also documented the psychological and physical health benefits of
the perception of being supported (Albrecht, Burleson, & Goldsmith, 1994; Barbee &
Cunningham, 1995; Burleson, Albrecht, & Sarason, 1994; Cohen, 1988; Cutrona & Russell,

A SOCIAL SUPPORT INTERVENTION ON TWITTER 27
1990; Leventhal, Diefenbach, & Leventhal, 1992). The mental buffer model and the main effect
model are two alternative theories, which offer insights into the positive impact of social support
on health benefits (Cohen, 1988; Cutrona & Russell, 1990; Gottlieb & Wagner, 1991; Gustafson
et al., 2008; Kornblith et al., 2001).
First, participation in social interactions can provide a mental buffer that indirectly
mitigates the unhealthful emotions precipitated by stressful events. Second, social support can
directly and positively influence one’s mental and physical well-being through various processes
and mechanisms, including serious physical and mental health outcomes such as morbidity,
mortality, and stress (Arntson, Droge, Albrecht, & Adelman, 1987; Burlingame, Fuhriman, &
Mosier, 2003; Cohen, 1988; Schaefer, Coyne, & Lazarus, 1981). The general principles behind
the main effect of social support on one’s health include empathy, validation, strategy and
outcome (Albrecht et al., 1994; Cella & Yellen, 1992).

Participation in online health-oriented social resources emulates similar health benefits, such as,
increased levels of comfort, assistance, esteem, efficacy, information, and advice. It is noted that
patients and caregivers need social interactions that provide a safe space to observe, compare,
and validate their experiences, thus fostering a sense of universality and normalcy (Festinger,
1954; Gottlieb & Wagner, 1991; Yalom, 1970). A sense of being supported in an anonymous
social media environment provides such safe channels to witness, internalize, and model after
various management strategies (Posluszny, Hyman, & Baum, 2002). In this study, social support
was proposed as a direct influence on the associations between participation in the SNS

A SOCIAL SUPPORT INTERVENTION ON TWITTER 28
intervention on Twitter and individual health outcomes. It focused on the role of social support
as a predictor and mediator of health outcomes as a result of the intervention.

H5b: Increased level of social support among participants will be significantly associated
with an increased level of health outcomes at the end of the social media
intervention.

The indirect effects of social support as a mediator were proposed as a means to mediate the
relationship between SNS usage and health outcomes. Similar to previous hypotheses on social
capital, the direct and pure effect of SNS usage, while controlling for the mediating influences of
social support, was organized into three categories.

H6a: Participants in the social media intervention will have a significantly increased level
of health outcomes at the end of the intervention compared to those who were
randomly assigned to the control condition through the mediated role of social
support.
H6b: Participants in the social media intervention will have a significantly increased level
of health outcomes at the end of the intervention compared to those who were
randomly assigned to the control condition while controlling for the mediated role
of social support.
H6c: Participants in the social media intervention will have a significantly increased level
of health outcomes at the end of the intervention compared to those who were

A SOCIAL SUPPORT INTERVENTION ON TWITTER 29
randomly assigned to the control condition through the direct and indirect role of
social support.

Self-efficacy

Bandura’s (2001) social cognitive theory defines self-efficacy as a person’s belief in his or her
ability to complete tasks and reach goals in specific situations. Self-efficacy is also viewed as an
individual’s sense of behavioral control related to the ease or difficulty of performing a particular
task (Ajzen, 1991). Self-efficacy influences the individual’s ability to face challenges, as well as
the decisions he/she makes in specific social circumstances (Bandura, 1977). Self-efficacy is
cultivated externally through social observations and interactions within one’s social groups with
regard to particular behaviors (Ajzen, 1991; Bandura, 1977, 2001; Etter, Bergman, Humair, &
Perneger, 2000). Self-efficacy has proven to be a central motivating factor influencing health-
related choices and behaviors (Bandura, 2001; Etter et al., 2000).
Studies have linked self-efficacy with direct health outcomes and behavioral changes
through various mechanisms. An integrative model of behavioral prediction and self-
determination theory associates self-efficacy with the sense of autonomy and identifies it as a
necessary antecedent to health-related behavioral change (Deci & Ryan, 1985; Fishbein &
Cappella, 2006; Phua, 2013). Participation in online health communities and SNSs enhances
self-efficacy beliefs, which, in turn, prompt positive health outcomes (Namkoong et al., 2010;
Shaw et al., 2006; K. Wright, 2000). In the present study, participation in a SNS intervention for

A SOCIAL SUPPORT INTERVENTION ON TWITTER 30
cancer patients is expected to be significantly and positively associated with greater self-efficacy
(H7a), which, in turn, will be significantly associated with health outcomes (H7b).

H7a: Participants in the social media intervention will have a significantly increased level
of self-efficacy at the end of the intervention compared to those who were randomly
assigned to the control condition.
H7b: Increased level of self-efficacy among participants will be significantly associated
with an increased level of health outcomes at the end of the social media
intervention.

The overall impact of the Twitter-based social media intervention on health outcomes was
proposed with and without the indirect, mediating effect of self-efficacy.

H8a: Participants in the social media intervention will have a significantly increased
level of health outcomes at the end of the intervention compared to those who were
randomly assigned to the control condition through the mediated role of self-
efficacy.
H8b: Participants in the social media intervention will have a significantly increased
level of health outcomes at the end of the intervention compared to those who were
randomly assigned to the control condition while controlling for the mediated role
of self-efficacy.
H8c: Participants in the social media intervention will have a significantly increased
level of health outcomes at the end of the intervention compared to those who were

A SOCIAL SUPPORT INTERVENTION ON TWITTER 31
randomly assigned to the control condition through the direct and indirect role of
self-efficacy.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 32
Methods

The study employed a pre- and post-intervention, within-subject design with 206 participants.
The study focused on utilizing the social media platform, Twitter, to evaluate the impact of
online social support and networking on an individual after a health intervention for cancer was
introduced. Each individual assigned to the social media intervention condition agreed to
complete a one-month long Twitter-based social media support and networking program. The
study also examined how individuals’ preexisting levels of key variables before the intervention
affected the levels of social media use on Twitter and the individual’s health afterward.

Recruitment

Participants were recruited through two sources. First, a recruitment ad was publicly posted on
Twitter and Facebook, and also on online pages that support cancer management and awareness.
A link to the screening survey and study information was posted on several existing online
cancer patient support communities, such as “PatientsLikeMe.com” and “DailyStrength.org.” An
ad was also posted on the Facebook pages dedicated to cancer-related topics and causes. The
study was publicly tweeted on Twitter through cancer-related communities, awareness groups,
and research institutes. In addition, participants were recruited from a paid ad on the internet that
appeared on the search results page. During the month of November 2012, this study was funded

A SOCIAL SUPPORT INTERVENTION ON TWITTER 33
through a $10 daily budget that was used for the ads at 50 cents per click for the key words,
“cancer support group.”
1

Inclusion criteria included having been diagnosed with cancer or being in a state of
survival. All types of cancers were eligible. Participants were required to be at least 18 years
old at the time of enrollment. They were also asked if they owned a smartphone or mobile
device with internet and app functionality. Lastly, each participant was required to be active on
social media and to use their phones or mobiles devices daily for text messages.

Participants

Overall 273 people were recruited through existing social support groups on social media
platforms and the paid ad between April and November 2012. Of the 368 applicants who took
the screening questionnaire, 337 satisfied the inclusion criteria. Thirty-one applicants who
completed the screening survey were not accepted into the study. Exclusion occurred mostly due
to the following reasons: no cancer diagnosis (22.6%); under 18 years old (12.9%); no mobile
1
Sponsors bid for the click every time the user clicks on the link from the paid search results in
the top of the sponsored search results on Google. Examples of the average number of monthly
searches on cancer-related search keywords and suggested bidding prices for each click were
acquired in September 2012 through Google Analytics (Appendix A). There were about 35,700
exposures, and among those, 2% clicked on the link to the recruitment page. Of the people who
clicked on the study ad, about 7% participated in the screening survey.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 34
device with internet and app functionality (16.1%); no social media use (29%) or no text
message use or text plan (19.4%). The acceptance rate was unbiased by gender. African
American (73%) and Hispanic applicants (79%) had lower acceptance rates than Caucasian
applicants (92%).

Research Design

Pre-intervention Procedure
Recruitment was an ongoing process for 9 months between June 2012 and January 2013.
An online advertisement was posted during the last month. Initial study materials were sent to
all applicants who satisfied the inclusion criteria using the online survey tool, “Qualtrics.”
Instructions, consent documents, and an online questionnaire were sent to the specified
participant’s email, Twitter, and Facebook accounts, and they were also delivered as a text
message using “Google Voice” (see Figure 1 for the study procedure).
Random Allocation to Intervention and Control Condition
After signing the consent documents and completing the pretest questionnaire,
participants were randomly allocated to the intervention or the control condition. Consolidated
Standards of Reporting Trials (CONSORT) guidelines were followed for the randomized
assignment (http://www.consort-statement.org/). Participants who were assigned to the
intervention condition received instructions on the study procedures and basic functionalities of
the Twitter group application. They were, then, assigned to a private Twitter-based cancer

A SOCIAL SUPPORT INTERVENTION ON TWITTER 35
support group, each with 17 to 20 recruited subjects. Participants began the Twitter intervention
program in roughly 5 cohorts due to various conditions and scheduling conflicts. Participants
assigned to the control group did not participate in the Twitter-based support program.
Intervention, Twitter Support Group
Protected Twitter groups were created on the virtual platform through a third-party
solution, “Grouptweet.”
2
Support group members participated in the group dashboards from
their personal Twitter accounts, either public or private, or through any existing Twitter clients,
such as “HootSuite,” “TweetDeck,” or “Seesmic.” Twitter group dashboards were set to private
and they displayed a stream of updates that were specific to the manually approved members
who signed up with the invitation link provided in the orientation package. Participants were
encouraged to tweet the group daily. Members were also able to compose member-specific
tweets by addressing the tweet to a particular member of the group using the “direct message”
(DM) function. They also had the ability to create a profile and post multimedia tweets. Group
2
Twitter is a broadcasting-based medium and does not support private groups. A third-party
mobile application, Group Tweet, provides users who are already on Twitter with access to
protected group and private messaging. The researcher created a group account on Twitter for
each support group, marked it “Protected” and activated the account on Grouptweet.com.
Participants were instructed to follow the group account on Twitter. The researcher manually
approved each follower and followed them back. Approved members participated in the group
through any existing Twitter clients. Twitter groups were also accessible through any Apple- or
Android-based mobile devices or through the web.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 36
Tweets and DMs to individual members were dispersed to the entire group but kept private from
those outside of the group.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 37
Figure 1. Flowchart of the study procedure.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 38
Facilitation by Periodic Tweet Assignments
The Twitter groups were facilitated to ensure active participation and engaged
communication among all group members. Throughout the duration of the study, the researcher
posted periodic tweet messages to the group that had open-ended and unstructured assignments.
Periodic tweet messages were adapted from a mobile social networking and narrative
intervention study for cancer survivorship (McLaughlin et al., 2012; Song et al., 2012), resources
on coping with cancer and supportive and palliative care by National Cancer Institute
(http://www.cancer.gov/cancertopics/coping), and cancer social support network communities by
the American Cancer Society, including WhatNext (http://www.whatnext.com/), Circle of
Sharing (http://circleofsharing.cancer.org), and Cancer Survivor’s Network
(http://csn.cancer.org). Narrative tweet assignments motivated members to engage with other
group members by covering a wide range of topics such as the following: coping skills used,
information gathering, communication with healthcare providers, and reflections on cancer
experiences or survivorship in general.
An example of the periodic tweet messages by the researcher that encouraged participants
to share their interests and backgrounds was, “Introduce yourself to the group! Post a photo of
yourself to share with the others and tweet the group a little bit about yourself.” An example of a
cancer-related topic was, “Pretend that you are talking to someone that you did not know before.
How would you share your experience battling with cancer?” Tweet assignments also consisted
of educational components that were intended to increase self-awareness and preparation for
potential stressors and risky behaviors, such as “Share with the group how you might deal with
the temptation to do things that are risky to your health, such as smoking or lying in the sun
without skin protection.”

A SOCIAL SUPPORT INTERVENTION ON TWITTER 39
Facilitating questions and comments posted on the Twitter groups were sent to the
control and experimental groups as email and text messages (see Appendix B for the complete
list of the facilitating tweet assignments and comments). In this way, the experimental and
control groups had the same conditions excepting participation in the Twitter intervention. A
total of 8,523 tweets and meta data for the study were collected between June 2012 and January
2013 using a Python-based script.
Post-Intervention Procedure
Participants received wrap-up materials after the intervention program. They were asked
to complete an online post-intervention questionnaire and encouraged to continue participating in
supportive activities on social media. The wrap-up package also included additional resources
and guidelines for cancer social media support from the National Cancer Institute (see Appendix
C for the complete list of the resources and guidance included in the wrap-up materials).

A SOCIAL SUPPORT INTERVENTION ON TWITTER 40
Figure 2. Image of Twitter Social Support Group

Participants

At baseline, the total sample (N=337; M
age
= 40.39; SD
age
= 11.45) consisted of 164 participants
allocated to the Twitter intervention condition and 173 to control. Retention at post-intervention

A SOCIAL SUPPORT INTERVENTION ON TWITTER 41
was high with rates of 65.2% (N = 107) for the intervention group and 57.2% (N = 99) for the
control group. Retention was not affected by age, gender, or ethnicity. Participants who were
lost at a higher rate after the intervention were those who had been treated for cancer for less
than 1 year (N=74) and colorectal cancer patients (N=33).
Table 1 shows the demographic profile of the 206 participants who passed the inclusion
criteria,and who were accepted into the study and completed the baseline and post-test survey.
The participants were predominantly Caucasian (72.8%) with a mean age of 38.9 years (SD =
14.6). Most participants had a relatively high level of education. Nearly two-thirds of the
participants had either a college or associate degree (69.9%), and almost half of the sample
(44.7%) had at least one parent who graduated from college. About one-third of the sample was
either working full-time (28.6%) or part-time (6.8%). The majority earned less than $50,000 per
year (53.4%). Almost half the sample started the treatment less than 1 year ago (54.9%).
Diagnosed cancers included the following: breast cancer (20.9%), colorectal cancer (20.4%),
prostate cancer (16.5%), lung cancer (14.6%), thyroid cancer (7.3%), pancreatic cancer (4.4%),
and Chronic Lymphocytic Leukemia (1.5%). Intervention and control groups were comparable
on these variables at baseline. Intervention and control groups were comparable on these
variables at baseline.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 42
Table 1. Demographic Characteristics of Sample.

Gender

Men 46 43.0%

46 46.5%

92 44.7%

Women 52 48.6%

47 47.5%

99 48.1%

Age

18-25 2 1.9%

2 2.0%

4 1.9%

26-35 12 11.2%

12 12.1%

24 11.7%

36-50 47 43.9%

43 43.4%

90 43.7%

51-65 6 5.6%

5 5.1%

11 5.3%

66-75 2 1.9%

2 2.0%

4 1.9%

76+ 0 0.0%

1 1.0%

1 0.5%

Mean (SD) 38.6 (16.5)

39.2 (12.7)

38.9 (14.6)

Race/Ethnicity

White 80 74.8%

70 70.7%

150 72.8%

Black 2 1.9%

3 3.0%

5 2.4%

A SOCIAL SUPPORT INTERVENTION ON TWITTER 43

Hispanic 1 0.9%

1 1.0%

2 1.0%

Asian 4 3.7%

5 5.1%

9 4.4%

Mixed 2 1.9%

2 2.0%

4 1.9%

Education

High School 7 6.5%

4 4.0%

11 5.3%

Some College 14 13.1%

16 16.2%

30 14.6%

University 52 48.6%

52 52.5%

104 50.5%

Graduate Schools 6 5.6%

4 4.0%

10 4.9%

Education of Parents

High School 16 15.0%

18 18.2%

34 16.5%

Some College 32 29.9%

27 27.3%

59 28.6%

University 15 14.0%

13 13.1%

28 13.6%

Graduate Schools 2 1.9%

3 3.0%

5 2.4%

Employment

Unemployed -
Seeking
5 4.7%

7 7.1%

12 5.8%

Unemployed - Not
Seeking
12 11.2%

9 9.1%

21 10.2%

Student 1 0.9%

1 1.0%

2 1.0%

Part Time 6 5.6%

8 8.1%

14 6.8%

A SOCIAL SUPPORT INTERVENTION ON TWITTER 44

Full Time 29 27.1%

30 30.3%

59 28.6%
Homemaker 15 14.0%

11 11.1%

26 12.6%
Retired 2 1.9%

3 3.0%

5 2.4%

Households Income

$25K or Less 25 23.4%

24 24.2%

49 23.8%
$25K ~ $50K 31 29.0%

30 30.3%

61 29.6%
$50K ~ $75K 12 11.2%

12 12.1%

24 11.7%
$75K ~ $100K 6 5.6%

8 8.1%

14 6.8%
$100K or More 7 6.5%

6 6.1%

13 6.3%

Type of Cancer

Breast 23 21.5%

20 20.2%

43 20.9%
Prostate 16 15.0%

18 18.2%

34 16.5%
Thyroid 8 7.5%

7 7.1%

15 7.3%
Lung 16 15.0%

14 14.1%

30 14.6%
Colorectal 20 18.7%

22 22.2%

42 20.4%
C.L.L. 1 0.9%

2 2.0%

3 1.5%
Pancreatic 5 4.7%

4 4.0%

9 4.4%

Time Since First
Treatment
Less than 1 year 58 54.2%

55 55.6%

113 54.9%

A SOCIAL SUPPORT INTERVENTION ON TWITTER 45

Measures

Bridging Social Capital
For social capital, bridging and bonding social capital measures were modified from Williams to
pertain specifically to cancer patients and survivors (2006). The bridging social capital scale was
composed of the following instructions and items: there are several other cancer patients or
survivors who make me: (1) interested in things that happen outside of my town; (2) want to try
new things; (3) interested in what people unlike me are thinking; (4) curious about other places
in the world; (5) feel like part of a larger community; (6) feel connected to the bigger picture; (7)
remind me that everyone in the world is connected; (8) want to spend time to support general
community activities; (9) connected to new people to talk to; (10) informed about new things to
try; and (11) connected to useful information and resources. Responses were solicited on a 1–5
point scale (“1 = Strongly disagree,” “5 = Strongly agree”). The obtained value of Chronbach’s
alpha for the bridging social capital scale was .90 (see Table 2 for the means and standard
deviation).
Bonding Social Capital
The bonding social capital scale contained the following items: there are several other
cancer patients and survivors that (1) I trust to help solve my problems; (2) I can turn to for

A SOCIAL SUPPORT INTERVENTION ON TWITTER 46
advice about making very important decisions; (3) I can talk to when I feel lonely; (4) would put
their reputations on the line for me; (5) would be good job references for me; (6) would share
their last dollar with me; and (7) would help me fight an injustice. Chronbach’s alpha was .86
for the bonding social capital.
Social Support
The measure of social support was the Multidimensional Scale of Perceived Social
Support (Zimet et al., 1988). The 12-item scale measured perceived adequacy of support from
family, friends, and significant others. A summative index assessed general perception of social
support in participants’ lives on a 7-point scale (“1 = Very strongly disagree” and “7” = Very
strongly agree”; Cronbach’s alpha =.93).
Self-efficacy
Self-efficacy was based on and extended from Jerusalem and Schwarzer’s (1992) 16-item
scale. Examples of the items were: “I am confident that I can find cancer-related health
information need”; “I am comfortable asking my doctor/nurse questions about my body and
health”; and “I am confident I can access the resources I need to stay healthy.” Responses were
measured on a four-point scale (“1 = Not at all true,” “2 = Barely true,” “3 = Moderately true,”
and “4 = Exactly true”). The value of Chronbach’s alpha was .90.
Health
In order to measure the health outcomes of participating in Twitter-based social support
groups, five questions measured cancer-related coping skills and quality of life (CACSQL),
adapted and modified from van Uden-Kraan et al.’s study (van Uden-Kraan et al., 2008). The

A SOCIAL SUPPORT INTERVENTION ON TWITTER 47
questions included :(1) whether each patient was confident with his/her cancer treatment, (2) had
become better at managing his/her cancer-related condition, (3) had become more optimistic
about his/her cancer-related condition, (4) had felt an increase in self-esteem, and (5) had
discovered an enhanced feeling of health and well-being. A summative index assessed general
health perceptions of cancer-related coping skills and quality of life on a 5-point Likert scale (“1
= strongly disagree” and “5” = strongly agree”).
Initial evidence for validity of the measure was evaluated using data from the 206
patients by examining the association between scores on the CACSQL and those of other similar
measures completed at the same time. The Center for Epidemiologic Studies depression scale
(CES-D) (Radloff, 1977) and the Brunel Mood Scale (POMS-A) were used to provide construct
validity. Scale scores correlated well with related, established scales. Measures of mood, the
Brunel Mood Scale (POMS-A), and depression, the CES-D, were substantially correlated with
the CACSQL (r
CACSQL/POM-S
= .67; r
CACSQL/CES-D
= -.75). Internal consistency reliability for the
measure was estimated with Chronbach’s alpha, which was .92. CACSQL was validated as a
brief, yet reliable construct for evaluating cancer-related coping skills and quality of life.
Participation on Twitter, Relational and Self-Contained Tweets
In addition to the pre- and post-test questionnaire data, two composite measures of
participants’ global participation on Twitter support groups were created from the individual
participation metrics. First, records of participants’ Twitter activity that was relational in nature
were obtained. A total number of retweets (RT), direct messages (DM), responses to the

A SOCIAL SUPPORT INTERVENTION ON TWITTER 48
narrative assignment tweets, and replies and mentions were used to measure relational tweets.
3

In addition to the count of relational tweets, the number of tweets that were self-contained and
did not engage other members’ tweet were also measured.
Perceived Benefits and Usability of Twitter
The questionnaire also contained items used to measure the individual’s general
satisfaction with the Twitter support group and intervention components. Benefits and usability
of Twitter was a composite index of thirty-seven 7-point scale items, including: “Twitter was
easy to use;” “I felt comfortable using Twitter;” “Overall, I am satisfied with the amount of time
it took to share on Twitter,” “I am satisfied with participating on Twitter social support;” “I
would recommend Twitter social support to other people;” “Overall, I have the impression that
participating in Twitter social support will increase the scope of patient care/management;”
“Overall, I have the impression that Twitter social support will increase patients' accessibility to
health information and support” (Chronbach’s alpha = .909).
Control Variables
To control for factors that may confound the relationship between theoretical constructs
and the benefits of the Twitter social support on health outcomes, this study controlled related
demographic variables examined in the literature: age, gender, type of cancer and length of time
since first treatment. Because education or income level and employment were not significantly
different taking into consideration the participants’ levels of bridging and bonding social capital,
3
A Retweet is a reposting of someone else's tweet and sharing it with others. “RT” at the
beginning of a tweet shows that the user is quoting someone else's tweet. A Mention or Reply is
a message sent to a specific member of the group.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 49
social support, and self-efficacy, and did not theoretically predict the health-related online
activities and health outcomes, these variables were removed from the analysis for parsimony.
Table 2 summarizes the descriptive statistics for key measures for the intervention and
control groups from pre- and post-intervention and correlation at baseline. Figure 4 illustrates
changes in mean levels and 95% confidence intervals of means for the all measures and health
outcomes, from baseline to post-treatment, separately for those who were allocated in the Twitter
group and those in the control group.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 50
Analyses

This study strove to investigate the influences of factors related to the effectiveness of the online
health intervention (i.e., bridging and bonding social capital, social support, and self-efficacy)
and the baseline differences of the results of the social media intervention and health outcomes.
The Analysis of Covariance (ANCOVA) was the fitted statistical test for this study. More
specifically in this study, a series of full-factorial univariate ANCOVA tests allowed for the
comparison of the dependent variable (health outcomes) between two groups (i.e., Twitter
intervention group vs. control group) while taking the influence of continuous covariates into
account (i.e., bridging social capital, bonding social capital, social support, and self-efficacy).
Furthermore, since a covariate is commonly used to capture the extraneous pretreatment
differences, individual differences in the level of subscales of social capital, social support, self-
efficacy and health status before the study were considered as appropriate covariates reflecting
the dispositional characteristic of individuals. The statistical software package, Mplus 7.11, was
used to model the ways in which health outcomes were affected by the Twitter social support
intervention while statistically controlling for initial levels of social capital, social support, self-
efficacy, and baseline health status and other pre-test variables as outcome mediators (Muthén &
Muthén, 2002).
The ANCOVA also examined the effects of the Twitter intervention directly on given
mediators at post-treatment. Baseline scores were controlled and included as covariate
predictors of the same variables in the ANCOVA tests. This possibility of interaction between
the treatment and the mediator has been emphasized in many influential pre- and post-

A SOCIAL SUPPORT INTERVENTION ON TWITTER 51
intervention studies (Chmura Kraemer, Kiernan, Essex, & Kupfer, 2008; Judd & Kenny, 1981;
VanderWeele & Vansteelandt, 2009). Therefore, this study added causally-defined direct and
indirect effects of the mediator and a baseline interaction effect on the outcome in conducting
ANCOVA mediation analyses in order to enhance the power to estimate the direct, indirect, and
total effects of the treatment (James & Brett, 1984; MacKinnon & Luecken, 2008; Pearl, 2001;
Preacher & Hayes, 2008; Robins, 2003; Robins & Greenland, 1992). Type of cancer, time since
first treatment, age, and gender were also included as additional covariates in the models.

Missing Data

Full information, maximum likelihood estimation was used to include participants with
incomplete data. For the pre- and post-test measures, missing data ranged from 0% for several of
the baseline variables to a high of 17.1% for the post-test mood scores. Most measures after the
intervention had between 5% and 15% missing data. The baseline measures with the most
missing data were the bonding social capital subscale and the self-efficacy scale at 4.8% and 5.2%
missing data each. For post-test variables, the level of missing values was higher than pre-test
data. 8.9% of the data for post-intervention variables was missing. The missing data for the pre-
and post-test measures were analyzed to be Missing Completely At Random (MCAR) according
to the MCAR test (Little, 1988).
The Kolmogorov-Smirnov goodness-of-fit tests were conducted to evaluate whether or
not the distribution of data in the sample conformed to a normal distribution with a specific mean
and standard deviation. Although some variables demonstrated skew, the deviation from

A SOCIAL SUPPORT INTERVENTION ON TWITTER 52
normality was not severe enough to warrant transformation. Homogeneity of variance was
evident for all variables. The results supported the normality assumption. Thus, data analyses
for this study were based on a series of full-factorial, one-way ANCOVA analyses, using
bridging and bonding social capital, social support, self-efficacy, and baseline scores as
covariates.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 53
Results

Descriptive Results

Participants allocated to the Twitter intervention condition (N=107) posted 8,523 total tweets.
This amounted to an average of 18.38 tweets per week per participant (SD=5.21). Across groups,
73.6% of participants sent at least one tweet whereas 26.4% were passive observers or lurkers.
Seventeen participants stopped following the group Twitter account. Among all tweets, 59.6%
were relational tweets, such as retweets, direct messages, mentions or replies, while 40.4% were
self-contained tweets. Each participant posted an average of 47.48 relational tweets and 19.17
self-contained tweets throughout the duration of the intervention. Participants reported tweeting
via mobile phone apps (38.2%), computer (37.1%), and/or texting (24.7%).
Participants rated the Twitter social support as easy (M = 4.03, SD = 1.59), comfortable
to use (M=4.8, SD=1.15), and were satisfied with the amount of time it took to share on Twitter
(M=4.12, SD=1.37). The mean score of perceived benefits and overall usability of the Twitter
intervention was relatively high at 4.06 (SD=1.45).

Tweet Content Results

Common tweet types were: (1) shared personal information (N=2301, 27%); (2) shared
emotional support (N=2215, 26%); (3) discussed commitment to treatment (N=1278, 15%); (4)

A SOCIAL SUPPORT INTERVENTION ON TWITTER 54
discussed an obstacle such as side effect, stigma and uncertainty (N=767, 9%); (5) discussed
strategies to counter an obstacle (N=511, 6%); (6) discussed a prognosis (N=426, 5%); (7)
discussed the treatment options (N=341, 4%); and (8) shared confidence about treatment (N=256,
3%).
Content that was less prevalent were: (9) evaluated the Twitter Intervention positively
(N = 93) or negatively (N = 74, total N = 167); (10) shared a stressful life event or
relationship (N=95); (11) shared another patient or survivor’s support of the treatment
(N=84); (12) discussed non-treatment (N=52); and (13) miscellaneous (N=28). A tweet
could be classified into more than one content category. Overall, 47.27% of the tweets
addressed recommended facilitation topics.
Independent samples t-tests were used to check for differences in the baseline measures
between the participants allocated to the Twitter intervention condition and the control condition
(Table 2). The results revealed no significant differences in the baseline scores between the
treatment and control groups.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 55
Table 2. Means and Standard Deviations of Key Variables and Intercorrelations at Baseline (N
=206).
Intervention (N=107) Control (N=99)
Pre

Post

Pre

Post
Bridging Social Capital 4.88 (0.92)

5.52 (0.90) *

4.72 (0.83)

4.75 (0.95)
Bonding Social Capital 4.29 (1.32)

4.54 (1.42) †

4.13 (1.05)

4.42 (1.78) †
Social Support 4.78 (1.07)

5.07 (1.37) †

4.69 (1.23)

4.72 (1.21)

Self-Efficacy 3.39 (0.50)

3.83 (0.41) *

3.24 (0.72)

3.32 (0.68)

Health Outcome 4.21 (0.09)

4.78 (0.98) *

4.26 (1.01)

4.34 (0.62)

Time Since First
Treatment
7.30 (3.37)

– –

8.20 (4.19)

– –

Total Tweet – –

79.66ₐ –

– –

– –

Relational Tweet – –

47.48ₐ –

– –

– –

Self-Contained Tweet – –

19.17ₐ –

– –

– –

Twitter Usability – – 4.06 (1.45) – – – –

Bridging
Social
Capital
Bonding
Social
Capital
Social
Support
Self-
Efficacy
Health
Outcome
Time
Since
Treatment
Bridging Social
Capital
– .19* .22* .26** .25** .16
Bonding Social
Capital
– .25* .18** .35** .09
Social Support

– .31** .29** .09*
Self-Efficacy

– .18** .13*
Health Outcome

– .08

A SOCIAL SUPPORT INTERVENTION ON TWITTER 56
Time Since First
Treatment
–
Total Tweet

Relational Tweet

Self-Contained
Tweet
Twitter Usability

Note. Means are given followed by standard deviation in parentheses. †p < .10. *p < .05. **p
< .01. a. Total number of tweets per person.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 57
Mediators of the Effect of Twitter Intervention on Health Outcomes

Bridging and Bonding Social Capital

The study conducted a series of full-factorial ANCOVA tests and mediation analyses to
empirically assess the “Twitter”-based social media intervention on health outcomes with the
subscales of social capital, social support, and self-efficacy as mediators. First, the effect of the
intervention directly on the given mediators, bridging and bonding social capital, was
hypothesized in H1 and H2 respectively. The ANCOVA results (column 1 and 2 from Table 3)
showed a significant univariate main effect of the intervention on bridging social capital (F (1,
193) =12.54, p < .05, η2 = .16). However, participants in the Twitter intervention did not accrue
significantly higher bonding social capital compared to those in the control condition (F (1, 188)
= 8.92, p = .12, see Figure 4).
The means depicted in the top two panels of Figure 4 indicate that the amount of bridging
and bonding social capital increased for both intervention and control groups, but only bridging
social capital increased significantly more for those who were allocated to the intervention group
(see Table 2 for the pair-wise comparison). Individuals from the intervention group were more
likely to have significantly higher bridging social capital (M = 5.52, SD = .90) than the control
group (M = 4.75, SD = .95). However, the top-right panel of Figure 4 illustrates how little the
Twitter intervention changed the levels of bonding social capital. The amount of bonding social
capital gain was similar in a comparison of the treatment and control groups. Participation in the
Twitter support program predicted an increased amount of bridging social capital, which
supported H1a, but not bonding social capital. Thus, H2a was rejected.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 58
Next, the ANCOVA tests estimated the mediator’s main effect on health outcomes. A
higher level of bridging and bonding social capital was theorized to be associated with better
health outcomes in H3 and H4 respectively. Results revealed statistically significant main
effects for both bridging (F (1, 178) =4.76, p < .001, η2 = .21) and bonding social capital F (1,
174) =8.54, p < .001, η2 = .24) on improving health at post-test (column 1 and 2 from Table 4).
In other words, an increased amount of bridging and bonding social capital predicted
improvement in overall health after the study regardless of the participation in the intervention.
Therefore, H1b and H2b were supported.
Pair-wise comparisons from Table 2 showed that the amount of bonding social capital
marginally increased for both the intervention and control groups and did not show significant
change for those who were allocated to the intervention group compared to the control group. It
was noteworthy that greater bonding social capital after the study, albeit not associated with the
Twitter treatment condition, did predict a higher level of overall health of participants in the
intervention and control groups alike (column 2 from Table 3 and Table 4). In other words,
participants in the intervention and control groups both increased in the amounts of bonding
social capital after the study independent of the Twitter social support intervention, which in turn,
contributed to a positive change in their health.

The next set of hypotheses pertained to the main outcome analyses of the overall effectiveness of
the Twitter intervention given the direct or indirect processes of mediation. The first predicted
an indirect effect of the intervention on health outcomes through the role of a given mediator.
The second predicted a direct effect of the intervention on health independent of the mediator.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 59
The final hypothesis combined the direct and indirect effects given the mediator and estimated
the total effects of the intervention on overall health after the study.
H3a and H4a proposed that the Twitter social media intervention has mediated effects on
health through bridging and bonding social capital. The mediated effects of the Twitter
intervention on the health outcomes, given the subscales of social capital as covariates, were
observed for bridging social capital (F (1, 178) = 8.21, p < .01, η2 = .17) but not for bonding
social capital (F (1, 174) = 7.19, p = .13, see column 1 and 2 from Table 5). The mediated role
of accumulated social capital on improving health was significant only for bridging social capital
but not for bonding social capital. H3a was supported. H4a was rejected.
The Twitter intervention was also expected to have a direct effect
4
on health outcomes
independent of the mediator. H3b and H4b proposed that participants in the Twitter intervention
would have significantly better health while controlling for the mediating role of bridging and
bonding social capital. The intervention was shown to have pure direct effects on the health
outcomes of participants independent of bridging social capital (F (1, 189) = 6.84, p < .01, η2
= .18) and bonding social capital (F (1, 191) = 8. 25, p < .01, η2 = .17). H3ab and H4b were
supported.
4
Using Mplus 7.11,the direct effect was computed not by holding the given mediator constant,
but instead, by allowing the mediator to vary over participants in the way it would vary if the
participants were allocated to the control group (see the Mediation Formula in Pearl (2012). In
other words, it calculated the expected difference, given the covariate, between the health
outcomes in the intervention and control group when the given mediator was held constant at the
values it would obtain for the control group.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 60
Finally, total effects of the Twitter intervention on improvement in health were estimated
by combining the direct and indirect effects given the mediator. The total effect
5
was the sum of
the direct and the indirect effect and was the total effect of the intervention on health outcomes,
both direct and indirect through the given mediator. H3c and H4c proposed the total effects of
the Twitter intervention on positive change in health among cancer patients specifically through
the direct and indirect roles of the subscales of social capital. Participation in the social media
intervention program had a significant total effect on health outcomes compared to those who
were randomly allocated to the control group through the direct and indirect mediated effect of
bridging social capital (F (1, 179) = 5.43, p < .01, η2 = .23) and bonding social capital (F (1, 191)
= 11.26, p < .01, η2 = .19). Therefore, H3c and H4c were supported.

5
The total effect was the conditional expectation, given the covariate, of the difference between
the health outcome in the Twitter intervention group when the given mediator changes from
values it would obtain in the intervention group to the values it would obtain in the control group
(Robins, 2003). It was also referred to as the natural indirect effect (Pearl, 2001; VanderWeele
& Vansteelandt, 2009).

A SOCIAL SUPPORT INTERVENTION ON TWITTER 61

A SOCIAL SUPPORT INTERVENTION ON TWITTER 62
Figure 3. Summary of findings related to (a) H1 and H3; (b) H2 and H4; (c) H5 and H6 and; (d)
H7 and H8. The independent variable (Twitter intervention) is indicated on the left-hand side of
the diagram. Arrows emerging from each signify the main effects on dependent variables while
those connecting independent variable and health outcomes signify their mediated, direct or total
effects for a given mediator. Relationships that are significant at p ≤ .05 are italicized and
indicated in bold.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 63

Table 3. ANCOVA Results From Testing Effects of Variables on Mediators With Adjusting for Baseline Scores.
Dependent Variable
Model Estimates
Effects on the Mediator at Post-Test
Outcome Variable: Post-Test Mediator
Source of Variation
Baseline Health -0.027 ( 0.049 ) 0.060 ( 0.053 ) -0.027 ( 0.043 ) -0.062 ( 0.031 )
Baseline Mediator 0.240 ( 0.050 ) ** 0.402 ( 0.054 ) ** 0.619 ( 0.038 ) ** 0.001 ( 0.032 ) **
Time Since First Treatment 0.011 ( 0.045 ) -0.034 ( 0.044 ) -0.049 ( 0.040 ) -0.008 ( 0.018 )
Gender 0.036 ( 0.045 ) -0.082 ( 0.048 ) † -0.045 ( 0.038 ) 0.125 ( 0.038 )
Age -0.068 ( 0.045 ) 0.069 ( 0.040 ) † 0.006 ( 0.037 ) -0.008 ( 0.049 )
Twitter Intervention 0.126 ( 0.046 ) ** 0.095 ( 0.043 ) 0.089 ( 0.039 ) * 0.073 ( 0.054 ) †
Note. Standardized Estimates (Standard Errors). † p < .10, *p < .05, **p < .01.
For Twitter intervention, 0 = control group, 1 = intervention group. For gender, 0 = female and 1 = male. All continuous variables are coded
such that higher scores reflect higher levels of the construct.
Bridging
Social Capital
Bonding
Social Capital
Self-Efficacy Social Support

A SOCIAL SUPPORT INTERVENTION ON TWITTER 64

Table 4. ANCOVA Results From Testing Mediators of the Effect of Twitter Intervention on Health With Adjusting for Baseline Scores and Initial Level
of Dependent Variable.
Mediators
Model Estimates
Effects on Post-Test Health Outcome Controlling for the Baseline Mediator
Outcome Variable: Post-Test Health Outcome
Source of Variation
Baseline Health -0.483 ( 0.048 ) ** 0.464 ( 0.050 ) ** -0.449 ( 0.049 ) ** -0.480 ( 0.046 ) **
Baseline Mediator -0.040 ( 0.040 ) 0.019 ( 0.048 ) -0.094 ( 0.056 ) † -0.009 ( 0.055 )
Post-Test Mediator 0.160 ( 0.043 ) * 0.228 ( 0.042 ) ** -0.060 ( 0.058 ) 0.209 ( 0.054 ) *
Time Since First Treatment 0.071 ( 0.036 ) -0.062 ( 0.035 ) † -0.186 ( 0.038 ) -0.070 ( 0.036 )
Gender 0.112 ( 0.039 ) -0.088 ( 0.038 ) † -0.110 ( 0.036 ) 0.109 ( 0.038 )
Age -0.129 ( 0.038 ) † 0.135 ( 0.036 ) 0.137 ( 0.038 ) -0.028 ( 0.045 )
Twitter Intervention 0.172 ( 0.031 ) ** 0.162 ( 0.033 ) ** 0.173 ( 0.035 ) ** 0.103 ( 0.030 ) **
Note. Standardized Estimates (Standard Errors). † p < .10, *p < .05, **p < .01.
For Twitter intervention, 0 = control group, 1 = intervention group. For gender, 0 = female and 1 = male. All continuous variables are coded
such that higher scores reflect higher levels of the construct.
Bridging
Social Capital
Bonding
Social Capital
Self-Efficacy Social Support

A SOCIAL SUPPORT INTERVENTION ON TWITTER 65

Table 5. Mediation Estimates of the Effect of Twitter Intervention on Health Outcome.
Mediators
Model Estimates
Effect of Twitter Intervention on Health Outcome
Mediated effect Through the Effect of the Mediator 0.021 ( 0.009 ) * 0.024 ( 0.014 ) 0.008 ( 0.011 ) 0.017 ( 0.011 ) †
Direct Effect Independent of the Effect of the Medi 0.175 ( 0.031 ) ** 0.162 ( 0.032 ) ** 0.171 ( 0.035 ) ** 0.103 ( 0.032 ) **
Total effect Combining Direct and Mediated Effect 0.002 ( 0.004 ) ** 0.005 ( 0.004 ) ** 0.002 ( 0.002 ) ** 0.000 ( 0.000 ) **
Note. Standardized Estimates (Standard Errors). † p < .10, *p < .05, **p < .01.
For Twitter intervention, 0 = control group, 1 = intervention group. For gender, 0 = female and 1 = male. All continuous variables are coded
such that higher scores reflect higher levels of the construct.
Bridging
Social Capital
Bonding
Social Capital
Self-Efficacy Social Support

A SOCIAL SUPPORT INTERVENTION ON TWITTER 66
Self-Efficacy

A set of full-factorial univariate ANCOVA tested the next set of hypotheses on the direct
and indirect effects of self-efficacy and social support as mediators. Hypotheses proposed a role
for self-efficacy and social support as mediators of the intervention on health outcomes.
Regarding self-efficacy, a significant effect of the Twitter intervention on post-treatment self-
efficacy was observed (F (1, 182) = 16.29, p < .05, η2 = .18, see column 3 from Table 3).
Individuals accumulated significantly more self-efficacy in the intervention condition (M=3.83,
SD=.41) than those in the control group (M=3.32, SD=.68).
The means, illustrated in the middle-left panel of Figure 4, shows a significant increase in
self-efficacy from the baseline for intervention group but almost a parallel line in the control
group. Results supported H5a in terms of the main effect of the Twitter intervention on self-
efficacy.
On the other hand, a higher level of self-efficacy was not shown to be associated with an
improvement in health after the study. The ANOVA results revealed no significant main effect
of self-efficacy on health outcomes at post-test (F (1, 168) = 11.21, p = .18; see column 3 from
Table 4). Participants who had a higher sense of self-efficacy were not necessarily the ones who
came out with better health outcomes at post-test. Therefore, H5b, which was about the main
effect of self-efficacy on health outcomes, was rejected.

H6a proposed an indirect effect of the Twitter intervention on health specifically through the
mediating role of self-efficacy. Mediated effects of the intervention on health outcomes, given

A SOCIAL SUPPORT INTERVENTION ON TWITTER 67
self-efficacy as a covariate, were not observed (F (1, 158) = 7.85, p = .13; see column 3 from
Table 5). The mediated role of self-efficacy on improving health status in the post-test was not
significant in this study. Therefore, H6a was rejected.
H6b predicted the direct effects of the Twitter intervention on positive change in health
independent of the mediating role of self-efficacy. The intervention revealed significant pure
effects directly on the post-test health outcomes of participants while controlling for the
influence of self-efficacy (F (1, 145) = 12.33, p < .01, η2 = .25). Participation in the Twitter
intervention was associated with gaining significant health outcomes through the mediated role
of self-efficacy. Therefore, H6b was supported.
Lastly, H6c hypothesized the total effects of the Twitter intervention on health status
through the direct and indirect effects combined, given the specific role of self-efficacy as a
mediator. Participation in the social media intervention program had a significant total effect on
health outcomes compared to those who were randomly allocated to the control group through
the direct and indirect mediation of the self-efficacy (F (1, 148) = 6.24, p < .01, η2 = .28). H6c
was supported.

Social Support

A significant result was shown for the main effect of the Twitter intervention on social
support (see column 4 from Table 3). Results indicated significant differences in the amount of
social support across the two conditions. Participants in the Twitter intervention group reported

A SOCIAL SUPPORT INTERVENTION ON TWITTER 68
slightly higher social support (M=5.07, SD=1.37) than their counterparts (M = 4.72, SD=1.21, t
(185) = 7.32, p < .05). H7a was marginally supported.
A pairwise t-test detected a significant difference in the expected direction in the pre- and
post-test social support scores between the intervention and control groups (Table 2). The means
shown in the middle-right panel of Figure 4 depicts different patterns of change in social support
following the intervention. Participation in the intervention did not exert a strong influence on
the change in the amount of social support. Individuals from both the intervention and control
groups experienced similar levels of social support gains after the study only with only slightly
more improvement from the baseline in the intervention group.
H7b proposed that an increased sense of social support would be related to health
outcomes. Results indicated statistically significant main effects for social support on health
outcomes (F (1, 164) = 7.45, p < .01, η2 = .26; see column 4 from Table 5). H7b was supported.
Interestingly, the relationships between the intervention and social support and also
between social support and health outcomes revealed similar patterns to the earlier findings for
bonding social capital. Participants with more social support or bonding social capital scored
significantly higher in their overall health indicators than those in the control group. However,
the effect of the Twitter intervention itself did not differentiate the amount of accrued bonding
social capital and only marginally affected the amount of social support. This trend was a
reversal of the case of self-efficacy, with the Twitter intervention producing significantly higher
self-efficacy, but an improved sense of self-efficacy did not affect the health outcomes at post-
test.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 69
H8a proposed that the Twitter intervention would have effects on health outcomes
specifically through the mediated effects of social support. The ANCOVA showed a significant
mediated effect of the Twitter intervention on health outcomes through social support. The
indirect effect of the Twitter intervention on the health outcomes through the mediated role of
social support was significant (F (1, 161) = 10.95, p < .01, η2 = .14; see column 4 from Table 5).
H8b proposed a direct effect of the Twitter intervention on health outcomes outside of the
direct or indirect influence of social support as a mediator. Pure indirect effects of the
intervention on the health outcomes of participants independent of social support were
statistically supported (F (1, 163) = 7.42, p < .01, η2 = .21). H8b was supported. Twitter
participants were better off in terms of health outcomes, even when the mediating role of social
support was controlled.
H8c proposed a significant total effect of the Twitter program on overall health through
the combined direct and indirect effects of social support. When the immediate and mediated
effects of social support on health outcomes were calculated, the intervention program on Twitter
had a statistically significant total effect on health outcomes compared to those who were
assigned to the control group (F (1, 158) = 7.22, p < .01, η2 = .28). H8c was supported.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 70

Figure 4.Changes in mean levels of health and the four mediators from baseline to post-treatment
separately for those who were allocated in the Twitter group and those in the control group.
Vertical bars show 95% confidence intervals of means.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 71
Discussion

This study used a randomized controlled design and mediational analysis to evaluate how
participation in a SNS-based intervention on Twitter affected health outcomes among cancer
patients. The study tested the overall impact of the Twitter intervention on the health outcomes
of cancer patients by examining how theoretical constructs and mediators impact the processes
and consequences of support sharing on Twitter and the health outcomes of cancer patients as a
consequence. Several traditional health communication concepts and social network theories,
including bridging and bonding social capital (Putnam, 2000; Williams, 2006), social support
(Zimet et al., 1988), and self-efficacy (Bandura, 1977), were integrated into the study design.
Results indicate that participation in support seeking and providing activities on a SNS
intervention, using Twitter as a case study, lead to higher levels of bridging social capital, self-
efficacy, and perceived social support among participating cancer patients. Increased levels of
bridging and bonding social capital and perceived social support from engagement in the SNS
community directly led to positive health outcomes among cancer patients. Furthermore,
bridging social capital and social support mediated the relationship between participation in the
SNS intervention and health outcomes among the participants, as indicated in the literature on
social network, social capital, and online patient social support. Beyond the theoretical support
proposed in the earlier section of the study, the Twitter-based SNS intervention affected the
physical and emotional health and the well-being of cancer patients, and influenced the processes
of supportive exchange in SNS in the ways described below.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 72
Bridging and Bonding Social Capital

First, participation in the Twitter-based SNS intervention program significantly increased
bridging social capital among participants. Bridging social capital is an important component of
social interactions because it is through bridging interactions with people outside one’s
immediate circle that people are able to tap into heterogeneous groups and access unique
resources (Afifi & Weiner, 2004; Arnst, 2008; Breckons et al., 2008; Eysenbach, 2008; Hofer &
Aubert, 2013; Jimison et al., 2008; Johnson & Meischke, 1993; J Sarasohn-Kahn, 2012). When
cancer patients engage in inclusive interactions in SNSs that draw users together across social
boundaries and geographic locations, this activity increases the individual’s bridging social
capital. Seeking health information and assisting one another in SNSs add an important resource
for building the friendship networks and expanding the social boundaries of cancer patients.
However, although active engagement and interaction in SNSs increased bridging social
capital, it did not exert the same positive impact on the level of bonding social capital. A SNS
intervention helped cancer patients build up more heterogeneous ties, connecting them with
people to whom they may not have been otherwise exposed, yet the strengthening ties between
close, intimate friends was not significantly associated with network activities in SNS.

The findings pertaining to the different impacts of SNS usage on subtypes of social capital
contradict the findings that linked using online social network sites and SNS activities with
enhanced bridging and bonding social capital (N. B. Ellison et al., 2007; Haythornthwaite, 2005).
With regard to bonding social capital and SNS usage, Ellison et al.(2007) suggest that SNSs
provide an effective mechanism for strengthening ties with close friends, thus bolstering bonding

A SOCIAL SUPPORT INTERVENTION ON TWITTER 73
social capital as a result. Contrary to the literature and the proposed hypothesis, SNS usage did
not help individuals deepen their established relationships or connect with more homogeneous
links.
In fact, the academic debates surrounding subtypes of social capital and their association
with health-related outcomes and online behavior have yet to be resolved (N. B. Ellison et al.,
2007; Hawe & Shiell, 2000; Morgan & Haglund, 2009; Muntaner, 2004; Norris, 2004; Pearce &
Davey Smith, 2003; Szreter & Woolcock, 2004; Valenzuela et al., 2009; Van Alstyne &
Brynjolfsson, 2005). This study adds plausible explanations for the complex mechanism of how
engagement in SNSs may influence bridging and bonding social capital.

The results from the study suggest that SNS activities on Twitter may promote weak-tie
connections, hence facilitating bridging social interactions, but they do not reinforce strong-tie
networks. That is, online social interactions in SNSs meet the conditions necessary for the
production of bridging social capital but may not be as effective in attracting participants with
strong motivations for bonding connections. Cancer patients with weak-tie network tendencies
and bridging interaction goals may be more actively engaged in SNS usage, therefore scoring
higher on bridging social capital assessments, compared to the strong-tie network participants
whose goal is to build stronger emotional bonds with others.
Another possible explanation comes from the literature that indicates that online social
interactions are, by nature, weak and do not significantly impact strong or bonding connections
(Haythornthwaite, 2005; Pickering & King, 1995; K. B. Wright et al., 2010). Weak-ties and
extended networks in SNSs are most suited and beneficial to building broad social connections
for patients faced with challenging physical or mental conditions (Haythornthwaite, 2005; Norris,

A SOCIAL SUPPORT INTERVENTION ON TWITTER 74
2004). As the Internet sustains communities for niche interests, online interaction is expected to
satisfy the preference and desire for diversity, and to serve the role of bridges for people from
diverse backgrounds (Kawachi, Subramanian, & Anderson, 2004; Pan & Jordan-Marsh, 2010;
Walther, 1996). Recent findings also provide evidence that online social network sites are ideal
for people who are interested in building and maintaining bridging social capital because they
offer a low cost way of keeping up to date on information and news from weak ties (N. B.
Ellison et al., 2007).

Several computer-mediated communication theories, including the social identity model of the
deindividuation (SIDE) theory, the hyperpersonal communication model, and the social
information processing (SIP) model, also provide plausible theoretical explanations for the weak
association between SNS participation and the cultivation of bonding social relations.
The SIDE theory suggests that the reduced amount of social cues available online helps
build more genuine and strong relationships based on common group identities (Postmes, Spears,
& Lea, 1998). In SNS, real-life identities are often articulated in the foreground in common
interactive features, like user profiles, news feeds, commenting and “liking,” and sharing. These
may actually hinder the process of private, anonymous, and idealized relationship building,
negatively affecting the deep trust building and emotional bonds among of those who value close
ties.
Similarly, the hyperpersonal communication model predicts that enhanced relational
outcomes are based on idealized impressions of each other, which are typical in online
interactions. Since online communication in general affords the ability to be more selective in
self-presentation, idealized impressions of communication partners are common online. With

A SOCIAL SUPPORT INTERVENTION ON TWITTER 75
regard to participants with strong, intimate network connection backgrounds and motives to
create predominantly homogeneous relationships, the ubiquity of the user profiles in SNS, which
establish members’ real-life identities in routine interactions, may prompt different processes of
social interaction than was the case in the early days of the largely text-based Internet.
Lastly, the limited availability of social cues on the Internet in general may hinder the
development of deep, long-lasting social relationships. Virtual communities often fail to engage
participants with strong motivations for bonding connections or close relationship building. The
SIP model predicts that the social relationships built online normally take more time and effort
than relationships built offline: this may deter participants with bonding-oriented relationship
goals (Walther, 1996). Geographically dispersed networks are also thought to contribute more to
network expansion than to in-depth relationship formation, which, in turn, provides strong
incentives for weak-tie connectors and bridging social relationships in SNS environments.
Another possible explanation may be related to the fact that bonding social capital
functions as a subcomponent of social capital. This broader construct reflects each individual’s
social network patterns and relational tendencies at individual or community levels, thus bonding
social capital may not serve as an independent mediator.

In sum, the results suggest that openly interacting with others with common health goals via
SNSs influences the ability to expand the breadth of one’s relationships with other members,
promoting more unique, heterogeneous social relations. Contrary to the proposed hypothesis,
which theorized that people whose focus is to build and bolster strong interpersonal ties or
generate more bonding social capital would be similarly motivated and involved in SNS
activities, interactions in SNSs were not linked to greater bonding social capital. Future research

A SOCIAL SUPPORT INTERVENTION ON TWITTER 76
needs to investigate the impact and consequences of regular SNS usage on bonding social
relationships, especially among SNS communities based on specialized mutual interests, such as
shared health goals.

In addition to the complex association between the Twitter intervention and subscales of social
capital, both bridging and bonding social capital significantly predicted overall better health
outcomes among cancer patients. Cancer patients who garnered greater bridging and bonding
social capital are more likely to feel emotionally and physically healthy regardless of their
participation in the Twitter intervention. It was noteworthy that higher levels of bonding social
capital, albeit not significantly linked with the Twitter treatment condition, predicted positive
health changes for participants in the intervention and control groups alike. Bridging social
capital, formed through usage of the Twitter social support group, also had a significant and
positive influence on the patients’ overall health status.
This finding adds further empirical evidence to Putman’s claim: “Of all the domains in
which I have traced the consequences of social capital, in none is the importance of social
connectedness so well established as in the case of health and well-being” (Putnam, 2000, p.
326). Study findings are also consistent with previous research that has attributed social capital
to a wide range of positive health-related outcomes, including emotional and psychological well-
being (Beaudoin & Tao, 2007), overall satisfaction with life (N. B. Ellison et al., 2007),
improved self-image (Phua & Jin, 2011), accumulation of health information and emotional
support resources (Papacharissi & Mendelson, 2011), and increased sense of trust at both the
individual and organizational level (N. B. Ellison et al., 2007; Valenzuela et al., 2009).

A SOCIAL SUPPORT INTERVENTION ON TWITTER 77
Direct, Mediated, and Total Effect of SNS Participation on Health Outcomes

Following the mediational analyses, this study’s conclusions confirmed previous findings related
to the direct influence of SNS activities on patients’ health outcomes while controlling for the
overall influence of bridging and bonding social capital as mediators (Chou et al., 2011; Chung,
2013; Greene et al., 2011; Tang & Yang, 2012). The results confirmed the overall effectiveness
of the interconnectedness in SNSs for achieving intended positive health outcomes.
Most importantly, the Twitter intervention alone had a direct, positive impact on
significantly improving the health of cancer patients independent of the role of influencing
mediators and moderators. Regardless of existing levels of bridging or bonding social capital at
the outset of the study or the amount of gain or loss of social network assets during the
intervention, cancer patients reported significantly better health outcomes when they participated
in the Twitter intervention and engaged in social networking with other patients.
In other words, it was empirically supported that, through engaging in encouraging,
active participation in SNS, patients experienced positive health change, without being affected
by the initial levels of homogeneous or heterogeneous tie strengths or the breadth and depth of
their relationships with others. Bonding social capital, either pre-existing or acquired, during the
intervention did not affect the difference in health outcomes as did the increased level of
bridging social capital. Even those patients who were poor in their relational capital at the
beginning of the intervention, or who were unsuccessful in building trust and emotional
connections with other patients during the intervention, shared in the positive change in their
overall health.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 78

This positive, direct impact of participation in SNSs on cancer-related health outcomes
was confirmed in all four mediational analyses that controlled for the mediating effects of
bridging and bonding social capital, social support, and self-efficacy respectively. After
controlling for self-efficacy and social support, participation in SNSs maintained a significant
association with positive health outcomes among cancer patients.
Participation in SNSs has the ability to offset an individual’s preexisting amount of social
support, his/her sense of control and behavioral efficacy over managing an illness, and the
breadth and depth of friendship networks among cancer patients. This influence allows SNS
participation to facilitate positive physical and emotional health and well-being. Patients’
performance in SNS-related activities and the outcomes of the social relationships in SNS,
including the post-intervention changed levels of perceived social support, self-efficacy, and
bridging and bonding social capital, did not preclude positive, significant associations with
positive health outcomes. The results of the mediational models connected with social support
and self-efficacy will be discussed in a later section.

This finding on the direct effects of SNS activities on health among cancer patients is consistent
with the numerous studies that have explored the associations between social networks and the
health of diverse populations.(Beusterien, Tsay, Gholizadeh, & Su, 2013; Chou et al., 2011;
Chung, 2013; S. C. Kim et al., 2013; Small, Taft, & Brown, 2011). One of the theoretical
supports for the positive impact of social networking activities on health is the diffusion of
medical knowledge through already-established connections among people (Coleman, 1989).

A SOCIAL SUPPORT INTERVENTION ON TWITTER 79
Another plausible interpretation comes from the literature on social identification theory
and subjective norms. Relevant research suggests that when a large number of people in a
community, even a virtual one, engages in formal group activities and trust each other, a strong
and durable social identity is forged (Ashforth & Mael, 1989; Ellemers, Kortekaas, & Ouwerkerk,
1999; Postmes, Spears, & Lea, 2000). Active social interaction on SNSs helps members to
activate and maintain positive self-concepts by conferring self-awareness of status as group
members. The members start to see themselves as “interchangeable exemplars” of the group,
rather than as individuals.
Additionally, greater social identification with other SNS members also significantly
influences members’ perceived similarities to the group’s subjective norms regarding cancer
management and adherence to treatment regimens. According to the literature on peer norms
(Ajzen, 1991; Perkins & Berkowitz, 1986; Rimal & Real, 2003; Yun & Silk, 2011), peer group
members operate under a kind of informal social control that prevents them from engaging in
socially undesirable, in this case health-threatening, actions (Fishbein & Cappella, 2006; Huang
et al., 2013; Morgan & Haglund, 2009). People engage in social observations within their social
groups, seeking clues as to the types of behavior that are appropriate and desirable in various
circumstances. They adhere to these implicit rules, in order to fit in with the social norms of the
group. SNSs enable patients to engage in social observation and behavioral modeling after in-
group members and, thus, foster health-inducing attitudes, values, and behaviors with regard to
illness treatment and management (Han et al., 2011; Jarvenpaa & Leidner, 1998; Walther &
Bunz, 2005; Yee, Bailenson, Urbanek, Chang, & Merget, 2007).

A SOCIAL SUPPORT INTERVENTION ON TWITTER 80
Another interesting finding of the study was that the intervention was proven to be effective and
beneficial in improving health among cancer patients through the direct or indirect role of
bridging social capital as a mediator. Bridging social capital exerted a strong, positive mediating
influence on the association between participation in SNSs and health outcomes, but it was
significant and positive only for bridging social capital, not for bonding social capital.
Health-related SNS activities significantly contribute to positive health changes among
cancer patients. More interestingly, the SNS participants who pursue close-tie relationships by
reaching out to the broader community and trying to widely connect with others, are more likely
to report significantly better health outcomes compared to those with lower bridging social
activities. In other words, bridging social capital and heterogeneous ties create a positive
feedback loop in the relationship between participation in SNSs and overall health outcomes.
When participants engage in broadening their social relations with others in the SNS network,
this increased capacity in their social network and relational assets is more likely to lead to
greater support from others and to more active engagement in online support-seeking behaviors
in SNS. This, in turn, results in better health outcomes. Study findings from the mediational
analyses confirm that the effects of the Twitter intervention in expanding weak-tie bridging
social capital is an essential element of the support- and information-sharing activities in SNS. It
is also a critical mediator which influences enhanced health outcomes through its direct and
indirect impact.

Finally, when the direct and indirect effects of the Twitter intervention, taking into consideration,
the influence of social capital were combined, the total effects of the intervention on overall
health outcomes were significant and positive. Participation in SNS-based social support

A SOCIAL SUPPORT INTERVENTION ON TWITTER 81
activities in community groups, medical groups, or friendship networks leads to positive health
change through the direct and indirect role of social capital.
This finding is consistent with the current literature, which highlights a social network
and relational view in explicating the effects of online communication and connection, with
bridging and reaching-out network participation being the most significant among various styles
of online health activities. Therefore, online health-related activities in SNS, such as
information-oriented or emotional support seeking for cancer patients, are conducive to building
broader, enduring social networks and are effective in translating the values and benefits of
social relations through human relationships.

Self-Efficacy

Participation in the Twitter-based SNS intervention led to positive change in self-efficacy among
cancer patients. Cancer patients who receive support from peers in SNSs were more likely to
feel confident in their ability to engage in cancer treatment, and they perceived more behavioral
control over their illnesses. By tapping into the unique information available through
heterogeneous social groups outside of the regular social boundaries and nurturing the intimate
and close social ties within their individual homogeneous groups, the SNS-intervention
participants gained a greater sense of control and efficacy over their medical conditions. This
also enhanced their ability to cope with challenging health issues (Fishbein & Cappella, 2006;
Judge, Erez, Bono, & Thoresen, 2002; Phua, 2013). Strong in-group identification can,
furthermore, prompt a stronger sense of self-efficacy and control over one’s surroundings
(Bandura, 2001; Tajfel & Turner, 2004). Engagement in SNSs makes the in-group identity

A SOCIAL SUPPORT INTERVENTION ON TWITTER 82
salient among members, facilitating the social learning processes by influencing each participant
to model other group members’ healthful behaviors. Within the SNS framework, members
cultivate emotional connections and commitment to their network groups through shared beliefs,
values, and attitudes. In turn, this may exert a significant impact on the sense of self-behavioral
control over health conditions.

In this study, the heightened self-efficacy from engaging in SNS social groups was not found to
foster positive health outcomes. Despite a significantly positive impact of the Twitter
intervention on self-efficacy, increased beliefs about individualized capacity to adhere to cancer
treatment and disease management did not predict positive health outcomes at post-test.
Similarly, the proposed mediational model between SNS activities and health outcomes with
self-efficacy as a mediator was not established. In other words, participation in the online
support-seeking and sharing behaviors in SNSs is strongly associated with gaining significant
health outcomes, but not specifically through the indirect influence of self-efficacy. The sense of
self-behavioral control and other efficacious feelings were found neither to intensify nor dilute
the significant association between SNS usage and the health outcomes of cancer patients.
It was unexpected that self-efficacy was not associated with positive health outcomes
despite the strong, positive influence of participation in SNSs on self-efficacy. The literature on
self-efficacy has shown that self-efficacy is a necessary antecedent to health-related behavioral
change, and it fosters positive health outcomes through various mechanisms (Deci & Ryan, 1985;
Fishbein & Cappella, 2006; Namkoong et al., 2010; Shaw et al., 2006; K. Wright, 2000).
However, the anticipated association between a heightened sense of behavioral control and
efficacy and an improvement in health did not emerge despite the proposed theoretical model

A SOCIAL SUPPORT INTERVENTION ON TWITTER 83
that SNS participants were more likely to acquire a higher level of self-efficacy than the control
group.
Various studies suggest that SNS users derive a heightened sense of closeness and
interconnectedness to other site members, thereby empowering them to take positive steps
toward exerting greater control over their disease conditions (Han et al., 2008; J. Kim et al., 2010;
Phua, 2013). Unfortunately, a greater sense of self-efficacy with regard to managing one’s
health conditions and participating in treatment was not translated into actual health and well-
being changes among the cancer patients in this study.

As with previous social support interventions for patients, self-efficacious behaviors related to
adherence to the treatment regimens for challenging health conditions were partly an outcome of
informational resources, tangible advice, and encouragement or advice contributed by other
members on the sites (Brendryen, Drozd, & Kraft, 2008; Etter et al., 2000; STODDARD et al.,
2005). Reporting and broadcasting adverse health events in SNSs has recently received
particular attention as a potentially significant mechanism of dissociation between self-efficacy
and health outcomes. SNS users frequently posted individual concerns about the adverse effects
of various cancer treatments or medications in a presumed attempt to see if their own
experiences correlated with that of others.
Similarly, dangerous, misleading, or risky self-medication behaviors or risk-
compensating behaviors supported by Twitter participants is another possible cause of the
dissociation between self-efficacy and actual health outcomes. Although this study was designed
for hypothesis testing and not for qualitative or exploratory analyses, a preliminary thematic
analysis of the tweet streams revealed that about one quarter (24%) of tweets shared sensitive

A SOCIAL SUPPORT INTERVENTION ON TWITTER 84
aspects of cancer management unlikely to be revealed in doctor-patient interactions. For
example, one series of tweets described how to enable extended alcoholic drinking sessions
without risking cancer recurrence. Another series of tweets discussed the metabolic needs of
triathletes between treatments, a highly specialized form of experiential knowledge not validated
by oncologists or endocrinologists. One form of the guidance provided by Twitter participants
was presented as narratives discrediting physicians, medical organizations, and books as limited
sources of information. This seemed to be an attempt to validate the use of SNSs as a superior
source of cancer information. Participants also frequently posted warnings for the newly
diagnosed about the routine difficulties of life as cancer patients and the social stigmatization of
cancer.
It is worth noting that potentially misleading, negative discussions on Twitter also arose
around the important issues of how to adjust expectations and make post-diagnosis lives
manageable. Taken from social identification and social norm theorization, patients may be
confronted with the possibility of negative consequences from being exposed to these tweets.
Within cancer-specific social support groups in SNS, other patients may be seen as “relevant
others,” as specific individuals whose perceived attitudes, beliefs, and behaviors with regard to
cancer treatments act as important clues to members seeking social and emotional approval. By
incorporating the negative social norms and adverse consequences of cancer treatment and
management by other members as their own, gains in self-efficacy to adhere to their treatment
regimens may be offset by the perceived similarity of the group’s unhealthful norms. Thus, the
gains may not be translated into positive health outcomes.
Taken as a whole, this discovery has implications for intervention development, as the
majority of interventions with cancer patients and survivors utilize a self-efficacy model to

A SOCIAL SUPPORT INTERVENTION ON TWITTER 85
improve overall health and well-being outcomes. Results suggest that the SNS processes
influencing the health and self-efficacy of cancer patients are complex and need further
investigation. Future research should explore this and other mediating relationships related to
the key constructs and other triggering and inhibiting variables.

Social Support

Furthermore, as the study found, SNS-specific information supplants or is integrated into users’
other forms of cancer-related support resources, significantly enhancing informational, emotional,
and tangible support. Compared to offline support groups, online social support groups have the
capacity to foster high levels of trust, intimacy, and self-disclosure within a relatively shorter
period of time (Walther & Bunz, 2005). It is clear from this study that SNS participants make
comparatively greater headway in developing supportive resources through Twitter
communication pathways than through traditional medical professional content pertaining to
condition management. Interactive features prevalent in SNS, for example, tweeting with others,
retweeting or mentioning someone’s tweets, asking and answering questions via direct messages,
engaging in health-specific discussion by generating hashtags, and so forth, directly correspond
with individual member’s informational and emotional support needs specific to their medical
conditions and psychological status. This study also indicated that perceived social support
among SNS users was significantly associated with cultivating better physical and emotional
health.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 86
In addition, this research concluded that perceived social support exerted a strong,
positive mediating influence on the association between SNS participation and members’ overall
health. As with previous online social support interventions, through active participation in
SNSs alone, cancer patients reported significantly better health outcomes, revealing a strong,
direct effect of participation in SNSs on positive health outcomes. When cancer patients share
personal information, disease-specific guidance, and feedback, and actually feel they are
supported in the SNS community, this heightened sense of social support makes them physically
and psychologically healthier and stronger compared with those who are less immersed in a
supportive community and who lacked a definable sense of social support.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 87
Limitations

This study has several limitations. The first limitation arose from the study design. Although
analyses adjusted for pre-intervention levels of key variables, there may be other important
sources that were not considered. Thus, future research should aim to more comprehensively
rule out potential confounding effects of the SNS activities on health outcomes. Next, health
outcomes measured cancer-related coping skills and general quality of life. These items captured
perceived rather than actual changes in cancer-related health outcomes following the intervention.
Thus, future research should utilize objective measures for a stronger measurement of the
physical and emotional health outcomes among cancer patients.
Another limitation was related to the sample scheme. Subjects recruited for this SNS
intervention may not be representative of SNS groups focusing on cancer. Even though a
concerted effort was made to recruit a wide range of cancer patients and survivors, the
participants who passed the screening survey were relatively young, white, well-educated people
with relatively recent diagnoses of cancer compared to the general demographic profile of cancer
patients in America. This may have contributed to the sampling bias.
Furthermore, cancer communities in SNSs contain a plurality of participants, not just
patients and survivors, but also family members, government or non-profit organizations,
advertisers, researchers, and news media outlets, all of which have divergent interests and modes
of communication. Further investigation should be pursued into how SNS environments
simultaneously serve as supportive communities, venues for public service announcements,
promotional spaces, and repositories of research subjects and data.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 88
Another weakness was this study’s focus on the social networking activities on Twitter.
Although Twitter is the world’s second largest social networking site, the study did not include
other general SNS sites, such as Facebook, Pinterest, and Instagram, or health-specific SNS sites,
such as PatientsLikeMe and I Had Cancer, which might yield alternate results. Future research
should include a wider range of general and health issue-specific SNSs to establish the
generalizability of the findings. The intervention was also conducted over a limited time period,
thus excluding the longitudinal and seasonal aspects of communication in SNS.
The present study did not analyze the effect of communication style and social influence
in SNSs, such as number of total tweets, number of followers, information in personal profiles,
and site activities in general. Future studies should examine how activity levels in SNSs and
social prominence affect the relationship between SNS activities and health outcomes among
cancer patients.
In addition, this study did not include qualitative or exploratory analyses, which provide a
rich source of data for elusive research subjects like cancer. For example, one plausible
explanation for the reduced impact of self-efficacy on the association between SNS usage and
health could be derived from a preliminary content analysis of the tweet stream on inaccurate,
misguided, or risky health behaviors and attitudes exchanged and promoted on Twitter.
Beyond a level of preliminary screening for obscenity and hate-speech, there are neither
editorial monitors nor fact-checkers on Twitter or Facebook. The inability to verify the source of
information or the identity of the posters, coupled with the prominent use of SNSs for
commercial purposes, may pose a significant issue to the trustworthiness of SNSs as an
indispensable source of informational and supportive sharing for cancer patients.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 89
Despite the limitations, this randomized, controlled SNS intervention with mediational
analyses allowed for the assessment of the impact of information- and support-sharing behaviors
among cancer using the Twitter social media site. This study incorporated key communication
theories as predictors and mediators, including bridging and bonding social capital, social
support, and self-efficacy.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 90
Implications

From this study, it is clear that a SNS-based cancer community could hold the potential to
significantly support, educate, and build connections between patients with similar conditions.
Such a community could also help clarify the pathways of mediation between key concepts, such
as bonding social capital and the issues of self-efficacy and behavioral control over illness
treatment.
SNS, as a mobilized social support community, is a critical conduit to positive health
change due to its capacity for facilitating social network development, sharing social support,
and aiding self-efficacy among cancer patients and survivors. However, further research should
concentrate on the processes and consequences of cancer patients’ increased sense of personal
capability and efficacy in terms of cancer management and treatment adherence. Additional
study should explore how bonding social capital affects changes in health.
Policymakers and health organizations should consider how to develop, deploy, and
evaluate SNS intervention programs to assure successful social networking and informational or
supportive sharing among cancer patients. The clear goal of such efforts would be to promote
physical and emotional health and well-being through social interconnectedness and network
building, health information and support exchange, and heightened awareness and senses of
behavioral control and self-efficacy.
Health professionals, clinicians, and researchers also need to utilize the strengths and
limitations of SNS usage when discussing information on cancer treatment or disease
management with patients.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 91
Conclusion

On several fronts, this study offers empirical support of the proposed health outcomes of SNSs in
the management of severe, chronic diseases like cancer. Overall, the theoretical mediational
model of influence proposed in this research, based on traditional paradigms of social
interconnectedness in SNS, was supported.
Participation in the Twitter-based SNS cancer social support intervention exerted a
significant impact by increasing levels of bridging social capital, providing a forum for seeking
and receiving social support, and fostering self-efficacy. Except self-efficacy, these factors, in
turn, significantly predicted improved health outcomes among cancer patients. Bonding social
capital, despite its lack of association with participation in SNS, also had a strong, positive
influence on acquiring positive health outcomes.
Participation in the Twitter-based SNS intervention also had a significant direct effect on
achieving positive health outcomes, even after controlling for the preexisting levels as well as the
changed amounts of all four predictors and mediators: bridging social support, bonding social
support, social support, and self-efficacy. The support for the proposed mediational model
reaffirms that regular use of SNSs as a community platform is a significant predictor for better
health results among cancer patients, with the preexisting levels and changed values of the key
variables being equal. This strong association between activities in SNSs and receiving health
outcomes was also mediated by the changed scores of bridging social capital and social support.
A few findings from the mediational analyses were unexpected. Despite positive
associations with health outcomes and contrary to the proposed hypotheses, bonding social
capital and self-efficacy were not found to mediate the relationship between participation in

A SOCIAL SUPPORT INTERVENTION ON TWITTER 92
SNSs and health outcomes. Also unexpected was the fact that self-efficacy was not associated
with positive health change nor did it mediate the relationship between SNS usage and health
outcomes despite the significant, positive effect of participation in SNSs on self-efficacy.
Several theoretical explanations and revised hypotheses were proposed. It is still unclear why a
greater sense of self-efficacy with regard to cancer management and adherence to treatment was
not manifested in the form of actual improved health and well-being among cancer patients in
this study. Future research should explore this and other relationships between key variables and
predictors so as to elucidate the processes and mechanisms of participation in SNSs and its
impact on cancer.
At the core of this structure, SNSs provide a forum for exchanging emotional and
logistical support, collectively building and expanding relational networks, and reinforcing self-
efficacy as related to disease management among patients affected by cancer. Collective
network building, emotional and informational support sharing, and the cognitive empowering of
members to take control of their disease management facilitated the achievement of greater
physical health and well-being among cancer patients.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 93
Appendix A. Average number of monthly searches on cancer-related search keywords and
suggested bidding price for each click on Google in September 2012

Ad group Keyword Avg. monthly searches Competition Suggested bid
Keywords like: Cervical Cancer Support Group cancer 135,000 $0.54 $2.76
Keywords like: Cervical Cancer Support Group prostate cancer 110,000 $0.95 $6.19
Pancreatic Cancer pancreatic cancer 110,000 $0.73 $1.68
American Cancer Society american cancer society 110,000 $0.48 $0.86
Lung Cancer lung cancer 90,500 $0.83 $3.59
Colon Cancer colon cancer 90,500 $0.86 $1.44
Ovarian Cancer ovarian cancer 74,000 $0.65 $1.26
Cancer thyroid cancer 74,000 $0.59 $2.51
Keywords like: Cervical Cancer Support Group chemotherapy 60,500 $0.53 $3.08
Cancer testicular cancer 60,500 $0.37 $1.37
Cancer liver cancer 49,500 $0.91 $1.32
Cancer bladder cancer 49,500 $0.80 $1.42
Cancer stomach cancer 49,500 $0.58 $0.77
Keywords like: Graviola oncology 49,500 $0.28 $1.34
Cancer Awareness breast cancer awareness 40,500 $0.75 $3.22
Cancer Treatment cancer treatment centers of america 33,100 $0.78 $1.37
Cancer bone cancer 33,100 $0.79 $1.42
Cancer cancer symptoms 27,100 $0.72 $1.41
Brain brain cancer 22,200 $0.82 $3.07
Keywords like: Graviola graviola 22,200 $0.88 $1.00
Cancer Center cancer centers of america 18,100 $0.88 $1.71
Cancer what is cancer 18,100 $0.31 $1.66
Keywords like: Cervical Cancer Support Group types of cancer 14,800 $0.72 $1.36
Cancer Cure cure for cancer 14,800 $0.56 $5.91
Cancer Institute national cancer institute 14,800 $0.30 $2.54
Cancer stages of cancer 12,100 $0.58 $1.33
Cancer cancer research 9,900 $0.72 $5.52
Cancer blood cancer 9,900 $0.70 $1.26
Cancer breast cancer facts 9,900 $0.80 $2.32
Cancer Treatment prostate cancer treatment 8,100 $0.96 $13.11
Cancer Cure cancer cure 8,100 $0.81 $6.28
Cancer cancer stages 6,600 $0.74 $1.01
Cancer cancer cell 6,600 $0.18 $3.60
Cancer what causes cancer 6,600 $0.39 $1.52
Cancer Treatment cancer treatment 5,400 $0.97 $8.39
Cancer Treatment breast cancer treatment 5,400 $0.95 $8.53
Cancer Survival prostate cancer survival rate 5,400 $0.86 $3.23
Cancer what is prostate cancer 5,400 $0.83 $3.90
Cancer cancer ribbons 5,400 $0.66 $1.41
Cancer Society cancer society 4,400 $0.84 $2.72
Cancer terminal cancer 4,400 $0.41 $3.14
Keywords like: Cervical Cancer Support Group support groups 3,600 $0.57 $2.21
Cancer Treatment lung cancer treatment 3,600 $0.96 $8.29
Cancer cancer statistics 3,600 $0.44 $3.09
Cancer wigs for cancer patients 3,600 $0.95 $1.94
Cancer how to prevent cancer 3,600 $0.30 $2.06
Cancer Awareness cancer awareness 2,900 $0.71 $2.18
Cancer Care cancer care 2,900 $0.72 $2.94
Cancer cancer facts 2,900 $0.78 $2.81
Cancer cancer diet 2,900 $0.88 $2.46
Cancer cancer prevention 2,900 $0.56 $3.09
Cancer cancers 2,900 $0.53 $1.50
Cancer causes of cancer 2,900 $0.48 $2.60
Cancer childhood cancer 2,900 $0.67 $2.76
Cancer bowel cancer 2,900 $0.81 $0.93
Cancer Treatment cancer treatments 2,400 $0.96 $8.73
Cancer Society breast cancer society 2,400 $0.72 $2.48
Cancer cancer types 2,400 $0.71 $1.08
Cancer cancer bracelets 2,400 $0.98 $2.59
Keywords like: Graviola radical prostatectomy 2,400 $0.36 $6.58
Keywords like: Cervical Cancer Support Group divorce support groups 1,900 $0.85 $1.52
Google Adwords Analytics

A SOCIAL SUPPORT INTERVENTION ON TWITTER 94
Appendix B. List of the facilitating tweet assignments and comments

1. Introduce yourself to the group! Share your story to tell the group a little about yourself.
2. Is there someone in your life who gave you courage and will to fight? If so, who inspired
you the most?
3. Anyone after treatment now suffer from mental health problems?
4. Hi everyone, how would you advise regarding do’s and don’ts for radiotherapy?
5. Hi everyone, we'd love to hear some of your favorite cancer-related books. What's at the
top of your list?
6. Is there anything you can recommend to help put our mind at ease so we can stay calm in
remission?
7. Has faith played a role in your fight with cancer? If so, how?
8. If you were working when you were diagnosed with cancer, have you returned to your
job? Why or why not?
9. Do any other survivors suffer from survivors’ guilt?
10. Has anyone lost close friends to cancer? and if so, how do you deal with it?
11. The holiday season can be difficult for cancer survivors and fighters. what do you do to
lift your loved one's spirits?
12. What is something you've done to celebrate the day that you kicked cancer to the curb?
13. Anyone get overwhelmed when going to the doctor or hospital? How do you deal with
these feelings?
14. What do you do to prevent nausea and vomiting?
15. What did you do when you felt lonely while dealing with cancer?

A SOCIAL SUPPORT INTERVENTION ON TWITTER 95
16. Share your funny moments during treatments
17. Is anyone more worried about returning to their normal life than worried about cancer?
18. Have you had trouble sleeping at night?
19. For those that have made the transition to being a cancer survivor: How do you
personally define what is your new normal?
20. Has anyone had family members just being so unsupportive when diagnosed with cancer?
21. Has anyone else experienced the emotional roller coaster?
22. Any long-time survivors have any good tips for paranoia?
23. Does anyone know the best diet for the day after chemo?
24. Is it weird to cry in front of a doctor?
25. How has cancer changed your views on what and who's important in your life?
26. Anyone else have family that won't even speak about or discuss the fact that you had
cancer?
27. What are some activities that supporters can do with survivors as the first steps to help
them be active and get their strength back?
28. Has Yoga helped anyone out? And how?
29. Are you open for alternative medicine for prevention and/or treatment? If you are, what
do you recommend to give a try?
30. Does anyone juice fresh fruits and vegetables?
31. What are some of the most common cancer causing foods/things in our life?
32. What are some of the best foods for cancer prevention?
33. What are some good online cancer support groups that you came across about cancer?
34. Has anyone had other major health issues before being diagnosed with cancer?

A SOCIAL SUPPORT INTERVENTION ON TWITTER 96
35. How do you deal with the side effects and what side effects do you have?
36. Does cancer run in families? Were you aware of that before you get diagnosed?
37. Where do you get advices on pros and cons of treatments and drugs?
38. How did you make a decision about care and treatment?
39. Has anyone out there used mental health therapy during their battle with cancer? If so-
how has it helped you?
40. Has anyone in your family had the same diagnosis?
41. Have any of you received financial help from cancer support institutions?
42. What would you say to someone who was just diagnosed?
43. Do you keep track of your medical & treatment record?
44. Have you stepped out of the workforce after your diagnosis? If so, how did you tell your
boss?
45. What makes you happy now?
46. Have you changed your life style to work on your physical and mental health?
47. Name 3 things you are grateful for.
48. What was your experience of feeling or not feeling "done" with cancer once your
treatment was over?

A SOCIAL SUPPORT INTERVENTION ON TWITTER 97
Appendix C. Wrap-Up Handout on Coping with Caner
Coping with Cancer: Supportive and Palliative Care
http://www.cancer.gov/cancertopics/coping

• Prognosis Resources: http://www.cancer.gov/cancertopics/coping/prognosis
After a cancer diagnosis, discussing prognosis (the likely course of the disease) can
help patients and their loved ones cope and make decisions.
• Managing Physical Effects: http://www.cancer.gov/cancertopics/coping/physicaleffects
Browse a list of common side effects of cancer or cancer treatment with links to
practical information for preventing or relieving these effects. Also find information
on maintaining proper nutrition during cancer treatment.
• Managing Emotional Effects:
http://www.cancer.gov/cancertopics/coping/emotionaleffects
Manage depression, anxiety, and other emotional effects, and learn how to support
people with cancer.
• For Caregivers, Family, and Friends:
http://www.cancer.gov/cancertopics/coping/familyfriends
Information to help caregivers cope while caring for a loved one with cancer, as well
as help someone with cancer talk about and cope with the illness.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 98
• About Children with Cancer: http://www.cancer.gov/cancertopics/coping/children-with-
cancer
Information for parents about children with cancer, as well as guides to help children
and teens cope when a family member has cancer.
• Finding Healthcare Services: http://www.cancer.gov/cancertopics/coping/healthservices
Tips for choosing a doctor or a treatment facility, healthcare options, assistance at
home, and hospice care.
• Financial, Insurance, and Legal Information:
http://www.cancer.gov/cancertopics/coping/financial-legal
Information about getting financial help, insurance coverage, and support
organizations.
• Survivorship - Living With and Beyond Cancer:
http://www.cancer.gov/cancertopics/coping/survivorship
Life and health after a cancer diagnosis and once treatment is over.
• Preparing for the End of Life: http://www.cancer.gov/cancertopics/coping/end-of-life
Information for patients, their families, and friends about grief and bereavement,
hospice care, and end-of-life planning.
• Supportive and Palliative Care Clinical Trials:
http://www.cancer.gov/cancertopics/coping/supportive-clinical-trials
Clinical trials to help people cope with symptoms and side effects of cancer.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 99
References
Afifi, Walid A, & Weiner, Judith L. (2004). Toward a theory of motivated information
management. Communication Theory, 14(2), 167-190.
Ajzen, Icek. (1991). The theory of planned behavior. Organizational behavior and human
decision processes, 50(2), 179-211.
Albrecht, Terrance L, & Adelman, Mara B. (1987). Communicating social support: Sage
Publications, Inc.
Albrecht, Terrance L, Burleson, Brant R, & Goldsmith, Dana. (1994). Supportive
communication. Handbook of interpersonal communication, 2, 419-449.
Armstrong, Natalie, & Powell, John. (2009). Patient perspectives on health advice posted on
Internet discussion boards: a qualitative study. Health Expectations, 12(3), 313-320.
Arnst, Catherine. (2008). Health 2.0: Patients as Partners. Business Week.
Arntson, Paul, Droge, David, Albrecht, TL, & Adelman, MB. (1987). Social support in self-help
groups: The role of communication in enabling perceptions of control. Communicating
social support, 148-171.
Ashforth, Blake E, & Mael, Fred. (1989). Social identity theory and the organization. Academy
of management review, 14(1), 20-39.
Balka, Ellen, Krueger, Guenther, Holmes, Bev J, & Stephen, Joanne E. (2010). Situating Internet
Use: Information‐ Seeking Among Young Women with Breast Cancer. Journal of
Computer‐ Mediated Communication, 15(3), 389-411.
Bandura, Albert. (1977). Self-efficacy: toward a unifying theory of behavioral change.
Psychological review, 84(2), 191.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 100
Bandura, Albert. (2001). Social cognitive theory: An agentic perspective. Annual review of
psychology, 52(1), 1-26.
Barak, Azy, Boniel-Nissim, Meyran, & Suler, John. (2008). Fostering empowerment in online
support groups. Computers in Human Behavior, 24(5), 1867-1883.
Barbee, AP, & Cunningham, MR. (1995). An experimental approach to social support
communications: Interactive coping in close relationships. Communication yearbook, 18,
381-413.
Beaudoin, Christopher E, & Tao, Chen-Chao. (2007). Benefiting from social capital in online
support groups: An empirical study of cancer patients. CyberPsychology & Behavior,
10(4), 587-590.
Bender, Jacqueline L, Jimenez-Marroquin, Maria-Carolina, & Jadad, Alejandro R. (2011).
Seeking support on facebook: a content analysis of breast cancer groups. Journal of
medical Internet research, 13(1).
Bender, Jacqueline L, Wiljer, David, To, Matthew J, Bedard, Philippe L, Chung, Peter, Jewett,
Michael AS, . . . Gospodarowicz, Mary. (2012). Testicular cancer survivors' supportive
care needs and use of online support: a cross-sectional survey. Supportive Care in Cancer,
20(11), 2737-2746.
Berkman, Lisa F, Glass, Thomas, Brissette, Ian, & Seeman, Teresa E. (2000). From social
integration to health: Durkheim in the new millennium. Social science & medicine, 51(6),
843-857.
Beusterien, Kathleen, Tsay, Sarah, Gholizadeh, Shadi, & Su, Yun. (2013). Real-world
experience with colorectal cancer chemotherapies: patient web forum analysis.
ecancermedicalscience, 7.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 101
Breckons, Matthew, Jones, Ray, Morris, Jenny, & Richardson, Janet. (2008). What do evaluation
instruments tell us about the quality of complementary medicine information on the
internet? Journal of medical Internet research, 10(1).
Brendryen, Håvar, Drozd, Filip, & Kraft, Pål. (2008). A digital smoking cessation program
delivered through internet and cell phone without nicotine replacement (happy ending):
randomized controlled trial. Journal of Medical Internet Research, 10(5).
Burleson, Brant R, Albrecht, Terrance L, & Sarason, Irwin G. (1994). Communication of social
support: Messages, interactions, relationships, and community: Sage Publications, Inc.
Burlingame, Gary M, Fuhriman, Addie, & Mosier, Julie. (2003). The differential effectiveness of
group psychotherapy: A meta-analytic perspective. Group Dynamics: Theory, Research,
and Practice, 7(1), 3.
Cacioppo, John T, Berntson, Gary G, Larsen, Jeff T, Poehlmann, Kirsten M, & Ito, Tiffany A.
(2000). The psychophysiology of emotion. Handbook of emotions, 2, 173-191.
Cella, DF, & Yellen, SB. (1992). Cancer support groups: the state of the art. Cancer Practice,
1(1), 56-61.
Cheng, Patricia, Gutierrez-Colina, Ana M, Loiselle, Kristin A, Strieper, Margaret, Frias, Patrick,
Gooden, Kevin, & Blount, Ronald L. (2013). Health Related Quality of Life and Social
Support in Pediatric Patients with Pacemakers. Journal of clinical psychology in medical
settings, 1-11.
Chmura Kraemer, Helena, Kiernan, Michaela, Essex, Marilyn, & Kupfer, David J. (2008). How
and why criteria defining moderators and mediators differ between the Baron & Kenny
and MacArthur approaches. Health Psychology, 27(2S), S101.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 102
Chou, Wen-ying Sylvia, Liu, Benmei, Post, Samantha, & Hesse, Bradford. (2011). Health-
related Internet use among cancer survivors: data from the Health Information National
Trends Survey, 2003–2008. Journal of Cancer Survivorship, 5(3), 263-270.
Chretien, Katherine C, Azar, Justin, & Kind, Terry. (2011). Physicians on twitter. JAMA: the
journal of the American Medical Association, 305(6), 566-568.
Christakis, Nicholas A, & Allison, Paul D. (2006). Mortality after the hospitalization of a spouse.
New England Journal of Medicine, 354(7), 719-730.
Christakis, Nicholas A, & Fowler, James H. (2008). The collective dynamics of smoking in a
large social network. New England journal of medicine, 358(21), 2249-2258.
Chung, Jae Eun. (2013). Social interaction in online support groups: Preference for online social
interaction over offline social interaction. Computers in Human Behavior, 29(4), 1408-
1414.
Cohen, Sheldon. (1988). Psychosocial models of the role of social support in the etiology of
physical disease. Health psychology, 7(3), 269.
Coleman, James S. (1989). Social capital in the creation of human capital: University of
Chicago Press.
Cutrona, Carolyn E, & Russell, Daniel W. (1990). Type of social support and specific stress:
Toward a theory of optimal matching.
De la Torre-Díez, Isabel, Díaz-Pernas, Francisco Javier, & Antón-Rodríguez, Míriam. (2012). A
content analysis of chronic diseases social groups on Facebook and Twitter. Telemedicine
and e-Health, 18(6), 404-408.
Deci, Edward L, & Ryan, Richard M. (1985). The general causality orientations scale: Self-
determination in personality. Journal of research in personality, 19(2), 109-134.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 103
Desai, Tejas, Shariff, Afreen, Shariff, Aabid, Kats, Mark, Fang, Xiangming, Christiano, Cynthia,
& Ferris, Maria. (2012). Tweeting the meeting: an in-depth analysis of Twitter activity at
Kidney Week 2011. PloS one, 7(7), e40253.
Digitome (Producer). (2011). Patient Social Networks. Retrieved from http://digito.me/patient-
social-networks/
Drentea, Patricia, & Moren‐ Cross, Jennifer L. (2005). Social capital and social support on the
web: the case of an internet mother site. Sociology of health & illness, 27(7), 920-943.
Duggan, Maeve, & Brenner, Joanna. (2013). The demographics of social media users, 2012 (Vol.
14): Pew Research Center's Internet & American Life Project.
Duggan, Maeve, & Smith, Aaron. (2013). Social media update 2013. Pew Internet and American
Life Project.
Durkheim, Emile. (1951). Suicide: A study in sociology (JA Spaulding & G. Simpson, Trans.).
Glencoe, IL: Free Press.(Original work published 1897).
Edelman. (2012). Edelman Trust Barometer 2012: Executive Summary: Edelman.
Elkin, Noah. (2008). How America searches: health and wellness. See http://www. icrossing.
com/icrossing-how-america-searches (last checked 21 February 2012).
Ellemers, Naomi, Kortekaas, Paulien, & Ouwerkerk, Jaap W. (1999). Self-categorisation,
commitment to the group and group self-esteem as related but distinct aspects of social
identity. European journal of social psychology, 29(23), 371-389.
Ellison, NB, & boyd, d. (2013). Sociality through social network sites. The Oxford handbook of
internet studies, 151-172.
Ellison, Nicole B. (2007). Social network sites: Definition, history, and scholarship. Journal of
Computer‐ Mediated Communication, 13(1), 210-230.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 104
Ellison, Nicole B, Steinfield, Charles, & Lampe, Cliff. (2007). The benefits of Facebook
“friends:” Social capital and college students’ use of online social network sites. Journal
of Computer‐ Mediated Communication, 12(4), 1143-1168.
Etter, Jean‐ François, Bergman, Manfred Max, Humair, Jean‐ Paul, & Perneger, Thomas V.
(2000). Development and validation of a scale measuring self‐ efficacy of current and
former smokers. Addiction, 95(6), 901-913.
Eysenbach, Gunther. (2008). Medicine 2.0: social networking, collaboration, participation,
apomediation, and openness. Journal of medical Internet research, 10(3).
Eysenbach, Gunther, Powell, John, Englesakis, Marina, Rizo, Carlos, & Stern, Anita. (2004).
Health related virtual communities and electronic support groups: systematic review of
the effects of online peer to peer interactions. Bmj, 328(7449), 1166.
Farmer, AD, Holt, CEM Bruckner, Cook, MJ, & Hearing, SD. (2009). Social networking sites: a
novel portal for communication. Postgraduate medical journal, 85(1007), 455-459.
Public Hearing on Promotion of FDA-Regulated Medical Products Using the Internet and Social
Media Tools (2009).
Festinger, Leon. (1954). A theory of social comparison processes. Human relations, 7(2), 117-
140.
Fishbein, Martin, & Cappella, Joseph N. (2006). The role of theory in developing effective
health communications. Journal of Communication, 56(s1), S1-S17.
Fogel, Joshua, Albert, Steven M, Schnabel, Freya, Ditkoff, Beth Ann, & Neugut, Alfred I. (2002).
Internet use and social support in women with breast cancer. Health Psychology, 21(4),
398.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 105
Fortinsky, Kyle J, Fournier, Marc R, & Benchimol, Eric I. (2012). Internet and electronic
resources for inflammatory bowel disease: a primer for providers and patients.
Inflammatory bowel diseases, 18(6), 1156-1163.
Fowler, James H, & Christakis, Nicholas A. (2008). Estimating peer effects on health in social
networks: A response to Cohen-Cole and Fletcher; Trogdon, Nonnemaker, Pais. Journal
of health economics, 27(5), 1400.
Fox, Susannah. (2011). Peer to Peer Health Care (I. Tech, Trans.): Pew Research.
Freimuth, Vicki S, & Quinn, Sandra Crouse. (2004). The contributions of health communication
to eliminating health disparities. American Journal of Public Health, 94(12), 2053.
Giordano, Giuseppe N, & Lindstrom, Martin. (2010). The impact of changes in different aspects
of social capital and material conditions on self-rated health over time: a longitudinal
cohort study. Social science & medicine, 70(5), 700-710.
Gotsis, Marientina, Wang, Hua, Spruijt-Metz, Donna, Jordan-Marsh, Maryalice, & Valente,
Thomas William. (2013). Wellness partners: Design and evaluation of a web-based
physical activity diary with social gaming features for Adults. JMIR Research Protocols,
2(1).
Gottlieb, Benjamin H, & Wagner, Fred. (1991). Stress and support processes in close
relationships The social context of coping (pp. 165-188): Springer.
Greene, Jeremy A, Choudhry, Niteesh K, Kilabuk, Elaine, & Shrank, William H. (2011). Online
social networking by patients with diabetes: a qualitative evaluation of communication
with Facebook. Journal of general internal medicine, 26(3), 287-292.
Gustafson, David H, Hawkins, Robert, McTavish, Fiona, Pingree, Suzanne, Chen, Wei Chih,
Volrathongchai, Kanittha, . . . Serlin, Ronald C. (2008). Internet‐ Based Interactive

A SOCIAL SUPPORT INTERVENTION ON TWITTER 106
Support for Cancer Patients: Are Integrated Systems Better? Journal of Communication,
58(2), 238-257.
Han, Jeong Yeob, Kim, Jung-Hyun, Yoon, Hye Jin, Shim, Minsun, McTavish, Fiona M, &
Gustafson, David H. (2012). Social and psychological determinants of levels of
engagement with an online breast cancer support group: posters, lurkers, and nonusers.
Journal of health communication, 17(3), 356-371.
Han, Jeong Yeob, Shah, Dhavan V, Kim, Eunkyung, Namkoong, Kang, Lee, Sun-Young, Moon,
Tae Joon, . . . Gustafson, David H. (2011). Empathic exchanges in online cancer support
groups: distinguishing message expression and reception effects. Health communication,
26(2), 185-197.
Han, Jeong Yeob, Shaw, Bret R, Hawkins, Robert P, Pingree, Suzanne, McTavish, Fiona, &
Gustafson, David H. (2008). Expressing positive emotions within online support groups
by women with breast cancer. Journal of Health Psychology, 13(8), 1002-1007.
Hawe, Penelope, & Shiell, Alan. (2000). Social capital and health promotion: a review. Social
science & medicine, 51(6), 871-885.
Hawn, Carleen. (2009). Take two aspirin and tweet me in the morning: how Twitter, Facebook,
and other social media are reshaping health care. Health affairs, 28(2), 361-368.
Haythornthwaite, Caroline. (2005). Social networks and Internet connectivity effects.
Information, Community & Society, 8(2), 125-147.
Heaivilin, N, Gerbert, B, Page, JE, & Gibbs, JL. (2011). Public health surveillance of dental pain
via Twitter. Journal of dental research, 90(9), 1047-1051.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 107
Hofer, Matthias, & Aubert, Viviane. (2013). Perceived bridging and bonding social capital on
Twitter: Differentiating between followers and followees. Computers in Human Behavior,
29(6), 2134-2142.
Høybye, Mette Terp, Johansen, Christoffer, & Tjørnhøj‐ Thomsen, Tine. (2005). Online
interaction. Effects of storytelling in an internet breast cancer support group. Psycho‐ Oncology, 14(3), 211-220.
Høybye, MT, Dalton, Susanne Oksbjerg, Deltour, I, Bidstrup, PE, Frederiksen, K, & Johansen, C.
(2010). Effect of Internet peer-support groups on psychosocial adjustment to cancer: a
randomised study. British journal of cancer, 102(9), 1348-1354.
Huang, Grace C, Unger, Jennifer B, Soto, Daniel, Fujimoto, Kayo, Pentz, Mary Ann, Jordan-
Marsh, Maryalice, & Valente, Thomas W. (2013). Peer influences: The impact of online
and offline friendship networks on adolescent smoking and alcohol use. Journal of
Adolescent Health.
Inc., Twitter. (2013). Annual Report 2013.
Jain, Sachin H. (2009). Practicing medicine in the age of Facebook. New England Journal of
Medicine, 361(7), 649-651.
James, Lawrence R, & Brett, Jeanne M. (1984). Mediators, moderators, and tests for mediation.
Journal of Applied Psychology, 69(2), 307.
Jarvenpaa, Sirkka L, & Leidner, Dorothy E. (1998). Communication and trust in global virtual
teams. Journal of Computer‐ Mediated Communication, 3(4), 0-0.
Jerusalem, Matthias, & Schwarzer, Ralf. (1992). Self-efficacy as a resource factor in stress
appraisal processes.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 108
Jimison, Holly, Gorman, Paul, Woods, Susan, Nygren, Peggy, Walker, Miranda, Norris, Susan,
& Hersh, William. (2008). Barriers and Drivers of Health Information Technology Use
for the Elderly, Chronically III, and Underserved.
Johnson, J David, & Meischke, Hendrika. (1993). A Comprehensive Model of Cancer‐ Related
Information Seeking Applied to Magazines. Human Communication Research, 19(3),
343-367.
Jordan-Marsh, Maryalice. (2010). Health technology literacy: A transdisciplinary framework for
consumer-oriented practice: Jones & Bartlett Publishers.
Judd, Charles M, & Kenny, David A. (1981). Process analysis estimating mediation in treatment
evaluations. Evaluation review, 5(5), 602-619.
Judge, Timothy A, Erez, Amir, Bono, Joyce E, & Thoresen, Carl J. (2002). Are measures of self-
esteem, neuroticism, locus of control, and generalized self-efficacy indicators of a
common core construct? Journal of personality and social psychology, 83(3), 693.
Kawachi, Ichiro, Kennedy, Bruce P, & Glass, Roberta. (1999). Social capital and self-rated
health: a contextual analysis. American journal of public health, 89(8), 1187-1193.
Kawachi, Ichiro, Kennedy, Bruce P, Lochner, Kimberly, & Prothrow-Stith, Deborah. (1997).
Social capital, income inequality, and mortality. American journal of public health, 87(9),
1491-1498.
Kawachi, Ichiro, Subramanian, SV, & Anderson, N. (2004). Social capital and health.
Encyclopedia on health and behavior. Thousand Oaks, CA: Sage Publications, 750-754.
Kim, Dohoon, & Chang, Hyejung. (2007). Key functional characteristics in designing and
operating health information websites for user satisfaction: An application of the

A SOCIAL SUPPORT INTERVENTION ON TWITTER 109
extended technology acceptance model. International Journal of Medical Informatics,
76(11), 790-800.
Kim, Junghyun, Han, Jeong Yeob, Shaw, Bret, McTavish, Fiona, & Gustafson, David. (2010).
The Roles of Social Support and Coping Strategies in Predicting Breast Cancer Patients’
Emotional Well-being Testing Mediation and Moderation Models. Journal of health
psychology, 15(4), 543-552.
Kim, Sojung Claire, Shah, Dhavan V, Namkoong, Kang, McTavish, Fiona M, & Gustafson,
David H. (2013). Predictors of Online Health Information Seeking Among Women with
Breast Cancer: The Role of Social Support Perception and Emotional Well‐ Being.
Journal of Computer‐ Mediated Communication, 18(2), 98-118.
Klemm, Paula. (2012). Effects of online support group format (moderated vs peer-led) on
depressive symptoms and extent of participation in women with breast cancer. Computers
Informatics Nursing, 30(1), 9-18.
Koch-Weser, Susan, Bradshaw, Ylisabyth S, Gualtieri, Lisa, & Gallagher, Susan S. (2010). The
Internet as a health information source: findings from the 2007 Health Information
National Trends Survey and implications for health communication. Journal of health
communication, 15(S3), 279-293.
Kontos, Emily Z, Emmons, Karen M, Puleo, Elaine, & Viswanath, K. (2010). Communication
inequalities and public health implications of adult social networking site use in the
United States. Journal of health communication, 15(S3), 216-235.
Kordzadeh, Nima, & Warren, John. (2013). Toward a typology of health 2.0 collaboration
platforms and websites. Health and Technology, 3(1), 37-50.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 110
Kornblith, Alice B, Herndon, James E, Zuckerman, Enid, Viscoli, Catherine M, Horwitz, Ralph I,
Cooper, M Robert, . . . Budman, Daniel. (2001). Social support as a buffer to the
psychological impact of stressful life events in women with breast cancer. Cancer, 91(2),
443-454.
Kubben, Pieter L. (2011). Twitter for neurosurgeons. Surgical neurology international, 2(1), 28.
Leonard, Rosemary, & Onyx, Jenny. (2003). Networking through loose and strong ties: an
Australian qualitative study. Voluntas: International Journal of Voluntary and Nonprofit
Organizations, 14(2), 189-203.
Leventhal, Howard, Diefenbach, Michael, & Leventhal, Elaine A. (1992). Illness cognition:
using common sense to understand treatment adherence and affect cognition interactions.
Cognitive therapy and research, 16(2), 143-163.
Leydon, Geraldine M, Boulton, Mary, Moynihan, Clare, Jones, Alison, Mossman, Jean,
Boudioni, Markella, & McPherson, Klim. (2000). Cancer patients' information needs and
information seeking behaviour: in depth interview study. Bmj, 320(7239), 909-913.
Lieberman, Morton A, & Goldstein, Benjamin A. (2006). Not all negative emotions are equal:
The role of emotional expression in online support groups for women with breast cancer.
Psycho‐ Oncology, 15(2), 160-168.
Linden, Wolfgang, & Vodermaier, Andrea. (2012). Mismatch of desired versus perceived social
support and associated levels of anxiety and depression in newly diagnosed cancer
patients. Supportive Care in Cancer, 20(7), 1449-1456.
Little, Roderick JA. (1988). A test of missing completely at random for multivariate data with
missing values. Journal of the American Statistical Association, 83(404), 1198-1202.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 111
MacKinnon, David P, & Luecken, Linda J. (2008). How and for whom? Mediation and
moderation in health psychology. Health Psychology, 27(2S), S99.
McLaughlin, Margaret, Nam, Yujung, Gould, Jessica, Pade, Courtney, Meeske, Kathleen A,
Ruccione, Kathleen S, & Fulk, Janet. (2012). A videosharing social networking
intervention for young adult cancer survivors. Computers in Human Behavior, 28(2),
631-641.
Meier, Andrea, Lyons, Elizabeth J, Rimer, Barbara K, Frydman, Gilles, & Forlenza, Michael.
(2007). How cancer survivors provide support on cancer-related Internet mailing lists.
Journal of Medical Internet Research, 9(2).
Morgan, Antony, & Haglund, Bo JA. (2009). Social capital does matter for adolescent health:
evidence from the English HBSC study. Health Promotion International, 24(4), 363-372.
Muntaner, Carles. (2004). Commentary: social capital, social class, and the slow progress of
psychosocial epidemiology. International Journal of Epidemiology, 33(4), 674-680.
Murray, Elizabeth, Burns, JSST, See, Tai S, Lai, R, & Nazareth, Irwin. (2005). Interactive
Health Communication Applications for people with chronic disease. Cochrane Database
Syst Rev, 4.
Muthén, Linda K, & Muthén, Bengt O. (2002). How to use a Monte Carlo study to decide on
sample size and determine power. Structural Equation Modeling, 9(4), 599-620.
Namkoong, Kang, McLaughlin, Bryan, Yoo, Woohyun, Hull, Shawnika J, Shah, Dhavan V, Kim,
Sojung C, . . . McTavish, Fiona M. (2013). The Effects of Expression: How Providing
Emotional Support Online Improves Cancer Patients’ Coping Strategies. JNCI
Monographs, 2013(47), 169-174.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 112
Namkoong, Kang, Shah, Dhavan V, Han, Jeong Yeob, Kim, Sojung Claire, Yoo, Woohyun, Fan,
David, . . . Gustafson, David H. (2010). Expression and reception of treatment
information in breast cancer support groups: How health self-efficacy moderates effects
on emotional well-being. Patient education and counseling, 81, S41-S47.
Norris, Pippa. (2004). The bridging and bonding role of online communities: Thousand Oaks,
California: Sage Publications.
Novotny, Paul J, Smith, Denise J, Guse, Lorna, Rummans, Teresa A, Hartmann, Lynn, Alberts,
Steven, . . . Sloan, Jeff A. (2010). A pilot study assessing social support among cancer
patients enrolled on clinical trials: a comparison of younger versus older adults. Cancer
management and research, 2, 133.
Owen, Jason E, Klapow, Joshua C, Roth, David L, & Tucker, Diane C. (2004). Use of the
internet for information and support: disclosure among persons with breast and prostate
cancer. Journal of behavioral medicine, 27(5), 491-505.
Pan, Shuya, & Jordan-Marsh, Maryalice. (2010). Internet use intention and adoption among
Chinese older adults: From the expanded technology acceptance model perspective.
Computers in human behavior, 26(5), 1111-1119.
Papacharissi, Zizi. (2010). A Networked self: identity, community, and culture on social network
sites: Routledge.
Papacharissi, Zizi, & Mendelson, Andrew L. (2011). Look at Us: Collective Narcissism in
College Student Facebook Photo Galleries. A Networked Self: Identity, Community, and
Culture on Social Network Sites, 251-271.
Pearce, Neil, & Davey Smith, George. (2003). Is social capital the key to inequalities in health?
American journal of public health, 93(1), 122-129.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 113
Pearl, Judea. (2001). Direct and indirect effects. Paper presented at the Proceedings of the
seventeenth conference on uncertainty in artificial intelligence.
Pearl, Judea. (2012). The causal mediation formula—a guide to the assessment of pathways and
mechanisms. Prevention Science, 13(4), 426-436.
Perkins, H Wesley, & Berkowitz, Alan D. (1986). Perceiving the community norms of alcohol
use among students: some research implications for campus alcohol education
programming*. Substance Use & Misuse, 21(9-10), 961-976.
Phua, Joe. (2013). Participating in Health Issue‐ Specific Social Networking Sites to Quit
Smoking: How Does Online Social Interconnectedness Influence Smoking Cessation
Self‐ Efficacy? Journal of Communication, 63(5), 933-952.
Phua, Joe, & Jin, Seung-A Annie. (2011). ‘Finding a home away from home’: the use of social
networking sites by Asia-Pacific students in the United States for bridging and bonding
social capital. Asian Journal of Communication, 21(5), 504-519.
Pickering, Jeanne M, & King, John Leslie. (1995). Hardwiring weak ties: Interorganizational
computer-mediated communication, occupational communities, and organizational
change. Organization Science, 6(4), 479-486.
Pinar, Gul, Okdem, Seyda, Buyukgonenc, Lale, & Ayhan, Ali. (2012). The relationship between
social support and the level of anxiety, depression, and quality of life of Turkish women
with gynecologic cancer. Cancer nursing, 35(3), 229-235.
Portier, Kenneth, Greer, Greta E, Rokach, Lior, Ofek, Nir, Wang, Yafei, Biyani, Prakhar, . . .
Mitra, Prasenjit. (2013). Understanding topics and sentiment in an online cancer survivor
community. JNCI Monographs, 2013(47), 195-198.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 114
Portnoy, David B, Scott-Sheldon, Lori AJ, Johnson, Blair T, & Carey, Michael P. (2008).
Computer-delivered interventions for health promotion and behavioral risk reduction: a
meta-analysis of 75 randomized controlled trials, 1988–2007. Preventive medicine, 47(1),
3-16.
Posluszny, Donna M, Hyman, Kelly B, & Baum, Andrew. (2002). Group Interventions in Cancer
Theory and research on small groups (pp. 87-105): Springer.
Postmes, Tom, Spears, Russell, & Lea, Martin. (1998). Breaching or building social boundaries?
SIDE-effects of computer-mediated communication. Communication research, 25(6),
689-715.
Postmes, Tom, Spears, Russell, & Lea, Martin. (2000). The formation of group norms in
computer‐ mediated communication. Human communication research, 26(3), 341-371.
Preacher, Kristopher J, & Hayes, Andrew F. (2008). Asymptotic and resampling strategies for
assessing and comparing indirect effects in multiple mediator models. Behavior research
methods, 40(3), 879-891.
Putnam. (2000). Bowling alone: The collapse and revival of American community: Simon and
Schuster.
Putnam. (2001). Social capital: Measurement and consequences. Canadian Journal of Policy
Research, 2(1), 41-51.
Radina, M Elise, Ginter, Amanda C, Brandt, Julie, Swaney, Jan, & Longo, Daniel R. (2011).
Breast cancer patients' use of health information in decision making and coping. Cancer
nursing, 34(5), E1-E12.
Radloff, Lenore Sawyer. (1977). The CES-D scale a self-report depression scale for research in
the general population. Applied psychological measurement, 1(3), 385-401.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 115
Rajani, R, Berman, DS, & Rozanski, A. (2011). Social networks—are they good for your health?
The era of Facebook and Twitter. QJM, 104(9), 819-820.
Resnick, Paul. (2001). Beyond bowling together: Sociotechnical capital. HCI in the New
Millennium(77), 247-272.
Richardson, A. (2012). Online public health interventions: A good strategy for those with mental
illness. J. Mass. Commun. Journalism, 2, e126.
Rimal, Rajiv N, & Real, Kevin. (2003). Understanding the influence of perceived norms on
behaviors. Communication Theory, 13(2), 184-203.
Rimer, Barbara K, Lyons, Elizabeth J, Ribisl, Kurt M, Bowling, J Michael, Golin, Carol E,
Forlenza, Michael J, & Meier, Andrea. (2005). How new subscribers use cancer-related
online mailing lists. Journal of medical internet research, 7(3).
Robins, James M. (2003). Semantics of causal DAG models and the identification of direct and
indirect effects. Highly structured stochastic systems, 70-81.
Robins, James M, & Greenland, Sander. (1992). Identifiability and exchangeability for direct and
indirect effects. Epidemiology, 3(2), 143-155.
Rodgers, Shelly, & Chen, Qimei. (2005). Internet community group participation: Psychosocial
benefits for women with breast cancer. Journal of Computer‐ Mediated Communication,
10(4), 00-00.
Sarasohn-Kahn, J. (2012). The wisdom of patients: Health care meets online social media.
California HealthCare Foundation (2008).
Sarasohn-Kahn, Jane. (2008). The wisdom of patients: Health care meets online social media:
California HealthCare Foundation.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 116
Sayers, Steven L, Riegel, Barbara, Pawlowski, Stephanie, Coyne, James C, & Samaha, Frederick
F. (2008). Social support and self-care of patients with heart failure. Annals of Behavioral
Medicine, 35(1), 70-79.
Schaefer, Catherine, Coyne, James C, & Lazarus, Richard S. (1981). The health-related functions
of social support. Journal of behavioral medicine, 4(4), 381-406.
Schneider, A, Jackson, R, & Baum, N. (2009). Social media networking: Facebook and Twitter.
The Journal of medical practice management: MPM, 26(3), 156-157.
Shapiro, Joseph. (2009). Patients turn to online community for help healing. National Public
Radio.
Shaw, Bret R, Hawkins, Robert, McTavish, Fiona, Pingree, Suzanne, & Gustafson, David H.
(2006). Effects of insightful disclosure within computer mediated support groups on
women with breast cancer. Health Communication, 19(2), 133-142.
Signorini, Alessio, Segre, Alberto Maria, & Polgreen, Philip M. (2011). The use of Twitter to
track levels of disease activity and public concern in the US during the influenza A H1N1
pandemic. PloS one, 6(5), e19467.
Skoric, Marko M, Ying, Deborah, & Ng, Ying. (2009). Bowling online, not alone: Online social
capital and political participation in Singapore. Journal of Computer‐ Mediated
Communication, 14(2), 414-433.
Small, Rhonda, Taft, Angela J, & Brown, Stephanie J. (2011). The power of social connection
and support in improving health: lessons from social support interventions with
childbearing women. BMC public health, 11(Suppl 5), S4.
Society, American Cancer. (2014). Cancer Facts & Figures 2014. Atlanta: American Cancer
Society.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 117
Song, Hayeon, Nam, Yujung, Gould, Jessica, Sanders, W Scott, McLaughlin, Margaret, Fulk,
Janet, . . . Ruccione, Kathleen S. (2012). Cancer survivor identity shared in a social media
intervention. Journal of Pediatric Oncology Nursing, 29(2), 80-91.
Stefanone, Michael A, Kwon, Kyounghee Hazel, & Lackaff, Derek. (2012). Exploring the
relationship between perceptions of social capital and enacted support online. Journal of
Computer‐ Mediated Communication, 17(4), 451-466.
STODDARD, JACQUELINE L, DELUCCHI, KEVIN L, MUÑOZ, RICARDO F, COLLINS,
NOAH M, PÉREZ Stable, ELISEO J, Augustson, Erik, & LENERT, LESLIE L. (2005).
Smoking cessation research via the internet: a feasibility study. Journal of health
communication, 10(1), 27-41.
Sugawara, Yuya, Narimatsu, Hiroto, Hozawa, Atsushi, Shao, Li, Otani, Katsumi, & Fukao,
Akira. (2012). Cancer patients on Twitter: a novel patient community on social media.
BMC research notes, 5(1), 699.
Szreter, Simon, & Woolcock, Michael. (2004). Health by association? Social capital, social
theory, and the political economy of public health. International Journal of Epidemiology,
33(4), 650-667.
Tajfel, Henri, & Turner, John C. (2004). The Social Identity Theory of Intergroup Behavior.
Tang, Xuning, & Yang, Christopher C. (2012). Ranking user influence in healthcare social media.
ACM Transactions on Intelligent Systems and Technology (TIST), 3(4), 73.
Thackeray, Rosemary, Neiger, Brad L, Smith, Amanda K, & Van Wagenen, Sarah B. (2012).
Adoption and use of social media among public health departments. BMC public health,
12(1), 242.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 118
Thompson, Lindsay A, Dawson, Kara, Ferdig, Richard, Black, Erik W, Boyer, J, Coutts, Jade, &
Black, Nicole Paradise. (2008). The intersection of online social networking with medical
professionalism. Journal of General Internal Medicine, 23(7), 954-957.
Tustin, Nupur. (2010). The role of patient satisfaction in online health information seeking.
Journal of Health Communication, 15(1), 3-17.
Valente, Thomas W, Unger, Jennifer B, & Johnson, C Anderson. (2005). Do popular students
smoke? The association between popularity and smoking among middle school students.
Journal of Adolescent Health, 37(4), 323-329.
Valenzuela, Sebastián, Park, Namsu, & Kee, Kerk F. (2009). Is There Social Capital in a Social
Network Site?: Facebook Use and College Students' Life Satisfaction, Trust, and
Participation1. Journal of Computer‐ Mediated Communication, 14(4), 875-901.
Van Alstyne, Marshall, & Brynjolfsson, Erik. (2005). Global village or cyber-balkans? Modeling
and measuring the integration of electronic communities. Management Science, 51(6),
851-868.
van Uden-Kraan, Cornelia F, Drossaert, Constance HC, Taal, Erik, Shaw, Bret R, Seydel, Erwin
R, & van de Laar, Mart AFJ. (2008). Empowering processes and outcomes of
participation in online support groups for patients with breast cancer, arthritis, or
fibromyalgia. Qualitative Health Research, 18(3), 405-417.
VanderWeele, Tyler, & Vansteelandt, Stijn. (2009). Conceptual issues concerning mediation,
interventions and composition. Statistics and its Interface, 2, 457-468.
Walther, Joseph B. (1996). Computer-mediated communication impersonal, interpersonal, and
hyperpersonal interaction. Communication research, 23(1), 3-43.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 119
Walther, Joseph B, & Boyd, Shawn. (2002). Attraction to computer-mediated social support.
Communication technology and society: Audience adoption and uses, 153188.
Walther, Joseph B, & Bunz, Ulla. (2005). The rules of virtual groups: Trust, liking, and
performance in computer‐ mediated communication. Journal of Communication, 55(4),
828-846.
Wasserman, Stanley, & Faust, K. Social Network Analysis. 1994. Cambridge University,
Cambridge.
Webb, Thomas L, Joseph, Judith, Yardley, Lucy, & Michie, Susan. (2010). Using the internet to
promote health behavior change: a systematic review and meta-analysis of the impact of
theoretical basis, use of behavior change techniques, and mode of delivery on efficacy.
Journal of medical Internet research, 12(1).
Wellman, Barry, & Berkowitz, Stephen D. (1988). Social structures: A network approach (Vol.
2): CUP Archive.
Williams, Dmitri. (2006). On and off the’Net: Scales for social capital in an online era. Journal
of Computer‐ Mediated Communication, 11(2), 593-628.
Winzelberg, Andrew J, Classen, Catherine, Alpers, Georg W, Roberts, Heidi, Koopman, Cheryl,
Adams, Robert E, . . . Taylor, C Barr. (2003). Evaluation of an internet support group for
women with primary breast cancer. Cancer, 97(5), 1164-1173.
Wright, Kevin. (2000). Perceptions of on‐ line support providers: An examination of perceived
homophily, source credibility, communication and social support within on‐ line support
groups. Communication Quarterly, 48(1), 44-59.

A SOCIAL SUPPORT INTERVENTION ON TWITTER 120
Wright, Kevin B. (2005). Researching Internet‐ based populations: Advantages and disadvantages
of online survey research, online questionnaire authoring software packages, and web
survey services. Journal of Computer‐ Mediated Communication, 10(3), 00-00.
Wright, Kevin B, & Bell, Sally B. (2003). Health-related support groups on the Internet: Linking
empirical findings to social support and computer-mediated communication theory.
Journal of Health Psychology, 8(1), 39-54.
Wright, Kevin B, Rains, Steve, & Banas, John. (2010). Weak‐ Tie Support Network Preference
and Perceived Life Stress Among Participants in Health‐ Related, Computer‐ Mediated
Support Groups. Journal of Computer‐ Mediated Communication, 15(4), 606-624.
Wuthnow, Robert. (2002). Religious involvement and status‐ bridging social capital. Journal for
the Scientific Study of Religion, 41(4), 669-684.
Yalom, ID. (1970). The therapy and practice of group psychotherapy: New York: Basic Books.
Yee, Nick, Bailenson, Jeremy N, Urbanek, Mark, Chang, Francis, & Merget, Dan. (2007). The
unbearable likeness of being digital: The persistence of nonverbal social norms in online
virtual environments. CyberPsychology & Behavior, 10(1), 115-121.
Yun, Doshik, & Silk, Kami J. (2011). Social norms, self-identity, and attention to social
comparison information in the context of exercise and healthy diet behavior. Health
communication, 26(3), 275-285.
Zhou, Tao. (2011). Understanding online community user participation: a social influence
perspective. Internet Research, 21(1), 67-81.
Zimet, Gregory D, Dahlem, Nancy W, Zimet, Sara G, & Farley, Gordon K. (1988). The
multidimensional scale of perceived social support. Journal of personality assessment,
52(1), 30-41.

Abstract (if available)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

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

PDF

Benefits of social networking in online social support groups

PDF

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

PDF

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

PDF

Social support in Cambodia: the role of peer educators in behavior change

PDF

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

PDF

The social groups approach to quitting smoking: An examination of smoking cessation online and offline through the influence of social norms, social identification, social capital and social support

PDF

Building personal wellness communities: meaningful play in the everyday life of a network society

PDF

The role of social control in promoting healthy eating behavior among Chinese immigrants: an ecological approach

PDF

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

PDF

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

PDF

The social meaning of sharing and geocoding: features and social processes in online communities

PDF

The influence of a humorous intervention on parent-child sexual communication

PDF

Using mobile health to improve social support for low-income Latino patients with diabetes: a randomized mixed methods feasibility trial of TExT-MED FANS

PDF

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

PDF

Communication and social support of parents of children treated at Childrens Hospital Los Angeles

PDF

Transfusional iron, anthracyclines and cardiac outcomes in childhood cancer survivors

PDF

Effectiveness of a heart plus brain health-focused nutrition intervention among Latinas: a randomized controlled trial of Buenos Hábitos Alimenticios Para Una Buena Salud (Good Eating Habits for ...

PDF

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

PDF

Immigrant health in context: a communication ecology approach to understanding health behavior

Asset Metadata

Creator Nam, Yujung (author)

Core Title 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

School Annenberg School for Communication

Degree Doctor of Philosophy

Degree Program Communication

Publication Date 08/27/2016

Defense Date 04/17/2014

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

Tag health,mediators,OAI-PMH Harvest,social media,social support

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor McLaughlin, Margaret L. (committee chair), Cody, Michael J. (committee member), Jordan-Marsh, Maryalice (committee member)

Creator Email namyujung@gmail.com,yujungna@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-466074

Unique identifier UC11287075

Identifier etd-NamYujung-2855.pdf (filename),usctheses-c3-466074 (legacy record id)

Legacy Identifier etd-NamYujung-2855.pdf

Dmrecord 466074

Document Type Dissertation

Format application/pdf (imt)

Rights Nam, Yujung

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

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