University of Southern California Dissertations and Theses

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

(USC Thesis Other)

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

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

MOBILE SELF-TRACKING FOR HEALTH:
VALIDATING PREDICTORS, EFFECTS, MEDIATOR, MODERATOR,
AND SOCIAL INFLUENCE

by

Chihwei (Selene) Hu

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

August 9 2016

Copyright 2016 Chihwei (Selene) Hu
i
TABLE OF CONTENTS
ACKNOWLEDGEMENTS..............................................................................................v
ABSTRACT…...................................................................................................................vi
INTRODUCTION.............................................................................................................1
CHAPTER ONE: THEORETICAL PERSPECTIVES ON MOBILE SELF-
TRACKING……………………………………………………………………………...5
Social Cognitive Theory…………………………………………………………..8
Theory of Planned Behavior………………………………………………………9
Self-Determination Theory………………………………………………………10
Social Action Theory…………………………………………………………….13
CHAPTER TWO: CONCEPTUALIZATION OF MOBILE SELF-TRACKING
AND ITS UNDERLYING PREDICTORS....................................................................15
Conceptualization of Self-tracking for Health…………………………………...15
Antecedents & Determinants of Mobile Self-tracking…………………………..21
Other Underlying Constructs…………………………………………………….31
CHAPTER THREE: OVERVIEW OF MOBILE HEALTH PROGRAMS……….35
Evidence-Based Strategies for Mobile Health Programs………………………..40
Characteristics in Relation to Effectiveness……………………………………..43
mHealth & Self-tracking on Weight Loss, Diet & Physical Activity……………44
Empirical Studies with Mobile Self-tracking……………………………………47
Research Questions & Hypotheses………………………………………………50
The Research Model……………………………………………………………..56
CHAPTER FOUR: RESEARCH DESIGN AND METHODS……………………....58
Study Design……………………………………………………………………..58
The Questionnaire………………………………………………………..58
Measures…………………………………………………………………………60
Demographics and Antecedents………………………………………….60
Predictors of Autonomous Motivation…………………………………...60
Mediator..….……………………………………………………………..65
Moderator.………………………………………………………………..65
Psychological Outcomes…………………………………………………66
Behavioral Outcomes…………………………………………………….67
Egocentric Network Measures…………………………………………...69
Analytic Strategies……………………………………………………………….70
CHAPTER FIVE: RESULTS………………………………………………………….72
Recruitment ……………………………………………………………………...72
Demographics …………………………………………………………………...72

ii
Main Findings …………………………………………………………………...78
Differences in Demographics……………………………………………78
Differences in Antecedents and Predictors………………………………79
Prediction of Motivation and Engagement………………………………82
Health Outcomes of Mobile Self-tracking….....…………………………84
Mediating Effects of Self-Determination…....………………………….103
Moderating Effects of EMI……………………………………………..107
Personal Network Influence…………………………………………….113
Modeling Mobile Self-tracking for Weight Management..…….………119
Mediating Effects: SEM Approach……………………………………..132
Multi-Group Analysis ………………………………………………….133
CHAPTER SIX: DISCUSSION AND CONCLUSION…………………………..…138
CHAPTER SEVEN: CONTRIBUTIONS AND LIMITATIONS………………….147
REFERENCES……………..……………………………………………………….....152
APPENDIX A: QUESTIONNAIRE………………………….....................................168

iii
LIST OF TABLES
Table 1. Means and Cronbach’s Alpha in Key Measures……………………………….69
Table 2. Demographic Information Across Groups………………………………………….73
Table 3. Means for Antecedents and Predictors (For All Participants)…………………..81
Table 4. Mean Differences in Antecedents & Predictors Between Groups with One-way
ANOVA…………………………………………………………………………………………….81
Table 5. Pearson Correlation Tests Between Covariates and Health Outcomes………..87
Table 6. Test for Mean Differences in Covariates Between Groups………………………88
Table 7. Homogeneity of Regression Slopes for Health Status and Group Membership..89
Table 8. Homogeneity of regression slopes for age and group membership……………..90
Table 9. Homogeneity of regression slopes for chronic disease and group
membership………………………………………………………………………………………..90
Table 10. Homogeneity of regression slopes for income and group membership……….90
Table 11. Homogeneity of regression slopes for education and group membership……91
Table 12. Multivariate Outcome Prior to Covariates Adjustment…………………………92
Table 13. Univariate Outcomes Prior to Covariates Adjustment………………………….92
Table 14. Main Effects on Health Outcomes Prior to Covariates Adjustment…………..93
Table 15. Levene’s Test of Equality of Error Variances in Health Outcomes Across
Groups ……………………………………………………………………………………………..94
Table 16. Results of Multivariate Analyses after Covariates Adjustment………….……...96
Table 17. Univariate Analyses After Applying Covariates: Main Effects on Health
Outcomes…………………………………………………………………………………………..97
Table 18. Post Hoc Analyses with Bonferroni Method………………………………….99
Table 19. Estimated Marginal Means Adjusted for the Effect of the Covariates………..100
Table 20. Comparison Between Mobile Self-trackers with Wearables and Mobile
phones…………………………………………………………………………………..103
Table 21. Estimated Marginal Means for Health Outcomes Between Groups……….....109
Table 22. Tests of Moderating Effect of EMI on Health Outcomes……………………….113
Table 23. Predictors of Same Tracking Behavior using Logistic Regression Analyses
Among all Tracker pairs (N=2347)…………………………………………………….119
Table 24. Modification Indices with Univariate Increment for Model Revision…….....123
Table 25. Results of the Confirmatory Model Specification……………………………….126
Table 26. Insignificant Paths to be Deleted in Modification 2………………….......….126
Table 27. Summary of Model Revisions for the Full Sample………………………………127
Table 28. Model Fit Indices for the Final Model…………………………………………....127
Table 29. Composite Reliability (CR) and Average Variance Extract (AVE) for the Final
Model………………………………………………………………………………………..…….128
Table 30. Square Root of Average Variance Extracted and Correlation Between Latent
Variables……………………………………………………………………………… 129
Table 31. Results of the Common Method Variance Test…………………………………..130
Table 32. Standardized Total Effects, Direct Effects, and Indirect Effects…………...…131
Table 33. Test of Mediation Effects using Bootstrapping Confidence Intervals with
AMOS……………………………………………………………………………………………..132
Table 34. Baseline Comparison: Assuming Model Unconstrained to be Correct……...134
Table 35. Pairwise Comparison Between Groups with Chi-square Difference Tests….136

iv
LIST OF FIGURES
Figure 1. Conceptualization of self-tracking for health and its underlying components..19
Figure 2. Multifaceted taxonomy of mobile self-tracking practices for health………….20
Figure 3. Underlying predictors associated with mobile self-tracking…………………..31
Figure 4. An integrated research model for mobile self-tracking on weight
management……………………………………………………………………………...56
Figure 5. Reasons for not self-tracking among non-trackers
(n=129)…………………………………………………………………………………...74
Figure 6. Alternatives to weight management among non-trackers (n=129)……………75
Figure 7. Motivational reasons among all self-trackers (mobile & traditional)…………76
Figure 8. Tracking themes among all self-trackers (mobile & traditional)……………...77
Figure 9. Model for mobile self-tracking
engagement………………………………………………………………………….…...82
Figure 10. Results of regression analyses for mobile self-tracking engagement………..84
Figure 11. Model for group comparison in health outcomes……………………………84
Figure 12. Research model of self-determination as a mediator……………………….104
Figure 13. The mediating effects of self-determination on health outcomes (a)……….106
Figure 14. The mediating effects of self-determination on health outcomes (b)………107
Figure 15. Research model of EMI as moderator of mobile self-tracking engagement on
health outcomes………………………………………………………………………...108
Figure 16. Research model of personal network influence on mobile self-tracking
adoption………………………………………………………………………………....113
Figure 17. Base model for mobile self-tracking for weight management……………...120
Figure 18. Results: Base model for mobile self-tracking with standardized estimates...122
Figure 19. Model modification for mobile self-tracking with standardized estimates…124
Figure 20. Confirmatory model specification with AMOS…………………………….125
Figure 21. Path-by-path comparison between groups using Chi-square difference
tests……………………………………………………………………………………..135

v
ACKNOWLEDGEMENTS
I feel lucky to have the pleasure of taking my academic journey at USC
Annenberg School for Communication with all the awesome faculty members, staff, and
colleagues here. I first came to Annenberg in 2008 for my masters in public relations with
the hope to pursue a Ph.D. here. Few years later, I am glad that I made it. As a mom and a
doctoral student, the past years were not easy for me. I know I wouldn’t have been able to
achieve this without the support from many people. I’d hence like to take this moment to
express my gratitude to them.
First of all, my sincere thanks go to my advisor, Dr. Margaret McLaughlin. I
wanted to thank her for being such a tremendous mentor, who has introduced me to the
amazing world of health communication and showed me the qualities to be a great
researcher. I also wanted to thank her for being so supportive, encouraging, and
understanding, especially at times when I lost faith in myself. It is her encouragement that
allowed me to grow, to find myself, and to explore my potential along the journey. Thank
you so much!
My sincere thanks also go to my other two wonderful dissertation committee
members, Dr. Michael Cody and Dr. Chih-Ping Chou. I wanted to thank them for serving
on my committee and for their invaluable advice on my research and career pursuit. I
appreciate their knowledge and guidance that have inspired me and cultivated my
expertise in social science. Their brilliant comments and suggestions made my
dissertation writing an enjoyable learning process. I also wanted to thank Dr. Lynn Miller
and Dr. Sheila Murphy who were on the committee for my qualifying exams. I truly
appreciate the time and efforts they took in guiding me throughout my doctoral studies.
Because of them, I was able to produce several quality papers and complete my
dissertation efficiently. Thank you!
Most importantly, I owe my deepest gratitude to my dear family. My sincere
appreciation goes to my husband, Jay, who has been a very supportive husband and
caring dad. Also, I owe my life to my dad, my mom, and my parents-in-law for their
immeasurable love, care, and constant support. Thank you for always being there for me,
and telling me not to give up. I wouldn’t have been able to make it without you. I also
wanted to thank my little sweetheart S who has always been my delight. Thank you for
being nice during the whole process. Mommy is done with her graduate studies! Yay!
My special thanks go out to a wonderful colleague and mentor, Nancy Chen, for
her kindness and generosity to always offer help when I was in need of it; to Dr. Tom
Goodnight, Dr. Larry Gross, Dr. Sarah Banet-Weiser, and Dr. Peter Monge for their
generosity and support in the past years; and to all my dear friends, who have been
supporting me throughout this process. Academic journey can be lonely, but mine is not
because of your company. Lastly, I wanted to thank Annenberg and USC for providing
such a resourceful academic environment for all of us! I am so proud to be a Trojan!

vi
ABSTRACT
This project seeks to explore self-guided mobile self-tracking for weight
management through an integrated theoretical lens. It looks at the correlational and causal
relationships between a set of psycho-social variables, as well as the effects of mobile
self-tracking by integrating classic communication and social psychology theories (i.e.,
social cognitive theory, theory of planned behavior) and emerging health behavior
theories (i.e., self-determination theory, social action theory) into a new model of mobile
self-tracking. Findings have shown that, among several underlying predictors, vigilance
coping, self-efficacy, and normative beliefs significantly predicted an individual’s
autonomous motivation for mobile self-tracking. More importantly, mobile self-trackers
were found to have better health outcomes than non-mobile self-trackers after controlling
for confounding factors. In addition, confirming theoretical perspectives, self-
determination was shown to be an important component that not only can be reinforced
through self-guided mobile self-tracking, but also can mediate the effects of mobile self-
tracking on health outcomes. Ecological momentary intervention was validated as a
significant moderator of the effects of mobile self-tracking engagement on healthy eating.
In addition, results revealed that social influence takes forms of network exposure, tie
strength, relation type, and role modeling in affecting individuals’ mobile self-tracking
adoption. Lastly, structural equation modeling was performed to validate the research
model. Contributions to the field of mobile health and implications for future health
interventions and health research are discussed.

1
INTRODUCTION
This work proposes and empirically tests a model of mobile self-tracking for
health in the domain of weight management. Mobile self-tracking has emerged as a sub-
segment of mobile health, stemming from the rapid rise of smartphone and wearable
applications. According to the National Institutes of Health, mobile Health (mHealth),
referring to “the use of mobile and wireless devices to improve health outcomes,
healthcare service, and health research” (“Definition of mHealth”, 2012), is growing
rapidly. There is public and professional interest in understanding mobile health in terms
of best use and effectiveness (Shaw & Bosworth, 2012; Rai, Chen, Pye, & Baird, 2013).
Among various endeavors, mobile self-tracking is gaining popularity not only in the field
of medicine but also in the field of communication research (Sweet, 2013). Reported in a
study on mobile health app publishing, the number of mHealth apps published on iOS
and Android has reached more than 100,000 in 2014 (Miller, 2014), and the top 10
mobile health apps generated up to 4 million free downloads and 300,000 paid downloads
per day in 2013 (Tode, 2013).
What these numbers tell us is that mobile health appears diffused in the
population, but whether mobile health applications are really beneficial remains
unknown. These numbers could be the result of normalization of self-tracking practices
in the media followed by a bandwagon effect among the public. As mobile self-tracking
has not received much examination by health researchers, there are only a few attempts
made either to characterize self-tracking practices or to examine the demographics and
motivations of self-trackers (Li, Dey, & Forlizzi, 2011; Butterfield, 2012; Crisostomo,
2013; Nißen, 2013). None of these studies has systematically reviewed relevant literature

2
on mobile health programs and provided a solid theoretical foundation for explaining the
phenomenon. Therefore, there seems to be a social imperative to understand the
effectiveness and potential effects associated with mobile self-tracking before we fully
embrace the opportunities it brings to us.
The current study, although exploratory in nature, serves as the first scholarly
endeavor to examine mobile health programs systematically (i.e., clinical trials and
interventions), provide theoretical foundations that are well-suited to the context of
mobile self-tracking, and conceptualize and re-define the behavior in question by
differentiating various kinds of self-tracking practices. It combines theoretical and
empirical interests, drawing on work in communication, social psychology, and public
health. It seeks to examine the correlational and causal relationships between a set of
psychometric measures as well as the effects of mobile self-tracking, by integrating
classic health behavior theories (i.e., social cognitive theory; theory of planned behavior)
and emerging theories in health promotion (i.e., self-determination theory, social action
theory) into a new model of mobile self-tracking. Lastly, the study takes an egocentric
network perspective to explore how social influence takes form in affecting individuals’
mobile self-tracking behavior adoption.
Chapter One provides an overview of theoretical perspectives on mobile self-
tracking for health from the disciplines of communication, social psychology, and public
health. Key concepts and premises are described. This chapter articulates how theoretical
perspectives are applied to the context of mobile self-tracking. This introductory chapter
sets up the context for more detailed literature reviews in subsequent chapters.

3
Chapter Two provides a systematic review of existing studies on mobile self-
tracking with attempts to characterize and conceptualizes self-tracking practices. This
chapter defines mobile self-tracking and differentiates it from other types of self-tracking
practices. It also articulates mobile self-tracking with underlying constructs that manifest
in behavior. It establishes the base for more specific research questions and hypotheses in
the subsequent chapter.
Chapter Three reviews the applications and effectiveness of contemporary mobile
health programs, including health interventions and experimental studies. It introduces to
readers how mobile technologies are currently incorporated into health promotion and
research, and summarizes evidence-based strategies and effective components found in
previous mobile health programs. This chapter addresses the role of ecological
momentary intervention (EMI) in health management⎯ a feature embedded in most
mobile applications, and provides rationales for examining underlying mediatosr and
moderators of mobile self-tracking on various health outcomes. The literature review
brings up the research model of mobile self-tracking as well as the research questions and
hypotheses examined in the present study.
Chapter Four outlines the research design and describes survey instruments, study
procedures, key measures in the research model, and analytic strategies used for data
analyses. Statistical methods used include chi-square test of independence, ANOVA,
multiple regression analysis, logistic regression analysis, MANOVA, MANCOVA, and
structural equation modeling.
Chapter Five reports findings from the statistical analyses including the
demographic characteristics of the participants and the main results. Mobile self-tracking

4
was found to be effective and conducive to various health outcomes compared to
traditional and non-tracking approaches to weight management. Self-determination was
found to be an important mediator of mobile self-tracking engagement on several health
outcomes. EMI moderated the effect of mobile self-tracking engagement on healthy
eating. This chapter also presents interesting findings on social influence.
Chapter Six concludes with the findings and discusses practical implications in
the field of mobile health. Chapter Seven addresses the study’s contributions, limitations,
and directions for future research.

5
CHAPTER ONE: THEORETICAL PERSPECTIVES ON MOBILE
SELF-TRACKING
According to Oxford Dictionaries, self-tracking for health refers to “the practice
of systematically recording information about one’s diet, health, or activities, typically by
means of a smartphone, so as to discover behavioral patterns that may be adjusted to help
improve one’s physical or mental well-being” (Oxford Dictionaries’ online dictionary,
n.d.). As specified, mobile self-tracking refers to the activities that “individuals [are]
using technologies to monitor, record, and evaluate various aspects of their bodily
functions and everyday habits, such as body weight, sleep pattern, exercises, and diet, etc.
(Guthridge, 2013), which is also known as “life-logging” (Guthridge, 2013). It is
assumed that people can be in a better position to manage their health with improvement
in self-knowledge by tracking their own health data (Guthridge, 2013). Mobile self-
tracking is believed to be able to encourage desired behaviors, increase awareness of
one’s behavior, and decrease undesired habits (Klasnja & Pratt, 2012).
The trend of self-tracking for health was facilitated by the “participatory features
of social media,” which encourages people to share their health data with others (Sweet,
2013). As low-cost and automated data collection tools have become available in the
market, self-monitoring of health has been increasing among the public in recent years. It
gradually becomes part of participatory health initiatives, and contributes to the “public
health ecosystem” where individuals use health social networks, self-experiment studies,
or smartphone applications to achieve a health goal and share self-data and knowledge
with online communities (Swan, 2012, p. 2). For example, a patient-organized
crowdsourcing platform, the Quantified Self, is a typical online self-tracking community

6
where enthusiasts are dedicating to collective knowledge on self-tracking and personal
wellness.
Self-tracking practices also underline a growing interest of individuals to exert
control over their own body and over their data with the hope of finding meaning in their
data (Sweet, 2013). As shown in a recent Pew survey on 3,014 adults in the Unites States
(Fox & Duggan, 2013), 69% of U.S. adults track a health indicator for themselves or for
others. The majority of them (49%) reported an informal way of keeping track of
progress, which is “in their heads,” 34% of people used non-technological methods (e.g.,
journals or notebooks), and 21% of them used some forms of technology (e.g., mobile
applications or devices). Regarding the impact of tracking, participants reported that
tracking behavior in turn has “changed their overall approach to maintaining health”
(46%), seeking further information from doctors (40%), and influenced a health decision
about a treatment or condition (34%)(Fox & Duggan, 2013). As to the theme of tracking,
the majority reported tracking their weight, diet, and exercise routine (60%), and 33%
tracked health indicators or symptoms (Fox & Duggan, 2013). In addition, Sweet (2013)
indicated that self-trackers reported that they feel more in control of their lives by
engaging in self-tracking practices such as losing weight, improving physical fitness and
sleeping patterns, or identifying the sources of stress in their lives.
A few types of people may benefit more from performing self-tracking, such as
patients diagnosed with medical problems for which conventional treatment offers
limited benefits, people who want to change behavior or to identify environmental,
dietary, contextual or social contributors to their symptoms, and those who want to be
more involved in their own health care (Norris, 2012). Norris (2012) also indicates that

7
self-tracking serves as “a behavioral modification driver” that helps change health
behavior patterns. Nonetheless, it has been argued that intense self-tracking may cause
“cyberchondria” and health anxiety (Pogorelc, 2013), leading to excessive anxiety about
one’s health state and the data collected (Lupton, 2013a). Mobile self-tracking is not
currently well regulated, and the quality, efficacy, and potential harm of health-related
mobile applications have also been questioned despite normalization of self-tracking
practices by social media (Pogorelc, 2013).
Above all, there is limited understanding toward this new trend, and the current
study is designed to explore the antecedents, determinants, and effects of mobile self-
tracking for health. In order to more accurately reflect monitoring aspects of self-
tracking, the scope examined was narrowed to mobile weight management because
weight, diet, and physical activities are found to be the most popular topics for self-
tracking over mobile devices (Fox & Duggan, 2013; Nißen, 2013).
The current study thus aims to explore five major areas through an integrated
theoretical lens discussed in the following section. The five major areas include: (1) What
are some particular characteristics associated with mobile self-trackers as opposed to
non-trackers? (2) What are the antecedents and predictors that motivate people to engage
in mobile self-tracking activities? (3) What are some positive and negative effects of
mobile self-tracking in terms of psychological and behavioral consequences? (4) How
does mobile self-tracking influence individuals’ self-determination, and how does self-
determination mediate individual health outcomes? And (5) how does social influence
come into play in this picture?

8
Social Cognitive Theory
Social cognitive theory (Bandura, 1986, 1989, 1991a, 1991b, 1992, 2002, 2004)
posits that health behavior is the product of a dynamic interplay of personal, behavioral,
and environmental influences with a focus on individuals’ ability to adapt to situations
and achieve goals. It not only concerns predicting health behaviors, but also provides
principles of promoting health behavior by specifying important self-regulatory
techniques (McAlister, Perry, & Parcel, 2008; Simons-Morton, McLeroy, & Wendal,
2012). The theory explains the dynamic interrelationships of self-regulatory processes
with crucial determinants of behavior, including self-efficacy, outcome expectations,
goals, social reactions to the behavior, self-evaluative reactions, and perceived facilitators
and obstacles, etc. (Bandura, 1989; 1992; 2004). It highlights the role of individualized
goals in influencing one’s responses to stimuli. Because individuals are self-directed
according to their goals, people manage their behavior and environment accordingly
through the self-regulatory processes involving goal-setting, self-monitoring, self-
instruction, self-reward, feedback, and enlistment of social support (Bandura, 1991a;
1991b). These processes are congruent with the core concept of mobile self-tracking for
weight management. Social cognitive theory thus serves as a meta-theory, explaining the
nature of self-tracking as a way to monitor and adjust one’s behavior in order to achieve a
desired goal (Norris, 2012). It is suitable in modeling mobile self-tracking for health for
several reasons. Firstly, people are becoming more serious about self-monitoring on what
they do, what they eat, and how they feel, wherein self-regulation is manifested. Self-
monitoring, as the core of self-tracking, entails systematic observation and recording of
one’s target behavior (Baker & Kirschenbaum, 1993). The use of mobile devices for

9
health, such as ubiquitous behavior assessment and tailoring information, manifests the
idea of self-regulation (Norris, 2012). In addition, mobile technologies involving constant
information seeking enhance individuals’ capabilities for managing their own health
through various functions in the form of endowing self-efficacy addressed in the theory
(Klasnja & Pratt, 2012; Riley et al., 2011). As Bandura (1986, 2002, 2004) indicates,
social cognitive means, such as self-management skills, sense of efficacy, and social
supports, can be very effective in health prevention and promotion. Situating the theory
in the explanations of goal-oriented self-regulation in the context of mobile self-tracking
is particularly suitable.
Theory of Planned Behavior
Based on expectancy-value theory and derived from the theory of reasoned action
(Fishbein & Ajzen, 1975), the theory of planned behavior (TPB) posits that behavioral
intention is influenced by attitudes, subjective norms, and perceived behavioral control
that reflect one’s various outcome expectations (Ajzen, 2002). Among various constructs,
intention is the most important determinant of behavior. With very specific constructs
and specified elements pertaining to the behavior in question (i.e., the action, the target,
the context, and time), the applications of TPB was found to yield larger effect sizes than
did studies using other theories (Webb et al., 2010). The theory lays out a well-developed
causal chain that links a set of beliefs to explain the likelihood of performing a health
behavior. It serves as a theoretical framework to examine mobile self-tracking behavior,
and it was chosen for being suitable for explaining deliberate behavior by “identifying,
measuring, and combining beliefs relevant to individuals or groups,” and allowing us to

10
“understand reasons that motivate the behavior of interest” (Montano & Kasprzyk, 2008,
p.76; Noar & Zimmerman, 2005).
Three essential predictors were highlighted in the theory of planned behavior:
attitudes, social norms, and self-efficacy. Attitude is defined as a person’s overall
favorableness or unfavorableness toward performing the behavior (Montano & Kasprzyk,
2008). It is the affective and instrumental evaluation of a target behavior, reflecting one’s
outcome expectancies toward performing a behavior (Bandura, 2004). Previous research
showed that individuals’ willingness to try out technology significantly predicted their
intention to adopt the technology (Rai et al., 2013). Social norms, defined as the social
pressure one feels to perform a behavior or not, includes a person’s normative beliefs
about what others are doing in one’s social network and his or her motivation to comply
with them (Montano & Kasprzyk, 2008). Social norms are indicators of social influence
in one’s social network. Self-efficacy is defined as one’s degree of confidence in the
ability to perform the behavior in the face of various obstacles, which is also addressed in
social cognitive theory (Bandura, 2004). By adopting the theoretical perspective, the
study seeks to answer the questions as to what beliefs/attitudes are most salient to
motivate mobile self-tracking, and what constructs are most important to target for
mobile self-tracking in health promotion. The theoretical framework provides guidance to
examine antecedents and determinants of motivation and behavior engagement that are
underlined in the context of mobile self-tracking.
Self-Determination Theory
Self-determination theory (SDT) provides a theoretical framework to examine
how mobile self-tracking affects health outcomes. The theory highlights the role of

11
autonomous motivation as a key to sustain health-related behaviors (Ryan, Patrick, Deci,
& Williams, 2008). It concerns satisfying basic psychological needs that are conducive to
health through the support of autonomy, competence, and relatedness (Ryan, Patrick,
Deci, & Williams, 2008). It assumes that one’s intrinsic motivation to perform a health
behavior can be maintained if his or her psychological needs are met, which will result in
behavior perseverance that is conducive to wellness and health (Broeck, Vansteenkiste,
De Witte, Soenens, & Lens, 2010; Chen, Van Assche, Vansteenkiste, Soenens, & Beyers,
2015; Chen et al, 2015; Fortier, Williams, Sweet, & Patrick, 2009).
To be more specific, autonomous motivation can be enhanced when a person is
feeling in control (i.e., autonomy), capable of doing something well (i.e., competency),
and feeling connected and being understood by others (i.e., relatedness). When these
needs are satisfied, a person is believed to be happier, functional, and motivated to
perform healthy behavior that leads to well-being (Simon-Morton, McLeroy, & Wendal,
2009).
Autonomy, referring to one’s sense of control on action taking in life, can be
enhanced when relevant information and rationales for action taking are provided without
pressures that detract individuals from a sense of choice. In other words, granting a sense
of control and options to a person can influence his or her way of pursuing goals in a
guided direction (Ryan, Patrick, Deci, & Williams, 2008). By doing so, a person will
come to value the behavior in question and align it with other lifestyle patterns if
rationales are internalized into his or her central values (Fortier, Williams, Sweet, &
Patrick, 2009). Competence, referring to one’s experiencing mastery of controlling
behaviors and outcomes, is developed through granting relevant skills, tools, and

12
feedbacks for change, etc. (Ryan, Patrick, Deci, & Williams, 2008). Relatedness,
referring to one’s connectedness with others in their social environment, is a similar
concept to perceived social support in facilitating health behavior. In the process of
gaining relatedness, a sense of being understood and cared for is essential to form the
connections and trust that allow internalization to occur (Ryan, Patrick, Deci, &
Williams, 2008). Fulfilling these three psychological needs is believed to enhance the
likelihood of enjoyment, persistence, and maintenance of a prescribed health behavior
(Fortier, Williams, Sweet, & Patrick, 2009).
Previous health behavior research based on Self-determination theory has shown
consistent findings, suggesting that when a person is experiencing autonomy,
competence, and relatedness, he or she is experiencing more “volitional engagement” in
the behavior and thus will “maintain outcomes better over time” (Ryan, Patrick, Deci, &
Williams, 2008, p.2). This pattern is believed to “hold for broad lifestyle changes” (Ryan,
Patrick, Deci, & Williams, 2008, p.2). To bring it into the mobile self-tracking context, it
is conceived that when a self-tracker is acquiring autonomy, competence, and relatedness
through performing self-tracking, he or she is experiencing more volitional engagement
in monitoring own health, which is conducive to health outcomes. The theory is thus
particularly suitable for studying self-tracking behavior because mobile self-tracking for
weight management is mostly self-initiated and self-guided without health professionals’
oversight as opposed to traditional weight management counseling in clinics. Plus, self-
monitoring and keeping records on weight-related indicators, involving perseverance of
physical activity, diet, calorie-deficit eating patterns, may not be inherently enjoyable
activities (Paddock, 2013). For many people, it’s difficult to stay motivated and to

13
maintain the behavior. Therefore, self-determination theory, which addresses sustaining
autonomous motivation by satisfying psychological needs, is especially relevant in the
context of mobile self-tracking for weight management.
Social Action Theory
Along the same lines, social action theory (SAT) addresses self-monitoring and
social interdependence in promoting health behavior. Ewart (1991, 2009) suggests that
action state analysis is effective in replacing unhealthy behavior with healthy ones. It
entails identifying the sequence of daily routines and environmental cues that trigger
unhealthy behaviors, as well as cues that encourage healthy behaviors. Action state
analysis also requires a person to identify and build health behaviors that serve similar
functions, and to identify social interaction in relation to social interdependence which
helps remove unhealthy habits that are rooted in one’s social relationships (Ewart, 2009).
These self-change processes can be activated through several motivational mechanisms,
including constructive thinking, problem-solving, generative capability, and social
interaction, etc. (Ewart, 1991, 2009). To be more specific, there are three effective
approaches to modify habits and facilitate prescribed health behavior, including (1)
identifying barriers under different contexts for the health habits (2) engaging in self-
monitoring that involves constructive thinking and problem-solving and (3) building
social and community-level self-regulatory resources that can bolster behavior change
(Ewart, 1991, 2009). These steps are congruent with the idea of mobile self-tracking for
health, which aims to gain knowledge on own health condition, identify patterns of
factors influencing health, and interact and exchange information with social ties and
online communities. In addition, the theory suggests the use of social power in

14
facilitating health behavior. For example, having patients in recovery who are supposed
to be more experienced in certain behavior to mentor novices who seek help to change
unhealthy behavior (Ewart, 2009). It is believed that, through social power, novices with
respect to a behavior can benefit from modeling effective coping, and mentors can benefit
from enhanced self-esteem as an exemplar, and both of them will sustain their behavior
changes, and this will also prevent them from having relapses (Ewart, 2009). This is
relevant to mobile self-tracking because connectivity and virtual communities allow
people to leverage social power in the forms of social supports and social capital to
achieve their goals. In addition, social action theory is compatible with self-determination
theory in the context of mobile self-tracking as both of them address the importance of
goals, skills, and motivation. The key components identified in social action theory are
overlapping with those in self-determination theory, including identifying barriers under
different contexts, constructive thinking, planning for strategies, and problem-solving.
Both of them provide solid theoretical foundations to explain the nature of mobile self-
tracking practices and set up an integrated framework to examine its associated effects.
To sum up, mobile self-tracking is theoretically manifested as a self-regulatory act
in order to pursue one’s health goal. Nonetheless, it has been criticized for causing health
anxiety, making people unduly attend to their self-data and become “narcissistic”
(Lupton, 2013a; 2013b). To fully understand the phenomenon, it is worth exploring the
predictors, effects, mediators, moderators, as well as social influence on mobile self-
tracking for health through an integrated theoretical lens.

15
CHAPTER TWO: CONCEPTUALIZATION OF MOBILE SELF-TRACKING
AND ITS UNDERLYING PREDICTORS
Conceptualization of Self-tracking for Health
A few recent studies provide descriptive analyses regarding self-tracking practices
in general (Crisostomo, 2013; Nißen, 2013). Crisostomo (2013) argues that self-tracking
for health is a behavior that lies at the “intersection of technology, wellness, and science,”
and it is motivated by several psychological and social factors such as one’s health
condition and well-being. As a matter of fact, informal self-tracking does not necessarily
require a smartphone or wearable devices because some self-trackers simply take a casual
approach to monitoring and keeping records for their health. It is thus important to
conceptualize self-tracking and properly define what constitutes mobile self-tracking and
what does not. The following section is dedicated to the discussion of conceptualization
of mobile self-tracking.
Self-tracking can be seen as a way of pursuing personal informatics, through
which individuals “collect personally relevant information for the purpose of self-
reflection and gaining self-knowledge” (Li, Dey, & Forlizzi, p.558). Crisostomo (2013)
described self-tracking as one’s “inward gaze” of pursuing the “perfect self” (p.32), a
sense-making of the self “through a measuring lens” (p.34), and a “democratizing
surveillance” targeted at the self (p.36). Self-tracking technology is viewed as a
“persuasive agent” that shapes how the users think, act, and live (p.39). Although the
discourse of self-tracking is not the focus of the current study, these viewpoints highlight
the importance of the technological revolution in modern life on health-related
monitoring.

16
The scope of self-tracking activities is broad; nonetheless, a few studies have
provided insight into the phenomenon that helps to conceptualize it in a different way.
Butterfield (2012) conducted an ethnographic research on Quantified Self community
members, and identified three axes with respect to personal experiments among self-
trackers. The three dimensions are: degree of technological involvement, level of
complexity, and types of goal. The first axis addresses the role of technology in self-
tracking based on the extent to which the tracker is dependent on it (e.g., tracking by
memory or by wearables, etc.). The second axis puts emphasis on the design of self-
tracking practices in terms of methodology. For example, a self-tracking practice can be
simply a “life-logging” activity to record certain information, or it could be an
experiment to look at correlations between different factors one is concerned about. The
third axis addresses the specificity of the goal one has in mind for self-tracking. For
example, some trackers may have a specific goal for performing self-tracking (e.g., to
lose weight or to be fit), while others may not be clear about their own goals and just
want to record whatever they do. Butterfield’s (2012) taxonomy of self-tracking practices
provides a fundamental understanding of self-tracking in general.
Nißen (2013) surveyed self-trackers within the Quantified Self community and
looked into their motivations to perform self-tracking from a functional perspective. Two
types of self-trackers were identified based on their purposes for self-tracking: “wellbeing
only” and “wellbeing and health” (p.35). The former refers to self-trackers who were
healthy in general and were tracking to enhance wellness. The latter refers to self-trackers
who suffered from chronic illnesses and were tracking to control or improve health
condition. Five motivational reasons were identified based on the function of self-

17
tracking. These motivations include entertainment (i.e., one is motivated by the “pleasure
bringing” aspect of self-tracking); self-association (i.e., one is motivated by the
community citizenship of self-tracking), self-design (i.e., one is motivated by the desire
to optimize one’s own health), self-discipline (i.e., one’s motivated by self-reward and
self-gratification due to self-tracking), and health style (i.e., one’s motivated to seeking
alternatives aside from the traditional health system) (Nißen, 2013, p.67-68). Among
these motivations, self-design was found to be the most important reason for performing
self-tracking (Nißen, 2013). It is consistent with the view of Gary Wolf, founder of the
Quantified Self, in that optimization, curiosity, and self-knowledge are the presumed
motivational reasons for participating in self-tracking activities (Wolf, 2010).
Functionally speaking, health self-tracking helps people better explore themselves
beyond “improvising, guessing, forgetfulness, and unconscious alterations” of their status
quos (Nißen, 2013, p.16). On top of this, self-tracking with mobile technology fulfills
individuals’ needs for an “enhanced self-knowledge” and greatly increases tracking
possibility and accuracy (Nißen, 2013, p. 16).
Li, Dey, and Forlizzi (2011) proposed the Stage-based Model of Personal
Informatics System to explain individuals’ self-tracking behavior as a way of
approaching personal health informatics. The model specifies five stages of health
information retrieving and usage, including preparation, collection, integration,
reflection, and action (behavior change). Li, Dey, and Forlizzi (2011) interviewed self-
trackers about their reasons to track their own health data from a stage-based perspective.
It was found that the majority reported a primary goal of changing or maintaining a
behavior with the personal informatics tools they adopted. In addition, six types of

18
information self-trackers seek were identified, including health status, health data history,
goals, discrepancies between one’s goal and current status, contexts of the indicators
collected, and factors influencing one’s goals (Li, Dey, and Forlizzi, 2011). Although the
types of tracking information “are not always clearly distinguishable” (Crisostomo, 2013,
p.29), self-trackers can be grouped into two phases in general that are associated with
differential information preferences: the discovery phase or the maintenance phase (Li,
Dey, & Forlizzi, 2011). Discovery phase refers to the stage where the tracker’s goal is
unclear and he or she is exploring a wide range of factors influencing the self. On the
other hand, maintenance phase refers to the stage where the tracker’s goal is clear, and
the factors affecting his or her goal have been identified. The distinction between the two
phases highlights different information needs among self-trackers, and provides a
framework to consider stages which trackers are in. Exploring self-trackers from a stage-
based perspective allows us to correlate their readiness with their subsequent outcomes,
and may produce more insight into the phenomenon.
Taken together, the conceptualization of self-tracking for health is outlined below
(see Figure 1). Three components are addressed, including self-tracking, health, and
technology. The intersection is the theme of the current study, where an emphasis is
placed on health-related self-tracking with mobile technology. The focus differentiates
itself from non-health-related self-tracking (e.g., one might self-track for environmental
factors or one’s living expenses, etc.), non-tech-assisted self-tracking, which has fairly
limited tracking objects (e.g., tracking in the head, counting by memory or pen and paper,
etc.), and e-health and mHealth that do not involve the idea of tracking of the self (e.g.,
Internet education programs, mobile games, etc.).

19

Figure 1. Conceptualization of self-tracking for health and its underlying components.
In addition, Figure 2 identifies types of people in light of mobile self-tracking
practice for health: non-self-trackers, non-mobile self-trackers, and mobile self-trackers.
Non-self-trackers refer to those who don’t regularly engage in self-monitoring activities.
Non-mobile self-trackers refer to those who use informal ways to monitor their health
regularly. Without the assistance of mobile technology, it is conceivable that their
tracking practices are relatively constrained. Mobile self-trackers refer to those who “use
(mobile) technologies to monitor, record, and evaluate various aspects of their bodily
functions and everyday habits” (Guthridge, 2013). Their tracking behavior and factors

20
associated with it are the focus of the current study.
Furthermore, the literature review also suggests that self-tracking practices can be
described by their aim, phase, and function, which are not mutually exclusive in
conceptualization (Nißen, 2013). The multi-faceted taxonomy allows us to look into
multiple dimensions of self-tracking and to examine how they correlate with underlying
factors and outcomes. The conceptualization and theoretical foundations framed the
questionnaire distributed to the participants.

Figure 2. Multifaceted taxonomy of mobile self-tracking practices for health.

21
Antecedents & Determinants of Mobile Self-tracking
Several underlying factors that are relevant in the context of mobile self-tracking
were drawn from theories and previous literature and are discussed in the following
section.
Health status. Previous studies suggest that engagement in e-health or mHealth is
associated with one’s health status and disease identity. Dutta and Feng (2007) found that
the use of online health community participation was largely guided by one’s disease-
triggered motivation. Specifically, one’s “disease-specific motivation in the realm of
perceived susceptibility to a disease or being detected with a disease” triggers
individuals’ participation in disease-specific online communities (Dutta & Feng, 2007,
p.181). In a similar vein, Dutta-Bergman (2004b) found that individuals’ online health
activities such as seeking drug and medication information as well as disease-specific
information were all driven by health information orientation and health beliefs (Dutta-
Bergman, 2004b). In addition, Hu, Bell, Kravitz, and Orrange (2012) examined factors
influencing patients’ information seeking on social media. They found that patients’ use
of online resources and social media support forums were greatly associated with their
illness identity, personal control and emotional representations. To be more specific,
patients who associated themselves with a larger number of symptoms with their illness,
held negative emotions toward their illness, or felt in control of their health, were more
likely to search for information on social media. Besides, Hu et al. (2012) found that
patients who had higher illness identity were more likely to turn online for health
information, while patients who had chronic diseases were more likely to seek support on
social media (Hu et al., 2012). Thackeray, Crookston, and West (2013) found that people

22
having chronic diseases were more likely to use online rankings and social networking
sites for health-related purposes. These findings all suggested that engaging in e-health is
affected by disease-oriented tendencies. Furthermore, a recent Pew Research report
indicates that health status plays an important role in the use of mobile health (Fox &
Duggan, 2013). The act of self-tracking is closely associated with chronic conditions, and
those who recently experienced a health crisis or significant changes in their physical
health were found to be more likely to look for health information via their cellphones
(Fox & Duggan, 2013). In Nißen (2013)’s recent study, there were more healthy and
wellness-oriented self-trackers than self-trackers with chronic diseases, nonetheless, it
was found that those suffering from chronic diseases tracked more parameters and spent
significantly more time on self-tracking related activities per day than their healthy
counterpart (Nißen, 2013). Results in previous studies suggest the influence of individual
health status on engaging in e-health and mobile health activities.
Health consciousness. Researchers have defined health consciousness as “a
tendency to focus attention on one’s health” (Iversen & Kraft, 2006, p.603). It’s an
individual’s psychological, “inner-state self-attention” to health alertness, involvement,
and self-monitoring of health that can be characterized as “the degree to which someone
attends to or focuses on his or her health” (Gould, 1990, p.228). The tendency is believed
to be relevant to personal engagement in health-related issues, and thus to prompt health
behaviors in response to one’s health information needs (Hu, 2013). Generally speaking,
it refers to “a pro-health mindset that directs one’s attention to health-related issues”, and
it “oftentimes manifests through engaging in health-related behaviors” (Hu, 2013, p.23).
Previous studies have shown that health consciousness is associated with health

23
behavioral intention and health behavior adoption, and facilitates various preventive
health behaviors (Hong, 2009; Hu, 2013, 2014, 2016; Kaynak & Eksi, 2011; Walker,
Volkan, Sechrist, & Pender, 1988). It was found to predict a wide variety of health-
related behaviors, including preventive health behavior (Gould, 1988; Jayanti & Burn,
1998), appraisal and coping with health messages (Iversen & Kraft, 2006), community
participation (Dutta-Bergman, 2004a), health information seeking (Dutta-Bergman,
2005), online engagement in health campaigns and advocacies (Hu, 2014, 2016), organic
food purchase (Kriwy & Mecking, 2012), and acceptance of health messages (Hong,
2011), etc. It is thus assumed that health consciousness is associated with self-tracking on
health because a tracker recognizes the importance of health and is concerned enough to
take action to monitor and track his or her own health indicators.
Health locus of control is defined as an individuals’ perceived control over his or
her own health (B. S. Wallston & Wallston, 1978; K. A. Wallston, Wallston, & DeVellis,
1978). In the present study, it specifically refers to one’s “internal” control over health as
opposed to “chance” and “powerful others” as the other two dimensions of health locus
of control discussed by Wallston and Wallston (1978). This concept is rooted in social
cognitive theory, suggesting that individuals’ beliefs about their own controllability over
events will affect their behaviors. Similarly to health consciousness that implies a sense
of responsibility, taking personal responsibility toward health is the core of internal
health locus of control. These two constructs are distinguishable in that, health
consciousness reflects “a sense of focus,” while internal health locus of control reflects “a
sense of control” (Gould, 1990; p.235), a belief that one can “control over health
behaviors or outcomes” (Wallston, Wallston, Smith, & Dobbins, 1987, p.1). Evidence

24
has shown that internal health locus of control better predicted health behaviors such as
exercise, quitting smoking, maintaining a diet, less sexually risky behavior, and
adherence to medication (P. Norman, Bennett, Smith, & Murphy, 1998; Victor & Haruna,
2012; B. S. Wallston & Wallston, 1978). In addition, health locus of control was found to
be negatively associated with visits to doctors (Gould, 1990; Walker et al., 1988; K. A.
Wallston et al., 1978). Health locus of control helps to explain the reason why health-
related predispositions (i.e., health motivation, knowledge, health value) may or may not
translate into certain health behaviors (Gould, 1990; Walker et al., 1988; K. A. Wallston
et al., 1978). Researchers suggested that using health locus of control in conjunction with
other health-related beliefs/expectancies (e.g., health consciousness, health value, self-
efficacy, perceived barriers) may account for more of the variance in health behavior (P.
Norman et al., 1998; B. S. Wallston & Wallston, 1978). It is relevant to mobile self-
tracking for weight management because it is one of the underlying health-related
dispositions that may encourage and facilitate health behavior. In other words, people
who internally feel responsible and obligated to control their own health may be more
likely to engage in mobile self-tracking as the behavior helps manage and regulate own
health. Therefore, health locus of control is proposed to be one of the underlying
predictors of individuals’ autonomous motivation to mobile self-track.
mHealth Literacy refers to a parallel concept of e-health literacy, which is
specifically defined as one’s level of literacy on mobile health. C. D. Norman and
Skinner (2006) defined e-Health literacy as the “ability to seek, find, understand, and
appraise health information from electronic sources and apply the knowledge gained to
addressing or solving a health problem”(e3.). E-health comprises six sub-dimensions,

25
including traditional literacy, media literacy, information literacy, computer literacy,
scientific literacy, and health literacy. mHealth literacy, working in a parallel way, is an
underlying capability of comprehending and interpreting the use of mobile applications
and the data generated through mobile health. Therefore, mHealth literacy is relevant in
the context of mobile self-tracking because it should be a determinant for a person to get
motivated to engage in mobile self-tracking. Without mHealth literacy, individuals will
not be aware of mobile health programs, let along taking advantage of any mobile
services and applications.
Vigilance Coping. Based on the Transactional Model of Stress and Coping,
dispositional coping style influences one’s coping strategies and efforts, as well as the
effects of coping efforts on coping outcomes (Glanz & Schwartz, 2008). Coping is a
broad concept within which important distinctions have been made by researchers
(Skinner, Edge, Altman, & Sherwood, 2003). One of these distinctions is between
problem-focused and emotion-focused coping (Lazarus & Folkman, 1984). Problem-
focused coping is directed at the stressor itself, which is characterized as taking action to
remove the stressor or diminish the impact of the stressor. Emotion-focused coping is
characterized as trying to minimize distress triggered by the stressor, which is directed at
emotional arousal (Carver & Connor-Smith, 2010). Another distinction of coping is made
between engagement and disengagement (Skinner, Edge, Altman, & Sherwood, 2003).
Engagement coping aims at dealing with threat or threat-related emotions, while
disengagement coping aims at avoiding the threat or escaping from threat-related
emotions (Moos & Schaefer 1993; Roth & Cohen 1986; Skinner, Edge, Altman, &
Sherwood, 2003). Engagement coping can take the form of both problem-focused and

26
emotion-focused strategies. For example, one can be engaged in coping with the distress
by seeking emotional support as well as advice from a social support group.
Disengagement coping is primarily emotion-focused, including denial, avoidance,
wishful thinking, and so forth (Carver & Connor-Smith, 2010). Krohne’s Model of
Coping Modes (MCM) (1989, 1993, 1996) provides a nuanced theoretical framework to
examine “monitors” and “blunters.” Based on the Theory of Uncertainty Management,
the MCM explains individual differences in responses to stressful events. The theory
posits that aversive stimulus and ambiguity are the two essential components in a
stressful situation that cause emotional arousal and uncertainty respectively. A person’s
coping mode is determined by his or her susceptibility to emotional arousal and
uncertainty. That is, when a person is more susceptible to uncertainty, he or she is likely
to habitually employ vigilance coping, which is characterized as active information
seeking. A person with a vigilance coping style is motivated to reduce uncertainty by
gaining more information and resources. In contrast, a person who is more susceptible to
emotional arousal is likely to habitually employ cognitive avoidance coping, which is
characterized as withdrawal from related cues of the threat and act as an information
avoider (Krohne, 1989, 1993). Therefore, vigilance coping in the present study was
characterized as an individual’s dispositional coping style, and referred to a generalized
way of responding to a stressor as an information seeker. It is a relatively stable
psychological tendency in the face of problems. It is relevant in the context of mobile
self-tracking because vigilance coping manifesting a problem-solving and information-
seeking tendency could be related to mobile self-tracking practices for health. That is,
individuals who display a vigilance coping style might be more prone to the idea of

27
mobile self-tracking, as it appears to align with vigilant behavior patterns, and thus have
greater autonomous motivation to perform it.
Personal innovativeness toward mobile service (PIMS). Previous studies
exploring determinants of consumer mobile health usage intention found that consumers’
intention to use mHealth, assimilation of mHealth, and channel preferences for health
services were affected by personal innovativeness toward mobile service, perceived
health conditions, health care availability, health care utilization, and demographic
variables (Lu, Yao, & Yu, 2005; Rai et al., 2013). Among these determinants, personal
innovativeness toward mobile service (PIMS) appears to be the most important predictor
of intentions to use mobile health services that interacted with health conditions (Lu,
Yao, & Yu, 2005). It is defined as “the degree to which an individual is willing to try out
any new mobile technology service” (Rai et al., 2013, p. 5). Specifically, participants
who scored both high in PIMS and high in perceived healthiness were significantly
associated with higher mHealth assimilation and mHealth use. Further, participants who
had both high PIMS and high vulnerability to chronic disease were significantly
associated with higher mHealth assimilation and mHealth use. These findings reveal that
mHealth use is not solely driven by health conditions. Instead, technology-related
attitudes are also important in predicting mHealth use behavior. This is consistent with
Venkatesh et al. (2003)’s proposition in the UTAUT model that performance expectancy
and effort expectancy are keys to predict the use of technology. Therefore, PIMS is
particularly relevant in the context of mobile self-tracking as it specifies a person’s
predisposition towards trying out different mobile applications, which should be applied
to mobile self-tracking behavior.

28
Self-efficacy. According to social cognitive theory, self-efficacy is defined as a
person’s conviction and assessment of his or her ability to undertake the behavior
required to produce certain outcomes, and it is one of the crucial determinants of
individual behavior (Bandura, 1991a; Simon-Morton, McLeroy, & Wendel, 2012).
Without it, people “have little incentive” to take action in the face of difficulties
(Bandura, 2002, p.3). For example, Rimal (2000) found that self-efficacy determined
whether individuals translated perceived risks into health information seeking and
whether they translated acquired knowledge into health behavioral practices. Self-
efficacy was widely used in health interventions aiming to increase participants’ self-
regulatory skills. For example, children were enrolled in video games and were told to
take care of virtual animals to prevent them from being diabetic, asthmatic, etc (Bandura,
2004). Self-efficacy is also outlined in several major health theories, such as the Health
Promotion Model (Pender, 2011), the Health Belief Model (Rosenstock, 1974), and the
theory of planned behavior (Ajzen, 2002), as an important factor influencing health
behavior. For example, Jackson, Tucker, and Herman (2007) found that health self-
efficacy was a significant predictor of one’s level of engagement in a health-promoting
lifestyle. In the context of mobile self-tracking for weight management, one’s own
confidence and assessment in using mobile technology to manage weight concerns are
theoretically associated with action taking in mobile-self-tracking. Although m-health
literacy may be inherently embedded in self-efficacy here because one’s perceived
capability to perform self-tracking via mobile technology is implied based on his e-health
literacy ⎯ the ability to “seek, find, understand, and appraise health information from
electronic sources and apply the knowledge gained to addressing or solving a health

29
problem” (C. D. Norman & Skinner, 2006) ⎯ self-efficacy focuses more on one’s own
perceived behavioral control over the behavior and the controllability in relation to one’s
surrounding environment (Ajzen, 2002). In other words, whether a person believe in own
capability, free will, and independence altogether to take on mobile self-tracking will
play an important role in determining his or her motivation and behavior.
Social Norms and Normative Beliefs. Social influence is identified as a substantial
determinant influencing health behavior in various theories, including social cognitive
theory (Bandura, 2004), the theory of reasoned action (TRA) (Fishbein & Ajzen, 1975),
the theory of planned behavior (TPB) (Ajzen, 2002), and social norms theory (Berkowitz,
2004), etc. Because health behavior is largely contextual, interpersonal influence plays an
important role in the process (Simon-Morton, Haynie, & Noelcke. 2009). Social
influence processes take the forms of socialization, selection, and social norms, and,
notably, social norms do not have to be accurate to be influential (Berkowitz, 2004).
Moreover, both the Unified Theory of Acceptance and Use of Technology (UTAUT) and
Health Promotion Model specify that social influence is one of the important
determinants of health behavior and technology adoption (Pender, 2011; Venkatesh,
Morris, Davis, & Davis, 2003). That is, health behaviors and technology adoption are
socially mediated through perceived social norms and interpersonal communication
(Heaney & Israel, 2008).
In addition, research has shown that one’s behavior is associated with his or her
perceived behavior of social ties no matter the perception was correct or incorrect in a
person’s assessment (Valente, 2010). Social networks studies have shown beneficial
effects on health of supportive social networks, implying that perceived social norm as a

30
form of social influence may be the key to understanding mobile self-tracking behavior
(Heaney & Israel, 2008, p.197). In other words, if a person believes that mobile self-
tracking is trendy and is endorsed in his or her social network, he or she will be more
likely to engage in the behavior than those who do not believe so.
Furthermore, social influence can also be examined through a micro-level
egocentric lens. From a network perspective, individuals’ health-related behavior is the
function of the beliefs, attitudes, and behaviors of their network associates (Valente,
2010). Social network theories posit that individuals’ social networks are important for
behavioral adoption for several reasons. Firstly, one’s social ties can usually inform or
provide resources and opportunities to individuals regarding a health behavior (Valente,
2010). Moreover, role modeling and vicarious learning are common phenomena through
which individuals acquire skills or learn behaviors in a given social network (Bandura,
2004). In other words, an individual’s action-taking is associated with his personal
network exposure (Valente, 1996), which can be derived from egocentric network data
(Valente, 2010, p.61). For example, studies have shown that social contagion exists in a
wide variety of human behaviors, and have shown the “spread” of emotions, health
problems, and health-related behaviors as network effects in one’s social network, such
as obesity, happiness, cancer screening, food consumption, and alcohol consumption, etc.
(Christakis & Fowler, 2013). Furthermore, relational factors, such as tie strength and
similarity, are important in understanding mobile self-tracking behavior. For some risky
behavior, such as drug use or syringe sharing, behavior adoption is associated with the
behavior found in people’s strong ties (Valente, 2010). As a result, it is conceivable that
people may be more likely to perform mobile self-tracking if their strong ties engage in

31
mobile self-tracking. A social network approach allows researchers to look at how social
influence comes into play at an interpersonal level.
See Figure 3 for a summary of four dimensions of underlying predictors
associated with mobile self-tracking.

Figure 3. Underlying predictors associated with mobile self-tracking.
Other Underlying Constructs
Autonomous motivation. According to Patty and Cacioppo (1986a, 1986b),
motivation refers to one’s underlying interest in an issue, and it arises from dispositional
and situational factors that serve as an antecedent to patterns of human behavior (Dutta &
Feng, 2007). Specified in many health behavior theories, motivation is conceived as an
important predictor of action taking, although it is conceptualized somewhat differently.
Social cognitive theory posits that individual motivation is “a product of reciprocal
interaction of the person, environment, and behavior”; the theory of planned behavior
measures motivation as the “intention” to perform a particular behavior (Simon-Morton,

32
McLeroy, & Wendal, 2012, p.238). Personality theories generally assume motivation as
an “internal preferences” of “tendencies” to take a particular action (Fortier, William,
Sweet, & Patrick, 2009; Simon-Morton, McLeroy, & Wendal, 2012, p.238).
Self-determination theory further differentiates different types of motivation and
highlights autonomous motivation, which is favored and posited as a key to influence
healthy behavior and to sustain behavior change (Fortier, Williams, Sweet, & Patrick,
2009; Simon-Morton, McLeroy, & Wendal, 2012). The theory makes a distinction
between intrinsic motivation and extrinsic motivation. Intrinsic motivation is formed by
identified regulation and integrated regulation whereby individuals perform a behavior
because of its perceived importance and consistency with goals and values (Ryan,
Patrick, Deci, & Williams, 2008). On the contrary, extrinsic motivation is formed by
external regulations and may not lead to voluntary behavior. The theory puts an emphasis
on the role of intrinsic motivation that is usually missing in most health behavior theories
(Ryan, Patrick, Deci, & Williams, 2008). Intrinsic motivation, also known as autonomous
motivation, is defined as one’s intrinsic interest and desire in and internal reason for
doing something (Ryan, Patrick, Deci, & Williams, 2008; Simon-Morton, McLeroy, &
Wendal, 2012). In the current study, it specifically refers to one’s internal motivation to
manage health concerns through the assistance of mobile technology. Those who perform
certain behavior out of autonomous motivation have identified with the behavior as
something meaningful that they value. For example, it has been shown that autonomous
motivation influences adherence to interventions among participants. Research indicated
that participants in physical activity interventions for a more internal reason (i.e.,
autonomous motivation) generally had greater success at adhering to a physical activity

33
regimen (Fortier, Williams, Sweet, & Patrick, 2009). It is unclear whether mobile self-
tracking practices will be most effective when used by self-guided delivery, under
provider oversight, or in conjunction with formal programs; nonetheless, the point
addressed here is that, according to Pagoto, Schneider, Jojic, DeBiasse, and Mann (2013),
self-guided use of mobile applications like mobile self-tracking for health “would seem to
have the broadest potential for reach, but perhaps the lowest potential for efficacy” as “it
may require a higher level of motivation” compared with other supervised methods
(p.580-581). Therefore, autonomous motivation should be highlighted when we examine
self-guided mobile applications relative to provider-supervised approaches (Paddock,
2013). It is thus argued that considering people’s autonomous motivation in relation to
their mobile self-tracking behavior is necessary and important.
In addition, the theory further explains how social environments influence
individuals’ interest an motivation in terms of autonomy, competence, and relatedness
(Fortier, Williams, Sweet, & Patrick, 2009). It is posited that individuals’ innate
psychological needs (i.e., autonomy, competence, and relatedness) are the bases of
intrinsic motivation, the keys to internalization and persistence of health behavior
(Fortier, Williams, Sweet, & Patrick, 2009). When these innate needs are fulfilled by
one’s social environment, individuals are able to function well and that leads to positive
outcomes, such as innate growth, integration of healthy lifestyle, and wellness (Fortier,
Williams, Sweet, & Patrick, 2009). Taking motivation into the mobile self-tracking
context, individuals who inherently come to value weight management issues and
endorse weight management activities are thus more likely to initiate and sustain mobile
self-tracking for their concerns. In turn, autonomous motivation to engage in mobile self-

34
tracking for weight management can be maintained or enhanced by fulfilling the needs
for autonomy, competence, and relatedness through their tracking experiences. Although
autonomous motivation and controlled motivations are not necessarily opposite
dimensions and individuals are likely to “report both autonomous and controlled
motivations for a given domain,” nonetheless, it is believed that individuals endorsing
more autonomous motivation than controlled motivation would lead to the most positive
outcomes (Ratelle, Guay, & Vallerand, 2007, p. 735; Ryan & Connell, 1989; Ryan, Plant,
& O’Malley, 1995).
The next chapter provides an overview of contemporary mobile health programs
and how they are used in the area of weight management for health.

35
CHAPTER THREE: OVERVIEW OF MOBILE HEALTH PROGRAMS
A wide range of mobile health programs and interventions have been developed
by researchers who have taken advantage of the technological capabilities of mobile
phones to enhance health monitoring (Klasnja & Pratt, 2012).
Klasnja and Pratt (2012) identified five mobile technologies that have been
widely used to support health monitoring, including text messaging (SMS), camera,
applications, automated sensing, and Internet access. Text messaging is a “push”
technology with minimum efforts on the receivers’ side (Gerber, Stolley, Thompson,
Sharp, & Fitzgibbon, 2009, p. 22; Klasnja & Pratt, 2012). It has been widely used for
sending reminders (Gold et al., 2011) and educational content (Chuang & Tsao, 2013;
Lim et al., 2012), and for maintaining users’ awareness of their health goals (Fjeldsoe,
Miller, & Marshall, 2010). Cameras are usually used to collect health-related information
such as food consumption, a condition, or a contextual factor. Mobile applications are
created by researchers and commercial companies to help individuals log their data on a
verity of health behaviors (e.g., diet, exercise, diabetes, blood glucose, etc.), or to teach
health-related skills in the form of mobile games. Most contemporary mobile phones are
equipped with built-in sensors such as GPS and accelerometers that enables detection of
users’ location and states without using external sensing devices (Klasnja & Pratt, 2012).
Built-in sensors also increase the acceptance of health interventions over mobile phones.
The capability of mobile phones to connect to the Internet makes it easy and fast for users
to upload and share data with others(Klasnja & Pratt, 2012).
Using mobile phones to track health-related parameters is at the core of mobile
health applications, which is often referred to self-monitoring⎯constant monitoring and

36
regulation on own behavior to accommodate social situations and to achieve goals
(Bandura, 1984, 1996, 2004; Ewart, 2009; Klasnja & Pratt, 2012). The concept of self-
monitoring in mobile self-tracking should not be confused with the concept of self-
monitoring of expressive behavior in social psychology proposed by Synder (1974),
which refers to a personality trait in which individuals adapt their self-presentation to
coincide with their impression of what others would like and what are socially
appropriate (Synder, 1974). Instead, the idea of self-monitoring in mobile self-tracking
concerns self-observation and self-regulation with respect to behavioral patterns and
behavior change in light of individual goals (Bandura, 2004). Self-tracking applications
attempt to reduce the efforts involved in these self-monitoring activities. For example,
some apps allow users to take pictures of their food intake or activities using phone
cameras to log activities (e.g., Wellness Diary), and these apps usually provide
visualization of these data over time to foster reflection on trends. Built-in sensors (e.g.,
embedded pedometers) can also reduce the burden associated with wearing additional
devices when tracking health. Empirically, mobile phones and their associated
functionalities have been widely used for health promotion in many health areas,
including medical and sexual health education (Chuang & Tsao, 2013; Guse et al., 2012),
adherence (Armstrong et al., 2009; Gold et al., 2011; McGillicuddy et al., 2013), smoking
cessation (Free, Whittaker, Knight, Abramsky, Rodgers, & Roberts, 2009), reducing
alcohol and tobacco use (Haug, 2013; McTavish, Chih, Shah, & Gustafson, 2012), mental
illness improvement (Depp et al., 2010), eating disorders treatment (Shaw & Bosworth,
2012), weight loss, diet, and physical activity (Gerber et al., 2009; Haapala, Barengo,
Biggs, Surakka, & Manninen, 2009; Joo & Kim, 2007; Lee, Chae, Kim, Ho, & Choi,

37
2010; Patrick, Raad, & Morman, 2009; G. M. Turner-McGrievy, Campbell, Tate,
Truesdale, Bowling, & Crosby, 2009), disease management (Mulvaney, Ritterband, &
Bosslet, 2011; Riley et al., 2011; Sirriyeh, Lawton, & Ward, 2010), and healthcare
communication efficacy (Wani, Rabah, Alfadil, Dewanjee, & Najmi, 2013). Studies have
shown that self-monitoring applications have positive effects on users’ health and
increase acceptance of interventions (Fox & Duggan, 2013; Klasnja & Pratt, 2012).
From the perspective of modality, mobile health promotion programs take
different forms, including short message service (SMS) for reminders or delivery of
educational material delivery (Armstrong et al., 2009; Chuang & Tsao, 2013; Fjeldsoe et
al., 2010; Gold et al., 2011), mobile games (Grimes, Kantroo, & Grinter, 2010; Lee et al.,
2010), mobile applications (Gerber et al., 2009; Haapala et al., 2009; Patrick et al., 2009;
Reid et al., 2011), as well as mixed methods including using emails, websites, printed
materials, pedometers, or group sections as complements (Joo & Kim, 2007;
Kristjansdottir et al., 2013; Lim et al., 2012; Newton, Wiltshire, & Elley, 2009; Shapiro
et al., 2010; G. M. Turner-McGrievy et al., 2009).
Among various modalities, the majority of mobile programs take advantages of
“push” technology through short message service (SMS) (Krishna, Boren, & Balas,
2009). Webb, Joseph, Yardley, and Michie (2010) indicate that text messages have been
used in several ways, such as promoting interaction with interventions, sending
motivational messages, challenging dysfunctional beliefs, or providing cues to action.
Wei, Hollin, and Kachnowski (2011) reviewed 24 studies on the use of text messaging
for clinical and healthy behavior interventions. It was found that text messaging was
acceptable and showed promising efficacy in most studies. It is generally found that

38
mobile personal contact via SMS helps to support behavior change (Webb, et al., 2010).
In terms of frequency, reviews on SMS mobile health promotion programs
indicate that the frequency of message delivery ranged from daily to once a week, and
varied by health areas. Text messages with intensive frequencies to participants were
usually tapped to diabetes, smoking cessation, asthma, and medical adherence, which
need constant self-monitoring and self-management (Krishna et al., 2009; Wei, Hollin, &
Kachnowski, 2011). Frequent contact with participants is shown to be necessary to
improve outcomes. Wei, Hollin, and Kachnowski (2011) also point out that the effects
on medication adherence appeared to be larger than the impact on clinical outcomes, and
had a higher retention rate.
In terms of outcomes, Cole-Lewis and Kershaw (2010) reviewed mobile health
interventions using SMS for disease prevention and management. It was found that
disease prevention studies targeted a wider range of topics including medication
adherence, weight loss, physical activity, and smoking cessation. Among them, diabetes
management was most advanced in using SMS within the disease management area, and
usually reported significant decrease in HbA1c. On the other hand, significant clinical
outcomes were observed for weight loss in obese adults, while behavioral outcomes were
relatively inconclusive (Cole-Lewis & Kershaw, 2010). For example, Patrick et al.
(2009)’s SMS-based intervention produced significant weight loss in the treatment group,
while Newton, Wiltshire, & Elley(2009)’s SMS trial on diabetic adolesents failed to
produce improvement in physical activity, and it was attributed to the burden of wearing
pedometers among the youths.

39
Taken together, it was found that text messaging was the most widely used
modality of mobile phones in mobile health promotion programs (Riley et al., 2011).
Further, many programs actually incorporated ecological momentary assessment
1
to
measure real-time data and to push out feedback messages. Nowadays, ecological
momentary assessment is built into almost all mobile self-tracking applications and
devices to ease the efforts for assessment and keeping records for health. For example,
Reid et al. (2011) developed an mobile app to help emotionally disordered patients
monitor their emotions. Participants’ real-time data was obtained and transmitted via
SMS to their general practitioners for medical review. The app helps to detect self-harm
behaviors based on the self-reported data and automatically makes alert calls to patients’
general practitioners.
Overall, there is evidence of a short-term effect regarding behavioral and clinical
outcomes observed for weight loss, smoking cessation, and diabetes management in
mobile SMS programs (Cole-Lewis & Kershaw, 2010). Krishna, Boren, and Balas(2009)
found that 84% (16 out of 19) of studies in their review reported significant behavior
changes, and 92% of studies reported significant change in clinical outcomes as a result
of a mobile intervention. Findings indicated that SMS alone is sufficient to produce
desired health outcomes. This is consistent with Heron and Smyth (2010)’s review on
ecological momentary interventions, which has shown that ecological momentary

1
Ecological Momentary Assessment (EMA) refers to diverse ambulatory assessment techniques, including
paper diaries, behavioral observation, self-monitoring systems, experience sampling, and ambulatory
monitoring of physiological information. Ecological Momentary Intervention (EMI) refers to interventions
that incorporated EMA that are delivered in people’s natural environments. Most of them are mobile-based
interventions. EMA and EMI overcome the drawbacks of retrospective self-report and have better
ecological validity and real-world generalizability (Heron & Smyth, 2010).

40
interventions can be effectively implemented for weight loss, anxiety, diabetes
management, healthy eating, physical activity, and smoking cessation interventions.
Evidence-Based Strategies for Mobile Health Programs
Social support and social influence are commonly used strategies in mobile
programs to enhance relatedness among users. Klasnja and Pratt (2012) identified three
types of social influence via mobile phones, which take forms of (1) facilitating peer-to
peer support or competition (2) facilitating support from significant others (e.g., family or
friends) (3) leveraging peer modeling. Social support has been shown to greatly influence
one’s self-care behavior and persistence (Klasnja & Pratt, 2012). For example, in
Fjeldsoe et al. (2010)’s MobileMum study, a nominated support person by each
participant received text messages as to how to help the participant. The intervention
produced a large effect size (d= 1.22). Moreover, many mobile applications have online
communities where users can seek help, advice, support, or exchange information with
like-minded users, such as MyFitnessPal, Runkeeper, etc. Generally speaking, creating a
sense of community can increase motivating interaction among peer users if group
members adhere to well-established principles of group formation and cohesion, such as
building group identity, sharing group activities, and working toward a shared goal
(Ewart, 2009). In this case, social influence is an effective means of self-regulatory
support, such as the Quantified Self community. Nonetheless, how the social component
is designed can greatly affect usefulness and acceptance among users.
Tailoring and real-time feedback. Feedback was another evidence-based strategy
used in mobile health programs, including content-tailored, time-tailored, or general
feedback sent via SMS to participants. Content-tailored messages in the mobile

41
interventions were either created through working with participants at the onset of
interventions (Fjeldsoe et al., 2010; Gerber et al., 2009), generated by computer using a
default algorithm (Free et al., 2013; Haapala et al., 2009), or written by researchers,
therapists, or physicians (Kristjansdottir et al., 2013). Time-tailored messages were pre-
arranged by researchers according to participants’ preference in receiving messages. For
example, in Partick et al. (2009)’s sutdy, a Web-based application was used to set user
prefernces, determine the appropriate timing and messages to send, and to process the
received replies. Studies which provided tailored feedback were generally very successful
in producing significant outcomes, as seen in the Kristjansdottir et al.(2013) study on
pain acceptance and Reid et al. (2011) study on emotional self-awareness. On the other
hand, studies which only provided general, non-tailored messages/feedback to
participants led to dissatisfaction. For example, participants in Gold et al. (2011)’s sexual
health and sun screen study requested tailored messages when there were only general
messages sent. In Joo and Kim (2007)’s mobile wieght loss study, about 30% of
participants were not very satisfied with the general messages sent to them in terms of
timing, effectiveness and helpfulness. As mobile techonolgies become more sophisticated,
it is anticipated that using EMA to assess physiological or environmental cues to tailor
intervention messages and feedback is increasingly essentail (Heron & Smyth, 2010). For
example, some mobile studies provided real-time intervention via text messages that have
shown effectiveness. In Free et al. (2009)’s smoking cessation study, a “crave” text can
be sent by participants to request an immediate, supportive message whenever they had a
craving for a cigarrette, and it was shown to be helpful. Similarly, Reid et al. (2011)
developed the Mobiletype program with mental modules sent several times a day to

42
assess participants’ mental states, and the mobile application incorporated automatic real-
time alerts sent to physicians when high-risk self-harm behaviors were identified among
emotionally disordered patients.
Interactivity and communication with professionals. Self-care advice by clinicians
or health professionals appears to be effective. For example, in a study by Kristjansdottir
et al. (2013), researchers as trained therapists provided personalized feedback to
participants based on their diaries describing feeling and symptoms. Reid et al. (2011)
sent reminders and make frequent contact with patients’ general practitioners, who
provided patients with clinical resources and referral pathways. Also, on-call
psychologists and counselors were automatically contacted when patients with mental
issues responded in a way signifying they were at risk of self-harm or suicide (Reid et al.,
2011). These elements significantly increased emotional self-consciousness and
decreased negative clinical outcomes among patients. In Fjeldsoe et al. (2010)’s
intervention, participants were provided with two consultations througout the intervnetion
with trained behavioral counselors to help formulate their goals, check progress and
identidy barriers. In a study by Patrick et al. (2009), participants received brief (5 to15
minutes) monthly phone calls from a trained health counselor to encourage continued
participation in the program and to work through any technical issues they might be
having with the intervention. Joo and Kim (2007) had a dietician assess participants’
diatary behavior and help them to set up diet goals in the initial meetings. In a study by
Shapiro et al. (2010), patients with bulimia nervosa undertook a weekly group therapy
intervention along with the SMS intervention, hence the role of theapists in their
intervention is essetital in help them make progress. Overall, mobile interventions

43
offering self-care advice and consultation from health care professionals (e.g., physicians,
therapists, counselors, dieticians, and psychologists, etc.) all have produced significant
improvement. It is worth noting that, despite the fact that interactivity remains an
important feature in many mobile interventions, the major use of mobile programs for
self-tracking is for extending self-care and self guided management, and thus to reduce
clinician time and burden (Depp et al., 2010). In most mobile self-tracking programs,
interactivity is provided by fancy interfaces and real-time mobile assessment protocols to
enhance user-system interaction.
Characteristics in Relation to Effectiveness
To summarize, the design of the mobile health programs and interventions
reviewed varied significantly in terms of theme, timing, and frequency of delivery.
Therefore, any generalizations about effectiveness and optimal use of mobile
device/applications from empirical studies are not warranted (Shaw & Bosworth, 2012).
However, it can be observed that mobile studies with larger effect sizes and positive
outcomes usually (1) provided some kind of interactivity where users’ input and the
system’s output are frequent; (2) provided individual review, advice, or feedback for
behavioral modifications by professionals or automated computer algorithm (3) provided
real-time interaction in the form of tailored message reminders or call alerts that serve as
cues to action (Simons-Morton, McLeroy, & Wendel, 2012, p.299). These characteristics
in relation to effectiveness can be easily fulfilled by contemporary mobile self-tracking
applications and mobile devices, through which interaction between users and the system
is facilated to provide trackers with rich information, cues to action, and feedback
regarding their activities and health concerns.

44
mHealth & Self-tracking on Weight Loss, Diet & Physical Activity.
Traditional behavioral weight management programs are reflected in evidence-
based lifestyle interventions, which usually involve intensive in-person visits, weight loss
counseling, a longer period of time (generally for a period of six months to one year or
more), and higher costs (Breton, Fuemmeler, & Abroms, 2011). Mobile health programs
leverage technology in order to minimize the burden on users. Specifically, evidence-
based strategies used in weight management programs usually include self-monitoring of
diet and weight, dietary strategies, regular physical activity, and social influence ( Breton,
Fuemmeler, & Abroms, 2011; Pagoto, Schneider, Jojic, DeBiasse, & Mann, 2013).
Technology-enhanced features afforded in mobile self-tracking applications and devices
can prompt engagement and reduce users’ burden. These features include barcode
scanners for nutrition information, built-in accelerometers for physical activity tracking,
online social networks, email or text messages as reminders, calendars for goal setting,
tracking negative thoughts and stress, and flags for lapses in dietary goal adherence, etc.
(Breton, Fuemmeler, & Abroms, 2011).
Although not all mobile health programs are designed to direct individuals to self-
tracking practices, the way most programs work is to some extent to form the habit of
constant self-monitoring among the users. Mobile weight management programs have
somewhat different approaches to their goals, including increasing physical activity,
reducing sedentary lifestyle, or diet and healthy eating (Shaw & Bosworth, 2012). Riley
et al. (2011) found that most mobile health promotion in weight loss, diet, and physical
activity areas heavily utilized SMS to deliver interventions, including non-tailored
informational text messages and customized feedback. Most mobile programs on diet,

45
weight management, and physical activity are embedded with ecological momentary
assessment that is used to collect and record accelerometer data. Goal-setting is another
frequent behavior strategy used in most mobile programs (G. J. Norman, Zabinski,
Adams, Rosenberg, Yaroch, & Atienza, 2007). Users can set up their goals of weight
management, and break down a big goal into smaller, visualized, achievable goals with
the assistance of technology. For example, a public-accessible, free-of-charge mobile
application, MyFitnessPal, is a currently available commercial mobile application, which
is designed to help users track calorie consumption, diet, and exercise in the form of a
mobile diary. The application has a large food database of over 3,000,000 and a built-in
barcode scanner for users to easily log foods they consume daily. The app also helps log
exercise, calculates daily calories consumed and burnt, and provides nutrition information
accordingly. It visually shows users their calorie balance and progress in relation to their
personal goal. By logging into the online forum, users can also participate in the online
community where current users seek help, advice, support, or exchange information.
Feedback proved to be an effective component across mobile health interventions and can
produce significant weight loss outcomes among participants (G. J. Norman et al., 2007).
A review of Internet-based interventions for weight management concluded that
Internet-based approaches are efficacious for improving behavioral outcomes with
tailored experience and easy self-monitoring (Breton, Fuemmeler, & Abroms, 2011).
Many of the reasons that make Internet-based approaches promising may apply to mobile
health promotion, while mobile programs add portability and availability regardless of
location and setting to interventions (Breton, Fuemmeler, & Abroms, 2011), and also
increase the ability of tailoring feedback and self-monitoring. From a functional

46
perspective, mobile health self-tracking programs not only help reduce the burden of
behavioral strategies, but also facilitate self-care and reduce clinicians’ burden.
Theoretically speaking, mobile self-tracking is able to enhance individuals’ competence
in achieving their goals, increase a sense of autonomy via non-supervised but informed
self-care, and increase a sense of relatedness via connecting to the mobile self-tracking
communities. The increased competence, autonomy, and relatedness can thus further lead
to adaptive coping and positive health outcomes as empirical studies indicated.
In terms of intervention duration of mobile programs, it was found that weight
management interventions ranged from two weeks to 12 months, and the majority were
less than 6 months (Shaw & Bosworth, 2012). G. J. Norman et al. (2007) found a range
from one to six months in their review. In the Riley et al. (2011) review on weight,
physical activity, and diet interventions, intervention duration was not provided, while the
authors indicated that the duration varied substantially across studies. Most of mobile
weight loss interventions appear to not have follow-ups after the post-test (Haapala et al.,
2009; Joo & Kim, 2007; Lee et al., 2010; Patrick et al., 2009; Sirriyeh et al., 2010; G. M.
Turner-McGrievy et al., 2009). This is consistent with the finding reported by Heron and
Smyth (2010), which indicated a lack of follow-up data in most mobile interventions, and
thus limit our understanding regarding the long-term effects of mobile interventions.
Nonetheless, it is observed that the goals of mobile weight management programs can be
reached within the range of a few weeks to several months.

47
Empirical Studies with Mobile Self-tracking
Lee et al. (2010) conducted a 6-week weight loss intervention using a mobile
phone-based game system, SmartDiet, which helps users track their weight and nutrition
intake, and promotes knowledge among obese people in Korea. Participants were
recruited from an obesity clinic. The mobile application features two functions, diet
planner and diet game. The frequency of using the app was up to participants. Body
composition was measured at baseline and after intervention at the clinic. Despite a
relatively short intervention duration, results showed that major outcomes (i.e., fat mass,
weight, and BMI) were significantly decreased for the intervention group. About 83% of
participants in the intervention group thought the application was useful.
Haapala et al. (2009) conducted a randomized controlled trial to evaluate a mobile
program, Weight Balance®, on weight loss among healthy overweight adults for 12
months. The program calculated diet information for users, sent automatically tailored
messages discouraging excessive daily calorie intake and encouraging physical activity
according to their reply. The program also had a password-protected website for
participants to keep diary records and track their weight loss in visual forms. The
experimental group had significantly more weight loss and greater reduction in waist
circumference. Satisfaction was 7.8 out of 10 on average at 12 months. The most useful
features reported by participants were the free app, regular reporting of weight,
immediate feedback, and short-term goal setting.
Turner-McGrievy and Tate (2011) conducted a weight-loss intervention
incorporating self-monitoring and social support. Participants were randomly assigned to

48
“podcast only” or the enhanced “podcast plus mobile tool” treatment group. The
intervention group was asked to download a diet and activity monitoring app, create a
Twitter account to read messages from the coordinators, and were encouraged to interact
with their cohorts on Twitter. Results show that the 6-month, minimal-contact
intervention was effective at helping participants achieve a mean weight loss of 2.7% of
their body weight, however, the study failed to produce clinically significal weight loss
outcomes (5%). Also, there were no significant differences found in all primary outcome
measures between groups, including changes in weight, physical activity, dietary intake,
self-efficacy, knowledge, and eating behavior at 3 and 6 months. A possible explanaiton
provided by the authors is that people in the control group were not prohibited from using
the self-monitoring application if they wanted. In addition, the self-monitoring app and
Twitter account were poorly used by participants in the intervention group, which is
possibly because they were poorly integrated into participants’ lives. The authors
concluded that participants may have found different ways of self-monitoring and gaining
social support outside the intervention, which was reflected in their reduced weight.
Cavallo et al. (2012) conducted a 12-week randomized controlled trial to examine
the effectiveness of a physical activity intervention incorporating Facebook social support
and a self-monitoring tool among female undergraduate students. The intervention group
had access to an education website designed by researchers (i.e., INSHAPE), and were
provided with a physical activity self-monitoring tool as well as social support through
invitation to a physical-activity-themed Facebook group. The control group only accessed
the limited version of INSHAPE. Findings revealed that both groups displayed increases
in perceived social support and physical activity over time, and there were no difference

49
between groups. In other words, the use of an online social networking group plus self-
monitoring did not produce greater perceptions of social support or physical activity as
compared to the education-only control condition. The authors explained that both
intervention and control groups were receiving social support from outside the study,
which contributed to their overall increases in perceived social support and physical
activity.
Grimes, Kantroo, and Grinter (2010) conducted a pilot study on a mobile casual
game to promote healthy eating among 12 adults for 3 weeks. The game, OrderUP!, was
invented by researchers and advised by a dietitian, targeting African Americans by
incorporating distinctive African American cuisine. Players interacted with the game
using buttons on the phone’s keypad, and they were asked to choose the healthiest foods
for their customers as waiters from the dishes displayed on the screen. Simple feedback
was given using a stoplight to notify players whether their choice was the healthiest or
not. In the end, participants reported having fun with the game, and more than half of the
participants applied what they learned. Findings suggested that the mobile game helped
raise consciousness, improve understanding of healthy eating, facilitate nutrition-related
thinking, form health habits and helping relationships, and engage in healthier lifestyles.
Taken together, previous studies unanimously agree that mobile phones are easily
integrated into people’s lifestyle and help reduce barriers to behavior adoption (Gerber et
al., 2009; Grimes, Kantroo, & Grinter, 2010; Haapala et al., 2009; Joo & Kim, 2007;
Krishna et al., 2009; Lee et al., 2010; Patrick et al., 2009). As social action theory posits,
health habit change is strongly associated with individual’s social environment. Mobile
self-tracking programs can help individuals identify personal and environmental factors

50
that influence their weight management-related practices, and further enhance adaptive
coping activities such as healthy eating or exercising among self-trackers. Despite the
claim by Pagoto et al. (2013) that behavioral strategies such as improving motivation,
problem solving toward goals, and stress-reducing were generally missing across weight
management mobile applications, the theoretical perspectives reviewed imply an indirect
influence of motivation on health outcomes by empowering individuals through mobile
self-tracking activities.
To sum up, it generally appears that mobile self-tracking can be used to guide
behavior change through spurring constant self-monitoring, consciousness raising, and
constructive thinking, which can in turn produce positive health outcomes in the context
of health interventions (Grimes, Kantroo, & Grinter, 2010). However, it is unclear
whether mobile self-tracking can be equally effective when it is self-guided and
unsupervised.
Research Questions & Hypotheses
Based on the literature review, the following research questions and hypotheses
were proposed regarding the predictors, effects, mediator, moderator, and social influence
associated with mobile self-tracking for health in the domain of weight management:
RQ
1
: How do mobile self-trackers differ from traditional trackers and non-trackers in
terms of demographics, antecedents, and personality traits? (Group Comparison)
RQ
2
: Following the previous question, how do personal and social factors predict
autonomous motivation to mobile self-track among mobile self-trackers? And how does

51
autonomous motivation predict mobile self-tracking engagement among mobile self-
trackers? (Predictors of Engagement)
H
1
: Autonomous motivation to mobile self-tracking engagement is positively
predicted by H
1a
) health consciousness, H
1b
) health locus of control, H
1c
) mHealth
literacy, H
1d
) vigilance coping, H
1e
) personal innovativeness for mobile service,
H
1f
) self-efficacy, H
1g
) social norms, and H
1h
) normative beliefs.
H
2
: Mobile self-tracking engagement is positively predicted by autonomous
motivation.
RQ
3
: What are the effects of self-guided mobile self-tracking for health on weight
management? Does group membership affect health outcomes? Do health outcomes
differ between mobile self-trackers, traditional self-trackers, and non-trackers after
controlling for health condition and demographics? (Effects; Group comparison)
H
3
: Mobile self-trackers have significantly higher H
3a
) health anxiety, H
3b
) social
support, H
3c
) healthy eating, H
3d
) physical activity, H
3e
) well-being, H
3f
)
satisfaction, H
6g
) goal attainment, and H
6h
) self-determination than traditional
self-trackers and non-trackers after controlling for the covariates.
In addition, given the theoretical and empirical evidence, the role of self-
determination is hypothesized as a mediator of the effects of mobile self-tracking on
health outcomes. In order to test it, the following conditions must be met to support the
mediating role of self-determination: (1) mobile self-tracking behavior should directly
influence health outcomes, (2) mobile self-tracking behavior should influence self-
determination, and (3) self-determination should influence health outcomes and reduce
the influence of the mobile self-tracking behavior when both are included as predictors in

52
the regression analysis. The following research question and hypotheses was proposed:
RQ
4:
How does mobile self-tracking engagement influence individuals’ self-
determination? And, how does self-determination mediate the effect of mobile self-
tracking on individuals’ psychological and behavioral outcomes? (Mediator-SD)
H
4
: Participants who engage more in mobile self-tracking will have significantly
higher self-determination on weight management.
H
5
: Mobile self-tracking engagement will have positive effects on H
5a
) health
anxiety, H
5b
) social support, H
5c
) healthy eating, H
5d
) physical activity, H
5e
) well-
being, H
5f
) satisfaction, and H
5g
) goal attainment.
H
6
: Self-determination mediates the effect of mobile self-tracking on H
6a
) health
anxiety, H
6b
) social support, H
6c
) healthy eating, H
6d
) physical activity, H
6e
) well-
being, H
6f
) satisfaction, and H
6g
) goal attainment.
RQ
5
: Does the amount of ecological momentary intervention (EMI) received during self-
tracking moderate the influence on health outcomes among mobile self-trackers?
(Moderator- EMI)
H
7:
Self-tracking engagement and EMI will interact in predicting health outcomes
such that self-tracking engagement is a more important predictor of health
outcomes for high-EMI than low-EMI mobile self-trackers.
Aside from the variability in the characteristics about respondents that may
predict mobile self-tracking, the present study also looks at social factors that influence
mobile self-tracking as an outcome variable. It is widely accepted that peer influence and
tie strength (Granovetter, 1973; Valente, Watkins, Jato, Der Straten, & Tsitsol, 1997;
Valente & Saba, 2001; Valente, 2010), role modeling (Bandura, 1986, 1997, 2004), and

53
socialization and similarity (Berkowitz, 2004; Valente, 2010) are important mechanisms
that influence individuals’ health behavior, including smoking (Alexander et al., 2001;
Urberg et al., 1997), syringe exchange (Valente & Vlahov, 2001), condom use (Peterson,
Rothenberg, Kraft, Beeker, & Trotter, 2007), and physical activities (Macdonald-Wallis,
Jago, & Sterne, 2012). Therefore, the next research question seeks to explore how social
influence affects individuals’ mobile self-tracking behavior. Specifically, the research
question seeks to examine how relationship characteristics influence mobile self-tracking
adoption. That is, the study also examines how relation type, strength of ties, role
modeling, and demographic similarity of social ties influence ones’ self-tracking
behavior between participants and their alters (i.e., their nominators).
First, personal network exposure plays an important role in determining whether
one will engage in a particular behavior or not. It is a “fundamental and critical variable”
that captures social influence “by measuring the extent one’s network engages in a
behavior” (Valente, 2010, p.66). It is defined as “the degree to which a focal individuals’
alters engage in a particular behavior” (Valente, 2010, p.65). The assumption is that the
more exposure to a particular behavior, the greater likelihood of a person to perform a
particular behavior. For example, in a study of contraceptive method adoption in Bolivia,
participants who reported using a contraceptive method had significantly greater personal
network exposure than those who did not report using a contraceptive method (Valente &
Saba, 2001). Similarly, it is conceived that mobile self-trackers might have greater
network exposure to mobile self-tracking in their social environment than those who do
not self-track.

54
Secondly, Valente (2010) suggests that strong ties are essential for behavior
adoption and behavior change at a micro level as individuals’ behavior is more likely to
be influenced by people with whom they have closer relationships than by those weakly
connected. On the other hand, weak ties may be more important in information
transmission rather than behavioral adoption (Valente, 2010). Therefore, relation type in
terms of closeness should be taken into account in predicting mobile self-tracking
adoption. Individuals should be influenced more by social ties with whom they have a
closer relationship. Along the same line, it is also assumed that the self-tracking decision
is influenced by social ties with whom individuals interact more frequently, and by social
ties whom individuals have known longer. It is based on the assumption that social
influence is most likely to occur between people who have a stronger and closer
relationship that allows persuasion to happen.
Third, social networks provide “role models” for behaviors (Valente, 2010, p.62).
As suggested by social cognitive theory, people adopt behaviors through social learning
processes and role modeling (Bendura, 1996, 2004). It is conceivable that individuals are
more likely to mimic a behavior if their social ties serve as positive role models because
individuals can observe positive rewards from the role models through vicarious learning.
In the context of mobile self-tracking, role modeling is most likely to occur between
those whom a subject perceived as being healthy and attractive because they serve as
testimonials of this particular behavior. In other words, an individual is more likely to
adopt a behavior endorsed by a healthy and attractive alter.
Finally, social network studies have shown that similarity is an underlying factor
of social influence. Egocentric network analysis allows us to look at whether

55
demographic homogeneity between an ego and his or her alters influences behavior
adoption. Specifically, previous studies have shown that a particular behavior is most
likely to be replicated between people who share demographic background, such as age,
gender, and ethnicity. It is thus important to examine whether similarity plays an
important role in mobile self-tracking adoption.
With respect to social influence, the following research question and hypotheses
are proposed:
RQ
6
: How do individuals’ social networks influence their adoption of mobile self-
tracking for weight management? (Personal Network Influence)
H
8
:

Participants who reported being mobile self-trackers had higher personal
network exposure than participants who were non-mobile self-trackers. (Network
exposure)
H
9
: Participants are more likely to have the same tracking behavior as their social
ties whom H
9a)
they contact more frequently, H
9)
they have known longer, and
H
9c)
they have closer relation types. (Strength of Tie: relation type, frequency,
& years known).
H
10
: Participants are more likely to have the same tracking behavior as their social
ties who are perceived to be healthy and attractive (Role Modeling: healthiness
& attractiveness).
H
11
: Participants are more likely to have the same self-tracking behavior as their
social ties who are of the same age, same gender, and same ethnicity (Similarity).

56
The Research Model

Figure 4. An integrated research model for mobile self-tracking on weight management.
Tracking health information to inform one’s own health-related knowledge is
becoming relatively easy with the growth of self-monitoring applications on phones
(Klasnja & Pratt, 2012). This trend is endorsed by one of the objectives in Healthy
People 2020 as “increasing the use of electronic health management tools” (Simons-
Morton, McLeroy, & Wendel, 2012, p.298), with which there is a great potential to
enhance peoples’ overall health. As discussed previously, the use of mobile phones
shows great promise in health care, although it has been accused of imposing threats to
health consumers in terms of increasing burden and health anxiety. There is a lack of
empirical studies on effects of technology-based self-tracking, and no prior study has
examined psychological and behavioral outcomes associated with self-guided mobile

57
self-tracking, as well as interpersonal influence on the behavior. Therefore, targeted at
people who are interested in weight management, the aim of this study is to explore the
underlying factors and outcomes associated with mobile self-tracking. The research
design is outlined in the next section.

58
CHAPTER FOUR: RESEARCH DESIGN AND METHODS
Study Design
A cross-sectional survey was developed to explore the research questions and
hypotheses. A questionnaire was created using Qualtrics online survey software.
Participants were recruited from a paid sampling service, Qualtrics Panel, which is a
large group consisting of participants all over the country whom researchers can recruit to
their surveys. Researchers who plan to use Qualitrics Panel to recruit participants need to
first specify the topic and the targeted population for whom the survey is intended.
Following this, a project manager in Qualtrics will work with the researcher to make sure
that everything is acceptable with respect to the questionnaire before it is distributed.
Finally, the survey will be distributed to a specific Qualtrics Panel pertaining to the
researchers’ interest and topic. Eligible participants in the Panel can participate and the
survey will be closed until the desired number of respondents is reached. Participants
who finish the survey and whose responses pass the quality checks (i.e., attention filters,
speeding check, and forced response questions) will be reimbursed through Qualtrics.
The Questionnaire
The questionnaire collects information on each participant’s demographic
characteristics, health-related psychometrics, and behavior outcomes based on their self-
reported data. There were a few quality control mechanisms embedded in the survey,
including attention filters, speeding check, and forced response questions to ensure
response quality. Three attention filter questions were randomly placed throughout the
survey asking participants to mark a specific point on the scale for the questions.
Participants who failed to respond correctly for these questions were automatically

59
disqualified as a way to exclude respondents who entered poor-quality responses. In
addition, a speeding check was added to the survey to automatically terminate
respondents who were not responding thoughtfully. A median time to completion came in
at 20 minutes during the soft launch with a speeding check measured as one-third the
median time, which was 392 seconds. After reviewing the data, the speeding check was
raised to 500 seconds to further ensure data quality. Forced responses were also
implanted throughout the survey except for the demographic segment.
Pre-determined quotas were specified for three targeted groups: 250 people for
mobile self-trackers, 125 people for traditional self-trackers, and 125 people for non-self-
trackers. The quotas were determined based on two pilot studies done previously and a
national report (Fox & Duggan, 2013) that mobile self-trackers were relatively fewer than
traditional trackers and non-trackers. As a result, a larger quota specified for mobile self-
trackers would satisfy the research purpose and facilitate later statistical analyses and
modeling. A drawback of this design would be a lack of ability to answer the prevalence
question on mobile self-tracking. Nonetheless, it can provide insight on current research
inquiries regarding mobile self-tracking and comparisons across groups.
In addition, in order to obtain social network data to examine social influence on
mobile self-tracking, a set of egocentric questions used in previous national surveys was
modified and used in the current survey (Burt, 1984; Valente, 2010). These questions
were included to measure participants’ personal network derived from their responses to
the name generator about whom they talk to about weight management issues and
concerns, referred to as “alters” in egocentric studies. Participants were asked to give
some identification for up to six people with whom they talk about weight management

60
issues. They were also asked to provide information on some demographic
characteristics, behavior information, and information on interaction with each alter they
nominated. The ego is the person whose mobile self-tracking behavior is being analyzed.
People to whom these respondents are linked serve as close social contacts in the domain
of weight management. Alters were assessed by type of relation to the ego, strength of
relationship, frequency of interaction, as well as some demographic and socioeconomic
characteristics and similarities. Once the personal networks data is collected, the study
can characterize participants’ close social ties and determine whether the network
characteristics are associated with mobile self-tracking behavior.
Measures
Demographics and Antecedents
Demographic information regarding gender, age, education, household income
level, employment, health status, chronic disease, ethnicity, marital status, having child,
smartphone use, social media use, and social media activeness were collected through
participants’ self-reported data.
Health status. Health status was measured by one item asking about the
individual’s perceived healthiness. The item was adapted from CDC’s Health Quality of
Life. “4. How would you describe your health status in general? My health condition is
___” with response options from “Awful” to “Excellent”.
Predictors of Autonomous Motivation
Health consciousness. Health consciousness is defined by Gould (1990) as “the
degree to which someone attends to or focuses on his or her health, and an inner state of

61
self-attention to self-relevant cues reflected in both thought and somatic feeling” (p.228).
It refers to a general, pro-health mindset that directs one’s attention to health-related
issues that has been shown to correlate with health-promoting behaviors in previous
studies. Because health consciousness oftentimes manifests through engaging in health-
related behaviors, concepts used to measure it involve individuals’ health-related
psychological traits and action taking (Hu, 2013). Four items adapted from Hu (2013),
Gould (1988), and Dutta-Bergman (2004b)’s health consciousness scales were used to
assess individual’s level of attention to their health conditions at baseline, post-treatment,
and follow-up at 4 months. Sample items are “I try my best to stay healthy” and “Living
life in best possible health is very important to me” with response options from “strongly
disagree” to “strongly agree.” The Cronbach’s α is 0.857.
Internal health locus of control. It is defined as an individual’s perceived control
over his or her own health (B. S. Wallston & Wallston, 1978; K. A. Wallston et al.,
1978). Items adapted from Wallson and Wallson’s (1978) Health Locus of Control
(HLC) were used in the current study. A sample item was “I believe that my health
heavily depends on how well I take care of myself” with response options from “strongly
disagree” to “strongly agree.” The Cronbach’s α was 0.734.
Personal innovativeness toward mobile service (PIMS). PIMS was measured with
three items adapted from Rai et al. (2013). Sample items were “If I heard about a new
mobile service, I would look for ways to experiment with it” and “Among my friends and
family, I am usually the first to try out new mobile service” with response options from
“strongly disagree” to “strongly agree.” The Cronbach’s α was 0.925.

62
mHealth literacy. mHealth literacy was measured using 3 items adapted from the
eHEALs scale proposed by Norman and Skinner (2006) to gauge participants’ literacy
level in using information technology for health (C. D. Norman & Skinner, 2006). A
sample item was “I know how to use the health information I get via my mobile phone
(s)” with response options from “strongly disagree” to “strongly agree.” The Cronbach’s
α was 0.964.
Vigilance coping. Vigilance coping is characterized as a dispositional coping
style, which is one’s generalized way of responding to a stressor (Krohne, 1993, 1996).
Although different situations bring out different coping approaches, it is believed that
dispositional coping strategy is a relatively stable psychological tendency (Carver,
Scheier, & Weintraub, 1989). According to the Transactional Model of Stress and
Coping, dispositional coping style will influence one’s coping efforts and influence the
effects of coping efforts on coping outcomes. Vigilance coping was measured in terms of
problem-solving and active coping with three items adapted from the COPE Inventory
(Carver, Scheier, & Weintraub, 1989). The scale measures how participants respond to
stressful events and difficulties in their lives with respect to their strategies of dealing
with problems. A sample item was “I take direct action to get around the problem” with
response options from “not at all” to “very much.” The Cronbach’s α was 0.748.
Self-efficacy. Self-efficacy refers to one’s own confidence and assessment of
having the skills and capabilities to perform a behavior. It determines whether individuals
translate perceived risks into information seeking and whether they translate acquired
knowledge into behavioral practices (Bandura, 1984). In the current study, self-efficacy
was defined as individuals’ general assessment of their own capability of assessing to

63
resources, managing own health, and dealing with health concerns. It was measured with
three items. Sample items were “I know what resources are available to me for my health
concerns,” “I know where to find helpful health resources for my health concerns,” and “I
have the skills I need to maintain or improve my health” with response options from
“strongly disagree” to “strongly agree.” The Cronbach’s α was 0.869.
Normative beliefs. Normative beliefs measure participants’ injunctive norms
about what is ought to be in relation to mobile self-tracking. It was assessed by three
items measuring one’s internalized perceptions about the proper ways to act, adapted
from Mackie, Moneti, Denny, and Shakya (2015)’s social influence scales. Sample items
were “I think I should engage in mobile self-tracking for health” and “Those people
important to me think I should do mobile self-tracking for health” with the responses
ranging from “strongly disagree” to “strongly agree.” The Cronbach’s α was 0.885.
Social norms. Based on the theory of planned behavior (Ajzen, 2002) and the
Unified Theory of Acceptance and Use of Technology (UTAUT) (Venkatesh, 2003) to
predict individual acceptance of technology, social influence is one of the important
determinants of health behavior and technology adoption. That is, health behaviors and
technology adoption are “socially mediated through perceived social norms” (Heaney &
Israel, 2008, p.197). A social norm refers to “what people in a group believe to be typical
and appropriate action in that group,” and it is maintained by “social influence” and
“one’s belief in the legitimacy of others’ expectations.” (Mackie, Moneti, Denny &
Shakya, 2015, p.10; Paluck & Ball, 2010). Therefore, If one perceives that people around
him are engaging in self-tracking for health purposes (i.e., descriptive norms), or think he
or she should engage in self-tracking for health purpose (i.e., injunctive norms), then he

64
or she is more likely to comply with the program. Social norms were measured by two
items asking what participants perceive that their significant others do and what their
significant others would expect them to do. Items included “How many of the people
important to you do mobile self-tracking for health purposes?” and “How many people
important to you think you do mobile self-tracking?” with response options from “none
of them” to “all of them.”
Autonomous motivation. Autonomous motivation was measured by five items
adapted from the Intrinsic Motivation Inventory (IMI; McAuley, Duncan, & Tammen,
1989) to access participants’ subjective interest, enjoyment and perceived value
associated with a given activity. Sample items were “I experience pleasure when I engage
in self-tracking” and “I am inherently interested in self-tracking for my health.” with
response options from “strongly disagree” to “strongly agree.” The Cronbach’s α was
0.933.
Self-tracking Engagement. In order to best reflect what self-tracking entails, self-
tracking behavior was measured as a composite variable in terms of behavior
engagement, which consists of four sub-concepts, including adherence, involvement,
intensity, and social share. The questions for adherence asked participants about their
own evaluation of self-tracking adherence. A sample item is “I would say my adherence
to my self-tracking routines is generally ___” with response options from “very poor” to
“very good.” The item for involvement asked participants about how involved they are
with their self-tracking routine: “I would say that my involvement in my self-tracking
routines is generally ___” with response options from “extremely low” to “extremely
high.” The item for intensity asked participants to indicate how frequently they self-track

65
for the health indicators they concern on a 7-point scale from “less than once a month” to
“several times a day.” The item for social share asked participants to indicate how often
they share their self-tracking data to people in their network on a scale from “never” to
“all of the time.” The Cronbach’s α of self-tracking engagement was 0.727.
Mediator
Self-determination. Self-determination theory posits that individuals’ self-
determination to engage in a healthy behavior is influenced by three psychological needs:
autonomy, relatedness, and competence. Therefore, self-determination was a composite
variable consisting of the three sub-components: autonomy, competence, and relatedness.
It was measured by items adapted from the Basic Psychological Need Satisfaction and
Frustration Scale (BPNSFP) (Chen, Van Assche, Vansteenkiste, Soenens, & Beyers,
2015; Chen et al., 2015). Sample items for autonomy, competence, and relatedness are
“Self-tracking helps build my own independence regarding my health,” “I feel more
competent in managing my health goals as I engage in self-tracking” and “I feel more
connected with other people when I am self-tracking.”
Moderator
Ecological momentary intervention (EMI). The purpose of examining EMI as a
moderator is to extend and validate the power of influence on mobile self-tracking
beyond psychosocial and health-related personality traits. As revealed in the literature
review, EMI serves as an important enhancer that allows people to track for various
physiological, emotional, and behavioral indicators. In mobile self-tracking, the
“treatment” is delivered during a user’s everyday life in real time and in natural settings.

66
As a result, the amount of EMI received by mobile self-trackers was measured by a
question asking participants to indicate various kinds of EMI modalities (i.e., instant
activity summary, instant feedback, reminders, alerts, other, etc.) they have received
during mobile self-tracking practices. Then the amount of EMI received by a respondent
was assessed by calculating the total number of real-time features the respondent received
during his or her mobile self-tracking practices.
Psychological Outcomes
Health anxiety. Health anxiety was measured by four items adapted from the
Health Anxiety Inventory (Salkovskis, Rime, Warwick, & Clark, 2002). Sample items
were “I have been bothered by fears about my health after my routine tracking activities”
and “I have been afraid that I have an illness/disease after my routine tracking activities”
with the response options ranging from “never” to “all of the time.” The Cronbach’s α
was 0.866.
Perceived social support. Perceived social support was measured by four items
adapted from the Multidimensional Scale of Perceived Social Support (MSPSS) (Zimet,
Dahlem, Zimet & Farley, 1988). The scale accesses people’s perceived support from
family, friends, and significant others. The validity and internal consistency has been
validated in previous studies (Pedersen, Spinder, Erdman, & Denollet, 2009). A sample
item was “I know there is a person with whom I can talk about my problems when I
engage in self-tracking” with response options from “strongly disagree” to “strongly
agree”. The Cronbach’s α was 0.893.
Well-being. Well-being was measured with four items created and adapted from

67
the Scale of Positive and Negative Experience developed by Diener, Wirtz, Tov, Kim-
Prieto, Choi, Oishi and Biswas-Diener (2010), and the Quality of Well-being Scale- Self
Administered (QWB-SA) developed by Seiber, Groessl, David, Ganiats, and Kaplan
(2008). The four items asked participants how they feel about their status quo and about
life in general since they started self-tracking for health. Sample items were “I have been
contented with my life since I engaged in self-tracking” and “I have been feeling happy
since I started self-tracking” with response options from “strongly disagree” to “strongly
agree.” The Cronbach’s α was 0.940.
Satisfaction. Satisfaction was defined as participants’ evaluation of their own
feelings toward their self-tracking outcomes, and was measured by one item asking
participants “in general how satisfied are you with the outcome(s) of your self-tracking
activities” with response options from “very dissatisfied” to “very satisfied.”
Goal attainment. Goal attainment was defined as participants’ own evaluation
about their goal achievement with respect to self-tracking. The question asked
participants whether they were able to achieve their health goal(s) through self-tracking
practices with response options “No,” ”Yes,” and “I am not sure”. Answers on “No” and
“I am not sure” were further coded and combined into one category as opposed to
responses on “Yes” for statistical analyses.
Behavioral Outcomes
Healthy eating for a healthy weight. Any healthy approaches to weight
management will include some kind of physical activity and behavior management
(Macera, 2003.). According to Centers for Disease Control and Prevention, healthy

68
weight can be achieved through many healthy lifestyle choices related to eating behavior.
For example, choosing a balanced diet, a healthy eating plan, controlling daily calorie
intake, and putting emphasis on fruits, vegetables, whole grains and fat-free or low-fat
foods (“Healthy eating for a healthy weight”, n.d.). As a result, items for measuring
healthy eating behavior were created based on the guidelines of the CDC for healthy
eating and healthy weight, which reflects one’s healthy practices in daily meals and food
consumption behavior. In addition, healthy eating was also measured by four items
adapted from Dutta-Bergman (2004b)’s healthy activities items and Kraft and Goodell
(1993)’s Wellness Scale. Sample items included “I increase the consumption of healthy
foods” and “I pay attention to ingredients, nutrients, and calories in foods” with response
options from “never” to “all of the time.” The Cronbach’s α was 0.808.
Physical activity. Physical activity was measured using four items adapted from
Walker, Sechrist, and Pender ’s Healthy People Lifestyle Profile (1987) and Kraft and
Goodell ’s Wellness Scale (1993). Sample items were “I increase exercise during usual
daily activities (such as walking during lunch, using stairs instead of elevators, parking
car away from destination and walking, light yard work, housekeeping, etc.)”, “I take part
in leisure time physical activities (such as swimming, dancing, bicycling, etc.)” and “I
workout more often than I used to” with response options from “never” to “always.” The
Cronbach’s α was 0.900.

69
Table 1. Means and Cronbach’s Alpha in Key Measures
Measures M SD Cronbach’s α
All Respondents (N=521)
Health Consciousness 5.71/7 0.94 0.857
Health Lotus of Control 5.67/7 0.89 0.734
M-Health Literacy 4.98/7 1.71 0.964
Vigilance Coping 3.54/5 0.87 0.748
PIMS 4.28/7 1.69 0.925
Social Norms 2.08/5 1.01 N/A
Normative Beliefs 4.62/7 1.48 0.885
Self-efficacy 5.45/7 1.06 0.869
Autonomous Motivation 5.04/7 1.39 0.933
Self-determination 5.09/7 1.00 0.732
Health Anxiety 2.23/7 0.92 0.866
Social Support 5.37/7 1.20 0.893
Well-being 5.24/7 1.27 0.940
Healthy Eating 3.61/5 0.78 0.808
Physical Activity 3.34/5 0.99 0.900
Satisfaction 4.74/7 1.48 N/A
Goal Attainment 0.62/1 0.48 N/A
Trackers Only (N=392)
Self-tracking Engagement 4.11/6 0.75 0.625

Egocentric Network Measures
Personal Network Influence. Egocentric data was collected using a name
generator. Participants were asked to provide up to six people with whom they talk about
weight concerns or weight management issues the most. Information about the named
“alters” was collected through the survey, including alters’ relations to the ego, the
strength of relationship, frequency of interaction, gender, age, ethnicity, healthiness,
attractiveness, and alters’ mobile self-tracking behaviors. These questions aim to examine
whether a person’s mobile self-tracking behavior is influenced by his or her social ties
that can be characterized in certain way. Firstly, individuals’ personal network exposure
was obtained by calculating the number of alters the participants indicated as mobile self-

70
trackers. Network exposure is defined as the “degree to which a focal individuals’ alters
engage in a particular behavior” (Valente, 2010, p.65). The influence of network
exposure is based on the assumption that individuals in the network observe the alters’
behavior, and then engage in the behavior. Valente (2010) suggests to test social
influence with personal network exposure by asking respondents to indicate whether they
were influenced by their alters. Personal Network Exposure was calculated by the
number of alters one indicated as mobile self-trackers. Strength of tie was obtained
through measuring relationship type (categorical), interaction frequency (ratio), and years
known (ratio). Role modeling was operationalized by two factors, perceived healthiness
and perceived attractiveness of the alter. Similarity was measured between each dyad in
terms of age, gender, and ethnicity.
Analytic Strategies
To investigate differences in demographic variables and personality
characteristics between groups, ANOVA and chi-square tests of independence were used.
Data was checked for normal distribution. Within the mobile self-trackers, bivariate
correlation and linear regression analyses were performed to assess the relationship
between antecedents, predictors, outcome variables, and the mediator. Hierarchical
regression analyses were used to examine the moderator. Multivariate analyses of
covariance (MANCOVA) was conducted to validate effects and differences in effects
between groups with control for covariates. Egocentric data was transformed into dyadic
data for analyses by transforming respondents and their nominators into pairs.
Characteristics are coded for each pair in terms of same tracking behavior, same age,
same gender, and same ethnicity. Logistic regression analyses were used to perform

71
egocentric network analyses. Finally, a structural equation modeling with AMOS was
conducted to confirm the causal relationships and paths in the model. Statistical analyses
were conducted using IBM SPSS Statistics 21 and AMOS 23.

72
CHAPTER FIVE: RESULTS
Recruitment
The questionnaire was developed using Qualtrics online survey software.
Participants were recruited from a paid sampling service, Qualtrics Panel. The researcher
first specified the topic and the targeted population for whom the survey is intended, and
the project manager in Qualtrics sent the survey to a specific Qualtrics Panel. Eligible
participants who finished the survey and whose responses passed the quality checks were
reimbursed through Qualtrics. Participants who were over 18 years old, residing in the
United States, and were interested in or currently undertaking weight management
regimens were eligible to participate in the survey.
This resulted in a total of 521 complete responses for subsequent analyses.
Responses in the study were examined for normal distribution, skewness, kurtosis, and
outliers where necessary. All variables in the study met the assumption of parametric
tests and they were outlier-free.
Demographics
Participants were 521 adults recruited from Qualtrics Panel. Two hundred and
sixty-three respondents were recruited as mobile self-trackers, 129 respondents were
recruited as traditional trackers, and 129 respondents were non-trackers. Males
constituted 21.1 % of the sample (n=110), and 78.5 % were female (n=409). The mean
age was 41.08 (SD= 13.81, ranging from 18 to 70 years old. About 85.6% (n=446) of
participants were White/Caucasian, 5.4% were Asian (n=28), 4.0% were Black/African
American (n=21), and 3.6% were Hispanic/Latino (n=19). Eighty three percent of

73
respondents were smartphone users (n=434), and 16.3% were not smartphone users
(n=85). See Table 2 for detailed information across groups.
Table 2. Demographic Information Across Groups
Whole Sample
(N=521)
Mobile Self-
trackers (n=263)
Traditional Self-
trackers (n=129)
Non-trackers
(n=129)
Age* M= 41.08
(SD= 13.81)
M= 37.06
(SD= 12.09)
M= 46.51
(SD=14.18)
M=43.84
(SD= 14.28)
Gender 21.1% Male
78.5% Female
19.4% Male
80.2% Female
23.3% Male
76.0% Female
19.4% Male
80.2% Female
Income* M=2.53 (SD=
1.22)
M=2.79
(SD=1.19)
M= 2.43
(SD=1.22)
M=2.10
(SD=1.14)
Health
Status*
M= 3.76 (SD=
0.76)
M= 3.84 (SD=
0.74)
M= 3.64
(SD=0.72)
M=3.71 (SD=
0.81)
Chronic*
Disease
75% No
25% Yes
79.1% No
20.9% Yes
65.9% No
34.1% Yes
76% No
24.0 Yes
Ethnicity 85.6% Caucasian
5.4% Asian
3.6% Latino
4.0% African
American
86.3% Caucasian
6.1% Asian
4.2% Latino
2.7% African
American
88.4% Caucasian
4.7% Asian
1.6% Latino
3.9% African
American
81.4% Caucasian
4.7% Asian
4.7% Latino
7.0% African
American
Marital
Status*
23.8% Single
13.6 % In a
relationship
52.2% Married
9.9% Other
17.1% Single
16.7 % In a
relationship
58.2% Married
7.2% Other
29.5% Single
9.3 % In a
relationship
48.1% Married
13.2% Other
31.8% Single
11.6 % In a
relationship
44.2% Married
11.4% Other
Children 52% Yes
47.6% No
55.9% Yes
43.7% No
48.1% Yes
51.9% No
48.1% Yes
51.2% No
Smartphone* 83.3% Yes
16.3 %No
95.8% Yes
3.8% No
72.1% Yes
27.1% No
69.0% Yes
31.0% No
Schooling* 2.1% Less than
high school
15.2% high school
34.9% Some
college
34.9% Bachelor’s
10.4% Master’s
1.7% Ph.D.
0.8% Less than
high school
11.4% high school
32.7% Some
college
38.8% Bachelor’s
12.5% Master’s
2.7% Ph.D.
2.3%Less than
high school
15.5% high school
35.7% Some
college
34.9% Bachelor’s
9.3% Master’s
1.6% Ph.D.
4.7% Less than
high school
22.5% high school
38.8% Some
college
27.1% Bachelor’s
7.0% Master’s
0% Ph.D.
Employee* 50.7 % Employed
8.3% Unemployed
11.7% Retired
6.0% Student
19.0%Homemaker
3.6% Other
58.2 % Employed
4.9%
Unemployed
5.3% Retired
7.2% Student
21.3%
Homemaker
2.7% Other
46.5% Employed
8.5% Unemployed
19.4% Retired
6.2% Student
13.2%
Homemaker
4.7% Other
39.8% Employed
14.8%
Unemployed
17.2% Retired
3.1% Student
20.3%
Homemaker
4.7% Other

74
Goal
Achieved*
62.2% Yes
17.5% No
20.3% Not sure.
70% Yes
12.9% No
17.1% Not sure.
66.7% Yes
18.6% No
14.7% Not sure.
41.9% Yes
25.6% No
32.6% Not sure.
Note: * Statistically significant differences between groups.
Barriers for Non-Trackers
A multiple-choice question was asked among 129 non-trackers as to why they are
not engaging in self-tracking in terms of awareness phases based on the Precaution
Adoption Process Model (Weinstein, Sandman, & Blalock, 2008). The majority of non-
trackers reported the reasons as “I’m undecided about self-tracking” (47%, n= 61),
”never heard about self-tracking” (23.3%, n= 30), and “have tried self-tracking before
and decided not to self-track for health” (9.3%, n=12). About 20% (n= 26) marked
“other”, indicating reasons including “laziness”, “no need”, “too much hassle”, “not
interested in it”, and “want to but just haven’t yet”, etc.

Figure 5. Reasons for not self-tracking among non-trackers (n=129).
To further explore potential barriers to self-tracking for non-trackers, a multiple-
choice question was asked about their perceived barriers to self-tracking. Among 129
non-trackers, 26.7% (n= 47) reported “do not have money to buy tracking devices or
Undecided about self-
tracking
Never heard about
self-tracking
Tried before but quit Other
Series1 61 30 12 26
47%
23.3%
9.3%
20%
0
10
20
30
40
50
60
70
Count
Reasons for Not Self-tracking among Non-Trackers (n=129)

75
apps”, 23.9% (n= 42) reported “I simply think there is no need to do it”, 18% (n= 32)
reported “I do not have time for it”, 14.8 %(N= 26) reported “I do not use a smartphone
or other devices”, 9.1% (n= 16) reported “I don’t like the idea of self-tracking”, and 7.4%
(n= 13) reported other barriers being foot problems that interfere with exercising,
forgetfulness, laziness, or being fretful.
Alternatives used by Non-trackers
Non-trackers were also asked to specify how they manage their weight for health
if not self-tracking. About 35.6% (n= 63) of non-trackers reported that “Do nothing
special”, 19.2% (n= 34) “Go to gym and workout”, 15.3%(n= 27) “Seek support from
family and friends”, 12.4% (n= 22) “Seek help from health professionals”, 7.3% (n=13)
“Purchased weight management programs or equipment”, 5.6% (n= 10) “Seek help from
online health communities”, and 4.5% (n= 8) “Others”.

Figure 6. Alternatives to weight management among non-trackers (n=129).

Do nothing Go to gym
Seek support
from F&F
Seek help
from health
professionals
Purchase
programs or
equipment
Seek help
from online
health
communities
Other
Series1 63 34 27 22 13 10 8
35.6%
19.2%
15.3%
12.4%
7.3%
5.6%
4.5%
0
10
20
30
40
50
60
70
Count
Alternatives for Non-Trackers (n=129)

76
Motivations for Self-Trackers
Previous literature indicates that self-trackers are motivated to self-track for
several reasons. A multiple choice question was asked among self-trackers (n=392) about
what motivated them to perform self-tracking. Results have shown that the majority of
self-trackers (69.6%, n= 273) were motivated by the desire of “optimizing own health”,
52% (n= 206) were motivated “by the pleasure of tracking own health”, 48.5% (n= 190)
were motivated by the desire to “regulate behavior through self-tracking”, 26.5% (n=
104) were motivated “by the pleasure of playing with tracking devices”, 20.4% (n= 80)
were motivated by “seeking alternative ways to understand health beyond the traditional
health system”, and 10.7% (n= 42) were motivated to “be part of an online community
where a common interest is shared”.

Figure 7. Motivational reasons among all self-trackers (mobile & traditional).
Following motivation, self-trackers were asked about what characteristics(s) they
are tracking for. About 75% (n= 294) of self-trackers track for weight, 62.2% (n= 244)
track for diet, foods, and calories, 59.7% (n= 234) track for physical activity, 54.8% (n=
215) track for fitness, 34.4% (n= 135) track for sleep, 12.2% (n= 48) track for menstrual
Optimizing
own health
Pleasure of
self-tracking
Regulating
behavior
Playing with
devices
Understanding
own health
beyond health
system
Being part of
an online
community
Series1 273 206 190 104 80 42
69.6%
52%
48.5%
26.5%
20.4%
10.7%
0
50
100
150
200
250
300
Count
Why Self-trackers Want to Self-track? (n=392)

77
cycle, 11.7% (n= 46) track for mood, and 11.7% (n= 46) track for blood sugar, 9.7%
(n= 38) track for cholesterol, 4.8% (n= 19) track for diabetes markers, and 4.1% (n= 16)
track for other things (e.g., blood pressure, stress, breathing, body temperature, red blood
cell control, etc.).

Figure 8. Tracking themes among all self-trackers (mobile & traditional).
We further conducted an independent sample t-test to compare tracking themes
between mobile self-trackers and traditional self-trackers. Findings revealed that there
were significantly more mobile self-trackers tracked for physical activity (t(390)=3.714,
p= 0.000, M
mobile
= .66, SD=.47; M
traditional
= .47, SD=.50) and fitness (t(390)=6.524, p=
0.000, M
mobile
= .66, SD=.48; M
traditional
= .33, SD=.47). The two groups did not differ in
themes including weight (t(390)=-1.88, p= 0.061), diet and foods (t(390)=0.729, p=
0.466), sleep (t(390)=1.012, p= 0.312), menstrual cycle (t(390)=1.316, p= 0.189), and
diabetes markers (t(390)=-1.655, p= 0.100).
On the other hand, there were significantly more traditional self-trackers who
tracked for mood (t(390)=-2.302, p= 0.022, M
mobile
= .09, SD=.29; M
traditional
= .17,
SD=.38), blood sugar (t(390)=-2.386, p= 0.018, M
mobile
= .09, SD=.28; M
traditional
= .18,
Weight
Diet,Food
s,Calories
Physical
Activity
Fitness Sleep
Menstrua
l cycle
Mood
Blood
sugar
Cholester
ol
Diabetes
markers
Other
Series1 294 244 234 215 135 48 46 46 38 19 16
75%
62.2%
59.7%
54.8%
34.4%
12.2% 11.7% 11.7%
9.7%
4.8% 4.1%
0
50
100
150
200
250
300
350
Count
Tracking Themes for Self-Trackers (n=392)

78
SD=.38), cholesterol (t(390)=-2.134, p= 0.034, M
mobile
= .07, SD=.26; M
traditional
= .15,
SD=.36), and other (t(390)=-2.907, p= 0.004, M
mobile
= .02, SD=.12; M
traditional
= .09,
SD=.29) than mobile self-trackers did.
Main Findings
Differences in Demographics
The first research question concerns the differences in demographics, antecedents,
and personality traits between mobile self-trackers, traditional self-trackers, and non-
trackers. First, multiple one-way ANOVAs were conducted to compare the mean
differences in age, income, education, and healthiness between the three groups. First, it
was found that mobile self-trackers were significantly younger (M= 37.06, SD= 12.09)
than traditional trackers (M= 46.51, SD=14.18) and non-trackers (M=43.84, SD= 14.28)
(F(2, 518) = 26.00, p < 0.000). Second, mobile self-trackers had significantly higher self-
reported healthiness (M= 3.84, SD= 0.74) than traditional trackers (M= 3.64, SD=0.72),
and marginally higher healthiness than non-trackers (M=3.71, SD= 0.81) (F(2, 518) =
3.58, p = 0.028). Third, mobile self-trackers had significantly higher annual household
income (M= 2.79, SD= 1.19) than traditional trackers (M= 2.43, SD=1.22) and non-
trackers (M=2.1, SD= 1.14) (F(2, 516) = 15.09, p < 0.000). In addition, mobile self-
tracker were found to have significantly higher education (M= 3.60, SD= 0.97) than
traditional trackers (M= 3.38, SD=0.99) and non-trackers (M=3.09, SD= 0.98) (F(2, 514)
= 11.56, p < 0.000).
Furthermore, a series of chi-square tests of independence were performed to
examine the association between group membership and demographics. First, it was

79
found that group membership was associated with marital status, Χ
2
(8, N=519)= 27.83, p
= 0.001. Mobile trackers were more likely to be married or in a relationship, while
traditional trackers and non-trackers were more likely to be single or separated/divorced.
Second, group membership was associated with smartphone usage, Χ
2
(2, N=519)=
61.58, p < 0.000. Mobile self-trackers were more likely to be people who are current
smartphones users. In addition, group membership was associated with employee status,
Χ
2
(10, N=517)= 42.39, p < 0.000. Mobile self-trackers were more likely to be employed
people, students, or housewives, while traditional trackers and non-trackers were more
likely to be unemployed or retired. Finally, group membership was associated with
chronic disease, Χ
2
(2, N=521)= 8.13, p =0.017. Mobile self-trackers were more likely to
be healthier people without a diagnosed chronic disease (79%) than traditional self-
trackers (65.9%) and non-trackers (76%).
There was no relation between type of trackers and gender (Χ
2
(2, N=519)= 0.98,
p = 0.61), having children (Χ
2
(2, N=519)= 3.21, p = 0.20), ethnicity (Χ
2
(8, N=519)=
10.07, p = 0.26), and phase of tracking (Χ
2
(1, N=392)= 0.05, p = 0.82).
To sum up, demographically speaking, mobile self-trackers appeared to be
younger, wealthier, and healthier people with greater educational attainment. They were
also more likely to be people who are smartphone users, in a committed relationship, and
currently employed compared to traditional trackers and non-trackers.
Differences in Antecedents and Predictors
The second part of the research question concerns the differences in antecedents
and personality traits between mobile self-trackers, traditional self-trackers, and non-
trackers. In order to answer this question, a one-way between subjects ANOVA was

80
conducted to compare the differences in antecedents and underlying predictors proposed
in the present study. Results have shown that all the examined antecedents and predictors
significantly differed between groups. A main effect of group membership was found for
social media frequency, F(2, 517) = 12.55, p < .000; social media activeness, F(2, 518) =
14.70, p < .000; health consciousness, , F(2, 518) = 6.68, p= .001; health locus of control,
F(2, 518) = 6.86, p= .001; mHealth literacy, F(2, 518) = 72.28, p< .000; vigilance
coping, F(2, 518) = 7.81, p< .000; PIMS, F(2, 518) = 52.171, p< .000; social norms, F(2,
518) = 83.73, p< .000; normative beliefs, F(2, 518) = 79.91, p< .000; and self-efficacy,
F(2, 518) = 21.90, p< .000.
Post Hoc tests for sources of differences have shown that, consistent with
expectations, mobile self-trackers exhibited higher social media use, social media
activeness, mHealth literacy, vigilance coping, PIMS, social norms, and normative
beliefs compared with traditional trackers and non-trackers. When contrasting mobile
self-trackers with only traditional self-trackers, it was found that mobile self-trackers had
significantly higher scores on all predictors except health consciousness, health locus of
control, and self-efficacy than traditional self-trackers. On the other hand, non-trackers,
compared to mobile self-trackers, were significantly lower on all underlying predictors.
See Table 4 for details.

81
Table 3. Means for Antecedents and Predictors (For All Participants)
Measures (N=521) M SD Cronbach’s α
Social Media Frequency 3.67/5 1.33 NA
Social Media Activeness 3.57/5 1.32 NA
Health Consciousness 5.71/7 0.94 0.857
Health Lotus of Control 5.67/7 0.89 0.734
M-Health Literacy 4.98/7 1.71 0.964
Vigilance Coping 3.54/5 0.87 0.748
PIMS 4.28/7 1.69 0.925
Social Norms 2.08/5 1.01 NA
Normative Beliefs 4.62/5 1.48 0.885
Self-efficacy 5.45/7 1.06 0.869

Table 4. Mean Differences in Antecedents & Predictors Between Groups with One-way
ANOVA
Measures (N=521) Mobile Traditional Non-Trackers Sig.
M SD M SD M SD p
Social Media
Frequency
b***,c***

3.95 1.18 3.37 1.38 3.40 1.44 .000
Social Media
Activeness
b***,c***

3.87 1.19 3.29 1.33 3.22 1.41 .000
Health Consciousness
c**,d**
5.77 0.93 5.85 0.77 5.46 1.07 .001
Health Lotus of Control
c***,d*
5.77 0.84 5.71 0.79 5.42 1.04 .001
M-Health Literacy
b***,c***
5.76 1.16 4.34 1.73 4.02 1.87 .000
Vigilance Coping
b*,c***
3.68 0.82 3.46 0.87 3.33 0.93 .000
PIMS
b***,c***
4.96 1.49 3.73 1.60 3.45 1.62 .000
Social Norms
a***
2.56 0.99 1.78 0.84 1.41 0.63 .000
Normative Beliefs
a***
5.30 1.12 4.26 1.36 3.62 1.54 .000
Self-efficacy
c***,d***
5.67 .094 5.51 1.01 4.95 1.17 .000
Trackers Only
Self-tracking Duration
b***
7.80 3.21 9.28 3.56 / / .000
Self-tracking Engagement
b**
4.19 0.75 3.94 0.73 / / .002
Note: Significance level: *= p<0.05; **= p<0.01; ***p<0.001
a= Significant between three groups.
b= Significant between mobile self-trackers and traditional self-trackers.
c= Significant between mobile self-trackers and non-trackers.
d= Significant between traditional self-trackers and non-trackers.

82
Prediction of Motivation and Engagement
The second research question concerns the relationship of autonomous motivation
to mobile self-tracking as well as the prediction of mobile self-tracking engagement.

Figure 9. Model for mobile self-tracking engagement.
Research question two examines how theoretically driven personal and social
factors influence autonomous motivation, and in turn, how motivation predicts mobile
self-tracking engagement. Guided by the theory of planned behavior and the UTAUT
model, individuals’ personality traits and propensities will affect their motivation to
engage in a certain behavior. Therefore, it was proposed that mobile trackers’ health-
related and technological propensities will affect their autonomous motivation to mobile
self-tracking. A Pearson correlation test revealed that autonomous motivation was
significantly associated with all of the proposed determinants. Following this, a multiple
regression analysis was performed to examine H
1
, which proposes that autonomous

83
motivation to mobile self-tracking was significantly and positively predicted by H
1a
)
health consciousness, H
1b
) health locus of control, H
1c
) mHealth literacy, H
1d
) vigilance
coping, H
1e
) personal innovativeness for mobile service, H
1f
) self-efficacy, H
1g
) social
norms, and H
1h
) normative belief.
Results indicated that the model was significant (R
2
=0.48, F (8, 262)=28.9,
p<0.000). The eight predictors explained 48.3% of the variance. Among them, vigilance
coping (β =0.21, p<0.000), normative beliefs (β =0.29, p<0.000), and self-efficacy (β
=0.24, p<0.000) significantly predicted individuals’ autonomous motivation to mobile
self-tracking. H
4d
(vigilance coping), H
4f
(self-efficacy), and H
4h
(normative belief) were
supported. Health locus of control was a marginally significant predictor (β =0.11,
p=0.06), while health consciousness (p=0.25), mHealth literacy (p=0.70), PIMS (p=0.69),
and social norms (p=0.72) were not significant predictors of autonomous motivation. H
4a

(health consciousness), H
4b
(health locus of control), H
4c
(mHealth literacy), H
4e
(PIMS),
and H
4g
(social norms) were rejected.
The second hypothesis proposes that autonomous motivation significantly
predicts mobile self-tracking engagement. A single regression analysis was performed to
examine this. Results have shown that the model is significant (R
2
=0.21, F(1,
262)=69.73, p<0.000), and autonomous motivation strongly predicted mobile self-
tracking behavior (β =0.46, p<0.000) and explained 21% of the variance of mobile self-
tracking engagement (See Figure 10).

84

Figure 10. Results of regression analyses for mobile self-tracking engagement.

Health Outcomes of Mobile Self-tracking
The third research question concerns the effects of mobile self-tracking as well as
the differences in health outcomes between groups.

Figure 11. Model for group comparison in health outcomes.

85
Health Outcomes Comparison Between Groups
To explore the effects of self-guided mobile self-tracking for health, and the
differences in health outcomes across groups, the following hypotheses were tested. H
3

states that mobile self-trackers would have significantly higher H
3a
) health anxiety H
3b
)
social support, H
3c
) healthy eating H
3d
) physical activity H
3e
) well-being, H
3f
)
satisfaction, and H
6g
) goal attainment than traditional self-trackers and non-trackers after
controlling for the covariates.
A multivariate analysis of covariance (MANCOVA) was performed to assess for
statistical differences on multiple continuous dependent variables by an independent
grouping variable (i.e., tracker category) while controlling for covariates which cannot be
randomly assigned to each group (Gill, 2001; Hair, Black, Babin, Anderson, & Tatham,
2006). By doing so, the analyses reduce error terms and “eliminate covariates’ effects on
the relationship between the independent grouping variable and the continuous dependent
variables” (“Multivariate Analysis of Covariance”, n.d.). Further, because the seven
outcome variables are conceptually related, it is appropriate to conduct one multivariate
significance test to account for the full variability in all dependent variables combined
and thus to have greater power.
In other words, adding covariates to the research model helps control for the
concomitant effects of covariates on the relationship between the grouping variable and
the outcome variables and hence reduces error terms (Huberty & Morris, 1989). In the
present study, participants’ self-reported health status, chronic disease, age, income, and
education were treated as covariates as they may have confounding effects on the studied
health outcomes. Therefore, health status, chronic disease, age, income and education

86
were controlled in the model to more accurately assess the effect of group membership on
health outcomes.
Prior Tests for Assumptions and Restrictions
Several diagnostic tests were run to ensure that the data does not violate important
assumptions of MANCOVA.
Pearson correlation tests. Firstly, we checked for the correlations between
covariates and dependent variables. One of the key assumptions that need to be addressed
before employing MANCOVA is reasonable correlation between the covariates and the
dependent variables (Mayers, 2013). A covariate must be significantly related to the
outcome in order to reduce error variance. Covariates that are not correlated with the
dependent variables can be excluded. In the present study, results of Pearson correlation
tests have showed that the correlation between the covariates (health status, chronic
disease, education, income, and age) and the dependent variables were significant and
reasonable, indicating that we have satisfied that requirement for the proposed covariates
(see Table 5). On the other hand, marital status, having children, employee status, and
smart phone use were not significantly associated with studied health outcomes, hence
they were excluded from the covariates list.

87
Table 5. Pearson Correlation Tests Between Covariates and Health Outcomes
Covariates HA SS Eating PA Well-being Satisfaction Goal
Attainment
Chronic
Disease
(N=521)
r .117
**
.001 .064 -.061 -.117
**
-.137
**
-.145
**

p .007 .985 .142 .166 .007 .002 .001
Health
Status
(N=521)
r -
.234
**

.218
**
.116
**
.283
**
.288
**
.264
**
.220
**

p .000 .000 .008 .000 .000 .000 .000
Edu
(N=517)
r -.014 .034 .151
**
.167
**
.126
**
.049 .029
p .754 .442 .001 .000 .004 .269 .507
Income
(N=519)
r -.030 .086 .097
*
.173
**
.144
**
.117
**
.017
p .497 .051 .027 .000 .001 .007 .698
Age
(N=521)
r -
.140
**

.083 .010 -.107
*
.011 -.066 -.059
p .001 .058 .828 .015 .796 .132 .179

Secondly, testing for independence of covariates. Covariates that are related to the
dependent variables were checked whether they are independent in respect of groups. A
covariate must not differ significantly in respect of the independent grouping variable.
Covariates that do not differ significantly across groups can help reduce error variance in
the model. Any covariate that is dependent on the grouping effect should be examined
separately to explore the extent of mediating or moderating effect that confounds the
main outcomes (Mayers, 2013). As age, health status, chronic disease, income, and
education all significantly differed between groups (λ = .824, F (2, 512) = 10.35, p <
.000, See Table 6), these covariates may not necessarily help reduce error variance.
Nonetheless, scholars suggest that we should not drop a covariate from the model when
this assumption is violated (Grace-Martin, n.d.). Instead, we should retain these
covariates in the model because by doing so we can have a more accurate estimate of the

88
real relationship between the categorical independent variable and the outcome
2
, because
the mean differences in outcomes are compared at any given value of the covariate. For
example, when age is added in the model as a covariate, the differences in the mean of
health outcomes for mobile self-trackers, traditional self-trackers, and non-trackers are
estimated for participants at the same age. If we drop the covariate, the difference in the
mean is estimated at the overall mean for each group. Similarly, when chronic disease is
included in the model as a covariate, the differences in the mean of health outcomes are
estimated for participants reporting having the same condition (i.e., yes or no). And when
health status is included in the model as a covariate, the differences in the mean of health
outcomes for the three groups are estimated for participants having the same health status
scores. The comparison would be more accurately reflecting the actual differences after
controlling for the covariates.
Table 6. Test for Mean Differences in Covariates Between Groups
Univariate Tests
Dependent Variable Sum of
Squares
df Mean
Square
F Sig.
Age Contrast 9010.337 2 4505.168 25.693 .000
Error 89776.545 512 175.345
Chronic
Disease
Contrast 1.497 2 .748 4.005 .019
Error 95.688 512 .187
Health
Status
Contrast 3.328 2 1.664 3.018 .050
Error 282.353 512 .551
Income Contrast 42.495 2 21.248 15.030 .000
Error 723.788 512 1.414
Edu Contrast 22.049 2 11.025 11.535 .000
Error 489.357 512 .956

Testing for normal distribution. The third step is to check for normal distribution
using skewness and kurtosis as all the measures in the study were using Likert scale. Data

2
When assumptions of ANCOVA are irrelevant. Retrieved from
http://www.theanalysisfactor.com/assumptions-of-ancova/

89
collected through a Likert scale is usually not normally distributed as it has ceiling and
floor effects on the scale (Clason & Dormody, 1994), hence the Shapiro-Wilk analyses
might not provide accurate assessment. The values for skewness and kurtosis between -2
and +2 are considered acceptable in order to satisfy the assumption of normal distribution
(George & Mallery, 2010). Results have shown that the skewness and kurtosis for each of
the covariates and health outcomes are within ±2, suggesting that all variables have met
the assumption of normal distribution. We proceeded to the next test.
Testing for homogeneity of regression slopes. We further checked if the
correlation between each covariate and the dependent variable differ across independent
variable groups. To satisfy the assumption, the interaction term of each covariate and the
grouping factor has to be nonsignificant. Health status, age, chronic disease, income, and
education were examined separately as they were shown to be dependent on the grouping
factor. Results have shown the assumption is satisfied because all the interaction terms
were nonsignificant (p>0.05), indicating the regression slopes on health outcomes were
homogenous between groups (See Table 7 ~ Table 11).
Table 7. Homogeneity of Regression Slopes for Health Status and Group Membership

TCategory *
Health
Status
DVs Type III
sum of
squares
df Mean
Square
F Sig.
HA .022 2 .011 .014 .986
SS .626 2 .313 .226 .798
Eating .045 2 .022 .039 .962
PA .844 2 .422 .493 .611
Well-being 6.919 2 3.460 2.420 .090
Satisfaction .563 2 .281 .145 .865
Goal
Attainment
.039 2 .019 .090 .914

90
Table 8. Homogeneity of regression slopes for age and group membership

TCategory
* Age
DVs Type III
sum of
squares
df Mean
Square
F Sig.
HA 1.796 2 .898 1.085 .339
SS .428 2 .214 .150 .861
Eating 5.614 2 2.807 4.949 .107
PA 3.377 2 1.688 1.841 .160
Well-being .289 2 .144 .093 .911
Satisfaction 2.120 2 1.060 .512 .599
Goal
Attainment
.418 2 .209 .938 .392

Table 9. Homogeneity of regression slopes for chronic disease and group membership

TCategory
* Chronic
Disease
DVs Type
III sum
of
squares
df Mean
Square
F Sig.
HA 2.628 2 1.314 1.568 .209
SS .841 2 .421 .290 .748
Eating .905 2 .453 .787 .456
PA .823 2 .412 .446 .640
Well-being 2.323 2 1.162 .755 .471
Satisfaction 2.108 2 1.054 .519 .595
Goal
Attainment
.048 2 .024 .109 .897

Table 10. Homogeneity of regression slopes for income and group membership

TCategory
* Income
DVs Type
III sum
of
squares
df Mean
Square
F Sig.
HA .236 2 .118 .140 .870
SS 2.509 2 1.254 .890 .411
Eating 1.601 2 .801 1.394 .249
PA 5.137 2 2.569 2.834 .060
Well-being 3.285 2 1.643 1.074 .342
Satisfaction 3.285 2 1.642 .806 .447
Goal
Attainment
.994 2 .497 2.238 .108

91
Table 11. Homogeneity of regression slopes for education and group membership

TCategory
*
Education
DVs Type III
sum of
squares
df Mean
Square
F Sig.
HA 1.029 2 .514 .615 .541
SS 1.347 2 .673 .464 .629
Eating 1.455 2 .727 1.274 .281
PA 4.486 2 2.243 2.475 .085
Well-being .219 2 .109 .070 .932
Satisfaction 5.963 2 2.982 1.443 .237
Goal Attainment .127 2 .063 .282 .754

As shown above, there was no significant interaction associated with the
covariates and health outcomes across independent variable groups. The requirement for
homogeneity of regression slopes was satisfied. All the assumptions were met for
multivariate analysis of covariance. Therefore, we proceed to interpret the main effects
prior to inclusion of covariates.
Main Effects of MANCOVA Prior to Covariate Adjustment
We firstly explored the main effects for all health outcomes (i.e., health anxiety,
social support, healthy eating, physical activity, well-being, satisfaction, and goal
attainment) in respect of tracking groups, prior to controlling for covariates. Based on
Wilk’s Lambda, we have a significant multivariate outcome prior to covariate adjustment
in respect of health anxiety, social support, healthy eating, physical activity, well-being,
satisfaction, and goal attainment across three groups, λ = .862, F (2, 518) = 5.634, p <
.000 (See Table 12).
Table 11 shows that there were significant univariate outcomes for healthy eating
(F (2, 518)= 16.522, p < .000), physical activity (F (2, 518)= 19.535, p < .000), well-
being (F (2, 518)= 11.558, p < .000), satisfaction (F (2, 518)= 15.405, p < .000) and goal

92
attainment (F (2, 518)= 16.121, p < .000) across group status, while there were no
significant outcomes for health anxiety (p = 0.275) and social support (p = .081).
Specifically, the three groups differed in healthy eating, physical activity, well-being,
satisfaction and goal attainment prior to the inclusion of covariates. Table 15 shows the
mean differences in each health outcome across groups.
Table 12. Multivariate Outcome Prior to Covariates Adjustment
Effect Value F Hypothesis df Error df Sig.
TCategory Pillai's Trace .141 5.559 14.000 1026.000 .000
Wilks' Lambda .862 5.634
b
14.000 1024.000 .000
Hotelling's Trace .156 5.708 14.000 1022.000 .000
Roy's Largest Root .128 9.412
c
7.000 513.000 .000

Table 13. Univariate Outcomes Prior to Covariates Adjustment

TCategory
DVs Type III
sum of
squares
df Mean
Square
F Sig.
HA 2.198 2 1.099 1.295 .275
SS 7.294 2 3.647 2.529 .081
Eating 19.041 2 9.521 16.522 .000
PA 35.969 2 17.985 19.535 .000
Well-being 35.845 2 17.923 11.558 .000
Satisfaction 63.520 2 31.760 15.405 .000
Goal
Attainment
7.179 2 3.589 16.121 .000

93
Table 14. Main Effects on Health Outcomes Prior to Covariates Adjustment
Health Outcomes Group N Mean SE
Health Anxiety Mobile self-trackers 263 2.19 .06
Traditional self-trackers 129 2.34 .08
Non-trackers 129 2.19 .08
Social Support Mobile self-trackers 263 5.48 .07
Traditional self-trackers 129 5.35 .11
Non-trackers 129 5.19 .11
Healthy
Eating***
Mobile self-trackers 263 3.74 .05
Traditional self-trackers 129 3.67 .07
Non-trackers 129 3.28 .07
Physical
Activity***
Mobile self-trackers 263 3.57 .06
Traditional self-trackers 129 3.28 .08
Non-trackers 129 2.93 .08
Well-being*** Mobile self-trackers 263 5.50 .08
Traditional self-trackers 129 5.03 .11
Non-trackers 129 4.93 .11
Satisfaction*** Mobile self-trackers 263 5.01 .09
Traditional self-trackers 129 4.75 .13
Non-trackers 129 4.16 .13
Goal
Attainment***
Mobile self-trackers 263 .70 .03
Traditional self-trackers 129 .67 .04
Non-trackers 129 .42 .04
Note: *p<0.05; **p<001; ***p<0.001

Multivariate Effects After Including Covariates
Next, a MANCOVA was performed using group membership as the independent
grouping variable and the health outcomes as dependent variables while controlling for
the aforementioned covariates including age, health status, chronic disease, income, and
education.
The Levene’s tests for homogeneity of variance are diagnostic tests of equal
variance in dependent variables across groups. Results have shown that between-group
variances across the three groups in respect of outcomes were not significant for health
anxiety (F (2, 512) = 1.754, p = 0.174), social support (F (2, 512) = .701, p = .496), and

94
healthy eating (F (2, 512) = 1.331, p = .265), and well-being (F (2, 512) = 1.372, p =
.255), while there were significant differences for physical activity (F (2, 512) = 3.475, p
= .032), satisfaction (F (2, 512) = 3.791, p = .023), and goal attainment (F (2, 512) =
4.486, p = .012). Nonetheless, analysis of covariance is robust to violations of the
assumption of homogeneity of variances when the ratio of the largest to smallest group
variance is less than 3.0 (“Solving Analysis of Covariance Problem”, n.d.). The ratio in
the current study was less than 3.0. Therefore, the assumption of equal variances was
satisfied (See Table 15).
Table 15. Levene’s Test of Equality of Error Variances in Health Outcomes Across
Groups
Levene’s Test of Equality of Error Variances
F df1 df2 Sig.
HA 1.754 2 512 .174
SS 1.701 2 512 .496
Eating 1.331 2 512 .265
PA 3.475 2 512 .032
Well-being 1.372 2 512 .255
Satisfaction 3.791 2 512 .023
Goal Attainment 4.486 2 512 .01

Main Results After Controlling for Covariates
The multivariate analysis of covariance after controlling for covariates yields
results showing that the multivariate outcome is much stronger subsequent to applying
the covariates compared with the outcome prior to inclusion of the covariates. It appears
that the covariates have reduced some of the error variance. There is a significant
multivariate effect (main effect) across three groups for the combined dependent
variables of health anxiety, social support, healthy eating, physical activity, well-being,
satisfaction, and goal attainment, λ = .880, F (2, 517) = 4.706, p < .000 (See Table 17).

95
Although the multivariate effect was significant, a large value of Wilks’ λ indicates that
the groups may have smaller differences (Patel & Bhavsar, 2013).
The main effects for healthy eating (F (2, 507) = 13.420, p < .000), physical
activity (F (2, 507) = 11.351, p < .000), well-being (F (2, 507) = 7.834, p < .000,
satisfaction (F (2, 507) = 12.951, p < .000), and goal attainment (F (2, 507) = 17.040, p <
.000) by group membership were significant, while the main effects for health anxiety (F
(2, 507) = 2.499, p = .083) and social support (F (2, 507) = 2.131, p = .120) by group
membership were not significant (See Table 14). We further use the Bonferroni pairwise
comparisons to identify the sources of differences (See Table 15).
According to the Post Hoc (Bonferroni) outcomes, it was shown that most of the
differences in outcomes exist between mobile self-trackers and non-trackers, as well as
between traditional trackers and non-trackers. Mobile self-trackers (M=2.14, SE=.06) had
marginally lower health anxiety than traditional trackers(M=2.37, SE=.08, p=0.07).
Mobile self-trackers also had significantly higher well-being (M=5.46, SE=.08) than
traditional self-trackers(M=5.05, SE=.11, p=0.01). Bonferroni comparisons also show
that mobile self-trackers had significantly higher healthy eating (M=3.73, SE=.05,
p<0.000), physical activity (M=3.50, SE=.06, p<0.000), well-being (M=4.97, SE=.11,
p=0.001), satisfaction (M=4.97, SE=.11, p<0.000), and goal attainment (M=0.692,
SE=.03, p<0.000) than non-trackers (See Table 18 for details). Table 19 shows the
adjusted mean for health outcomes in respect of groups.

96
Table 16. Results of Multivariate Analyses after Covariates Adjustment
Effect Value F Hypothesis
df
Error df Sig.

Intercept
Pillai's Trace .384 44.565
b
7.000 501.000 .000
Wilks' λ .616 44.565
b
7.000 501.000 .000
Hotelling's Trace .623 44.565
b
7.000 501.000 .000
Roy's Largest Root .623 44.565
b
7.000 501.000 .000

Tracker
Category
Pillai's Trace .122 4.657 14.000 1004.000 .000
Wilks' λ .880 4.706
b
14.000 1002.000 .000
Hotelling's Trace .133 4.756 14.000 1000.000 .000
Roy's Largest Root .108 7.778
c
7.000 502.000 .000
Age Pillai's Trace .053 4.035
b
7.000 501.000 .000
Wilks' λ .947 4.035
b
7.000 501.000 .000
Hotelling's Trace .056 4.035
b
7.000 501.000 .000
Roy's Largest Root .056 4.035
b
7.000 501.000 .000
Chronic Pillai's Trace .033 2.450
b
7.000 501.000 .018
Wilks' λ .967 2.450
b
7.000 501.000 .018
Hotelling's Trace .034 2.450
b
7.000 501.000 .018
Roy's Largest Root .034 2.450
b
7.000 501.000 .018
Health
Status
Pillai's Trace .135 11.124
b
7.000 501.000 .000
Wilks' λ .865 11.124
b
7.000 501.000 .000
Hotelling's Trace .155 11.124
b
7.000 501.000 .000
Roy's Largest Root .155 11.124
b
7.000 501.000 .000
Education Pillai's Trace .020 1.455
b
7.000 501.000 .181
Wilks' λ .980 1.455
b
7.000 501.000 .181
Hotelling's Trace .020 1.455
b
7.000 501.000 .181
Roy's Largest Root .020 1.455
b
7.000 501.000 .181
Income Pillai's Trace .021 1.500
b
7.000 501.000 .165
Wilks' λ .979 1.500
b
7.000 501.000 .165
Hotelling's Trace .021 1.500
b
7.000 501.000 .165
Roy's Largest Root .021 1.500
b
7.000 501.000 .165

97
Table 17. Univariate Analyses After Applying Covariates: Main Effects on Health
Outcomes
Tests of Between-Subjects Effects
Source Dependent
Variable
Type III
Sum of
Squares
df Mean
Square
F Sig.
Corrected
Model
HA 38.927
a
7 5.561 7.292 .000
SS 40.057
b
7 5.722 4.204 .000
Eating 29.542
c
7 4.220 7.514 .000
PA 77.071
d
7 11.010 12.928 .000
Well-being 89.982
e
7 12.855 8.856 .000
Satisfaction 133.875
f
7 19.125 9.952 .000
Goal Attainment 13.713
g
7 1.959 9.223 .000
Intercept HA 93.544 1 93.544 122.659 .000
SS 114.821 1 114.821 84.355 .000
Eating 52.106 1 52.106 92.767 .000
PA 25.105 1 25.105 29.478 .000
Well-being 95.201 1 95.201 65.585 .000
Satisfaction 105.035 1 105.035 54.658 .000
Goal Attainment 1.253 1 1.253 5.899 .015
Tracker
Category
HA 3.811 2 1.905 2.499 .083
SS 5.802 2 2.901 2.131 .120
Eating 15.075 2 7.538 13.420 .000
PA 19.335 2 9.667 11.351 .000
Well-being 22.744 2 11.372 7.834 .000
Satisfaction 49.775 2 24.888 12.951 .000
Goal Attainment 7.239 2 3.619 17.040 .000
Age HA 11.323 1 11.323 14.847 .000
SS 4.339 1 4.339 3.188 .075
Eating .156 1 .156 .278 .598
PA 3.723 1 3.723 4.372 .037
Well-being 2.358 1 2.358 1.625 .203
Satisfaction 1.297 1 1.297 .675 .412
Goal Attainment .028 1 .028 .131 .717
Chronic HA 2.145 1 2.145 2.812 .094
SS 2.371 1 2.371 1.742 .187
Eating 3.690 1 3.690 6.570 .011
PA 1.961 1 1.961 2.303 .130
Well-being .441 1 .441 .304 .582
Satisfaction 2.502 1 2.502 1.302 .254
Goal Attainment .746 1 .746 3.513 .061
Health
Status
HA 15.572 1 15.572 20.418 .000
SS 24.036 1 24.036 17.658 .000
Eating 4.402 1 4.402 7.838 .005
PA 29.897 1 29.897 35.105 .000
Well-being 34.744 1 34.744 23.935 .000
Satisfaction 49.384 1 49.384 25.698 .000
Goal Attainment 4.083 1 4.083 19.222 .000

98
Education HA .130 1 .130 .170 .680
SS .626 1 .626 .460 .498
Eating 2.179 1 2.179 3.879 .049
PA 1.197 1 1.197 1.406 .236
Well-being .985 1 .985 .679 .410
Satisfaction 2.897 1 2.897 1.507 .220
Goal Attainment .117 1 .117 .548 .459
Income HA 1.134 1 1.134 1.487 .223
SS .289 1 .289 .213 .645
Eating .015 1 .015 .026 .871
PA .830 1 .830 .975 .324
Well-being .205 1 .205 .141 .707
Satisfaction .243 1 .243 .126 .722
Goal Attainment .815 1 .815 3.838 .051
Error HA 386.654 507 .763
SS 690.110 507 1.361
Eating 284.775 507 .562
PA 431.784 507 .852
Well-being 735.948 507 1.452
Satisfaction 974.292 507 1.922
Goal Attainment 107.693 507 .212
Total HA 2962.375 515
SS 15626.313 515
Eating 7003.125 515
PA 6248.313 515
Well-being 14957.688 515
Satisfaction 12697.000 515
Goal Attainment 319.000 515
Corrected
Total
HA 425.581 514
SS 730.167 514
Eating 314.317 514
PA 508.855 514
Well-being 825.930 514
Satisfaction 1108.167 514
Goal Attainment 121.406 514
a. R Squared = .091 (Adjusted R Squared = .079)
b. R Squared = .055 (Adjusted R Squared = .042)
c. R Squared = .094 (Adjusted R Squared = .081)
d. R Squared = .151 (Adjusted R Squared = .140)
e. R Squared = .109 (Adjusted R Squared = .097)
f. R Squared = .121 (Adjusted R Squared = .109)
g. R Squared = .113 (Adjusted R Squared = .101)

99
Table 18. Post Hoc Analyses with Bonferroni Method
Pairwise Comparisons
Dependent
Variable
(I)
TCategory
(J) TCategory Mean
Differenc
e (I-J)
SE Sig.
b

HA Mobile Traditional -.223 .100 .078
Non-trackers -.081 .101 1.000
Traditional Mobile .223 .100 .078
Non- trackers .143 .111 .595
Non-T Mobile .081 .101 1.000
Traditional -.143 .111 .595
SS Mobile Traditional .145 .134 .831
Non-trackers .274 .134 .126
Traditional Mobile -.145 .134 .831
Non- trackers .128 .148 1.000
Non-T Mobile -.274 .134 .126
Traditional -.128 .148 1.000
Eating Mobile Traditional .063 .086 1.000
Non- trackers .432
*
.086 .000
Traditional Mobile -.063 .086 1.000
Non- trackers .369
*
.095 .000
Non-T Mobile -.432
*
.086 .000
Traditional -.369
*
.095 .000
PA Mobile Traditional .145 .106 .512
Non-trackers .503
*
.106 .000
Traditional Mobile -.145 .106 .512
Non- trackers .358
*
.117 .007
Non-T Mobile -.503
*
.106 .000
Traditional -.358
*
.117 .007
Well-being Mobile Traditional .409
*
.138 .010
Non- trackers .495
*
.139 .001
Traditional Mobile -.409
*
.138 .010
Non- trackers .086 .153 1.000
Non-T Mobile -.495
*
.139 .001
Traditional -.086 .153 1.000
Satisfaction Mobile Traditional .129 .159 1.000
Non- trackers .789
*
.160 .000
Traditional Mobile -.129 .159 1.000
Non- trackers .660
*
.176 .001
Non-T Mobile -.789
*
.160 .000
Traditional -.660
*
.176 .001
Goal
Attainment
Mobile Traditional .004 .053 1.000
Non- trackers .287
*
.053 .000
Traditional Mobile -.004 .053 1.000
Non- trackers .283
*
.058 .000
Non- T Mobile -.287
*
.053 .000
Traditional -.283
*
.058 .000

100
Table 19. Estimated Marginal Means Adjusted for the Effect of the Covariates
*significant results shown in the multivariate analysis after MANCOVA adjustment.
To sum up, MANCOVA analyses subsequent to controlling for age, health status,
chronic disease, income, and education showed a significant multivariate effect for
healthy eating, physical activity, well-being, satisfaction, and goal attainment across
tracking groups (λ = .880, F (2, 517) = 4.706, p < .000). Initial results without controlling
for covariates have showed that mobile self-trackers reported higher healthy eating,
physical activity, well-being, satisfaction, and goal attainment than traditional self-
trackers and non-trackers (λ = .862, F (2, 518) = 5.634, p < .000). These were confirmed
across dependent variables including healthy eating (F (2, 518)=16.522), physical activity
(F (2, 518)=19.353), well-being (F (2, 518)=11.558), satisfaction (F (2, 518)=15.405),
and goal attainment (F (2, 518)=16.121). When age, chronic disease, health status,
income, and education were added as covariates in the model, theses effects remain
significant (λ = .880, F (2, 517) = 4.706, p < .000). Post hoc (Bonferroni) analyses of the
univariate outcomes (adjusted for age, chronic disease, health status, income and
education) revealed that mobile self-trackers had significantly greater well-being than
traditional self-trackers (p=0.01) and non-trackers (p=0.001), and had marginally lower
health anxiety than traditional self-trackers (p=0.07). H
3
was partially supported.

Dependent Variables
HA SS Eating* PA* Well-being* Satisfaction* Gaol*
Group N mean SE mean SE mean SE mean SE mean SE mean SE mean SE
Mobile 263 2.14 .06 5.48 .08 3.73 .05 3.50 .06 5.46 .08 4.97 .09 .692 .03
Traditional 129 2.37 .08 5.34 .11 3.67 .07 3.35 .08 5.05 .11 4.84 .13 .688 .04
Non-T 129 2.22 .08 5.21 .11 3.30 .07 3.00 .08 4.97 .11 4.18 .13 .405 .04

101
Interpretation of Covariates
Table 16 shows the multivariate effects for age (p<0.000), chronic disease (p=
0.018), and health status (p < .000), indicating significant covariate effects on health
outcomes. Education (p = 0.181), and income (p = 0.165) were not significant covariates.
In the present study, covariates were added to control for variables that rival the
independent variable of interest and are used to explain a portion of the variance. These
covariates play the roles of unique relationships to the dependent variables whether they
are significant or not. The covariate relationship may not be a predictor of interest, and
results have revealed that these covariates altered the strength of the original outcomes by
slightly reducing the error variance and illustrating the extent to which they confounded
the original health outcomes.
Among Mobile Self-trackers
Interestingly, when we further dichotomized mobile self-trackers into two
categories⎯ self-trackers with mobile phones vs. self-trackers with wearables ⎯ the
differences in predictors and health outcomes had revealed interesting findings. First of
all, mobile self-trackers using wearables (e.g., smart watches, etc.) had significantly
higher social norms (M=2.72, SD=1.00) than mobile self-trackers with apps on mobile
phones (M=2.39, SD=0.90) (t=-2.734, p=0.007). That implies that action-taking on
purchasing a wearable for self-tracking purposes was heavily influenced by individuals’
perception of whether their significant social network ties are engaging in mobile self-
tracking. As revealed previously, mobile self-trackers were more technology savvy in
general, and the majority of them were smartphone users (i.e., 95.8%). That is to say,

102
what motivated them to get a wearable in addition to their smartphones for self-tracking
purpose was shown to be their perception of the extent to which their significant others
were doing mobile self-tracking. This finding suggests that, despite the prevalence of
smartphone ownership among self-trackers, using wearables for mobile self-tracking
seem to be a more salient result of social norms.
In addition, when we compared the differences in health outcomes between
mobile self-trackers with smartphones and mobile self-trackers with wearables, findings
have shown that mobile self-trackers using smartphones had significantly higher healthy
eating (M=3.84, SD=0.67) than mobile self-trackers with wearables (M=3.63, SD=0.83)
(t=2.137, p=0.029), while there were no outcome differences in health anxiety, social
support, physical activity, well-being, satisfaction, goal attained, and self-determination
between the two groups. This may be the reason that healthy eating requires more manual
input from users such as taking notes on foods consumed, diet diaries, etc. It would be
easier to track and record information on a smartphone than on a wearable for eating
behavior.
Another interesting finding revealed the differences in the themes of self-tracking
between the two types of mobile self-trackers. An independent sample t-test was
conducted to compare tracking themes between mobile self-trackers with mobile phones
and those who with wearables. Findings were shown that those who self-tracked with
wearables were significantly higher in fitness (t(242)=-3.58, p= 0.000, M
wearables
= .79,
SD=.41; M
mphones
= .58, SD=.50) and sleep (t(242)=-5.22, p= 0.000, M
wearbles
=.53, SD=.50;
M
mphone
=.22, SD=.42), and lower in weight (t(242)=2.043, p= 0.042, M
wearbles
=.65,
SD=.48; M
mphone
=.77, SD=.42) and diet and food (t(242)=2.399, p= 0.017, M
wearbles
=.55,

103
SD=.50; M
mphone
=.70, SD=.46). These findings suggest that wearbales may afford greater
capabilities for tracking activities that are constant and requires minimum interventions
when they are worn on the wrist or clipped on the body, while mobile phones afford
greater functions to manage weight, diet, and foods-related tracking through goal-specific
applications. This might also explain why mobile self-trackers with mobile phones were
shown to have significantly higher healthy eating behavior.
Table 20. Comparison Between Mobile Self-trackers with Wearables and Mobile phones
MS with Mobile Phones MS with Wearbles
Predictors
Social norms** M=2.39, SD=0.90 M=2.72, SD=1.00
Health Outcomes
Healthy Eating* M=3.84, SD=0.67 M=3.63, SD=0.83
Tracking Themes
Fitness*** M=0.58, SD=0.50 M=0.79, SD=0.41
Sleep*** M=0.22, SD=0.42 M=0.53, SD=0.50
Weight* M=0.77, SD=0.42 M=0.65, SD=0.48
Diet & Foods* M=0.70, SD=0.46 M=0.55, SD=0.50

Mediating Effects of Self-Determination
The fourth research question concerns the mediating effect of self-determination
on health outcomes among mobile self-trackers (See Figure 12).

104

Figure 12. Research model of self-determination as a mediator.
An earlier MANOVA has shown that group membership had a statistically
significant effect on self-determination (F (2, 518) = 10.13, p < .0000), where mobile
self-trackers had significantly higher self-determination (M=5.29, SD=0.99) than
traditional self-trackers (M=4.88, SD=0.98, p<0.000) and non-trackers (M=4.91,
SD=0.97, p=0.001). In order to further examine the role of self-determination in the
context of mobile self-tracking, three steps were performed to explore the following
hypothesis. Firstly, H
4
proposed that participants who engaged more in mobile self-
tracking had significantly higher self-determination on weight management. A single
regression analysis was conducted to examine the effect of mobile self-tracking
engagement on self-determination. It was shown that mobile self-tracking significantly
and strongly predicted self-determination (R
2
=0.254, F (1,262)=88.95, p<0.000; β
=0.504, p<0.000). H
4
was supported.

105
Secondly, H
5
proposed that mobile self-tracking engagement had positive effects
on H
5a
) health anxiety, H
5b
) social support H
5c
) healthy eating H
5d
) physical activity H
5e
)
well-being H
5f
)satisfaction and H
5g
) goal attainment. A series of single regression
analyses were conducted to examine the effects of mobile self-tracking engagement on
several proposed health outcomes. It was found that mobile self-tracking engagement had
significant effects on all outcomes but health anxiety (p=0.26), including social support
(R
2
=0.095, F(1,262)=27.43, p<0.000; β =0.308, p<0.000), healthy eating (R
2
=0.112,
F(1,262)=33.02, p<0.000; β =0.335, p<0.000), physical activity (R
2
=0.099,
F(1,262)=28.82, p<0.000; β =0.315, p<0.000), well-being (R
2
=0.08, F(1,262)=23.54,
p<0.000; β =0.288, p<0.000), satisfaction (R
2
=0.15, F(1,262)=46.05, p<0.000; β =0.387,
p<0.000), and goal attainment (R
2
=0.10, F(1,262)=27.65, p<0.000; β =0.310, p<0.000).
H
5a
was rejected, while H
5b
, H
5c
, H
5d
, H
5e
, H
5f
, and H
5g
were all supported.
Thirdly, a series of single regression analyses were conducted to examine the
effect of self-determination on the proposed outcomes. Results have shown that self-
determination had significant effects on all outcome variables. Therefore, when the three
aforementioned relationships were significant, we can further examine whether self-
determination reduces the influence of mobile self-tracking engagement on health
outcomes when it was included in each of the outcome regression models.
Results have shown that self-determination mediated the effects of mobile self-
tracking engagement on social support (R
2
=0.308, F(2,262)=57.98, p<0.000; β
ms
=0.039,
p=0.52; β
sd
=0.535, p< 0.000), healthy eating (R
2
=0.216, F(2,262)=35.76, p<0.000; β
ms

=0.147, p=0.021; β
sd
=0.372, p< 0.000), physical activity (R
2
=0.246, F(2,262)=42.48,
p<0.000; β
ms
=0.09, p=0.14; β
sd
=0.44, p< 0.000), well-being (R
2
=0.196, F(2,262)=31.62,

106
p<0.000; β
ms
=0.09, p=0.16; β
sd
=0.389, p< 0.000), satisfaction (R
2
=0.17,
F(2,262)=26.59, p<0.000; β
ms
=0.305, p<0.000; β
sd
=0.163, p= 0.013), and goal
attainment (R
2
=0.12, F(2,262)=17.47, p<0.000; β
ms
=0.222, p=0.001; β
sd
=0.175, p=
0.01).
To be more specific, when self-determination was included in the regression
model as a predictor of each of the outcomes, it reduced the influence of mobile self-
tracking engagement on healthy eating, well-being, satisfaction, and goal attainment,
which indicates a partial mediating effect of self-determination between mobile self-
tracking engagement and these health outcomes. In addition, self-determination also
mitigated the influence of mobile self-tracking engagement on social support, physical
activity, and well-being as the relationships became non-significant after self-
determination came into play in the model, which indicating a fully mediating effect of
self-determination on these outcomes (See Figure 13 & 14).

Figure 13. The mediating effects of self-determination on health outcomes (a).

107

Figure 14. The mediating effects of self-determination on health outcomes (b).

Moderating Effects of EMI
Literature suggests that ubiquitous health behavior assessment enabled by mobile
technologies, also known as ecological momentary assessment (EMA)/ecological
momentary intervention (EMI), plays an important role in health management among
users (Norris, 2012; Klasnja & Pratt, 2012; Riley et al., 2011). It has shown to be
effectively implanted for behavior interventions on a wide array of health issues (Heron
& Smyth, 2010). As EMA and EMI make it possible to deliver real-time treatment and
feedback in people’s natural environment, it is believed to be beneficial in facilitating
health behavior and managing health goals (Heron & Smyth, 2010). Therefore, the next
research question concerns the moderating effect of ecological momentary intervention
(EMI) on health outcomes among mobile self-trackers, that is, whether the amount of

108
EMI received by mobile self-trackers moderates their health outcomes. H
7
proposed that
self-tracking engagement and EMI will interact in predicting health outcomes such that
self-tracking engagement is a more important predictor of health outcomes for high-EMI
than low-EMI mobile self-trackers (See Figure 15).

Figure 15. Research model of EMI as moderator of mobile self-tracking engagement on
health outcomes.
A few steps were taken to examine the hypothesis. First, mobile self-trackers were
dichotomized into two groups using a cut-off point of the mean of EMI (M=1.054)
reported by participants: high EMI group and low EMI group. Following this, a
MANOVA was conducted to compare means between high-EMI and low EMI mobile
groups on studied health outcomes. The multivariate F tests have shown that the two
groups differed on health anxiety (p=0.016) and healthy eating (p=0.025), suggesting that

109
EMI (high vs. low) might have a moderating effect on two health outcomes (See Table
21).
Table 21. Estimated Marginal Means for Health Outcomes Between Groups
Dependent
Variable
EMI Group Mean SE p
HA* Low EMI 2.023 .096 0.016
High EMI 2.324 .079
SS Low EMI 5.335 .112 0.127
High EMI 5.558 .092
Eating* Low EMI 3.606 .068 0.025
High EMI 3.804 .056
PA Low EMI 3.432 .086 0.053
High EMI 3.647 .071
Well-being Low EMI 5.383 .114 0.245
High EMI 5.555 .094
Satisfaction Low EMI 5.048 .121 0.701
High EMI 4.988 .100
Goal
Attainment
Low EMI .730 .043 0.464
High EMI .688 .036
Note: *=Significant between groups.
Following this, in order to examine whether EMI moderates the relationship
between mobile self-tracking engagement and the studied outcomes, a series of
hierarchical regression analyses were conducted to validate if adding the interaction term
of EMI and mobile self-tracking engagement results in a significant change in R
2
for each
health outcome. If so, then we can determine that EMI is a moderator of that particular
health outcome.
The moderating effect analyses for each outcome model were as followed: The
overall model for healthy anxiety was significant, R
2
= 0.03, F(3, 259) = 2.55, p = .056.
EMI itself accounted for a significant amount of additional variance in health anxiety,
ΔR
2
= .023, ΔF(1, 260) = 6.10, p = .014, b = 0.16, t(260) = 2.47, p = .014. Next, an
interaction term between EMI (centered) and mobile self-tracking engagement (centered)

110
was created, which did not account for a significant proportion of the variance in health
anxiety, suggesting that there is not a moderating effect of EMI on health anxiety, ΔR
2
=
.001, ΔF(1, 259) = 0.24, p = .62, b = 0.41, t(259) = -0.10, p = .62.
The overall model for social support was significant, R
2
= 0.104, F(3, 259) =
10.06, p <0.000. EMI itself did not account for a significant amount of additional
variance in social support, ΔR
2
= .008, ΔF(1, 260) = 2.24, p = .136, b = 0.09, t(260) =
1.50, p = .136. In the final step of the regression analysis, an interaction term between
EMI (centered) and mobile self-tracking engagement (centered) was created, which did
not account for a significant proportion of the variance in healthy eating, indicating that
there is no moderating effect of EMI on social support, ΔR
2
= .001, ΔF(1, 259) = 0.02, p
=0.51, b = 0.13, t(259) = 0.662, p =0.51.
The overall model for healthy eating was significant, R
2
= 0.14, F(3, 259) =
14.31, p < .000. EMI itself accounted for a significant amount of additional variance in
healthy eating, ΔR
2
= .015, ΔF(1, 260) = 4.49, p = .035, b = 0.13, t(260) = 2.22, p = .035.
In the final step of the regression analysis, an interaction term between EMI (centered)
and mobile self-tracking engagement (centered) was created, which accounted for a
significant proportion of the variance in healthy eating, ΔR
2
= .015, ΔF(1, 259) = 4.48, p
= .035, b = 0.41, t(259) = 2.12, p = .035. The result suggested a moderating effect of EMI
on healthy eating.
The overall model for physical activity was significant, R
2
= 0.12, F(3, 259) =
11.43, p < .000. EMI itself accounted for a marginally significant amount of additional
variance in physical activity, ΔR
2
= .012, ΔF(1, 260) = 3.47, p = .06, b = 0.11, t(260) =
1.86, p = .06. In the final step of the regression analysis, an interaction term between EMI

111
(centered) and mobile self-tracking engagement (centered) was created, which did not
account for a significant proportion of the variance in physical activity, indicating that
there is not a moderating effect of EMI on physical activity, ΔR
2
= .006, ΔF(1, 259) =
1.65, p = .020, b = 0.25, t(259) = 1.28, p = .20.
Next, the overall model for well-being was significant, R
2
= 0.09, F(3, 259) =
11.35, p < .000. EMI itself did not account for a significant amount of additional variance
in well-being, ΔR
2
= .004, ΔF(1, 260) = 1.22, p = .27, b = 0.07, t(260) = 1.11, p = .27. In
the final step of the regression analysis, an interaction term between EMI (centered) and
mobile self-tracking engagement (centered) was created, which did not account for a
significant proportion of the variance in well-being, indicating that there is no moderating
effect of EMI on well-being, ΔR
2
= .004, ΔF(1, 259) = 1.07, p = .30, b = 0.21, t(259) =
1.03, p = .30.
The overall model for satisfaction was significant, R
2
= 0.15, F(3, 259) = 15.29, p
< .000. EMI itself did not account for a significant amount of additional variance in
satisfaction, ΔR
2
= .000, ΔF(1, 260) = 0.146, p = .70, b = -0.02, t(260) = -0.382, p = .70.
In the final step of the regression analysis, an interaction term between EMI (centered)
and mobile self-tracking engagement (centered) was created, which did not account for a
significant proportion of the variance in satisfaction, indicating that there is no
moderating effect of EMI on satisfaction, ΔR
2
= .000, ΔF(1, 259) = 0.004, p = .95, b = -
0.013, t(259) = -0.066, p = .95.
Finally, the overall model for goal attainment was significant, R
2
= 0.10, F(3,
259) = 1.855, p < .000. EMI itself did not account for a significant amount of additional
variance in goal attainment, ΔR
2
= .002, ΔF(1, 260) = 0.457, p = .50, b = -0.041, t(260) =

112
-0.676, p = .50. In the final step of the regression analysis, an interaction term between
EMI (centered) and mobile self-tracking engagement (centered) was created, which did
not account for a significant proportion of the variance in goal attainment, indicating that
there is no moderating effect of EMI on goal attainment, ΔR
2
= .003, ΔF(1, 259) = 0.947,
p = .33, b = -0.193, t(259) = -0.973, p = .33.
To sum up, the initial MANOVA test has shown that the high-EMI group had
significantly higher health anxiety and healthy eating. The following hierarchical
regression analyses revealed that EMI itself was found to significantly and positively
predict health anxiety and healthy eating. The findings confirmed the literature review,
indicating that EMI can be conducive to adaptive behavior while simultaneously
increasing health anxiety among mobile self-trackers. Nonetheless, the interaction term of
EMI and mobile self-tracking only significantly predicted healthy eating in the sample,
revealing that the moderation effect only exists in healthy eating. H
7
was partially
supported. The result implies that EMI may be most effective in promoting healthy eating
behavior among mobile self-trackers (See Table 22).

113
Table 22. Tests of Moderating Effect of EMI on Health Outcomes
Moderating
Effects
N=263 R
2
F p value
No Health Anxiety R
2
= 0.03
ΔR
2
= .001
F(3, 259) = 2.55
ΔF(1, 259) = 0.24
p = .056
p = .62
No Social Support R
2
= 0.104
ΔR
2
= .001
F(3, 259) = 10.06
ΔF(1, 259) = 0.02
p <.000
p =0.51
Yes Healthy Eating* R
2
= 0.14
ΔR
2
= .015
F(3, 259) = 14.31
ΔF(1, 259) = 4.48
p<.000
p = .035
No Physical Activity R
2
= 0.12
ΔR
2
= 0.006
F(3, 259) = 11.43
ΔF(1, 259) = 1.65
p < .000
p = .20
No Well-being R
2
= 0.09
ΔR
2
= .004
F(3, 259) = 11.35
ΔF(1, 259) = 1.07
p < .000
p = .30
No Satisfaction R
2
= 0.15
ΔR
2
= .000
F(3, 259) = 15.29
ΔF(1, 259) = 0.004
p < .000
p = .95
No Goal Attainment R
2
= 0.10,
ΔR
2
= .003
F(3, 259) = 1.855
ΔF(1, 259) = 0.947
p < .000
p = .33

Personal Network Influence
The sixth research question examines how individuals’ social networks influence
their self-tracking behavior for weigh management.

Figure 16. Research model of personal network influence on mobile self-tracking
adoption.

114
Network exposure. The first hypothesis concerns the effect of network exposure,
proposing that participants who reported as mobile self-trackers have higher personal
network exposure than participants who were not mobile self-trackers. First, self-tracking
network exposure and mobile self-tracking network exposure were calculated according
to participants’ egocentric data on their reported alters’ tracking behavior. One-way
ANOVA was conducted to examine this hypothesis. Results have indicated a main effect
of personal network exposure on mobile self-tracking behavior. Mobile self-trackers had
significantly greater personal network exposure to mobile self-tracking (F(2, 510) =
45.19, p < .000) (M = 1.87, SD = 1.83) than either traditional self-trackers (M = 0.75, SD
= 1.09) or non-trackers (M = 0.50, SD = 0.96). Further, mobile self-trackers had
significantly greater personal network exposure to self-tracking (F(2, 510) = 25.34, p <
.000; M = 1.97, SD = 1.78) than either traditional self-trackers (M = 1.50, SD = 1.44) or
non-trackers (M = 0.76, SD = 1.17). A logistic regression analysis was further performed
using mobile self-tracking behavior as the dependent variable and mobile tracking
exposure as independent variables among all participants (N=521). Results indicated that
the relationship between exposure to mobile self-tracking and mobile self-tracking
adoption is significant, X
2
(1, N = 521) = 86.51, p <0.000. Nagelkerke’s R
2
of 0.208
indicated a moderate amount of variance explained by the individual’s network exposure
to mobile self-tracking. Prediction success overall was 66.3 %, which showed a small
improvement on the 50.9% correct classification with the constant model. The Wald
criterion demonstrated that exposure to mobile self-tracking made significant
contributions to prediction (β= 0.63, SE= 0.08, Wald= 60.46, df=1, p<0.000, Odds=
1.877). The Hosmer and Lemeshow chi-square Test has a significance of 0.439, which

115
means that we accepted the null hypothesis that there is no difference between observed
and model-predicted values. The findings suggest that mobile self-tracking adoption was
influenced by personal social network exposure. Specifically, when individual exposure
increases by one unit, the odds ratio is 1.877 times as large and therefore participants are
1.877 more times likely to adopt mobile self-tracking. H
8
was supported.
Strength of ties. The next hypothesis concerns the effect of strength of tie on
individuals’ tracking behavior in terms of years known, interaction frequency, and
relationship type. The hypotheses propose that participants are more likely to have the
same mobile self-tracking behavior as their social ties whom a) they contact more
frequently, b) have known longer, and c) with whom they have closer relationship types.
In order to test the hypotheses, a series of logistic regression analyses were conducted
using same tracking behavior as the dependent variable, and years known, interaction
frequency, and relation type as independent variables respectively among all self-
trackers. Results have shown the relationship between strength of ties and same tracking
behavior is significant, X
2
(1, N = 2347) = 53.67, p <0.000. Specifically, relation type
(p=0.027) and interaction frequency (p=0.000) significantly predicted same tracking
behavior, while years known did not (p= 0.856). H
9a
was supported, indicating that
participants were more likely to have same self-tracking behavior as their social ties
whom they contact more frequently. H
9b
was rejected as years known did not predict
same tracking behavior. H
9c
was also supported, that certain relation types significantly
predicted same tracking behavior. Specifically, colleague/schoolmate (p=0.018) and
celebrity (p=0.04) are the sources of self-tracking influence among self-trackers. See
Table 23 for details.

116
Role modeling. The next hypothesis concerns the effect of role modeling in terms
of healthiness and attractiveness, proposing that participants are more likely to have
similar mobile self-tracking behavior as their social ties whom they perceived as healthy
or attractive. Two chi-square tests of independence were conducted to examine the
relationship between individuals’ tracking behavior and perceived healthiness and
perceived attractiveness of their nominators. The relationship between same tracking
behavior and healthiness was significant,

X
2
(1, N = 2801) = 15.11, p <0.000. Same
tracking behavior was more likely to happen between the participant and his or her alter
who is perceived as healthy by the participant than between the participants and an alter
who is not perceived as healthy. In addition, the relationship between same tracking
behavior and attractiveness was also significant, X
2
(1, N = 2801) = 5.76, p =0.016. Same
tracking behavior was more likely to happen between the participant and his or her alter
who is perceived as attractive by the participant than between the participants and an alter
who is not perceived as attractive. A logistic regression analysis was further conducted
using same tracking behavior as the dependent variable and perceived healthiness and
perceived attractiveness as independent variables. A test of the full model against a
constant only model was statistically significant, indicating that the predictors as a set
reliably distinguished between same tracking behavior or not (X
2
= 31.83, p < .000 with df
= 2). Nagelkerke’s R
2
of 0.02 indicated a small amount of variance explained by the two
variables. Prediction success overall was 69.5%. The Wald criterion demonstrated that
both attractiveness (p=0.037) and healthiness (p<0.000) made significant contributions to
prediction. H
10a
and H
10b
were supported.
Similarity. The next hypothesis concerns the influence of similarity in terms of

117
age, gender, and ethnicity on self-trackers’ tracking behavior. The hypothesis proposes
that participants are more likely to have similar mobile self-tracking behavior as their
social ties who are H
11a
) the same age, H
11b
) the same gender, and H
11c
) from the same
ethnic groups. Firstly, alters’ age were calculated and transformed into “same aged
group,” “older aged group,” and “younger aged group” according to the provided
information about alters’ age. A chi-square test of independence was performed to check
if same tracking behavior is dependent on same age. The result has shown that the
relationship between age and same tracking behavior was not significant, X
2
(1, N =
2790) = 1.36, p =0.507, indicating that same tracking behavior was not associated with
age groups. H
11a
was rejected.
Second, a chi-square test of independence was performed to examine the
relationship between gender and same tracking behavior. The relationship between
gender and same tracking behavior was not significant either, X
2
(1, N = 2784) = 0.030, p
=0.872. This indicates that same tracking behavior was not associated with the same
gender group. H
11b
was rejected.
Another chi-square test of independence was performed to examine the
relationship between ethnicity and same tracking behavior. The relationship between
ethnicity and same tracking behavior was not significant either, X
2
(1, N = 2772) = 3.27,
p =0.07. In addition, when all the three similarity measures (i.e., age, gender, ethnicity)
were included in the logistic regression model, the model was not significant (X
2
(1, N =
2347) = 6.60, p =0.086 with df=3), indicating that these similarity measures as a set do
not distinguish between same tracking behavior and different tracking behavior. H
11a
,
H
11b,
and H
11c
were all rejected.

118
Finally, a multiple predictors logistic regression analysis was performed to predict
same tracking behavior between all trackers and their alters (N=2347) using relation type,
interaction frequency, years known, healthiness, attractiveness, same ethnicity, same
gender, and same age as independent variables. A test of the full model against a constant
only model was statistically significant, indicating that the predictors as a set reliably
distinguished between same tracking behavior or not (X
2
= 88.24, p < .0000 with df = 14).
Nagelkerke’s R
2
of 0.06 indicated a small amount of variance explained by the set of
predictors. Prediction success overall was 69.1%. The Wald criterion demonstrated that,
perceived healthiness (p<0.000, Odds= 1.666), perceived attractiveness (p=0.038, Odds=
1.410), interaction frequency (p<0.000, Odds= 1.381), and relation type (p=0.013) made
significant contributions to prediction. Specifically, among various relation types,
colleague/schoolmate (p=0.027, Odds= 2.173) and celebrity (p=0.049, Odds= 3.512)
made significant contributions to prediction. The Hosmer and Lemeshow chi-square Test
had a significance of 0.646, which means that we accepted the null hypothesis that there
is no difference between observed and model-predicted values. This implies that the
model’s estimates fit the data at an acceptable level. In other words, the model is an
adequate fit to the data, and the model prediction does not significantly differ from the
observed.

119
Table 23. Predictors of Same Tracking Behavior using Logistic Regression Analyses
Among all Tracker pairs (N=2347)
Independent variables b se Wald Prob. Odds
Relation Type / / 16.22 0.013 /
Colleague/Schoolmate* 0.776 0.351 4.894 0.027 2.173
Celebrity* 1.256 0.639 3.861 0.049 3.512
Interaction Frequency*** 0.323 0.057 31.97 0.000 1.381
Healthiness*** 0.511 0.139 13.53 0.000 1.666
Attractiveness* 0.343 0.166 4.297 0.038 1.410
(Insignificant IVs)
(Years Known) -0.002 0.004 0.287 0.592 0.998
(Same Ethnicity) 0.139 0.153 0.825 0.364 1.149
(Same Gender) -0.131 0.107 1.503 0.220 0.877
(Age Group) / / 2.191 0.334 /

Modeling Mobile Self-tracking for Weight Management
Structural equation modeling was performed using AMOS 23.0 to further
integrate several research models examined in the previous sections as well as evaluate
the overall model fit for a refined model for mobile self-tracking (Anderson & Gerbing,
1988). The proposed model was a six-factor model combining measurements models and
path models for mobile self-trackers. Initial model specification was guided by the
original research model and regression results conducted previously. Health outcomes
examined in the present study were further categorized into two latent factors ⎯ adaptive
behavior and positive psychology ⎯ in order to more accurately reflect the nature of the
concept (See Figure 17). Adaptive behavior consists of two behavioral outcomes in the
present study, healthy eating and physical activity. Positive psychology, defined as “how
people develop and sustain positive characteristics, such as hope, courage, spirituality,
wisdom, future mindedness, and perseverance in the face of stress,” indicates a general
healthy mindset of having a sense of meaningfulness in life, internal locus of control, and

120
purpose of life, etc. (Glanz & Schwartz, 2008, p.223). This latent factor aims to represent
the general healthy mindset as an outcome of self-tracking, including well-being,
satisfaction, and perceived social support.

Figure 17. Base model for mobile self-tracking for weight management.
Assessment of Normality
Before proceeding to the model estimation, a test of normality was performed to
ensure the data is normally distributed and is suitable for Maximum Likelihood Estimates
method. Firstly, skewness and kurtosis of univariate variables in the model were checked.

121
Results have shown that all the values of skewness for each univariate variable is within
|2|, and all the values of kurtosis for each variable is within |7|. Results indicate that these
univariate variables in the model meet the assumption of univariate normality (Kline,
2005, p.49-52). In addition, the multivariate c.r. value for the sample is 29.21. A value of
multivariate c.r. less than 50 is appropriate for using Maximum Likelihood Estimates to
generate robust estimates (Gao, Mokhtarian & Johnston, 2007). Therefore, Maximum
Likelihood Estimates method is appropriate and was used to for model specification and
evaluation.
Model Fit
The overall goodness-of-fit results of the Base Model are described below. The chi-
square for the base model is 714.503 with degree of freedom of 221. The p value is
significant (p=0.000), and the model fit indices are acceptable but not ideal (CMIN/df=
3.233, RMSEA= .092, CFI= .811, GFI= .787, RMR=0.107, See Figure 18). Further
model modifications were made based on the Modification Indices suggested.

122

Figure 18. Results: Base model for mobile self-tracking with standardized estimates

Model Modification- Correlated Error Terms
According to the Modification Indices, several correlations between error terms
were suggested in order to improve the model fit. Although correlating errors may not be
a recommended way to improve model fit, scholars suggest that it is acceptable when a).
there are theoretical or methodological justifications, b) correlating error terms does not
significantly change the estimates of regression weights in the model, and c) correlating

123
error terms do not significantly change factor loadings of the observed variables (Fornell,
1983; Chin, 1998; Bagozzi, 1983). Correlated error terms in measurement models
represent the assumption that the unique variances of the associated observed variables
overlap. In other words, any two variables in the model could possibly measure
something in common other than the latent constructs that are represented in the model.
Hence, error correlation can be seen as “an unanalyzed association” representing “a
specific nature of a shared something (that) is unknown”
3
(“Correlated Error Terms”,
2007). After reviewing each suggested correlation between error terms, seven correlated
errors among observed variables were justified and added into the research model for
theoretical considerations for potentially overlapping variance (see Table 24). By
releasing the correlations between these error terms, the overall model fit was improved
(v.83) (chi-square= 508.160, df= 215, p= .0000, CMIN/df=2.364, RMSEA= 0.072, GFI=
.853, CFI= 0.888, RMR=0.09).
Table 24. Modification Indices with Univariate Increment for Model Revision
Error terms Error terms M.I. Par Change
e8(Involvement) <--> e9 (Adherence) 62.896 0.309
e4 (Relatedness) <--> e34 (Social share) 51.19 0.652
e9 (Adherence) <--> e30 (Perseverance) 38.444 0.196
e7 (Intensity) <--> e9 (Adherence) 21.304 0.24
e8 (Involvement) <--> e30 (Perseverance) 21.223 0.196
e1 (Social norms) <--> e4 (Relatedness) 19.11 0.323
e7 (Intensity) <--> e30 (Perseverance) 17.115 0.234

3
http://zencaroline.blogspot.com/2007/04/correlated-error-terms.html

124

Figure 19. Model modification for mobile self-tracking with standardized estimates.

Confirmatory Model Specification
Results of the revised model showed that all paths were significant but three,
including the relationship between self-determination to adaptive behavior, self-tracking
engagement to positive psychology, and social influence to mobile self-tracking exposure

125
(See Figure 20). A confirmatory model specification with AMOS was thus performed to
check the best model specification. Results of the confirmatory model specification
showed that the best model is the one without the three nonsignificant paths previously
reported. The model (Model 51) had a BIC value of 0.000 and lowest value of BCC,
indicating that it had “positive” evidence that Model 51 would be the best model for the
data among alternative models (See Table 25).

Figure 20. Confirmatory model specification with AMOS.

126

Table 25. Results of the Confirmatory Model Specification
Model Name Params df χ
2
C-df BCC 0 BIC 0 c/df p
51 Default
model
58 218 509.562 291.562 29.596 0.000 2.337 0.000
41 Default
model
57 219 518.203 299.203 36.034 3.068 2.366 0.000
52 Default
model
58 218 513.904 295.904 33.938 4.342 2.357 0.000
61 Default
model
59 217 508.431 291.431 30.666 4.441 2.343 0.000

Model Revision
Another structural equation modeling was performed without the three
nonsignificant paths suggested to be removed from the model. The obtained fit indices
were compared to those with the paths remained in the model. This resulted in a decrease
in chi-squares by 19.137, but the CFI (=.888) remained the same while GFI (=.852),
RMR(=.092) and RMSEA (=.076) worsened. Results suggested that there might be no
room for further model improvement.
Table 26. Insignificant Paths to be Deleted in Modification 2
Factor Factor Estimate S.E. C.R. P
Adaptive Behavior <--- Self-determination .125 .108 1.19 .251
Positive Psychology <--- Self-tracking
Engagement
.214 .481 .445 .656
Mobile Self-tracking
Exposure
<--- Social Influence .026 .092 .288 .773

127
Table 27. Summary of Model Revisions for the Full Sample
Paths added
(Univariate Increment)
χ
2
df CMIN/DF p RMSEA CFI GFI RMR
Base Model

714.503 221 3.233 .000 .092 .811 .787 .107
Revision 1:
Correlated errors
508.160 215 2.364 .000 .072 .888 .853 .090
Revision 2:
Removing Insignificant
Paths
489.023 196 2.495 .000 .076 .888 .852 .092

The solution ( χ
2
(215) = 508.160, p= 0.000, CMIN/DF= 2.364, CFI=.888, GFI =
.853, RMSEA = 0.07, RMR= 0.09, and cmin/df = 2.337) suggests, with a significant p
value, that the structural equation model was not consistent with the data distribution of
the sample. However, as suggested by Jöreskog & Sörbom (1993), χ
2
should not be the
only index to evaluate goodness of fit since it is unduly influenced by large sample sizes.
Tanaka (1993) and Maruyama (1998) suggest that p value tends to be significant when
the sample size is over 200. Therefore, accessing other model fit indices to evaluate the
model fit is appropriate. We can use CFI and GFI to evaluate the model because they are
independent of model complexity and sample size (Cheung & Rensvold, 2002). Taking
into consideration other indices in this study suggests that the research model should be
acceptable and should not be abandoned.
Table 28. Model Fit Indices for the Final Model
Model Fit χ
2
(215) = 508.160, p= 0.000, df= 215
Goodness-of-fit Recommended cut-off criterion Final Model
CFI >0.95 (Hu and Bentler, 1999) 0.888
GFI >0.80 (Etezadi-Amolo and Farhoomand, 1996) 0.853
RMSEA <0.1 0.072
RMR <0.05 0.090
CMIN/df ~0 2.364

128
Construct Validity
Construct reliability (CR) and Avaerage Variance Extract (AVE) were calculated for
each of the latent factors to assess their internal quality. As shown in Table_, the
construct reliability were all above 0.6 (Fornell & Larcker, 1981) except self-tracking
engagement, suggesting a good internal validity of the model. In addition, average
variance extract (AVE) was calculated for each latent factor to examine the extent to
which the variance explained by the latent factor is from the observed variables. Results
have shown that three of the latent factors have an AVE value greater than 0.5 (Jöreskog
& Sörbom, 1993). The greater AVE suggests a greater amount of variance explained by
the latent variable is from the observed variables. The AVE for self-tracking engagement
is not ideal among all.
Table 29. Composite Reliability (CR) and Average Variance Extract (AVE) for the Final
Model
Latent Factor CR (> = 0.6,
(Fornell & Larcker, 1981)
AVE (> = 0.5)
(Jöreskog & Sörbom, 1993)
Motivation 0.81 0.43
Social Influence 0.60 0.45
Self-tracking Engagement 0.51 0.19
Self-determination 0.77 0.61
Adaptive Behavior 0.72 0.56
Positive Psychology 0.74 0.50

Discriminant Validity
In addition, discriminant validity was evaluated with Bootstrapping method to
obtain confidence intervals of the correlations (Torkzadeh, Koufteros, & Pflughoeft,
2003), and average variance extracted criterion (Fornell & Larcker, 1981). First,
bootstrapping confidence intervals have shown that all the confidence intervals of paired

129
correlation did not include “1”, indicating that all the correlations among any two factors
in the model were significantly less than “1” (p<0.05). In other words, any paired factors
in this model are distinct and non-identical, indicating that discriminant validity is
supported.
In addition, the average variance extracted method was also used (Fornell &
Larker, 1981). It is believed that evidence of discriminant validity is shown if the average
variance extracted (AVE) is greater than the square of the factor’s correlations with the
other factors. Results have shown that most of the AVEs appearing in the diagonal are
larger than any correlation between the associated latent construct and any other
construct. It is considered a more stringent method to examine discriminant validity thus
evidence from most AVEs is sufficient. Both the bootstrapping method and average
variance extracted method have shown that the model displays discriminant validity
(Chin, 1998; Majchrzak, Beath, Lim, & Chin, 2005).
Table 30. Square Root of Average Variance Extracted and Correlation Between Latent
Variables
Factor
Motivation Social
Influence
Self-tracking
Engagement
Self-
determination
Adaptive
Behavior
Positive
Psychology
Motivation
0.66 - - - - -
Social
Influence
0.66 0.67 - - - -
Self-tracking
Engagement
0.67 0.48 0.44 - - -
Self-
determination
0.82 0.68 0.54 0.78 - -
Adaptive
Behavior
0.78 0.62 0.56 0.72 0.75 -
Positive
Psychology
0.65 0.46 0.44 0.64 0.67 0.71

Taken together, the construct reliability (CR) for each factor is mostly larger than
0.6, and the average variance extracted (AVE) for each factor is mostly closer or greater

130
than 0.5. This indicates that the model has a good internal consistency and convergent
validity. Therefore, the final model should be considered acceptable and should not be
abandoned.
Common Method Variance Test
Common Method Variance refers to the systematic measurement error in the
study that can bias the estimates of the true relationship among theoretical constructs, and
can either inflate or deflate observed relationship between constructs, and in turn increase
Type I or Type II errors (Eichhorn, 2014). In order to detect whether there is a common
method variance in the model, a CFA estimate method (Lindell & Whitney, 2001) and a
CFA comparison method were used. Firstly, the CFA estimate method calculated the
average of factor loadings from a single-factor CFA with a hypothesized latent variable
of “common method variance.” The average of the factor loadings from all the observed
variables on the latent variable represents the influence of the common method. The
value in the current model is 0.05, indicating a negligible influence of a common method.
In addition, a chi-square difference test was performed to confirm whether a common
method variance exists. Results have shown that the single factor (common method
variance) model significantly differed from the multi-factor model (See Table 29). This
suggests that a common method variance does not exist in the current study.
Table 31. Results of the Common Method Variance Test
Model χ
2
df Δχ
2
Δdf p
Single-Factor 1219.819 252 627.304 37 0.000
Multi-Factor 592.515 215

131
Table 32. Standardized Total Effects, Direct Effects, and Indirect Effects
Factor Motivation
(MO)
Social
Influence (SI)
Self-tracking
(ST)
Adaptive
Behavior (AB)
Self-
determination
(SD)
Positive
Psychology
(PP)
Effects Di. In. To. Di. In. To. Di. In. To. Di. In. To. Di. In. To. Di. In
.
To
.
Self-Tracking
.77 . .77 .30 . .30 . . . . . . . . . . . .
Self-
determination . .66 .66 . .25 .25 .85 . .85 . . . . . . . . .
Adaptive
Behavior . .62 .62 . .24 .24 .66 .14 .80 . . . .16 . .16 . . .
Positive
Psychology . .50 .50 . .19 .19 .08 .57 .65 .41 . .41 .28 .07 .35 . . .
Internalizatio
n . .52 .52 . .20 .20 . .69 .68 . . . .80 . .80 . . .
Share
. .25 .25 . .10 .10 .32 . .32 . . . . . . . . .
NB
. . . .84 . .84 . . . . . . . . . . . .
PA_
. .47 .47 . .18 .18 . .60 .60 .76 . .76 . .12 .12 . . .
Perseverance
. .20 .20 . .08 .08 .25 . .25 . . . . . . . . .
SE
.80 . .80 . . . . . . . . . . . . . . .
HC
.72 . .72 . . . . . . . . . . . . . . .
HLoC
.68 . .68 . . . . . . . . . . . . . . .
mHLiteracy
.74 . .74 . . . . . . . . . . . . . . .
VCope
.48 . .48 . . . . . . . . . . . . . . .
PIMS
.46 . .46 . . . . . . . . . . . . . . .
Satisfaction
. .24 .24 . .09 .09 . .31 .31 . .20 .20 . .16 .16 .47 . .47
SS
. .38 .38 . .15 .15 . .49 .49 . .31 .31 . .26 .26 .76 . .76
Well-being
. .41 .41 . .16 .16 . .53 .53 . .34 .34 . .28 .28 .82 . .82
Eating
. .46 .46 . .18 .18 . .61 .60 .75 . .75 . .12 .12 . . .
Adherence
. .36 .36 . .14 .14 .47 . .47 . . . . . . . . .
Involvement
. .48 .48 . .18 .18 .62 . .62 . . . . . . . . .
Intensity
. .31 .31 . .12 .12 .40 . .40 . . . . . . . . .
Competence
. .59 .59 . .23 .23 . .77 .76 . . . .90 . .90 . . .
Related
. .35 .35 . .14 .14 . .46 .46 . . . .54 . .54 . . .
Autonomy
. .54 .54 . .21 .21 . .71 .71 . . . .83 . .83 . . .
SN
. . . .46 . .46 . . . . . . . . . . . .

132
Mediating Effects: SEM Approach
The final research model has shown that the effects of mobile self-tracking
engagement on positive psychology are mediating by self-determination. A
Bootstrapping Confidence Intervals with AMOS was performed to validate the mediating
effects on adaptive behavior and positive psychology. Results have shown that the direct
effect of self-tracking engagement on adaptive behavior was significant, while the
indirect effect through self-determination was insignificant. This suggests that self-
determination did not mediate the relationship between self-tracking engagement and
adaptive behavior. On the other hand, the total effect and indirect effect of self-tracking
engagement on positive psychology was significant, but the direct effect was not. This
suggests that the influence on positive psychology was mainly indirect, which was fully
mediated by both self-determination and adaptive behavior. Results revealed interesting
findings on mobile self-tracking, suggesting the engagement in mobile self-tracking can
help develop and sustain healthy mindsets through endowing self-determination and
facilitating adaptive behavior change.
Table 33. Test of Mediation Effects using Bootstrapping Confidence Intervals with AMOS
Bootstrapping
Bias-Corrected 95% CI Percentile 95% CI
Path Estimate SE Lower Upper p Lower Upper p
STàAB
(Total)
1.490 .255 1.072 2.029 .001 1.133 1.970 .001
(Direct) 1.236 .380 0.602 2.058 .001 0.701 1.937 .001
(Indirect) 0.254 .306 -.258 .964 .33 -0.222 0.794 0.417
STàPP
(Total)
1.712 .281 1.265 2.396 .001 1.336 2.250 .001
(Direct) 0.214 .844 -1.395 1.983 .747 -1.198 1.496 .843
(Indirect) 1.497 .821 .001 3.189 .000 0.349 2.974 .038
Note: 2,000 bootstrap samples with 95% confidence interval.

133
Multi-Group Analysis
Finally, a multi-group analysis in SEM was performed. The goal of testing for
measurement and structure invariance is to determine if the same SEM model is
applicable to both mobile self-trackers and traditional self-trackers. The general
procedure is to compare the unconstrained model for all groups combined, and then for a
model with constrained parameters in which all parameters are constrained to be equal
between the two groups. If the chi-square difference statistic is significant between the
unconstrained and the constrained models, then we conclude that the two groups are
different at the model level, and this SEM model is not applicable for all self-trackers. On
the contrary, it the chi-square difference statistic is nonsignificant, then we conclude that
the two groups are not different at the model level and the SEM model can apply to all
self-trackers.
A multi-group analysis between mobile self-trackers and traditional self-trackers
with AMOS was performed to test group invariance. The chi-square difference between
the unconstrained and constrained models is 10.856 with df=17, p=0.864. The result
suggests that imposing the additional restrictions of equal factor loadings across the two
groups did not result in a statistically significant worsening of overall model fit. In other
words, AMOS assumed that the unconstrained model is true, and the equal-loading model
that specified a group-invariant factor pattern was supported by the sample data.
Therefore, we accepted the equal-factor-loading model which suggests the two groups are
invariant at the model level (See Table 32 for details). That is, mobile self-trackers and
traditional self-trackers did not significantly differ in factor loadings at the model level.

134
Table 34. Baseline Comparison: Assuming Model Unconstrained to be Correct
Model df CMIN P NFI IFI RFI TLI
Delta-1 Delta-2 rho-1 rho2
Structural
weights
8 14.708 0.065 0.004 0.004 -0.001 -0.001
Structural
covariances
11 19.455 0.053 0.005 0.005 -0.001 -0.002
Structural
residuals
14 29.784 0.008 0.007 0.008 0 0
Measurement
residuals
44 204.823 0.000 0.049 0.055 0.026 0.03
Assuming model Structural weights to be correct:

Model df CMIN P NFI IFI RFI TLI
Delta-1 Delta-2 rho-1 rho2
Structural
covariances
3 4.747 0.191 0.001 0.001 -0.001 -0.001
Structural
residuals
6 15.076 0.02 0.004 0.004 0 0
Measurement
residuals
36 190.115 0.000 0.045 0.051 0.027 0.031

Assuming model Structural covariance to be correct:

Model df CMIN P NFI IFI RFI TLI
Delta-1 Delta-2 rho-1 rho2
Structural
residuals
3 10.329 0.016 0.002 0.003 0.001 0.001
Measurement
residuals
33 185.368 0.000 0.044 0.05 0.027 0.031

Assuming model Structural residuals to be correct:

Model df CMIN P NFI IFI RFI TLI
Delta-1 Delta-2 rho-1 rho2
Measurement
residuals
30 175.038 0.000 0.042 0.047 0.027 0.03

The finding of group invariance between mobile self-trackers and traditional self-
trackers suggests that the overall SEM model fits both tracker groups. Nonetheless, there

135
might be difference in a specific path between the two groups and additional procedures
are required to check for path differences. Therefore, a series of chi-square difference
tests were performed for path-by-path comparison. Results have shown that, among all
the paths in the model, only the path from adaptive behavior to positive psychology
differed between the two groups (chi-square difference test: p< 0.05)(See Table 33).
Specifically, the path was significant in the mobile self-tracking group, but is not
significant in the traditional self-tracking group (See Figure 21). Findings suggest that,
for mobile self-trackers, acquired sense of self-determination and adaptive behavior can
be further translated into positive psychology, while it is not the case among traditional
self-trackers. This demonstrates the evidence that mobile self-tracking could be a more
effective approach to weight management for health. Conclusions and implications were
discussed in the next section.

Figure 21. Path-by-path comparison between groups using Chi-square difference tests.

136
Table 35. Pairwise Comparison Between Groups with Chi-square Difference Tests
Mobile Self-tracking Traditional Self-tracking Pairwise

χ
2
Differen
ce Test
Path Estim
ate
S.E. C.R. P Estima
te
S.E. C.R. P c.r. p
STßSI 0.115 0.042 2.724 0.006 0.033 0.045 0.728 0.466 -1.328 ns
STßMo 0.427 0.074 5.786 *** 0.679 0.133 5.118 *** 1.661 ns
SDßST 2.039 0.277 7.348 *** 1.817 0.352 5.158 *** -0.495 ns
ABßST 1.236 0.305 4.057 *** 0.894 0.305 2.932 0.003 -0.795 ns
ABßSD 0.125 0.109 1.148 0.251 0.248 0.109 2.278 0.023 0.803 ns
PPßSD 0.309 0.157 1.965 0.049 0.537 0.6 0.895 0.371 0.368 ns
PPßST 0.214 0.481 0.445 0.656 3.409 2.413 1.413 0.158 1.298 ns
PPßAB 0.582 0.216 2.698 0.007 -1.947 2.335 -0.83 0.404 1.078 p<0.05
SNßSI 0.473 0.081 5.876 *** 0.54 0.155 3.482 *** 0.381 ns
MNEßSI 0.026 0.092 0.288 0.773 -0.055 0.133 -0.42 0.676 -0.507 ns
AutoßSD 1 1
RelaßSD 0.93 0.102 9.116 *** 0.804 0.139 5.791 *** -0.726 ns
CompßS
D
0.956 0.055 17.52
7
*** 0.861 0.072 12.01
7
*** -1.065 ns
IntenßST 1.446 0.247 5.857 *** 1.186 0.33 3.597 *** -0.632 ns
InvoßST 1.973 0.215 9.187 *** 1.915 0.279 6.851 *** -0.166 ns
AdhßST 1 1
SatßPP 0.637 0.091 7.038 *** 0.793 0.202 3.936 *** 0.707 ns
PIMSßM
O
1.032 0.146 7.084 *** 1.088 0.27 4.027 *** 0.184 ns
Vcopeß
MO
0.586 0.08 7.317 *** 0.687 0.147 4.673 *** 0.604 ns
mHLßM
O
1.277 0.114 11.17
1
*** 0.888 0.293 3.033 0.002 -1.238 ns
HLoCßM
O
0.852 0.082 10.39
8
*** 0.734 0.134 5.484 *** -0.752 ns
HCßMO 1 1
SEßMO 1.119 0.093 12.06
5
*** 1.085 0.171 6.338 *** -0.176 ns
PersßST 0.532 0.122 4.363 *** 0.321 0.165 1.942 0.052 -1.024 ns

137
EatßAB 0.798 0.077 10.42
7
*** 0.835 0.131 6.395 *** 0.246 ns
SSßPP 0.919 0.086 10.69 *** 0.622 0.153 4.068 *** -1.693 ns
WBßPP 1 1
NBßSI 1 1
ShareßS
T
1.015 0.226 4.498 *** 0.69 0.226 3.06 0.002 -1.018 ns
PAßAB 1 1 ns
Internalß
SD
0.954 0.064 15.00
5
*** 0.849 0.085 9.998 *** -0.985 ns

138
CHAPTER SIX: DISCUSSION AND CONCLUSION
The present study has revealed several key findings associated with mobile self-
tracking. First of all, it describes mobile self-trackers by identifying them as people who
are younger, wealthier, healthier, and had higher educational attainment compared to
traditional self-trackers and non-trackers. This finding is in line with earlier studies,
suggesting that mobile self-tracking is primarily taken for the reason of optimizing health
instead of illness management (Nißen, 2013; Wolf, 2010). Mobile self-trackers were also
more likely to be smartphone users, in a committed relationship, and currently employed.
It might be the reason that self-tracking behavior is salient and interdependent when
individuals are engaging with their partners or coworkers. It echoed findings on the
sources of social influence that “colleague” is a significant source of influence on mobile
self-tracking behavior adoption. Mobile self-trackers were also significantly higher in
social media use and social media activity than traditional self-trackers and non-trackers,
suggesting that they are people whose predisposition was typically in favor of using
communication technologies.
Second, consistent with theoretical perspectives, mobile self-trackers scored
higher on the four proposed dimensions (i.e., health-oriented, technology-oriented, self-
monitoring-oriented, and socially oriented factors) than either traditional self-trackers or
non-trackers. To be more specific, the three groups significantly differed in social norms
and normative beliefs, where mobile self-trackers had the greatest perceived social norms
and normative beliefs and non-trackers had the lowest. Social influence is shown to be
one of the key factors that affect individuals’ decision-making on mobile self-tracking. It
was also validated in the SEM model. Nonetheless, it is worth noting that mobile self-

139
trackers did not significantly differ from traditional self-trackers in health consciousness,
health locus of control, and self-efficacy, but the two groups differed in technology-
oriented traits (i.e, mHealth literacy, PIMS) and vigilance coping. This suggests that both
kinds of self-trackers may be equally health conscious, feeling responsible for their
health, and having confidence in managing their health compared to non-trackers, while
mobile self-trackers are more vigilant and technology-savvy by nature. This finding
confirmed the UTAUT model (Venkatesh, Morris, Davis & Davis, 2003), suggesting that
individuals’ acceptance of technology are heavily influenced by their attitudes and
expectancy. For example, PIMS is parallel to the attitude toward the performance
expectancy, which is believed to play an important role in predicting health behavior with
new technology (Haapala et al., 2009); mHealth literacy is associated with effort
expectancy, involving the degree of ease associated with the use of mobile technologies.
In addition, findings confirmed the theory of planned behavior (Ajzen, 2002),
indicating that vigilance coping, self-efficacy, and normative beliefs played important
roles in predicting one’s autonomous motivation to mobile self-tracking, and motivation
in turn predicted mobile self-tracking engagement. Based on the expectancy-value
assumption, the theory of planned behavior proposes that intention as a proxy of
motivation is influenced by attitudes, subjective norms, and perceived control. In this
project, results have shown that autonomous motivation for self-tracking is most likely to
occur when a person has a dispositional monitoring tendency (i.e., vigilance coping),
sufficient confidence (i.e., self-efficacy), and internalized perceptions about mobile self-
tracking as a right thing to do (i.e., normative beliefs). This demonstrates that
dispositional coping style and situation-specific capability outweighed other health-

140
related and technology-oriented factors in determining autonomous motivation to mobile
self-tracking. It implies that, to promote mobile self-tracking, it may be more effective to
address the development of self-esteem in a specific context. In addition, vigilance by
nature was identified as a salient factor in motivating mobile self-tracking behavior. It has
practical implications for health promotion. For example, matching health messages to
“monitor-blunter” coping style would be effective in encouraging mobile self-tracking
adoption, and mechanisms such as information-processing, need for cognition, and
proactive problem-solving might be particularly effective techniques for potential mobile
self-trackers (Miller, 1987; Williams-Piehota, Pizarro, Schneider, Mowad, & Salovey,
2005).
On the other hand, consistent with self-determination theory, findings have shown
that autonomous motivation significantly predicted behavior engagement, demonstrating
that it is conducive to the initiation of health behavior. As many people engage in weight
management activities only because of external regulation such as doctor’s warning,
parents’ enforcement, and so forth, it is hard for people to take on and sustain a regimen
if they are not inherently enjoying it (Ryan et al., 2008). Findings highlight the
importance of autonomous motivation in that individuals must come to believe in own
ability to perform the behavior and internalize the behavior as something they are
intrinsically interested in through social learning in order to initiate the behavior
engagement.
Of particular interest is to examine the effects associated with self-guided mobile
self-tracking for weight management. The multivariate analyses of covariance revealed
the effects of group membership on various health outcomes. Results have shown that

141
mobile self-trackers had significantly higher well-being and marginally lower health
anxiety than traditional self-trackers after controlling for confounders. This serves as
evidence to counterbalance the criticism about mobile self-tracking as being the cause of
undue anxiety about one’s own health. On the contrary, participants who constantly self-
monitored and acquired health information through mobile self-tracking had lower health
anxiety and higher psychological well-being than traditional self-trackers. This might be
the case in real life where more accurate data collected through mobile self-tracking
actually help reduce uncertainties and health concerns among habitual trackers.
Furthermore, mobile self-trackers have shown significantly greater healthy eating
behavior, physical activity, higher well-being, satisfaction, and goal attainment than non-
trackers. Results suggest that mobile self-tracking is probably a superior way to other
kinds of practices for weight management when used in a private, non-supervised setting.
Furthermore, results confirmed self-determination theory, indicating that self-
determination mediated the effects of self-tracking engagement on various health
outcomes. The theory addresses acquiring basic psychological needs that are conducive
to health behavior through the support of autonomy, competence, and relatedness. That
is, developing a sense of autonomy, competence, and relatedness is critical to the
processes of internalization and sustaining motivation to a particular behavior, which will
in turn lead to well-being. When engaging in a behavior simultaneously satisfies these
psychological needs, a person will come to internally self-regulate and sustain the
behavior that produces health outcomes (Ryan et al., 2008). This effect was empirically
validated in self-guided mobile self-tracking. Self-determination was found to fully
mediate the effects of mobile self-tracking engagement on social support, physical

142
activity, and well-being, and partially mediate the effects on healthy eating, satisfaction,
and goal attainment. It might be the reason that, during mobile self-tracking, relevant
information and rationales for weight management are provided without external control
that detracts individuals from a sense of choice. Plus, multiple features allow individuals
to decide how they want to go about self-tracking and align it with their own lifestyle
patterns and central values. A sense of autonomy is gained. In addition, with enhanced
sensors and functions, mobile self-tracking is designed to identify behavior patterns and
barriers that might be incongruent to one’s health goals. A sense of competence is likely
to grow. Particularly, when a barrier is overcome with mobile self-tracking, it creates a
sense of “effectance motivation” that further motivates individuals to continue on (Ryan,
Patrick, Deci, & Williams, 2008). Finally, a sense of relatedness (i.e., being understood,
cared for, and a sense of belonging) is likely to form because of enhanced connectivity
over mobile devises that enable acquisition of social support and social capitals. Taken
together, the theoretical perspectives were supported by the finding that engaging in
mobile self-tracking led to heightened self-determination, consisting of autonomy,
competence, and relatedness, that theoretically allows internalization to occur and
produces health outcomes (Rooksby, Rost, Morrison, & Chalmers, 2015).
In addition to significant findings on self-determination as a mediator, this
research also revealed a moderating effect of ecological momentary intervention on the
relationship between mobile self-tracking engagement and healthy eating. Literature
review has suggested that the effectiveness of mobile health programs on weight
management is generally associated with interactivity, tailored messages, real-time
feedback, and cues to action, which can be further characterized as ecological momentary

143
intervention (Breton et al., 2011; Shaw & Bosworth, 2012). The study empirically
validated it by examining under what conditions mobile self-tracking engagement lead to
better health outcomes. The results of MANOVA have shown that health anxiety and
healthy eating differed between high- and low-EMI groups. EMI itself accounted for a
significant amount of additional variance in health anxiety and healthy eating.
Nonetheless, results of the hierarchical regression analyses have shown that EMI and
mobile self-tracking engagement only interacted in predicting healthy eating. That is,
adding the interaction term into the regression model only resulted in a statistically
significant change in the explained variance (R
2
) for healthy eating. Mobile self-tracking
engagement was found to be a more important predictor of healthy eating for high EMI
than low-EMI mobile self-trackers. Findings suggest that ecological momentary
intervention may be particularly useful in assisting and promoting healthy eating as it
especially requires constant monitoring. EMI may not be equally important in producing
physical activity and other kinds of psychological health outcomes.
Another interesting finding revealed in this research is the sources of social
influence on mobile self-tracking behavior. It was found that network exposure, strength
of tie, and role modeling are key factors in social influence. First, consistent with our
expectation, mobile self-tracking exposure significantly predicted mobile self-tracking
adoption. That is, the more mobile self-trackers one is exposed to in his or her social
network, the more likely one will imitate the behavior. In addition, relation type and
interaction frequency significantly predicted same self-tracking behavior. Specifically,
colleague/schoolmate and celebrity are the main sources of influence on self-tracking.
This might be the reason that workplaces and schools are the places where adults and

144
college students spend the most time during a day so they are exposed to and observe
new behavior from colleagues and schoolmates the most. Besides, celebrity as a
significant source of social influence highlights the importance of mass media,
entertainment channels, and social media, in representing and promoting mobile self-
tracking. This suggests that mobile health promotion may start from workplace and mass
media as individuals were found to be heavily influenced by their colleagues,
schoolmates, and celebrities.
Further, self-tracking behavior is likely to be influenced by social ties whom a
person contacts more frequently, but not necessarily by those who one has known longer.
This makes sense as frequent communication allows persuasion to occur, while years
known does not guarantee trust and closeness between people. Role modeling was also
shown to be a significant factor in predicting self-tracking adoption. It was found that
same self-tracking behavior (e.g., both the participant and the alter were mobile self-
trackers, or both were traditional self-trackers) was more likely to occur between a person
and an alter who is perceived as attractive and healthy by the person. This finding is in
line with underlying mechanisms in persuasion theories (Cialdini, 2009), and has
commercial implications for promoting mobile self-tracking. For instance, choosing an
attractive spokesperson or building an image of pursuing wellness as opposed to
managing illness is likely to appeal to potential users and inspire a greater mobile self-
tracking adoption rate. It is worth noting that similarity was not a significant predictor of
self-tracking behavior. Contrary to some social network studies, same-aged, same-
gendered, and same ethnicity did not predict same self-tracking behavior. It might be the
reason that demographic similarities are not necessarily required for assimilation in the

145
context of mobile self-tracking. Instead, as revealed earlier, individuals are more likely to
be influenced through exposure and vicarious learning in their networks (i.e., social
rewards and punishments in the forms of praise, compliments, and popularity, etc.). This
finding also provides evidence that mobile self-tracking has probably penetrated the
population regardless of age, gender, and ethnicity, as the behavior is not particularly
replicated within same demographic groups.
Finally, the structural equation modeling confirmed the theory-driven research
model, suggesting the model is applicable for both mobile self-trackers and traditional
self-trackers. It is observed that motivation played a relatively more essential role than
social influence in affecting mobile self-tracking engagement. Among several sub-
elements of motivation, self-efficacy weighted more than other factors. This has
implications for mobile health promotion. For instance, interventions incorporating a self-
tracking component should emphasize acquiring self-efficacy, building self-esteem, and
providing training programs in order to increase individuals’ autonomous motivation to
take advantage of mobile self-tracking. Another interesting finding observed in this
research is the path-by-path comparison between mobile self-trackers and traditional self-
trackers. Results from the SEM model have shown that mobile self-tracking engagement
was effective in facilitating adaptive behavior and enhancing self-determination, and in
turn led to positive psychology, however, the same effects were not observed among
traditional self-trackers. It has thus become apparent from the present study that mobile
self-tracking, compared with traditional self-tracking, advances the tracking experience
by empowering the users in terms of autonomy, ability, connectivity, and triggers to
action, which are theoretically required for behavior change and health (Ewart, 2009;

146
Ryan, Patrick, Deci, & Williams, 2008). These components might explain why enhanced
self-determination and adaptive behavior acquired can be further translated into a sense
of fulfilling among mobile self-trackers but not among traditional self-trackers.
To sum up, the present study is one of the initial attempts seeking to understand,
describe, and determine the impact of self-guided mobile self-tracking. Mobile
technology-assisted self-tracking is shown to hold great promise in self-guided weight
management by producing positive health outcomes through self-determination and
adaptive behavior. It is also shown to be a more effective approach than other non-mobile
self-tracking approaches to health. With this in mind, we are more confident in
encouraging the use of mobile self-tracking for weight management among the public,
particularly among those who have less access to health care.

147
CHAPTER SEVEN: CONTRIBUTIONS AND LIMITATIONS
The current study contributes to the field of communication and mobile health in
several ways. Firstly, findings in the present study help build upon health behavior
theories by bringing new perspectives and integrating underlying theories in social
psychology and public health. Specifically, the study contributes to the field of mobile
health by integrating four major theories into a new model of mobile self-tracking and
empirically validating it. Scholars in communication have suggested a need for
recognition of the similarity of theoretical constructs in health behavior theories through
empirical validation across theories. Described by Noar and Zimmerman (2005), a
problem in the area of health behavior research is “no clear or conclusive findings on
what models, variables are more influential than others, nor do we know what behaviors
or situations are understood better than others” (p.324). A lack of consensus requires us
to “refine theory” and “move toward consensus in the field where possible” (Noar &
Zimmerman, 2005, p.287). In the same vein, Weinstein (1993) indicated a lack of
empirical comparisons across health behavior theories, which leads to fragmentation of
the area and hinders the growth and progress in cumulative knowledge. Noar and
Zimmerman (2005) suggest that the “most compelling direction to move in” would be
comparing, testing, and integrating theories (p.281). These can be done by, for example,
testing more than one theory in a study, using regression, SEM or meta-analysis to test
the theory as a whole, or by testing multiple dependent variables in a study, etc. (Noar &
Zimmerman, 2005, p. 281). This study responded to the call for integration of theories in
health communication by integrating and empirically testing key concepts across major
theories in a new model of mobile self-tracking. This allows us to better understand the

148
big picture of mobile self-tracking and theorize on mobile health practices.
Secondly, the present study adds insight into the mobile health industry and
regulation. For example, by learning about the relationships between individual
characteristics, their differential mobile self-tracking behaviors, and the subsequent
effects, software development companies could know how to best address the needs and
wants of different kinds of self-trackers to help achieve their weight management goals
more efficiently and effectively. For example, it was found that mobile self-trackers who
were using mobile phones for tracking activities had significantly higher healthy eating
than mobile self-trackers who were tracking with wearables. This suggests that
wearables, although providing ubiquitous sensing functions, might not be as convenient
as mobile phones in assisting healthy eating practices, which involve logging foods, diets,
calories, and so forth. This finding implies a possible direction for improvement in
developing wearables.
In addition, findings from the current study offer a rich description of the
antecedents, predicators, mediator, and moderator of mobile self-tracking behavior,
which aid in the development of future health programs and interventions that address the
use of mHealth to enhance public health. For instance, the current study takes into
consideration health-oriented, technology-oriented, cognition/personality-driven, and
socially driven factors, and it was shown that cognition/personality-oriented factors
outweighed other factors in the model of mobile self-tracking engagement. This provides
a clear direction for health professionals and the mobile industry in promoting and
marketing mobile self-tracking for health.

149
On top of it, inasmuch as mobile self-tracking is not currently well regulated, the
quality, efficacy, and potential harm of health-related mobile applications are subject to
question despite the normalization of self-tracking practices by advocates (Pogorelc,
2013). The present study empirically validated the effects of self-guided mobile self-
tracking and compared effects between different self-tracking approaches to weight
management. Results are inspiring as they have shown that the popularity of self-tracking
movement is not simply a bandwagon effect followed by the normalization of self-
tracking practices in the media; instead, mobile self-tracking is shown to be beneficial
and conducive to health, and can actually help reduce health anxiety individuals have
about their health concerns.
Limitations
Of course, the present study has several limitations. Firstly, the results were based
on a cross-sectional survey, which limits its ability to examine the dynamic and long-term
effects of mobile self-tracking for health. Although the questionnaire used in the study
was carefully designed and reviewed to elicit what the researcher intended to ask, the
nature of the design prohibits it from looking at the temporal sequence of constructs.
Mobile self-tracking could be a fad, and hence its long-term effects require further
examination. Future studies should consider conducting a longitudinal study or
incorporating the model for stages of change into the instruments to produce richer
findings on the temporal relationships between key concepts.
Secondly, the causal inferences can be further strengthened by adding objective
measures such as mobile outputs and medical records to avoid the problem of solely

150
relying on self-reported data. In addition, the sample in the present study was collected
through purposive sampling using Qualtrics Panel. It is not a random sampling method
and thus the results may not be as representative as is random sampling in terms of
generalizability. However, purposive sampling is to focus on the particular characteristics
of a population that are of interest, which will best enable the researcher to answer the
research questions. As the topic in question concerns the issue of health and weight
management, a purposive sampling technique through the Qualtrics Health Panel is
considered appropriate. Future studies may consider using random sampling to obtain
results that more accurately reflect diverse demographic information, and to compare
findings on mobile self-tracking with different sampling techniques. In addition, the pre-
determined quota for each groups may have prohibited the study from answering the
“prevalence” question about mobile self-tracking; however, this approach satisfied the
statistical requirement for producing accurate estimates and employing appropriate
analytical strategies to explore the addressed research questions.
Furthermore, the present study does not particularly look into gender- and
ethnicity-related issues associated with mobile self-tracking given the majority in the
studied sample were female and Caucasians. Future research on mobile self-tracking can
take into account cultural perspectives and gender differences, and can further examine
associated factors as well as self-tracking practices in different populations. Adding
gender and cultural perspectives into mobile health applications would yield more
interesting findings and advance our understanding about mobile health practices.
Last but not least, future research on mobile self-tracking can extend this project
by adding qualitative components to examine discourses about mobile self-tracking in

151
multiple media landscapes, such as normalization of self-tracking, narcissism, the self
and the body in mass media and on social media, etc. A mix-method approach will yield
richer insights into the behavior, and will allow us to explore mobile self-tracking with a
more critical lens.

152
REFERENCES
Ajzen, J. (2002). Perceived behavioral control, self-efficacy, locus of control and the
theory of planned behavior. Journal of Applied Social Psychology, 32(4), 665-
683.
Anderson, J., & Gerbing, D. (1988). Structural equation modeling in practice: A review
and recommended two-step approach. Psychological bulletin, 103(3), 411-423.
Armstrong, A. W., Watson, A. J., Makredes, M., Frangos, J. E., Kimball, A. B., &
Kvedar, J. C. (2009). Text-message reminders to improve sunscreen use: a
randomized, controlled trial using electronic monitoring. Arch Dermatol, 145(11),
1230-1236. doi: 10.1001/archdermatol.2009.269
Bagozzi, R. P. (1983). Issues in the application of covariance structure analysis: A further
comment. Journal of Consumer Research, 9, 449- 450.
Baker, R. C., & Kirschenbaum, D. S. (1993). Self-monitoring may be necessary for
successful weight control. Behavior Therapy, 24, 377-394.
Bandura, A. (1986a). Social foundations of thought and action: A social cognitive theory.
Englewood Cliffs, NJ: Prentice-Hall.
Bandura, A. (1989). Social cognitive theory. In R. Vasta (Ed.), Annals of child
development, 6. Six theories of child development (pp. 1-60). Greenwich, CT: JAI
Press.
Bandura, A. (1991a). Self-efficacy mechanism in physiological activation and health-
promoting behavior. In J. Madden, IV (Ed.), Neurobiology of learning, emotion
and affect (pp. 229-270). New York: Raven, 1991.
Bandura, A. (1991b). Self-regulation of motivation through anticipatory and self-
regulatory mechanisms. In R. A. Dienstbier (Ed.), Perspectives on motivation:
Nebraska symposium on motivation (Vol. 38, pp. 69-164). Lincoln: University of
Nebraska Press.
Bandura, A. (1992). Social cognitive theory and social referencing. In S. Feinman (Ed.),
Social referencing and the social construction of reality in infancy (pp. 175-208).
New York: Plenum Press.
Bandura, A. (2002). Growing primacy of human agency in adaptation and change in the
electronic era. European Psychologist, 7 (1), 2-16.
Bandura, A. (2004). Health promotion by social cognitive means. Health Educ Behav,
31(2), 143-164. doi: 10.1177/1090198104263660

153
Berkowitz, A.D. (2004). The social norms approach: Theory, research, and annotated
bibliography. Retrieved from
http://www.alanberkowitz.com/articles/social_norms.pdf
Breton, E. R., Fuemmeler, B. F., & Abroms, L. C. (2011). Weight loss-there is an app for
that! But does it adhere to evidence-informed practices? Transl Behav Med, 1(4),
523-529. doi: 10.1007/s13142-011-0076-5
Broeck, A., Vansteenkiste, M., De Witte, H., Soenens, B., & Lens, W. (2010). Capturing
autonomy, competence, and relatedness at work: Construction and initial
validation of the Work-related Basic Need Satisfaction scale. Journal of
Occupational and Organizational Psychology, 83, 981-1002.
Butterfield, A.D. (2012). Ethnographic assessment of quantified self meetup groups.
Master Thesis. San JOSÉ State University.
Burt, R. S. (1984). Network items and the general social survey. Social Networks, 6, 293-
339.
Carver, C. S., Scheier, M. F., & Weintraub, J. K. (1989). Assessing coping strategies: A
theoretically based approach. Journal of Personality and Social Psychology, 56,
267-283.
Carver, C. S., & Connor-Smith, J. (2010). Personality and coping. Annual Review of
Psychology, 61, 679-704. doi: 10.1146/annurev.psych.093008.100352.
Cavallo, D. N., Tate, D. F., Ries, A. V., Brown, J. D., DeVellis, R. F., & Ammerman, A.
S. (2012). A social media-based physical activity intervention: a randomized
controlled trial. Am J Prev Med, 43(5), 527-532. doi:
10.1016/j.amepre.2012.07.019
Chen, B., Van Assche, J., Vansteenkiste, M., Soenens, B. & Beyers, W. (2015). Does
psychological need satisfaction matter when environmental or financial safety are
at risk? Journal of Happiness Studies, 16(3), 745-766.
Chen, B., Vansteenkiste, M., Bayers, W., Boone, L., Deci, E. L., Van der Kaap-Deeder,
J.,… & Verstuyf, J. (2015). Basic psychological need satisfaction, need
frustration, and need strength across four cultures. Motivation and Emotion, 39,
216-236.
Cheung, G. W., & Rensvold, R. B. (2002). Evaluting Goodness-of-fit for testing
measurement invariance. Structural Equation Modeling, 9 (2), 233-255.
Chin, W.W. (1998). The partial least squares approach to structural equation modeling. In
G.A. Marcoulides [ed.]. Modern Methods for Business Research (pp. 295-336).
Mahwah, NJ: Lawrence Erlbaum Associates.

154
Christakis, N. A., & Fowler, J. H. (2013). Social contagion theory: examining dynamic
social networks and human behavior. Stat Med, 32(4), 556-577. doi:
10.1002/sim.5408
Chuang, Y. H., & Tsao, C. W. (2013). Enhancing nursing students' medication
knowledge: The effect of learning materials delivered by short message service.
Computers & Education, 61, 168-175. doi: 10.1016/j.compedu.2012.09.013
Cialdini, R. (2009). Influence: Science and Practice, ePub (5th Edition). Pearson HE,
Inc.. Kindle Edition.
Clason, D. L., & Dormody, T. J. (1994) Analyzing data measured by individual Likert-
type items. Journal of Agricultural Education, 35(4), 31- 35.
Cole-Lewis, H., & Kershaw, T. (2010). Text messaging as a tool for behavior change in
disease prevention and management. Epidemiol Rev, 32(1), 56-69. doi:
10.1093/epirev/mxq004
Correlated error terms. (2007, April 19). Retrieved from
http://zencaroline.blogspot.com/2007/04/correlated-error-terms.html
Crisostomo, A. (2013). The quest for happiness in self-tracking mobile technology.
Master's Thesis. University of Amsterdam.
Definition of mHealth. (January 5, 2012). Mobile Health IT. Retrieved from
http://www.himss.org/ResourceLibrary/GenResourceDetail.aspx?ItemNumber=2
0221
Depp, C. A., Mausbach, B., Granholm, E., Cardenas, V., Ben-Zeev, D., Patterson, T. L., .
. . Jeste, D. V. (2010). Mobile interventions for severe mental illness: design and
preliminary data from three approaches. J Nerv Ment Dis, 198(10), 715-721. doi:
10.1097/NMD.0b013e3181f49ea3
Diener, E., Wirtz, D., Tov, W., Kim-Prieto, C., Choi. D., Oishi, S., & Biswas-Diener, R.
(2010). New well-being measures: Short scales to assess flourishing and positive
and negative feelings. Social Indicators Research, 97 (2), 143-156.
Dutta, M. J., & Feng, H. (2007). Health Orientation and Disease State as Predictors of
Online Health Support Group Use. Health Communication, 22(2), 181-189.
Dutta-Bergman, M. J. (2004a). An alternative approach to social capital- exploring the
linkage between health consicousness. Health Communication, 16(4), 393-409.
Dutta-Bergman, M. J. (2004b). Health attitudes, health cognitions, and health behaviors
among Internet health information seekers: population-based survey. J Med
Internet Res, 6(2), e15. doi: 10.2196/jmir.6.2.e15

155
Dutta-Bergman, M. J. (2005). Developing a profile of consumer intention to seek out
additional information beyond a doctor: The role of communicative and
motivation variables. Health Communication, 17(1), 1-16.
Eichhorn, B. R. (2014). Common Method Variance Techniques. Retrieved from
http://www.mwsug.org/proceedings/2014/AA/MWSUG-2014-AA11.pdf
Ewart, C. (1991). Social action theory for a public health psychology. American
Psychology, 46(9), 931-946.
Ewart, C. (2009). Changing our unhealthy ways: Emerging perspectives from social
action theory. DiClemente, Ralph, Crosby, Richard, Kegler, Michelle (Eds).
Emerging Theories in Health Promotion Practice and Research. (pp. 359-386).
(2nd ed.). San Francisco: Josseys-Bass.
Fishbein, M., & Ajzen, I. (1975). Belief, Attitude, Intention, and Behavior: An
Introduction to Theory and Research. Reading, MA: Addison-Wesley.
Fjeldsoe, B. S., Miller, Y. D., & Marshall, A. L. (2010). MobileMums: a randomized
controlled trial of an SMS-based physical activity intervention. Ann Behav Med,
39(2), 101-111. doi: 10.1007/s12160-010-9170-z.
Fornell, C., & Larcker, D. (1981). Evaluating structural equation models with
unobservable variables and measurement error. Journal of marketing research,
18, 39-50.
Fortier, M. S., Williams, G.C., Sweet, S. N., & Patrick, H. (2009). Self-determination
theory: Process models for health behavior change. In R. J. DiClemente, R. A.
Crosby, & M. C. Kegler (Eds), Health promotion, practice, and research (pp.
157-184). San Francisco, Jossey-Bass.
Fox, S. & Duggan, M. (2013). Tracking for health. Pew Research Internet Project.
Retrieved from http://www.pewinternet.org/Reports/2013/Tracking-for-
Health.aspx
Free, C., Phillips, G., Watson, L., Galli, L., Felix, L., Edwards, P., . . . Haines, A. (2013).
The effectiveness of mobile-health technologies to improve health care service
delivery processes: a systematic review and meta-analysis. PLoS Med, 10(1),
e1001363. doi: 10.1371/journal.pmed.1001363
Free, C., Whittaker, R., Knight, R., Abramsky, T., Rodgers, A., & Roberts, I. G. (2009).
Txt2stop:A pilot randomised controlled trial of mobile phone-based smoking
cessation support. Tob Control, 18(2), 88-91.
Gao, S., Mokhtarian, P. L., & Johnston, R. A. (2007). Exploring the causal connections
among job accessibility, employment, income, and auto ownership with structural
equation modeling. The Annals of Regional Science, 42 (2), 341-356.

156
Gerber, B. S., Stolley, M. R., Thompson, A. L., Sharp, L. K., & Fitzgibbon, M. L. (2009).
Mobile phone text messaging to promote healthy behaviors and weight loss
maintenance: a feasibility study. Health Informatics J, 15(1), 17-25. doi:
10.1177/1460458208099865
George, D., & Mallery, M. (2010). SPSS for Windows Step by Step: A Simple Guide and
Reference (10a ed.). Boston: Pearson.
Gill, J. (2001). Generalized Linear Models: A Unified Approach. Thousand Oaks, CA:
Sage Publications.
Glanz, K., & Schwartz, M. (2008). Stress, coping and health behavior. In K. Glanz, B.
Rimer, & K. Viswanath (Eds.), Health Behavior and Health Education: Theory,
Research and Practice (pp. 210-236). San Francisco: Jossey Bass.
Gold, J., Aitken, C. K., Dixon, H. G., Lim, M. S., Gouillou, M., Spelman, T., . . . Hellard,
M. E. (2011). A randomised controlled trial using mobile advertising to promote
safer sex and sun safety to young people. Health Educ Res, 26(5), 782-794. doi:
10.1093/her/cyr020
Gould, S. J. (1988). Consumer attitudes toward health and health care: A differential
perspective. Journal of Consumer Affairs, 22(1), 96-118.
Grace-Martin, K. (n.d.). When assumptions of ANCOVA are irrelevant. The Analysis
Factor. Retrieved from http://www.theanalysisfactor.com/assumptions-of-ancova/
Granovetter (1973).The strength of weak ties. American Journal of Sociology, 78 (6),
1360-1380.
Grimes, A., Kantroo, V., & Grinter, R. E. (2010). Let's Play! Mobile health games for
adults. Paper presented at the UbiComp’10, Copenhagen, Denmark.
Guse, K., Levine, D., Martins, S., Lira, A., Gaarde, J., Westmorland, W., & Gilliam, M.
(2012). Interventions using new digital media to improve adolescent sexual
health: a systematic review. J Adolesc Health, 51(6), 535-543. doi:
10.1016/j.jadohealth.2012.03.014.
Guthridge, L. (April 8, 2013). Improving your self-awareness through self-tracking.
Retrieved from http://connectconsultinggroup.com/improve-your-self-awareness-
through-self-tracking/
Haapala, I., Barengo, N. C., Biggs, S., Surakka, L., & Manninen, P. (2009). Weight loss
by mobile phone: a 1-year effectiveness study. Public Health Nutr, 12(12), 2382-
2391. doi: 10.1017/S1368980009005230
Hair, J. F., Black, W. C., Babin, B. J., Anderson, R. E., & Tatham, R. L. (2006).
Multivariate data analysis (6th ed.). New Jersey : Prentice-Hall.

157
Haug, S. (2013). Mobile phone text messaging to reduce alcohol and tobacco use in
young people – a narrative review. Smart Homecare Technology and
TeleHealth, 11. doi: 10.2147/shtt.s43222

Healthy eating for a healthy weight. (n.d.). Centers for Disease Control and Prevention.
Retrieved from http://www.cdc.gov/healthyweight/healthy_eating/index.html

Heaney, C.A., & Israel, B.A. (2008). Social networks and social support. In Glanz K,
Rimer BK, Viswanath K, Eds. (4
th
ed). Health Behavior and Health Education:
Theory, Research, and Practice (pp.189-210). San Francisco: Jossey-Bass.
Heron, K. E., & Smyth, J. M. (2010). Ecological momentary interventions: incorporating
mobile technology into psychosocial and health behaviour treatments. Br J Health
Psychol, 15(Pt 1), 1-39. doi: 10.1348/135910709X466063
Hong, H. (2009). Scale development for measuring health consciousness: Re-
conceptualization. Paper presented at the 12th Annual International Public
Relations Research Conference: Research That Matters to the Practice, FL:
University of Miami.
Hong, H. (2011). An extension of the extended parallel process model (EPPM) in
television health news: the influence of health consciousness on individual
message processing and acceptance. Health Commun, 26(4), 343-353. doi:
10.1080/10410236.2010.551580
Huberty, C. J., & Morris, J. D. (1989). Multivariate analysis versus multiple univariate
analyses Psychological Bulletin, 105(2), 302-308.
Hu, C. (2013). A new measure for health consciousness: Development of a health
consciousness conceptual model. Paper presented at the 99
th
Annual Conference
of the National Communication Association, Washington, DC. November 21-24,
2013.
Hu. C. (2014). Health slacktivism on social media: Effects and predictors. Conference
proceedings of the Human Computer Interaction International (HCI) 2014, June
22-27, 2014. Crete, Greece.
Hu, C. (2016). Understanding health slacktivism: A structural equation modeling
approach to health slacktivism on social media. Paper presented at the 66
th

International Communication Association Annual Conference, Fukuoka, Japan. June
9-13, 2016.
Hu, X., Bell, R. A., Kravitz, R. L., & Orrange, S. (2012). The prepared patient:
information seeking of online support group members before their medical
appointments. J Health Commun, 17(8), 960-978. doi:
10.1080/10810730.2011.650828

158
Iversen, A. C., & Kraft, P. (2006). Does socio-economic status and health consciousness
influence how women respond to health related messages in media? Health Educ
Res, 21(5), 601-610. doi: 10.1093/her/cyl014
Jackson, E. S., Tucker, C. M., & Herman, K. C. (2007). Health value, perceived social
support, and health self-efficacy as factors in a health-promoting lifestyle. J Am
Coll Health, 56(1), 69-74. doi: 10.3200/JACH.56.1.69-74

Jayanti, R. K., & Burn, A. C. (1998). The antecedents of preventive health care behavior:
An empirical studies. Journal of the Academy of Marketing Science, 26(1), 6-15.
Jöreskog, K., & Sörbom, D. (1993). LISREL 8: Structural equation modeling with the
SIMPLIS command language. Chicago, IL: Scientific Software International Inc.
Joo, N. S., & Kim, B. T. (2007). Mobile phone short message service messaging for
behaviour modification in a community-based weight control programme in
Korea. J Telemed Telecare, 13(8), 416-420. doi: 10.1258/135763307783064331
Kaynak, R., & Eksi, S. (2011). Ethnocentrism, Religiosity, Environmental and Health
Consciousness: Motivators for Anti-Consumers. Eurasian Journal of Business
and Economics, 4(8), 31-50.
Klasnja, P., & Pratt, W. (2012). Healthcare in the pocket: mapping the space of mobile-
phone health interventions. J Biomed Inform, 45(1), 184-198. doi:
10.1016/j.jbi.2011.08.017
Kline, R. B. (2005). Principles and Practice of Structural Equation Modeling (2nd ed.).
New York: Guilford
Kraft, F. B., & Goodell, P. W. (1993). Identifying the health conscious comsumer.
Journal of Health Care Marketing, 13(3), 18-25.
Krishna, S., Boren, S. A., & Balas, E. A. (2009). Healthcare via cell phone: A systematic
review. Telemed J E Health, 15(3), 231-240.
Kristjansdottir, O. B., Fors, E. A., Eide, E., Finset, A., Stensrud, T. L., van Dulmen, S., . .
. Eide, H. (2013). A smartphone-based intervention with diaries and therapist-
feedback to reduce catastrophizing and increase functioning in women with
chronic widespread pain: randomized controlled trial. J Med Internet Res, 15(1),
e5. doi: 10.2196/jmir.2249
Kriwy, P., & Mecking, R.-A. (2012). Health and environmental consciousness, costs of
behaviour and the purchase of organic food. International Journal of Consumer
Studies, 36(1), 30-37. doi: 10.1111/j.1470-6431.2011.01004.x

159
Krohne, H. W. (1989). The concept of coping modes: Relating cognitive person variables
to actual coping behavior. Advances in Behaviour Research and Therapy, 11,
235-248.
Krohne, H. W. (1993). Vigilance and cognitive avoidance as concepts in coping research.
In H. W. Krohne (Ed.), Attention and avoidance: Strategies in coping with
aversiveness (pp. 19-50). Seattle,WA: Hogrefe & Huber.
Krohne, H. W. (1996). Individual differences in coping. In M. Zeidner & N. S. Endler
(Eds.), Handbook of coping: Theory, research, applications (pp. 381-409). New
York: Wiley.
Lee, W., Chae, Y. M., Kim, S., Ho, S. H., & Choi, I. (2010). Evaluation of a mobile
phone-based diet game for weight control. J Telemed Telecare, 16(5), 270-275.
doi: 10.1258/jtt.2010.090913
Lazarus, R. S., & Folkman, S. (1984). Stress, appraisal, and coping. New York:
Springer.
Li, I., Dey, A. K., & Forlizzi, J. (2011). Understanding my data, myself: Supporting self-
reflection with ubicomp technologies. Proceedings from UbiComp '11: The 13th
international conference on Ubiquitous computing.New York, NY: ACM.
Lim, M. S., Hocking, J. S., Aitken, C. K., Fairley, C. K., Jordan, L., Lewis, J. A., &
Hellard, M. E. (2012). Impact of text and email messaging on the sexual health of
young people: a randomised controlled trial. J Epidemiol Community Health,
66(1), 69-74. doi: 10.1136/jech.2009.100396
Lindell, M. K. & Whitney, D. J. (2001). Accounting for common method variance in
cross-sectional research design. Journal of Applied Psychology, 86 (1), 114-121.
Lu, J., Yao, J. E., & Yu, C.-S. (2005). Personal innovativeness, social influences and
adoption of wireless Internet services via mobile technology. The Journal of
Strategic Information Systems, 14(3), 245-268. doi: 10.1016/j.jsis.2005.07.003
Lupton, D. (2013a). Living the quantified self: The realities of self-tracking for health.
Retrieved from http://simplysociology.wordpress.com/2013/01/11/living-the-
quantified-self-the-realities-of-self-tracking-for-health/
Lupton, D. (2013b). The rise of the quantified self as a cultural phenomenon. Retrieved
from http://simplysociology.wordpress.com/2013/08/13/the-rise-of-the-
quantified-self-as-a-cultural-phenomenon/
Macdonald-Wallis, K., Jago, R., & Sterne, J. A. (2012). Social network analysis of
childhood and youth physical activity: A systematic review. American Journal of
Preventive Medicine, 43 (6), 636-642. doi: 10.1016/j.amepre.2012.08.021.

160
Macera, C.A. (2003). Promoting healthy eating and physical activity for a healthier
nation. In Centers for Disease Control and Prevention, Promising practices in
chronic dis- ease prevention and control: A public health framework for action
(pp. 7-1 through 7-22). Atlanta, GA: Centers for Disease Control and Prevention.
Mackie, G., Moneti, F., Denny, E., & Shakya, H. (2015). What are social norms? How
they are measured? UNICEF/UCSD Center on Global Justice Project
Cooperation Agreement.
Majchrzak, A., Beath, C. M., Lim, R. A., & Chin, W. W. (2005). Managing client
dialogues during information systems design to facilitate client learning. MIS
Quarterly, 29(4), 653-672.
Mayers, A. (2013). Introduction to statistics and SPSS in psychology. Harlow, United
Kingdom: Pearson.
Maruyama, G. (1998). Basics of Structural Equation Modeling. Thousand Oaks CA:
Sage.
McAlister, A. L., Perry, C. L., & Parcel, G. S. (2008). How individ- uals, environments,
and health behaviors interact: Social cogni- tive theory. In K. Glanz, B. K. Rimer,
& K. Viswanath (Eds.), Health behavior and health education: Theory, research,
and practice (4th ed., pp. 169-188). San Francisco, CA: Jossey- Bass.
McAuley, E., Duncan, T., & Tammen, V. V. (1989). Psychometric properties of the
Intrinsic Motivation Inventory in a competitive sport setting: A confirmatory
factor analysis. Research Quarterly for Exercise and Sport, 60, 48-58.
McGillicuddy, J. W., Gregoski, M. J., Weiland, A. K., Rock, R. A., Brunner-Jackson, B.
M., Patel, S. K., . . . Treiber, F. A. (2013). Mobile Health Medication Adherence
and Blood Pressure Control in Renal Transplant Recipients: A Proof-of-Concept
Randomized Controlled Trial. JMIR Res Protoc, 2(2), e32. doi:
10.2196/resprot.2633

McTavish, F. M., Chih, M. Y., Shah, D., & Gustafson, D. H. (2012). How Patients
Recovering From Alcoholism Use a Smartphone Intervention. J Dual Diagn,
8(4), 294-304. doi: 10.1080/15504263.2012.723312

Miller, J. A. (2014). The future of mHealth goes well beyond fitness apps. CIO.
Retrieved from http://www.cio.com/article/2855047/healthcare/the-future-of-
mhealth-goes-well-beyond-fitness-apps.html
Miller, S. M. (1987). Monitoring and blunting: Validation of a questionnaire to assess
styles of information seeking under threat. Journal of Personality and Social
Psychology, 52(2), 345-353.

161
Montano, D.E., & Kasprzyk. (2008). Theory of reasoned action, theory of planned
behavior, and the integrated behavioral model. In Glanz K, Rimer BK, Viswanath
K, Eds. (4
th
ed). Health Behavior and Health Education: Theory, Research, and
Practice. San Francisco: Jossey-Bass. pp 67-96.
Moos, R. H., & Schaefer, J. A. (1993). Coping resources and processes: Current concepts
and measures. In: Goldberger L, Breznitz S, editors. Handbook of stress:
Theoretical and clinical aspects (pp. 234–257). New York: Free Press.
Multivariate analysis of covariance (MANCOVA). (n.d.). Statistics Solutions. Retrieved
from http://www.statisticssolutions.com/multivariate-analysis-of-covariance-
mancova/
Mulvaney, S. A., Ritterband, L. M., & Bosslet, L. (2011). Mobile intervention design in
diabetes: review and recommendations. Curr Diab Rep, 11(6), 486-493. doi:
10.1007/s11892-011-0230-y
Newton, K. H., Wiltshire, E. J., & Elley, C. R. (2009). Pedometers and text messaging to
increase physical activity: randomized controlled trial of adolescents with type 1
diabetes. Diabetes Care, 32(5), 813-815. doi: 10.2337/dc08-1974
Nißen, M. (2013). Quantified Self: An exploratory study on the profiles and motivations
of self-tracking. (Bachelor Thesis). Karlsruhe Institute of Technology (KIT),
Germany.
Noar, S. M., & Zimmerman, R. S. (2005) Health Behavior Theory and cumulative
knowledge regarding health behaviors: are we moving in the right direction?
Health Education Research, 20, 275–290.
Norman, C. D., & Skinner, H. A. (2006). eHealth Literacy: Essential Skills for Consumer
Health in a Networked World. J Med Internet Res, 8(2), e9. doi:
10.2196/jmir.8.2.e9
Norman, G. J., Zabinski, M. F., Adams, M. A., Rosenberg, D. E., Yaroch, A. L., &
Atienza, A. A. (2007). A review of eHealth interventions for physical activity and
dietary behavior change. Am J Prev Med, 33(4), 336-345. doi:
10.1016/j.amepre.2007.05.007
Norman, P., Bennett, P., Smith, C., & Murphy, S. (1998). Health locus of control and
health behaviour. J Health Psychol, 3(2), 171-180. doi:
10.1177/135910539800300202

Norris, J. (2012, October 5). Self-tracking may become key element of personalized
medicine. University of California San Francisco. Retrieved from
https://www.ucsf.edu/news/2012/10/12913/self-tracking-may-become-key-
element-personalized-medicine

162
Paluck, E. L., & Ball, L. (2010). Social norms marketing aimed at gender based violence:
A literature review and critical assessment. New York: International Rescue
Committee. Retrieved from
https://static1.squarespace.com/static/5186d08fe4b065e39b45b91e/t/52d1f24ce4b
07fea759e4446/1389490764065/Paluck+Ball+IRC+Social+Norms+Marketing+L
ong.pdf.
Pender, N. (2011). The health promotion model manual. Retrieved
https://deepblue.lib.umich.edu/bitstream/handle/2027.42/85350/HEALTH_PROM
OTION_MANUAL_Rev_5-2011.pdf

Paddock, C. (2013, August 15). How self-monitoring is transforming health. Medical
News Today. Retrieved from
http://www.medicalnewstoday.com/articles/264784.php

Pagoto, S., Schneider, K., Jojic, M., DeBiasse, M., & Mann, D. (2013). Evidence-based
strategies in weight-loss mobile apps. Am J Prev Med, 45(5), 576-582. doi:
10.1016/j.amepre.2013.04.025
Patel, S., & Bhavsar, C.D. (2013). Analysis of pharmacokinetic data by wilk's lambda
(An important tool of MANOVA). International Journal of Pharmaceutical
Science Invention, 2 (1), 36-44.
Patty, R.E., & Cacioppo, J. T. (1986a). The elaboration likelihood model of persuasion.
In L. Berkowitz (Ed.), Advances in experiemental social psychology, 19, 123-205
. New York: Academic Press.
Patty, R.E., & Cacioppo, J. T. (1986b). Communication and persuasion: Central and
peripheral routes to attitude change. New York: Springer-Verlag.
Peterson, J. L. Rothernberg, R., Kraft, J. M., Beeker, C., & Trotter, R. (2007). Perceived
condom norms and HIV risks among social and sexual networks of young African
American men who have sex with men. Health Education Research, 24(1), 119-
127.
Patrick, K., Raad, F., & Morman, G. J. (2009). A text message-based intervention for
weight loss: Randomized controlled trial. Journal of Medical Internet Research,
11(1): e1. doi: 10.2196/jmir.1100.
Pender, N. (2011). The health promotion model manual. Retrieved from
http://deepbluelib.umich.edu/bitstream/2027.42/85350/1/heal
Pedersen, S.S., Spinder, H., Erdman, R M. & Denollet, J. (2009). Poor perceived social
support in implantable cardioverter defibrillator (ICD) patients and their partners:
Cross-validation of the multidimensional scale of perceived social support.
Psychosomatics, 50(5), 461-467.

163
Pogorelc, D. (2013, January 1). Are we headed toward the over-quantified self? MedCity
News. Retrieved from http://medcitynews.com/2013/01/are-we-headed-toward-
the-over-quantified-self/
Rai, A., Chen, L., Pye, J., & Baird, A. (2013). Understanding determinants of consumer
mobile health usage intentions, assimilation, and channel preferences. J Med
Internet Res, 15(8), e149. doi: 10.2196/jmir.2635
Ratelle, C. F., Guay, F. & Vallerand, R. J. (2007). Autonomous, controlled, and
amotivated types of academic motivation: A person-oriented analysis. Journal of
Educational Psychology, 99 (4), 734- 746.
Reid, S. C., Kauer, S. D., Hearps, S. J., Crooke, A. H., Khor, A. S., Sanci, L. A., &
Patton, G. C. (2011). A mobile phone application for the assessment and
management of youth mental health problems in primary care: a randomised
controlled trial. BMC Fam Pract, 12, 131. doi: 10.1186/1471-2296-12-131
Riley, W. T., Rivera, D. E., Atienza, A. A., Nilsen, W., Allison, S. M., & Mermelstein, R.
(2011). Health behavior models in the age of mobile interventions: are our
theories up to the task? Transl Behav Med, 1(1), 53-71. doi: 10.1007/s13142-011-
0021-7
Rimal (2000). Closing the knowledge-behavior gap in health promotion: The mediating
role of self-efficacy. Health Communication, 12 (3), 219-237.
Rooksby, J., Rost, M., Morrison, A., & Chalmers, M. (2015). Pass the ball: Enforced turn
taking in activity tracking. In Proceedings of the 33rd Annual ACM Conference
on Human Factors in Computing Systems (CHI '15). ACM, New York, NY, USA,
2417-2426. doi:10.1145/2702123.2702577
Rosenstock, I. M. (1974). The health belief model and preventive health behavior. Health
Education & Behavior, 2(4), 354-386. doi: 10.1177/109019817400200405

Roth, S., Cohen, L. J. (1986). Approach, avoidance, and coping with stress. American
Psychologist, 41, 813-819.
Ryan, R. M., & Connell, J. P. (1989). Perceived locus of causality and internalization:
Examining reasons for acting in two domains. Journal of Personality and Social
Psychology, 57, 749–761.
Ryan, R. M., Plant, R. W., & O’Malley, S. (1995). Initial motivations for alcohol
treatment: Relations with patient characteristics, treatment in- volvement and
dropout. Addictive Behaviors, 20, 279–297.
Ryan, R. M., Patrick, H., Deci, E. L., & Williams, G. C. (2008). Facilitating health
behavior change and its maintenance: Interventions based on self-determination
theory. The European Health Psychologist, 10(1), 2-5.

164
Salkovskis, P.M., Rimes, K.A., Warwick, H.M.C. & Clark, D.M. (2002). The health
anxiety inventory: The development and validation of scales for the measurement
of health anxiety & hypochondriasis. Psychological Medicine, 32, 843-853.
Self-tracking. (n.d.). In Oxford Dictionaries. Retrieved April 20, 2016, from
http://www.oxforddictionaries.com/us/definition/american_english/self-tracking.
Sieber, W. J., Groessl, E. J., David, K. M., Ganiats, T. G., & Kaplan, R. M. (2008).
Quality of Well-being Self-Administered (QWB-SA) Scale User's Manual.
Retrieved from https://hoap.ucsd.edu/qwb-info/QWB-Manual.pdf.
Shapiro, J. R., Bauer, S., Andrews, E., Pisetsky, E., Bulik-Sullivan, B., Hamer, R. M., &
Bulik, C. M. (2010). Mobile therapy: Use of text-messaging in the treatment of
bulimia nervosa. Int J Eat Disord, 43(6), 513-519. doi: 10.1002/eat.20744
Shaw, R., & Bosworth, H. (2012). Short message service (SMS) text messaging as an
intervention medium for weight loss: A literature review. Health Informatics J,
18(4), 235-250. doi: 10.1177/1460458212442422
Simons-Morton, B. G., Haynie, D., & Noelcke, E. (2009). Social influences: The effects
of socialization, selection, and social normative processes on health behavior. In
R. J. DiClemente, R.A. Crosby, & M. C. Kegler (Eds.), Emerging theories in
health promotion, practice, and research (pp. 65-95). San Francisco: Jossey-Bass.
Simon-Morton, B., McLeroy, K. R., & Wendel, M. L. (2012). Self-determination theory
and motivational interviewing. In Behavior theory in health promotion, practice,
and research (pp. 235-256). London: Jones and Bartlett Learning.
Simon-Morton, B., McLeroy, K. R., & Wendel, M. L. (2012). Self-regulation and social
cognitive theory. In Behavior theory in health promotion, practice, and research
(pp. 127-152). London: Jones and Bartlett Learning.
Sirriyeh, R., Lawton, R., & Ward, J. (2010). Physical activity and adolescents: an
exploratory randomized controlled trial investigating the influence of affective
and instrumental text messages. Br J Health Psychol, 15(Pt 4), 825-840. doi:
10.1348/135910710X486889
Skinner, E.A., Edge, K., Altman, J., & Sherwood, H. (2003). Searching for the structure
of coping: A review and critique of category systems for classifying ways of
coping. Psychology Bulletin, 129, 216-269.
Solving analysis of covariance problems. (n.d.). Retrieved from
http://www.learningace.com/doc/1725755/e32a011a8078ec2d062a19fa95da8a13/
solving-analysis-of-covariance-problems
Swan, M. (2012). Crowdsourced health research studies: an important emerging
complement to clinical trials in the public health research ecosystem. J Med
Internet Res, 14(2), e46. doi: 10.2196/jmir.1988

165
Sweet, M. (Dec. 23, 2013). In these times of “data-utopia”, what questions should we be
asking about the rise of self-tracking? Retrieved from
http://blogs.crikey.com.au/croakey/2013/12/23/in-these-times-of-data-utopia-
what-questions-should-we-be-asking-about-the-rise-of-self-tracking/
Synder, M. (1974). Self-monitoring of expressive behaivor. Journal of Personality and
Social Psychology, 20 (4), 526-537.
Tanaka, J. S. (1993). Multifaceted conceptions of fit in structural equation models. In
K.A. Bollen, & J.S. Long (eds.), Testing structural equation models. Newbury
Park, CA: Sage.
Thackeray, R., Crookston, B. T., & West, J. H. (2013). Correlates of health-related social
media use among adults. J Med Internet Res, 15(1), e21. doi: 10.2196/jmir.2297
Tode, C. (2013). Mobile health app marketplace to take off, expected to reach $26B by
2017. Mobile Marketer. Retrieved from
http://www.mobilemarketer.com/cms/news/research/15023.html
Torkzadeh, G., Koufteros, X., & Pflughoeft, K. (2003). Confirmatory analysis of
computer self-efficacy. Structural Equation Modeling, 10(2), 263-275.
Turner-McGrievy, G. M., Campbell, M. K., Tate, D. F., Truesdale, K. P., Bowling, J. M.,
& Crosby, L. (2009). Pounds Off Digitally study: a randomized podcasting
weight-loss intervention. Am J Prev Med, 37(4), 263-269. doi:
10.1016/j.amepre.2009.06.010
Turner-McGrievy, G., & Tate, D. (2011). Tweets, Apps, and Pods: Results of the 6-
month Mobile Pounds Off Digitally (Mobile POD) randomized weight-loss
intervention among adults. J Med Internet Res, 13(4), e120. doi:
10.2196/jmir.1841
Valente, T. W. (1996). Social network thresholds in the diffusion of innovations. Social
Networks, 18, 69-89.
Valente, T.W., Watkins, S.C., Jato, M. N., Der Straten, A. V., & Tsitsol, L-P. M. (1997).
Social network associations with contraceptive use among Cameroonian women
in voluntary associations. Social Science & Medicine, 45(5), 677-687.
Valente, T.W. & Saba, W.P. (2001). Campaign exposure and interpersonal
communication as factors in contraceptive use in Bolivia. Journal of Health
Communication: International Perspectives, 6(4), 303-322.
Valente, T.W. & Vlahov, D. (2001) Selective risk taking among needle exchange
participants: Implications for supplemental interventions. American Journal of
Public Health, 91(3), 406-411.

166
Valente, T. W. (2010). Social networks and health: Models, methods, and applications.
New York: Oxford.
Venkatesh, V., Morris, M. G., Davis, G. B., & Davis, F. D. (2003). User acceptance of
information technology: Toward a unified view. MIS Quarterly, 27(3), 425-478.
Victor, E., & Haruna, K. (2012). Relationship between health locus of control and sexual
risk behaviour. Retrovirology, 9(Suppl 1), P62. doi: 10.1186/1742-4690-9-s1-p6

Wani, S. A., Rabah, S. M., Alfadil, S., Dewanjee, N., & Najmi, Y. (2013). Efficacy of
communication amongst staff members at plastic and reconstructive surgery
section using smartphone and mobile WhatsApp. Indian J Plast Surg, 46(3), 502-
505. doi: 10.4103/0970-0358.121990

Walker, S. N., Sechrist, K. R., & Pender, N. J. (1987). The health-promoting lifestyle
profile: Development and psychometric characteristics. Nursing Research, 36(2),
76-81.
Walker, S. N., Volkan, K., Sechrist, K. R., & Pender, N. J. (1988). Health-promoting life
styles of older adults: comparisons with young and middle-aged adults, correlates
and patterns. ANS: Advance in Nursing Science, 11(1), 76-90.
Wallston, B. S., & Wallston, K. A. (1978). Locus of Control and Health: A Review of the
Literature. Health Education & Behavior, 6(1), 107-117. doi:
10.1177/109019817800600102
Wallston, K. A., Wallston, B. S., & DeVellis, R. (1978). Development of the
Multidimensional Health Locus of Control (MHLC) Scales. Health Education &
Behavior, 6(1), 160-170. doi: 10.1177/109019817800600107

Wallston, K. A., Wallston, B.S., Smith, S. & Dobbins, C.J. (1987). Perceived control and
health. Current Psychological Research & Reviews, 6(1), 5-25

Weinstein, N.D. (1993). Testing four competing theories of health-proective behavior.
Healthy Psychology, 12(4), 324-333.

Weinstein, N. D., Sandman, P. M., & Blalock, S. J. (2008). The Precaution Adoption
Process Model. In K. Glanz, B. K. Rimer, & K. Viswanath (Eds.), Health
Behavior and Health Education (pp. 123–147). San Francisco: Jossey-Bass.

Webb, T. L., Joseph, J., Yardley, L., & Michie, S. (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. J Med Internet Res, 12(1), 1-20.
Wei, J., Hollin, I., & Kachnowski, S. (2011). A review of the use of mobile phone text
messaging in clinical and healthy behaviour interventions. J Telemed Telecare,
17(1), 41-48. doi: 10.1258/jtt.2010.100322

167
Williams-Piehota, P., Pizarro, J., Schneider, T. R., Mowad, L. & Salovey, P. (2005).
Matching health messages to monitor-blunter coping styles to motivate screening
mammography. Health Psychology, 24(1), 58-67.
Wolf, G. (2010, April 28). The data-driven life. The New York Times Magazine.
Retrieved from http://www.nytimes.com/2010/05/02/magazine/02self-
measurement-t.html
Zimet, G.D., Dahlem, N.W., Zimet, S.G. & Farley, G.K. (1988). The multidimensional
scale of perceived social support. Journal of Personality Assessment, 52, 30-41.

168
APPENDIX A: QUESTIONNAIRE
Mobile Self-Tracking For Weight Management And Health Survey
Thank you for participating in the survey! Please carefully read the
following qualification and definition before you start.

Qualification

Participants who are interested in weight management-related issues (ex: healthy weight,
weight control, weight maintenance, weight loss, diet, healthy eating, physical activity,
fitness, workout, etc.), and who are currently or have been undertaking any weight
management-related regimens are eligible to participate.

What Type of Self-Tracker Are You?

Ø TYPE 1: Mobile Self-Tracker: If you routinely self-monitor and keep records of
any health indicators for weight management/health purpose(s) (ex: tracking for
exercises, physical activities, diets, nutrition, food diaries, water, sleep, etc.), and you
engage in self-tracking activities with the assistance of any kinds of mobile
technologies, such as mobile phones and apps, (ex: MyFitnessPal, Runkeepers,
etc.), wearables (ex: Apple Watches, Fitbit, Jawbone, Nike+ FuelBand, etc.), or other
mobile devices, you are a mobile self-tracker.

Ø TYPE 2: Non-Mobile Self-Tracker: If you routinely self-monitor and keep records
of any indicators for weight management/health purpose(s) without the assistance of
any mobile technologies (ex: taking notes with pen and paper, memorizing tracking
data in your head, keeping diaries, etc.), you are a traditional self-tracker.

Ø TYPE 3: Non-Tracker: If you are NOT engaging in any routine self-tracking
activities for weight management/health purpose(s), or you had self-tracked before
but not anymore now, then you are a non-tracker.

No matter which category you belong to, we want to learn from you! We want to know
your true feelings, thoughts, and experiences. All your personal information and answers
will remain anonymous and will not be revealed to anyone. Please press next to start the
survey. Thank you for your participation!

Chihwei Hu
Annenberg School for Communication & Journalism
University of Southern California

169
PART 1. Screening Questions

1. [Age] How old are you? [If Q1 Is Less Than 18, Then Skip To End of Block]

2. Are you a self-tracker? Do you regularly track and keep records for any health
indicators for weight management purposes (For example, track for exercise, diet,
physical activity, calories, weight changes, water, sleep, blood pressure, blood sugar,
etc.)?
m YES [=> Self-tracker] (Go to 3)
m NO [=> Non-tracker] (Go to 8, 9, 10)

3. Do you self-track with the assistance of any mobile technologies?
m Yes, I am a mobile self-tracker. I use mobile technologies to assist my
tracking activities (e.g., smartphones, mobile applications, wearables, Fitbit,
Nike+, Apple Watch, etc.). (Go to 4, 5, 6 ,7)
m No, I self-track in a traditional way without mobile technologies (i.e., using
pan and paper, taking notes, keeping diaries, etc.). (Skip 4 &5, Go to 6 & 7)
m No, I track my health activities “in my head”. (Skip 4 &5, Go to 6 & 7)
m Other. Please specify if none of the above is applicable to you:
____________________ (Skip 4 &5, Go to 6 & 7)

4. What is the major mobile tracking app/device do you use?
m Please specify: _______________________________

5. [Ecological Momentary Intervention] Please think about the mobile device(s)/app(s)
you use to assist your self-tracking activities. Are there any following features that
provide you with real-time (immediate) support? Please check all that apply:
o Instant activity summary (1)
o Instant feedback/comments (2)
o Reminders (for tracking activities or goals, etc.) (3)
o Alerts (for relapses or unwanted behavior, etc.) (4)
o Other real-time functions. Please specify: (5) ____________________

170
Trackers Only

6. [Tracking Themes] What theme(s) are your tracking for? Check all that apply.
o Physical activities (1)
o Fitness (2)
o Diet, food, calorie counter (3)
o Mood (4)
o Sleep (5)
o Blood sugar (6)
o Period or menstrual cycle (7)
o Weight (8)
o Cholesterol (9)
o Diabetes markers (10)
o Other. Please specify: (11) ____________________

7. [Tracking Reasons] Which of the following statements best describe(s) why you are
motivated to do self-tracking? (Check all that apply)
o I am motivated by the pleasure of playing with self-tracking devices (or apps,
etc.). (1)
o I am motivated by the pleasure of tracking my own health. (2)
o I am motivated by the desire to be part of an online community or an offline
network where a common interest is shared (e.g., the Quantified Self,
WebMD, weight management communities, etc.). (3)
o I am motivated by the desire to optimize my own health. (4)
o I am motivated by the desire to regulate my behavior through self-tracking.
(5)
o I am motivated by the desire to understand myself beyond the traditional
health system. (6)
o None of the above. Please specify your motivation if none of the above
applies to you: (7) ____________________

171
Non-Trackers Only

8. [Reasons] Which of the following statements best explain(s) the reason why you don't
self-track for weight management/health purpose(s)?
m I’ve never heard about self-tracking. (1)
m I’m undecided about self-tracking. (2)
m I’ve tried it before and decided I don’t want to self-track. (3)
m Please specify a reason if none of the above applies to you: (4)
____________________

9. [Barriers] What are the major barriers, constraints, or difficulties that prohibited you
from engaging in self-tracking for weight management/health purpose(s)? Check all that
apply.
o I do not have time. (1)
o I do not use a smartphone or other similar devices. (2)
o I do not have money to buy tracking devices or apps. (3)
o I don't like the idea of self-tracking. (4)
o I think there is no need to do it. (5)
o If none of the above applies to you, please briefly specify a reason that prohibited
you from engaging in self-tracking: (6) ____________________

10. [Alternatives] Following the last question, as you are not engaging in self-tracking
activities for your weight management/health purposes, what do you do to deal with your
weight management/health concerns? Check all that apply.
o Seek help from health professionals (ex: doctors, nurses, specialists, personal
coaches, etc.) (1)
o Seek help from people in online communities (ex: PatientLikeMe, WebMD,
MedHelp, CureTogether, etc.) (2)
o Purchase weight management programs or relevant equipment (ex: DVD, online
courses, exercise tools, books, etc.) (3)
o Go to gyms/workout (4)
o Seek support from family and friends (5)
o Do nothing special. (6)
o Other. Please specify (7) ____________________

172
PART 2. Predictors of Autonomous Motivation

In this section, you will be asked 8 sets of questions regarding how you feel about
yourself and your significant others in different aspects. For each statement, please
choose the answer which best reflects your feeling, perception, or situation. There are
no right or wrong answers. Please make your answers as true for you as possible.

1. [Health Consciousness]
Strongly
Disagree
(1)
Disagree
(2)
Somewh
at
Disagree
(3)
Neither
Agree nor
Disagree
(4)
Somewh
at Agree
(5)
Agree (6) Strongly
Agree (7)
A. I try my best
to stay healthy.
m m m m m m m
B. Living life in
best possible
health is very
important to me.
m m m m m m m
C. I actively do
something to
prevent
diseases or
illness.
m m m m m m m
D.I constantly
examine my
health.
m m m m m m m

2. [Internal Health Lotus of Control]
Strongly
Disagree
(1)
Disagree
(2)
Somewh
at
Disagree
(3)
Neither
Agree nor
Disagree
(4)
Somewh
at Agree
(5)
Agree (6) Strongly
Agree (7)
A. I believe that
my health
heavily depends
on how well I
take care of
myself.
m m m m m m m
B. I believe I am
in control of my
health.
m m m m m m m
C. If I get sick,
my behavior
determines how
soon I get well.
m m m m m m m

173
3. [Personal Innovativeness toward Mobile Service (PIMS)]

What's your attitude toward new mobile services? Please read the following statements
and indicate to what extent you agree or disagree with each of the statements.
Strongly
Disagree
(1)
Disagree
(2)
Somewh
at
Disagree
(3)
Neither
Agree nor
Disagree
(4)
Somewh
at Agree
(5)
Agree (6) Strongly
Agree (7)
A. If I heard
about a new
mobile service, I
would look for
ways to
experiment with
it.
m m m m m m m
B. Among my
friends and
family, I am
usually the first
to try out new
mobile services.
m m m m m m m
C. I like to
experiment with
new mobile
services or
functions.
m m m m m m m

4. [mHealth Literacy]
Strongly
Disagree
(1)
Disagree
(2)
Somewh
at
Disagree
(3)
Neither
Agree nor
Disagree
(4)
Somewh
at Agree
(5)
Agree (6) Strongly
Agree (7)
A. I know how
to use the
health
information I get
via my mobile
phone(s).
m m m m m m m
B. I know how
to make use of
my mobile
phone(s) to find
health
resources I
need.
m m m m m m m
C. I know how
to make use of
my mobile
phone(s) to
answer my
health
questions.
m m m m m m m

174
5. [Vigilance Coping]

Please think about the times when you were faced with difficult or stressful events in
your life. We are interested in what you generally do and feel when you experience
stressful events. As you read through the following statements, please answer
them based on your own experiences.

Not at all (1) A little (2) Somewhat
(3)
Much (4) Very much
(5)
A. I concentrate my efforts
on doing something about
the situation.
m m m m m
B. I focus on dealing with
the problem, and if
necessary let other things
slide a little.
m m m m m
C. I take direct action to get
around the problem.
m m m m m
D. This is an attention filter.
Please select "Very much"
for this statement.
m m m m m

6. [Self-Efficacy]
Strongly
Disagree
(1)
Disagree
(2)
Somewh
at
Disagree
(3)
Neither
Agree nor
Disagree
(4)
Somewh
at Agree
(5)
Agree (6) Strongly
Agree (7)
A. I know what
resources are
available to me
for my health
concerns.
m m m m m m m
B. I know where
to find helpful
health
resources for
my health
concerns.
m m m m m m m
C. I have the
skills I need to
maintain or
improve my
health.
m m m m m m m

175
7. [Normative Beliefs]
Strongly
Disagree
(1)
Disagree
(2)
Somewha
t
Disagree
(3)
Neither
Agree nor
Disagree
(4)
Somewha
t Agree
(5)
Agree (6) Strongly
Agree (7)
A. I think I
should engage
in mobile self-
tracking for
health.
m m m m m m m
B. I think the
people
important to
me should do
mobile self-
tracking for
their health.
m m m m m m m
C. Those
people
important to
me think I
should do
mobile self-
tracking for
health.
m m m m m m m

8. [Social Norms]
Please think about your "significant others," who are the people "close and important" to
you. Please read the following statements and indicate to what extent you agree or
disagree with each of the statements.
None of
Them (1)
A few of them
(2)
About half of
them (3)
More than
half of them
(4)
All of them
(5)
A. How many of the
people important to
you do mobile self-
tracking for health
purposes?
m m m m m
B. How many people
important to you think
you do mobile self-
tracking?
m m m m m

176
PART 3. Module for Autonomous Motivation

1. [Autonomous Motivation] Why did you decide to do self-tracking?

It is because…
Strongly
Disagree
(1)
Disagree
(2)
Somewh
at
Disagree
(3)
Neither
Agree nor
Disagree
(4)
Somewh
at Agree
(5)
Agree (6) Strongly
Agree (7)
A. I experience
pleasure when I
engage in self-
tracking.
m m m m m m m
B. I believe self-
tracking for
health could be
beneficial to me.
m m m m m m m
C. I am
inherently
interested in
self-tracking for
my health.
m m m m m m m
D. I personally
find it important
to self-track for
my health.
m m m m m m m
E. It is
consistent with
my life goals.
m m m m m m m
F. This is an
attention filter.
Please select
"Strongly
Disagree" for
this statement.
m m m m m m m

177
PART 4. Module for Self-Determination
In this section, we would like to know how you generally feel when you engage in mobile
self-tracking activities for health. Please read the following statements and indicate to
what extent you agree or disagree with each of them.

1. [Autonomy] When I engage in self-tracking, I feel that...
Strongly
Disagree
(1)
Disagree
(2)
Somewh
at
Disagree
(3)
Neither
Agree nor
Disagree
(4)
Somewh
at Agree
(5)
Agree (6) Strongly
Agree (7)
A. Self-tracking
helps build my
own
independence
regarding my
health.
m m m m m m m
B. Self-tracking
makes me
behave in an
assertive
manner
regarding my
health goals.
m m m m m m m
C. Self-tracking
makes me feel
in control of my
health concerns.
m m m m m m m
D. Self-tracking
makes me feel
that I have the
opportunity to
make choices
with respect to
the way I want
to manage my
health.
m m m m m m m

178

2. [Relatedness] When I engage in self-tracking,....

Strongly
Disagree
(1)
Disagree
(2)
Somewh
at
Disagree
(3)
Neither
Agree nor
Disagree
(4)
Somewh
at Agree
(5)
Agree (6) Strongly
Agree (7)
A. I feel more
connected with
other people
when I am self-
tracking.
m m m m m m m
B. There are
people I can
share my
thoughts when I
am engaging in
self-tracking.
m m m m m m m
C. I feel there
are people who
understand me
and know what I
am doing as I
engage in self-
tracking.
m m m m m m m
D. I can get
support from
people when I
need it as I
engage in self-
tracking.
m m m m m m m

3. [Competence] When I engage in self-tracking,…
Strongly
Disagree
(1)
Disagree
(2)
Somewh
at
Disagree
(3)
Neither
Agree nor
Disagree
(4)
Somewh
at Agree
(5)
Agree (6) Strongly
Agree (7)
A. I feel more
competent in
managing my
health goals as I
engage in self-
tracking.
m m m m m m m
B. I feel more
confident with
my health
decision(s) as I
use information
obtained
through self-
tracking
activities.
m m m m m m m
C. I have the
feeling that I can
m m m m m m m

179
accomplish my
health goal(s)
as I engage in
self-tracking.

PART 5. Module for Self-tracking Behavior

1. [Adherence] To what extent do you adhere to your self-tracking routines? Please
choose your answer from the scale that best reflects your adherence level.
Very poor
(1)
Poor (2) Fair (3) Good (4) Very good
(5)
A. I would say my
adherence to my self-
tracking routines is
generally _______.
m m m m m
Always (1)
Most of the
Time (2)
Sometimes
(3)
Rarely (4) Never (5)
B. Did you ever forget to
perform your routine self-
tracking activities?
m m m m m
C. Did you ever cut back or
stop self-tracking for no
reasons?
m m m m m
D. Did you ever feel
hassled about sticking to
your self-tracking
activities?
m m m m m

2. [Involvement] To what extent do you involve in your self-tracking routines?
Extremel
y
Low (1)
Low (2) Somewh
at Low
(3)
Moderate
(4)
Somewh
at High
(5)
High (6) Extremel
y High
(7)
Generally, I would
say that my
involvement in my
self-tracking
routines is
________.
m m m m m m m

3. [Intensity] How frequently do you self-track for your weight management/health
purpose(s)?
Less than
Once a
Month (1)
Once a
Month (2)
2-3 Times
a Month
(3)
Once a
Week (4)
2-3
Times a
Week (5)
Once a
day (6)
Several
times a
day (7)
My self-tracking
frequency
is___________.
m m m m m m m

180
4. [Social Share] How often do you share your self-tracking results/ health data with
people (i.e., family, friends, online communities, etc.). For example, posting
status/reports online, submitting tracking results on social media, sharing on a
discussion forum, or phoning a friend or family members?
Never (1) Rarely (2) Sometimes
(3)
Often (4) All of the
Time (5)
How often do you
share your self-
tracking
results/health data
with people?
m m m m m

PART 6. Egocentric Network Measures

Please provide the initials or nicknames of up to 6 people with whom you talk about your
weight management issues or concerns (To protect privacy, you don’t have to provide
the real names of these people). Please answer the following 11 questions for each of
your nominators by scrolling the horizontal bar from the left to the right and checking out
each drop-down list to choose your answers. All the information will NOT be revealed to
anyone, and your nominators’ identities will stay anonymous and confidential. They will
NOT be contacted or bothered for any reason. All the information is collected only for
research purposes.

1. How did you know this person?

Family/
Relative
(1)
Friend (2)
Colleague/
Schoolmat
e (3)
Health
profession
al
(4)
Celebrity
(5)
Online
Communit
y (6)
Neighbor
(7)
Nominator
1:
m m m m m m m
Nominator
2:
m m m m m m m
Nominator
3:
m m m m m m m
Nominator
4:
m m m m m m m
Nominator
5:
m m m m m m m
Nominator
6:
m m m m m m m

181
2. What is the gender of this person?
Male (1) Female (2)
Nominator 1: m m
Nominator 2: m m
Nominator 3: m m
Nominator 4: m m
Nominator 5: m m
Nominator 6: m m

3. What is the age of this person?
Nominator 1: Age (1)
Nominator 2: Age (2)
Nominator 3: Age (3)
Nominator 4: Age (4)
Nominator 5: Age (5)
Nominator 6: Age (6)

4. What is this person's ethnicity/race?
Caucasian (1) Asian (2) Hispanic (3)
African
American (4)
Other (5)
Nominator 1: m m m m m
Nominator 2: m m m m m
Nominator 3: m m m m m
Nominator 4: m m m m m
Nominator 5: m m m m m
Nominator 6: m m m m m

5. How long have you known this person? Please write down the number(s) in YEARS. You can
use point to indicate length less than a year. EX: "0.5" (year) = 6 months.
Nominator 1: ________year(s) known (1)
Nominator 2: ________year(s) known (2)
Nominator 3: ________year(s) known (3)
Nominator 4: ________year(s) known (4)
Nominator 5: ________year(s) known (5)
Nominator 6: ________year(s) known (6)

182
6. How frequently do you interact with this person?

On a daily
basis (1)
On a weekly
basis (2)
On a monthly
basis (3)
On a yearly
basis (4)
Very rare (5)
Nominator 1: m m m m m
Nominator 2: m m m m m
Nominator 3: m m m m m
Nominator 4: m m m m m
Nominator 5: m m m m m
Nominator 6: m m m m m

7. Is this person a self-tracker for any health-related purpose(s)?
Yes (1) No (2) I don't know (3)
Nominator 1: m m m
Nominator 2: m m m
Nominator 3: m m m
Nominator 4: m m m
Nominator 5: m m m
Nominator 6: m m m

8. Is this person using any mobile apps/devises to track for health indicators?
Yes (1) No (2) I don't know (3)
Nominator 1: m m m
Nominator 2: m m m
Nominator 3: m m m
Nominator 4: m m m
Nominator 5: m m m
Nominator 6: m m m

9. Did this person recommend/suggest mobile self-tracking to you?
Yes (1) No (2)
Nominator 1: m m
Nominator 2: m m
Nominator 3: m m
Nominator 4: m m
Nominator 5: m m
Nominator 6: m m

183
10. In your opinion, is this person considered healthy in his or her age?
Yes (1) No (2)
Nominator 1: m m
Nominator 2: m m
Nominator 3: m m
Nominator 4: m m
Nominator 5: m m
Nominator 6: m m

11. In your opinion, is this person considered attractive in his or her age?
Yes (1) No (2)
Nominator 1: m m
Nominator 2: m m
Nominator 3: m m
Nominator 4: m m
Nominator 5: m m
Nominator 6: m m

PART 7. Module for Health Outcomes
For each statement below, please select the answer that best describes how you
generally have felt after your routine self-tracking activities.

1. [Health Anxiety]

Never (1) Rarely (2) Sometimes
(3)
Often (4) All of the
Time (5)
A. I have been bothered by
fears about my health after
my routine tracking
activities.
m m m m m
B. I have been concerned
about my health after my
routine tracking activities.
m m m m m
C. I have been afraid that I
have an illness/disease
after my routine tracking
activities.
m m m m m
D. If I have a bodily
sensation or change, I am
eager to know what it
means after my tracking
activities.
m m m m m

184

2. [Perceived Social Support]
How have you generally felt about social support around you since you started self-
tracking for weight management/health purposes?

Strongly
Disagree
(1)
Disagree
(2)
Somewh
at
Disagree
(3)
Neither
Agree nor
Disagree
(4)
Somewh
at Agree
(5)
Agree (6) Strongly
Agree (7)
A. I can get
support from
people around
me when I am in
need.
m m m m m m m
B. I know there
is a person with
whom I can talk
about my
problems when I
engage in self-
tracking.
m m m m m m m
C. There is a
person in my life
who cares about
my feelings.
m m m m m m m
D. I get help I
need from
people when I
engage in self-
tracking.
m m m m m m m

185
3. [Well-being]
How have you been generally feeling about your life since you started self-tracking
for weight management/health purposes?

Strongly
Disagre
e
(1)
Disagree
(2)
Somewh
at
Disagree
(3)
Neither
Agree
nor
Disagree
(4)
Somewh
at Agree
(5)
Agree (6) Strongly
Agree (7)
A. I have been
contented with my
life since I engaged
in self-tracking.
m m m m m m m
B. I have been
feeling happy since
I started self-
tracking.
m m m m m m m
C. I have thought
positively and
constructively since
I started self-
tracking.
m m m m m m m
D. The things I do
in my life have
been worthwhile.
m m m m m m m

4. [Satisfaction]
Please think about your goal(s) of weight management. How satisfied are you with
the outcome(s) in general?
m Very Dissatisfied (1)
m Dissatisfied (2)
m Somewhat Dissatisfied (3)
m Neutral (4)
m Somewhat Satisfied (5)
m Satisfied (6)
m Very Satisfied (7)

5. [Goal Attainment]
Please think about your goal(s) of weight management. Would you say that you were
able to achieve your goal(s) of weight management through engaging in self-tracking
activities?
m No (0)
m Yes (1)
m I am not sure. (2)

186
In the following section, we are interested in your eating and physical activity behavior.
Please indicate how true each of the statements below reflects your typical behavior
since you started engaging in self-tracking activities.

6. [Healthy Eating]
Are there any eating behavior changes since you started engaging in self-tracking
activities?

Never (1) Rarely (2) Sometimes
(3)
Often (4) All of the
Time (5)
A. I increase the
consumption of
healthy foods.
m m m m m
B. I remind myself to
drink enough water
that I need for a day.
m m m m m
C. I avoid unhealthy
foods or foods that are
not good for my
health.
m m m m m
D. I pay attention to
ingredients, nutrients,
and calories in foods.
m m m m m

7. [Physical Activity]
Are there any physical activity changes since you started engaging in self-tracking
activities?

Never (1) Rarely (2) Sometimes
(3)
Most of the
Time (4)
Always (5)
A. I increase exercise
during usual daily activities
(such as walking during
lunch, using stairs instead
of elevators, parking car
away from destination and
walking, light yard work,
housekeeping, etc.)
m m m m m
B. I take part in leisure
time physical activities
(such as swimming,
dancing, bicycling, etc.)
m m m m m
C. I workout more often
than I used to.
m m m m m
D. I take exercise more
often than I used to.
m m m m m

187
PART8. Demographics

1. [Gender] What is your gender?
m Male (1)
m Female (2)

2. [Marital Status] What is your marital status?
m Single. (1)
m In a relationship. (2)
m Married. (3)
m Widowed. (4)
m Divorced/ Separate. (5)

3.[Health Status]. How would you describe your health status in general?

Awful (1) Poor (2) Fair (3) Good (4) Excellent (5)
My health
condition is
__________.
m m m m m

4. [Chronic Disease] Do you suffer from any prescribed chronic disease(s)?
m NO (1)
m YES. Please tell us what is it? (2) ____________________

5. [Education] What is the highest grade or level of schooling you completed?
m Less than high school diploma or GED (1)
m High school diploma or equivalent. (2)
m Some college or associates degree (3)
m Bachelors Degree (4)
m Masters Degree / J.D. (5)
m Ph.D./ M.D. (6)

6. [Child] Do you have children?
m Yes. Please specify the number of children you have: (1)
____________________
m No (2)

7. [Social Media Use] How frequently do you engage in social media (e.g., Facebook,
Twitter, MySpace, Tumblr, Instagram, etc.)?
m Never (1)
m Occasionally (2)
m Fairly Many Times (3)
m Very Often (4)
m Always (5)

188
8. [Social Media Activeness] Do you consider yourself an active user on social media
(such as Facebook, Instagram, Youtube, Twitter, LinkedIn, Tumblr, Google+, etc.)?
m Absolutely NOT. (1)
m I am not very active on social media. (2)
m I am slightly active on social media. (3)
m I am moderately active on social media. (4)
m I am very active on social media. (5)

9. [Smartphone Use] Are you a smartphone user?
m Yes, I am. (1)
m No, I am not. (2)
m
10. [Employment] What is your employee status:
m Employed or self-employed. (1)
m Unemployed (2)
m Retired (3)
m Student (4)
m Housewife (5)
m Other. Please specify: (6) ____________________

11. [Ethnicity] What is your ethnicity?
m White/Caucasian (1)
m Asian (2)
m Hispanic/Latino (3)
m Black/African American (4)
m Other (Please Specify) (5) ____________________

12. [Income] What is your annual household income in US dollars?
m Under $30,000 USD (1)
m $30,001 - $60,000 USD (2)
m $ 60,001- $90,000 USD (3)
m $ 90,001-$120,000 USD (4)
m Over $120,001 (5)

***The End***

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

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

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

PDF

Network influences in health initiatives: multimedia games for youth in Peru

PDF

Contagious: social norms about health in work group networks

PDF

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

PDF

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

PDF

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

PDF

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

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

Social network influences on depressive symptoms among Chinese adolescents

PDF

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

PDF

Developing and testing a heuristic-systematic model of health decision making: the role of affect, trust, confidence and media influence

PDF

Understanding normative influence of neighborhoods: a multilevel approach to promoting Latinas’ cervical cancer prevention behaviors in urban ethnic communities

PDF

Mindfulness-based self-regulation for psychotic disorders: a feasibility study

PDF

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

PDF

Social self-control and adolescent substance use

PDF

The role of protective factors on outcomes for Latinos with schizophrenia

PDF

Applications of mindfulness toward promoting psychosocial health in diverse communities

PDF

Community integration of individuals with serious mental illness: a network perspective from India and United States

Asset Metadata

Creator Hu, Chihwei (Selene) (author)

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

School Annenberg School for Communication

Degree Doctor of Philosophy

Degree Program Communication

Publication Date 07/27/2016

Defense Date 03/08/2016

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

Tag communication technology,EMI,health consciousness,mHealth,mobile self-tracking,OAI-PMH Harvest,SEM,social influence,social network

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor McLaughlin, Margaret (committee chair), Chou, Chih-Ping (committee member), Cody, Michael (committee member)

Creator Email huchihwe@usc.edu,vestamomo@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-282497

Unique identifier UC11281097

Identifier etd-HuChihweiS-4647.pdf (filename),usctheses-c40-282497 (legacy record id)

Legacy Identifier etd-HuChihweiS-4647.pdf

Dmrecord 282497

Document Type Dissertation

Format application/pdf (imt)

Rights Hu, Chihwei (Selene)

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