University of Southern California Dissertations and Theses

Understanding the dynamic relationships between physical activity and affective states using real-time data capture techniques

(USC Thesis Other)

Understanding the dynamic relationships between physical activity and affective states using real-time data capture techniques

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

UNDERSTANDING THE DYNAMIC RELATIONSHIPS BETWEEN
PHYSICAL ACTIVITY AND AFFECTIVE STATES USING
REAL-TIME DATA CAPTURE TECHNIQUES

by

Yue Liao

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

August 2015

Copyright 2015 Yue Liao

i
DEDICATION
To my family.
To people who practice good science.

ii
ACKNOWLEDGMENTS
It is my honor to finally get to this point in life and take this moment to show my
gratitude to everyone who has shown great support and love along the way.
I would like to thank my mentor and committee chair Genevieve Dunton for inspiration,
encouragement, and great belief in me. You are the role model for young females in the
academic world. I would also like to thank Chih-Ping Chou, my co-mentor for the past five
years, for sharing your wisdom and invaluable experience, and providing me opportunities to
collaborate with other great minds. I am fortunate to have the two of you as my advisors, which
made my PhD life a lot easier than many others.
Thanks to my dissertation committee members, Jimi Huh and Adam Leventhal, for
always being available and offering insightful feedback whenever I need help. Thanks
Maryalice Jordan-Marsh for generously serving as an outside committee member, showing me
perspectives beyond my normal realm.
Special thanks to Marny Barovich, my counsel for everything, for creating a positive
support system that makes all the impossible become possible. Your calm voice, charming smile,
and witty emails have uplifted my spirit throughout these years.
I would also like to thank my lovely peers and colleagues for their support and
encouragement. Thanks to Eleanor Tate Shonkoff for being a jolly cheerleader and an amazing
collaborator; Eldin Dzubur for serving me a clean dataset; and Stephanie Raye Dyal, Trevor
Pickering, Lauren Martinez, Cheng K Wen, Chris Warren, Jennifer Tsai, and Angelica Delgado
for always attending my practice talks and providing helpful feedback on my work. Thank you
to my dearest Facebook support team, Rebecca Lee, Scher Mama, Donna Spruijt-Metz, and
Steve Sussman. You have shown me tremendous cheering and love both online and offline.

iii
I am also grateful for the superb staff members of the Project MOBILE and the MATCH
Study: Keito Kawabata, Cesar Aranguri, Frank Cedeno, Lissette Ramirez, and all the student
interns. Your hard work has ensured the high quality of data, which made this dissertation study
possible.
Finally, I would like to thank my parents, for their investment in my education with their
love, support, and patience. Huge thanks to Mingju Cao, my partner in life, for everything.

iv
Table of Contents
DEDICATION ................................................................................................................................. I
ACKNOWLEDGMENTS ............................................................................................................... II
LIST OF TABLES ...................................................................................................................... VII
LIST OF FIGURES ................................................................................................................... VIII
ABSTRACT ................................................................................................................................ IX
CHAPTER 1: INTRODUCTION .................................................................................................... 1
BACKGROUND & SIGNIFICANCE .................................................................................................... 1
Scope of the Problem ............................................................................................................... 1
Definition of Affect ................................................................................................................... 3
Possible Mechanisms Linking Physical Activity and Affect .......................................................... 4
Measuring Mood and Affect ...................................................................................................... 6
Real-Time Data Capture Techniques ......................................................................................... 7
Methodological Limitations of Previous Research on Physical Activity and Affect ......................... 9
OVERVIEW OF DISSERTATION STUDIES ........................................................................................ 11
CHAPTER 2: EXAMINING THE ACUTE BI-DIRECTIONAL RELATIONSHIPS BETWEEN
PHYSICAL ACTIVITY AND AFFECTIVE STATE IN FREE-LIVING SITUATIONS ................... 13
ABSTRACT ................................................................................................................................ 13
INTRODUCTION .......................................................................................................................... 15
Acute Effects of Affective States on Subsequent Physical Activity ................................................ 16
Acute Effects of Physical Activity on Subsequent Affective States ................................................ 16
Current Research Gaps .......................................................................................................... 19
Study Aims ............................................................................................................................ 20
METHODS ................................................................................................................................. 21
Participants .......................................................................................................................... 21
Study Protocols ..................................................................................................................... 22
Measures .............................................................................................................................. 23
Real-time Window ................................................................................................................. 24
Statistical Analysis ................................................................................................................. 24
Power Analysis ...................................................................................................................... 28
RESULTS ................................................................................................................................... 30
Descriptive Statistics .............................................................................................................. 30
Affective State and Subsequent Physical Activity ....................................................................... 31
Physical Activity and Subsequent Affective State ....................................................................... 33
DISCUSSIONS ............................................................................................................................. 33
Affective State Predicting Immediate Subsequent Physical Activity ............................................. 34
Physical Activity Predicting Immediate Subsequent Affective State ............................................. 36
Limitations ............................................................................................................................ 38
Implications .......................................................................................................................... 39
Future Directions .................................................................................................................. 40
Conclusions .......................................................................................................................... 41
CHAPTER 3: EXAMINING WHETHER AFFECTIVE RESPONSES DURING PHYSICAL
ACTIVITY PREDICT CURRENT AND FUTURE DAILY PHYSICAL ACTIVITY LEVELS
AMONG ADULTS ....................................................................................................................... 48
ABSTRACT ................................................................................................................................ 48
INTRODUCTION .......................................................................................................................... 50
Predictors of Physical Activity ................................................................................................ 50

v
Affective Responses to Physical Activity ................................................................................... 50
Change in Affect after Physical Activity Predicting Future Physical Activity Behavior ................. 51
Affective Responses During Physical Activity Predicting Future Physical Activity Behavior ......... 52
Perceived Affective Benefits as a Potential Mediator of the Relationship Between Affective
Response and Future Physical Activity .................................................................................... 53
Research Gaps ...................................................................................................................... 54
Research Questions and Hypotheses ........................................................................................ 56
METHODS ................................................................................................................................. 57
Data Source .......................................................................................................................... 57
Measures .............................................................................................................................. 57
Statistical Analysis ................................................................................................................. 59
Power Analysis ...................................................................................................................... 61
RESULTS ................................................................................................................................... 61
Data Availability ................................................................................................................... 61
Affective Responses During Physical Activity and Current Physical Activity Level ....................... 63
Affective Responses During Physical Activity and Future Physical Activity Level ........................ 63
Perceived Affective Benefits as a Mediator between Affective Responses During Physical
Activity and Future Physical Activity Level .............................................................................. 64
DISCUSSIONS ............................................................................................................................. 64
Limitations ............................................................................................................................ 67
Implications .......................................................................................................................... 68
Conclusions .......................................................................................................................... 68
CHAPTER 4: EXPLORING THE DYADIC RELATIONSHIPS BETWEEN PHYSICAL
ACTIVITY AND AFFECTIVE STATES IN MOTHER-CHILD PAIRS ......................................... 71
ABSTRACT ................................................................................................................................ 71
INTRODUCTION .......................................................................................................................... 73
Physical Inactivity in Parents and Children .............................................................................. 73
Parental Influences on Children’s Physical Activity .................................................................. 73
Parental Stress as a Predictor of Children’s Behavior ............................................................... 74
Parent-child Mutual Affect ..................................................................................................... 75
Current Research Gaps .......................................................................................................... 76
Research Question and Hypothesis .......................................................................................... 77
METHODS ................................................................................................................................. 77
Participants .......................................................................................................................... 77
Study Protocols ..................................................................................................................... 78
Measures .............................................................................................................................. 79
Statistical Analyses ................................................................................................................ 80
RESULTS ................................................................................................................................... 82
Data Availability ................................................................................................................... 82
Descriptive Statistics .............................................................................................................. 83
Multilevel Models .................................................................................................................. 83
DISCUSSIONS ............................................................................................................................. 85
Limitations ............................................................................................................................ 88
Implications .......................................................................................................................... 89
CHAPTER 5: DISCUSSION AND CONCLUSIONS ................................................................... 94
IMPLICATIONS ........................................................................................................................... 96
Methodological Implications................................................................................................... 96
Theoretical Implications ......................................................................................................... 97
Intervention Implications ........................................................................................................ 99
FUTURE RESEARCH DIRECTIONS ............................................................................................... 100

vi
REFERENCES ......................................................................................................................... 103
APPENDIX ................................................................................................................................ 125

vii
List of Tables
Table 1. Participant demographic characteristics (N=110). ................................................... 42
Table 2. Associations between affective state and subsequent minutes in
moderate-to-vigorous physical activity (MVPA) .......................................................... 43
Table 3. Associations between affective state and subsequent minutes in light physical
activity (LPA) .......................................................................................................... 44
Table 4. Associations between physical activity
1
and subsequent affective states .................... 45
Table 5. Associations between affective response during physical activity and future
physical activity level ................................................................................................ 69
Table 6. Mother’s affective state when with their children and children’s subsequent
affective state (N = 80) .............................................................................................. 91
Table 7. Total effect of mother’s affective state on child’s subsequent physical activity .......... 91
Table 8. Associations between mother’s affective state and child’s subsequent
moderate-to-vigorous physical activity (MVPA), mediated by child’s affective state ....... 92
Table 9. Associations between mother’s affective state and child’s subsequent light physical
activity (LPA), mediated by child’s affective state ....................................................... 93

viii
List of Figures

Figure 1. EMA items assessing current affective state. ........................................................ 46
Figure 2. Illustration of 15-minute time windows summarizing total minutes spent in
moderate-to-vigorous physical activity (MVPA) before and after each random EMA
prompt in one day. .................................................................................................... 47
Figure 3. Power plot for affective states predicting physical activity. .................................... 70
Figure 4. A proposed multilevel framework for factors that might influence the
bi-directional relationship between physical activity and affective state in real-life
situations ............................................................................................................... 102

ix
ABSTRACT
This dissertation examined the relationships between physical activity and affective states
using real-time data capture techniques. Specifically, the (1) acute effects (i.e., the bi-directional
relationships at the moment-to-moment level), (2) longitudinal effects (i.e., how the affective
responses during physical activity might predict future physical activity behavior), and (3)
dyadic effects (e.g., whether mothers’ affective states may influence their children’s subsequent
affective states and physical activity levels) were tested and explored using data collected from
mobile phone apps and accelerometers. The unique characteristics of real-time data capture
methods allow researchers to minimize participants’ recall biases and improve a study’s external
and ecological validity. Results from this dissertation study show that a more positive affective
state was associated with more physical activity both short-term and long-term. Further,
engaging in more physical activity led to an immediate improvement in physical feeling state.
However, this study did not find any significant relationship between one person’s (i.e., mothers)
affective states and another person’s (i.e., children) subsequent physical activity levels. Overall,
this dissertation study demonstrated the use of real-time data to examine the relationships
between physical activity and affective states in free-living settings. Findings from this study
could offer directions for future studies (e.g., explore potential moderators and mediators under a
ecological framework in free-living environments) and insights for intervention development
(e.g., target negative affective feelings as a barrier for engaging in daily physical activity).

1
CHAPTER 1: INTRODUCTION
Background & Significance
Scope of the Problem
There is strong evidence for the health benefits of physical activity including reduced
rates of heart disease, metabolic syndrome, breast and colon cancers, depression, and increased
cardiorespiratory and muscular fitness and improved cognitive function (Haskell, Blair, & Hill,
2009; Warburton, Charlesworth, Ivey, Nettlefold, & Bredin, 2010). However, available data
suggest that 31% of the world’s population is not meeting the minimum recommendations for
physical activity (Hallal et al., 2012). Although the United States released the first-ever national
guidelines for physical activity in 2008, there are currently more than 80% of adults and
adolescents do not do enough aerobic physical activity to meet these guidelines (U.S.
Department of Health and Human Services, 2013). Worldwide, physical inactivity causes 6-10%
of the major chronic diseases and 9% of premature mortality (Lee et al., 2012). In the U.S.,
physical inactivity accounts for nearly 1 in 10 deaths (Danaei et al., 2009). The high prevalence
of physical inactivity and its well-established negative health consequences put the promotion of
regular physical activity as a global public health priority.
There is also evidence of mental health benefits from engaging in physical activity.
Regular physical activity may decrease depression and anxiety symptoms (e.g., Mead et al.,
2009; Rethorst, Wipfli, & Landers, 2009; Wipfli, Rethorst, & Landers, 2008). Further, acute
physical activity (i.e., a single bout of moderate-to-vigorous activity) is associated with a
decrease in negative affect and/or increase in positive affect (e.g., Landers & Arent, 2001; Reed
& Ones, 2006; Yeung, 1996;) in both clinical and community populations. These feel-good
effects are shown to be both acute (i.e., immediately following physical activity) and long-term
(i.e., over days or even months; Yeung, 1996; Annesi, 2002a). Although these studies offer

2
support for the potential emotional benefits, it is not clear how people’s feelings before, during,
and after engaging in physical activity might influence their decisions to perform this behavior in
their daily lives.
According to behavioral theories, people will engage in a behavior when pleasure results
from this behavior (e.g., the greatest happiness principle; Bentham, 1962); or when they
anticipate some positive affective response from engaging in a behavior (e.g., the subjective
expected pleasure theory; Mellers, 2000). In addition, neurobiological theories, such as the
somatic-marker hypothesis, attempt to explain how emotions could influence people’s decision
making about their behaviors to maximize reward and minimize punishment (Naqvi, Shiv, &
Bechara, 2006). Reward is a biological mechanism mediating behavior, motivated by events that
are associated with pleasure. From a neurobiological perspective, pleasure is a function of
reward and motivation circuits that is imbedded in the central nervous system (Esch & Stefano,
2010). There are two types of motivation, namely appetitive and aversive motivation.
Appetitive motivation involves behaviors that are associated with moving towards positive
hedonic states whereas aversive motivation involves getting away from unpleasant conditions
(Bozarth, 1994). Therefore, when studying how affect might influence people’s decision making
about engaging in physical activity, it is important to consider both positive emotions (i.e., the
appetitive motivation) and negative emotions (i.e., the aversive motivation). In summary,
improved knowledge of the effects of affect on physical activity in naturalistic, free-living
settings could help us understand why people engage or do not engage in physical activity in
their daily lives.

3
Definition of Affect
Affect, mood, feeling, and emotion. Previous studies on this topic usually use a variety
of terms such as “affective response,” “feeling state,” “mood state,” and “mood” to describe
people’s psychological experience. Some psychologists and psychiatrists use “affect” as a
broader term that encompasses “mood,” “feeling states,” and “emotion” (e.g., Rahe, Rubin, &
Gunderson, 1972; Schwarz & Clore, 2007). Yet others consider that “emotion” to be a more
generic term than “affect” (e.g., Lazarus, 1991) since it could include other attributes such as
physiological changes (e.g., Drever, 1952). Researchers from other fields (e.g., behavioral
science and public health) sometimes use these words interchangeably. In this dissertation study,
“emotion” and “feeling” are used synonymously as broader terms to describe emotional
experiences and subjective feelings. “Affect” refers to immediate emotional states that a person
feels at any given moment while “mood” implies a more sustained feeling that could last a few
days or longer duration.
Discrete versus dimensional approach. In psychology, researchers have different ways
of conceptualizing emotion. Some define emotion as discrete emotional reactions, such as
pleasure, fear, anger, and happiness (e.g., Lazarus, 1991; Ortony, Clore, & Foss, 1987).
Researchers consider each emotion to be unique since they correspond to different reactions to
the perceived impact of an event. Others argue that emotions are best described in terms of their
common properties, or dimensions (e.g., Russell, 1980; Watson & Tellegen, 1985). According
to Russell’s “circumplex” model, emotions can be defined as dimensions of valence (i.e.,
pleasant-unpleasant) and arousal (i.e., high-low). Then each emotion can be classified along
these two dimensions. For example, feeling excited involves high arousal and high pleasure,
feeling relaxed involves low arousal and high pleasure, and feeling sad involves low arousal and

4
low pleasure. In Watson and Tellegen’s work (1985), they have shown two major dimensions
arise from analyses of emotions, namely positive and negative affect. Positive affect refers to
feelings such as alertness and activeness, whereas negative affect refers to unpleasant affective
states such as anxious and depressed.
Possible Mechanisms Linking Physical Activity and Affect
Although why and how physical activity is associated with affective states are not yet
fully understood, there are some possible biochemical, physiological, and psychological
mechanisms that might explain the links between the two.
Biochemical and physiological mechanisms. Physical activity influences various body
systems (e.g., hormones, monoamine neurotransmitters), which may all lead to changes in
affective states. There are several biochemical and physiological based hypotheses regarding the
mechanisms by which physical activity could change affective states.
The thermogenic hypothesis posits that the elevation in the body’s temperature can cause
changes in affect following physical activity (Koltyn, 1997). For instance, the increase in
temperature of specific brain regions (e.g., the brain stem) can lead to reduction in muscular
tension and feeling of relaxation (deVries, Wiswell, Bulbulian, & Moritani, 1981).
The endorphin hypothesis postulates that the increase in β-endorphin production
following physical activity leads to a positive change in affective states (Hoffmann, 1997).
There are also studies suggesting connections between this and the thermogenic hypothesis. For
example, the increase in plasma β-endorphin, together with the physical activity induced heat
load, lead to higher opioid levels in the cerebrospinal fluid. This mechanism suggests the
thermogenic hypothesis enhances the endorphin response and increases the cerebral action of
plasma β-endorphin (Dubnov & Berry, 2000).

5
Another promising physiologic mechanism is the monoamine hypothesis, which states
that physical activity could lead to an increase in neurotransmitters responsible for pleasure in
the brain (e.g., serotonin, dopamine, and norepinephrine; Chaouloff, 1989). The decrease in
these neurotransmitters has been shown to relate with depression and negative affect (Ruhé
Mason, & Schene, 2007).
Overall, there are complex interrelationships among physical activity, neurohumoral, and
metabolic activity, which could all influence affective states. For example, activation of the
stress system during physical activity initiates several metabolic cascades, resulting in changes in
neurotransmitter actions. The released neurohormone β-endorphin correlates with blood acidosis
(an indicator of effort), and produces calming and anxiolytic effects, which is mediated through
modifying dopamine and noradrenaline neurotransmission in brain areas (Dubnov & Berry,
2000).
Psychological mechanisms. Several psychological mechanisms have also been
proposed. The distraction hypothesis suggests that engaging in physical activity can distract
one’s mind from everyday worries and stressors, which leads to an elevated positive affect
(Morgan, 1985). The mastery/self-efficacy hypothesis states that engaging in physical activity
may introduce an achievement sensation resulting in improved affect (North, McCullagh, &
Tran, 1990). Lastly, the social interaction hypothesis proposes that the social relationships
developing from engaging in physical activity contributes to the positive effects on people’s
subjective feelings (Ransford, 1982; Vilhjalmsson & Thorlindsson, 1992).
Summary. While it is hard to directly test each of these mechanistic hypotheses, it is
likely that the combination of biological, psychological, and sociological factors influence the
relationship between physical activity and affect. In addition, there may be individual variations

6
in the mechanisms or combination of mechanisms of this relationship. For example, for people
who just starting physical activity (i.e., in the adoption phase), the psychological mechanisms
might play a more important role since these people have not yet adapted to the physical activity
stimulus physiologically. For people who are in the maintenance phase of physical activity, both
psychological and physiological mechanisms could be equally important (Boutcher, 1993).
Measuring Mood and Affect
The majority of studies that investigate mood and affect in relation to physical activity
have used self-assessment scales such as Profile of Mood States (POMS), and Positive and
Negative Affect Schedule (PANAS; Biddle, 2001). Criticisms of these scales include failing to
capture positive feelings of well-being (for POMS) and not being able to differentiate more
specific types of emotion (for PANAS). In an effort to address these limitations, Exercise-
induced Feeling Inventory (EFI) was developed to capture four distinct feeling states in exercise
“at this moment:” revitalization, tranquility, positive engagement, and physical exhaustion
(Gauvin & Rejeski, 1993). EFI has been shown as a valid instrument for both adults and
children, and can be easily used in field assessments of physical activity affect (Biddle, 2001).
However, there is one major limitation of these recall-based assessments. As discussed
earlier, mood reflects a person’s global feeling over a longer period, which may not reflect a
specific feeling in response to, or results from, one particular behavior. For example, according
to Kahneman and colleagues’ research (1997) on hedonic psychology, people’s retrospective,
global evaluations of a behavior differ from the aggregate of instantaneous, affective experiences
that happen during the course of the behavior. In particular, when reporting global mood, people
tend to focus on the most intense and most recent affective experiences that occurred during the
behavior, rather than aggregating of all affective experiences and weighing them equally

7
(Fredrickson & Kahneman, 1993; Redelmeier & Kahneman, 1996). Therefore, Kahneman and
other have recommended the use of real-time measures of instant affective response to a
behavior to avoid the biases of memory and evaluation that affect retrospective judgments
(Kahneman, Wakker, & Sarin, 1997).
Real-Time Data Capture Techniques
Real-time data capture (RTDC) refers to collecting data as it “naturally unfolds in a
person’s life” (Stone & Broderick, 2007). RTDC includes self-reported techniques (e.g.,
ambulatory monitoring, experience sampling) and objective measurements (e.g., through sensors
and monitors) for physiological (e.g., heart rate, blood pressure), behavioral (e.g., physical
activity, sedentary behavior), and contextual (e.g., location) data. RTDC differs from traditional
retrospective data collection methods for its root in sampling “snapshots” of people’s lives to
more accurately capture the variability of experiences. Because of its real-time feature, RTDC
can reduce memory and other biases that are associated with recall of experiences and events
(Schwarz, 2007).
One of the RTDC techniques to collect self-reported behavioral and cognitive processes
in real-world settings is Ecological Momentary Assessment (EMA). EMA methods are
characterized by assessing people’s current or recent states, repeatedly over time, in their natural
environments (Shiffman, Stone, & Hufford, 2008; Stone & Shiffman, 1994). With these
features, EMA studies are able to focus on within-subject changes in behavior and experiences
over time and across contexts. The use of EMA methods not only allows for more ecologically
valid observations, but could also examine temporal sequences of events and/or experiences,
which is helpful to address the antecedents and consequences of events and behaviors.

8
EMA encompasses different delivery methods ranging from paper-pencil diaries,
telephones (e.g., using interactive voice response technology, see Mundt, Perrine, Searles, &
Walter, 1995), and portable electronic devices such as palm-top computers and more recently,
mobile phones. Because of the large amounts of repeated-measurements nature of EMA, the use
of electronic devices to deliver and record momentary assessments are shown to be a more
reliable method than paper-pencil diaries, especially in terms of higher compliance (Green,
Rafaeli, Bolger, Shrout, & Reis, 2006; Piasecki, Hufford, Solhan, & Trull, 2007). More
importantly, electronic devices are able to give an exact time stamp of each assessment and to
ensure that the assessment is completed when it should have been (e.g., issuing reminder signals
when an assessment is missed). Electronic EMA also has the flexibility in designing sampling
schedules and survey questions. For example, EMA assessment could be interval- (e.g.,
information recorded every 90 minutes), event- (e.g., information recorded during/after a certain
behavior), or signal-contingent (e.g., information recorded when being prompted randomly),
with adjustable levels of prompting, and questions could appear in fixed or random order (Barret
& Barret, 2001).
Activity monitors such as accelerometers are another example of RTDC techniques.
Accelerometers are wearable devices using sensors to detect body movements in terms of
acceleration in orthogonal planes (e.g., anteropsterior, mediolateral, and vertical), which can then
be used to estimate the intensity of activity over time (Chen & Bassett, 2005). Signals detected
from accelerometry sensors are converted to digital numbers by the devices, which are called
“raw counts.” Then these “raw counts” can be summed for each pre-determined time period
(i.e., epoch; usually in 1 minute or 30 seconds) to give the “activity counts” that is generally used
by researchers to assess and categorize activity types and intensities.

9
In summary, RTDC techniques have the advantages of collecting large amounts of data in
real-time, real life. RTDC techniques not only enhance the data quality by reducing memory and
recall biases, their repeated measures also allow researchers to better investigate within-person
variability and associations. More importantly, RTDC methods (especially for data collected via
electronic devices) provide the information of when things happen. Knowing when things
actually occur is crucial for both empirical testing of causal relationships and development of
theoretical statements (Mitchell & James, 2001).
Methodological Limitations of Previous Research on Physical Activity and Affect
Most studies investigating the acute associations between physical activity and affective
states have been conducted in controlled laboratory settings. These studies usually asked
participants to rate their current affective states before, during, and/or after performing treadmill
or pedaling exercise in a lab (see Petruzzello, Landers, Hatfield, Kubitz, & Salazar, 1991 and
Reed & Ones, 2006 for reviews). For example, in a series of studies conducted by Ekkekakis
and colleagues, undergraduate students were invited to a laboratory, where they were fitted with
heart rate monitors (Ekkekakis, Hall, VanLanduyt, & Petruzzello, 2000). Participants were taken
to a treadmill and were facing a barren wall. A clipboard was placed on the control board of the
treadmill for participants to mark their responses to questions regarding affective states.
Affective states were assessed before, during (i.e., at the 8th minute of the 15-minute treadmill
walking session), immediately following, post 10-minute, and post 15-minute of the treadmill
walking.
Although lab studies allow researchers to have a precise control of the physical activity
session (i.e., the intensity and duration), and an exact timing of when to assess affective states
(e.g., every 5 minutes during a session, upon completion of a session), the external validity of

10
findings from these studies is questionable. For example, studies have shown that behaviors
performed under laboratory settings (e.g., walking) are not representative of the same behavior
performed in daily life outside the laboratory (Bussmann, Ebner-Priemer, & Fahrenberg, 2009).
Further, emotion reactions could also differ dramatically between lab-based assessments and
feelings experienced in naturalistic settings (Gunes, Piccardi, & Pantic, 2008). One of the
reasons is that in the laboratory, conditions are prescribed to participants, while in the real world,
individuals have natural preferences and choices about situations they seek and avoid (Wilhelm
& Grossman, 2010). In addition, activity intensity and duration are usually self-selected in real-
world settings. Therefore, conclusions about the relationships between physical activity and
affective states from lab studies might not translate well to the dynamic relationships between
physical activity and affective states in everyday life.
Even for the few studies that used real-time measures for assessing feeling states, several
major methodological limitations still exist (Liao, Tate, & Dunton, unpublished). For example,
most studies did not use an objective measure of physical activity. Self-reported physical
activity in general has a low-to-moderate association with the objectively measured physical
activity due to individuals’ tendency of over-estimating both activity intensity and duration
(Prince et al., 2008; Troiano et al., 2008). Although most studies used repeated-measures of
affective states, the monitoring period was relatively short in some studies (e.g., monitoring
period only lasted 1 or 2 days). Thus, these studies might fail to capture a more representative
picture of people’s daily experiences. Further, most of the studies used a convenience sample
(e.g., undergraduate students in Psychology classes, members of fitness center). This sampling
technique poses great selection bias, and findings from these studies might not be generalized to
populations from the general community, or high-risk populations for physical inactivity. In

11
summary, while there are some studies attempting to examine the acute relationships between
physical activity and affective state in people’s daily lives, these studies still face several major
methodological limitations. Thus, there is a need for using a stronger methodology to elucidate
the complex relationships between physical activity and affective states in real world.
Overview of Dissertation Studies
The methodological and substantive limitations of previous work in understanding the
dynamic relationships between physical activity and affective states can be partially addressed
through the combined use of electronic EMA and accelerometer. Electronic EMA can capture
current affective states as people are going through their normal daily behaviors; accelerometer
can continuously record people’s activity levels without relying on their recalls. Linking real-
time information from these two sources together, we could investigate what factors might
trigger (i.e., antecedents) physical activity, as well as the immediate mental benefits (i.e.,
affective responses) from engaging in physical activity (i.e., consequences). Better knowledge
on this topic could provide critical information and practical strategies for future physical
activity intervention design.
The first study aims to test the bi-directional acute relationships between physical activity
and affective states. The second study aims to test whether affective responses during physical
activity predict current and future physical activity levels. The third study takes an innovative
approach to analyze real-time dyadic data in mother-child pairs (i.e., both mothers and children
were assessed by electronic EMA and accelerometer in their daily lives). This dyadic approach
not only provides an opportunity to examine the acute physical activity and affective states
relationships under a specific social context (i.e., when mothers and their children are together),

12
but also allows us to explore the interpersonal effects (i.e., how mothers’ affective states might
influence her children’s activity level).

13
CHAPTER 2: EXAMINING THE ACUTE BI-DIRECTIONAL RELATIONSHIPS
BETWEEN PHYSICAL ACTIVITY AND AFFECTIVE STATE IN FREE-LIVING
SITUATIONS
Abstract
Purpose: There are several plausible mechanisms for the bi-directional relationships between
physical activity and affective states. However, limited studies have examined these
relationships in real-time, free-living settings. This study used ecological momentary assessment
(EMA), a real-time self-report strategy, to collect information about people’s affective states
during their everyday lives and linked with activity data measured by accelerometer.
Methods: Data from 110 adults were collected through mobile phones and accelerometers across
4 days. Electronic EMA surveys were randomly prompted up to 8 times each day asking about
current affective feeling states (e.g., positive affect, negative affect, energy, and fatigue). All
participants also wore accelerometers during this period to objectively measure light physical
activity (LPA) and moderate-to-vigorous physical activity (MVPA). Multilevel regression
model was used to test the bi-directional relationships between physical activity and affective
states at within-person and between-person level.
Results: More positive affective states (i.e., higher positive affect, lower negative affect, and
lower fatigue) than one’s usual level at the moment led to more MVPA minutes within a
subsequent 15-minute window. Feeling more negative than the average participant in this study
was associated with more MVPA minutes. Feeling more negative than one’s usual level at the
moment was associated with more LPA minutes up to 30 minutes later. Feeling more energetic
was associated with more LPA minutes at both within-person and between-person level. Lastly,
engaging in more physical activity, regardless of the intensity and duration, was associated with
feeling more energetic subsequently. Engaging in more LPA minutes than one’s usual level was
additionally associated with subsequent feelings of more negative and less tired.

14
Conclusions: Individuals might be more likely to be active when they are in a mentally positive
state. Negative affective state could be a potential barrier for people to be active in their daily
lives. Interventions could offer strategies to encourage people to be more active even when in a
negative feeling state. Especially being more active could eventually lead to an enhancement in
mental well-being (i.e., increase in feeling of energy, as found in the current study).

15
Introduction
There are biological and psychological bases for the bi-directional relationships between
physical activity and affective states. On one hand, people’s current affective states reflect their
physical and mental readiness for engaging in a behavior (Seo, Barrett, & Bartunek, 2004;
Schwarz, 1990). Further, when people are uncertain about the outcome of their behaviors, the
somatic-marker hypothesis, a neurobiological theory, states that emotions bias decision making
toward choices that maximize reward and minimize punishment (Naqvi, Shiv, & Bechara, 2006).
On the other hand, the increase in body temperature following physical activity can lead to
reduction in muscular tension and feeling of relaxation (deVries et al., 1981); the increase in β-
endorphin production following physical activity is responsible for the feeling of euphoria
(Tuson & Sinyor, 1993); and physical activity could lead to an increase in production of
neurotransmitters of the monoamine family (i.e., dopamine, serotonin, and noradrenaline), which
has been shown to relate with decreases in feeling of stress and increase in feeling of relaxation
and vigor (Dubnov & Berry, 2000). Some popular psychological hypothesis regarding the
mechanisms by which physical activity could lead to affective changes include the mastery
hypothesis (i.e., engaging in physical activity may increase people’s sense of mastery or
achievement and thereby resulting in improved affect; Norris, Carroll, & Cochrane, 1990), the
distraction hypothesis (i.e., engaging in physical activity can provide a temporary break from life
stresses or worries, which is responsible for improved affect; Morgan, 1985), and the social
interaction hypothesis (i.e., the social relationships developing from engaging in physical activity
contributes to the positive effects on people’s subjective feelings; Vilhjalmsson & Thorlindsson,
1992).

16
Acute Effects of Affective States on Subsequent Physical Activity
A review of current evidence suggests that only a few studies have examined whether
affective state predicts subsequent physical activity in people’s everyday lives (Liao et al.,
unpublished). Assessing participants’ affective state randomly (up to 10 times per day) for 5
days, Wichers and colleagues found that positive affect was significantly higher about 180
(versus 90) minutes before increased activity was observed, thus concluded that a decrease in
positive affect may trigger the onset of physical activity (Wichers et al., 2012). However,
physical activity for this study was assessed by one self-report question asking about physical
activity level on a 7-point Likert scale. Therefore, validity and reliability of this self-reported
physical activity may be questionable. Another study using similar methods (e.g., affective state
was randomly assessed throughout the day up to 8 times per day for 8 days, physical activity was
self-reported) found no association between affective state and subsequent physical activity in 90
minutes (Mata et al., 2012). While these studies were among the first attempts to examine
whether affective states predict immediate physical activity in free-living situations, their
findings are limited by their methodological design (e.g., self-report physical activity, affective
states were assessed at least 90 minutes after physical activity).
Acute Effects of Physical Activity on Subsequent Affective States
Several review studies provide the evidence for increased positive affect and/or reduced
negative affect post exercise across a variety of populations (e.g., Arent, Landers, & Etnier,
2000; Petruzzello et al., 1991; Reed & Ones, 2006). These exercise-induced affective changes
often peak within 5 minutes post-exercise (e.g., Petruzello, Hall, & Ekkekakis, 2001; Petruzzelo,
Jones, & Tate, 1997; Steptoe & Cox, 1988; Steptoe, Kearsley, & Walters, 1993), and these
affective changes could remain significant for 10 to 30 minutes (e.g., Bixby, Spalding, &

17
Hatfield, 2001; Ekkekakis, Hall, VanLanduyt, & Petruzzello, 2000; Steptoe & Cox, 1988).
Findings from these review studies also suggest that intensity of the activity might play an
important role in the acute affective responses to physical activity.
There are mixed results and theories regarding how activity intensity might influence
immediate affective response. Some empirical studies suggest that low intensity activity
increased positive affect (e.g., Ekkekakis et al., 2000) while high intensity activity increased
negative affect (e.g., Pronk, Crouse, & Rohack, 1995). Other studies suggest that emotional
responses to physical activity are non-linear, and the optimal psychological benefits occur
following moderate, but not low or high intensity activity (e.g., the inverted-U curve, Ojanen,
1994). Some researchers have argued that high intensity physical activity may lead to affective
improvement through a “rebound” effect. For example, people feel improved affective states
(e.g., increased positive affect, decreased negative affect) after a high intensity activity because
of the affective decline during that activity (e.g., the opponent-process theory, Solomon, 1980;
the dual-mode theory, Ekkekakis, 2003). Nevertheless, results from meta-analytic reviews
showed that all levels of intensity were associated with positive improvement in affective
response (i.e., increased positive affect and/or decreased negative affect), although the effect size
for low intensity activity was larger than moderate or high intensities (Reed & Ones, 2006). In
summary, even though researchers have several different theoretical hypotheses regarding the
dose-response relationship between activity intensity and acute affective responses, empirical
evidence does not yield consistent results supporting these theories.
As discussed above, a number of studies have attempted to elucidate the acute affective
response to physical activity. However, most of these previous studies were carried out in
controlled laboratory settings (i.e., people were asked to perform prescribed activities in a

18
laboratory, or in any settings that are not part of their normal daily lives; e.g., Bixby et al., 2001;
Ekkekakis et al., 2000; Petruzello et al., 2001). Lab studies may have the advantage of having a
precise control over experimental conditions (e.g., intensity of the exercise, exact start and stop
time point for each exercise session) and accurate and comprehensive measurements (e.g.,
multiple movement sensors on the body, heart rate monitors, electroencephalograms) to monitor
physiological and biological changes. Nevertheless, one of the main concerns about results from
these lab studies is their potential low external and ecological validity. Lab studies tend to make
assumptions that the activity being measured under laboratory conditions is representative of the
behavior performed in daily life outside the laboratory. However, the artificial setting and the
fact that a person is well aware of being extensively watched could induce very different
behavior, physiological, and psychological processes from those which would otherwise occur in
normal daily life (Bussmann et al., 2009). For example, Horemans and colleagues found that the
heart rate while walking at a self-preferred speed on a closed, indoor, oval-shaped track was
significantly lower than walking in daily life (Horemans, Bussmann, Beelen, Stam, & Nollet,
2005). Further, for people that are unfamiliar with laboratory settings and/or exercise equipment,
unpleasant emotions such as anxiety could arise (e.g., Kerr & Kuk, 2001; McAuley, Mihalko, &
Bane, 1996). Therefore, in order to more accurately understand whether physical activity and
affective states acutely influence each other in people’s daily lives, both physical activity and
affective states need to be assessed in free-living settings.
In recent years, there have been an increasing number of studies that aim to examine the
acute affective response to physical activity during the course of people’s everyday lives. To
measure physical activity, these studies either asked participants to self-record the type, duration,
and intensity every time they exercised (e.g., Carles, Coit, Young, & Berger, 2007; Guérin,

19
Fortier, & Sweet, 2013) or used objective measurement (i.e., accelerometry) to continuously
record people’s movement throughout the day (e.g., Kanning, 2012; Schwerdtfeger, Eberhardt,
Chmitorz, & Schaller, 2010). To measure affective states, participants were either instructed to
record their affective state after they completed each exercise session (e.g., Guérin et al., 2013),
or were randomly prompted to report their current affective state several times a day (e.g.,
Schwerdtfeger et al., 2010), or both (e.g., Gauvin, Rejeski, & Norris, 1996). An increase in
positive affect was found immediately after (e.g., Carles et al., 2007; Gauvin et al., 1996), and
10-, 15-, and 30-minute after (e.g., Kanning, 2012; Schwerdtfeger et al., 2010) engaging in
physical activity in people’s daily lives. For negative affect, some studies found a significant
decrease immediately after engaging in physical activity (e.g., Gauvin et al., 1996), some found a
significant increase 15- and 30-minute after the end of the physical activity bout (e.g.,
Schwerdtfeger et al., 2010), and some did not find a significant association around 90 minutes
after a self-report physical activity episode (e.g., Mata et al., 2012) or an increase in physical
activity (e.g., Wichers et al., 2012). Overall, these studies offer some initial evidence of affective
responses to physical activity in daily lives, although each study varied in how physical activity
and/or affect was being measured.
Current Research Gaps
To date, there are limited studies examining the acute effects of affective states on
subsequent physical activity. A better understanding of this effect could shed light on people’s
decision-making regarding their daily physical activity. Meanwhile, learning the acute effects of
physical activity on subsequent affective states could help explain people’s willingness of future
engagement in physical activity. Nevertheless, present knowledge about acute affective
responses to physical activity largely comes from studies conducted in controlled laboratory

20
settings. Although these studies offer some important insights about the acute relationships
between physical activity and affective states (e.g., relationships might vary by intensity of the
activity), they bear potentially low external and ecological validity since people’s behaviors and
emotions differ between lab vs. free-living settings. Further, it is difficult for lab studies to test
whether affective states could trigger physical activity engagement (i.e., making a decision about
whether to engage in a behavior or not) in real-life, everyday situations. While there have been
several recent studies investigating the acute relationships between physical activity and
affective states in free-living settings, these studies also had a number of methodological
limitations (e.g., Gauvin et al., 1996; Carles, et al., 2007; Kanning, 2012; Mata et al., 2012;
Wichers et al., 2012). First, most studies relied on participants’ self-reported physical activity,
which might have low reliability (especially for studies that only used one item to assess physical
activity level). Second, although majority of these studies used ecological momentary
assessment (EMA) to assess affective state, not all of them utilized electronic EMA methods.
Electronic EMA in general yields a higher compliance rate than paper-pencil diaries, and is able
to give an exact time stamp when each assessment is completed. The later feature is especially
useful when linking EMA data with other types of real-time data (e.g., accelerometer data).
Study Aims
To address the present research gaps in this topic, the present study used real-time data
that was collected in free-living environments to examine the acute bi-directional relationship
between physical activity and affective states. Participants’ affective states were assessed using
electronic EMA methods, and physical activity was objectively measured via accelerometer.
Thus, these real-time data collection methods allow a more ecologically valid and reliable way to
test the acute associations between physical activity and affective states. As discussed earlier,

21
since affective responses to physical activity could be influenced by the intensity of the activity,
physical activity (LPA) and moderate-to-vigorous activity (MVPA) were examined separately.
This study aimed to answer the following research questions:
1. Do affective states acutely predict subsequent physical activity in free-living
situations?
It was hypothesized that having a more positive affective state at the beginning of a time
window would lead to more minutes spent in LPA/MVPA within that time window, and having a
more negative affective state at the beginning of a time window would lead to less minutes spent
in LPA/MVPA within that time window.
2. Is physical activity acutely associated with subsequent affective states in free-living
situations?
It was hypothesized that more minutes spent in LPA/MVPA within each time window
would lead to a more positive affective state at the end of that time window.
Methods
Participants
This study used baseline data from Project Measuring Our Behaviors in Living
Environments (MOBILE), which investigated the effects of environmental and intrapersonal
factors on health behavior decision-making processes. Participants were low-active (i.e.,
engaged in <150 minutes/week physical activity) adults living in Chino, California, or a
surrounding community. Individuals were excluded for participating in the study if they (a) did
not speak and read English fluently; (b) had annual household income greater than $210,000; (c)
had physical disabilities limiting physical activity. All participants were also required to be able
to answer electronic EMA surveys while at work. A total of 117 participants were recruited to
participate in the study.

22
Study Protocols
Data collection. Eligible participants were scheduled for a data collection appointment
at a local community site or their home. Participants received monitoring equipment with verbal
and written instructions. Height and weight were measured by study staff using an electronically
calibrated digital scale (Tanita WB-110A) and professional stadiometer (PE-AIM-101).
Participants also filled out a paper-pencil survey, which assessed their demographic information.
EMA. Electronic EMA surveys were delivered through an HTC Shadow mobile phone
(T-Mobile USA, Inc.). A custom software program (MyExperience) was installed in each phone
as a platform to randomly prompt the EMA survey and store the survey responses. All other
functions of the mobile phone were disabled. Eight EMA surveys were prompted each day from
Saturday to Tuesday (up to 32 total surveys total). Each EMA survey was prompted at a random
time within eight pre-programmed windows (between 6:30 am to 10:00 pm) to ensure adequate
sampling spacing across the day. EMA surveys were prompted using an auditory signal. Upon
receiving the signal, participants were instructed to complete a short question sequence on the
display screen. If a survey prompt was not answered (i.e., no response entry was made), the
mobile phone emitted up to three reminder signals at 5-minute intervals. After the third
reminder, the EMA survey became inaccessible until the next prompt. Each prompted EMA was
time-stamped.
Accelerometry. The Actigraph, Inc., GT2M model (firmware v06.02.00) accelerometer
was used as an activity monitor to objectively measure participants’ physical activity. This
device was attached to an adjustable belt and placed on participants’ right hip. Participants were
instructed to wear this belt during their waking hours across the 4-day EMA monitoring period.

23
This device continuously recorded participants’ activity intensity (as expressed in activity
counts) every 30-second. Each accelerometer recording was time-stamped.
Measures
Affective states. The current study used EMA items that assessed current positive affect,
negative affect, energy, and fatigue. To reduce participant burden, each set of question(s) was
designed to appear in a randomly programmed 6 out of the 8 daily question sequences (75% of
sequences). The positive affect scale contains three questions assessing current feelings of
happy, cheerful, and relaxed (Cronbach’s α = .837). The negative affect scale contains four
questions assessing current feelings of stressed, angry, anxious, and depressed (Cronbach’s α =
.865). Energy was assessed through one item asking about how energetic they were feeling.
Fatigue was assessed through one item asking about how tired they were feeling. Response
choices for these items were “1=not at all, 2=a little, 3=moderately, 4=quite a bit, 5=extremely”
(see Figure 1 for sample screenshots for EMA items). Therefore, “improved/better affective
states” refer to higher positive affect, lower negative affect, higher energy, and lower fatigue.
Physical activity. Activity counts from the accelerometer were converted to minutes
spent in LPA and MVPA. The cut-point for MVPA was defined as 2,020 activity counts per
minute, which is consistent with national surveillance studies (Belcher et al., 2010; Troiano et
al., 2008). LPA was defined as time that was not spent in MVPA and sedentary activity (i.e.,
less than 100 counts per minute; Healy et al., 2008).
Weight category. Body mass index was calculated as kg/m
2
. Weight category was
classified as normal weight (BMI<25), overweight (25≤BMI<30), and obese (BMI≥30).
EMA time variables. In addition to the exact time when the participant answered a
prompt, each EMA prompt was also coded for day of the week (i.e., weekdays vs. weekend

24
days), and time of day (i.e., morning [6:30 am to 11:59 am], afternoon (12:00 pm to 5:59 pm),
vs. evening [6:00 pm to 10:00 pm]).
Real-time Window
The internal clocks of the mobile phone and accelerometer were synchronized to the
same computer each time before giving out to participants. Thus, data points from the two
devices could be time-matched. To test the acute relationships between physical activity and
affective states, one 15-minute and one 30-minute real-time window were created before and
after each answered EMA survey. Activity data was then summarized within each of the time
windows. Thus, the physical activity measure was expressed as: (1) total minutes spent in LPA
within the 15 minutes before each EMA prompt; (2) total minutes spent in LPA within the 30
minutes before each EMA prompt; (3) total minutes spent in MVPA within the 15 minutes
before each EMA prompt; (4) total minutes spent in MVPA within the 30 minutes before each
EMA prompt; (5) total minutes spent in LPA within the 15 minutes after each EMA prompt; (6)
total minutes spent in LPA within the 30 minutes after each EMA prompt; (7) total minutes spent
in MVPA within the 15 minutes after each EMA prompt; (8) total minutes spent in MVPA
within the 30 minutes after each EMA prompt (see Figure 2 for illustration of the 15-minute time
window summarizing the total MVPA minutes within each window).
Statistical Analysis
Only answered EMA prompts were included in the analysis, and their time stamps (i.e.,
the exact time when participants answered the EMA prompt) were used to create the time
windows. A total of zero activity counts within 60 minutes surrounding (i.e., 30-minute before
and 30-minute after) each EMA prompt was considered as accelerometer non-wear (i.e., invalid
accelerometer data), and that EMA entry was excluded from analyses.

25
Research question 1. To test whether affective states acutely predict subsequent
physical activity, multilevel linear regression analysis was used with physical activity level (i.e.,
total LPA/MVPA minutes within the 15-/30-minute window after each EMA prompt) as the
outcome, and affective state at each prompt as the predictor. In summary, there were a total of 4
outcomes: (1) total LPA minutes within the 15-minute window after the EMA prompt, (2) total
MVPA minutes within the 15-minute window after the EMA prompt, (3) total LPA minutes
within the 30-minute window after the EMA prompt, and (4) total MVPA minutes within the 30-
minute window after the EMA prompt; and 4 predictors: (1) positive affect, (2) negative affect,
(3) energy, and (4) fatigue. Each pair of outcome and predictor was tested in separate multilevel
models, one at a time.
Although affective states were assessed across multiple days, they were not expected to
vary as a function of time (i.e., positive affect increased from day 1 to day 4). Therefore, it is
necessary to remove the linear trend from the data, if any. To do this, the mean of the prompt-
specific affective states within each individual (i.e., 𝐴 𝐹𝐹𝐸𝐶 𝑇 � � � � � � � � � � �
𝑖 ) was computed, which is denoted
as 𝐴 𝐹𝐹𝐸𝐶 𝑇 𝐵𝑃
; then the person-specific mean was subtracted from each individual’s prompt-
specific affective states to obtain 𝐴 𝐹𝐹𝐸𝐶 𝑇 𝑊𝑃 . Next, 𝑃𝑟 𝑜𝑚𝑝𝑡
𝑡 𝑖 was used to represent the
measure of time, which is the chorological order of each prompt (ranged from 1 to 32 in our
case). Finally, the 𝐴 𝐹𝐹𝐸𝐶 𝑇 𝑊 𝑃 was used as the level-1 predictor, controlling for 𝑃𝑟 𝑜𝑚𝑝𝑡
𝑡 𝑖 . This
way, the prompt-level affective states used in the regression model represented the deviation
with respect to the individual-specific regression linking the affective states and time (Curran &
Bauer, 2011).
Further, since time-varying variables (e.g., the prompt-level affective states) could vary
over individuals and prompts, they simultaneously contain both within-person and between-

26
person variability. Thus, it was necessary to disaggregate the between- and within-person effects
within the multilevel model. To do this, both 𝐴 𝐹𝐹𝐸𝐶 𝑇 𝐵𝑃
and 𝐴 𝐹𝐹𝐸𝐶 𝑇 𝑊𝑃 were used as
predictors in the regression model. This way, the between-person (i.e., how individuals were
different from each other) and within-person (i.e., how individuals fluctuated within him/herself
across prompts) effects were disaggregated (Curran & Bauer, 2011).
Additionally, all models controlled for the total LPA/MVPA minutes within the 15-/30-
minute window before the EMA prompt. This way, activity levels prior to answering the EMA
prompt could be accounted for (e.g., if someone already engaged in physical activity prior to
answering the EMA prompt). This also allowed examining effects of affective states on change
in activity levels. Lastly, potential confounders for physical activity were screened for
significance in all models, one at a time. These potential confounders were determined a prior,
which included both person-level variables (i.e., age, gender, ethnicity, annual household
income, and weight category) and prompt-level variables (i.e., day of week and time of day).
Significant confounders were retained as a covariate in the final model.
The following equation represents a generic multilevel linear regression model as
outlined above:
level-1
𝑃𝐴 𝑡 𝑖 = 𝛽 0 𝑖 + 𝛽 1 𝑖 𝐴 𝐹𝐹𝐸𝐶 𝑇 𝑊𝑃 + 𝛽 2 𝑖 𝑃𝑟 𝑜𝑚𝑝𝑡
𝑡 𝑖 + 𝛽 3 𝑖 𝑃𝐴 𝑏𝑒𝑓𝑜 𝑟𝑒
+ 𝛽 4 𝑖 𝐶𝑂 𝑉 𝑡 𝑖 + 𝑟 𝑡 𝑖
level-2
𝛽 0 𝑖 = 𝛾 0 0
+ 𝛾 0 1
𝐴 𝐹𝐹𝐸𝐶 𝑇 𝐵𝑃
+ 𝛾 0 2
𝐶𝑂 𝑉 𝑖 + 𝑢 0 𝑖
𝛽 1 𝑖 = 𝛾 1 0
+ 𝛾 1 1
𝐶𝑂 𝑉 𝑖
where 𝑃𝐴 𝑡 𝑖 is the physical activity level after prompt t’s for individual i, 𝛽 0 𝑖 and 𝛽 1 𝑖 represent
the intercept and linear slope for individual i, 𝑃𝐴 𝑏𝑒𝑓𝑜 𝑟𝑒
is the total LPA/MVPA minutes within

27
the 15-/30-minute window before the EMA prompt, 𝛽 4 𝑖 represents the between-person effect for
a prompt-level covariate such as time of day and day of week, 𝑟 𝑡 𝑖 represents the random error
associated with the prompt t for individual i, 𝛾 0 0
represents the sum of an overall mean at the
individual level (i.e., grand mean), 𝛾 0 1
is the between-person effect, 𝛾 1 0
is the overall mean
slope, 𝑢 0 𝑖 represents a series of random deviations from that mean, and 𝐶𝑂 𝑉 𝑖 is a person-level
covariate such as age, gender, race/ethnicity, and weight category.
Note that the total MVPA minutes within the 15- and 30-minute windows were not
normally distributed, therefore, log-transformation was performed for these two outcomes.
However, since a large amount of observations had zero MVPA minutes, which led to missing
data after log-transformation. In this case, a two-piece model was necessary so that the first
model (i.e., Piece 1 Model) was a multilevel logistic regression analysis to predict the probability
of engaging in no MVPA (i.e., zero MVPA minutes) versus some MVPA (i.e., non-zero MVPA
minutes); then the second model (i.e., Piece 2 Model) was a multilevel linear regression analysis
to predict the log-transformed non-zero MVPA minutes. Mplus was used to run this two-piece
model. For LPA minutes, since log-transformation was unnecessary, SAS PROC MIXED was
used to run the multilevel linear regression models.
Research question 2. To test whether physical activity acutely associated with
subsequent affective states, similar multilevel linear regression models were used as detailed
above. The outcome variable was current affective state at the end of a 15-/30-minute window.
The predictor variable was total LPA/MVPA minutes within the 15-/30-minute window before
the EMA prompt. Therefore, there were a total of 4 outcomes: (1) positive affect, (2) negative
affect, (3) energy, and (4) fatigue; and 4 predictors: (1) total LPA minutes within the 15-minute
window before the EMA prompt, (2) total MVPA minutes within the 15-minute window before

28
the EMA prompt, (3) total LPA minutes within the 30-minute window before the EMA prompt,
and (4) total MVPA minutes within the 30-minute window before the EMA prompt. Each pair
of outcome and predictor was tested in separate multilevel models. Note that only negative
affect was not normally distributed, therefore, it was log-transformed before fitting into the
multilevel linear regression models. All models were fitted using SAS PROC MIXED.
Similar to controlling for prior physical activity in research question 1, prior affective
state for this research question needs to be controlled for in order to determine change. Since the
time between two EMA prompts varied due to various reasons (i.e., the affect questions were not
asked during all EMA prompt sequences, missing prompt by participants), time between current
EMA and the closest prior EMA prompt was also controlled for in all models. Further, to keep
the “acute” nature of the research question, if the closest prior EMA prompt fell into the day
before (e.g., the morning prompt’s closest prior prompt would be the one occurred the night
before), that EMA prompt was excluded from the analyses.
Power Analysis
Power analysis was conducted using G*Power (version 3.1). Linear bivariate regressions
(i.e., relationships between physical activity and affective states) were applied with α level at .05
and power at .80. The prompt-level standard deviations ranged from .67 to 1.13 for affective
states; from 1.21 to 3.25 for physical activity after each EMA prompt; and from 1.36 to 3.60 for
physical activity before each EMA prompt. Therefore, assuming a sample size of 110 (i.e., each
participant only has one observation) gives estimated slopes range from .28 to 1.26 for affective
states predicting subsequent physical activity (research question 1); and from .05 to .22 for
physical activity predicting subsequent affective states (research question 2).

29
Since each individual had multiple observations/prompts in the current study, the data is
in a hierarchical structure that involved clustering within individuals. In research designs that
use random samples, statistical power depends on: (1) the significance level of the test (i.e., the α
level), (2) the expected effect size (e.g., the slope of a linear bivariate regression), and (3) the
sample size (Cohen, 1988). However, in multilevel studies, power also depends on the extent of
the clustering effects, which are typically measured by intraclass correlation coefficients (ICC;
Hedges & Rhoads, 2010). Values of ICC range from zero to one. If the ICC is close to zero, it
implies that the within-cluster variance is much greater than the between-cluster variance, which
means there is little clustering within clusters. In this case, the statistical power will be close to
that of a study that used simple random sampling and the same total sample size (i.e., all prompts
were treated as independent observations). If the ICC is near one, it shows that there is no
correlation of responses within a cluster, and therefore, the statistical power will be close to that
of a study that used a simple random sampling and a total sample size equal to the number of
clusters.
SAS PROC MIXED was used to calculate ICC for the prompt-level physical activity and
affective states variables. The ICC for physical activity variables were fairly low (ranged from
.06 to .11), suggesting that there was little clustering within individuals. The ICC for affective
states were moderate (ranged from .25 to .35), suggesting that there was some degrees of
clustering within individuals.
For a two-level hierarchical structure data, the total actual sample size is
𝑁 = 𝑚𝑛

30
where m is the number of participants, and n is the prompt-level observation for each participant.
The following two equations define the operational effect size (OES) and effective sample size
(ESS):
𝑂𝐸𝑆 = 𝛿 �
𝑛 1+(𝑛 −1)𝜌
𝐸𝑆𝑆 =
𝑚𝑛
1+(𝑛 −1)𝜌
where 𝛿 is the actual effect size, and 𝜌 is the ICC.
If each participant had an average of 13 valid observations in this study, we would have
an effective sample size of 650 (vs. the actual sample size of 1430) assuming the ICC is .10. A
sample size of 650 would give us over .80 power to detect a slope of .35 for research question 1
and a slope of .07 for research question 2 (see Appendix for power plots using G*Power).
Results
Descriptive Statistics
Of the 117 participants recruited for the study, EMA data was unavailable for 3
participants due to data downloading problems. Of the remaining 114 participants,
accelerometer device was lost for 2 participants, and data downloading problems occurred for 2
participants. Therefore, a total of 110 participants had available EMA and accelerometer data
(see Table 1 for their demographic characteristics). On average, participants answered 82%
(range 25 – 100%) of the EMA prompts. Of these answered EMA prompts, 86% had valid
accelerometer data. The likelihood of answered vs. unanswered EMA prompts did not vary as a
function of day of the week, time of day, gender, age, race/ethnicity, or weight category.
However, female participants were less likely to have accelerometer non-wear than males (12 %
vs. 19% non-wear rate; beta = -.54, p = .01), and obese participants were more likely to have

31
accelerometer non-wear compared to normal weight participants (17% vs. 11%; beta = .54, p
= .03). Further, compared to evening prompts (9% non-wear rate), accelerometer non-wear was
more likely to occur in morning prompts (27% non-wear rate; beta = 1.35, p < .01), and less
likely to occur in afternoon prompts (5% non-wear rate; beta = -.69, p < .01).
The person-level average for positive affect was 3.06 (SD = 0.63) on a 5-point scale; for
negative affect was 1.44 (SD = 0.41); for energy was 2.63 (SD = 0.67); for fatigue was 2.04 (SD
= 0.61). On average, during the 30-minute window before each answered prompt, participant
spent 6.78 (SD = 2.54) minutes in LPA and 0.72 (SD = 0.85) minutes in MVPA. During the 15-
minute window before each prompt, participants spent an average of 4.12 (SD = 1.35) minutes in
LPA and 0.36 (SD = 0.43) minutes in MVPA. During the 30-minute window after each prompt,
participants on average spent 7.19 (SD = 2.36) minutes in LPA and 0.70 (SD = 0.83) minutes in
MVPA. Participants on average spent 3.49 (SD = 1.25) minutes in LPA and 0.33 (SD = 0.41)
minutes in MVPA during the 15-minute window after each answered prompt.
Affective State and Subsequent Physical Activity
Table 2 shows the results from the two-piece models using MVPA minutes as the
outcome and affective states as the predictors, controlling for prior activity level, time in study,
and significant covariates as indicated for each specific model. In summary, at both within-
person and between-person level, affective state was not associated with the probability of
engaging in some MVPA minutes versus no MVPA minutes in the subsequent 15-minute
window. For the subsequent 30-minute window, only feeling more energetic led to a higher
probability in engaging in some MVPA minutes than no MVPA minutes (coef. = .135, SE =
.064, p = .034).

32
For participants who did engage in some MVPA minutes within the subsequent 15-
minute window, feeling more positive than one’s usual level (i.e., the within-person effect) was
associated with more MVPA minutes (coef. = .093, SE = .038, p = .014); feeling more tired than
one’s usual level (i.e., the within-person effect) was associated with less MVPA minutes (coef. =
-.084, SE = .039, p = .030) during that 15-minute window. Interestingly, feeling more negative
than one’s usual level (i.e., the within-person effect) was associated with less subsequent MVPA
minutes within the 15-minute window (coef. = -.161, SE = .068, p = .017); however, participants
who on average, felt more negative compared to other people in the study, engaged in more
MVPA minutes (i.e., the between-person effect; coef. = .188, SE = .091, p =.039). For
participants who did engage in some MVPA minutes within the subsequent 30-minute window,
all the effects observed in the 15-mintue window became non-significant, except for the positive
between-person effect for negative affect (coef. = .336, SE = .097, p = .001).
Table 3 shows the results from the multilevel linear regression models using LPA
minutes as the outcome and affective states as the predictors, controlling for prior activities, time
in study, and significant covariates as indicated for each specific model. In summary, feeling
more energetic than one’s usual level (i.e., the within-person effect) was associated with more
subsequent LPA minutes in the 15-minute window (beta = .261, SE = .075, p = .001); further,
participants who on average, felt more energetic compared to other people in the study, engaged
in more subsequent LPA minutes (i.e., the between-person effect; beta = .264, SE = .116, p =
.025). These positive effects of feeling energetic on subsequent LPA minutes were also found
for the 30-minute window. In addition, feeling more negative than one’s usual level (i.e., the
within-person effect) was associated with engaging in more subsequent LPA minutes within both
15- and 30-minute windows (beta = .324, SE = .130, p = .012; beta = .633, SE = .244, p = .010;

33
respectively). No significant effect was found for positive affect or fatigue on subsequent LPA
minutes.
Physical Activity and Subsequent Affective State
Table 4 shows the results from the multilevel linear regression models using affective
state as the outcome and physical activity as the predictor, controlling for prior affective state,
time between current and prior affective state, time in study, and significant covariates as
indicated for each specific model. In summary, more MVPA minutes than one’s usual level
during a 15-minute window (i.e., the within-person effect) was associated with feeling more
energetic at the end of this window (beta = .067, SE = .027, p = .013). This effect was also
found for the 30-minute window (beta = .034, SE = .015, p = .021). MVPA minutes were not
associated with any other subsequent affective state for both 15- and 30-minute windows.
More LPA minutes than one’s usual level during a 15-minute window (i.e., the within-
person effect) was associated with feeling more negative (beta = .008, SE = .003, p = .003), more
energetic (beta = .021, SE = .009, p = .021), and less tired (beta = -.019, SE = .008, p = .022) at
the end of this window. The significant within-person effects for energetic and tired were also
found for the 30-minute window. However, the positive within-person effect for negative affect
became marginally significant for the 30-minute window (beta = .003, SE = .002, p = .068).
LPA minutes was not associated with subsequent positive affect for both 15- and 30-minute
windows.
Discussions
The current study used real-time data capture methods, electronic EMA surveys
combined with accelerometer, to examine the bi-directional acute relationships between physical
activity and affective state in adults’ daily lives. Because of the repeated-measurement, this

34
study was able to examine the effects of between-person differences and within-person variations
of those relationships, which could have different implications. For example, by examining the
between-person and within-person effects, we could answer the question of whether happier
people are usually more active than less happy people (i.e., the between-person effect) or feeling
happier than one’s usual level at the moment leads to being more active subsequently (i.e., the
within-person effect). Further, in addition to MVPA, this study also examined the effects of light
activities (i.e., LPA) separately. Engaging in LPA might have different practical implications
from MVPA (e.g., taking a break from sedentary activities by standing up and moving around vs.
making effort to engage in more intense activities).
Affective State Predicting Immediate Subsequent Physical Activity
As hypothesized, feeling more positive, less negative, and less tired than one’s usual level
at the moment led to more MVPA minutes within a subsequent 15-minute window. However,
these effects were not found for the 30-minute window. These results suggest that while more
positive affective states might predict higher physical activity level, such effect may not last very
long. This might partially explain why previous studies that examined a much longer time
window (i.e., 90 minutes) found no association between affective states and subsequent physical
activity (Mata et al., 2012).
Interestingly, results from this study suggest that people who in general felt more
negative actually engaged in more MVPA minutes than people who felt less negative. This
between-person effect of negative affect was in the opposite direction of the within-person effect,
which demonstrated the importance of separating these two effects when analyzing time-
intensive multilevel data. While the within-person effect of negative affect is along the direction
as hypothesized (i.e., feeling less negative led to more subsequent physical activity), the

35
between-person effect is against the cross-sectional and longitudinal evidence of more stressed
people are less physically active (e.g., Paluska & Schwenk, 2000; Sherwood & Jeffery, 2000;
Mouchacca, Abbott, & Ball, 2013). Nevertheless, most of these studies measured people’s
overall perceived stress (i.e., a “static” measure) while our current study captured people’s
negative feeling state “at the moment”. It is possible that people who experienced more daily
hassles and stressful events might more often use physical activity as their stress management
technique, given the fact that physical activity is often recommended as a strategy for stress
coping and is demonstrated to be an effective stress coping techniques (Nguyen-Michel, Unger,
Hamilton, & Spruijt-Metz, 2006; Austin, Shah, & Muncer, 2005; Berger, 1994).
Although feeling more energetic at the moment was not associated with subsequent
MVPA minutes, it increased the probability of engaging in at least some MVPA minutes vs. no
MVPA minutes at all. Further, feeling more energetic was associated with more subsequent
LPA minutes for both 15- and 30-minutes windows. Together, these results imply that although
feeling more energetic at the moment might not lead to higher activity level, it may predict less
time in sedentary activity up to 30 minutes later. Previous studies have shown a positive
relationship between physical activity and feeling of energy, even though most of these studies
were cross-sectional and a temporal relationship cannot be established (Puetz, 2006). Further,
the positive relationship between energy and LPA found in this study was also observed at the
between-person level. This finding suggests that people who on average feel more energetic
during their everyday lives might spend less time in sedentary activity than people who feel less
energetic.
Opposite to the findings for MVPA, results from this study suggest that feeling more
negative at the moment than one’s usual level led to more LPA minutes within the subsequent

36
15- and 30-minute windows. Since higher negative affect is associated with higher frequency of
daily hassles and stressful events (e.g., van Eck, Nicolson, & Berkhof, 1998; Almeida,
Wethington, & Kessler, 2002; Jacobs, et al., 2007), it is possible that people would need to
physically attend to those events (e.g., running errands, moving around back and forth) so that
they were not in a sedentary position (e.g., sitting down) nor engaging in activities that hit the
moderate intensity. Further, unlike MVPA, momentary positive affect and fatigue were not
associated with subsequent LPA minutes, which suggests that effect of these two specific
affective states might only be relevant with higher intensity of activities.
Physical Activity Predicting Immediate Subsequent Affective State
Results from this study suggested an increase in energy after engaging in more physical
activity than one’s usual level. This positive relationship between energy and physical activity
was found for both MVPA and LPA, and for both 15- and 30-minute windows. In other words,
as long as people spent less sedentary minutes than what they usually did for the past 15- or 30-
minute, they would feel more energetic. This finding is consistent with previous free-living
studies that showed a significant increase in energy following physical activity bouts (Kanning,
2012; Kanning, Ebner-Priemer, & Brand, 2012; Gauvin et al., 1996). In addition, the current
study showed that spending more time in LPA than one’s usual level within the past 15- and 30-
minute period led to feeling of less tired. Nevertheless, no association was found between
fatigue and MVPA. This null finding is consistent with previous studies that examined change in
physical exhaustion before and after self-reported physical activity bouts (Gauvin et al., 1996).
Therefore, it is likely that the effect of decrease in fatigue might only exist for less intense
activities, or the so-called “non-sedentary, non-exercise activity”. More recently, the non-
exercise activity thermogenesis (NEAT) has drawn increasing attention because of its potential

37
health benefits (e.g., lower risk of metabolic syndrome; Uemura et al., 2013). Results from the
current study further indicate that the non-sedentary, non-exercise activity seemed to produce an
overall positive physical feeling (i.e., increase in energy and decrease in fatigue).
Spending more minutes in MVPA than one’s usual level was not associated with
subsequent positive and negative affect. These null findings are consistent with several other
studies that also examined the affective response from physical activity in free-living settings
(Mata et al., 2012; Wichers et al., 2012; von Haaren et al., 2013). These findings from free-
living studies are contradictory to those lab-based studies where exercise induced affective
changes were observed (e.g., Bixby, Spalding, & Hatfield, 2001; Ekkekakis, Hall, VanLanduyt,
& Petruzzello, 2000; Petruzello, Hall, & Ekkekakis, 2001; Steptoe & Cox, 1988). It is possible
that when in a lab setting, people are more aware of their changes in affective state simply
because they are being monitored. Consequently, they might be able to feel and recognize subtle
changes in positive and negative affect under lab settings but not free-living settings. The
discrepancy between lab-based and free-living studies could also due to the type of physical
activity being performed. For lab-based studies, some most common forms of exercise include
treadmill running/walking and stationary bike pedaling. While in free-living settings, people
could exercise from a much wider range of choices, such as walking in the neighborhood,
running in a park, aerobic exercising at a gym, weightlifting at home, etc. It is possible that the
diverse types of physical activity that people engage in during their daily lives contribute to non-
consistent results in affective response from physical activity. Therefore, although evidence
from lab studies supports acute “feel-good” effects from physical activity, people might not
experience the same mood-enhancing effects when they are active in their daily lives. Notably,
the current study found that more LPA minutes in the past 15-minute was associated with an

38
increase in negative affect. Again, as discussed earlier, it is possible that engaging in light
activities might imply running errands, dealing with hassles, and handling other stressful events
that happen in people’s daily lives, which may result in an increase in negative affect.
Limitations
Despite the combined use of electronic EMA to assess current affective states and
accelerometer to objectively measure physical activity in free-living settings, this study has
several limitations. First, we captured physical activity by real-time windows, rather than by
activity bouts/episodes. It is possible that we captured only part of one physical activity session
(e.g., the last 15 minutes of a 60-minute running session), which would imply that we might have
included affective responses during physical activity in our analysis. It is also possible that we
captured activity bouts that would not be considered as “physical activity” by participants (e.g.,
brisk walk for an errand). However, the EMA method used in the study was not designed to
capture the physical activity that participants engaged in throughout the day, rather, it was
designed to randomly capture people’s behaviors and affective states (i.e., snapshots) as they
occurred.
Secondly, even though multiple items were used to create a composite score for positive
affect and negative affect, energy and fatigue were measured using only a single EMA item.
However, due to the repeated-measure nature of the EMA method, it is necessary to keep each
assessment relatively short (i.e., so participant can finish each EMA assessment in less than 2-3
minutes) in order to minimize participant burden. The current study examined effects of the four
affective states (i.e., positive affect, negative affect, energy, and fatigue) in separate models.
Since people’s emotional state can be multi-dimensional, it is possible that people could feel
happy and anxious at the same time. By examining the affective states separately, we ignored

39
the potential confounding effects from the opposite dimension of feeling state that people might
experience at that moment. Nevertheless, bivariate correlations at the prompt level showed that
these four affective states were moderately correlated with each other (absolute value for
Pearson’s r ranged from .3 to .6, with the exception of negative affect and energy, where
Pearson’s r = -.20). Further, the current study did not capture other affective feelings that might
be related to physical activity (e.g., enjoyment, discomfort, boredom).
Thirdly, this study only collected data over the course of 4 days. Although these 4 days
encompassed both weekdays and weekend days, this short monitoring period might not be fully
representative of adults’ usual daily behaviors and affective states. Additionally, since 2 of the 4
days were weekend days, weekday behaviors might be underrepresented in this study.
Furthermore, we found systematically missing patterns for accelerometer non-wear. Missing
accelerometer data was more likely to occur in males than females, in obese participants than
normal weight participants, and in the mornings than in the evenings. Thus, results from this
study might not be representative for males, obese people, and activities that occurred in the
morning. Lastly, all of the study participants were inactive adults (i.e., engaged in <150
minutes/week physical activity). Therefore, findings from this study might not be generalizable
to more active adults.
Implications
The current study demonstrated using real-time data capture techniques (i.e., electronic
EMA and accelerometer) to create various time windows (i.e., snapshots) in people’s everyday
lives. By creating this time windows, temporal relationships between physical activity and
affective states could be tested. This data structure can be a helpful way to study antecedents and
consequences of people’s daily behaviors. Further, the time-intensive longitudinal data (i.e.,

40
collecting information repeatedly) allows researchers to examine the between-person effect and
within-person effect of a relationship. As results from this study suggest, sometimes these two
effects could be in opposite directions (i.e., the effect of negative affect on subsequent MVPA
minutes). Fail to disaggregate the within-person from the between-person effects might lead to
potential “ecological fallacy” (i.e., making conclusions about relationships within individuals on
the basis of differences between individuals; Fleeson, 2009). Therefore, it is important to
separate the within-person and between-person effects when analyzing time-intensive
longitudinal data.
Overall, results from this study indicate that having a more positive (e.g., feeling more
positive such as happy and relaxed) and less negative (e.g., feeling less negative such as stressed
and anxious) affective state was associated with being more active subsequently. This finding
might reflect the emotional state that prepares people (e.g., mentally ready) to engage in daily
physical activity. This also implies that a more negative affective state might be an emotional
barrier for people to become more active in their everyday lives. Thus, interventions could
potentially offer strategies to help people overcome this emotional barrier.
Future Directions
This study showed momentary affective states could predict subsequent physical activity
level. Future studies might explore some potential mediators of this relationship. For example,
affective states might influence an individual’s cognitive states (e.g., motivation, self-efficacy,
intention), which then affect the subsequent physical activity level (Rhodes & Nigg, 2011; Loehr,
Baldwin, Rosenfield, & Smits, 2014). Future studies could also further examine whether
physical (e.g., indoor vs. outdoor) and social context (e.g., alone vs. with friends) might
moderates people’s affective change following physical activity. Lastly, the current study

41
suggested that feeling more energetic led to being more physically active; and being more
physical active led to feeling more energetic. Nevertheless, in order to test whether this positive
feedback loop exists, a more systematic statistical approach is needed. For example, the
dynamical system modeling method would be able to examine the potential time-varying,
nonlinear relationship between physical activity and affective states (Riley et al., 2011).
Conclusions
In summary, by analyzing real-time data from electronic EMA and accelerometer, this
study showed that individuals in this study were more likely to be physically active if they were
in an emotionally positive state. And although engaging in physical activity might not produce
an immediate mood-enhancing effect, it might give people a feeling of energy boost.

42
Table 1. Participant demographic characteristics (N=110).
Age Mean (SD)

40.4 (9.74)
Gender
Male
n (%)
30 (27.5)
Female 79 (72.5)
Race/Ethnicity
Hispanic/Latino 33 (30.3)
Non-Hispanic/Latino 77 (69.7)
Annual Household Income
Less than $40,000 25 (23.6)
$40,000 - $70,000 24 (22.6)
$70,001 - $90,000 27 (25.5)
Above $90,000 30 (28.3)
Weight Category
Underweight/Normal Weight 42 (38.2)
Overweight 34 (30.9)
Obese 34 (30.9)

43

Table 2. Associations between affective state and subsequent minutes in moderate-to-vigorous physical activity (MVPA)
15 Minute Window

30 Minute Window
Piece 1 Model
(Some vs. no MVPA minutes)
1
Piece 2 Model
(MVPA minutes)
2
Piece 1 Model
(Some vs. no MVPA minutes)
1

Piece 2 Model
(MVPA minutes)
2

p Coefficient
Estimate (SE)
p Coefficient
Estimate (SE)
p Coefficient
Estimate (SE)
p Coefficient
Estimate (SE)
Positive
Affect
WP Effect -0.117 (0.089)

0.191 0.093 (0.038) 0.014 -0.073 (0.088)
c
0.407 0.052 (0.037)
c
0.162
BP Effect -0.105 (0.137)
ab
0.442 -0.023 (0.068)
ab
0.735 -0.063 (0.130) 0.627 -0.058 (0.064) 0.365

Negative
Affect
WP Effect 0.175 (0.115) 0.130 -0.161 (0.068) 0.017 0.186 (0.109)
c
0.087 -0.109 (0.068)
c
0.113
BP Effect -0.005 (0.169) 0.976 0.188 (0.091) 0.039 -0.085 (0.237)
de
0.718 0.336 (0.097)
de
0.001

Energy
WP Effect 0.087 (0.074) 0.237 0.072 (0.042) 0.087 0.135 (0.064) 0.034 0.054 (0.032) 0.090
BP Effect 0.022 (0.122)
b
0.855 0.039 (0.062)
b
0.534 0.101 (0.107) 0.348 0.012 (0.057) 0.826

Fatigue
WP Effect -0.080 (0.073) 0.276 -0.084 (0.039) 0.030 -0.089 (0.061) 0.148 -0.054 (0.040) 0.182
BP Effect -0.187 (0.187) 0.318 -0.024 (0.062) 0.691 -0.191 (0.167) 0.252 0.027 (0.065) 0.681
Note: WP = within-person, BP = between-person. All models controlled for prior activity (i.e., MVPA minutes in the prior
corresponding 15-/30-minute window) and the chorological order of each prompt (i.e., time in study).
1
Multilevel logistic regression model predicting the probability of engaging in some MVPA minutes vs. no (i.e., zero) MVPA minutes.
2
Multilevel linear regression model predicting the log-transformed non-zero MVPA minutes.
a
Indicates the model additionally controlled for gender.
b
Indicates the model additionally controlled for ethnicity.
c
Indicates the
model additionally controlled for time of day.
d
Indicates the model additionally controlled for household income.
e
Indicates the
model additionally controlled for weight category.

44

Table 3. Associations between affective state and subsequent minutes in light physical activity
(LPA)
15 Minute Window 30 Minutes Window
p Beta (SE) p Beta (SE)
Positive Affect
WP Effect 0.041 (0.091)
a
0.653 0.120 (0.174)
a
0.490
BP Effect 0.244 (0.123) 0.050 0.367 (0.233) 0.118

Negative Affect
WP Effect 0.324 (0.130) 0.012 0.633 (0.244)
a
0.010
BP Effect -0.220 (0.189) 0.248 -0.383 (0.376) 0.311

Energy
WP Effect 0.261 (0.075)
a
0.001 0.492 (0.144)
a
0.001
BP Effect 0.264 (0.116) 0.025 0.497 (0.218) 0.025

Fatigue
WP Effect -0.115 (0.077) 0.134 -0.194 (0.145) 0.181
BP Effect -0.187 (0.130)
b
0.153 -0.417 (0.243) 0.089
Note: WP = within-person, BP = between-person. All models controlled for prior activity (i.e.,
LPA minutes in the prior corresponding 15-/30-minute window) and the chorological order of
each prompt (i.e., time in study).
a
Indicates the model additionally controlled for time of day.
b
Indicates the model additionally
controlled for weight category.

45

Table 4. Associations between physical activity
1
and subsequent affective states
Positive Affect Negative Affect
2
Energy Fatigue
p Beta (SE) p Beta (SE) p Beta (SE) p Beta (SE)
Total MVPA Minutes
in 15-Minute Window
WP Effect 0.040 (0.023) 0.082 -0.008 (0.008)
a
0.327 0.067 (0.027)
a
0.013 0.020 (0.024)
a
0.400
BP Effect -0.155 (0.140) 0.270 -0.014 (0.057) 0.802 -0.103 (0.144) 0.477 0.122 (0.138)
b
0.380

Total MVPA Minutes
in 30-Minute Window
WP Effect 0.019 (0.013) 0.123 -0.006 (0.005)
a
0.211 0.034 (0.015)
a
0.021 0.012 (0.014)
a
0.396
BP Effect -0.048 (0.070) 0.499 -0.028 (0.029) 0.335 -0.025 (0.073) 0.735 0.022 (0.069)
b
0.753

Total LPA Minutes
in 15-Minute Window
WP Effect -0.007 (0.007) 0.348 0.008 (0.003) 0.003 0.021 (0.009)
a
0.021 -0.019 (0.008)
a
0.022
BP Effect 0.017 (0.040) 0.665 -0.017 (0.016) 0.282 0.019 (0.041) 0.657 -0.045 (0.039)
b
0.249

Total LPA Minutes
in 30-Minute Window
WP Effect 0.001 (0.004) 0.889 0.003 (0.002)
a
0.068 0.016 (0.005)
a
0.003 -0.017 (0.005) 0.001
BP Effect 0.014 (0.021) 0.517 -0.011 (0.008) 0.196 0.012 (0.022) 0.578 -0.026 (0.021) 0.231
Note: WP = within-person, BP = between-person. All models controlled for prior affect, time between current affective state and prior
affective state, and the chorological order of each prompt (i.e., time in study).
1
Physical activity refers to either total minutes spent in moderate-to-vigorous physical activity (MVPA) or light physical activity
(LPA) within a 15-minute or a 30-minute window prior to the assessment of affective states.
2
Log transformation was applied.
a
Indicates the model additionally controlled for time of day.
b
Indicates the model additionally controlled for annual household
income.

46

Figure 1. EMA items assessing current affective state.

47
15 minutes 15 minutes

Figure 2. Illustration of 15-minute time windows summarizing total minutes spent in moderate-
to-vigorous physical activity (MVPA) before and after each random EMA prompt in one day.

48
CHAPTER 3: EXAMINING WHETHER AFFECTIVE RESPONSES DURING
PHYSICAL ACTIVITY PREDICT CURRENT AND FUTURE DAILY PHYSICAL
ACTIVITY LEVELS AMONG ADULTS
Abstract
Purpose: Although there are theoretical foundations for a link between affective responses
during physical activity and future physical activity behavior, there is currently no study to test
this relationship in free-living settings. This study used ecological momentary assessment
(EMA) to capture affective responses during free-living physical activity, and used these
affective responses to predict moderate-to-vigorous physical activity (MVPA) minutes at 6
months and 12 months.
Methods: Electronic EMA surveys were randomly prompted across 4 days asking about current
activities (e.g., physical or sedentary activity) and affective states (i.e., positive affect, negative
affect, energy, and fatigue). Affective states reported during physical activity EMA prompts
were considered as affective responses. Data was available from 82 adults. These participants
also wore accelerometers for 7 consecutive days during the EMA monitoring period (i.e.,
baseline), as well as at 6 months and 12 months. Perceived affective benefits from physical
activity was assessed through a paper-pencil survey at baseline, and was tested as a mediator of
the relationship between affective responses during physical activity and future physical activity
levels.
Results: Affective responses during physical activity were not related to current MVPA
minutes. However, feeling more energetic during physical activity was associated with more
MVPA minutes at both 6 months and 12 months. Feeling less negative during physical activity
was associated with more MVPA minutes at 12 months only. Perceived affective benefits from
physical activity was not a significant mediator the relationship between affective responses
during physical activity and MVPA minutes at 6 and 12 months.

49
Conclusions: This study demonstrated the use of EMA to capture affective responses during
free-living physical activity. Results suggest that feelings of more energetic and less negative
were associated with more future physical activity. Although the perceived affective benefits
was not a significant mediator of this relationship, future studies could explore other potential
mediators (e.g., activity intensity, cognitive states) and moderators (e.g., gender, weight status,
social and physical context).

50
Introduction
Predictors of Physical Activity
Given the significant potential benefits of regular physical activity (e.g., as a protective
factor for heart disease, type 2 diabetes, obesity, and clinical depression) for both the general
populations and at-risk individuals, physical activity promotion has become a priority for public
health agencies throughout the world (Heath et al., 2012). To aid in the design of effective
physical activity interventions, a comprehensive understanding of the correlates and
determinants of physical activity is critical. Research on this topic has mostly focused on
individual-level factors (e.g., demographic factors such as age and gender, cognitive factors such
as self-efficacy and motivation), social factors (e.g., social support, cultural norms), and
environmental factors (e.g., social environment such as traffic and organizational practices, built
environment such as community design and public transport (Bauman et al., 2012; Trost, Owen,
Bauman, Sallis, & Brown, 2002). Although substantial efforts have been invested in
correlational studies, only a small number of the factors have been identified as consistent
predictors of physical activity (Bauman et al., 2012).
Affective Responses to Physical Activity
In contrast to the cognitive, social, and environmental factors, fewer studies have focused
on affective processes as determinants of adoption and maintenance of physical activity.
Hedonic principle provides a framework for understanding how affective responses to physical
activity could relate to physical activity adherence. According to the hedonic principle, people
seek to enhance or prolong pleasure and avoid or minimize pain (Freud, 1955). Kahneman and
colleagues have shown that affective responses to a behavior could influence decisions about
whether to repeat that behavior in the future (Kahneman, Fredrickson, Schreiber, & Redelmeier,
1993). Based on the hedonic principle, Ekkekakis and colleagues postulated that the unpleasant

51
affective responses from physical activity might reduce the intrinsic motivation and thus lead to
decreased adherence to regular physical activity (Ekkekakis, Hall, & Petruzzello, 2005).
Change in Affect after Physical Activity Predicting Future Physical Activity Behavior
Although there are theoretical foundations for the relationship between acute affective
responses to physical activity and future physical activity, only a few studies have examined this
association empirically. In an intervention study conducted by Klonoff and colleagues,
participants (23 women) were allowed to attend as many exercise sessions as they chose. Affect
was measured before and after a baseline exercise session. However, this difference score
between the before and after assessments in affect was not associated with exercise adherence
(i.e., number of exercise session attended) during the subsequent 10 weeks (Klonoff, Annechild,
& Landrine, 1994). Annesi conducted a series of studies using new members of community
fitness centers (2002b, 2002c, 2006). Participants’ affect was measured before and after each
biweekly exercise session for 15 weeks. Differences in pre-post affect score were aggregated
over time and used to predict attendance at the fitness centers during the 15 weeks. In two
studies, the difference scores in positive feelings and negative feelings were not associated with
exercise adherence (Annesi, 2002b, 2002c). In one study, the aggregated change in feelings (i.e.,
combined score of the positive and negative feelings, which showed a strong negative
correlation) during cardiovascular exercise (i.e., tredmills, rowing machines, stationary bicycles,
and elliptical training machines) was positively associated with exercise adherence (Annesi,
2006).
While these studies represent some of the early attempts to examine the relationship
between affective response to physical activity and/or physical activity maintenance, they
measured change in affect before and after the exercise session, rather than during the exercise.

52
Previous studies have shown that affective states tend to rebound immediately post exercise,
especially for high intensity activities (e.g., Ekkekakis, Hall, & Petruzzello, 2008; Sheppard &
Parfitt, 2008). Therefore, pre-post comparisons usually show a positive improvement in affect
(e.g., Yeung, 1996). However, significant inter-individual variability in acute affective state
during exercise has been found. For example, participants were likely to feel better (e.g.,
increased positive affect, decreased negative affect) during low-to-moderate physical activity but
felt worse during high-intensity activity (e.g., Backhouse, Ekkekakis, Biddle, Foskett, &
Williams, 2007; Ekkekakis et al., 2000). Thus, some believe affective state during physical
activity rather than the pre-post affective change that predicts physical activity adherence in the
future (Ekkekakis, Hall, & Petruzzello, 2004).
Affective Responses During Physical Activity Predicting Future Physical Activity Behavior
Several recent studies have found that physical activity induced affective states (i.e.,
affective response during physical activity) predict higher levels of physical activity
participation. In a sample of 37 previously sedentary adults, greater positive affective state
(measured by a single-item 11-point bipolar measure of valence, i.e., the Feeling Scale)
measured during a moderate-intensity physical activity bout (i.e., treadmill task until the
participant reached 85% of his/her age predicted maximum heart rate) was associated with
greater physical activity both 6 and 12 months later (measured by self-reported physical activity
recall during the past 7 days; Williams et al., 2008). In another study with 192 healthy
adolescents, participants were asked to perform a 30-minute moderate-intensity exercise bout on
cycle ergometer in the lab. Affective state (also measured by the Feeling Scale) was assessed
before the exercise, 10 minutes into the exercise, and 20 minutes into the exercise. An average
score (of ratings at 10 and 20 minutes) was created to indicate affective state during physical

53
activity. Results showed that adolescents who had improved affect during physical activity (as
compared to the affective state before exercise) engaged in more daily physical activity
(measured by accelerometer for 7 days) than adolescents who experienced no change, and those
who experienced a decline in affective valence (Schneider, Dunn, & Cooper, 2009).
Perceived Affective Benefits as a Potential Mediator of the Relationship Between Affective
Response and Future Physical Activity
Perceived affective benefits from physical activity refer to what people think they would
feel if they engaged in physical activity, which might also influence people’s decision making
about their future physical activity engagement. Perceived affective benefits may describe past
(e.g., if I engaged in physical activity yesterday, I would have felt …), current (e.g., if I engage in
physical activity now, I would feel …), or future (e.g., if I engage in physical activity next week, I
will feel …) behaviors. Researchers usually use the term anticipated affective response to define
people’s perceived affect about their future behaviors. Several psychological models have
sought to explain how anticipated affective response may influence behavior. For instance,
response expectancy theory and expected pleasure theory state that the expected affective
responses to a behavior will determine whether the behavior will be repeated (Kirsh, 1997;
Mellers, 2000). There is empirical evidence supporting perceived affect predicting future
physical activity levels. For example, Dunton and Vaughan (2008) found that anticipated
positive affect at baseline predicted future physical activity adoption among initially inactive
adults after 90 days.
In addition to predicting future behaviors, perceived affective responses may also directly
influence actual affective responses. According to response expectancy theory, the anticipated
affective response to a behavior may influence one’s actual affective response when performing

54
the behavior. In expected pleasure theory, the affective response to a behavior partly depends on
the counterfactuals (i.e., one’s perceived probability of the subjective value of the outcomes not
being received). Therefore, it is possible that perceived affective benefits to physical activity
might act as a mediator of the relationship between affective response to physical activity and
future physical activity engagement. Indeed, Williams (2008) proposed an integrative model of
exercise intensity, affective response, and exercise adherence. According to this model, acute
affective response to exercise, which could be influenced by cognitive (e.g., perceived
autonomy) and interoceptive (e.g., exercise intensity) factors, is associated with exercise
adherence via anticipated affective response to future exercise.
Research Gaps
Although the two studies mentioned above offer empirical supports to the link between
affective response during physical activity and overall physical activity participation, each study
bears some major methodological limitations. Both studies used a single-item one-dimensional
measure to capture affective states. Even though this measure (i.e., the Feeling Scale) has been
widely used in physical activity studies (Ekkekakis, 2003), and has been shown to be related to
enjoyment of acute exercise (Robbins, Pis, Pender, & Kazanis, 2004) and other measures of
affective valence (Hall, Ekkekakis, & Petruzzello, 2002), it has at least two limitations. First, the
Feeling Scale only measures valence, yet other aspects of affect are also closely related to
physical activity (e.g., revitalization and fatigue as in Annesi’s studies). Second, people’s
affective state can be multidimensional, and therefore this measure might not be able to fully
capture the multifaceted affect that people might experience during physical activity (e.g., people
can feel both tired and excited at the same time). Therefore, these two studies only provide
evidence for one aspect of affect (i.e., valence) rather than a more comprehensive picture of

55
affective states in relation to physical activity participation. Further, Williams’ study (2008)
used a self-reported measure of physical activity, which is prone to recall bias, and although
Schneider and colleagues (2009) used objective measure of physical activity, her study did not
examine the relationship between affect and physical activity longitudinally. Lastly, both studies
measured physical activity induced affective state in a laboratory environment, and used affect
measured in this way to predict physical activity participation in naturalistic settings (i.e.,
people’s everyday life). Nevertheless, the affective states captured in a controlled lab setting
could be very different from the ones people experience in their daily lives. Therefore, one of
the major unanswered questions is will results from these lab-based studies be generalizable to
naturalistic settings?
In summary, there is a need to better understand the relationship between affective
responses during physical activity in naturalistic settings and future physical activity
engagement. This knowledge could be helpful to answer the question of why individuals may or
may not engage in physical activity in their daily lives. An advantage of the EMA method is that
it can collect people’s feelings in naturalistic settings when people are engaging in their daily
physical activity. Although some studies have examined affective response to physical activity
via EMA (e.g., Kanning, 2012; Mata et al, 2012; Schwerdtfeger et al., 2010; as discussed in
Chapter 2), to date, no EMA study has examined the link between affective responses during
physical activity and future physical activity level. The current study aimed to address this
important research gap by analyzing a longitudinal dataset with affective states measured via
EMA and objective daily physical activity measured by accelerometer.

56
Research Questions and Hypotheses
This study aimed to answer the following research questions:
1. What is the strength and direction of the association between average affective
responses during physical activity and current overall physical activity level (i.e.,
daily MVPA minutes)?
Based on the hedonic principle and findings from previous research, it was hypothesized
that having a more positive affective response to physical activity would be positively associated
with one’s current overall physical activity level, and having a more negative affective response
to physical activity would be negatively associated with one’s current overall physical activity
level.
2. What is the strength and direction of the association between average affective
responses during physical activity and future overall physical activity level (i.e.,
daily MVPA minutes after 6 months, and 12 months)?
Based on the hedonic principle and findings from previous research, it was hypothesized
that having a more positive affective response to physical activity would be positively associated
with one’s future overall physical activity level, and having a more negative affective response to
physical activity would be negatively associated with one’s future overall physical activity level.
3. Do perceived affective benefits from physical activity mediate the relationships
between affective responses during physical activity and future overall physical
activity levels?
Based on the response expectancy theory and expected pleasure theory, it was
hypothesized that having a more positive affective response to physical activity would be
positively associated with one’s perceived affective benefits from physical activity, which would

57
lead to higher overall physical activity level in the future. Having a more negative affective
response to physical activity would be negatively associated with one’s perceived affective
benefits from physical activity, which would lead to lower overall physical activity level in the
future.
Methods
Data Source
This study used the longitudinal data from Project MOBILE, which included 3 waves of
data with each about 6 months apart. Details about the participants and study protocol have been
described elsewhere (i.e., Chapter 2). At each wave, participants were given a mobile phone that
randomly delivered EMA survey questions up to 8 times a day for 4 consecutive days.
Participants also wore an accelerometer on their hip during waking hours for 7 consecutive days,
which encompassed the 4 EMA monitoring days. When participants received their monitoring
devices at each wave, they also filled out a paper-pencil survey that assessed the demographic
variables, and other psychosocial measures related to physical activity.
Measures
Affective responses during physical activity. To capture affective response during
physical activity, only EMA responses indicating physical activity as the current main activity
was used. Each EMA survey sequence asked about participants’ current activity: “What were
you DOING right before the beep went off? (Choose your main activity)”. The response choices
were “Reading/Computer”, “Watching TV/Movies”, “Eating/Drinking”, “Physical
Activity/Exercise”, and “Other”. In a follow-up question, participants were also asked: “Were
you?” The response choices were “Sitting”, “Standing”, “Walking”, and “Jogging/Running”.
Taking these two questions together, EMA responses with the choice of “Physical
Activity/Exercise” or “Jogging/Running” indicated the participants were engaging in physical

58
activity at the moment when answering the EMA surveys. Therefore, the affective states
recorded for those EMA responses represent affective states during physical activity. Positive
affect, negative affect, energy, and fatigue were examined separately. For participants who had
multiple EMA responses for affective states across the four monitoring days, an average score
was used to indicate the average affective response during physical activity.
Perceived affective benefits from physical activity. Two items from the perceived
benefits of regular physical activity on the paper-pencil survey were used to create a composite
score to represent perceived affective benefits from physical activity. The two questions were:
“If I participate in regular physical activity or sports, then I will feel less depressed and/or bored”
and “If I participate in regular physical activity or sports, then I will feel less tension and stress”
(Cronbach’s α = .654). The response choices were on a 5-point Likert scale ranging from
“strongly disagree” to “strongly agree.” The first question was reverse-coded so that a higher
score indicates a more positive affective response to physical activity.
Physical activity level. Data (i.e., activity counts) from the accelerometer (Actigraph
Model GT1M) was downloaded and processed through MeterPlus (version 4.3). Consecutive
records of zero activity counts for 60 minutes were considered as device non-wear, and marked
as an invalid hour. A minimum of 10 valid hours a day was considered as a valid day. Each
participant needs to have at least 4 valid days out of the 7 monitoring days in order to be
considered as having valid accelerometer data. The MVPA cut-point for adults is 2,020 activity
counts per minute (equivalent to three METs; Troiano et al., 2008). Average daily MVPA
minutes across valid days at each wave were used as the outcome variables for this study.

59
Statistical Analysis
Research question 1. To test the strength and direction of the association between
average affective responses during physical activity and current overall physical activity level,
linear regression modeling was used. Average daily MVPA minutes at wave 1 was the outcome,
and average affective states during EMA reported physical activity at wave 1 was the predictor.
Since the outcome variable was not normally distributed, a log transformation was performed. A
set of covariates (i.e., gender, age, ethnicity, weight category, and average affective states during
non-physical activity episodes) were added to the regression model separately, one at a time.
Significant covariates were kept in the final adjusted model. All linear regression models were
fitted using SAS PROC REG.
To additionally explore the effects of multidimensionality of affective responses during
physical activity, average positive affect and average negative affect were entered in one model,
and average energy and average fatigue were entered in one model to see any effects found from
the previous step still exist when controlling for the opposite dimension of the affective state.
Research question 2. To test whether the average affective responses during physical
activity were associated with average daily MVPA minutes after 6 months, change in average
daily MVPA minutes (i.e., average daily MVPA minutes at wave 2 minus average daily MVPA
minutes at wave 1) was used as the outcome, and average affective state during EMA reported
physical activity at wave 1 was used as the predictor. The affective states (i.e., average positive
affect, average negative affect, average energy, and average fatigue) were tested in separate
models, one at a time. Significant covariates were kept in the final adjusted model.
The same procedures described above were repeated using average daily MVPA minutes
at wave 3 to create the change score. Thus, the linear regression models tested whether the

60
average affective responses during physical activity are associated with average daily MVPA
minutes after 12 months.
Research question 3. To test whether perceived affective benefits from physical activity
mediate the relationship between affective responses during physical activity and future overall
physical activity level, two sets of meditation models were fitted: one used average daily MVPA
minutes at wave 2 as the outcome, the other used average daily MVPA minutes at wave 3 as the
outcome. For each set of the meditation models, four predictors (i.e., average positive affect,
average negative affect, average energy, and average fatigue at wave 1) were tested separately.
All mediation models were fitted using Mplus.
The following regression equations demonstrate the mediation model using average daily
MVPA minutes at wave 2 as the outcome (denoted as 𝑀 𝑉𝑃𝐴 𝑤 2
), average positive affect during
physical activity at wave 1 as the predictor (denoted as 𝐴 𝐹𝐹𝐸𝐶 𝑇 𝑤 1
), and perceived affective
benefits from physical activity at wave 1 as the mediator (denoted as 𝑃𝐸𝑅𝐶
𝑤 1
):
𝑀 𝑉𝑃𝐴 𝑤 2
= 𝑖 1
+ 𝑐′ 𝐴 𝐹𝐹𝐸𝐶 𝑇 𝑤 1
+ 𝑏 𝑃𝐸𝑅𝐶
𝑤 1
+ 𝑒 1

𝑃𝐸𝑅𝐶
𝑤 1
= 𝑖 2
+ 𝑎𝐴 𝐹𝐹𝐸𝐶𝑇 + 𝑒 2

where c’ is the effect of average positive affect during physical activity at wave 1 on average
daily MVPA minutes at wave 2 controlling for perceived affective benefits from physical
activity at wave 1, b is the effect of the perceived affective benefits from physical activity at
wave 1 on average daily MVPA minutes at wave 2, and a is the effect of average positive affect
during physical activity at wave 1 on perceived affective benefits from physical activity at wave
1 (see Figure 3 for diagram of this model). The indirect effect of positive affect to MVPA at
wave 2 is equal to the product of a and b (i.e., the Sobel method; Muthén, 2011).

61
Further, moderated-mediation was explored for gender and weight category at wave 1.
The moderated-mediation refers to investigating whether a mediated relation holds across the
moderating variable (e.g., gender and weight category). However, models that simultaneously
examine mediation and moderation effects often have very low power because of the multiple
interaction terms and estimation of indirect effects (Fairchild & MacKinnon, 2009). Due to the
small sample size, the moderated-mediation model would only be a secondary analysis for
research question 3.
Power Analysis
Power analysis was conducted using G*Power (version 3.1). Linear bivariate regressions
were applied with α level at .05 and power at .80. The person-level standard deviations ranged
from .41 to .66 for affective states. The standard deviation for average daily MVPA was 22.57 at
wave 1, 12.12 at wave 2, and 15.91 at wave 3. Therefore, having a sample size of 65 (i.e., the
most conservative estimation) would give estimated slopes range from 11.38 to 18.32 for
affective states predicting physical activity at wave 1 (research question 1); from 6.11 to 9.84 for
affective states predicting physical activity at wave 2 (research question 2); and from 8.02 to
12.91 for affective states predicting physical activity at wave 3 (research question 2).
Figure 3 shows the power plot for a range of slopes for affective states predicting
physical activity (assuming σ
x
= .41 and σ
y
= 12.12). Have a sample of 50 would provide .80
power to detect a slope of 11. However, we have limited power to detect smaller slopes (e.g.,
only .6 power to detect a slope of 8 for a sample of 65).
Results
Data Availability
There were a total of 117 participants recruited for Project MOBILE at wave 1. Of these
participants, 3 did not have EMA data due to an EMA data download problem. Of the 114

62
participants with EMA data at wave 1, 85 reported engaging in physical activity when answering
the EMA surveys at least once across the 4 monitoring days. Among these 85 participants, 69
answered positive affect and energy questions during physical activity prompts, and 66 answered
negative affect and fatigue questions during physical activity prompts. For these participants,
less than 20 had more than two EMA entries for affective state questions during physical
activity. Specifically, participants on average had 1.83 (SD = 1.12) EMA entries for positive
affect during physical activity, 1.85 (SD = 1.13) EMA entries for negative affect during physical
activity, 1.83 (SD =1.12) EMA entries for energy, and 1.91 (SD = 1.21) entries for fatigue during
physical activity. The small number of EMA entries per participant could be due to (1) they did
not have multiple EMA entries for physical activity, or (2) questions about affective state were
not assessed when they reported to be physically active, since the affective state questions only
appeared 60% of the time according to the EMA protocol. Three of the 85 participants did not
have any affect assessment for their physically active EMA prompts. Therefore, the analytical
sample size was 82. The excluded participants did not different from the analytical sample in
age, gender, race/ethnicity, or weight category.
Of the 82 participants eligible for data analysis based on their EMA responses, 10 did not
return for wave 2 and wave 3 assessments. An additional 3 participants missed their wave 2
assessment only, and 5 missed their wave 3 assessment only. In summary, the retention rate at
wave 2 was 84%, and at wave 3 was 82%.
Of the 82 eligible participants at wave 1, 79 (96%) had valid (i.e., had at least 4 valid
days of) accelerometer data; of the 69 participants that came back at wave 2, 65 (94%) had valid
accelerometer data; of the 67 participants that came back at wave 3, 62 (93%) had valid

63
accelerometer data. There were no significant differences in age, gender, race/ethnicity, and
weight category between participants with valid vs. non-valid accelerometer data.
Affective Responses During Physical Activity and Current Physical Activity Level
On average, positive affect during physical activity at wave 1 was 3.39 (SD = 0.63) on a
5-point scale. The average negative affect during physical activity was 1.51 (SD = 0.72),
average energy was 3.37 (SD = 1.09), and average fatigue was 1.63 (SD = 0.81). The average
daily MVPA minutes at wave 1 was 26.94 (SD = 22.57). Results from linear regression analysis
showed that none of the affective response during physical activity was associated with the
average daily MVPA minutes at wave 1.
Affective Responses During Physical Activity and Future Physical Activity Level
The average daily MVPA minutes at wave 2 was 19.78 (SD = 12.20), and at wave 3 was
22.54 (SD = 15.91). Table 5 shows the results from the linear regression models using change in
MVPA minutes at wave 2 and wave 3 as the outcome and affective response during physical
activity at wave 1 as the predictor. Change in MVPA minutes between wave 1 and wave 2 was
positively associated energy during physical activity at wave 1(beta = 5.05, SE = 1.81, p = .01),
suggesting that every 1 point increase in feeling of energetic during physical activity will lead to
an increase of 5 minutes of MVPA per day at 6-month later. This positive association was also
found for change in MVPA minutes between wave 1 and wave 3 (beta = 5.44, SE = 2.43, p =
.03). Further exploratory analysis show that energy remains as a significant predictor of change
in MVPA minutes between wave 1 and wave 2 even after controlling for fatigue (beta = 4.63, SE
= 2.15, p = .04). However, after controlling for fatigue, energy became marginally significant in
predicting change in MVPA minutes between wave 1 and wave 3 (beta = 4.99, SE = 2.80, p =
.08).

64
Negative affect during physical activity at wave 1 was negatively associated with change
in MVPA minutes between wave 1 and wave 3 (beta = -14.20, SE = 6.13, p = .02), suggesting
that every 1 point increase in feeling of negative during physical activity will lead to an decrease
of 14 minutes in MVPA at 12-month later. Further exploratory analysis showed that this
negative association became non-significant after controlling for positive affect. Positive affect
and fatigue during physical activity were not associated with future physical activity level.
Perceived Affective Benefits as a Mediator between Affective Responses During Physical
Activity and Future Physical Activity Level
The average perceived affective benefits from physical activity was 3.10 (SD = 0.53) on a
5-point scale. Mediation analysis showed a marginally positive effect of positive affect during
physical activity at wave 1 on perceived affective benefits (beta = .15, SE = .09, p = .08),
suggesting that people who experienced more positive affect during physical activity perceived
more affective benefits from physical activity. All other paths in the mediation models were
non-significant for physical activity level at both wave 2 and wave 3. Therefore, further
moderated-mediation analysis was not performed.
Discussions
The current study used EMA to capture affective state during physical activity among
adults’ daily lives, and examined its relationship with current and future physical activity level.
This EMA-assessed affective response to physical activity could partly address previous
research’s (i.e., assessed affective responses during physical activity in a lab setting) limitations
on external and ecological validity. Further, physical activity was measured via accelerometer in
this study, which provides a more reliable source compared to self-reported physical activity as
used in previous studies (e.g., Williams et al., 2008). Results from this study did not find a

65
significant relationship between affective responses during physical activity and current daily
physical activity level. However, feeling more energetic during physical activity predicted more
future daily physical activity level, even after controlling for the opposite affective state (i.e.,
fatigue). The perceived affective benefits was not a significant mediator of the relationship
between affective response and future physical activity as hypothesized.
The current study found that affective responses during physical activity were not related
to current physical activity level. One possible explanation for this finding could be that the
assessment periods for EMA and physical activity did not map onto each other perfectly. The
current physical activity level was operationalized as the average of daily MVPA minutes across
7 days. However, affective responses during physical activity were randomly assessed across 4
days of the 7-day monitoring period. Further, most participants only had 1-2 affective responses
reported, which possibly came from 1-2 days across the 7-day period. Therefore, it is possible
that for some participants, the affective responses during physical activity were captured on day
1 of the study only, which implies that for these participants, we actually examined the effect of
affective responses during physical activity predicting the immediate future (i.e., the next 7 days)
physical activity levels.
Contrary to previous lab-based studies that found a positive relationship between positive
affective state and future physical activity (e.g., Williams et al., 2008, Schneider, Dunn, &
Cooper, 2009), this study did not find any association between positive affect during physical
activity and change in daily MVPA minutes in 6 and 12 months later. This finding potentially
suggests that the affective responses during physical activity might be different between lab-
based exercise (e.g., treadmill task, cycle ergometer) and free-living exercise (e.g., walking in the
neighborhood). Thus, it will be crucial for future study to keep in mind that findings based on

66
affective responses during lab-based exercise might not be applied to real-life scenarios. Further,
our results indicated a significant negative relationship between negative affect during physical
activity and change in daily MVPA minutes at 12 months. This finding suggests that negative
affect and positive affect might represent two different feeling states that people could
experience during physical activity, and may have distinct effects on future physical activity
level. Therefore, studies that use a single item to capture affective response during physical
activity (e.g., the Feeling Scale) might fail to distinguish these separate effects.
This study also found that individuals who reported feeling more energetic during
physical activity engaged in more MVPA in both 6 and 12 months later than individuals who
reported feeling less energetic during physical activity. This finding is consistent with results
from previous cross-sectional studies that found a positive relationship between feeling of energy
and physical activity (e.g., Puetz, 2006; Yoon, Buckworth, Focht, & Ko, 2013). Further, results
from this study also suggest that this positive relationship holds significant after controlling for
the opposite affective state (i.e., fatigue). This implies that even if a person feels tired when
doing an exercise (e.g., intense running), as long as s/he also feels energetic during that exercise,
then s/he will likely to be more active in the future.
Lastly, this study did not find perceived affective benefits mediate the relationships
between affective responses during physical activity and future physical activity levels. This is
possibly due to that perceived affective benefits was only marginally associated with positive
affect during physical activity, and positive affect was not related to future physical activity
levels. This result fails to support the potential pathway from affective response to exercise
adherence that was proposed by Williams (2008), which represents an integration of the dual-
mode model and hedonic theory. Nevertheless, the current study did not examine the effect of

67
exercise intensity (i.e., the part of Willians’ proposed pathway that reflects the dual-mode model),
which might influence the affective responses to exercise. Since this study found that negative
affect and energy were associated with future physical activity, future studies could explore other
variables that might be more relevant to these two affective responses. For example, self-
efficacy was found to be partially mediated the relationship between feeling of energy and
exercise (Yoon et al., 2013).
Limitations
Given the small sample size of this study, there was limited power to detect the proposed
relationships between affective responses during physical activity and future physical activity
levels. For example, the current sample size only achieved 11.4% power when testing the
relationship between positive affect during physical activity wave 1 and change in MVPA
minutes between wave 1 and wave 2. In addition, since affective states were not assessed during
each EMA sequence, more than half of the participants only had one physical activity prompt
with reported affective states. Therefore, the affective response during physical activity we
captured might not be representative of people’s general affective response during physical
activity. Further, this study did not measure other affective feelings that might be relevant to
physical activity, such as enjoyment, boredom, and discomfort. Moreover, activity intensity
would be an important confounder for the affective response during physical activity. However,
the EMA self-reported physical activity did not assess activity intensity. Lastly, characteristics
of this study’s participants (e.g., adults engaged in <150 minutes/week physical activity, mostly
females, ability to read English) might make the results have limited generalizability to other
populations.

68
Implications
The current study is among the first attempts to examine whether affective responses
during physical activity in people’s daily lives would predict future physical activity levels. It
demonstrated that the use of electronic EMA via mobile phones could be a practical method to
capture affective responses during free-living exercise among adults. Results from this study
show that although affective responses during physical activity were not related to current
physical activity levels, feeling less negative and more energetic were associated with an
increase in physical activity level up to 12 months later. Since participants of this study were
inactive adults (i.e., not meeting the recommendation of engaging in at least 30 minutes MVPA
on average each day), findings from this study could potentially offer some insights about how to
encourage more physical activity for this particular population. For example, interventions could
try to identify factors that are associated with negative affective responses to physical activity
(e.g., the activity is too intense, the environment is not comfortable), and then offer strategies to
overcome these factors (e.g., break activity into shorter bouts, try to exercise at a more pleasant
location such as outdoor at a park).
Conclusions
Favorable affective responses (i.e., feeling of higher energy and less negative) during
free-living physical activity were positively associated with future physical activity levels.
Nevertheless, perceived affective benefits did not mediate the relationship between affective
responses and future physical activity. Future studies could examine other dimensions of
affective responses that might relate to physical activity, and explore alternative mechanisms that
may explain the relationship between affective responses during physical activity and long-term
physical activity maintenance.

69
Table 5. Associations between affective response during physical activity and future physical
activity level
Change in MVPA Minutes
between Wave 1 and Wave 2
Change in MVPA Minutes
between Wave 1 and Wave 3
p Beta (SE) p Beta (SE)
Positive Affect 2.52 (2.50) 0.32 0.81 (3.29)
a
0.81
Negative Affect -4.38 (4.36) 0.32 -14.20 (6.13)
b
0.02
Energy 5.05 (1.81) 0.01 5.44 (2.43) 0.03
Fatigue -0.49 (3.34) 0.88 1.25 (3.75) 0.74
Note:
a
Indicates the model controlled for ethnicity.
b
Indicates the model controlled for affect
during non-active prompts.

70

Figure 3. Power plot for affective states predicting physical activity.

71
CHAPTER 4: EXPLORING THE DYADIC RELATIONSHIPS BETWEEN PHYSICAL
ACTIVITY AND AFFECTIVE STATES IN MOTHER-CHILD PAIRS
Abstract
Purpose: Parents could have significant influences on children’s daily physical activity.
However, although previous studies have examined the effect of parental stress on children’s
health-related behavior, there is no study investigating whether parents’ affective states might
influence children’s subsequent affective states and physical activity levels.
Methods: This study used ecological momentary assessment (EMA) via a smartphone app to
assess mothers and their children’s affective states across a 7-day monitoring period. Children
also wore accelerometers during this period to objectively measure their light physical activity
(LPA) and moderate-to-vigorous physical activity (MVPA). Multilevel mediation model was
used to test whether mothers’ affective states were associated with children’s subsequent
physical activity levels, mediated by children’s affective states.
Results: There was no significant total effect, direct effect, or indirect effect of mothers’
affective states on children’s subsequent physical activity levels. However, results revealed a
positive association between mothers and children’s positive affect at both within-person and
between-person level; and a positive association between mothers and children’s negative affect
at within-person level. Further, children who had higher positive affect than the average child in
this study engaged in more MVPA minutes; and higher negative affect than a child’s usual level
was associated with more concurrent MVPA minutes. Children’s affective states were not
associated with their concurrent LPA minutes.
Conclusion: Mothers’ affective states could affect their children’s subsequent affective states,
although not their children’s physical activity levels. This study demonstrated the use of EMA to
collect real-time self-report data on a dyadic level. Future studies could use this methodology to

72
explore whether parental affective feelings might influence children’s feeling states and
behaviors across different social and physical context in free-living settings.

73
Introduction
Physical Inactivity in Parents and Children
According to the recently released national report card on physical activity for children
and youth, the U.S. received a grade of D- in the category of overall physical activity, indicating
that only one quarter of U.S. children and youth meet the physical activity recommendation of at
least 60 minutes of moderate-to-vigorous physical activity (MVPA) per day (National Physical
Activity Plan Alliance, 2014). Meanwhile, less than 5% of U.S. adults meet the physical activity
recommendation of at least 30 minutes of MVPA per day based on objective measurement
(Troiano et al., 2008). Therefore, one can easily assume that in most American households, both
parents and children are not physically active enough in their daily lives. In fact, a study
conducted by Dunton and colleagues showed that, of the 4 hours that parents and children spent
together each day (during non-school waking hours and not counting commute time), about 1.5
hours were spent engaging in sedentary behavior together, and only about 2.5 minutes were spent
performing MVPA together (Dunton et al., 2012).
Parental Influences on Children’s Physical Activity
Empirical evidence shows that parents’ physical activity level is highly correlated with
their children’s physical activity level. Self-report data shows that children who had physically
active parents were more likely to engage in structured physical activity as compared with
children who had one or both inactive parents (e.g., Eriksson, Nordqvist, & Rasmussen, 2008;
Wagner et al., 2004). Objectively measured data (i.e., physical activity measured by
accelerometers) also supports that greater parental MVPA was associated with increased child
MVPA (e.g., Fuemmeler, BAnderson, & Mâsse, 2011; Oliver, Schofield, & Schluter, 2010).
Aside from genetic influence (e.g., Bouchard, & Malina, 1983), it is generally believed that

74
parents may exert influences on their children’s physical activity through social learning.
According to the Social Learning Theory, children learn their habits and attitudes toward
physical activity by observing and imitating their parents (Sallis & Nader, 1998). From the
theoretical perspective, parents can influence their children’s physical activity behavior through
(1) role modeling, which includes parents showing interests in physical activity and parents’
active engagement in physical activity, and (2) parental support, which includes parental
encouragement, involvement (i.e., participating in physical activity with children together), and
facilitation (i.e., providing opportunities for children to be active; see Welk, Wood, & Morss,
2003). Therefore, in an effort to promote physical activity among children, a growing body of
research has investigated in how parental role modeling and parental support might influence
children’s activity level, and a majority of these studies demonstrated positive associations
between parental measures and children’s behavior (e.g., Beets, Cardinal, & Alderman, 2010;
Edwardson & Gorely, 2010).
Parental Stress as a Predictor of Children’s Behavior
Relatively few studies have investigated how parental psychological states (e.g., general
mood, daily affective states) might influence children’s physical activity level. One notable
exception is the studies of examining how maternal stress may predict children’s obesity risk.
According to the theory of household production, families could “produce” children’s health
outcome by allocation or lack of parental resources (Foster, 2002). Factors such as parental
stress might influence a family’s adjustment, and such stressors can be transmitted to children
(e.g., through diminished parenting, lack of time with children), which then could lead to higher
levels of stress among children (Lohman, Stewart, Gundersen, Garasky, & Eisenmann, 2009).
Higher stress and more negative emotions in children are associated with more sedentary

75
behaviors (Brodersen, Steptoe, Williamson, & Wardle, 2005). Previous studies also showed that
parent-perceived stress was positively related to child’s fast-food consumption (Parks et al.,
2012), and higher parenting stress was associated with lower levels of child’s health-related
behaviors including oral health, personal hygiene, safety practices, nutrition, and physical
activity (Park & Walton-Moss, 2012). However, most of these studies focused on stress, which
only represents one dimension of people’s negative emotional experience in their daily lives.
Other emotions such as sadness and hostility are some of the negative affect that people could
experience in their everyday lives, and each represents meaningful and differentiable
psychological constructs (Watson & Clark, 1992). Therefore, more research is needed to better
understand whether parental negative affect might affect children’s daily affective states and
behaviors such as physical activity.
Parent-child Mutual Affect
Parents and children are important relationship partners that can influence each other’s
emotions and behaviors (Reis, Collins, & Berscheid, 2000). To better describe these types of
emotional interactions between parents and children, Kochanska (1997) has proposed a construct
of mutually responsive orientation – a positive, close, mutually binding, and cooperative
relationship which encompasses responsiveness and shared positive affect. In this construct,
shared positive affect refers to the pleasurable interactions infused with positive emotions shared
by the parent and the child. This mutuality construct, while correlated with the positive affect
experienced by each individual, is believed to be distinct and represents aspects of co-regulated
affect and behavior at the level of the dyad (Deater-Deckard & O'Connor, 2000). Since the
affective states of the parent and the child are linked, parent affect and child affect should occur
in cycles of reciprocal causality in theory (e.g., one giving rise to the other).

76
Several studies investigated how this parent-child mutuality might influence children’s
social-emotional outcomes such as social competence, self-regulation, and prosocial behavior.
Kochanska’s studies of young children revealed that the shared positive affect within the dyad
was associated with children’s positive behavioral outcomes (e.g., willingness and eagerness to
accept rules and norms) several years later (Kochanska & Murray, 2000). Using a sample of
school-aged children (mean age = 8.16), Deater-Deckard and Petrill (2004) found that lower
levels of parent-child dyadic mutuality (e.g., co-occurring positive affect) are associated with
higher levels of child behavior problems. Therefore, parent-child mutual positive affect could be
a potential important factor to consider when examining the relations between parent-child
affective states and physical activity, especially given the evidence that positive affective states
are associated with higher physical activity level in children (e.g., Dunton et al., 2014).
Current Research Gaps
In summary, although there are some studies examining how parental stress might
influence children’s physical activity, no known studies have investigated whether parental
positive emotions may affect children’s physical activity. More importantly, previous research
has not examined these effects as they occurred in daily lives. Further, there could be within-
person variations in parents’ and children’s affective experiences and behaviors, which could
only be captured using intensive repeatedly measurements. These day-to-day, or moment-to-
moment variations in affective states experienced by parents and children might reflect some of
the important family characteristics (e.g., stressors experienced by the family, family resources,
parenting practice, family rules) that are potential correlates of children’s physical activity levels.
Therefore, investigating the relationships between parent-child affective states could potentially

77
lead to intervention strategies about how to increase physical activity levels for both parents and
children.
Research Question and Hypothesis
This study used ecological momentary assessment (EMA), a real-time self-report
strategy, to collect data from mother-child pairs and answer the following research question: Do
mothers’ affective states influence children’s immediate subsequent physical activity level,
mediated by children’s concurrent affective states?
It is hypothesized that mothers’ positive affective state would be positively associated
with children’s immediate subsequent positive affective state, which would lead to higher
concurrent physical activity level in children; and mother’s negative affective state would be
positively associated with children’s immediate subsequent negative affective state, which would
lead to lower concurrent physical activity level in children.
Methods
Participants
This study used data from mother-child pairs participating in the Mothers’ and Their
Children’s Health (MATCH) Study. Participants of the MATCH Study were mothers and their 8
to 12 years old children. Additional inclusion criteria for MATCH Study were: (1) mothers need
to have at least 50% of child’s custody, and (2) both mother and child need to be able to read in
English or Spanish. Individuals were excluded for participating in the study if (a) currently
taking medications for a psychological condition; (b) having health issues that limit physical
activity; (c) enrolled in special education programs; (d) currently using oral or inhalant
corticosterioids, and (e) the mother is currently pregnant. Participants were recruited from local

78
community centers (e.g., Boys and Girls Club), after-school programs, and public elementary
schools in Los Angeles County.
Study Protocols
EMA. Electronic EMA surveys were delivered through an Android (Google USA, Inc.)
based smartphone with a custom software program (i.e., the MATCH app) installed. Mothers
used their own Android phone if they had one, and the MATCH app was installed on their
phones. The MATCH app was a platform to randomly prompt the EMA survey and store the
survey responses. When the smartphone was connected to the Internet, the MATCH app would
transfer the stored survey responses wirelessly to a cloud server. The EMA monitoring period
was 7 days, and participants were asked to proceed with their daily routines as normal. Overall,
both mothers and children received 3 prompts from afternoon to evening during weekdays, and 7
prompts throughout the day during weekend days. EMA surveys were prompted at a random
time within a pre-set schedule to ensure adequate spacing across the day. In addition, each of
children’s EMA prompts was delivered within a 30-60 minutes window after each mother’s
EMA prompt to ensure proper time-lags between children’s and mothers’ EMA survey. If a
survey prompt was not answered (i.e., no response entry was made), the smartphone would
signal up to two reminders at 3 minutes intervals. Ten minutes after the initial prompt, the EMA
survey would become inaccessible until the next prompt. Each prompted EMA was time-
stamped.
Accelerometry. The Actigraph, Inc., GT3M+ model accelerometer was used as an
activity monitor to objectively measure mothers’ and children’s physical activity throughout the
day. Participants wore the device on their right hip during their waking hours across the 7-day
EMA monitoring period. Accelerometer continuously recorded participants’ activity intensity

79
(as expressed in activity counts) in 30-second epochs. Each accelerometer recording was time-
stamped.
Measures
Affective states. During each EMA survey sequence, mothers and children answered 5
questions assessing their current affective states. Children were asked about how happy, joyful,
mad, sad, and stressed they were feeling right before the phone went off. Mothers were asked
about how happy, calm/relaxed, frustrated/angry, sad/depressed, and stressed they were feeling
right before the phone went off. Response choices for these items were “0=not at all, 1=a little,
2=quite a bit, 3=extremely”. For children, items of happy and joyful were combined together to
create a composite positive affect score (Cronbach’s α = 0.92); and mad, sad, and stressed were
combined together to create a composite negative affect score (Cronbach’s α = 0.73). For
mothers, items of happy and calm/relaxed were combined together to create a composite positive
affect score (Cronbach’s α = 0.74); and frustrated/angry, sad/depressed, and stressed were
combined together to create a composite negative affect score (Cronbach’s α = 0.71). In
addition, feelings of happy and stressed were examined separately since these two items were
worded the same way for both mothers and children.
Social context. During each EMA survey sequence, mothers were asked whether they
were with their child just before the phone went off. The response choices were either “Yes” or
“No”.
Physical activity for children. Physical activity levels were summarized as current light
physical activity (LPA) and moderate-to-vigorous physical activity (MVPA) minutes, defined as
total minutes in LPA and MVPA within the 30-minute window around (i.e., ± 15 minutes of)
each child’s answered EMA survey. An age-specific cut-point was used for defining LPA and

80
MVPA, which are generated from the Freedson prediction equation (Freedson, Pober, & Janz,
2005). A total of zero activity counts within the 30-minute window was considered as
accelerometer non-wear.
Statistical Analyses
To test whether mothers’ affective states influence children’s immediate subsequent
physical activity level, mediated by children’s affective states, mediation analysis was used. For
this analysis, only prompts when mothers reported they were with their children were included.
Based on the EMA protocol, children’s surveys were prompted 30-60 minutes after mothers’
surveys for each EMA time window. Therefore, the temporal sequence of mothers’ and
children’s affect was clear and this enabled us to test the relationship between mothers’ current
affect and children’s subsequent affect and activity level.
For the mediation model, mothers’ prompt-level affective state was used as the predictor,
children’s prompt-level affective state was used as the mediator, and children’s subsequent
physical activity level was used as the outcome. Since the predictor, mediator, and outcome
were all measured at level-1 (i.e., prompt-level), and prompts were nested in level-2 (i.e.,
mother-child pairs), multilevel mediation (1-1-1) model was used. Prompt-level positive affect,
negative affect, stress, and happiness was tested in separate models. Mplus was used to run all
multilevel mediation using structural equation model (i.e., MSEM) framework (Preacher, Zyphur,
& Zhang, 2010). The MSEM method implemented in Mplus can accommodate missing data,
unbalanced cluster size, and does not require the assumption of normality (Muthén &
Asparouhov, 2008).
Similar to disaggregating the within-person and between-person effects for multilevel
linear regression analysis, the within- and between-pair mediation effects for 1-1-1 mediation

81
models also needed to be separated. This way, we were able to examine (1) whether children’s
average affective states (i.e., 𝑐𝐴 𝐹𝐹𝐸𝐶 𝑇 � � � � � � � � � � � � �
𝑖 , represents 𝑐𝐴 𝐹𝐹𝐸𝐶 𝑇 𝐵𝑃
, which is pair-mean at level-2)
mediate the influence of mothers’ average affective states (i.e., 𝑚 𝐴 𝐹𝐹𝐸𝐶 𝑇 � � � � � � � � � � � � � �
𝑖 , represents
𝑚 𝐴 𝐹𝐹𝐸𝐶 𝑇 𝐵𝑃
, which is pair-mean at level-2) on children’s immediate subsequent physical
activity at the mother-child pair-level (i.e., the between-pair mediation effect); (2) whether
children’s relative affective states at each prompt (i.e., 𝑐𝐴 𝐹𝐹𝐸𝐶 𝑇 𝑡 𝑖 − 𝑐𝐴 𝐹𝐹𝐸𝐶 𝑇 � � � � � � � � � � � � �
𝑖 , denoted as
𝑐𝐴 𝐹𝐹𝐸𝐶 𝑇 𝑊𝑃 ) mediate the relationship between mothers’ relative affective states (i.e.,
𝑚 𝐴 𝐹𝐹𝐸𝐶 𝑇 𝑡 𝑖 − 𝑚 𝐴 𝐹𝐹𝐸𝐶 𝑇 � � � � � � � � � � � � � �
𝑖 , denoted as 𝑚 𝐴 𝐹𝐹𝐸𝐶 𝑇 𝑊𝑃 ) on children’s immediate subsequent
physical activity (i.e., the within-pair mediation effect). Mean centering predictor and mediator
within each mother-child pair (i.e., 𝑐𝐴 𝐹𝐹𝐸𝐶 𝑇 𝑊𝑃 and 𝑚 𝐴 𝐹𝐹𝐸𝐶 𝑇 𝑊𝑃 ) and use of pair-level means
on predictor (i.e., 𝑚 𝐴 𝐹𝐹𝐸𝐶 𝑇 𝐵𝑃
) and mediator (i.e., 𝑐𝐴 𝐹𝐹𝐸𝐶 𝑇 𝐵𝑃
) in the level-2 model of
outcome can “deconflate” the level-1 effects with their level-2 component (Zhang, Zyphur, &
Preacher, 2009).
The following equations demonstrate the 3 steps for a “deconflated” 1-1-1 mediation
model:
Step 1
level-1 𝑐 𝑃𝐴 𝑡 𝑖 = 𝛽 0 𝑖 (1)
+ 𝛽 1 𝑖 (1)
𝑚 𝐴 𝐹𝐹𝐸𝐶 𝑇 𝑊𝑃 + 𝑟 𝑡 𝑖 (1)

level-2 𝛽 0 𝑖 (1)
= 𝛾 0 0
(1)
+ 𝛾 0 1
(1)
𝑚 𝐴 𝐹𝐹𝐸𝐶 𝑇 𝐵𝑃
+ 𝑢 0 𝑖 (1)

𝛽 1 𝑖 (1)
= 𝛾 1 0
(1)

Step 2
level-1 𝑐𝐴 𝐹𝐹𝐸𝐶 𝑇 𝑡 𝑖 = 𝛽 0 𝑖 (2)
+ 𝛽 1 𝑖 (2)
𝑚 𝐴 𝐹𝐹𝐸𝐶 𝑇 𝑊𝑃 + 𝑟 𝑡 𝑖 (2)

level-2 𝛽 0 𝑖 (2)
= 𝛾 0 0
(2)
+ 𝛾 0 1
(2)
𝑚 𝐴 𝐹𝐹𝐸𝐶 𝑇 𝐵𝑃
+ 𝑢 0 𝑖 (2)

82
𝛽 1 𝑖 (2)
= 𝛾 1 0
(2)

Step 3
level-1 𝑐𝑃𝐴 𝑡 𝑖 = 𝛽 0 𝑖 (3)
+ 𝛽 1 𝑖 (3)
𝑚 𝐴 𝐹𝐹𝐸𝐶 𝑇 𝑊𝑃 + 𝛽 2 𝑖 (3)
𝑐𝐴 𝐹𝐹𝐸𝐶 𝑇 𝑊𝑃 + 𝑟 𝑡 𝑖 (3)

level-2 𝛽 0 𝑖 (3)
= 𝛾 0 0
(3)
+ 𝛾 0 1
(3)
𝑚 𝐴 𝐹𝐹𝐸𝐶 𝑇 𝐵𝑃
+ 𝛾 0 2
(3)
𝑐𝐴 𝐹𝐹𝐸 𝐶𝑇
𝐵𝑃
+ 𝑢 0 𝑖 (3)

𝛽 1 𝑖 (3)
= 𝛾 1 0
(3)

𝛽 2 𝑖 (3)
= 𝛾 2 0
(3)

Therefore, the level 1 mediation effect is
𝛾 1 0
(2)
× 𝛾 2 0
(3)

and the level 2 mediation effect is
𝛾 0 1
(2)
× 𝛾 0 2
(3)

In addition, a multilevel regression analysis was conducted to test the total effect of
mother’s affect on child’s physical activity. Nevertheless, a significant total effect is not a
necessary step before assessing the mediation effect (Rucker, Preacher, Tormala, & Petty, 2011).
Results
Data Availability
This study analyzed data from the first 100 mother-child dyads in the MATCH study.
Among these 100 dyads, 4 mothers had missing EMA data and 3 mothers did not report to be
with their children during any EMA prompts. Of the 93 mothers that can be matched with their
children’s EMA prompts, 5 children had missing EMA data and 8 children did not have any
valid accelerometer data during those EMA prompts. Therefore, there were a total of 80 mother-
child dyads included for analysis. Demographic information was only available for children’s
gender and age. Children in the analytical sample were 52.6% girls with an average age of 10.6

83
years (SD = .80). Children excluded from the analysis did not differ from children in the
analytical sample in terms of gender and age.
Descriptive Statistics
On average, mother received 28 EMA prompts (range = 7 – 35) across the 7-day
monitoring period. The average compliance rate was 75.0% (SD = 0.24). Mothers reported to
be with their children 52.8% (SD = 0.18) of the answered EMA prompts. Children received an
average of 23 EMA prompts (range = 2 – 32), with an average compliance rate of 80.0% (SD =
0.23). Table 6 shows the person-level average of mothers’ affective state when they reported to
be with their children, and their children’s subsequent affective state. During the 30-minute
window surrounding (i.e., ±15 minutes) the EMA prompts answered by children, they spent an
average of 1.28 minutes (SD = 0.98) in MVPA and 9.16 minutes (SD = 2.73) in LPA.
Multilevel Models
Table 7 shows the total effect of mothers’ current affective state on children’s subsequent
physical activity level. Results indicated that neither positive affect nor negative affect that
mothers experienced when they were with their children was associated with children’s
subsequent MVPA and LPA minutes.
As shown in Table 8, there was no significant direct nor indirect effect observed from the
multilevel mediation models, suggesting that mothers’ affective state was not related to
children’s subsequent MVPA minutes. Nevertheless, the mediation models indicated that
mothers’ affective state was associated with their children’s subsequent affective state.
Specifically, mothers’ positive affect was positively associated with children’s subsequent
positive affect, both at within-person level (beta = 0.117, SE = 0.041, p = 0.005) and at between-
person level (beta = 0.464, SE = 0.192, p = 0.015). This result was consistent when feeling of

84
happy was examined separately (WP beta = 0.079, SE = 0.037, p = 0.030; BP beta = 0.439, SE =
0.165, p = 0.008). These findings suggest that if mothers felt happier than her usual level when
with their children, then their children would also feel happier than their usual level at 30-60
minutes later (i.e., the WP effect); and for mothers who in general felt happier than the average
mother in the sample when with their children, their children would also feel happier than the
average children in the sample (i.e., the BP effect). Interestingly, mothers’ negative affect was
negatively associated with children’s subsequent negative affect at within-person level (beta = -
0.084, SE = 0.041, p = 0.040), suggesting that when mothers felt more negative than her usual
level when with their children, their children would feel less negative at 30-60 minutes later.
However, mothers’ feeling of stress was not associated with children’s subsequent feeling of
stress. Further, children’s positive affect was positively associated with their current MVPA
minutes at the between-person level, controlling for mothers’ positive affect (beta = 0.453, SE =
0.220, p = 0.039). This result suggests that children who in general felt more positive than the
average children in the sample engaged in more MVPA minutes. This effect became marginally
significant when examining feeling of happiness by itself. The mediation model also indicated
that children’s negative affect was positively associated with their current MVPA minutes at the
within-person level, controlling for mothers’ negative affect (beta = 0.821, SE = 0.414, p =
0.048). This result suggests that children were more active when they felt more negative than
their usual level.
Table 9 shows results from the multilevel mediation models testing the associations
between mother’s affective state and child’s subsequent LPA minutes, mediated by child’s
affective state. Results indicated that none of the direct or indirect effects was significant,
suggesting that mothers’ affective state was not related to children’s subsequent LPA minutes.

85
In addition, children’s affective state was not associated with their current LPA minutes,
controlling for mother’s affective state.
Discussions
This study used real-time EMA data from mother-child pairs and linked with children’s
accelerometer data to investigate whether mothers’ affective state is associated with children’s
subsequent physical activity level, mediated by children’s affective state. This novel
methodology allowed repeated assessments of mothers and children’s affect as they experienced
in their daily lives, and time-matched the responses at the mother-child dyad level. Further, the
use of accelerometer to continuously monitor children’s activity level throughout the day gives
the flexibility to time-link the activity data with EMA data, and helps minimizing potential bias
from self-report based physical activity assessment.
The current study is among the first to test whether maternal positive and negative affect
influenced children’s subsequent physical activity on a daily basis. Although previous studies
have suggested that maternal stress might influence children’s health-related behaviors (e.g.,
Parks et al., 2012, Park & Walton-Moss, 2012), this study did not find any significant total
effect, direct effect, or indirect effect of mothers’ affective state on children’s subsequent
physical activity level. One distinct feature of the current study is capturing mothers’ affective
state randomly during their daily lives. This approach is different from most of the previous
studies where maternal stress was measured as a “static” or “trait” variable (e.g., mother’s
overall stress level in the past year; Parks et al., 2012). The positive and negative affect that was
assessed in this study were “momentary” feeling state that mothers experienced as they went
through their daily lives. These momentary feelings that mothers experienced during their
everyday lives might have different effects on their children compared to the longer-term

86
maternal mood variables such as depression and subjective well-being, as many studies have
shown the differential effects of emotional state vs. trait on health and health-related behaviors
(e.g., Salovey, Rothman, Detweiler, & Steward, 2000; Pressman & Cohen, 2005). This potential
difference in emotional state vs. trait might also explain why the current study did not find any
association between mothers’ momentary stress and children’s momentary stress, although
previous studies found that parental stress and children’s stress are positively correlated (e.g.,
Ge, Conger, Lorenz, & Simons, 1994; Goodman et al., 2011).
The intensive longitudinal data also allowed the current study to examine the between-
person differences and within-person variations in relationships between mothers and children’s
affective state. Results from this study suggested that in general, for mothers who had higher
positive affect than the “average” mothers in this study, their children also had higher positive
affect than the “average” children in this study (i.e., the between-person effect); and if mothers
felt happier than her usual level, their children would also feel happier than their usual level
subsequently (i.e., the within-person effect). These results suggested a strong and consistent
positive relationship between mothers and children’s positive affective state. This finding is
contrary to previous studies, which found no relationship between mothers and children’s
positive affect (e.g., Ben-Zur, 2003) nor between parents and children’s positive affect (e.g.,
Casas et al., 2012; Bedin & Sarriera, 2014). Nevertheless, these studies measured the trait
positive affect (i.e., feelings and emotions over the past week; feelings about life in general),
which might represent different characteristics from state positive affect that was measured in
this study. Further, the current study revealed a negative association between mothers’
momentary negative affect and children’s subsequent negative affect at the within-person level
while the Ben-Zur’s study (2003) found no relationship between mothers and children’s negative

87
affect. The negative association at the within-person level found in this study suggests that when
mothers felt more negative than her usual level at the moment, her child would feel less negative
than his/her usual level 30-60 minutes later. Because of the time-lagged feature of this study, it
is possible that mothers and children experienced the negative affect together, and mothers
helped the children relieve such negative event, which led to a decrease in negative affect in
children. Nevertheless, when examining stress separately, there was no significant association
between mothers and children’s feeling of stress. This implies that the negative affect that
mothers and children reported did not encompass daily stress (e.g., feelings of mad and sad).
Results from this study suggests that children who in general felt more positive than
others engaged in more MVPA minutes (i.e., the between-person effect), although feeling more
positive than children’s usual level was only marginally related to more MVPA minutes (i.e., the
within-person effect). The positive between-person association between positive affect and
physical activity is consistent with previous findings that individuals with high positive affect
and emotional wellbeing tend to be more active (e.g., Lyubomirsky, King, & Diener, 2005;
Steptoe & Butler, 1996). However, feeling more positive at the moment might not be associated
with current activity level, as suggested by the null within-person effect. This finding is similar
to previous findings in children that feeling more positive at the moment did not lead to more
MVPA minutes in the next 30 minutes (Dunton et al., 2014). Interestingly, the current study
revealed a positive association between negative affect and MVPA minutes at the within-person
level, suggesting that children were more active when they felt more negative than their usual
level. This finding is different from Dunton’s study where negative affect was not related to
subsequent MVPA minutes. However, the current study examined children’s affective state with
current MVPA minutes, which combined MVPA minutes during the 15-minute before and 15-

88
minute after the EMA prompt. Therefore, it is possible that the increased negative affect could
be due to the engagement of MVPA, as found in Dunton’s (2014) study. Lastly, this study did
not find any association between children’s momentary affective state and current LPA minutes.
It is likely that activity intensity is a potential moderator of the relationship between affect and
physical activity for children, such that children’s affective state might only relate to higher
intensity activity but not lower intensity activity.
Limitations
This study examined how mothers and children’s affective state using a time-lagged
method (e.g., mother’s affective states predicating children’s subsequent physical activity).
Although this method creates clear temporal sequence, we did not capture the mutual affective
states (i.e., prompting mothers and children to record their affect at the exact same time). We
also did not control for children’s prior affective states, and therefore we were unable to examine
children’s change in affective states. In addition, this study only assessed affective states, but not
the “event” that mothers and children were engaging in (e.g., we do not know what behaviors or
situations that were associated with mothers and children’s positive or negative affect).
Although we used two items to create the composite score for positive affect and three items for
negative affect, we might still miss some other aspects of these two core affective states (e.g.,
excited, enthusiastic, nervous, scared). We also did not measure any physical feelings (e.g.,
energy and fatigue). For children’s physical activity levels, we used minutes in MVPA within a
30-minute window. However, this variable is not normally distributed, and in 45.5% of the
prompts, children did not engage in any MVPA minutes. Nevertheless, the MSEM method we
used for analysis does not require the assumption of normality (Muthén & Asparouhov, 2008),
and we also used LPA minutes as another measure for physical activity levels. Further, we did
not control for the potential effect of a “third person” that might be present (e.g., father, siblings),

89
which may influence both children’s affect and physical activity level. Lastly, we examined
children’s concurrent physical activity (i.e., ± 15 minutes of the assessment of affective states)
instead of their subsequent physical activity (i.e., activity levels after the assessment of affective
states. Therefore, the temporal relationship between children’s affective states and their physical
activity levels could not be established.
Implications
This study demonstrated one way to capture real-time information at the dyadic level. By
carefully design the EMA prompting schedule (e.g., time-lagged or concurrently), researchers
could have the flexibility to study how one person might influence another person’s feelings,
attitudes, perceptions, and behaviors. EMA can also potentially be used at the group level (e.g.,
all members in a family, all students in a classroom) to explore interpersonal relationships and
behavioral interactions at the social network scale.
Although this study did not find a relationship between mothers’ affective states and
children’s physical activity level, results suggested that mothers’ affective states were related to
children’s subsequent affective states, and children’s affective states were associated with their
concurrent physical activity levels. Future studies could further explore whether this potential
mediation effect might differ by children’s gender. For example, previous studies have found
that parents’ feeling of stress were correlated with girls but not boys, and parents’ feeling of
excitement was correlated with boys but not girls (Bedin & Sarriera, 2014). In addition to
children’s gender, future studies could also examine the effects of other potential moderators
such as social and physical context of physical activity (Dunton, Liao, Intille, Wolch, & Pentz,
2011). Further, compared to adults, relatively few studies investigated the association between
light physical activity and affective state among children (Birkeland, Torsheim, & Wold, 2009).

90
Nevertheless, engaging in more light physical activity could imply less time being spent in
sedentary activity and is found to be negatively associated with children’s overweight status
(Butte, Puyau, Adolph, Vohra, & Zakeri, 2007). Lastly, while the current study examined
children’s concurrent physical activity levels, future studies could investigate whether children’s
affective states predict their subsequent physical activity (e.g., in the next 30 minutes),
controlling for mothers’ affective states.
In summary, this study showed a promising novel method to understand the parental
influence on children’s daily physical activity level. More research is needed in this area to
better determine the best strategy to create an environment that can encourage active living for
both generations.

91

Table 6. Mother’s affective state when with their children and children’s subsequent affective
state (N = 80)
Mothers Children
Mean (SD)
Positive Affect
Mean (SD)
1.71 (0.49) 2.02 (0.74)
Happiness 1.82 (0.54) 2.10 (0.74)
Negative Affect 0.40 (0.16) 0.36 (0.15)
Stress 0.49 (0.41) 0.30 (0.41)

Table 7. Total effect of mother’s affective state on child’s subsequent physical activity
Light Physical Activity
(LPA)

Moderate-to-Vigorous Physical Activity
(MVPA)

p Beta Estimate (SE) p Beta Estimate (SE)
Positive Affect
WP Effect -0.076 (0.382) 0.843 0.150 (0.159) 0.345
BP Effect -0.329 (0.772) 0.670 0.073 (0.309) 0.814

Happiness
WP Effect 0.209 (0.349) 0.549 0.144 (0.125) 0.246
BP Effect -0.246 (0.649) 0.704 0.199 (0.300) 0.507

Negative Affect
WP Effect -0.292 (0.937) 0.755 -0.075 (0.468) 0.873
BP Effect 1.921 (1.666) 0.249 0.210 (1.022) 0.837

Stress
WP Effect -0.158 (0.411) 0.700 -0.238 (0.178) 0.181
BP Effect 1.184 (0.788) 0.133 0.506 (0.335) 0.131

92

Table 8. Associations between mother’s affective state and child’s subsequent moderate-to-vigorous physical activity (MVPA),
mediated by child’s affective state
a Path
(Mother’s Affect on
Child’s Affect)

b Path
(Child’s Affect on
Child’s MVPA Minutes)

Direct Effect Indirect Effect
p Beta Estimate
(SE)
p Beta Estimate
(SE)
p Beta Estimate
(SE)
p Beta Estimate
(SE)
Positive
Affect
WP Effect 0.117 (0.041) 0.005 0.356 (0.196) 0.070 0.109 (0.158) 0.491 0.042 (0.026) 0.103
BP Effect 0.464 (0.192) 0.015 0.453 (0.220) 0.039 -0.188 (0.361) 0.602 0.210 (0.140) 0.133

Happiness
WP Effect 0.079 (0.037) 0.030 0.362 (0.195) 0.063 0.116 (0.122) 0.343 0.029 (0.018) 0.120
BP Effect 0.439 (0.165) 0.008 0.389 (0.212) 0.067 -0.013 (0.330) 0.969 0.170 (0.115) 0.138

Negative
Affect
WP Effect -0.084 (0.041) 0.040 0.821 (0.414) 0.048 -0.006 (0.468) 0.990 -0.069 (0.049) 0.157
BP Effect 0.180 (0.116) 0.121 0.328 (0.890) 0.713 0.126 (1.046) 0.904 0.059 (0.169) 0.726

Stress
WP Effect 0.032 (0.045) 0.485 0.414 (0.283) 0.143 -0.251 (0.174) 0.150 0.013 (0.023) 0.571
BP Effect 0.000 (0.128) 0.998 0.259 (0.249) 0.298 0.464 (0.336) 0.168 0.000 (0.033) 0.998

93

Table 9. Associations between mother’s affective state and child’s subsequent light physical activity (LPA), mediated by child’s
affective state
a Path
(Mother’s Affect on
Child’s Affect)

b Path
(Child’s Affect on
Child’s LPA Minutes)

Direct Effect Indirect Effect
p Beta Estimate
(SE)
p Beta Estimate
(SE)
p Beta Estimate
(SE)
p Beta Estimate
(SE)
Positive
Affect
WP Effect 0.118 (0.041) 0.004 0.381 (0.407) 0.349 -0.121 (0.383) 0.753 0.045 (0.052) 0.388
BP Effect 0.464 (0.192) 0.015 -0.522 (0.584) 0.371 -0.030 (0.705) 0.966 -0.242 (0.302) 0.422

Happiness
WP Effect 0.080 (0.036) 0.029 0.403 (0.401) 0.315 0.177 (0.351) 0.614 0.032 (0.036) 0.369
BP Effect 0.439 (0.165) 0.008 -0.497 (0.589) 0.399 0.020 (0.598) 0.973 -0.218 (0.288) 0.449

Negative
Affect
WP Effect -0.084 (0.041) 0.042 -0.250 (0.908) 0.783 -0.313 (0.926) 0.735 0.021 (0.080) 0.792
BP Effect 0.180 (0.116) 0.121 -2.288 (1.535) 0.136 2.464 (1.736) 0.156 -0.413 (0.398) 0.300

Stress
WP Effect 0.033 (0.045) 0.466 0.486 (0.335) 0.148 -0.174 (0.409) 0.670 0.016 (0.025) 0.519
BP Effect 0.000 (0.128) 0.998 -0.776 (0.569) 0.172 1.289 (0.801) 0.108 0.000 (0.100) 0.998

94
CHAPTER 5: DISCUSSION AND CONCLUSIONS
The goals of this dissertation study were to (1) demonstrate the use of real-time data
capture techniques to assess affective states (via electronic Ecological Momentary Assessment)
and physical activity (via accelerometer) in individual’s everyday life, and (2) examine the
associations between affective states and physical activity in free-living settings. The first study
used cross-sectional data to explore the bi-directional relationships between affective states (i.e.,
positive affect, negative effect, energy, and fatigue) and activity levels (i.e., light physical
activity and moderate-to-vigorous physical activity) among adults. Because of the nature of
repeated-measure of real-time data capture methods, it was possible to test the between-person
differences and within-person variations of those relationships. The second study used
longitudinal data to examine whether and how affective states during physical activity might
predict future activity levels. This is the first known study that examined the long-term effect of
affective responses to free-living physical activity using real-time data capture methods. The
third study used dyadic data from mother-child pairs to investigate whether mothers’ affective
states might associate with children’s subsequent physical activity levels through influencing
children’s affective states. This study demonstrated how to apply real-time data capture methods
at the dyadic level, and thus allowed the opportunity to explore the interpersonal relationships
between affective states and physical activity.
Results from Study 1 showed that a more positive affective feeling state led to an
immediate increase in moderate-to-vigorous physical activity. Nevertheless, this increase in
physical activity might not last beyond 15 minutes. When examining individuals’ affective
feeling state during self-report physical activity only, Study 2 found that a more positive
affective response during physical activity was associated with an increase in moderate-to-

95
vigorous physical activity up to 12 months later. These results suggest that affective feeling state
could be a potential predictor of both shorter-term physical activity behavior and longer-term
physical activity maintenance.
While Study 1 found a negative association between negative affect and subsequent
moderate-to-vigorous physical activity at the within-person level among adults, Study 3 found
the opposite. Results from Study 3 suggest a positive association between negative affect and
concurrent moderate-to-vigorous physical activity at the within-person level among children.
This discrepancy in direction of effect could be due to the different temporal relationship being
examined: Study 1 tested the subsequent physical activity levels while Study 3 tested the
concurrent physical activity levels. Further, Study 1 did not find any significant changes in
negative affect following physical activity. These findings imply that the relationship between
physical activity and affective states could differ depending on whether the assessment of affect
was done before, during, or after the physical activity.
When examining the relationships between affective states and light physical activity,
Study 1 found a positive association between negative affect and light activity, and a positive
association between physical feeling states (i.e., energy and fatigue) and light activity among
adults. However, Study 3 did not find any significant relationship between affective states and
light physical activity among children. It is possible that affective states might only influence
light physical activity among adults but not children. Nevertheless, Study 1 tested the direct
relationship between affective states and light physical activity while Study 3 tested this
relationship in a mediation model (i.e., effect of affective states and light physical activity among
children, controlling for mothers’ affective states). Therefore, more research is needed before a
solid conclusion can be made.

96
Implications
Methodological Implications
Although real-time assessment of everyday affect could be done through paper-pencil
diary (e.g., Annesi, 2002b), the use of electronic device (e.g., beeper alert) can help with careful
timing (either at fixed intervals or at random times) of each assessment (e.g., Gauvin, Rejeski, &
Norris, 1996). Nevertheless, it is hard to tell whether participants actually write down their
responses at each scheduled time when using the paper-pencil method (e.g., it is possible that an
individual can write down all the answers at once). More recently, researchers have
implemented the EMA protocol using portable electronics such as PDA (e.g., Mata et al., 2012;
Schwerdtfeger et al., 2010) and loaned mobile phones (e.g., Study 1 and Study 2). The use of
electronic EMA could give researchers greater confidence that the assessment was completed
when it should have been. However, these portable electronics are usually an extra device that
participants need to carry around with them at all times, it is possible that participants might
forget to bring the device to work or forget to recharge the device when it is out of battery.
Therefore, to decrease participant burden and increase compliance, it might be advantageous to
implement the EMA protocol to a device that participants already normally use, for example,
their personal smartphones (e.g., Study 3). According to the most recent survey conducted by
Pew Research Center, 64% of American adults own a smartphone as of 2014, up from 35% just
3 years ago in 2011 (Smith, 2015). At the same time, more researchers have devoted their efforts
to develop EMA apps for both Android-based (e.g., Kini, 2013) and IOS-based (e.g., Runyan et
al., 2013) smartphones. Thus, using EMA via personal smartphones could be considered as a
more convenient method for future technology-based assessment in behavioral research studies.

97
This dissertation study also demonstrated the creation of various time windows, which
could be used as one of the data reduction strategies for continuously self-recording data (e.g.,
accelerometer data). Data from accelerometer was aggregated within various real-time windows,
which were created by linking with EMA data. Since each EMA data point could contain rich
self-report information regarding individual’s current status (e.g., affective state, cognitive state,
perceptions, activity type, physical and social context), it could complement the objectively
measured data where such information would be hard to capture. Therefore, instead of
aggregating thousands of continuous measured data points (e.g., activity level, GPS location,
heart rate, glucose level) into person-level or day-level, these data points could be time-matched
with EMA data (e.g., random times of individuals’ daily lives, specific daily events) and
aggregated at much smaller time intervals (e.g., average glucose level 30-minute before and 10-
minute after each eating episode). Since temporal sequences can be established, these real-time
windows could be used to explore the possible predictors and consequences of people’s daily
health-related behaviors.
Theoretical Implications
The relationship between physical activity and affect has intrigued researchers for
decades. There are many theories attempting to explain the link between the two, including from
both biological perspectives (e.g., the thermogenic hypothesis, deVries et al., 1981; the
monoamine hypothesis, Chaouloff, 1989) and psychological perspectives (e.g., the distraction
hypothesis, Morgan, 1985; the mastery hypothesis, North, McCullagh, & Tran, 1990). While
these are all plausible mechanisms, it is hard to directly test each of these hypotheses. There are
several theories addressing the importance of activity intensity on relationship between physical
activity and affective states (e.g., the inverted-U curve model, Ojanen, 1994; the dual mode
model, Ekkekakis, 2003). Nevertheless, currently evidence that show support for these theories

98
are all from lab-based studies (e.g., Welch, Hulley, Ferguson, & Beauchamp, 2007; Ekkekakis,
Parfitt, & Petruzzello, 2011).
While it might be easier to control the conditions (e.g., duration, intensity) of physical
activity under a lab setting for theory testing purpose, these types of physical activity could be
very different from what people actually engage in during their everyday lives. Consequently,
the relationship found between physical activity and affect from lab-based studies might not be
extended to real-life settings. For example, although many lab-bases studies found a positive
relationship between physical activity and positive affect (e.g., Reed & Ones, 2006), this
dissertation study did not find positive affect to be associated with current activity level,
immediate subsequent activity level, or future activity level. Some possible reasons of this
discrepancy could be that while lab-based physical activity could be uniformed (e.g., all
participants engage in the same type of exercise), free-living physical activity could vary greatly
across individual from type of exercise to physical and social context of exercise (e.g., Dunton,
Whalen, Jamner, & Floro, 2007). And these different conditions could potentially influence
people’s affective feelings. For instance, people were in a greater positive mood when with
friends compared with alone (Larson, 1990); greater mental benefits were found for outdoor
physical activity than indoors (Thompson Coon et al., 2011). Some recent studies also showed
physical and social context as a moderator of affective responses to physical activity (Dunton,
Liao, Intille, Huh, & Leventhal, in press). Therefore, theories need to consider the conditions
and variables that might impact the relationship between free-living physical activity and
affective state in order to be more translational to real-life situations.
Figure 4 demonstrates a proposed multilevel framework that outlines some key factors
that might influence the relationships between physical activity and affective states in real-life

99
settings. Adapted from the socio-ecological model (Sallis, Owen, & Fisher, 2008), this
framework outlines how social and physical environment might interact with activity and
personal characteristics to impact the bi-directional relationship between physical activity and
affective states. For example, results from Study 3 suggest that when children were with their
mothers (a specific social context), their positive and negative affective state was associated with
their mothers’. Although Study 3 did not find an effect of mothers’ affective state on children’s
subsequent physical activity, it is possible that such effect might vary by their physical location
(e.g., home indoor vs. outside at a park), and/or with some children’s demographic
characteristics (e.g., gender, weight status).
Intervention Implications
Results from this dissertation study could help inform future interventions that aim to
promote physical activity and enhance mental well-being in people’s daily lives. For example,
this study found that engaging in physical activity, regardless of the length and intensity of the
activity, led an increase in feeling of energy. Engaging in light physical activity also led to a
decrease in feeling of fatigue. Therefore, interventions could encourage people taking a short
break from prolonged sedentary activity (e.g., sitting for long hours at work) in order to maintain
a high level of energy. The positive affective changes from physical activity could be used as an
intrinsic motivation to engage in physical activity in the future (e.g., “I engage in physical
activity because I want to feel more energetic”). Further, since negative affect was negatively
associated with subsequent and future moderate-to-vigorous physical activity, interventions
could try to help individual overcome this affective barrier by making activities more fun and
enjoyable. For previously inactive individuals, interventions might want to start them with less
intense activity so that they might be more likely to experience positive affective change from
the activity.

100
Future Research Directions
With the advancement in technology, it is more feasible and affordable to apply real-time
data capture techniques in behavioral research than ever before. Examining the potential
predictors and consequences of health-related behaviors under free-living settings can help
researchers to better understand those behavioral mechanisms and determinants. This knowledge
will ultimately lead to design of more effective behavioral interventions that promote healthier
lifestyle. Therefore, future studies should consider verifying the ecological validity of existing
lab-based theoretical models in free-living settings.
Recognizing the complex factors that might influence an individual’s daily behavior, the
relationship between physical activity and affective state might involve multiple variables
beyond the characteristics of physical activity (e.g., intensity and duration) and individuals (e.g.,
gender and age) themselves. Under the proposed multilevel framework, future research could
further explore how physical environment (e.g., location, built environment features) and social
environment (e.g., social companion, presence of pets, support from family and friends) might
interact with activity and personal characteristics to influence the relationship between physical
activity and affective state. A better understanding of this could help researchers to identify the
situations where individuals might achieve the maximal mental benefits from physical activity.
For example, running with a dog in a park with well-maintained trails and lots of trees might be a
more pleasurable experience for inactive overweight adults than running on a treadmill at gym
by themselves.
Future studies could also explore the conditions where individuals are more likely to
naturally be active, especially for non-structured short activity bouts (e.g., brisk walking from
office to parking lot, taking stairs instead of elevators). Shorter activity bouts might be easier to

101
integrate into individuals’ daily routine than structured physical activity (e.g., a predetermined
running session). Further, engaging in short activity bouts (e.g., <10 minutes) and accumulating
higher intensity physical activity are both strongly associated with several biologic health
outcomes (Loprinzi & Cardinal, 2013; Fan et al., 2013). The use of objective activity monitor
(such as accelerometer as used in this dissertation study) could help with capturing these short
activity bouts. Electronic EMA could be used to assess the social and physical environment, as
well as individual factors such as affective state and cognitive state. Linking data from these two
sources could allow researchers to explore the interaction and meditational effect of contextual
factors, personal factors, and spontaneous activity bouts.
In summary, the use of real-time data capture techniques offers a great opportunity for
researchers moving beyond the lab settings to assess and study individuals’ daily behaviors and
factors that are related to those behaviors. This dissertation study demonstrated the combined
use of electronic EMA and accelerometer to examine the effects of personal factor (i.e.,
individuals’ affective states) and interpersonal factor (i.e., mothers’ affective states) on daily
physical activity levels. Methods used in this dissertation study could be used to guide future
research that is interested in adopting real-time data capture techniques.

102

Figure 4. A proposed multilevel framework for factors that might influence the bi-directional relationship between physical activity
and affective state in real-life situations

Physical Activity

Affective State
Activity Characteristics
Social Environment
Physical Environment
Activity Type Activity Intensity
Family/Peer Support
Physical Context Environmental Feature
Social Context
Individual Characteristics
Cognitive Factors

103
REFERENCES
Almeida, D. M., Wethington, E., & Kessler, R. C. (2002). The daily inventory of stressful events
an interview-based approach for measuring daily stressors. Assessment, 9(1), 41-55.
Annesi, J. J. (2002a). Relation of rated fatigue and energy after exercise and over 14 weeks in
previously sedentary women exercisers. Perceptual and Motor Skills, 95, 719-727.
Annesi, J. J. (2002b). Relationship between changes in acute exercise-induced feeling states,
self-motivation, and adults' adherence to moderate aerobic exercise. Perceptual and
Motor Skills, 94(2), 425-439.
Annesi, J. J. (2002c). Self-motivation moderates effect of exercise-induced feelings on
adherence. Perceptual and motor skills, 94(2), 467-475.
Annesi, J. J. (2006). Relations of self-motivation, perceived physical condition, and exercise-
induced changes in revitalization and exhaustion with attendance in women initiating a
moderate cardiovascular exercise regimen. Women & Health, 42(3), 77-93.
Arent, S., Landers, D. M., & Etnier, J. L. (2000). The effects of exercise on mood in older adults:
A meta-analytic. Journal of Ageing and Physical Activity, 8, 407-430.
Austin, V., Shah, S., & Muncer, S. (2005). Teacher stress and coping strategies used to reduce
stress. Occupational Therapy International, 12(2), 63-80.
Backhouse, S. H., Ekkekakis, P., Biddle, S. J., Foskett, A., & Williams, C. (2007). Exercise
makes people feel better but people are inactive: Paradox or artifact?. Journal of Sport
and Exercise Psychology, 29(4), 498.
Barrett, L. F., & Barrett, D. J. (2001). An introduction to computerized experience sampling in
psychology. Social Science Computer Review, 19(2), 175-185.

104
Bauman, A. E., Reis, R. S., Sallis, J. F., Wells, J. C., Loos, R. J., & Martin, B. W. (2012).
Correlates of physical activity: why are some people physically active and others not?.
The Lancet, 380(9838), 258-271.
Bedin, L. M., & Sarriera, J. C. (2014). Dyadic analysis of parent-children subjective well-being.
Child Indicators Research, 7(3), 613-631.
Beets, M. W., Cardinal, B. J., & Alderman, B. L. (2010). Parental social support and the physical
activity–related behaviors of youth: a review. Health Education & Behavior, 37(5), 621-
644.
Belcher, B. R., Berrigan, D., Dodd, K. W., Emken, B. A., Chou, C. P., & Spruijt-Metz, D.
(2010). Physical activity in US youth: effect of race/ethnicity, age, gender, and weight
status. Medicine & Science in Sports & Exercise, 42(12), 2211-2221.
Ben-Zur, H. (2003). Happy adolescents: The link between subjective well-being, internal
resources, and parental factors. Journal of Youth and Adolescence, 32(2), 67-79.
Bentham, J. (1962). The collected works of Jeremy Bentham: An introduction to the principles
of morals and legislation. New York: Oxford University Press.
Berger, B. G. (1994). Coping with stress: The effectiveness of exercise and other techniques.
Quest, 46(1), 100-119.
Biddle, S. J. H. (2001). Emotion, mood and physical activity. In S. J. H. Biddle, K. R. Fox, & S.
H. Boutcher (Eds.), Physical activity and psychological well-being (pp. 63-87). London:
Routledge.
Birkeland, M. S., Torsheim, T., & Wold, B. (2009). A longitudinal study of the relationship
between leisure-time physical activity and depressed mood among adolescents.
Psychology of Sport and Exercise, 10(1), 25-34.

105
Bixby, W. R., Spalding, T. W., & Hatfield, B. D. (2001). Temporal dynamics and dimensional
specificity of the affective response to exercise of varying intensity: differing pathways to
a common outcome. Journal of Sport & Exercise Psychology, 23(3), 171-190.
Bouchard, C., & Malina, R. M. (1983). Genetics of physiological fitness and motor performance.
Exercise & Sport Sciences Reviews, 11(1), 306-339.
Boutcher, S. H. (1993). Emotion and aerobic exercise. In R. Singer, M. Murphy, & L. Tennant
(Eds.), Handbook on research in sport psychology (pp. 799-814). New York: Macmillan.
Bozarth, M. A. (1994). Pleasure systems in the brain. In D. M. Wartburton (Ed.), Pleasure: The
politics and the reality (pp. 5-14). New York: Wiley & Sons.
Brodersen, N. H., Steptoe, A., Williamson, S., & Wardle, J. (2005). Sociodemographic,
developmental, environmental, and psychological correlates of physical activity and
sedentary behavior at age 11 to 12. Annals of Behavioral Medicine, 29(1), 2-11.
Bussmann, J. B. J., Ebner-Priemer, U. W., & Fahrenberg, J. (2009). Ambulatory activity
monitoring: Progress in measurement of activity, posture, and specific motion patterns in
daily life. European Psychologist, 14(2), 142-152.
Butte, N. F., Puyau, M. R., Adolph, A. L., Vohra, F. A., & Zakeri, I. (2007). Physical activity in
nonoverweight and overweight Hispanic children and adolescents. Medicine and Science
in Sports and Exercise, 39(8), 1257-1266.
Carels, R. A., Coit, C., Young, K., & Berger, B. (2007). Exercise makes you feel good, but does
feeling good make you exercise?: An examination of obese dieters. Journal of Sport &
Exercise Psychology, 29(6), 706-722.

106
Casas, F., Coenders, G., González, M., Malo, S., Bertran, I., & Figuer, C. (2012). Testing the
relationship between parents’ and their children’s subjective well-being. Journal of
Happiness Studies, 13(6), 1031-1051.
Chaouloff, F. (1989). Physical exercise and brain monoamines: a review. Acta Physiologica
Scandinavica, 137(1), 1-13.
Chen, K. Y., & Bassett, D. R. (2005). The technology of accelerometry-based activity monitors:
current and future. Medicine & Science in Sports and Exercise, 37(S11), S490-S500.
Cook, W. L., & Kenny, D. A. (2005). The Actor–Partner Interdependence Model: A model of
bidirectional effects in developmental studies. International Journal of Behavioral
Development, 29(2), 101-109.
Curran, P. J., & Bauer, D. J. (2011). The disaggregation of within-person and between-person
effects in longitudinal models of change. Annual Review of Psychology, 62, 583-619.
Danaei, G., Ding, E. L., Mozaffarian, D., Taylor, B., Rehm, J., Murray, C. J., & Ezzati, M.
(2009). The preventable causes of death in the United States: Comparative risk
assessment of dietary, lifestyle, and metabolic risk factors. PLoS Medicine, 6(4),
e1000058.
Deater-Deckard, K., & O'Connor, T. G. (2000). Parent–child mutuality in early childhood: Two
behavioral genetic studies. Developmental Psychology, 36(5), 561-570.
Deater‐ Deckard, K., & Petrill, S. A. (2004). Parent–child dyadic mutuality and child behavior
problems: An investigation of gene–environment processes. Journal of Child Psychology
and Psychiatry, 45(6), 1171-1179.

107
deVries, H. A., Wiswell, R. A., Bulbulian, R., & Moritani, T. (1981). Tranquilizer effect of
exercise. Acute effects of moderate aerobic exercise on spinal reflex activation level.
American Journal of Physical Medicine, 60(2), 57-66.
Drever, J. (1952). A dictionary of psychology. Oxford: Penguin Books.
Dubnov, G., & Berry, E. M. (2000). Physical activity and mood. The endocrine connection. In N.
Constantini & A. C. Hackney (Eds), Endocrinology of physical activity and sport (pp.
405-415). New York: Springer Science.
Dunton, G. F., Huh, J., Leventhal, A. M., Riggs, N., Spruijt-Metz, D., Pentz, M. A. & Hedeker,
D. (2014). Momentary assessment of affect, physical feeling states, and physical activity
in children. Health Psychology, 33(3), 255-263.
Dunton, G. F., Liao, Y., Almanza, E., Jerrett, M., Spruijt-Metz, D., Chou, C. P., & Pentz, M. A.
(2012). Joint physical activity and sedentary behavior in parent-child pairs. Medicine &
Science in Sports & Exercise, 44(8), 1473-1480.
Dunton, G. F., Liao, Y., Intille, S. S., Huh, J., & Leventhal A. (in press). Contextual influences
on affective response during physical activity. Health Psychology.
Dunton, G. F., Liao, Y., Intille, S., Wolch, J., & Pentz, M. A. (2011). Physical and social
contextual influences on children's leisure-time physical activity: An ecological
momentary assessment study. Journal of Physical Activity and Health, 8(1), S103-108.
Dunton, G. F., & Vaughan, E. (2008). Anticipated affective consequences of physical activity
adoption and maintenance. Health Psychology, 27(6), 703-710.
Dunton, G. F., Whalen, C. K., Jamner, L. D., & Floro, J. N. (2007). Mapping the social and
physical contexts of physical activity across adolescence using ecological momentary
assessment. Annals of Behavioral Medicine, 34(2), 144-153.

108
Edwardson, C. L., & Gorely, T. (2010). Parental influences on different types and intensities of
physical activity in youth: A systematic review. Psychology of Sport and Exercise, 11(6),
522-535.
Ekkekakis, P. (2003). Pleasure and displeasure from the body: Perspectives from exercise.
Cognition & Emotion, 17(2), 213-239.
Ekkekakis, P., Hall, E. E., & Petruzzello, S. J. (2004). Practical markers of the transition from
aerobic to anaerobic metabolism during exercise: rationale and a case for affect-based
exercise prescription. Preventive Medicine, 38(2), 149-159.
Ekkekakis, P., Hall, E. E., & Petruzzello, S. J. (2005). Variation and homogeneity in affective
responses to physical activity of varying intensities: an alternative perspective on dose–
response based on evolutionary considerations. Journal of Sports Sciences, 23(5), 477-
500.
Ekkekakis, P., Hall, E. E., & Petruzzello, S. J. (2008). The relationship between exercise
intensity and affective responses demystified: to crack the 40-year-old nut, replace the
40-year-old nutcracker!. Annals of Behavioral Medicine, 35(2), 136-149.
Ekkekakis, P., Hall, E. E., VanLanduyt, L. M., & Petruzzello, S. J. (2000). Walking in (affective)
circles: can short walks enhance affect?. Journal of Behavioral Medicine, 23(3), 245-275.
Ekkekakis, P., Parfitt, G., & Petruzzello, S. J. (2011). The pleasure and displeasure people feel
when they exercise at different intensities. Sports Medicine, 41(8), 641-671.
Ekkekakis, P., & Petruzzello, S. J. (1999). Acute aerobic exercise and affect: Current status,
problems and prospects regarding dose–response. Sports Medicine, 28(5), 337–374.

109
Eriksson, M., Nordqvist, T., & Rasmussen, F. (2008). Associations between parents' and 12-
year-old children's sport and vigorous activity: the role of self-esteem and athletic
competence. Journal of Physical Activity & Health, 5(3), 359-373.
Esch, T., & Stefano, G. B. (2010). Endogenous reward mechanisms and their importance in
stress reduction, exercise and the brain. Archives of Medical Science, 6(3), 447-455.
Fairchild, A. J., & MacKinnon, D. P. (2009). A general model for testing mediation and
moderation effects. Prevention Science, 10(2), 87-99.
Fan, J. X., Brown, B. B., Hanson, H., Kowaleski-Jones, L., Smith, K. R., & Zick, C. D. (2013).
Moderate to vigorous physical activity and weight outcomes: does every minute count?.
American Journal of Health Promotion, 28(1), 41-49.
Fleeson, W. (2009). Studying personality processes: Explaining change in between-persons
longitudinal and within-person multilevel models. In R. W. Robins, R. C. Fraley, & R. F.
Krueger (Eds.). Handbook of Research Methods in Personality Psychology (pp. 523-
542). New York: Guilford Press.
Foster, E. M. (2002). How economists think about family resources and child development.
Child Development, 73(6), 1904-1914.
Fredrickson, B. L., & Kahneman, D. (1993). Duration neglect in retrospective evaluations of
affective episodes. Journal of Personality and Social Psychology, 65(1), 45-55.
Freedson, P., Pober, D., & Janz, K. F. (2005). Calibration of accelerometer output for children.
Medicine & Science in Sports & Exercise, 37(11 Suppl), S523-530.
Freud, S. (1955). Beyond the pleasure principle. In J. Strachey (Ed.). The complete psychological
works of Sigmund Freud, volume XVIII (1920-1922): Beyond the pleasure principle,
group psychology and other works (pp. 1-64). London: Hogarth Press.

110
Fuemmeler, B. F., Anderson, C. B., & Mâsse, L. C. (2011). Parent-child relationship of directly
measured physical activity. International Journal of Behavioral Nutrition & Physical
Activity, 8(1), 17.
Gauvin, L., & Rejeski, W. J. (1993). The exercise-induced feeling inventory: Development and
initial validation. Journal of Sport and Exercise Psychology, 15(4), 403-423.
Gauvin, L., Rejeski, W. J., & Norris, J. L. (1996). A naturalistic study of the impact of acute
physical activity on feeling states and affect in women. Health Psychology, 15(5), 391-
397.
Ge, X., Conger, R. D., Lorenz, F. O., & Simons, R. L. (1994). Parents' stressful life events and
adolescent depressed mood. Journal of Health and Social Behavior, 28-44.
Goodman, S. H., Rouse, M. H., Connell, A. M., Broth, M. R., Hall, C. M., & Heyward, D.
(2011). Maternal depression and child psychopathology: a meta-analytic review. Clinical
Child and Family Psychology Review, 14(1), 1-27.
Guérin, E., Fortier, M. S., & Sweet, S. N. (2013). An experience sampling study of physical
activity and positive affect: investigating the role of situational motivation and perceived
intensity across time. Health Psychology Research, 1(2), e21.
Gunes, H., Piccardi, M., & Pantic, M. (2008). From the Lab to the real world: Affect recognition
using multiple cues and modalities. In J. Or (Ed.). Affective computing: Focus on emotion
expression, synthesis, and recognition (pp. 185-218). Vienna, Austria: I-Tech Education
and Publishing.
Green, A. S., Rafaeli, E., Bolger, N., Shrout, P. E., & Reis, H. T. (2006). Paper or plastic? Data
equivalence in paper and electronic diaries. Psychological Methods, 11(1), 87-105.

111
Hall, E. E., Ekkekakis, P., & Petruzzello, S. J. (2002). The affective beneficence of vigorous
exercise revisited. British Journal of Health Psychology, 7(1), 47-66.
Hallal, P. C., Andersen, L. B., Bull, F. C., Guthold, R., Haskell, W., & Ekelund, U. (2012).
Global physical activity levels: surveillance progress, pitfalls, and prospects. The Lancet,
380(9838), 247-257.
Haskell, W. L., Blair, S. N., & Hill, J. O. (2009). Physical activity: health outcomes and
importance for public health policy. Preventive Medicine, 49(4), 280-282.
Healy, G. N., Dunstan, D. W., Salmon, J., Cerin, E., Shaw, J. E., Zimmet, P. Z., & Owen, N.
(2008). Breaks in Sedentary Time Beneficial associations with metabolic risk. Diabetes
Care, 31(4), 661-666.
Heath, G. W., Parra, D. C., Sarmiento, O. L., Andersen, L. B., Owen, N., Goenka, S., ... &
Brownson, R. C. (2012). Evidence-based intervention in physical activity: lessons from
around the world. The Lancet, 380(9838), 272-281.
Hedges, L. V., & Rhoads, C. (2010). Statistical power analysis in education research. (NCSER
2010-3006). Washington, DC: National Center for Special Education Research, Institute
of Education Sciences, U.S. Department of Education.
Hoffmann, P. (1997). The endorphin hypothesis. In W. P. Morgan (Ed.), Physical activity and
mental health (pp. 163-178). Washington, DC: Taylor & Francis.
Horemans, H., Bussmann, J., Beelen, A., Stam, H., & Nollet, F. (2005). Walking in
postpoliomyelitis syndrome: the relationships between time-scored tests, walking in daily
life and perceived mobility problems. Journal of Rehabilitation Medicine, 37(3), 142-
146.

112
Jacobs, N., Myin-Germeys, I., Derom, C., Delespaul, P., Van Os, J., & Nicolson, N. A. (2007). A
momentary assessment study of the relationship between affective and adrenocortical
stress responses in daily life. Biological Psychology, 74(1), 60-66.
Kahneman, D., Fredrickson, B. L., Schreiber, C. A., & Redelmeier, D. A. (1993). When more
pain is preferred to less: Adding a better end. Psychological Science, 4(6), 401-405.
Kahneman, D., Wakker, P. P., & Sarin, R. (1997). Back to Bentham? Explorations of
experienced utility. The Quarterly Journal of Economics, 112(2), 375-406.
Kanning, M. (2012). Using objective, real-time measures to investigate the effect of actual
physical activity on affective states in everyday life differentiating the contexts of
working and leisure time in a sample with students. Frontiers in Psychology, 3, 602.
Kanning, M., Ebner-Priemer, U., & Brand, R. (2012b). Autonomous regulation mode moderates
the effect of actual physical activity on affective states: An ambulant assessment approach to
the role of self-determination. Journal of Sport & Exercise Psychology, 34(2), 260-269.
Kenny, D. A. (1995). The effect of nonindependence on significance testing in dyadic research.
Personal Relationships, 2(1), 67-75.
Kenny, D. A., Kashy, D. A., & Bolger, N. (1998). Data analysis in social psychology. In D.
Gilbert, S. Fiske, and G. Lindzey (Eds.), Handbook of social psychology (4th ed., pp.
233-265). New York: McGraw-Hill.
Kerr, J. H., & Kuk, G. (2001). The effects of low and high intensity exercise on emotions, stress
and effort. Psychology of Sport and Exercise, 2(3), 173-186.
Kincaid, C. (2005, April). Guidelines for selecting the covariance structure in mixed model
analysis. In Proceedings of the Thirtieth Annual SAS Users Group International
Conference (No. 198-30). Cary, NC: SAS Institute Inc.

113
Kini, S. (2013). Please take my survey: Compliance with smartphone-based EMA/ESM studies
(Dartmouth Computer Science Technical Report TR2013-734). Hanover, NH:
Department of Computer Science, Dartmouth College.
Kirsch, I. (1997). Response expectancy theory and application: A decennial review. Applied and
Preventive Psychology, 6(2), 69-79.
Klonoff, E. A., Annechild, A., & Landrine, H. (1994). Predicting exercise adherence in women:
the role of psychological and physiological factors. Preventive Medicine, 23(2), 257-262.
Kochanska, G. (1997). Mutually responsive orientation between mothers and their young
children: Implications for early socialization. Child Development, 68(1), 94-112.
Kochanska, G., & Murray, K. T. (2000). Mother–child mutually responsive orientation and
conscience development: From toddler to early school age. Child Development, 71(2),
417-431.
Koltyn, K. F. (1997). The thermogenic hypothesis. In W. P. Morgan (Ed.), Physical activity and
mental health (pp. 213-226). Washington, DC: Taylor & Francis.
Krull, J. L., & MacKinnon, D. P. (1999). Multilevel mediation modeling in group-based
intervention studies. Evaluation Review, 23(4), 418-444.
Larson, R. W. (1990). The solitary side of life: An examination of the time people spend alone
from childhood to old age. Developmental Review, 10(2), 155-183.
Lazarus, R. S. (1991). Progress on a cognitive-motivational-relational theory of emotion.
American Psychologist, 46(8), 819-834.
Lee, I. M., Shiroma, E. J., Lobelo, F., Puska, P., Blair, S. N., & Katzmarzyk, P. T. (2012). Effect
of physical inactivity on major non-communicable diseases worldwide: an analysis of
burden of disease and life expectancy. Lancet, 380(9838), 219-229.

114
Liao, Y., Tate, E., & Dunton, G. F. (unpublished). Bi-directional acute relationships between
physical activity and affective states in daily life: A systematic review of evidence.
Liu, C., Cao, D., Chen, P., Zagar, T., & Lilly, E. (2007). RANDOM and REPEATED statements
- How to use them to model the covariance structure in proc mixed. In SAS Conference
Proceedings: Midwest SAS User Group.
Loehr, V. G., Baldwin, A. S., Rosenfield, D., & Smits, J. A. (2014). Weekly variability in
outcome expectations: Examining associations with related physical activity experiences
during physical activity initiation. Journal of Health Psychology, 19(10), 1309-1319.
Lohman, B. J., Stewart, S., Gundersen, C., Garasky, S., & Eisenmann, J. C. (2009). Adolescent
overweight and obesity: links to food insecurity and individual, maternal, and family
stressors. Journal of Adolescent Health, 45(3), 230-237.
Loprinzi, P. D., & Cardinal, B. J. (2013). Association between biologic outcomes and objectively
measured physical activity accumulated in≥ 10-minute bouts and< 10-minute bouts.
American Journal of Health Promotion, 27(3), 143-151.
Lyubomirsky, S., King, L., & Diener, E. (2005). The benefits of frequent positive affect: Does
happiness lead to success?. Psychological Bulletin, 131(6), 803.
Mata, J., Thompson, R. J., Jaeggi, S. M., Buschkuehl, M., Jonides, J., & Gotlib, I. H. (2012).
Walk on the bright side: Physical activity and affect in major depressive disorder. Journal
of Abnormal Psychology, 121(2), 297-308.
McAuley, E., Mihalko, S. L., & Bane, S. M. (1996). Acute exercise and anxiety reduction: Does
the environment matter?. Journal of Sport and Exercise Psychology, 18, 408-419.
Mead, G. E., Morley, W., Campbell, P., Greig, C. A., McMurdo, M., & Lawlor, D. A. (2009).
Exercise for depression. Cochrane Database Systematic Review, 3.

115
Mellers, B. A. (2000). Choice and the relative pleasure of consequences. Psychological Bulletin,
126(6), 910-924.
Mitchell, T. R., & James, L. R. (2001). Building better theory: Time and the specification of
when things happen. Academy of Management Review, 26(4), 530-547.
Morgan, W. P. (1985). Affective beneficence of vigorous physical activity. Medicine & Science
in Sports & Exercise, 17(1), 94-100.
Mouchacca, J., Abbott, G. R., & Ball, K. (2013). Associations between psychological stress,
eating, physical activity, sedentary behaviours and body weight among women: a
longitudinal study. BMC Public Health, 13(1), 828.
Mundt, J. C., Perrine, M. W., Searles, J. S., & Walter, D. (1995). An application of interactive
voice response (IVR) technology to longitudinal studies of daily behavior. Behavior
Research Methods, Instruments, & Computers, 27(3), 351-357.
Muthén, B. (2011). Applications of causally defined direct and indirect effects in mediation
analysis using SEM in Mplus. Retrieved from
http://www.statmodel2.com/download/causalmediation.pdf
Muthén, B., & Asparouhov, T. (2008). Growth mixture modeling: Analysis with non-Gaussian
random effects. In G. Fitzmaurice, M. Davidian, G. Verbeke, & G. Molenberghs (Eds.),
Longitudinal data analysis (pp. 143-165). Boca Raton, FL: Chapman & Hall.
National Physical Activity Plan Alliance. (2014). 2014 United States report card on physical
activity for children and youth. Columbia SC.
Naqvi, N., Shiv, B., & Bechara, A. (2006). The role of emotion in decision making: A cognitive
neuroscience perspective. Current Directions in Psychological Science, 15(5), 260-264.

116
Nguyen-Michel, S. T., Unger, J. B., Hamilton, J., & Spruijt-Metz, D. (2006). Associations
between physical activity and perceived stress/hassles in college students. Stress and
Health, 22(3), 179-188.
North, T. C., McCullagh, P., & Tran, Z. V. (1990). Effect of exercise on depression. Exercise &
Sport Sciences Reviews, 18(1), 379-416.
Norris, R., Carroll, D., & Cochrane, R. (1990). The effects of aerobic and anaerobic training on
fitness, blood pressure, and psychological stress and well-being. Journal of
Psychosomatic Research, 34(4), 367-375.
Oliver, M., Schofield, G. M., & Schluter, P. J. (2010). Parent influences on preschoolers’
objectively assessed physical activity. Journal of Science and Medicine in Sport, 13(4),
403-409.
Ojanen, M. (1994). Can the true effects of exercise on psychological variables be separated from
placebo effects?. International Journal of Sport Psychology, 25(1), 63-80.
Ortony, A., Clore, G. L., & Foss, M. A. (1987). The referential structure of the affective lexicon.
Cognitive Science, 11(3), 341-364.
Paluska, S. A., & Schwenk, T. L. (2000). Physical activity and mental health. Sports Medicine,
29(3), 167-180.
Park, H., & Walton-Moss, B. (2012). Parenting style, parenting stress, and children's health-
related behaviors. Journal of Developmental & Behavioral Pediatrics, 33(6), 495-503.
Parks, E. P., Kumanyika, S., Moore, R. H., Stettler, N., Wrotniak, B. H., & Kazak, A. (2012).
Influence of stress in parents on child obesity and related behaviors. Pediatrics, 130(5),
e1096-e1104.

117
Petruzzello, S. J., Hall, E. E., & Ekkekakis, P. (2001). Regional brain activation as a biological
marker of affective responsivity to acute exercise: influence of fitness. Psychophysiology,
38(01), 99-106.
Petruzzello, S. J., Jones, A. C., & Tate, A. K. (1997). Affective responses to acute exercise: a test
of opponent-process theory. The Journal of Sports Medicine and Physical Fitness, 37(3),
205-212.
Petruzzello, S. J., Landers, D. M., Hatfield, B. D., Kubitz, K. A., & Salazar, W. (1991). A meta-
analysis on the anxiety-reducing effects of acute and chronic exercise. Sports medicine,
11(3), 143-182.
Piasecki, T. M., Hufford, M. R., Solhan, M., & Trull, T. J. (2007). Assessing clients in their
natural environments with electronic diaries: rationale, benefits, limitations, and barriers.
Psychological Assessment, 19(1), 25-43.
Preacher, K. J., Zyphur, M. J., & Zhang, Z. (2010). A general multilevel SEM framework for
assessing multilevel mediation. Psychological Methods, 15(3), 209.
Pressman, S. D., & Cohen, S. (2005). Does positive affect influence health?. Psychological
Bulletin, 131(6), 925.
Prince, S. A., Adamo, K. B., Hamel, M. E., Hardt, J., Gorber, S. C., & Tremblay, M. (2008). A
comparison of direct versus self-report measures for assessing physical activity in adults:
a systematic review. International Journal of Behavioral Nutrition and Physical Activity,
5(1), 56.
Pronk, N. P., Crouse, S. F., & Rohack, J. J. (1995). Maximal exercise and acute mood response
in women. Physiology and Behavior, 57(1), 1-4.

118
Puetz, T. W. (2006). Physical activity and feelings of energy and fatigue. Sports Medicine, 36(9),
767-780.
Rahe, R. H., Rubin, R. T., Gunderson, E. K. (1972). Measures of subjects’ motivation and affect
correlated with their serum uric acid, cholesterol and cortisol. Archives of General
Psychiatry, 26(4), 357-359.
Ransford, C. P. (1981). A role for amines in the antidepressant effect of exercise: A review.
Medicine & Science in Sports & Exercise, 14(1), 1-10.
Redelmeier, D. A., & Kahneman, D. (1996). Patients' memories of painful medical treatments:
Real-time and retrospective evaluations of two minimally invasive procedures. Pain,
66(1), 3-8.
Reed, J., & Ones, D. S. (2006). The effect of acute aerobic exercise on positive activated affect:
A meta-analysis. Psychology of Sport and Exercise, 7(5), 477-514.
Reis, H. T., Collins, W. A., & Berscheid, E. (2000). The relationship context of human behavior
and development. Psychological Bulletin, 126(6), 844-872.
Rethorst, C. D., Wipfli, B. M., & Landers, D. M. (2009). The antidepressive effects of exercise:
A meta-analysis of randomized trials. Sports Medicine, 39(6), 491-511.
Rhodes, R. E., & Nigg, C. R. (2011). Advancing physical activity theory: a review and future
directions. Exercise and sport sciences reviews, 39(3), 113-119.
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?. Translational Behavioral Medicine, 1(1), 53-71.

119
Robbins, L. B., Pis, M. B., Pender, N. J., & Kazanis, A. S. (2004). Exercise self-efficacy,
enjoyment, and feeling states among adolescents. Western Journal of Nursing Research,
26(7), 699-715.
Rucker, D. D., Preacher, K. J., Tormala, Z. L., & Petty, R. E. (2011). Mediation analysis in
social psychology: Current practices and new recommendations. Social and Personality
Psychology Compass, 5(6), 359-371.
Ruhé, H. G., Mason, N. S., & Schene, A. H. (2007). Mood is indirectly related to serotonin,
norepinephrine and dopamine levels in humans: a meta-analysis of monoamine depletion
studies. Molecular Psychiatry, 12(4), 331-359.
Runyan, J. D., Steenbergh, T. A., Bainbridge, C., Daugherty, D. A., Oke, L., & Fry, B. N.
(2013). A smartphone ecological momentary assessment/intervention “app” for collecting
real-time data and promoting self-awareness. PloS One, 8(8), e71325.
Russell, J. A. (1980). A circumplex model of affect. Journal of Personality and Social
Psychology, 39(6), 1161-1178.
Sallis, J. F., & Nader, P. R. (1998). Family determinants of health behaviors. In D. S. Gochman
(Ed), Health behavior: Emerging research perspectives (pp. 107-124). New York:
Springer.
Sallis, J. F., Owen, N., & Fisher, E. B. (2008). Ecological models of health behavior. In K. Glanz,
B. K. Rimer, K. Viswanath (Eds), Health behavior and health education: Theory,
research, and practice (pp. 465-482). San Francisco: Jossey-Bass.
Salovey, P., Rothman, A. J., Detweiler, J. B., & Steward, W. T. (2000). Emotional states and
physical health. American Psychologist, 55(1), 110.

120
Schneider, M., Dunn, A. L., & Cooper, D. (2009). Affective, Exercise and Physical Activity
among Healthy Adolescents. Journal of Sport & Exercise Psychology, 31(6), 706.
Schwarz, N. (1990). Feelings as information: Informational and motivational functions of
affective states. In E. T. Higgins & R. M. Sorrentino (Eds.), Handbook of motivation and
cognition: Foundations of social behavior, (vol. 2, pp. 527-561). New York, NY:
Guilford Press.
Schwarz, N. (2007). Retrospective and concurrent self-reports: The rationale for real-time data
capture. In A. A. Stone, S. S. Shiffman, A. Atienza, & L. Nebeling (Eds.), The science of
real-time data capture: Self-reports in health research, (pp. 11-26). New York: Oxford
University Press.
Schwarz, N., & Clore, G. L. (2007). Feelings and phenomenal experiences. In A. Kruglanski &
E. T. Higgins (Eds.), Social psychology: Handbook of basic principles (2nd ed.; pp. 385-
407). New York: Guilford.
Schwerdtfeger, A., Eberhardt, R., Chmitorz, A., & Schaller, E. (2010). Momentary affect
predicts bodily movement in daily life: An ambulatory monitoring study. Journal of
Sport & Exercise Psychology, 32(5), 674-693.
Seo, M. G., Barrett, L. F., & Bartunek, J. M. (2004). The role of affective experience in work
motivation. Academy of Management Review, 29(3), 423-439.
Sheppard, K. E., & Parfitt, G. (2008). Acute affective responses to prescribed and self-selected
exercise intensities in young adolescent boys and girls. Pediatric Exercise Science, 20(2),
129.
Sherwood, N. E., & Jeffery, R. W. (2000). The behavioral determinants of exercise: implications
for physical activity interventions. Annual Review of Nutrition, 20(1), 21-44.

121
Shiffman, S., Stone, A. A., & Hufford, M. R. (2008). Ecological momentary assessment. Annual
Review of Clinical Psychology, 4, 1-32.
Singer, J. D. (1998). Using SAS PROC MIXED to fit multilevel models, hierarchical models,
and individual growth models. Journal of Educational and Behavioral Statistics, 23(4),
323-355.
Smith, A. (2015). U.S. Smartphone Use in 2015. Retrieved from
http://www.pewinternet.org/2015/04/01/us-smartphone-use-in-2015/
Solomon, R. L. (1980). The opponent-process theory of acquired motivation: the costs of
pleasure and the benefits of pain. American Psychologist, 35(8), 691-712.
Steptoe, A. S., & Butler, N. (1996). Sports participation and emotional wellbeing in adolescents.
The Lancet, 347(9018), 1789-1792.
Steptoe, A., & Cox, S. (1988). Acute effects of aerobic exercise on mood. Health Psychology,
7(4), 329.
Steptoe, A., Kearsley, N., & Walters, N. (1993). Acute mood responses to maximal and
submaximal exercise in active and inactive men. Psychology and Health, 8(1), 89-99.
Stone, A. A., & Broderick, J. E. (2007). Real‐ time data collection for pain: Appraisal and current
status. Pain Medicine, 8(S3), S85-S93.
Stone, A. A., & Shiffman, S. (1994). Ecological momentary assessment (EMA) in behavorial
medicine. Annals of Behavioral Medicine, 16(3), 199-202.
Thompson Coon, J., Boddy, K., Stein, K., Whear, R., Barton, J., & Depledge, M. H. (2011).
Does participating in physical activity in outdoor natural environments have a greater
effect on physical and mental wellbeing than physical activity indoors? A systematic
review. Environmental Science & Technology, 45(5), 1761-1772.

122
Troiano, R. P., Berrigan, D., Dodd, K. W., Mâsse, L. C., Tilert, T., & McDowell, M. (2008).
Physical activity in the United States measured by accelerometer. Medicine & Science in
Sports & Exercise, 40(1), 181-188.
Trost, S. G., Owen, N., Bauman, A. E., Sallis, J. F., & Brown, W. (2002). Correlates of adults'
participation in physical activity: review and update. Medicine & Science in Sports &
Exercise, 34(12), 1996-2001.
Uemura, H., Katsuura-Kamano, S., Yamaguchi, M., Nakamoto, M., Hiyoshi, M., & Arisawa, K.
(2013). Abundant daily non-sedentary activity is associated with reduced prevalence of
metabolic syndrome and insulin resistance. Journal of Endocrinological Investigation,
36(11), 1069-1075.
U.S. Department of Health and Human Services. (2013). Healthy People 2020: Physical Activity.
Retrieved from
http://www.healthypeople.gov/2020/topicsobjectives2020/overview.aspx?topicid=33
van Eck, M., Nicolson, N. A., & Berkhof, J. (1998). Effects of stressful daily events on mood
states: Relationship to global perceived stress. Journal of Personality and Social
Psychology, 75(6), 1572-1585.
Vilhjalmsson, R., & Thorlindsson, T. (1992). The integrative and physiological effects of sport
participation: A study of adolescents. The Sociological Quarterly, 33(4), 637-647.
von Haaren, B., Loeffler, S. N., Haertel, S., Anastasopoulou, P., Stumpp, J., Hey, S., & Boes, K.
(2013). Characteristics of the activity-affect association in inactive people: an ambulatory
assessment study in daily life. Frontiers in Psychology, 4, 163.

123
Wagner, A., Klein-Platat, C., Arveiler, D., Haan, M. C., Schlienger, J. L., & Simon, C. (2004).
Parent-child physical activity relationships in 12-year old French students do not depend
on family socioeconomic status. Diabetes & Metabolism, 30(4), 359-366.
Warburton, D., Charlesworth, S., Ivey, A., Nettlefold, L., & Bredin, S. S. (2010). A systematic
review of the evidence for Canada’s Physical Activity Guidelines for Adults.
International Journal of Behavioral Nutrition and Physical Activity, 7(1), 39.
Watson, D., & Clark, L. A. (1992). Affects separable and inseparable: On the hierarchical
arrangement of the negative affects. Journal of Personality and Social Psychology, 62(3),
489.
Watson, D., & Tellegen, A. (1985). Toward a consensual structure of mood. Psychological
Bulletin, 98(2), 219-235.
Welch, A. S., Hulley, A., Ferguson, C., & Beauchamp, M. R. (2007). Affective responses of
inactive women to a maximal incremental exercise test: A test of the dual-mode model.
Psychology of Sport and Exercise, 8(4), 401-423.
Welk, G. J., Wood, K., & Morss, G. (2003). Parental influences on physical activity in children:
An exploration of potential mechanisms. Pediatric Exercise Science, 15, 19-33.
Wichers, M., Peeters, F., Rutten, B. P. F., Jacobs, N., Derom, C., Thiery, E., … van Os, J.
(2012). A time-lagged momentary assessment study on daily life physical activity and
affect. Health Psychology, 31(2), 135-144.
Wilhelm, F. H., & Grossman, P. (2010). Emotions beyond the laboratory: Theoretical
fundaments, study design, and analytic strategies for advanced ambulatory assessment.
Biological Psychology, 84(3), 552-569.

124
Williams, D. M. (2008). Exercise, affect, and adherence: An integrated model and a case for self-
paced exercise. Journal of Sport & Exercise Psychology, 30(5), 471.
Williams, D. M., Dunsiger, S., Ciccolo, J. T., Lewis, B. A., Albrecht, A. E., & Marcus, B. H.
(2008). Acute affective response to a moderate-intensity exercise stimulus predicts
physical activity participation 6 and 12 months later. Psychology of Sport and Exercise,
9(3), 231-245.
Wipfli, B. M., Rethorst, C. D., & Landers, D. M. (2008). The anxiolytic effects of exercise: A
meta-analysis of randomized trials and dose-response analysis. Journal of Sport &
Exercise Psychology, 30(4), 392-410.
Yeung, R. R. (1996). The acute effects of exercise on mood state. Journal of Psychosomatic
Research, 40(2), 123-141.
Yoon, S., Buckworth, J., Focht, B., & Ko, B. (2013). Feelings of energy, exercise-related self-
efficacy, and voluntary exercise participation. Journal of Sport & Exercise Psychology,
35(6), 612-624.
Zhang, Z., Zyphur, M. J., & Preacher, K. J. (2009). Testing multilevel mediation using
hierarchical linear models problems and solutions. Organizational Research Methods,
12(4), 695-719.

125
Appendix

Figure A1. Power plot for affective states predicting subsequent physical activity ( Study 1 research
question 1).

Figure A2. Power plot for physical activity predicting subsequent affective states (Study 1 research
question 2).

Abstract (if available)

Abstract This dissertation examined the relationships between physical activity and affective states using real‐time data capture techniques. Specifically, the (1) acute effects (i.e., the bi‐directional relationships at the moment‐to‐moment level), (2) longitudinal effects (i.e., how the affective responses during physical activity might predict future physical activity behavior), and (3) dyadic effects (e.g., whether mothers’ affective states may influence their children’s subsequent affective states and physical activity levels) were tested and explored using data collected from mobile phone apps and accelerometers. The unique characteristics of real‐time data capture methods allow researchers to minimize participants’ recall biases and improve a study’s external and ecological validity. Results from this dissertation study show that a more positive affective state was associated with more physical activity both short‐term and long‐term. Further, engaging in more physical activity led to an immediate improvement in physical feeling state. However, this study did not find any significant relationship between one person’s (i.e., mothers) affective states and another person’s (i.e., children) subsequent physical activity levels. Overall, this dissertation study demonstrated the use of real‐time data to examine the relationships between physical activity and affective states in free‐living settings. Findings from this study could offer directions for future studies (e.g., explore potential moderators and mediators under a ecological framework in free‐living environments) and insights for intervention development (e.g., target negative affective feelings as a barrier for engaging in daily physical activity).

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Understanding the methodological limitations In the ecological momentary assessment of physical activity

PDF

The acute relationship between affective states and physiological stress response, and the moderating role of moderate-to-vigorous physical activity

PDF

The acute and longitudinal associations between sedentary behaviors, affective states, and emotional disorder symptoms among youth

PDF

Sleep health and variability in youth: a real-time data capture study to examine influences on daily dietary intake patterns and longitudinal weight trajectories

PDF

Developing and testing novel strategies to detect inattentive responding in ecological momentary assessment studies

PDF

Effects of the perceived and objectively assessed environment on physical activity in adults and children

PDF

Affect, digital media use, physical activity, and ADHD in youth

PDF

Using a structural model of psychopathology to distinguish relations between shared and specific features of psychopathology, smoking, and underlying mechanisms

PDF

Effects of parent stress on weight-related parenting practices and child obesity risk

PDF

Influences of specific environmental domains on childhood obesity and related behaviors

PDF

System dynamics of in-home family eating behavior: insights from intensive longitudinal data using Ecological Momentary Assessment and wearable sensors

PDF

The Internet activities, gratifications, and health consequences related to compulsive Internet use

PDF

Motivation and the meanings of health behavior as factors associated with eating behavior in Latino youth

PDF

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

PDF

Addressing federal pain research priorities: drug policy, pain mechanisms, and integrative treatment

PDF

Cultural risk and protective factors for tobacco use behaviors and depressive symptoms among American Indian adolescents in California

PDF

Energy drink consumption, substance use and attention-deficit/hyperactivity disorder among adolescents

PDF

Effects of near-roadway air pollution exposure on obesity, obesity-related behavior, and neurobehavioral deficits during peripuberty

PDF

Problematic alcohol use in Hispanic emerging adults: the role of perceived discrimination, cultural identity, and salient cultural values

PDF

Investigating daily effects of activity space-based built environment exposures on physical activity behaviors in Hispanic women during pregnancy and early postpartum

Asset Metadata

Creator Liao, Yue (author)

Core Title Understanding the dynamic relationships between physical activity and affective states using real-time data capture techniques

School Keck School of Medicine

Degree Doctor of Philosophy

Degree Program Preventive Medicine (Health Behavior)

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

Tag affect,ecological momentary assessment,free‐living,intensive longitudinal data,OAI-PMH Harvest,physical activity

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Dunton, Genevieve F. (committee chair), Chou, Chih-Ping (committee member), Huh, Jimi (committee member), Leventhal, Adam M. (committee member)

Creator Email yueliao.usc@gmail.com,yueliao@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-588570

Unique identifier UC11302048

Identifier etd-LiaoYue-3567.pdf (filename),usctheses-c3-588570 (legacy record id)

Legacy Identifier etd-LiaoYue-3567.pdf

Dmrecord 588570

Document Type Dissertation

Format application/pdf (imt)

Rights Liao, Yue

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