Citation
Integrating meaning, purpose, and self-determination theory as predictors of physical activity maintenance

Material Information

Title:
Integrating meaning, purpose, and self-determination theory as predictors of physical activity maintenance
Creator:
Hooker, Stephanie Ann ( author )
Place of Publication:
Denver, Colo.
Publisher:
University of Colorado Denver
Publication Date:
Language:
English
Physical Description:
1 electronic file : ;

Thesis/Dissertation Information

Degree:
Doctorate ( Doctor of philosophy)
Degree Grantor:
University of Colorado Denver
Degree Divisions:
Department of Psychology, CU Denver
Degree Disciplines:
Clinical health psychology

Subjects

Subjects / Keywords:
Autonomy (Psychology) ( lcsh )
Exercise ( lcsh )
Autonomy (Psychology) ( fast )
Exercise ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references.
System Details:
System requirements: Adobe Reader.
Statement of Responsibility:
Stephanie Ann Hooker.

Record Information

Source Institution:
University of Colorado Denver Collections
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
982959962 ( OCLC )
ocn982959962
Classification:
LD1193.L645 2016d H66 ( lcc )

Downloads

This item has the following downloads:


Full Text
INTEGRATING MEANING, PURPOSE, AND SELF-DETERMINATION THEORY AS
PREDICTORS OF PHYSICAL ACTIVITY MAINTENANCE
by
STEPHANIE ANN HOOKER M.P.H., University of Colorado Denver, 2015 M.S., Syracuse University, 2011 B.A.S., University of Minnesota Duluth, 2009
A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Doctor of Philosophy Clinical Health Psychology Program
2016


2016
STEPHANIE ANN HOOKER ALL RIGHTS RESERVED
11


This thesis for the Doctor of Philosophy degree by Stephanie Ann Hooker has been approved for the Clinical Health Psychology Program by
Krista W. Ranby, Chair Kevin S. Masters, Advisor James Grigsby James O. Hill
Date: May 14, 2016
m


Hooker, Stephanie Ann (Ph.D., Clinical Health Psychology)
Integrating Meaning, Purpose, and Self-Determination Theory as Predictors of Physical Activity Maintenance
Dissertation directed by Professor Kevin S. Masters
ABSTRACT
Despite the widely known benefits of physical activity (PA), most adults are insufficiently active and have difficulty maintaining new exercise programs. Understanding factors related to PA in previously sedentary exercise initiates may improve interventions designed to increase consistent PA. One factor might be the extent to which individuals experience meaning in their lives on a daily basis. Individuals living lives with more experienced meaning, or meaning salience, may be more likely to engage in health behaviors including PA. This process is consistent with Self-Determination Theory (SDT), a theory of behavior regulation and motivation that states that more internally regulated behaviors are more likely to be maintained. This study examined processes (i.e., meaning in the context of SDT) that may be associated with PA maintenance. Previously sedentary exercise initiates (A=160; Mage=43.3 years, SD=l 1.4 years; 76.9% female) participated in a randomized controlled trial of a 4-week daily self-monitoring of meaning, mood, and PA condition compared to a random survey control as they began a self-initiated exercise program. Participants completed surveys at baseline, 4-weeks, and 12-weeks and fitness assessments at baseline and 12-weeks. Multilevel mixed models and path analytic methods were used to analyze data. Results revealed no significant differences between the self-monitoring and control groups on PA or M/P. Within day analyses revealed that greater daily meaning salience was significantly related to greater daily PA duration, /?=. 21,/K.0001, and intensity, P=. 2 l,p<.0001, and greater likelihood of visiting the fitness center, Odds Ratio=1.48 (95%
IV


CI=1.18,1.86),/)=.0008. Longitudinal path analytic models suggest that SDT variables accounted for a significant proportion of the variance in PA. Modified models revealed that baseline meaning significantly predicted greater change in PA from baseline to 4-weeks,
/?=. 20,/>=. 02. However, meaning was not related to change in PA from 4-weeks to 12-weeks. Results suggest that meaning salience plays an important role in PA participation in previously sedentary exercise initiates. Findings related to global ratings of meaning are mixed, suggesting that global ratings of meaning may not be as important as meaning salience. Future research should examine meaning salience over longer observation periods and meaning-based interventions in exercise initiates.
The form and content of this abstract are approved. I recommend its publication.
Approved: Kevin S. Masters
v


ACKNOWLEDGEMENTS
This research was funded by the American Heart Association Pre-doctoral Fellowship grant (14PRE18710033) and by the Psychology Department at the University of Colorado Denver. This project was supported by the NIH/NCRR Colorado CTSI Grant Number UL1 RR025780. Its contents are the authors sole responsibility and do not necessarily represent official NIH views.
I would like to thank Kaile Ross, Kaylae Nakamura, Jean Wood, and Emma Lyons for their assistance with data collection and study management. Additionally, I would like to thank Dr. James Hill and the staff at the Anschutz Health and Wellness Center for allowing me to recruit their members and providing the resources to conduct this study. Without their support, this study would not have been possible. I would like to thank Dr. Krista Ranby for her assistance with reviewing statistical models and suggestions for data analysis. I would also like to thank Dr. Kevin Masters for his wonderful mentorship and support over my graduate school career and for reading and providing feedback on countless drafts of manuscripts, proposals, grants, and essays. Finally, I would like to thank my husband, Tom Showalter, Jr., for his endless support throughout my undergraduate and graduate careers.
vi


TABLE OF CONTENTS
CHAPTER
I. INTRODUCTION............................................................1
Meaning and Purpose.....................................................2
Self-Determination Theory...............................................6
Meaning and Mood.......................................................12
Self-Monitoring of M/P, Mood, and PA as a Possible Intervention........13
Purpose of the Study...................................................14
II. METHOD.................................................................15
Overview...............................................................15
Participants and Recruitment...........................................15
Randomization..........................................................17
Procedure..............................................................18
Measures...............................................................20
Daily mood..........................................................20
Daily M/P salience..................................................22
PA..................................................................22
24-hour activity recall.............................................23
Psychological needs satisfaction....................................23
Behavioral regulations motivation...................................24
M/P.................................................................25
Meaning in life..................................................25
Purpose in life..................................................26
vii


PA........................................................................26
Psychological well-being..................................................27
Subjective vitality....................................................27
Life satisfaction......................................................27
Depressive symptoms.......................................................28
Physical fitness..........................................................29
M/P and PA connection.....................................................29
Power Analysis................................................................30
Aim 1. Effect M/P, mood, and PA self-monitoring on PA.....................30
Aim 2. Relationship of daily M/P salience to daily PA.....................30
Aim 3. Examining the SDT Process Model of PA adoption with and without M/P ..........................................................................30
Data Analysis.................................................................31
Aim 1. Effect of M/P, mood, and PA self-monitoring on PA.................32
Aim 2. Relationship of daily M/P salience to daily PA.....................33
Aim 3. Examining the SDT process model of PA adoption with and without M/P ..........................................................................33
Exploratory analyses......................................................34
III. RESULTS.....................................................................35
Study Flow & Missing Data Analysis............................................40
Aims 1 & 3: Baseline, 4-week, and 12-week missing data....................40
Aim 2 missing data........................................................41
Study Variables at Baseline, 4-weeks, and 12-weeks............................41
Aim 1. Effect of M/P, Mood, and PA Self-Monitoring on PA......................45
PA........................................................................45
viii


M/P..........................................................................48
SDT mediators................................................................49
Psychological well-being.....................................................51
Aim 2. Relationship of Daily M/P Salience to Daily PA............................54
Aim 3. Examining the SDT Process Model of PA Adoption with and without M/P ..56
Model la: Cross-sectional baseline model.....................................56
Model lb: Modified cross-sectional baseline model............................58
Model 2a: Cross-sectional 4-week model.......................................59
Model 2b: Modified cross-sectional 4-week model..............................60
Model 3a: Cross-sectional 12-week model......................................61
Model 3b: Modified cross-sectional 12-week model.............................62
Model 4a: Longitudinal model predicting absolute levels of PA at 12-weeks...63
Model 4b. Modified longitudinal model predicting absolute levels of PA at 12-weeks........................................................................64
Model 5a: Longitudinal model predicting change in PA at 4-weeks and 12-weeks .............................................................................64
Model 5b: Modified longitudinal model predicting change in PA at 4-weeks and 12-weeks.....................................................................66
Exploratory Outcomes.............................................................67
Average and variability of meaning salience in relation to PA and fitness outcomes.....................................................................67
Relationships between global M/P and PA and objective outcomes...............70
Relationships between M/P and PA connection, M/P, PA, and objective outcomes .............................................................................72
IV. DISCUSSION.....................................................................74
Aim 1. Effect of M/P, Mood, and PA Self-Monitoring on PA.........................74
Aim 2. Relationship of Daily M/P Salience to Daily PA............................77
ix


Aim 3. Examining the SDT Process Model of PA Adoption with and without M/P ..78
Strengths.................................................................85
Limitations...............................................................85
Future Directions.........................................................86
Conclusions...............................................................87
REFERENCES......................................................................88
APPENDIX
A. Daily Quotes for the Self-Monitoring Condition............................98
B. Screening and Eligibility Form...........................................100
C. Baseline-Only Measures...................................................103
D. Baseline, 4-week, and 12-week Measures...................................106
E. Baseline and 12-week Measures............................................114
F. Daily Measures...........................................................115
x


LIST OF TABLES
TABLE
1. Measures Collected in This Study..................................................21
2. Baseline Demographics for the Entire Sample and Between Groups....................37
3. Descriptive Statistics of Study Variables at Baseline, 4-weeks, and 12-weeks......43
4. Means and Standard Deviations of Study Variables by Group.........................47
5. Descriptive Statistics of Variables in Aim 2......................................55
6. Pearson Correlations among Variables in Aim 2.....................................55
7. Pearson Correlations among Aim 3 Model Variables..................................57
8. Model Fit Statistics..............................................................58
9. Pearson Correlations between the TOMS and Other Study Outcomes....................69
10. Regression Models with Meaning Salience Predicting 12-week Outcomes..............70
11. Pearson Correlations between Global M/P, PA, and Objective Outcomes..............71
12. Pearson Correlations between M/P and PA Connection, M/P, PA, and Objective
Outcomes.........................................................................73
xi


LIST OF FIGURES
FIGURE
1. The SDT Internalization Continuum..................................................7
2. Model Integrating SDT and M/P to Increase Activity................................11
3. Study Timeline....................................................................19
4. CONSORT Diagram...................................................................42
5. Trajectories of PA Over Time......................................................45
6. Mean PA by Condition from Baseline to 12-weeks....................................46
7. Mean Fitness Center Visits by Condition from Baseline to 12-weeks.................48
8. Mean Purpose in Life by Condition from Baseline to 12-weeks.......................49
9. Mean Basic Psychological Needs Satisfaction by Condition from Baseline to 12-weeks 50
10. Mean Autonomous Regulation by Condition from Baseline to 12-weeks................51
11. Mean Vitality by Condition from Baseline to 12-weeks.............................52
12. Mean Life Satisfaction by Condition from Baseline to 12-weeks....................53
13. Mean Depressive Symptoms by Condition from Baseline to 12-weeks..................54
14. Cross-sectional SDT Process Model of Behavior Change at Baseline.................58
15. Cross-sectional Modified SDT Process Model of Behavior Change at Baseline........59
16. Cross-sectional SDT Process Model of Behavior Change at 4-weeks..................60
17. Cross-sectional Modified SDT Process Model of Behavior Change at 4-weeks.........61
18. Cross-sectional SDT Process Model of Behavior Change at 12-weeks.................62
19. Cross-sectional Modified SDT Process Model of Behavior Change at 12-weeks........63
20. Longitudinal SDT Process Model of Behavior Change Predicting Absolute PA at 12-
weeks............................................................................64
21. Modified Longitudinal SDT Process Model of Behavior Change Predicting Absolute PA
at 12-weeks......................................................................65
xii


22. Longitudinal SDT Process Model of Behavior Change Predicting Change in PA at 4-
weeks and 12-weeks.............................................................66
23. Modified Longitudinal SDT Process Model of Behavior Change Predicting Change in PA
at 4-weeks and 12-weeks........................................................68
xiii


LIST OF ABBREVIATIONS
ACT Acceptance and Commitment Therapy
AHWC Anschutz Health and Wellness Center
BMI Body Mass Index
BPNES Basic Psychological Needs in Exercise Scale
BREQ-2 Behavioral Regulations in Exercise Questionnaire 2nd version
CVD Cardiovascular disease
COMIRB Colorado Multiple Institutional Review Board
DMS Daily Meaning Scale
EMA Ecological Momentary Assessment
IPAQ-SF International Physical Activity Questionnaire Short Form
LET Life Engagement Test
MET Metabolic Equivalent of Task
M/P Meaning and purpose
MILQ Meaning in Life Questionnaire
PA Physical activity
PAR-Q Physical Activity Readiness Questionnaire
PANAS Positive and Negative Affect Schedule
REDCap Research Electronic Data Capture
SDT Self-Determination Theory
svs Subjective Vitality Scale
SWLS Satisfaction with Life Scale
TOMS Thoughts of Meaning Scale
TTM Transtheoretical Model
UCD University of Colorado Denver


CHAPTER I
INTRODUCTION
Approximately 2400 US adults die from cardiovascular disease (CVD) every day (Lloyd-Jones et al., 2009). Given the significant burden of CVD, primary prevention of CVD is essential to improve health and well-being and to reduce costs associated with chronic illness (Probst-Hensch, Tanner, Kessler, Burri, & Kiinzli, 2011). Regular physical activity (PA) is known to reduce the risk of cardiovascular disease and stroke (Physical Activity Guidelines Advisory Committee, 2008). Sustained long-term PA reduces risk of incident CVD in men and CVD-related and all-cause mortality in both men and women (Shortreed, Peeters, & Forbes, 2013). Individuals who meet the national guidelines for PA (>150 minutes of moderate-to-vigorous PA per week; Physical Activity Guidelines Committee, 2008) also have significantly better CVD risk factor profiles than those who do not meet national guidelines (Glazer et al., 2013). Because the evidence that PA is beneficial for cardiovascular health is well known, the American Heart Association has listed meeting national guidelines for PA as one of the metrics for good cardiovascular health (Lloyd-Jones et al., 2010).
Despite the benefits of engaging in PA for cardiovascular health, a vast majority of US adults do not engage in regular PA. In fact, estimates from national self-report surveys indicate that only 45% to 65% of the population meets guidelines for PA (Adabonyan, Loustalot, Kruger, Carlson, & Fulton, 2007; Macera et al., 2005) but when measured via objective accelerometer assessment a mere 5% meet the guidelines (Troiano et al., 2008). Interventions to increase PA generally demonstrate short-term success but fail to show longterm maintenance (Marcus, Owen, Forsyth, Cavill, & Fridinger, 1998; Marcus et al., 2000).
1


Identifying and utilizing psychological and behavioral processes that enhance long-term maintenance of PA is desperately needed.
Meaning and Purpose
One psychological process likely to be related to long-term maintenance of PA is a sense of personal meaning (i.e., feeling that ones existence is significant) and purpose (i.e., pursuit and attainment of behavioral goals consistent with ones subjective values and life goals) (M/P; Reker, Peacock, & Wong, 1987; Ryff & Singer, 1998; Steger, Frazier, Oishi, & Kaler, 2006). The sense of meaning in life is considered a multidimensional construct with (1) affective (feelings of fulfillment or satisfaction that come from the conviction that life is worth living); (2) cognitive (making sense of and finding value and purpose in life events, circumstances, or encounters); and (3) motivational (the pursuit and attainment of personal goals that are consistent with ones subjective values, needs and wants) components (Reker, 2000; Wong, 1989). Although meaning and purpose are often used interchangeably in the literature, purpose is considered the motivational component of meaning that stimulates goals and influences behavior (McKnight & Kashdan, 2009). Several studies have demonstrated that M/P is positively related to many different dimensions of health and well-being, including increased longevity (Boyle, Barnes, Buchman, & Bennett, 2009; Cohen, Bavishi,
& Rozanski, 2016; Krause, 2009), reduced risk for cardiovascular events (Cohen et al.,
2016), improved physical health (Pinquart, 2002), better self-rated health and functioning (Holahan, Holahan, & Suzuki, 2008; Krause, 2009; Krause & Shaw, 2003; OConnor & Vallerand, 1998; Takkinen, Suutama, & Ruoppila, 2001), greater positive affect (Hicks & King, 2008, 2009; King, Hicks, Krull, & Del Gaiso, 2006), and better psychological and
2


physical well-being (Reker et al., 1987; Scheier et al., 2006; Steger et al., 2006). Thus, through some as yet unidentified mechanisms, M/P is associated with health and well-being.
A potential mechanism linking M/P to better health and well-being is greater engagement in healthy behaviors, and specifically, PA. Ryff and Singer (1998) suggest that individuals who have a sense of purpose in their lives may be more likely to practice health-promoting behaviors, and that taking good care of oneself in terms of daily health practices presupposes a life that is worth taking care of (p. 22). This postulates that individuals who have a lesser awareness of M/P in their lives may not practice health-promoting behaviors, at least in part, because they do not see the value of effortfully supporting a life lacking M/P.
On the other hand, if individuals live with awareness of personal M/P, this may motivate them to engage in healthier behaviors.
Scheier and colleagues (2006) argue that purpose in life is a part of behavioral self-regulation, as it provides the why to engaging in certain behaviors. When individuals live with daily awareness of what makes their lives meaningful and gives them purpose, they may be able to make routine choices that support their M/P. Adults are frequently faced with choices that can either be healthy (e.g., get up early to exercise before going to work) or hedonically pleasurable (e.g., skipping the workout because I do not feel like it). Making the easier choice (i.e., skipping the workout) is more likely when individuals do not live with awareness of their personal M/P. Choosing the more difficult behavior that is congruent with ones personal values, goals, and purpose is more likely when one is aware, in the moment, of those values, goals, and purpose. Those who connect engaging in healthy behaviors, such as PA, with their sense of M/P may be more likely to engage in those behaviors on a daily basis. Thus, M/P can be motivational and contribute to self-regulation of healthy behavior.
3


Moreover, when individuals have a strong sense of M/P they likely also have several longterm life goals. Without good health, individuals are likely to have difficulty reaching those goals. Thus, engagement in healthy behaviors, and especially in PA, can be seen as a secondary goal and value in support of reaching ones ultimate life purposes.
Cross-sectional research on M/P supports this claim and observational findings show that greater M/P is related to greater engagement in PA (Holahan et al., 2008; Holahan & Suzuki, 2006; Homan & Boyatzis, 2010; Hooker & Masters, 2014; Ruuskanen & Ruoppila, 1995; Takkinen et al., 2001). Several of these studies show that a global sense of personal M/P (e.g., I have a good sense of what makes my life meaningful; Steger et al., 2006) is related to self-reported PA. However, one recent study examined purpose in relation to objectively-measured (with accelerometers) PA (Hooker & Masters, 2014). This study demonstrated that after controlling for several demographic (age, gender, socioeconomic status, race/ethnicity, and marital status), affective (depressive symptoms and positive affect), and cognitive (optimism and self-mastery) confounds, purpose remained a robust predictor of objectively-measured lifestyle movement, and to a lesser extent, moderate-to-vigorous exercise. These results support the hypothesis that awareness of personal M/P is related to greater engagement in PA. Furthermore, it is hypothesized that behaviors and goals (i.e., PA and healthy lifestyles) that are explicitly integrated with ones life meaning and purpose are more likely to be maintained.
Indeed, some interventions have been designed to explicitly connect what is valuable or meaningful to a person with their behavior. Notably, acceptance and commitment therapy (ACT; Hayes, Strosahl, & Wilson, 1999) proposes that commitment to value-directed activities is important for behavior change. In one pilot study, an ACT intervention designed
4


to increase value-directed behaviors (as well as other acceptance-based strategies such as willingness to experience distress) and PA was more effective than an education-based control at increasing visits to a fitness center over 8 weeks (Butryn, Forman, Hoffman, Shaw, & Juarascio, 2011). However, it is not clear how much the results were driven by value-directed behaviors versus other acceptance-based approaches. Another approach, taken from the Disconnected Values Model (Anshel, 2010), suggests that individuals are more likely to engage in healthy behaviors when they realize there is a discrepancy (or disconnect) between what they value (e.g., health, family) and how they behave (e.g., lack of exercise). Anshel and colleagues (Anshel, Brinthaupt, & Kang, 2010; Anshel, Kang, & Brinthaupt, 2011) have tested their 10-week intervention (including an orientation based on the Disconnected Values Model, an exercise prescription, and weekly personal training) in university employees and have shown that the intervention increases physical fitness and well-being. However, both tests of the intervention did not have a control group, and the extent to which the intervention results are due to the theoretical model rather than the structured personal training components of the program is not clear. Whereas these interventions provide evidence that connecting what is meaningful or valuable to a person to their health behaviors may improve engagement in those behaviors, none of these interventions have explicitly examined the proposed theoretical mediators (e.g., value-directed behavior, values connection) to their outcomes. Moreover, none of these interventions have examined personal M/P in relation to PA adoption or maintenance.
The primary limitation of the current literature is that all current studies of M/P and PA are cross-sectional in nature; to my knowledge, no studies have specifically incorporated life M/P into a longitudinal observational study of PA adoption. Therefore, this study will
5


address a large gap in the research by studying the relationship between M/P and PA over time in a group of previously sedentary new exercisers. Additionally, no studies have tested the hypothesis that greater salience of M/P on a daily basis is related to PA on that particular day. This investigation will address that hypothesis in a daily diary study of PA and M/P. Moreover, no studies have incorporated a theoretical model that attempts to explain how M/P may influence PA over time. Thus, this study will attempt to integrate M/P theory with an existing model of health behavior change, i.e., Self-Determination Theory (SDT). Self-Determination Theory
It is often cited that M/P provides motivation to confront lifes problems (Sagy, Antonovsky, & Adler, 1990) and to develop well-formed organized goals (McKnight & Kashdan, 2009). This is consistent with the assertion that behaviors integrated with M/P are more likely to be maintained. SDT (Ryan & Deci, 2000) is a relatively young and particularly promising theory of motivation that posits that behaviors that are internalized or intrinsically motivated are more likely to be maintained than behaviors that are externalized or extrinsically motivated. Deci and Ryan (2000) define intrinsic motivation as the active engagement with tasks that people find interesting and that, in turn, promote growth (p.
233). SDT also explains why people may engage in behaviors that are not interesting and enjoyable. The SDT sub-theory called organismic integration theory postulates that motivation exists on a continuum of behavioral regulations: (a) amotivation (lacking intention to act); (b) external regulation (driven by external rewards and punishments), (c) introjected regulation (motivated by guilt or need to comply); (d) identified regulation (driven by personal importance or conscious valuing); (e) integrated regulation (motivated by
6


making healthy choices. In a systematic review of 66 studies examining the relations between SDT constructs and exercise and PA, 91% of the studies found a positive association between autonomous regulations and exercise behavior (Teixeira, Carra9a, Markland, Silva, & Ryan, 2012). The vast majority of the evidence comes from cross-sectional studies demonstrating that people who report more self-determined or internally regulated motivations also report greater participation in PA and exercise and more positive psychological outcomes of exercise participation (Chatzirantis & Hagger, 2007; Hall, Rodgers, Wilson, & Norman, 2010; Ingledew & Markland, 2004; Ingledew & Markland, 2008; Landry & Soloman, 2004; Markland & Ingledew, 2007; Mullan & Markland, 1997; Sebire, Standage, & Vansteenkiste, 2009; Standage, Sebire, & Loney, 2008; Wilson & Rodgers, 2002; Wilson, Rodgers, Fraser, & Murray, 2004).
There are several prospective longitudinal studies (range of 4-24 weeks) that demonstrate that basic psychological needs and autonomous motivation at time 1 are predictive of PA behavior at time 2 (Barbeau, Sweet, & Fortier, 2009; Gunnell, Crocker, Mack, Wilson, & Zumbo, 2014; Hagger, Chatzirantis, & Harris, 2006a,b; Kwan, Caldwell Hooper, Magnan, & Bryan, 2009). However, none of these prospective studies examined SDT constructs in a behavior change context; participants were either already active or were not explicitly recruited because they intended to change their activity levels. One study of middle-aged women recruited from the community specifically examined autonomous regulation in relation to behavioral intentions to engage in PA and change in PA from time 1 to time 2 (6 months later; Fortier, Kowal, Lemyre, & Orpana, 2009). Participants were eligible if they stated they intended to increase their PA over the next 6 months. Autonomous regulation at baseline was positively related to intentions to increase PA, which were
8


positively related to PA behavior change. However, contrary to the hypotheses, autonomous motivation was not significantly related to behavior change. Fortier and colleagues (2009) hypothesized this was because there was limited variability in autonomous motivation, and the sample was highly internally regulated already. Another possible hypothesis is that the majority of participants were meeting PA guidelines at baseline, which may limit the ability to predict increases in PA behavior. Fortier et al. (2009) suggest that work examining previously sedentary individuals desiring to increase their PA would add to the SDT literature.
Research suggests that new (previously sedentary) exercisers, in non-SDT-based exercise interventions, tend to naturally decrease in external regulations and increase in more internalized motivations over time (Rodgers, Hall, Duncan, Pearson, & Milne, 2010) and more self-determined motivations are associated with greater exercise persistence (Gorin et al., 2008; Hagger & Chatzirantis, 2008; Teixeira, Silva, Mata, Palmeira, & Markland, 2012). In exercise initiates starting new exercise programs, identified and integrated forms of regulation tend to increase more quickly over time than intrinsic motivation (Rodgers et al., 2010; Teixeira, Carraqa et al., 2012). This suggests that new exercisers may be more likely to continue engaging in exercise because exercise becomes increasingly consistent with their identities or values, rather than because they experience increased enjoyment of exercise. Based on cross-sectional, prospective, and intervention research, there is ample evidence to support the hypothesis that when exercise is more internalized, the behavior is more likely to be maintained. However, little is known about the natural progression of the SDT motivation internalization process in previously sedentary exercise initiates not participating in formalized exercise interventions.
9


How does a behavior become self-determined or internally motivated? The SDT Process Model of Behavior Change (Fortier, Duda, Guerin, & Teixeira, 2012) and a SDT sub-theory called the Basic Needs Theory (Ryan & Deci, 2000) suggest that social environments that support the three basic psychological needs of autonomy (feeling behavior is self-organized, accompanied by a sense of volition), relatedness (feeling connected to others), and competence (feeling capable of achieving goals) foster the internalization of motivation and facilitate behavior change. Evidence suggests that increases in psychological needs satisfaction for PA over three years are positively related to PA behavior changes in adolescents (Gunnell, Belanger, & Brunet, 2016). Moreover, in exercise initiates in non-SDT based interventions, basic psychological needs satisfaction (autonomy and competence) is predictive of increases in autonomous regulation (e.g., Edmunds, Ntoumanis, & Duda, 2007). Support for this model has also been observed in studies of PA interventions incorporating SDT, in that they have shown success in increasing internalized regulation and increasing PA by targeting these three basic needs (Fortier et al., 2007; Fortier et al., 2011; Jolly et al.,
2009; Rouse, Duda, Ntoumanis, Jolly, & Williams, 2010; Silva et al., 2008; Silva et al.,
2010; Silva et al., 2009). Despite the promise of the SDT approach, none of these studies have explicitly examined M/P in the context of SDT as an added value to the model.
Given previous hypotheses that M/P helps individuals maintain behaviors that are integrated with their most important values and life goals (Scheier et al., 2006), it is likely that those who live with awareness of their personal M/P also have more internalized or integrated forms of behavioral regulation. Figure 2 illustrates the SDT Process Model of Behavior Change (Fortier et al., 2012) with the important addition of M/P. Fortier and colleagues (2012) model includes psychological needs satisfaction, behavioral regulations
10


motivation, PA behavior, and psychological well-being. However, I have modified the model to include M/P.
Figure 2. Model Integrating SDT and M/P to Increase PA and Well-being.
The SDT Process Model of Behavior Change suggests that behavior change occurs in social environments where individuals perceive their three basic psychological needs are being met. Individuals must feel (1) more competent in their abilities to perform certain behaviors (competence); (2) supported by their social group (relatedness); and (3) that they have the opportunity to choose how and when the behaviors will be performed (autonomy). When all three needs are satisfied, individuals tend to engage in more autonomously regulated or internalized (e.g., identified, integrated, and intrinsic) behaviors. More internalized behaviors are more likely to be maintained; thus, PA that is more autonomously regulated will be more likely to be maintained. Autonomously regulated behavior leads to improved psychological well-being. I suggest that M/P will add to this model in three specific ways:
11


1. When the three basic psychological needs are met, the environment supports individuals in developing their own personal M/P. Thus, there will be a positive relationship between psychological needs satisfaction and M/P.
2. When individuals have greater awareness of their personal M/P, they will likely act in ways that are more consistent with their M/P, i.e., more internally regulated; therefore, individuals who live with greater awareness of their M/P will also be more internally regulated to engage in PA. An implicit assumption is that the participants in this study will value PA, and PA is consistent with their overall M/P. Because these individuals will be voluntarily starting an exercise program, it is likely that this assumption is true.
3. Finally, previous evidence suggests that M/P is directly and positively related to psychological well-being (e.g., Scheier et al., 2006; Steger et al., 2006).
Meaning and Mood
A possible rival hypothesis to the M/P as motivation theory is that M/P is a consequence of improved mood as a result of activity. Evidence suggests that increased PA is related to reduced depressive symptoms (Goodwin, 2003; Hooker, MacGregor, Funderburk, & Maisto, 2013; Taliaferro, Rienzo, Pigg, Miller, & Dodd, 2008) and increased positive affect (Cards, Coit, Young, & Berger, 2007). The Feelings as Information theory proposed by Schwarz (2001) suggests that individuals use positive affect as information to make judgments about other areas of their lives. For instance, sedentary individuals who start exercising more will experience improved mood (Annesi, 2004). Because they are experiencing more positive affect, they will use their mood to make judgments about how
12


meaningful their lives are. If they are feeling more positive, then they will rate their meaning in life as being greater as well.
Some evidence suggests that positive mood is related to global evaluation of M/P (Hicks & King, 2008; 2009; King et al., 2006). However, when experimentally primed with images of religious commitment (Hicks & King, 2008) or social relatedness (Hicks & King, 2009), the relationship between M/P judgments and positive affect declines. This suggests that positive moods may predispose individuals to feel that their lives are meaningful; however, individuals use additional factors to judge how meaningful their lives actually are. Moreover, in a study of M/P and objectively assessed PA (Hooker & Masters, 2014), positive affect and depressive symptoms were not significantly related to PA over the course of three consecutive days, and controlling for both positive affect and depressive symptoms did not significantly reduce the relationship between M/P and PA. Although positive affect seems to have important relations with M/P and PA, it is not clear that it is a confounder of the relationship between M/P and PA. However, in the present study, mood and depressive symptoms are measured to account for this potential confound.
Self-monitoring of M/P, Mood, and PA as a Possible Intervention
One goal of this study is to examine M/P salience in relation to PA, controlling for daily mood. It is possible that having participants record their daily M/P and PA could, in fact, be an intervention in itself. Indeed, a recent meta-analysis found that self-monitoring of PA is an effective technique to increase PA self-efficacy and PA behavior (Olander et al., 2013). However, given the limited literature on the association between M/P and PA, it is not clear whether monitoring M/P, mood, and PA all together is, in fact, an intervention. Thus, in
13


the current study, participants were randomized to the daily self-monitoring of M/P, mood, and PA condition or to a random-survey assessment-control.
Purpose of the Study
Interventions that enable individuals to maintain PA have proven difficult for behavioral scientists, and novel approaches to behavior change are needed. Given that previous research has only examined this relationship between M/P and PA in cross-sectional studies, a longitudinal study of these associations as adults begin PA programs may help us better understand the relationships of M/P with PA. Moreover, no studies have explicitly examined whether salience of M/P is related to PA on a daily basis. To fill the gaps in the research, this study addresses three aims:
(1) Determine whether self-monitoring of daily M/P, mood, and PA results in increased PA compared to an assessment-control random-survey condition;
(2) Examine the relationship between daily ratings of M/P and PA over time adjusting for daily mood; and
(3) Examine whether M/P predicts PA and psychological functioning beyond an existing and supported model of health behavior change, the SDT Process Model of Behavior Change.
The primary hypotheses of this study are that:
(1) self-monitoring will not increase PA compared to the control condition;
(2) daily awareness of M/P will be positively related to greater engagement in PA during that same day; and
(3) M/P will predict increased engagement of PA over time (12 weeks) beyond the SDT Process Model of Behavior Change.
14


CHAPTER II
METHOD
Overview
This study used a randomized controlled trial design (RCT) and was registered with clinicaltrials.gov (NCT02538068). Participants were randomized to a daily self-monitoring of M/P, mood, and PA condition for 4-weeks or to a random-survey assessment-control. The relationship between daily M/P and PA was examined for the self-monitoring group only. A conceptual model of the relationship of M/P with PA and psychological well-being was also tested. The model, presented in Figure 2, was adapted from the work of Fortier and colleagues (2012) and was enhanced by adding the crucial M/P component. The study examined whether the basic psychological needs posited by SDT were related to M/P and internalization of PA motivation (behavioral regulations motivation). In turn, internalized motivation and M/P were hypothesized to be related to greater engagement in PA and greater psychological well-being. Additionally, it was hypothesized that M/P would be directly related to psychological well-being as studies have shown that people who engage in intrinsically meaningful life activities that are consistent with their innermost values and aspirations experience greater life satisfaction and well-being (Steger, Kashdan, & Oishi, 2008).
Participants and Recruitment
Participants were recruited from the Anschutz Medical Campus and the greater Denver area in a variety of ways. First, some individuals were informed about the study when they began a new membership at the Anschutz Health and Wellness Center (AHWC). Center staff asked potentially eligible individuals if they were interested in having a research assistant contact them about the study. Second, one recruitment email was sent to new
15


members within one week of joining the center to determine if they were interested in participating in the study. Other methods of recruitment, including flyers on the campus and in the community, email announcements to University of Colorado Denver (UCD) faculty and staff, posting of the study on the University Clinical Trials website and the AHWC website, and word of mouth were also used.
Participants were screened for eligibility to participate. The eligibility criteria were chosen to form a sample of sedentary adults who were joining a fitness center. To be included, participants had to be (1) at least 30 years of age; (2) able to read and understand English; (3) sedentary (engaging in < 60 minutes of moderate-to-vigorous exercise per week) for the last 3 months; and (4) joining the AHWC. Individuals were excluded from the study if they (1) had medical or physical contraindications to participate in PA; (2) had an existing diagnosis of cardiovascular disease; or (3) were pregnant. The rationale for including adults age 30 and older was that M/P becomes more salient for adults who enter middle and older age (Steger, Oishi, & Kashdan, 2009) and PA declines for both genders and most ethnic groups at middle age (Hawkins et al., 2009). Therefore, middle to older age is a particularly important time to understand the processes that predict long-term maintenance of PA and is also particularly apropos for studying M/P in this context. The rationale for enrolling adults who read and understood English was that the majority of the measures had only been validated in English. Participants had to be previously sedentary because the study was designed to examine the process of PA adoption. Finally, participants had to be joining the AHWC because a secondary outcome was fitness center attendance. Moreover, because the study was observational and did not involve a formalized intervention, it was thought that participants joining a fitness center would be more likely to attempt to become regularly
16


physically active. Participants who had medical or physical contraindications to participate in PA (defined as a positive Physical Activity Readiness Questionnaire [PAR-Q; Thomas, Reading, & Shephard, 1992] score) were excluded. Participants who were unable to engage in certain forms of PA due to illness or injury would artificially skew the averages of PA in the sample. Additionally, this study was designed as a study of primary prevention of cardiovascular disease; therefore, those with CVD were excluded. Finally, it may not have been safe for pregnant women who were not already active to start an exercise program without supervision from a physician; thus, they were also excluded from this study. Randomization
To assess the effect of self-monitoring daily M/P and PA, participants were randomized 1:1 to the self-monitoring or the control condition. A blocked randomization sequence was created using a random number generator in Microsoft Excel. The randomization included eight blocks so that the number of participants randomized to each group would be even for every 20 participants. Participants in the self-monitoring condition completed daily surveys for the first four weeks (28 days total). In addition, at the end of the daily survey, a quote appeared to help prime participants to think about their personal M/P (see Appendix A). Participants randomized to the control condition completed eight surveys delivered at blocked random intervals over the first 4 weeks. The surveys asked participants to recall their activities for the last 24 hours but explicitly did not ask participants to record their PA or rate their M/P. Participants were randomized to get one survey on a weekend day (Saturday or Sunday) and one survey on a weekday (Monday Friday) for each of the four weeks.
17


Procedure
The study procedures were reviewed and approved by the Colorado Multiple Institutional Review Board (COMIRB). Interested individuals completed a phone or in-person screening prior to enrollment. Research assistants acquired verbal consent for screening from potential participants, and asked questions in the screening questionnaire (see Appendix B) to determine if the individual was eligible to participate. If eligible, research assistants scheduled an in-person meeting to complete consent and baseline measurements. Participants were offered $20 for each month they participated in the study, for a total of $60 for the three months. This was available either as a discounted rate off their AHWC memberships or in the form of a gift card, which they received at the end of the study. Additionally, participants were entered into a drawing for a variety of gift cards every time they completed a survey for the study.
Participants were observed as they began new PA programs for 12 weeks. They completed self-report measures at baseline (within 2 weeks of joining the AHWC), 4-weeks and 12-weeks (see Figure 3 for study timeline; see Appendices C, D, and E for questions). During the baseline visit, research assistants reviewed the full COMIRB-approved consent form with participants. Participants completed the baseline questionnaire, and were randomized to either the self-monitoring or control condition. Research assistants showed participants how to complete the daily measures that were to be sent to them via email. After participants were given an overview of what to expect for the next 12 weeks, research assistants conducted a brief fitness assessment to gather measures of body size (height, weight, body mass index [BMI], body fat percent, and waist circumference) and physical fitness (cardiovascular, strength, and flexibility). The baseline visit took, on average, between
18


45-60 minutes to complete. Participants randomized to the self-monitoring condition were asked to complete daily measures (at 8:00 PM each day) of M/P, mood, and PA for the four weeks after baseline (see Appendix E). Participants in the control condition completed a survey that included a 24-hour recall of daily activities (e.g., sleeping, eating, working, etc.) on eight random days over the first four weeks. Participants in the self-monitoring condition also received the 24-hour recall of daily activities on eight random days. For the purposes of this study, the control condition was considered an assessment-control and was not designed to encourage self-monitoring. Both the self-monitoring and control surveys were designed to take less than 10 minutes to complete.
Daily measures via email for 4 weeks and 8 random surveys
OR
8 random surveys via email over 4 weeks
1 1

Screening Baseline & 4-week Follow-up 12-week Follow-up
Randomization
Figure 3. Study Timeline.
Study data were collected and managed using REDCap electronic data capture tools hosted at the Elniversity of Colorado Denver (Harris et al., 2009). REDCap (Research Electronic Data Capture) is a secure, web-based application designed to support data capture for research studies, providing: 1) an intuitive interface for validated data entry; 2) audit trails for tracking data manipulation and export procedures; 3) automated export procedures for seamless data downloads to common statistical packages; and 4) procedures for importing data from external sources.
19


Participants received email notifications sent directly from REDCap prompting them to complete the surveys at the specified times. At the end of the 12 weeks, participants completed a second in-person fitness assessment. Measures for each time point are described in Table 1.
Measures
As part of their participation in this study, participants completed a variety of self-report measures. These surveys were all completed online through the survey software REDCap. All measures were chosen to satisfy the aims of this project. At baseline, demographics (age, race, ethnicity, education, religious affiliation, living situation, family status, and income) and health histories were collected to use as covariates in analyses. Additionally, measures of body size (weight, height, BMI, waist circumference, body fat percent) and physical fitness were collected at the baseline and 12-week follow-up as objective measures of cardiovascular health risk. An overview of the measures collected is represented in Table 1.
To examine Aim 2, participants in the self-monitoring condition completed daily measures of mood, M/P, and self-reported PA via email for four weeks. Participants in both conditions completed 24-hour recalls of daily activities on eight random days over the first four weeks.
Daily mood. Daily positive and negative mood were measured using a positive and negative affect scale previously used in a study of daily M/P and daily mood (Steger et al., 2008). Eight items measured positive affect (relaxed, proud, excited, appreciative, enthusiastic, happy, satisfied, and curious) and five items measured negative affect (sluggish, afraid, sad, anxious, and angry). Participants rated their mood using a 5-point Likert-type
20


scale ranging from 1 (very slightly or not at all) to 5 {extremely). Sums of the positive and negative affect scales have been shown to be positively correlated with their respective scales on the Positive and Negative Affect Schedule (PANAS; Watson, Clark, & Tellegen, 1988) (r
Table 1
Measures Collected in this Study
Aim Primary Outcome Secondary Outcomes Primary Predictor Secondary Predictors/ Mediators Covariates
Aim 1 PA (IPAQ-SF at 4-week and 12-weeks) Fitness center attendance, M/P, SDT mediator, Psychological well-being (Subjective Vitality, Life Satisfaction) Selfmonitoring vs. Control N/A N/A
Aim 2 Daily self-reported minutes of PA Fitness center attendance, Self-reported PA intensity Daily M/P N/A Daily mood
Aim 3 PA (IPAQ-SF at 4-weeks and 12-weeks) Psychological well-being (Subjective Vitality, Life Satisfaction), Body size, Physical Fitness Meaning in Life (MILQ) Purpose in life (LET; Secondary), Basic Psychological Needs in Exercise, Behavioral Regulations in Exercise Demographics, health history, depressive symptoms
Note. IPAQ-SF = International PA Questionnaire Short Form; LET = Life Engagement Test; M/P = Meaning and purpose; MILQ = Meaning in Life Questionnaire; PA = Physical Activity; SDT = Self-Determination Theory.
21


= .55 .57) and to have very high internal consistency (a = .97 .98). In this study, the internal consistency was very high for positive affect (as = .86 .94) and acceptable to high for negative affect (as = .70 .86) over the 28 days.
Daily M/P salience. Daily M/P salience was measured using the Thoughts of Meaning Scale (TOMS), which was developed for this study. The TOMS included the 2-item Daily Meaning Scale (DMS; Steger et al., 2008). Participants rated the extent to which the two statements were true for them at the moment (How meaningful does your life feel? and How much do you feel your life has a purpose?). To increase the potency of the measurement self-monitoring as a possible intervention and to more directly assess M/P salience, eight additional items were added to this scale. Participants rated the extent to which they thought about M/P that day (e.g., How much have you thought about your purpose in life today?). Participants completed these items using Likert-type rating scales ranging from 1 {not at all) to 7 {absolutely [DMS] or quite a bit [TOMS]). The DMS has been shown to have very strong reliability (a = .98) and to be positively related to daily engagement in eudaimonic behaviors (i.e., behaviors that are consistent with values; Ryan & Deci, 2001) {y = .13) in daily studies of M/P (Steger et al., 2008). In this study, the internal consistency of all 10 items over the 28 days was very high (as = .88 .96).
PA. Both self-report and objective measures of PA were gathered. Participants reported the type of exercise activities they engaged in (e.g., jogging, swimming, yoga, etc.), the duration of the activity (min/day), and intensity of engagement (using Borgs Category-Ratio exertion scale, also known as the Ratings of Perceived Exertion Scale; Borg, 1998). To rate the average intensity of their activities, participants were shown a scale ranging from 6 to 20, where 6 corresponded to no exertion at all, 11 corresponded to light activity, 15
22


corresponded to hard (heavy) activity, and 20 corresponded to maximal exertion. This method was used in a previous daily diary study of exercise (Lutz, Stults-Koehmainen, & Bartholomew, 2010) and allowed for analysis of different components of PA (duration, frequency [days/week], and intensity of PA). In addition to daily self-report of PA, the number of days that participants attended the fitness center was obtained from fitness center records.
24-hour activity recall. A 24-hour recall of activities was delivered on 8-random days over the first 4-weeks to participants in both conditions. Participants were asked to recall their activities from a list of 46 common activities (e.g., sleeping, driving, working, eating, etc.) for every half hour period over the previous 24 hours (from 8:00 p.m. the night before to 7:30 p.m. the night the survey was delivered). Participants could report up to two activities for each half hour period. There were also two other categories that participants could specify if they were engaging in an activity that was not on the list.
To examine Aim 3, participants completed measures of the model in Figure 2 at baseline, 4, and 12 weeks, and those were as follows:
Psychological needs satisfaction. Satisfaction of the three basic needs in SDT (autonomy, relatedness, and competence) was measured using the Basic Psychological Needs in Exercise Scale (BPNES) (Vlachopoulos & Michailidou, 2006). The BPNES has 11-items, with four items corresponding to autonomy (e.g., The way I exercise is in agreement with my choices and interests. ), four items corresponding to competence (e.g., I am able to meet the requirements of my exercise program. ), and three items corresponding to relatedness (e.g., My relationships with the people I exercise with are close. ). Participants rated their needs satisfaction on a 5-point Likert-type scale ranging from 1 (I dont agree at
23


all) to 5 (/ completely agree). The BPNES has demonstrated strong internal consistency (as ranging from .81 to .92), high test-retest reliability over four weeks (ICC = .97), and construct validity (Vlachopoulos & Michailidou, 2006). Perceived competence has been shown to be positively related to exercise frequency (r = .21; Vlachopoulos & Michailidou, 2006). The internal consistency of the whole scale was high at baseline (a = .85), 4 weeks (a = .90), and 12-weeks (a = .92).
Behavioral regulations motivation. Motivation internalization for exercise was measured using the Behavioral Regulation in Exercise Questionnaire-2 (Markland & Tobin, 2004; Mullan, Markland, & Ingledew, 1997; BREQ-2). The 19-item BREQ-2 measured motivations for exercise on the SDT continuum (see Figure 1). The 5 subscales were: amotivation (e.g., / dont see why I should have to exercise.); external regulation (e.g., / exercise because other people say I should)', introjected regulation (e.g., /feel guilty when I dont exercise.); identified regulation (e.g., Its important to me to exercise regularly)', and intrinsic motivation (e.g., I exercise because its fun). Since the development of the BREQ-2, four items used to assess integrated regulation for exercise were developed (e.g., I exercise because it is consistent with my life goals Wilson, Rodgers, Loitz, & Scime,
2006). These items were added to the existing BREQ-2. Participants rated the extent to which they engaged in exercise (or did not engage in exercise) for each of the reasons on a 5-point Likert-type scale ranging from 0 {not true for me) to 4 (very true for me). The six subscales were combined using the bifurcation approach outlined by Wilson and colleagues (Wilson, Sabiston, Mack, & Blanchard, 2012) into two scaled scores: autonomous and controlled regulation. Autonomous regulation was the average of the intrinsic, integrated, and identified scales whereas controlled regulation was the average of the external and introjected scales.
24


The BREQ-2 has been shown to have good psychometric properties, including good internal consistency (as range from .78 .93; Markland & Tobin, 2004; Wilson et al., 2006). Additionally, more internalized forms of regulation (identified, integrated, and intrinsic) have been shown to be positively related to psychological needs satisfaction and to self-reported PA (Wilson et al., 2006). In this study, the internal consistency of the autonomous regulation score was very high at all three time points (as = .92 .94). The controlled regulation score was acceptable at all three time points (as = .78 .82).
M/P. M/P was measured using The Meaning in Life Questionnaire (MILQ; Steger et al., 2006) and the Life Engagement Test (LET; Scheier et al., 2006).
Meaning in life. The 10-item MILQ measured the extent to which a person perceives his or her life as meaningful and searches for meaning in life. This measure included two subscales: presence measured the sense that ones life is meaningful (e.g., My life has a clear sense of purpose ) whereas search measured the drive and orientation towards finding a sense of meaning in life (e.g., I am always looking to find my lifes purpose ). Participants rated the extent each statement was true for them on a 7-point Likert-type scale ranging from 1 {absolutely untrue) to 7 {absolutely true). Responses were summed for each subscale.
Steger and colleagues (2006) found that both scales demonstrate good internal consistency (as = .86 and .92 for presence and search, respectively) and have moderate test-retest reliability over one month (rs = .70 and .73 for presence and search, respectively). Evidence for convergent and discriminant validity indicated that the presence subscale was positively associated with health and well-being indicators (i.e., life satisfaction, joy) and negatively associated with depressive symptoms and negative emotionality. Conversely, the search subscale was not associated with health and well-being indicators but was positively
25


associated with depressive symptoms and negative emotionality. In this study, the internal consistency was very high at all three time points (as = .89 .94).
Purpose in life. The 6-item LET measured purpose in life and the extent to which individuals believed their activities were valuable and important (e.g., To me, the things I do are all worthwhile. ). Participants rated the extent to which they agreed with each item on a 5-point Likert-type scale ranging from 1 {strongly disagree) to 5 {strongly agree). Odd items were reverse-scored and the six items were summed for a total score. Scheier and colleagues (2006) examined the test-retest reliability of this measure in four different samples and found that the LET was moderately stable (rs ranged from .61 .76) over 4 months. Additionally, they tested the convergent and discriminant validity of the measure and found that it was positively associated with many health and well-being indicators (i.e., life satisfaction, general physical health) and negatively associated with depressive symptoms. The internal consistency for this measure was high in a previous sample of adult community members (a = .85; Hooker & Masters, 2014). In this study, the internal consistency was high to very high at all three time points (as = .85 .93).
PA. In addition to the PA measures described above, at baseline, 4-weeks, and 12-weeks, participants completed the International Physical Activity Questionnaire Short Form (IPAQ-SF), a self-report measure of activity in the last 7 days (Craig et al., 2003). The short form consists of seven items and asks participants to estimate the amount of moderate and vigorous exercise, walking, and sedentary activity they engaged in during the past 7 days. Items were intended to provide separate scores on walking, moderate-intensity, and vigorous-intensity activity and computation of a total score requires the multiplication of the duration (in minutes) times the frequency (days) of walking, moderate-intensity, and
26


vigorous-intensity activities). Categorical (Low, Medium, and High activity) and continuous (minutes, metabolic equivalent of task [MET] minutes of PA) scores were calculated for each participant. Evidence suggests that the IPAQ-SF has demonstrated acceptable measurement properties, on par with other self-report measures of PA. Test-retest reliability over 8 days has been shown to be very good, with rs ranging from .71-.91, and the self-report version is moderately correlated with objectively measured (accelerometer) PA over the same 7-day period (rs ranging from .26 to .47; Craig et al., 2003).
Psychological well-being. Psychological well-being was measured using two measures: The Subjective Vitality Scale (SVS; Ryan & Frederick, 1997) and The Satisfaction with Life Scale (SWLS; Diener, Emmons, Larsen, & Griffin, 1985).
Subjective vitality. The 7-item SVS measured feeling active, alive, enthusiastic, and energetic (e.g., Ifeel alive and vital). Participants rated the extent they generally felt this way on a 7-point Likert-type scale ranging from 1 {not at all true) to 7 (very true). Item 2 ( I dont feel very energetic ) was dropped (per author instructions and confirmatory factor analysis results, see Bostic, Rubio, & Hood, 2000) and the rest of the items were summed for a total SVS score. Previous psychometric studies demonstrated that the SVS correlated positively with other positive measures of well-being (e.g., self-esteem, self-actualization, satisfaction with life) and negatively with poor psychological well-being (e.g., depression, anxiety, psychopathology), and was internally consistent (a = .84-.86) (Ryan & Frederick, 1997). The internal consistency of the SVS in this study was very high at all three time points ( Life satisfaction. On the SWLS, participants rated their agreement with five statements on a 7-point Likert scale ranging from 1 {strongly disagree) to 7 {strongly agree).
27


Responses were summed so that higher scores corresponded with greater satisfaction with life. Diener and colleagues (1985) conducted a thorough assessment of the SWLSs reliability and validity. Previous assessments of reliability indicated that this measure is internally consistent (a = .87) and had good two-month temporal stability (r = .82). Evidence for construct validity indicated that the SWLS correlated positively with other measures of subjective well-being and negatively correlated with measures of personality psychopathology. Evidence for criterion validity indicated that the measure correlated highly with interviewers rating of the individuals satisfaction with life (Diener et al., 1985). In the current study, the scale demonstrated high internal consistency at all three time points (as = .86 .90).
Depressive symptoms. Depressive symptoms were used as a covariate in the path analysis model. The 8-item Patient Health Questionnaire-8 (PHQ-8; Kroenke & Spitzer, 2002) was used to measure depressive symptoms. Participants rated the extent to which they were bothered by a series of eight problems (e.g., little interest or pleasure in doing things ) over the past 2 weeks on a scale ranging from 0 {not at all) to 3 {nearly every day).
If they experienced any of the eight problems, they were asked to rate how difficult those problems had been for them {not difficult at all, somewhat difficult, very difficult, or extremely difficult). The PHQ-8 was derived from the PHQ-9, but the ninth item assessing suicidal ideation was omitted as per recommendations from the authors because this study (1) used self-administered surveys; and (2) depression was assessed as a secondary measure (Kroenke & Spitzer, 2002). Evidence for validity for this scale indicated that higher scores on the PHQ were related to greater likelihood of being diagnosed with any depressive disorder
28


(Kroenke & Spitzer, 2002). The internal consistency of the PHQ-8 was high at all three time points in this study (as = .82 .85).
Physical fitness. Physical fitness was assessed using the standardized fitness test available at AHWC. Participants completed three physical fitness tests, including one for aerobic fitness (Young Mens Christian Association [YMCA] 3-minute step test; Golding, 2000), one for strength (grip strength as assessed by Takei Hand Grip Dynamometer A5401; Takei Scientific Instruments, Tokyo, Japan), and one for flexibility (sit-and-reach test; Holt, Pelham, & Burke, 1999). Additionally, participants resting heart rates were gathered at the beginning of the fitness test. Raw scores for each test (resting heart rate, step test, grip strength, and sit and reach) were recoded into categorical scores based on fitness targets by age and gender, with scores ranging from 1 (poor fitness) to 5 (excellent fitness). The fitness targets were based on the guidelines from the American College of Sports Medicine (ACSM; Percia, Davis, & Dwyer, 2012). The four scores were summed for a total score (ranging from 4 to 20) and then recoded into a fitness percent score ranging from 0-100%.
M/P and PA connection. As an exploratory measure, two items were embedded in the BREQ-2 to assess whether participants made an active connection between M/P and reasons to do PA. Participants rated the two items on the same scale as the BREQ-2, or as the extent to which the two statements were true for them (Exercise gives me more energy to do the things that really matter to me in life. and Engaging in regular exercise helps me reach my life goals.). The two items were summed for a total score, and had acceptable internal consistency at each of the three time points (a = .69, .73, and .70 at baseline, 4-weeks, and 12-weeks, respectively).
29


Power Analysis
The sample size was determined based on power analyses as well as feasibility. GPower 3.1 (Faul, 2007) was used to estimate power for the three aims.
Aim 1. Effect of M/P, mood, and PA self-monitoring on PA. Power for Aim 1 was estimated for a two group, repeated measures design with three time points. The group-by-time interaction was of primary interest. Assuming a two-tailed test with a = .05, a moderate correlation between the three time points (r = .50), and power = .80, a sample size of 160 was needed to detect a small difference (f= .10) between the groups.
Aim 2. Relationship of daily M/P salience to daily PA. Power for a within-subjects repeated measures design with 25 measurements per subject, allowing for three days of missed data collection per subject (out of 28 total days) was estimated. Additionally, a small-moderate correlation within the individual of PA duration across days (r = .30) was used to estimate the correlation of the repeated measures over time. A sample size of 42 would have 95% power to detect a small to moderate relationship (f= .15) between daily M/P and PA given an alpha of .05. Because participants were randomly assigned to this condition in a 1:1 fashion and the sample size requirements for Aim 1 and Aim 3 were 160 (see below), 80 subjects were randomized to this condition, allowing for 39 subjects to dropout while still maintaining this level of power.
Aim 3. Examining the SDT process model of PA adoption with and without M/P.
Because path analysis is essentially a series of multiple regressions, power for the path analysis was calculated using the largest regression in the model (6 predictors in the model plus 7 covariates). A sample size of 131 was needed to detect a small-moderate relationship (f= .15) between the 13 predictors and the outcome (PA) with 80% power. However, given
30


the longitudinal nature of the study, a 20% attrition rate was expected. Therefore, a target sample size of 160 was recruited for the project.
Data Analysis
Analyses were conducted in SAS 9.4 (SAS Institute, Inc., 2015) and Mplus 1A (Muthen & Muthen, 2015). SAS was used to manage the data and calculate descriptive statistics. Descriptive statistics were used to examine distributions and assess potential violations of assumptions. Variables that did not meet assumptions were transformed prior to analysis. Specifically, for PA, square-root transformations were used to normalize the data. Baseline differences between the self-monitoring and control group were examined using the appropriate statistical tests (y2 for categorical data and /-tests for continuous data). Longitudinal PA patterns were graphed using a spaghetti plot to determine if there were any prototypical patterns of change. Additionally, patterns of missing data were examined to determine whether missing data were missing at random, missing completely at random, or missing not at random. Individuals who missed the 4-week or 12-week follow-up assessments were compared to those who did not miss those assessments on baseline variables using the appropriate tests (y2 for categorical data and /-tests for continuous data). The proportion of completed daily surveys was calculated and correlated with baseline variables. Non-random missing data patterns were taken into account when interpreting the results. Specifically, if the missing data analysis suggested that missing data were related to the relationship between the primary predictor and the outcome (e.g., participants who were missing more follow-up data also had lower M/P and PA at the beginning of the study), then these patterns were noted in the limitations of the study. Finally, bivariate correlations among all study predictors, covariates, and outcomes were calculated to examine patterns in the data.
31


Aim 1. Effect of M/P, mood, and PA self-monitoring on PA. Mixed repeated measures models were used to examine the differences between treatment groups in PA and M/P over time (at baseline, 4-weeks and 12-weeks). Each model included three predictors: treatment group (between factor), time (within factor), and the group-by-time interaction. Models were estimated using unstructured covariance matrices and restricted maximum likelihood (REML) estimation, which is recommended for estimating the covariance structure of the data (Fitzmaurice, Laird, & Ware, 2011). PA, measured by the IPAQ-SF, was the primary outcome. To examine secondary outcomes, the model was repeated with fitness center attendance (number of visits per week at each time point) and different self-reported PA intensities (vigorous, moderate, walking) as the dependent variables. Additionally, the model was repeated with hypothesized mediators, including M/P (measured by the MILQ and LET), basic psychological needs satisfaction (BPNES), and autonomous regulation as dependent variables. Finally, the models were repeated with psychological well-being and depression variables as exploratory outcomes. The group-by-time interactions were of primary interest. Group means over time were graphed using time plots.
Aim 2: Relationship of daily M/P salience to daily PA. Multi-level models were used to examine the relationship between daily M/P and daily self-reported PA (frequency, intensity, duration) controlling for mood and clustering at the time level. Multi-level models controlled for repeated measures within subjects to ensure that standard errors were appropriately estimated and Type 1 error rates were not inflated. Models were run in Mplus, which controlled for clustering of data and used cases that have partially complete data through full information maximum likelihood estimation. As a secondary outcome, daily M/P was examined in relation to fitness center visits as a measure of objective PA to examine
32


consistency with the results of the self-report. A multilevel model with the same predictors (meaning, positive affect, and negative affect) was used to predict whether or not participants visited the fitness center that day (coded as a binary outcome: l=visit; 0=no visit).
Aim 3: Examining the SDT process model of PA adoption with and without M/P. All SDT and M/P process variables were assessed at the three main assessments. Path analysis examined variations of the model in Figure 2. Mplus was used for analysis so that full information maximum likelihood accounted for any missing data. Models were tested cross-sectionally at each assessment (baseline, 4-week, and 12-week) and then longitudinally to examine the change from baseline to 12-weeks. The initial model (base SDT Process Model of Behavior Change without M/P) was tested for goodness of fit using several fit indices, including Akaikes Information Criterion (AIC), Bayesian Information Criterion (BIC), % test of model fit, comparative fit index (CFI), root mean square error approximation (RMSEA), and standardized root mean residual (SRMR). Hu and Benders (1999) recommended cut-points for each fit index were used to evaluate the fit of each model, including a non-significant %, RMSEA < .06, CFI > .95, and SRMR < .08. All models controlled for six covariates on PA (gender, age, employment [full-time vs. not full-time], income [> $40,000 annually v. < $40,000 annually], ethnicity [white v. non-white], and marital status [married v. not married]).
In the cross-sectional models, basic psychological needs satisfaction predicted autonomous regulation, which predicted self-reported PA (assessed by the IPAQ-SF), which in turn predicted psychological well-being (combined vitality and life satisfaction). Basic psychological needs satisfaction then directly predicted psychological well-being. As a second step, meaning in life (the presence subscale of the MILQ) was added to the models.
33


Specifically, basic psychological needs satisfaction predicted meaning in life, which then predicted PA and psychological well-being. The original SDT Process Model of Behavior Change was compared to the modified model using comparative fit statistics (AIC and BIC) and change in variance in PA accounted for by the model (R2).
To simplify the longitudinal models, psychological well-being was not included. Two types of longitudinal models were examined. The first was designed to predict absolute levels of PA at 12-weeks. The original SDT Process Model of Behavior Change was modeled first, with baseline basic psychological needs satisfaction predicting 4-week autonomous regulation, and 4-week autonomous motivation predicting 12-week PA. To modify the model, baseline meaning in life was added to the model. Basic psychological needs satisfaction at baseline was used to predict baseline meaning in life, which was then used to predict 12-week PA.
A second type of longitudinal model was used to predict residualized change in PA from baseline to 4-weeks and from 4-weeks to 12-weeks. In the base SDT Process Model of Behavior Change, psychological needs satisfaction at time 1 predicted change in autonomous regulation (between baseline and 4-weeks), which predicted change in self-reported PA (assessed by the IPAQ-SF from baseline to 4-weeks and from 4-weeks to 12-weeks). The modified SDT model with meaning in life added was then assessed for goodness of fit. Because meaning in life was considered relatively stable, baseline meaning in life was used to predict 4-week and 12-week changes in PA. The modified SDT model with the addition of meaning in life was compared to the model without meaning in life to determine whether the addition of meaning in life improved the prediction of PA. Comparative model fit statistics
34


(AIC and BIC) were used to compare the overall models and a change in R2 determined the extent to which prediction of PA was improved.
Exploratory analyses. To examine the effects of meaning salience on other 12-week outcomes (e.g., PA as measured by the IPAQ-SF, body size and composition, physical fitness, and psychological well-being), meaning salience over the 28 days was averaged for each person in the self-monitoring arm. Additionally, the standard deviation in meaning salience was calculated for each person. Pearson correlations were used to examine the relationships between the average and standard deviation of meaning salience in relation to these outcomes. Subsequently, a series of linear regression models were used to examine the relationships between meaning salience (average and variability) and fitness center visits, total PA, and M/P (MILQ and LET) at 12-weeks. All models controlled for gender, age, race, employment status, marital status, and income.
Pearson correlations were also used to assess the relationship of global M/P (MILQ and LET, respectively) in relation to body size and composition, physical fitness, PA intensity, and the number of visits at the fitness center. Finally, Pearson correlations were used to assess the relationships between the exploratory M/P and PA connection variable to the same outcomes.
35


CHAPTER III
RESULTS
A total of 160 adults consented to participate in the study and completed baseline questionnaires. Demographic and health characteristics of the entire sample and by arm are presented in Table 2. On average, participants were 43.3 years old (SD = 11.4, range = 30-72). The sample was predominantly female (76.9%) and highly educated; nearly one half having completed a graduate or professional degree (48.7%) and an additional third had completed a 4-year college degree (34.8%). Additionally, the vast majority of the sample was employed (80.6% employed full-time) and reported high annual household income (57.5% reported an annual income of $80,000 or more). The sample was not very religious; nearly one third of the sample (31.3%) reported that their religious affiliation was I consider myself spiritual, but not religious.
Nearly one third (31.3%) of the sample reported that it had been 6 months or less since they were regularly physically active, and only a small percentage (1.9%) reported that they had never been regularly physically active. The majority of participants (90.0%) reported that there had been times in their lives when they were regularly physically active for at least 2 months and subsequently not physically active for at least 3 months. When asked how many times in their adult lives that had happened, the median response was 4 {range =1-50 times). The most endorsed reason for stopping regular PA was lack of time due to work (51.9%), followed by lack of interest in PA (28.8%) and personal stress (28.8%). More than two thirds (68.8%) reported that they had participated in athletics in their lifetimes.
36


Table 2
Baseline Demographics for the Entire Sample and Between Groups
Variable All (N= 160) n (%) Self- Monitoring (n = 80) n (%) Control (n = 80) n (%) x\p
Female 123 (76.9) 62 (77.5) 61 (76.3) 0.04, .85
Race/Ethnicity 1.55, .82
African American 12(7.5) 7(8.8) 5 (6.3)
Asian 9 (5.6) 6(7.5) 3 (3.8)
Hispanic or Latino/a 10(6.3) 5 (6.3) 5 (6.3)
White, non-Hispanic 118 (73.8) 57(71.3) 61 (76.3)
Mixed or other 11 (6.9) 5 (6.3) 6(7.5)
Marital Status (n= 159) (n = 79) 10.00, .04
Single, never married 41 (25.8) 24 (30.4) 17(21.3)
Currently married 82 (51.6) 32 (40.5) 50 (62.5)
Cohabiting or in a long- 13 (8.2) 10(12.7) 3 (3.8)
term relationship
Divorced 18 (11.3) 11 (13.9) 7(8.8)
Widowed 6(3.1) 2(2.5) 4(5.0)
Employment Status 7.37, .39
Employed full-time 129 (80.6) 62 (77.5) 67 (83.8)
Employed part-time 9 (5.6) 5 (6.3) 4(5.0)
Retired 6 (3.8) 2(2.5) 4(5.0)
Partially disabled 2(1.3) 1(1.3) 1(1.3)
Unemployed 2(1.3) 2(2.5) 0 (0.0)
Student 10(6.3) 6(7.5) 4(5.0)
Homemaker 2(1.3) 2(2.5) 0 (0.0)
Education (n= 158) (n = 78) 4.90, .30
High school or equivalent 5 (3.2) 2(2.5) 3 (3.9)
Some college 13 (8.2) 7(8.8) 6(7.7)
2-year college degree 8(5.1) 4(5.0) 4(5.1)
4-year college degree 55 (34.8) 34 (42.5) 21 (26.9)
Graduate or professional 77 (48.7) 22 (41.3) 44 (56.4)
degree
Annual household income 11.06, .05
Less than $20,000 7 (4.4) 7(8.8) 0 (0.0)
$20,000 $39,999 18 (11.3) 12(15.0) 6(7.5)
$40,000 $59,999 25 (15.6) 12(15.0) 13 (16.3)
$60,000 $79,999 18 (11.3) 9 (11.3) 9 (11.3)
$80,000 $99,999 31 (19.4) 15 (18.8) 16 (20.0)
$100,000 or more 61 (38.1) 25 (31.3) 36 (45.0)
37


Table 2 cont.
Variable All (N= 160) n (%) Self- Monitoring (n = 80) n (%) Control (n = 80) II (%) I.P
Religious Affiliation
Catholic 25 (15.6) 17(21.3) 8 (10.0) 3.84, .05
Protestant Christian 36 (22.5) 17(21.3) 19(23.8) 0.14, .71
Latter Day Saint 1 (0.6) 0 (0.0) 1(1.3) 1.00, .32
Jewish 9 (5.6) 1(1.3) 8 (10.0) 5.77, .02
Muslim 2(1.3) 2(2.5) 0 (0.0) 2.03, .15
Hindu 1 (0.6) 0 (0.0) 1(1.3) 1.00, .32
Buddhist 1 (0.6) 0 (0.0) 1(1.3) 1.01, .32
Atheist 4 (2.5) 2(2.5) 2(2.5) 0.00, 1.00
I consider myself spiritual 50 (31.3) 23 (28.8) 27 (33.8) 0.47, .50
but not religious
None 25 (15.6) 15 (18.8) 10(12.5) 1.19, .28
Other 12(7.5) 7(8.8) 5 (6.3) 0.36, .55
Family History of 78 (48.8) 45 (56.3) 43 (53.8) 1.60, .21
Coronary Heart Disease
Family History of Stroke (n= 159) (n = 79) 0.05, .82
67 (42.1) 34(43.0) 33 (41.3)
Family History of 96 (60.0) 45 (56.3) 51 (63.8) 0.94, .33
Hypertension
Family History of Diabetes 76 (47.5) 38 (47.5) 38 (47.5) 0.00, 1.0
Family History of Obesity 74 (46.3) 31 (38.8) 43 (53.8) 3.62, .06
Smoking Status 3.06, .22
Non-smoker 151 (94.4) 74 (92.5) 77 (96.3)
Former smoker (quit in the 3(1.9) 3 (3.8) 0 (0.0)
last 6 months)
Current smoker 6 (3.8) 3 (3.8) 3 (3.8)
Time Since Last Regular PA 4.01, .68
Less than 6 months 50 (31.3) 27 (33.8) 23 (28.8)
6 months 1 year 39 (24.4) 20 (25.0) 19(23.8)
1-2 years 24 (15.0) 13 (16.3) 11 (13.8)
2-5 years 30 (18.8) 15 (18.8) 15 (18.8)
5-10 years 8 (5.0) 3 (3.8) 5 (6.3)
More than 10 years 6 (3.8) 1(1.3) 5 (6.3)
Never regularly active 3(1.9) 1(1.3) 2 (2.5)
Why stopped most recent
activity
Lack of time because of 83 (51.9) 37 (46.3) 46 (57.5) 2.03, .15
work
Lack of time because of 22 (13.8) 12(15.0) 10(12.5) 0.21, .65
household duties
38


Table 2 cont.
Variable All (N= 160) n (%) Self- Monitoring (n = 80) n (%) Control (n = 80) II (%) I.P
Why stopped most recent activity
Lack of time because of 29 (18.1) 15 (18.8) 14(17.5) 0.04, .84
children
Lack of time because of 12(7.5) 6(7.5) 6(7.5) 0.00, 1.0
social activities
Lack of time because of 12(7.5) 5 (6.3) 7(8.8) 0.36, .55
spouse
Lack of money 26 (16.3) 15 (18.8) 11 (13.8) 0.73, .39
Lack of facilities 15 (9.4) 10(12.5) 5 (6.3) 1.84, .18
Lack of PA partner 26 (16.3) 12(15.0) 14(17.5) 0.18, .67
Lack of interest in PA
Health problems 48 (30.0) 19(23.8) 29 (36.3) 2.98, .08
Injury
Season or weather change 17 (10.6) 8 (10.0) 9 (11.3) 0.07, .80
Personal stress 25 (15.6) 14(17.5) 11 (13.8) 0.43, .51
Other 34 (21.3) 17(21.3) 17(21.3) 0.00, 1.0
46 (28.8) 22 (27.5) 24 (30.0) 0.12, .73
18 (11.3) 5 (6.3) 13 (16.3) 4.01, .04
Participated in athletics 110(68.8) 59 (73.8) 51 (63.8) 1.86, .17
Note. Sample size is only reported in a cell if there are missing data.
There were no differences in age between the self-monitoring (M= 43.4, SD = 10.7) and control (M= 43.1 ,SD= 12.0) groups at baseline, t (158) = 0.20,/) = .84. However, the control group was significantly more likely to be married (p = .04) and report a Jewish religious affiliation (p = .02) and less likely to report a Catholic religious affiliation (p = .05) than the self-monitoring group. There was also a trend towards the control group reporting a greater annual income than the self-monitoring group (p = .05). There were no other significant differences between the self-monitoring and control groups at baseline.
39


Study Flow & Missing Data Analysis
Of the 160 participants that consented and completed baseline, 140 (87.5%) participants completed the 4-week follow-up survey and 140 (87.5%) participants completed the 12-week follow-up survey and visit (see Figure 4 for CONSORT diagram). Seven participants (4.3%) withdrew from the study. Two stated that they did not have enough time, one did not provide a reason, and four stated other reasons for withdrawal (e.g., stressful life events, disliking the online survey questions or format). Eleven participants (6.8%) were lost to follow-up and could not be reached to complete the study after multiple contact attempts. Participants in the self-monitoring group were not significantly more likely to miss the 4-week assessment, %2 (1) = 1.11 ,p= .29, or the 12-week assessment, %2 (1) = 0.25,p= .62, compared to the control group.
Aims 1 & 3: Baseline, 4-week, and 12-week missing data. Participants who missed the 4-week assessment were compared to participants who completed the 4-week survey. Those who did not complete the survey reported significantly greater perceived competence in PA at baseline (M= 11.7, SD = 3.5) than those who did complete the survey (M= 9.6, SD = 3.3, t [154] = -2.84,/> = .005). Moreover, those who did not complete the survey were significantly less satisfied with their lives at baseline (M= 19.6, SD = 8.0) than those who completed the survey (M= 23.8, SD = 6.5, t [157] = 2.35,p = .019). Those who did not complete the survey reported marginally more total PA at baseline (square-rooted METs, M = 40.0, SD = 25.2) compared to those who completed the survey (M= 31.3, SD = 17.7, t [158] = 1.78,/) = .07). There was also a trend that those who did not complete the 4-week survey were more likely to make less than $40,000 per year than those who completed the survey, %2 (1) = 3.23,p = .07.
40


The patterns for missing 12-week surveys were also examined. Those who did not complete the 12-week survey were more likely to be older (M= 49.8 years, SD = 11.4 years) than those who completed the survey (M= 42.4, SD = 11.1, t [158] = 2.66,p = .009). They were also reported significantly less vitality (M= 20.7, SD = 7.1) and life satisfaction (M= 19.5, SD = 8.0) at baseline than those who completed the survey (Vitality: M= 24.3, SD =
6.6, t [156] = 2.18,p= .03; Life Satisfaction: M= 23.9, SD = 6.5, t [157] = 2.61,p = .01). There was a trend that those who did not complete the survey were less likely to be married than those who completed the survey, %2 (1) = 3.53,p= .06.
Aim 2: Daily missing data. With 80 participants randomized to receive the daily surveys, there were 2240 possible surveys distributed. Of these, 1813 (80.9%) surveys were accessed (some questions were answered), and 1691 (75.5%) were completed (all questions were answered). Each participant completed a median of 24 of the 28 daily surveys (85.7%), with a range of 2 to 28 surveys completed.
Study Variables at Baseline, 4 weeks, and 12 weeks
Study variables were examined at baseline, 4-weeks, and 12-weeks to determine if there were longitudinal patterns or trends over time. Descriptive statistics on these variables are presented in Table 3. Of note, PA variables were all significantly positively skewed and leptokurtic; thus, a square root transformation was used to normalize the data. Trajectories of individual participants PA over the 12 weeks is plotted in Figure 5. The spaghetti plot suggested that that average person increases in PA from baseline to 4-weeks slightly, and then PA levels flatten from 4-weeks to 12-weeks. However, some individuals show drastic increases in PA from baseline to 4-weeks and drastic decreases from 4-weeks to 12-weeks. The plot demonstrated that the greatest variability in PA was at 4-weeks.
41


Figure 4. CONSORT Diagram.
42


Table 3
Descriptive Statistics of Study Variables at Baseline, 4-weeks, and 12-weeks
Baseline
Variable a n M SD Min Max Skew Kurt
Total PA1 160 32.2 18.6 0.1 97.2 0.95 1.35
Vigorous PA1 160 15.2 17.1 0.1 69.3 0.87 -0.02
Moderate PA1 160 10.2 11.8 0.1 60.0 1.38 2.31
Walking Activity1 160 20.3 14.3 0.1 64.5 1.11 1.34
Fitness Center Attendance 159 2.4 1.7 0 7 0.49 -0.32
Physical Fitness 153 48.1 21.0 0 93.8 -0.12 -0.61
Meaning in Life .92 159 26.6 5.8 5 35 -0.86 0.71
Purpose in Life .93 157 21.5 8.1 12 30 -0.28 -0.82
Basic Psychological Needs .85 154 29.8 8.1 11 48 0.00 -0.29
Satisfaction
Autonomy .80 160 11.9 3.6 4 20 0.02 -0.45
Competence .75 156 9.5 3.4 4 20 0.35 -0.32
Relatedness .88 158 8.5 3.5 3 15 0.01 -0.97
Autonomous Regulation .92 157 9.2 3.2 0 16 -0.17 -0.11
Controlled Regulation .79 159 4.4 2.7 0 11 0.41 -0.47
Psychological Well-being2 .92 157 47.2 12.3 11 72 -0.34 -0.35
Subjective Vitality .88 158 23.9 6.7 6 41 -0.03 -0.05
Life Satisfaction .90 159 23.4 6.8 5 35 -0.58 -0.46
Depressive Symptoms .85 154 5.6 4.4 0 24 1.29 2.40
Body Mass Index 156 29.9 7.6 17.9 68.1 1.63 4.74
Body Fat % 151 33.4 8.9 9.8 50.0 -0.28 -0.56
Physical Fitness 153 48.1 21.0 0 93.8 -0.12 -0.61
4 Weeks
Variable a n M SD Min Max Skew Kurt
Total PA1 135 42.7 22.2 0.1 129.5 0.45 1.35
Vigorous PA1 140 23.7 20.4 0.1 91.7 0.47 -0.09
Moderate PA1 138 16.7 14.5 0.1 71.0 1.25 2.86
Walking Activity1 137 24.5 16.9 0.1 64.5 0.85 0.31
Fitness Center Attendance 159 1.4 1.5 0 7 1.22 1.57
Meaning in Life .94 139 26.6 6.2 5 35 -0.34 -0.63
Purpose in Life .85 138 24.4 4.0 12 30 -1.09 1.17
Basic Psychological Needs .90 138 33.8 8.91 17 53 -0.03 -0.82
Satisfaction
Autonomy .80 143 13.0 3.4 4 20 -0.07 -0.48
Competence .87 140 11.6 3.8 4 20 0.08 -0.56
Relatedness .86 142 9.2 3.3 3 15 -0.25 -0.75
Autonomous Regulation .94 129 9.9 3.3 1.67 16 -0.02 -0.71
Controlled Regulation .78 135 4.1 2.4 0 11 0.31 -0.46
Psychological Well-being2 .93 138 51.6 12.6 18 77 -0.39 -0.58
43


Table 3 cont.
Variable a n M SD Min Max Skew Kurt
Subjective Vitality .91 136 27.3 7.1 9 42 -0.12 -0.63
Life Satisfaction .90 138 24.3 6.4 9 35 -0.62 -0.30
Depressive Symptoms .85 138 4.0 3.9 0 18 1.38 1.79
12 Weeks
Variable a n M SD Min Max Skew Kurt
Total PA1 140 42.1 21.4 0.1 116.0 0.52 0.60
Vigorous PA1 141 22.8 19.9 0.1 100.4 0.70 0.96
Moderate PA1 141 15.9 14.2 0.1 64.8 0.98 1.22
Walking Activity1 140 24.2 16.4 0.1 64.5 0.62 0.01
Fitness Center Attendance 159 1.1 1.5 0 7 1.44 1.69
Physical Fitness 127 52.7 20.2 12.5 100 -0.07 -0.55
Meaning in Life .89 141 27.4 5.3 9 35 -0.94 0.96
Purpose in Life .87 141 25.1 4.1 12 30 -1.21 1.42
Basic Psychological Needs Satisfaction .92 137 34.7 9.7 13 55 -0.41 -0.54
Autonomy .87 139 13.4 3.8 4 20 -0.39 -0.63
Competence .87 142 12.0 3.8 4 20 -0.20 -0.03
Relatedness .88 140 9.3 3.4 3 15 -0.26 -0.89
Autonomous Regulation .92 136 10.4 2.9 0.67 16 -0.37 0.53
Controlled Regulation .82 139 4.2 2.6 0 12 0.47 -0.29
Psychological Well-being2 .91 139 54.5 11.2 24 74 -0.53 -0.18
Subjective Vitality .89 141 28.4 6.7 7 42 -0.29 -0.07
Life Satisfaction .86 139 26.1 5.5 10 35 -0.75 0.04
Depressive Symptoms .82 132 3.8 3.5 0 20 1.50 3.30
Body Mass Index 135 29.7 7.2 19.0 64.9 1.55 4.45
Body Fat % 131 33.1 8.5 8.7 49.2 -0.29 -0.47
Physical Fitness 127 52.7 20.2 12.5 100 -0.07 -0.55
Square-root transformed metabolic units. Combined subjective vitality and life satisfaction scales. Note: n = number of complete; M= Mean; SD = Standard Deviation; Min = Minimum; Max = Maximum; Skew = Skewness; Kurt = Kurtosis.
44


Spaghetti Plot of Physical Activity
Figure 5. Trajectories of PA Over Time.
Aim 1. Effect of a Self-monitoring of M/P, Mood, and PA Self-Monitoring on PA.
PA. A mixed model examining differences between the self-monitoring and control groups over the 12 weeks on PA, as measured by the IPAQ-SF, revealed that the group-by-time interaction was not significant, F (1, 136) = 0.03, p = .97. Similarly, there were no significant differences in total PA by study group, F (1, 143) = 0.00, p = .99. There was a significant increase in PA in both groups over the first four weeks, F (1, 134) = 14.55, p = .0002. On average, the sample increased by 128.1 minutes of PA per week from baseline to 4 weeks. From four weeks to 12 weeks, there was no significant difference in PA, F (1, 135) =
45


0.12,p= .69. Thus, the gains in PA from baseline to four weeks were, on average, maintained by the sample (see Figure 6 and Table 4).
Time Plot of Physical Activity
By Treatment Group
50 1
45
30
25 -i.........................................................................................................I
0 4 8 12
Time (in Weeks)
Group Assignment o o o Self-Monitoring * Control
Figure 6. Mean PA by Condition from Baseline to 12-weeks.
Secondary analyses revealed there were no significant group differences on type of activity, including vigorous activity, A (1, 148) = 0.13,p= .72, moderate activity, A (1, 148) = 1.67, p = .20, and walking, F (1, 148) = 0.23, p = .63. Similar patterns to total PA were observed for vigorous activity, moderate activity, and walking over time. There were significant increases from baseline to 4-weeks (ps < .01) and those gains were maintained from 4-weeks to 12-weeks (ps > .32).
46


Table 4
Means and Standard Deviations of Study Variables by Condition
Self-Monitoring
Baseline 4 weeks 12 weeks
Total PA* 32.4(19.6) 41.9(42.3) 42.3 (21.5)
Vigorous PA* 15.4(18.4) 23.0(20.1) 24.1 (22.2)
Moderate PA* 9.2 (11.3) 16.3 (14.2) 13.9(12.6)
Walking Activity* 20.5 (15.1) 23.0(15.8) 24.0(15.8)
Fitness Center Attendance 2.5 (1.7) 1.6 (1.6) 1.3 (1.7)
Physical Fitness 50.2 (20.5) 53.9(19.8)
Meaning in Life 26.2 (5.9) 26.2 (6.2) 26.9 (5.8)
Purpose in Life 24.6 (3.9) 24.2 (4.3) 25.1 (4.3)
Basic Psychological Needs Satisfaction 31.1 (7.6) 34.5 (9.8) 35.5 (10.1)
Autonomous Regulation 9.4 (3.2) 9.8 (3.4) 10.2 (3.2)
Subjective Vitality 24.3 (6.8) 27.4 (7.5) 28.4 (6.7)
Life Satisfaction 23.6 (6.5) 24.0 (6.4) 25.5 (6.1)
Depressive Symptoms 5.1 (3.7) 3.9 (4.0) 3.7 (3.4)
Control
Baseline 4 weeks 12 weeks
Total PA* 32.0(17.7) 43.5 (23.9) 41.9(21.5)
Vigorous PA* 15.0(15.9) 24.4 (20.7) 21.4(17.2)
Moderate PA* 11.1 (12.3) 17.0(14.8) 17.8 (15.5)
Walking Activity* 20.1 (13.6) 26.0(17.6) 24.5 (17.2)
Fitness Center Attendance 2.3 (1.8) 1.2 (1.3) 0.9 (1.2)
Physical Fitness 46.0(21.5) 51.6(20.7)
Meaning in Life 27.0 (5.8) 26.9 (6.2) 28.0 (4.8)
Purpose in Life 24.2 (3.8) 24.5 (3.8) 25.1 (3.8)
Basic Psychological Needs Satisfaction 28.5 (8.5) 33.2 (8.1) 33.8 (9.2)
Autonomous Regulation 9.0 (3.2) 9.9 (3.2) 10.6 (2.6)
Subjective Vitality 23.5 (6.7) 27.2 (6.9) 28.5 (6.4)
Life Satisfaction 23.1 (6.1) 24.6 (6.4) 26.6 (4.9)
Depressive Symptoms 6.1 (5.1) 4.1 (3.9) 3.9 (3.5)
* Square-root transformed variable. Note. PA = Physical activity.
The group-by-time interaction was not significant for fitness center attendance, F (2, 156) = 0.79, p = .46. However, there was a significant reduction in fitness center visits from the first week (M= 2.4, SD = 1.7) to the fourth week (M= 1.4, SD = 1.5), F (l, 157) = 39.58, p < .0001, and again from the fourth week to the twelfth week (M= 1.1 ,SD= 1.5), F (1, 157) = 4.05,/) = .046 (see Figure 7). Similarly, there was not a significant group-by-time
47


interaction on physical fitness, F (1, 132) = 0.44, p = .51. There was a significant increase in physical fitness over the 12-weeks, F (1, 134) = 7.20, p = .008. On average, participants increased their physical fitness wellness scores by 4.6%.
Time Plot of Fitness Center Visits
By Treatment Group
8
7
6
Time (in Weeks)
Group Assignment o o o Self-Monitoring * Control
Figure 7. Mean Fitness Center Visits by Condition from Baseline to 12-weeks.
M/P. Meaning in life did not differ by group at any of the three time points, F (1,
146) = 2.08, p = .15, or over time, F (2, 137) = 1.52,p = .22. Similarly, there were no differences between the groups in purpose in life, F (1, 152) = 0.03, p = .87. However, there was a significant increase in purpose in life from baseline (M= 24.4, SD = 3.8) to 12-weeks (M= 25.1, SD = 4.1) for both groups, F (2, 139) = 3.38,p= .04 (see Figure 8).
48


Time Plot of Purpose in Life
By Treatment Group
28
o ]
a. I
3
CL
22 -
2 1....................................................................................................1
0 4 8 12
Time (in Weeks)
Group Assignment o o o Self-Monitoring * Control
Figure 8. Mean Purpose in Life by Condition from Baseline to 12-weeks.
SDT mediators. There was no difference between groups in basic psychological needs satisfaction over time, F (2, 142) = 0.20, p = .82. However, basic psychological needs satisfaction significantly increased in both groups over the first four weeks, F (1, 145) = 14.44, p = .0002. There was no significant difference in basic psychological needs satisfaction from 4-weeks to-12 weeks, F (1, 132) = 0.58, p = 0.57. Thus, gains made in the first four weeks were, on average, maintained from 4 weeks to 12 weeks (see Figure 9). A similar pattern was observed with changes in autonomous regulation. There were no significant differences between groups, F (1, 159) = 0.01,/? = .91; however, autonomous
49


Basic Psychological Needs
regulation significantly increased from baseline to 4 weeks, F (1, 142) = 16.49,/? < .0001 and marginally increased from 4 weeks to 12 weeks, F(1, 131) = 3.76,/? = .05 (see Figure 10).
Time Plot of Basic Psychological Needs
By Treatment Group
40
25
20-......................................................................... ..................................
0 4 8 12
Time (in Weeks)
Group Assignment o o o Self-Monitoring * Control
Figure 9. Mean Basic Psychological Needs Satisfaction by Condition from Baseline to 12-weeks.
50


Time Plot of Autonomous Regulation
By Treatment Group
15 -
13
o
jf 11
3
0 4 8 12
Time (in Weeks)
Group Assignment o o o Self-Monitoring * Control
Figure 10. Mean Autonomous Regulation by Condition from Baseline to 12-weeks.
Psychological well-being. There were no significant differences between groups in vitality, F(l, 154) = 0.08, p =.77, life satisfaction, F (1, 152) = 0.18,/? = .67, or depressive symptoms, F (l, 155) = 0.31 ,/? = .58, over the 12-weeks. However, vitality significantly increased over the first four weeks, F (1, 137) = 28.78,/? < .0001, and marginally increased from 4-weeks to 12-weeks, F (1, 136) = 3.33,/? = .07 (see Figure 11). Life satisfaction marginally increased from baseline to 4-weeks, F (1, 140) = 3.68,/? = .06, and significantly increased from 4-weeks to 12-weeks, F (l, 136)= 15.0,/? = .0002 (see Figure 12).
Depressive symptoms significantly decreased from baseline to 4-weeks, F (1, 147) = 16.23,/?
51


Vitality
< .0001. There was no difference in depressive symptoms from 4-weeks to 12-weeks, F (1, 136) = 0.52, p = A1 (see Figure 13).
Time Plot of Subjective Vitality
By Treatment Group
30
15-1 ............................................................................................................................ I
0 4 8 12
Time (in Weeks)
Group Assignment o o o Self-Monitoring * Control
Figure 11. Mean Vitality by Condition from Baseline to 12-weeks.
52


Satisfaction with Life
Figure 12. Mean Life Satisfaction by
Condition from Baseline to
12-weeks.
53


Time Plot of Depression
By Treatment Group
Time (in Weeks)
Group Assignment o o o Self-Monitoring * Control
Figure 13. Mean Depressive Symptoms by Condition from Baseline to 12-weeks.
Aim 2. Relationship of Daily M/P Salience to Daily PA.
For Aim 2, analyses included the 80 participants who were randomized to record daily meaning salience, mood, and PA for 4 weeks (28 days). Descriptive statistics for variables in Aim 2 are presented in Table 5, and correlations among the variables are presented in Table 6.
54


Table 5
Descriptive Statistics of Variables in Aim 2
Variable n M(SD) Min-Max a
Thoughts of Meaning Scale 1762 45.2 (12.5) 10-70 .94
Positive Affect 1752 24.9 (6.7) 8-40 .91
Negative Affect 1764 8.5 (3.6) 5-23 .80
PA Minutes 1795 41.2 (51.6) 0-480
PA Intensity 1280 12.1 (2.6) 6-19
Table 6
Pearson Correlations among the Variables in Aim 2
1 2 3 4
1. Thoughts of Meaning Scale
2. Positive Affect .66*
3. Negative Affect -.15* -.36*
4. PA Minutes .13* .11* -.14*
5. PA Intensity .05 .03 -.10* .33*
*p < .001
In a mixed model with days nested within participants and controlling for daily mood, meaning salience was significantly and positively associated with daily minutes of PA (J3 = 21, p < .0001). Participants who reported greater daily meaning salience also reported more minutes of activity on the same days. Positive affect was not significantly associated with daily PA when controlling for both meaning salience and negative affect (J3 = -.01, p = .94). Controlling for positive affect and meaning salience, negative affect was negatively associated with PA (J3= -A4,p < .0001); on days when participants experienced more
negative affect, they also engaged in fewer minutes of PA. The mixed model was repeated with PA intensity as the dependent variable. On days that participants reported engaging in
55


PA and controlling for positive and negative affect, meaning salience was associated with increased intensity of PA (J3= .21 ,P< .0001). Thus, greater salience of meaning was associated with more intense levels of PA.
In a multilevel logistic regression model predicting fitness center attendance and controlling for positive and negative mood, daily meaning salience was related to significantly greater likelihood of visiting the fitness center, Odds Ratio (OR) (1608) = 1.48 (95% Cl = 1.18, 1.86),p = .0008. For every standard deviation increase in daily meaning salience, participants were 48% more likely to visit the fitness center on that day. Neither positive affect, OR = 1.004 (95% Cl = 0.81, 1.24),/> = .97, or negative affect, OR = 0.86 (95% Cl = 0.72, 1.02),p= .08, were significantly related to same day fitness center visits. Aim 3. Examining the SDT Process Model of PA Adoption with and without M/P.
Prior to examining the path analytic models, bivariate Pearson correlations were used to examine the associations among Aim 3 variables. Correlations are presented in Table 7.
Model la: Cross-sectional baseline model. At baseline, the SDT Process Model of Behavior Change had acceptable fit (see Table 8 for model fit statistics). The model and standardized path coefficients are presented in Figure 14. Basic psychological needs satisfaction was significantly and positively related to autonomous regulation for PA, which was significantly and positively related to PA. PA was not significantly related to psychological well-being, but psychological needs satisfaction was significantly and positively related to PA. The base SDT Process Model accounted for 16% of PA and 18% of psychological well-being at baseline.
56


Table 7
Pearson Correlations among Aim 3 Model Variables
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Baseline 1. BPNES
2. AR .51
3. MIL .17 .22
4. PA .27 .33 .02
5. PWB 4-week .31 .33 .63 .14
6. BPNES .23 .13 .18 .12 .07
7. AR .28 .74 .15 .22 .13 41
8. MIL .03 .19 .73 .12 .54 .23 .28
9. PA .03 .03 .19 .27 .02 .23 .12 .07 --
10. PWB 12-week .15 .28 .60 .17 .73 .30 .36 .73 .12
11. BPNES .33 .25 .03 .22 .03 .63 .36 .03 .17 .21
12. AR .36 .70 .06 .34 .09 .39 .81 .06 .18 .24 .55
13. MIL .02 .13 .57 .15 .45 .20 .16 .69 .20 .63 .17 .10
14. PA .06 .09 .00 26 -.02 .20 .15 .11 .42 .17 .30 .25 .06
15. PWB .13 .16 .43 .21 .62 .28 .22 .54 .11 .77 .40 .26 .64 .19
Note. BPNES = Basic Psychological Needs in Exercise Scale; AR = Autonomous Regulation; MIL = Meaning in Life; PA = PA; PWB = Psychological Well-Being. Significant (p < .05) correlations are in bold.
57


Table 8
Model Fit Statistics
Model 3 X df P RMSEA (90% Cl) CFI SRMR AIC BIC
Cross-sectional Models
la. Baseline 16.44 9 .06 .07 (.00-.13) .92 .03 3242.25 3315.14
lb. Mod. Baseline 20.98 16 .18 .05 (.00-.09) .97 .04 4153.63 4244.74
2a. 4-week 19.13 9 .02 .09 (.03-.15) .79 .04 2814.33 2884.05
2b. Mod. 4-week 25.20 16 .07 .07 (.00-.11) .93 .04 3591.68 3678.83
3 a. 12-week 19.32 9 .02 .09 (.03-.15) .89 .04 2835.89 2905.97
3b. Mod. 12-week 31.31 16 .01 .08 (.04-.13) .91 .05 3613.32 3700.92
Longitudinal Models
4a. Absolute PA 9.56 8 .30 .04 (.00-. 11) .94 .03 1848.56 1886.80
4b. Mod. absolute 15.66 16 .48 .00 (.00-.07) 1.00 .04 2825.68 2877.31
PA
5a. Change in PA 30.24 21 .09 .05 (.00-.09) .96 .04 4958.78 5077.22
5b. Mod. change 37.13 30 .17 .04 (.00-.08) .97 .04 5932.20 6068.87
in PA
Note. y2 = chi-square test of model fit; df= degrees of freedom; p =/i-value; RMSEA = root mean square error of approximation; CFI = comparative fit index; SRMR = standardized root mean square residual; AIC = Akaikes Information Criterion; BIC = Bayesian Information Criterion; Mod. = Modified; PA = Physical Activity
Basic Autonomous
Psychological 5 Y *** Regulation
Needs Satisfaction R2 = .26
***p < .001; **p < .01; *p < .05
Figure 14. Cross-sectional SDT Process Model of Behavior Change at Baseline.
Model lb: Modified cross-sectional baseline model. Meaning was added to the
model, and the model fit the data very well (see Table 8). The modified model is presented in Figure 15. The associations found in the base SDT Process Model were maintained. Basic
58


psychological needs satisfaction was positively related to experience of meaning in life. Meaning was not significantly related to PA at baseline, but it was significantly and positively associated with psychological well-being. There was no change in R2 from the base model in PA, but the variance in psychological well-being accounted for by the model significantly increased to 46%. Thus, meaning improved the prediction of psychological well-being but not of PA at baseline. ***
***p < .001; **p < .01; *p < .05
Figure 15. Cross-sectional Modified SDT Process Model of Behavior Change at Baseline.
Model 2a: Cross-sectional 4-week model. The base SDT Process Model of Behavior Change at week four did not fit the data well (see Table 8). The path model is presented in Figure 16. Similar to the model at baseline, basic psychological needs satisfaction was positively and significantly related to both autonomous regulation and to psychological well-being. However, autonomous regulation was not significantly related to
59


PA at 4-weeks, and PA was not significantly related to psychological well-being. One possible reason that the model did not fit the data well is that basic psychological needs satisfaction was significantly correlated with PA at 4-weeks (see Table 7), but this association was not fully accounted for by the mediation through autonomous regulation. The model accounted for 12% of the variance in PA (with no significant predictors of PA) and 13% of the variance in psychological well-being.
***p < .001; **p < .01; *p < .05
Figure 16. Cross-sectional SDT Process Model of Behavior Change at 4-weeks.
Model 2b: Modified cross-sectional 4-week model. The modified model is presented in Figure 17, and the model fit the data well. Similar to baseline, basic psychological needs satisfaction was positively associated with meaning, but meaning was not significantly related to PA. Thus, the model did not significantly account for more variance in PA. Meaning was strongly and positively associated with psychological wellbeing, and the variance in psychological well-being accounted for by the model increased substantially to 58%.
60


***p < .001; **p < .01; *p < .05
Figure 17. Cross-sectional Modified SDT Process Model of Behavior Change at 4-weeks.
Model 3a: Cross-sectional 12-week model. The base SDT Process Model of Behavior Change at 12-weeks was a poor fit to the data. The path model is presented in Figure 18. Basic psychological needs satisfaction was significantly and positively related to autonomous regulation, which was significantly and positively related to PA. PA was not significantly related to psychological well-being. Similar to the cross-sectional 4-week model, one possible reason that the model did not fit the data well is that basic psychological needs satisfaction was significantly correlated with PA at 12-weeks (see Table 7), but again this association was not fully accounted for by the mediation through autonomous regulation. The model accounted for 31% of the variance in autonomous regulation, 18% of the variance in PA, and 20% of the variance in psychological well-being.
61


***p < .001; **p < .01; *p < .05
Figure 18. Cross-sectional SDT Process Model of Behavior Change at 12-weeks.
Model 3b: Modified cross-sectional 12-week model. Meaning in life was added to a modified 12-week model (see Figure 19), and the model fit was acceptable. Basic psychological needs satisfaction was significantly and positively associated with presence of meaning, and presence of meaning was significantly and positively associated with psychological well-being. However, meaning was not significantly related to PA. The modified model accounted for 31% of the variance in autonomous regulation, 21% of the variance in PA, and 54% of the variance in psychological well-being. Thus, the modified model increased the proportion of variance in PA accounted for by the predictors by 3%; however, this change is not significant because the association between meaning in life and PA was not significant.
62


***p < .001; **p < .01; *p < .05
Figure 19. Cross-sectional Modified SDT Process Model of Behavior Change at 12-weeks.
Model 4a: Longitudinal model predicting absolute levels of PA at 12-weeks.
Because PA was not significantly associated with psychological well-being in the cross-sectional models, psychological well-being was not included in the longitudinal model. The base SDT Process Model was an excellent fit to the data (see Table 8 and Figure 20). Basic psychological needs satisfaction at baseline was significantly and positively related to autonomous regulation at 4-weeks. Those who reported greater autonomous regulation at 4-weeks also reported significantly more PA at 12-weeks. The model accounted for 18% of the variance in PA at 12-weeks.
63


***p < .001; **p < .01
Figure 20. Longitudinal SDT Process Model of Behavior Change Predicting Absolute PA at 12-weeks.
Model 4b: Modified longitudinal model predicting absolute levels of PA at 12-weeks. The modified longitudinal SDT Process Model is presented in Figure 21 and was an excellent fit to the data. The relationships found in the base SDT longitudinal model were maintained. Meaning in life at baseline was significantly related to basic psychological needs satisfaction at baseline. However, baseline meaning in life was not significantly related to PA at 12-weeks. Adding meaning in life to the model did not significantly increase the amount of variance in 12-week PA accounted for the model.
Model 5a: Longitudinal model predicting change in PA at 4-weeks and 12-weeks. A longitudinal path model was used to assess the SDT Process Model of Behavior Change and predict residualized change in PA from baseline to 4-weeks and from 4-weeks to 12-weeks (see Figure 22). The model was a good fit to the data (Table 8). Basic psychological needs satisfaction at baseline was significantly and positively related to autonomous regulation at baseline, which was significantly and positively related to PA at baseline. Basic psychological needs satisfaction was significantly and negatively related to the change in autonomous regulation from baseline to 4-weeks. Thus, those with lower basic psychological needs satisfaction at baseline experienced greater change in autonomous regulation over the first 4-weeks. Autonomous regulation at 4-weeks was not significantly
64


related to change in PA from baseline to 4-weeks. However, autonomous regulation at 4 weeks was positively and significantly related to change in PA from 4-weeks to 12-weeks. Those with higher autonomous regulation in the first 4-weeks also experienced a greater increase in PA from 4-weeks to 12-weeks. ***
***p < .001; **p < .01; *p < .05
Figure 21. Modified Longitudinal SDT Process Model of Behavior Change Predicting Absolute PA at 12-weeks.
65


***p < .001; **p < .01
Figure 22. Longitudinal SDT Process Model of Behavior Change Predicting Change in PA at 4-weeks and 12-weeks.
Model 5b: Modified longitudinal model predicting change in PA at 4-weeks and 12-weeks. The model was modified to include meaning in life at baseline as a predictor of change in PA at 4-weeks and 12-weeks (see Figure 23). The modified model was an excellent fit to the data. The relationships modeled in the base SDT model remained the same in the modified model. Basic psychological needs satisfaction at baseline was significantly and positively related to presence of meaning in life at baseline. Baseline meaning in life was significantly and positively related to the change in PA from baseline to 4-weeks. Those who
66


had a greater sense of meaning reported a greater change in PA behavior from baseline to 4-weeks. Adding meaning in life increased the amount of variance in change in PA accounted for by the model from 29% to 32%. However, baseline meaning in life was not significantly related to change in PA from 4-weeks to 12-weeks.
Exploratory Outcomes
Average and variability of meaning salience in relation to PA and fitness outcomes. Pearson correlations were used to examine the relationship of the average and standard deviation of the daily meaning salience score in relation to PA, physical fitness, body composition and size, and psychological well-being outcomes. In addition, the relationship between global meaning and purpose and the meaning salience was also assessed. Pearson correlations are presented in Table 9. Those with greater average meaning salience reported higher global meaning in life and purpose in life, greater subjective vitality, better life satisfaction, and fewer depressive symptoms. They also reported more minutes of moderate PA at 4-weeks. Conversely, they reported fewer minutes of vigorous PA at 4-weeks.
67


***p < .001; **p < .01
Figure 23. Modified Longitudinal SDT Process Model of Behavior Change Predicting Change in PA at 4-weeks and 12-weeks.
Those with greater variability in meaning salience during the first 28 days had fewer visits to the fitness center during the first week and during the twelfth week. They also reported fewer minutes of vigorous activity at 12-weeks and more depressive symptoms at baseline, 4-weeks, and 12-weeks. There were no significant relationships between meaning salience and objective outcomes, including BMI, body fat percent, and physical fitness.
68


Table 9
Pearson Correlations between the TOMS and Other Study Outcomes
Thoughts of Meaning Scale (TOMS)
Baseline Week 4 Week 12
Measure M SD M SD M SD
Number of Fitness Center Visits -.14 -.26 .01 -.18 -.01 -.33
Total PA .00 .07 -.04 .04 .06 -.13
Vigorous PA -.15 .08 -.26 -.14 -.04 -.23
Moderate PA .14 .03 .25 -.14 .02 -.11
Walking Activity .11 .04 .21 .20 .22 .13
Body Mass Index -.02 .02 -.01 .03
Body Fat Percent .13 .03 .11 .05
Physical Fitness -.01 .03 .00 -.05
Meaning in Life .43 -.10 .61 -.10 .53 -.04
Purpose in Life .54 .00 .58 -.09 .51 -.18
Subjective Vitality .44 -.11 .54 -.09 .38 -.08
Satisfaction with Life .44 -.04 .53 -.09 .47 -.03
Depressive Symptoms -.19 .25 -.25 .36 -.35 .30
Note.M= Average TOMS over 28 days; SD = Standard Deviation of TOMS over 28 days; PA = Physical Activity. Significant correlations are in bold.
Four multiple regression models were used to examine the relationships between meaning salience and 12-week outcomes, including fitness center visits, total PA, and M/P. Results of these models are presented in Table 10. After controlling for demographics, the variability of meaning salience in the first 28 days was a significant predictor of visits to the fitness center at 12-weeks. Those who had, on average, lower variability of meaning salience had more visits to the fitness center during the 12th week. Conversely, meaning salience was not a significant predictor of total PA at 12-weeks. After controlling for relevant demographics, average meaning salience in the first 28 days was a significant predictor of both global ratings of meaning in life (MILQ) and purpose in life (LET) at 12-weeks. Interestingly, those who reported an annual income of $40,000 or greater reported significantly less meaning in life than those who reported a lower income.
69


Table 10
Regression Models with Meaning Salience Predicting 12-week Outcomes
Predictor Dependent Variable (f >)
Fitness Center Visits Total PA MILQ LET
Female .10 -.04 .02 -.06
Age .11 .10 -.06 .03
White -.07 .09 .07 -.04
Work full-time .33 -.10 .15 .01
Income (>$40,000) -.19 .17 -.42 -.16
Married .17 -.15 .19 .04
Meaning Salience Average .03 .03 .53 .47
Meaning Salience Variability -.40 -.12 .10 -.06
F 2.51 0.67 5.22 3.24
P .019 .716 <.0001 .0038
R2 .22 .08 .40 .29
Note. MILQ = Meaning in Life Questionnaire; LET = Life Engagement Test; PA = Physical Activity. Significant correlations are in bold.
Relationships between global M/P, PA, and objective outcomes. Pearson correlations were used to examine the relationships between global meaning and purpose and objective outcomes (see Table 10). There was some evidence that greater meaning and purpose was related to greater PA prospectively. As previously noted, baseline meaning in life was significantly related to total PA at four weeks, and this was primarily in more minutes of moderate activity. Meaning in life at 4-weeks was also significantly and positively related to fitness center visits during the fourth week, and meaning in life at 12-weeks was positively related to moderate PA at 12-weeks. Those who reported a greater sense of purpose in life at 4-weeks also reported more minutes of walking activity at 12-weeks.
70


Table 11
Pearson Correlations between GlobalM/P, PA, and Objective Outcomes
Baseline Week 4 Week 12
Measure MILQ LET MILQ LET MILQ LET
Baseline
Number of Fitness Center Visits .04 -.06 .00 -.08 .02 -.07
Total PA .02 .08 .12 .14 .15 .22
Vigorous PA -.01 .03 .09 .08 .07 .15
Moderate PA .02 .07 .13 .11 .08 .18
Walking Activity .10 .08 .13 .17 .21 .15
Body Mass Index -.21 -.26 -.17 -.13 -.18 -.09
Body Fat Percent -.11 -.08 -.05 .02 -.11 .01
Physical Fitness .10 .10 -.05 .00 -.01 -.09
Week 4
Number of Fitness Center Visits .15 .09 .18 .11 .12 .13
Total PA .19 .07 .07 .12 .20 .11
Vigorous PA .15 .05 .03 .02 .08 .04
Moderate PA .17 .06 .15 .10 .15 .08
Walking Activity .08 .03 .01 .14 .21 .12
Week 12
Number of Fitness Center Visits .07 .00 -.01 -.08 .04 .10
Total PA .00 .01 .11 .09 .06 .15
Vigorous PA -.11 -.01 -.08 -.07 -.02 .12
Moderate PA .03 -.07 .14 .06 .17 .07
Walking Activity .03 .02 .15 .18 .04 .11
Body Mass Index -.30 -.32 -.21 -.17 -.21 -.09
Body Fat Percent -.21 -.16 -.12 -.05 -.16 .00
Physical Fitness .12 .14 .09 .14 .13 .06
Note. MILQ = Meaning in Life Questionnaire; LET = Life Engagement Test; PA = Physical Activity. Significant correlations are in bold.
There was also evidence that engaging in PA may lead to increased meaning and purpose in life. Those who reported more total PA and moderate activity at baseline reported greater purpose in life at 12-weeks, and those who reported more walking at baseline reported greater purpose in life at 4-weeks and greater meaning in life at 12-weeks. Walking at 4-weeks was also significantly and positively related to meaning in life at 12-weeks.
71


Finally, there was evidence that those with greater BMI reported less meaning and purpose in life at all three time points. Meaning in life at baseline was also significantly and negatively related to body fat percent at 12-weeks. Those with a greater sense of meaning in life when they started their PA programs had a lower body fat percentage 12-weeks later.
Relationships between M/P and PA connection, M/P, PA, and objective outcomes. The exploratory variable examining the strength of participants experienced connections between M/P and PA was assessed in relation to other study variables. Correlations between these measures are presented in Table 12. There were small positive associations between stronger M/P and PA connection and global evaluations of meaning and purpose cross-sectionally but not prospectively. Similarly, M/P and PA connection was positively associated with several PA variables cross-sectionally, including total PA, vigorous PA, and moderate PA. Baseline PA was prospectively and positively associated with a greater M/P connection. The more activity individuals were doing at the beginning of the study, the stronger they rated the connection between M/P and PA. Baseline ratings of M/P and PA connection were negatively and significantly associated with BMI and body fat percent at baseline and at the end of the study.
72


Table 12
Pearson Correlations between M/P and PA Connection, M/P, PA, and Objective Outcomes
Measure Baseline Week 4 Week 12
Meaning Salience Average .24 .21 .18
Meaning Salience Variability -.04 .09 -.06
Baseline
Meaning in Life .16 .11 -.02
Purpose in Life .24 .15 .08
Number of Fitness Center Visits -.14 .14 .08
Total PA .22 .22 .23
Vigorous PA .19 .14 .06
Moderate PA .17 .18 .15
Walking Activity .13 .09 .22
Body Mass Index -.14 -.11 -.02
Body Fat Percent -.17 -.10 .01
Physical Fitness .03 .00 -.03
Week 4
Meaning in Life .18 .26 .04
Purpose in Life .11 .15 .11
Number of Fitness Center Visits -.10 .08 .16
Total PA .00 .06 .09
Vigorous PA -.05 .07 .05
Moderate PA .08 .10 .12
Walking Activity -.01 .00 .08
Week 12
Meaning in Life .11 .10 .11
Purpose in Life .15 .09 .18
Number of Fitness Center Visits -.01 .10 .26
Total PA .00 .17 .28
Vigorous PA .12 .13 .33
Moderate PA -.04 .11 .17
Walking Activity -.10 .07 .11
Body Mass Index -.19 -.10 -.04
Body Fat Percent -.22 -.08 -.04
Physical Fitness -.02 -.02 .04
Note. Significant correlations are in bold.
73


CHAPTER IV
DISCUSSION
The purpose of the study was to examine the associations between M/P and PA, in the context of SDT, in previously sedentary adults beginning new exercise programs. The first aim was to test whether self-monitoring of M/P, mood, and PA daily for the first four weeks increased exercise more than a random-survey control. Results suggest that the self-monitoring condition had no effect on PA or secondary outcomes of M/P, SDT mediators, psychological well-being, body size/composition, or physical fitness. The second aim examined whether daily salience of M/P was related to daily PA in the first four weeks of starting an exercise program. Results revealed that controlling for daily mood, on days when participants reported greater salience of meaning, they also reported more minutes of PA, were more likely to visit the fitness center, and reported greater intensity of PA when they did exercise. The final aim of the study was to examine whether adding M/P to the SDT Process Model of Behavior Change predicted PA above and beyond the base SDT model. Results of this aim were mixed, with M/P at baseline predicting greater change in PA from baseline to 4-weeks. However, M/P was not related to PA cross-sectionally at any of the three time points or to PA at 12-weeks. Thus, it appears that M/P salience is important during the behavior change process, but further research needs to examine the role of global evaluations of M/P during PA adoption and maintenance.
Aim 1. Effect of M/P, Mood, and PA Self-Monitoring on PA.
A major question of this study was whether just asking people to report on their daily M/P and also their daily mood and PA would be an intervention in itself. Previous PA interventions have shown that self-monitoring is an effective tool for PA behavior change
74


(Olander et al., 2013). Consistent with the hypothesis, results suggest that just asking people to report on M/P salience, mood, and PA did not improve PA adoption or maintenance in the first 12 weeks of making a behavior change. The daily surveys, were in fact, framed as such, and were not labeled an intervention to participants. Perhaps giving participants the rationale for self-monitoring would increase the likelihood that the self-monitoring would have had an effect on PA or other outcomes (e.g., calling the daily surveys an intervention and noting that self-monitoring of behavior is one strategy to improve behavior change). Another possible reason the daily surveys may not have impacted PA is that participants reported at the end of the day. Self-monitoring may work best when it is an ongoing process that occurs throughout the day, because it can remind individuals to engage in the behavior that they are monitoring. For instance, participants could go about their day, while not actively thinking about engaging in PA or what is meaningful to them, and then complete the self-monitoring task in the evening. By the time they completed the self-monitoring task, it may have been too late to engage in PA for the day, and any potential impact of the monitoring on PA dissipated overnight.
On the other hand, it may take a stronger intervention to increase the pairing of M/P salience to PA in order to increase PA. Indeed, it was hypothesized that those who connect their personal sense of M/P to the reasons they want to make a PA behavior change, keep those reasons salient or in the forefront of their minds as they go about their days, and then make decisions based on that salience would be most successful. It is likely that just asking people to report on M/P salience and PA at the end of the day did not make a strong enough connection for them. Instead, asking people to actively connect what is meaningful to them or consistent with their sense of purpose in life to the reasons why they should make the
75


behavior change would likely be a more effective intervention. Once participants make the connection, then providing timely reminders to engage in PA and think about M/P could increase their activity. For example, individuals could engage in a task to increase their connection between their what gives their lives meaning and their goals to engage in healthy behaviors. Individuals could first identify what areas give their lives the most meaning (e.g., work, family, relationships, spirituality, etc.). After they identify these areas, they would be asked to make connections to how being healthier (e.g., by engaging in PA) supports their values and goals in the previously identified areas. For instance, one person might state that engaging in PA gives them more energy at work, which is a key area that gives their lives meaning. After identifying several points of connection, an interventionist can design tailored messages to be delivered at key choice points for PA (e.g., a person plans to walk at lunch time and a reminder would be delivered a few minutes prior to lunch) that remind the person of their previously made connection between meaning and PA. This message could bring the salience of meaning to the forefront of their minds, and thus, the choice to engage in PA may be more likely.
Not surprisingly, one of the biggest implications of these findings is that PA and M/P are not easily changed by measurement alone. Thus, the purpose of this aim was more of a methodological question than it was an intervention efficacy question. As researchers, we are highly concerned about measurement reactivity, and this is especially true when PA is the primary outcome. The results of this study suggest that repeated daily measurement of M/P, mood, and PA does not directly influence several outcomes, including PA, M/P, psychological well-being, and SDT mediators.
76


Aim 2. Relationship of Daily M/P Salience to Daily PA.
As hypothesized, on days when participants thought more about what is meaningful or valuable to them, they also engaged in more minutes of PA and were more likely to visit the fitness center. This relationship existed even after controlling for positive and negative mood, suggesting that meaning salience is an important predictor of PA. Previous research has highlighted the importance of mood for predicting PA (Cards et al., 2004), but this is the first study to show that after controlling for mood, meaning salience is a predictor of PA. Somewhat surprisingly, meaning salience was also associated with greater PA intensity on days on which participants reported engaging in PA. It may be that individuals who frequently think about what makes their lives meaningful and make decisions based on their personal sense of M/P also more consciously engaged in exercise versus PA. Exercise is a generally more intentional and intense behavior than PA, and thus individuals who make more intentional decisions to engage in exercise may be making those decisions because they are consistent with their personal M/P.
Meaning salience seems to be a more robust predictor of PA than global ratings of M/P (see below for further discussion). Several studies have shown that global ratings of M/P are positively associated with PA (Holahan et al., 2008; Holahan & Suzuki, 2006; Homan & Boyatzis, 2010; Hooker & Masters, 2014; Ruuskanen & Ruoppila, 1995; Takkinen et al., 2001). M/P seems important for PA but only to the extent to which M/P is salient for an individual on a daily basis. Global ratings of M/P are quite stable (e.g., Steger et al., 2006), and most of our empirical base of the relations between M/P and PA have been from static glimpses of individuals reported M/P and PA behavior. Spring and colleagues (Spring, Gotsis, Paiva, & Spruijt-Metz, 2013) note that the majority of health behavior change
77


theories involve causal constructs [that] are intrapsychic, conscious, somewhat vaguely specified processes whose quantitative relationship to health behavioral change is only imprecisely specified (p. 35). Global ratings of M/P fall in this category; it is difficult to understand how this global sense of M/P would be related to engaging in PA on a day-to-day basis, especially if individuals do not think about this other than when they are asked about it on a survey. In contrast, meaning salience is a dynamic process and could be used as motivation to engage in healthy behavior. Indeed, Frankl states this nicely, For the meaning of life differs from man to man, from day to day and from hour to hour. What matters, therefore, is not the meaning of life in general but rather the specific meaning of a persons life at a given moment (Frankl, 1985, p. 108). The conscious process of thinking about what is meaningful to an individual when faced with everyday decisions, such as the decision about whether or not to engage in PA that day, could be uniquely applied to just-in-time behavioral interventions to increase engagement in PA. Indeed, the advent of mobile technology allows researchers to design investigations that can tap into these dynamic processes to (1) increase knowledge of proximal predictors of behavior in context and (2) intervene at apropos times to improve health behavior adoption and maintenance (Spruijt-Metz, et al., 2015). A better understanding of these proximal predictors of behavior could change health behavior theory, and subsequently health behavior interventions, dramatically. Meaning salience is one such predictor.
Aim 3. Examining the SDT Process Model of PA Adoption with and without M/P.
For the third aim, the models largely supported the SDT Process Model of Behavior Change. Individuals that reported greater psychological needs satisfaction also reported greater autonomous regulation, PA, and psychological well-being. Those who reported more
78


autonomous regulation also engaged in more PA at baseline and at 12-weeks. In contrast to SDT theory, autonomous regulation was not associated with PA at 4-weeks. Longitudinal models also supported the SDT Process Model, suggesting that the SDT significantly explained variance in absolute PA at 12-weeks and the change in PA from baseline to 4-weeks and from 4-weeks to 12-weeks. Of note, the 4-week and 12-week longitudinal models did not adequately represent the data. One postulate of the SDT Process Model of Behavior Change is that basic psychological needs satisfaction is associated with greater autonomous regulation, and in turn, autonomous regulation is positively associated with PA. Moreover, the relationship between basic psychological needs satisfaction and PA is fully mediated by autonomous regulation. However, the full mediation was not observed at 4-weeks and 12-weeks after starting a PA program. This could be because changes in basic psychological needs satisfaction early in a new PA program do not quickly translate into changes in autonomous motivation. For example, one could imagine that a new, previously sedentary exercise initiate may experience rapid increases in competence. However, just because individuals are more competent in engaging in PA (e.g., knowing what exercises to do or how to do the exercises), does not mean they quickly become more integrated within their values or become more enjoyable. SDT principles do seem to account for a significant proportion of the variance in PA behavior during PA adoption; however, it does seem that adjustments to the SDT Process Model may be needed to more fully understand PA adoption in previously sedentary exercise initiates (e.g., accounting for meaning salience or other health behavior theory constructs).
These findings significantly advance the SDT literature in the area of PA adoption and maintenance. Of the longitudinal studies of these processes, only two have had more than
79


two time points (Edmunds et al., 2007; Gunnell et al., 2016), only two have examined the processes in previously sedentary individuals (Edmunds et al., 2007; Rodgers et al., 2010), and only one has observed these processes in individuals intending to become more active outside the context of an exercise intervention (including both non-SDT interventions and SDT-interventions; Fortier et al., 2009). This study adds to the current literature by providing an observational study of SDT hypothesized behavior change constructs in previously sedentary adults attempting to increase their PA.
Many of the previous studies have examined SDT processes in samples that included non-exercisers, regular exercisers, and everyone in between. Because the motivation continuum ranges from amotivation to intrinsic motivation, SDT may do a better job distinguishing regular exercisers from non-exercisers than predicting PA adoption in exercise initiates. For example, in this study, the controlled regulation composite score of the BREQ-2 demonstrated very low means and a restricted range of scores across all three time points. This suggests that participants did not strongly agree with the controlled regulation statements. Conversely, participants may not have endorsed intrinsic motivation, or engaging in exercise because of enjoyment, as high as regular exercisers. Previous research supports this hypothesis. Across stages of change, those in the pre-preparation stages endorse the highest levels of controlled regulation (external, introjected) and the lowest levels of autonomous regulation (identified, intrinsic) compared to those in more advanced stages; individuals in the maintenance stage endorse the highest levels of autonomous regulation and the lowest levels of controlled regulation (Mullan & Markland, 1997). A group of individuals in the midst of a behavior change may be somewhere in the middle. Exercise initiates do generally become more self-determined in their exercise regulations during the first 6 months
80


of behavior change; however, they may never fully reach the levels of autonomous motivation of regular exercisers, or those who have exercised at least three times per week for a minimum of 6 months (Rodgers et al., 2010). Moreover, their endorsement of self-determined regulation is more variable than that of regular exercisers, suggesting that their levels of motivation, as least in the first 6 months, are not stable (Rodgers et al., 2010). Exercise initiates likely see the value of exercise and it may be consistent with their goals, but still need the extrinsic forms of motivation to encourage the behavior. Indeed, previously sedentary exercise initiates may never experience sustained intrinsic motivation to engage in PA, and the goal of behavior change may need to be integrated regulation of the behavior, or engaging in the behavior because it is consistent with ones innermost values, identity, and self-concept (I am an exerciser). Others have started to call for this approach (e.g., Stevens & Bryan, 2015), and suggest that capitalizing on the more integrated forms of external regulation may be the best inroad to long-term maintenance of behavior change in this group.
One of the interesting findings in this study was that the cross-sectional SDT models did not accurately predict PA behavior at 4-weeks, but it was more predictive of behavior at baseline and 12-weeks. This suggests that engaging in PA at 4-weeks after starting a new exercise program may require more strategies than just the motivation regulation suggested by SDT, and broader models of health behavior change may be necessary. Specifically, time-dependent models of health behavior change may help predict behavior adoption and maintenance. The Transtheoretical Model (TTM; also known as Stages of Change; Prochaska & Velicer, 1997), one such time-dependent model of health behavior change, has been created in order to match strategies of intervention to the stage of change of the participant. However, the TTM lumps all behavior between preparation and maintenance into the
81


action stage. The action stage may be the most important stage for understanding how individuals go from starting a behavior to maintaining a behavior. Understanding the psychological factors that occur during the action stage, including the variability of self-determined regulations and meaning in life, may help us better explain behavior during this time. It is unclear what psychological processes are changing on a day-to-day basis in the midst of behavior change, likely because these factors are transient and difficult to capture with our current methods of measurement.
Interestingly, those who reported greater levels of meaning in life at baseline also reported more PA at 4-weeks, and meaning in life at baseline was one of the only predictors of change in PA from baseline to 4-weeks. This suggests that those who felt their lives were more meaningful at the beginning of a behavior change were engaging in more PA 4-weeks later. Further analyses revealed this was primarily through increased moderate activity at 4-weeks. This suggests that a strong sense of meaning in life may be an important factor that helps individuals in the early stages of behavior change to adopt a new healthy behavior, such as PA. As Ryff and Singer (1998) noted, believing that ones life is meaningful is a prerequisite to taking care of oneself with healthy behaviors. Individuals with a stronger sense of meaning in life also endorsed a greater connection between M/P and exercise at baseline and 4-weeks, and that connection was associated with more PA at baseline and 4-weeks. Thus, exercise initiates with a stronger sense of M/P tend to also believe that exercise supports their achievement of life goals.
In contrast to previous research (Holahan et al., 2008; Holahan & Suzuki, 2006; Homan & Boyatzis, 2010; Hooker & Masters, 2014; Ruuskanen & Ruoppila, 1995; Takkinen et al., 2001), M/P was largely not cross-sectionally related to PA, with a couple exceptions
82


(greater meaning in life at 4-weeks was related to more visits to the fitness center, meaning in life at 12-weeks was related to more moderate PA at 12-weeks). This may be for a number of reasons. First, this is the first study to examine this relationship in a group of previously sedentary exercise initiates. The variability in PA behavior was likely restricted at baseline (individuals were sedentary) and then became much more variable over time. In contrast, global evaluations of meaning in life were relatively stable over the 12-weeks of starting an exercise program. A second reason is that the relationship between a global sense of meaning in life and PA may not be perfectly linear. For example, Ryff and Singers (1998) supposition that engaging in health behaviors requires a necessary level of M/P in life may be true, but does having much more meaning above and beyond the necessary level translate to increased engagement in healthy behaviors? Meaning salience, or the extent to which individuals think about and are aware of that meaning, seems to have a stronger relationship with PA than global ratings of M/P, as was observed in this study.
Thus, meaning in life at baseline may predict PA at 4-weeks above and beyond the SDT Process Model of Behavior Change because of the sense that engaging in PA is important for supporting health, sense of M/P, and life goals. This hypothesis was supported in the exploratory analyses, which revealed that the extent to which participants endorsed a meaning and exercise connection was related to global meaning and purpose as well as PA at baseline. However, these associations seemed to weaken over the course of the study.
Perhaps individuals are most motivated to engage in PA behavior at the beginning of starting an exercise program; they view the behavior as consistent with their goals and sense of purpose, and they imagine that their future selves are happier after they achieve their goals.
83


Then the real work of having to engage in a regular PA routine starts, and motivation likely wanes, especially for those who do not maintain the activity.
Another interesting finding was that PA at earlier time points was longitudinally related to greater M/P at later time points. This raises the question which comes first? The hypotheses in this study revolve around the idea that meaning is motivation for engaging in healthy behaviors, but perhaps there is a bi-directional relationship. For some, engaging in PA could increase a sense of meaning, which may be related to the findings that PA improves mood (Annesi, 2004) and executive functioning (Etnier & Chang, 2009). Meaning is a combination of affective, motivational, and cognitive factors (Reker, 2000; Wong, 1989), and the sense of coherence that stems from meaning (e.g., life makes sense) could be related to better executive functioning. Other evidence suggests that models of change in both psychological factors and behavior are relatively equivalent in terms of which comes first -behavior or psychological constructs (Gunnell et al., 2016). Perhaps in some cases, the behavior comes first, and then individuals attempt to rationally explain the behavior (e.g., I exercise because it is consistent with my goals or because my life must be meaningful). Unfortunately, the type of analysis required for this hypothesis (latent growth curve modeling) requires larger samples than that was recruited for this project.
The main findings of this study suggest that modifications to health behavior change theories are needed. Indeed, the process of health behavior change seems to be a dynamic process, and drilling down to the daily decisions of individuals in the process of making a change may be necessary to fully understand the psychological constructs predicting changes. Moreover, meaning in life and meaning salience need to be considered in the modification of health behavior change theories. A sense of meaning is closely tied to an
84


individuals identity, which is likely related to long-term maintenance of behavior change. For a behavior to become a habit, it must be fully integrated with the self, and meaning in life and autonomous regulation may be key factors to make PA behavior more habitual. Thus, theoretical modifications, and resulting interventions, are needed to fully understand and improve engagement in health behavior.
Strengths
This study has several strengths. First, this was the first study of previously sedentary exercise initiates that examine M/P in the context of behavior change. Second, objective assessments of PA (fitness center visits), body size (BMI, body fat, and waist circumference), and physical fitness (cardiovascular, strength, and flexibility) were gathered to support the primary outcome of self-reported PA. Third, the study used a RCT design to examine the effects of a measurement intervention on primary and secondary outcomes. Finally, the studys emphasis on meaning salience, rather than global ratings of M/P, is an innovative contribution to the field.
Limitations
The primary limitation of this study is reliance on self-report measures of PA. Self-report PA measures are known to have poor associations with objective PA measures (Troiano, Pettee Gabriel, Welk, Owen, & Sternfeld, 2011), but are considered reliable for rank ordering PA behavior (Masse & de Niet, 2011). Thus, the self-reported PA measures should be reliable in distinguishing those who do more activity from those who do less activity, but the absolute levels of PA may not be representative of actual behavior. A second and important limitation is that the sample was predominantly female, highly educated, and able to afford a membership to a fitness center. Thus, these results may not generalize to less
85


educated and less affluent samples, although it may be representative of those who join private fitness centers. Also, as with all longitudinal studies, missing data are of concern. It is likely that participants who were lost to follow-up or dropped out were less likely to continue to engage in PA. However, the proportion of participants lost in this study (12.5%) was much less than what was expected (20%). Further, missing data analyses suggested that there were very few differences between those who completed the follow-up assessments at 4-weeks and 12-weeks and those who did not complete those assessments.
Of note, participants retrospectively reported their daily meaning salience and PA at the end of the day. Thus, this study cannot determine the within-day time order of meaning salience and PA. It could be that participants engaged in PA, which increased their meaning salience, rather than meaning salience increasing the likelihood of engaging in PA. A different ecological momentary assessment (EMA) design may be able to assess the question of time-order more effectively. A final limitation to note is that although this study is able to establish time-order relationships across the 12-weeks, it is not able to establish causality. A sense of meaning in life may not be fully subjected to experimental manipulation, i.e., we cannot assign individuals to have meaning or to not have meaning, but interventions to increase meaning salience may be an appropriate strategy to test these ideas.
Future Directions
There are several possible future directions for this research. One would be to recruit a larger sample from the general community, without the requirement that the participants be joining a fitness center, or to recruit from several different fitness centers, to increase the generalizability of the findings and replicate the results. Secondly, examining these processes over longer time periods (more than three months) and in larger samples would allow for
86


more complex linear growth curve modeling to examine the changes in the predictors (meaning in life, SDT mediators) and PA behavior. An EMA design with multiple within-day measurements and objective monitoring of PA would improve the assessment of timeordering of meaning salience and PA behavior. For example, the objective PA monitor could capture bouts of PA and then examine whether there was a stronger relationship between meaning salience prior to a bout of PA behavior or between PA and meaning salience after a bout of activity. Finally, interventions incorporating meaning salience to increase PA in sedentary adults with intentions to be more active would be a logical experimental step. It would be hypothesized that individuals who were primed to think more about meaning salience and to integrate their sense of M/P with their reasons to be more active would be more likely to maintain PA over time.
Conclusions
PA is an important health behavior for preventing many chronic illnesses, including CVD, as well as for increasing mental health, vitality, and longevity (Physical Activity Guidelines Advisory Committee, 2008). Most adults do not engage in regular PA, and determining factors related to long-term PA maintenance in previously sedentary exercise initiates is vital to our understanding of behavior change. This study provides evidence that basic psychological needs satisfaction and autonomous regulation from the SDT Process Model of Behavior Change and meaning salience are related to PA adoption. Future research should continue to explore these factors in exercise initiates, in combination with other health behavior change theories, in order to advance health behavior theory and design better interventions to improve PA adoption and maintenance.
87


Full Text

PAGE 1

INTEGRATING MEANING, PURPOSE, AND SELF DETERMINATION THEORY AS PREDICTORS OF PHYSICAL ACTIVITY MAINTENANCE b y STEPHANIE ANN HOOKER M.P.H., University of Colorado Denver 2015 M.S., Syracuse University, 2011 B.A.S., University of Minnesota Duluth, 2009 A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Doctor of Philosophy Clinical Health Psychology Program 2016

PAGE 2

ii 2016 STEPHANIE ANN HOOKER ALL RIGHTS RESERVED

PAGE 3

iii This thesis for the Doctor of Philosophy degree by Stephanie Ann Hooker has been approved for the Clinical Health Psychology Program by Krista W. Ranby, Chair Kevin S. Masters, Advisor James Grigsby James O. Hill Date: May 14, 2016

PAGE 4

iv Hooker, Stephanie Ann (Ph.D., Clinical Health Psychology) Integrating Meaning, Purpose, and Self Determination Theory as Predictors of Physical Activity Maintenance Dissertation directed by Professor Kevin S. Masters ABSTRACT Despite the widely known benefits of physical activity (PA) most adults are insufficiently active and have difficulty maintaining new exercise programs. Understanding facto rs related to PA in previously sedentary exercise initiates may improve interventions designed to increase consistent PA One factor might be the extent to which individuals experience meanin g in their lives on a daily basis. Individuals living lives with more experienced meaning, or meaning salience, may be more likely to engage in health behaviors inc luding PA This process is consistent with Self Determination Theory (SDT), a theory of beha vior regulation and motivation that states that more internally regulated behaviors are more likely to be maintained This study examined processes (i.e., meaning in the context of SDT) that may be associated with PA maintenance. P reviously sedentary exercise initiates ( N = 160; M age= 43.3 years SD = 11.4 years; 76.9% female) participated in a randomized controlled trial of a 4 week daily self monitoring of meaning, mood, and PA condition compared to a random survey control as they began a self initiated exercise program. Participants completed surveys at baseline, 4 weeks, and 12 weeks and f itness assessments at baseline and 12 weeks. Multilevel mixed models and path analytic methods were used to analyze data. Results revealed no significant differences b etween the self monitoring and control groups on PA or M/P. Within day analyses revealed that greater daily meaning salience was significantly related to greater daily PA duration, =.21, p <.0001, and intensity, =.21, p <.0001 and greater likelihood of visiting the fitness center, Odds Ratio=1.48 (95%

PAGE 5

v CI=1.18,1.86), p =.0008 Longitudinal path analytic models suggest that SDT variables accounted for a significant proportion of the variance in PA Modified models revealed that b aseline m eaning significantly predicted greater change in PA from baseline to 4 weeks, =.2 0 p =.02. However, meaning was not related to change in PA from 4 weeks to 12 weeks. Results suggest that meaning salience plays an important role in PA participatio n in previously sedentary exercise initiates. Findings related to global ratings of meaning are mixed suggesting that global ratings of meaning may not be as important as meaning salience. Future research should examine meaning salience over longer observ ation periods and meaning based interventions in exercise initiates The form and content of this abstract are approved. I recommend its publication. Approved: Kevin S. Masters

PAGE 6

vi ACKNOWLEDGEMENTS This research was funded by the American Heart Association Pre doctoral Fellowship grant (14PRE18710033) and by the Psychology Department at the University of Colorado Denver. This project was supported by the NIH/NCRR Colorado CTSI Grant Number UL1 RR025780. Its contents are the author's sole responsibility and do not necessarily represent official NIH views. I would like to thank Kaile Ross, Kaylae Nakamura, Jean Wood, and Emma Lyons for their assistance with data collection and study management. Additionally, I would like to thank Dr. James Hill and the staff at the Anschutz Health and Wellness Center for all owing me to recruit their members and providing the resources to conduct this study. Without their support, this study would not have been possible. I would like to thank Dr. Krista Ranby for her assistance with reviewing statistical models and suggestions for data analysis. I would also like to thank Dr. Kevin Masters for his wonderful mentorship and support over my graduate school career and for reading and providing feedback on countless drafts of manuscripts, proposals, grants, and essays. Finally, I wo uld like to thank my husband, Tom Showalter, Jr., for his endless support throughout my undergraduate and graduate careers.

PAGE 7

vii TABLE OF CONTENTS CHAPTER I. INTRODUCTION ................................ ................................ ................................ ......... 1 Meaning and Purpose ................................ ................................ ................................ ..... 2 Self Determination Theory ................................ ................................ ............................ 6 Meaning and Mood ................................ ................................ ................................ ...... 12 Self Monitoring of M/P, Mood, and PA as a Possible Intervention ............................ 13 Purpose of the Study ................................ ................................ ................................ .... 14 II. METHOD ................................ ................................ ................................ .................... 15 Overview ................................ ................................ ................................ ...................... 15 Participants and Recruitment ................................ ................................ ....................... 15 Randomization ................................ ................................ ................................ ............. 1 7 Procedure ................................ ................................ ................................ ..................... 1 8 Measures ................................ ................................ ................................ ...................... 2 0 Daily m ood ................................ ................................ ................................ ............. 2 0 Daily M/P salience ................................ ................................ ................................ 2 2 PA ................................ ................................ ................................ .......................... 2 2 24 hour activity recall ................................ ................................ ............................ 23 Psychological n eeds s atisfaction ................................ ................................ ............ 2 3 Behavioral r egulations m otivation ................................ ................................ ......... 2 4 M/P ................................ ................................ ................................ ......................... 25 M eaning in life ................................ ................................ ................................ 2 5 Purpose in life ................................ ................................ ................................ .. 26

PAGE 8

viii PA ................................ ................................ ................................ .......................... 2 6 Psychological w ell b eing ................................ ................................ ....................... 2 7 Subjective vitality ................................ ................................ ............................ 2 7 Life satisfaction ................................ ................................ ................................ 2 7 Depressive s ymptoms ................................ ................................ ............................ 2 8 Physical f itness ................................ ................................ ................................ ....... 2 9 M/P and PA c onnection ................................ ................................ ......................... 29 Power Analysis ................................ ................................ ................................ ............ 30 Aim 1. Effect M/P, m ood, and PA s elf m onitoring on PA ................................ ... 30 Aim 2. Relationship of d aily M/P s alience to d aily PA ................................ ........ 30 Aim 3. Examining the SDT P rocess M odel of PA a doption with and without M/P ................................ ................................ ................................ ................................ 30 Data Analysis ................................ ................................ ................................ ............... 3 1 Aim 1. Effect of M/P, m ood, and PA s elf m onitoring on PA ............................... 3 2 Aim 2. Relationship of d aily M/P s alience to d aily PA ................................ ......... 3 3 Aim 3. Examining the SDT p rocess m odel of PA a doption with and without M/P ................................ ................................ ................................ ................................ 3 3 Exploratory analyses ................................ ................................ .............................. 3 4 III. RESULTS ................................ ................................ ................................ .................... 3 5 Study Flow & Missing Data Analysis ................................ ................................ ......... 40 Aims 1 & 3: Baseline, 4 week, and 12 week m issing d ata ................................ ... 4 0 Aim 2 m issing d ata ................................ ................................ ................................ 4 1 Study Variables at Baseline, 4 weeks, and 12 weeks. ................................ ................. 4 1 Aim 1. Effect of M/P, Mood, and PA Self Monitoring on PA ................................ ... 4 5 PA ................................ ................................ ................................ .......................... 4 5

PAGE 9

ix M/P ................................ ................................ ................................ ......................... 4 8 SDT m ediators ................................ ................................ ................................ ....... 49 Psychological w ell b eing ................................ ................................ ....................... 5 1 Aim 2. Relationship of Daily M/P Salience to Daily PA ................................ ............ 5 4 Aim 3. Examining the SDT P rocess M odel of PA A doption with and without M/P .. 5 6 Model 1a: Cross sectional b aseline m odel ................................ ............................ 56 Model 1b: Modified c ross sectional b aseline m odel ................................ ............. 5 8 Model 2a: Cross sectional 4 week m odel ................................ .............................. 5 9 Model 2b: Modified c ross sectional 4 week m odel ................................ ............... 60 Model 3a: Cross sectional 12 week m odel ................................ ............................ 61 Model 3b: Modified cr oss sectional 12 week m odel ................................ ............. 6 2 Model 4a: L ongitudinal m o del p redicting a bsolute l evels of PA at 12 weeks ...... 6 3 Model 4b. Modified l ongitudinal m odel p redicting a bsolute l evels of PA at 12 weeks ................................ ................................ ................................ ...................... 6 4 Model 5a: L ongitudinal m odel p redicting c hange in PA at 4 weeks and 12 weeks ................................ ................................ ................................ ................................ 6 4 Model 5b: Modified l ongitudinal m odel p redicting c hange in PA at 4 weeks a nd 12 weeks ................................ ................................ ................................ ................ 6 6 Exploratory Outcomes ................................ ................................ ................................ 6 7 Average and v ariability of m eaning s alience in r elation to PA and f itness o utcomes ................................ ................................ ................................ ................ 6 7 Relationships between g lobal M/P and PA and o bjective o utcomes ..................... 70 Relationships between M/P and PA c onnection, M/P, PA and o bjective o utcomes ................................ ................................ ................................ ................................ 7 2 IV. DISCUSSION ................................ ................................ ................................ .............. 7 4 Aim 1. Effect of M/P, Mood, and PA Self Monitoring on PA ................................ ... 7 4 Aim 2. Relationship of Daily M/P Salience to Daily PA ................................ ........... 7 7

PAGE 10

x Aim 3. Examining the SDT P rocess M odel of PA A doption w ith and without M/P .. 7 8 Strengths ................................ ................................ ................................ ...................... 8 5 Limitations ................................ ................................ ................................ ................... 8 5 Future Directions ................................ ................................ ................................ ......... 8 6 Conclusions ................................ ................................ ................................ .................. 8 7 REFERENCES ................................ ................................ ................................ ........................ 8 8 APPENDIX A. Daily Quotes for the Self Monitoring Condition ................................ ......................... 9 8 B. Screening and Eligibility Form ................................ ................................ .................. 100 C. Baseline Only Measures ................................ ................................ ............................ 1 0 3 D. Baseline, 4 week, and 12 week Measures ................................ ................................ 1 0 6 E. Baseline and 12 week Measures ................................ ................................ ................ 1 1 4 F. Daily Measures ................................ ................................ ................................ .......... 1 1 5

PAGE 11

xi LIST OF TABLES TABLE 1. Measures Collected in This Study ................................ ................................ ..................... 2 1 2. Baseline Demographics for the Entire Sample and Between Groups ................................ 3 7 3. Descriptive S tatistics of S tudy V ariables at B aseline, 4 weeks, and 12 weeks ................. 4 3 4. Mean s and S tandard Deviations of S tudy V ariables by G roup ................................ ......... 4 7 5. Descriptive S tatistics of V ariables in Aim 2 ................................ ................................ ...... 5 5 6. Pearson Corr elations among V ariables in Aim 2 ................................ ............................... 5 5 7. Pearson Correlati ons among Aim 3 Model Variables ................................ ....................... 5 7 8. Model Fit S tatistics ................................ ................................ ................................ ............ 5 8 9. Pearson Correlations between the TOMS and O ther Study O utcomes ............................. 6 9 10. Regression M odels with M eaning Salience P redicting 12 week O utcomes ..................... 70 11. Pearson Correlations between G lobal M/P PA and O bjective O utcomes ....................... 71 12. Pearson Correlations between M/P and PA Connection, M/P PA and Objective O utcomes ................................ ................................ ................................ ........................... 7 3

PAGE 12

xii LIST OF FIGURES FIGURE 1. The SDT I nternalization C ontinuum ................................ ................................ ................... 7 2. Model I ntegrating SDT and M/P to I ncrease A ctivity ................................ ....................... 11 3. Study T imeline ................................ ................................ ................................ ................... 19 4. CONSORT D iagram ................................ ................................ ................................ .......... 42 5. Trajectories of PA O ver T ime ................................ ................................ ............................ 4 5 6. Mean PA by C ondition from B aseline to 12 weeks ................................ .......................... 4 6 7. Mean F itness C enter V isits by C ondition from B aseline to 12 weeks .............................. 4 8 8. Mean P urpose in L ife by C ondition from B aseline to 12 weeks ................................ ....... 49 9. Mean B asic P sychological N eeds S atisfaction by C ondition from B aseline to 12 weeks 50 10. Mean A utonomous R egulation by C ondition from B aseline to 12 weeks ........................ 5 1 11. Mean V itality by C ondition from B aseline to 12 weeks ................................ ................... 5 2 12. Mean L ife S atisfaction by C ondition from B aseline to 12 weeks ................................ ..... 5 3 13. Mean D epressive S ymptoms by C ondition from B aseline to 12 weeks ............................ 5 4 14. Cross sectional SDT Process Model of Behavior Change at Baseline .............................. 5 8 15. Cross sectional Modified SDT Process Model of Behavior Change at Baseline .............. 5 9 16. Cross sectional SDT Process Model of Behavior Change at 4 weeks .............................. 60 17. Cross sectional Modified SDT Process Model of Behavior Change at 4 weeks .............. 61 18. Cross sectional SDT Process Model of Behavior Change at 12 weeks ............................ 6 2 19. Cross sectional Modified SDT Process Model of Behavior Change at 12 weeks ............ 6 3 20. Longitudinal SDT Process Model of Behavior Change Predicting Absolute PA at 12 weeks. ................................ ................................ ................................ ................................ 6 4 21. Modified Longitudinal SDT Process Model of Behavior Change Predicting Absolute PA at 12 weeks. ................................ ................................ ................................ ....................... 6 5

PAGE 13

xiii 22. Longitudinal SDT Process Model of Behavior Change Predicting Change in PA at 4 weeks and 12 weeks ................................ ................................ ................................ .......... 66 23. Modified L ongitudinal SDT Process Model of Behavior Change Predicting Change in PA at 4 weeks and 12 weeks ................................ ................................ ................................ ... 68

PAGE 14

xiv LIST OF ABBREVIATIONS ACT Acceptance and Commitment Therapy AHWC Anschutz Health and Wellness Center BMI Body Mass Index BPNES Basic Psychological Needs in Exercise Scale BREQ 2 Behavioral Regulations in Exercise Questionnaire 2 nd version CVD Cardiovascular disease COMIRB Colorado Multiple Institutional Review Board DMS Daily Meaning Scale EMA Ecological Momentary Assessment IPAQ SF International Physical Activity Questionnaire Short Form LET Life Engageme nt Test MET Metabolic Equivalent of Task M/P Meaning and purpose MILQ Meaning in Life Questionnaire PA Physical activity PAR Q Physical Activity Readiness Questionnaire PANAS Posit ive and Negative Affect Schedule REDCap Research Electronic Data Capture SDT Self Determination Theory SVS Subjective Vitality Scale SWLS Satisfaction with Life Scale T O MS Thoughts of Meaning Scale TTM Transtheoretical Model UCD University of Colorado Denver

PAGE 15

1 CHAPTER I INTRODUCTION Approximately 2400 US a dults die from cardiovascular disease (CVD) every day (Lloyd Jones et al., 2009) Given the significant burden of CVD, primary prevention of CVD is essential to improve health and well being and to reduce costs associated with chronic illness (Probst Hensch, Tanner, Kessler, Burri, & KŸnzli, 2011) Regular p hysical activity (PA) is known to reduce the risk of cardiovascular disease and stroke (Physical Activity Guidelines Advisory Committee, 2008) Sustained long term PA reduces risk of incident CVD in men and CVD related and all cause mortality in both men and women (Shortreed, Peeters, & Forbes, 2013) Individuals who meet the national guidelines for PA ( !150 minutes of moderate to vigorous PA per week ; Physical Activity Guidelines Committee, 2008 ) al so have significantly better CVD risk factor profiles than those who d o not meet national guidelines (Glazer et al., 2013). Because the evidence that PA is beneficial for cardiovascular health is well known, the American Heart Association has listed meetin g national guidelines for PA as one of the metrics for good cardiovascular health (Lloyd Jones et al., 2010). Despite the benefits of engaging in PA for cardiovascular health, a vast majority of US adults do not engage in regular PA In fact, estimates fr om national self report surveys indicate that only 45% to 65% of the population meets guidelines for PA (Adabony a n, Loustalot, Kruger, Carlson, & Fulton, 2007; Macera et al., 200 5 ) but when measured via objective accelerometer assessment a mere 5% meet the guidelines (Troiano et al., 2008) Interventions to increase PA generally demonstrate short term success but fail to show long term maintenance (Marcus, Owen, Forsyth, Cavill, & Fridinger, 1998; Marcus et al., 2000)

PAGE 16

2 Identifying and utilizing psychologica l and behavioral processes that enhance long term maintenance of PA is desperately needed. Meaning and Purpose One psychological process likely to be related to long term maintenance of PA is a sense of personal meaning (i.e., feeling that one's existence is significant) and purpose (i.e., pursuit and attainment of behavioral goals consistent with one's subject ive values and life goals) ( M/P; Reker, Peacock, & Wong, 1987; Ryff & Singer, 1998; Steger, Frazier, Oishi, & Kaler, 2006 ) The sense of meaning in life is considered a multidimensional construct with (1) affective (feelings of fulfillment or satisfaction that come from the conviction that life is worth living) ; (2) cognitive (making sense of and find ing value and purpose in life events, circumstances or encounters); and (3) motivational (the pursuit and attainment of personal goals that are consistent with one's subjective values, needs and wants) components (Reker, 2000; Wong, 1989). Although meaning and purpose are often used interchangeably in the literature, purpose is considered the motivational component of meaning that stimulates goals and influences behavior (McKnight & Kashdan, 2009). Several studies have demonstrated that M/P is positively related to many different dimensions of health and w ell being, including increased longevity (Boyle, Barnes, Buchman, & Bennett, 2009; Cohen, Bavishi, & Rozanski, 2016; Krause, 2009), reduced risk for cardiovascular events (Cohen et al., 2016), improved physical health (Pinquart, 2002), better self rated he alth and functioning (Holahan, Holahan, & Suzuki, 2008; Krause, 2009; Krause & Shaw, 2003; O'Connor & Vallerand, 1998; Takkinen, Suutama, & Ruoppila, 2001), greater positive affect (Hicks & King, 2008, 2009; King, Hicks, Krull, & Del Gaiso, 2006 ), and bett er psychological and

PAGE 17

3 physical well being (Reker et al., 1987; Scheier et al., 2006; Steger et al., 2006). Thus, through some as yet unidentified mechanisms, M/P is associated with health and well being. A potential mechanism linking M/P to better health an d well being is greater engagement in healthy behaviors, and specifically, PA Ryff and Singer (1998) suggest that individuals who have a sense of purpose in their lives may be more likely to practice health promoting behaviors, and that "taking good care of oneself in terms of daily health practices presupposes a li fe that is worth taking care of (p. 22). This postulates that individuals who have a lesser awareness of M/P in their lives may not practice health promoting behaviors, at least in part, becaus e they do not see the value of effortfully supporting a life lacking M/P. On the other hand, if individuals live with awareness of personal M/P, this may motivate them to engage in healthier behaviors. Sch eier and collea g u es (2006) argue that purpose in life is a part of behavioral self regulation, as it provides the "why" to engaging in certain behaviors. When individuals live with daily awareness of what makes their lives meaningful and gives them purpose, they may be abl e to make routine choices that support their M/P. Adults are frequently faced with choices that can either be healthy (e.g., get up early to exercise before going to work) or hedonically pleasurable (e.g. skipping the workout because "I do not feel like i t"). Making the easier choice ( i.e ., skipping the workout) is more likely when individuals do not live with awareness of their personal M/P. Choosing the more difficult behavior that is congruent with one's personal values, goals, and purpose is more likel y when one is aware in the moment, of those values goals, and purpose Those who connect engaging in healthy behaviors, such as PA with their sense of M/P may be more likely to engage in those behaviors on a daily basis. Thus, M/P can be motivational an d contribute to self regulation of healthy behavior.

PAGE 18

4 Moreover, when individuals have a strong sense of M/P they likely also have several long term life goals. Without good health, individuals are likely to have difficulty reaching those goals. Thus, engag e ment in healthy behaviors, and especially in PA can be seen as a secondary goal and value in support of reaching one's ultimate life purposes. Cross sectional research on M/P supports this claim and observational findings show that greater M/P is related to greater engagement in PA (Holahan et al. 2008; Holahan & Suzuki, 2006; Homan & Boyatzis, 2010; Hooker & Masters 2014 ; Ruuskanen & Ruoppila, 1995; Takkinen et al. 2001) Several of these studies show that a global sense of personal M/P (e.g., "I have a good sense of what makes my life meaningful"; Steger et al., 2006) is related to self reported PA However, one recent study examined purpose in relation to objectively measured (with accelerometers) PA (Hooker & Masters, 2014 ). This study demonstrated that after controlling for several demographic (age, gender, socioeconomic status, race/ethnicity, and marital status), affective (depressive sym ptoms and positive affect), and cognitive (optimism and self mastery) confounds, purpose remained a robust predictor of objectively measured lifestyle movement, and to a lesser extent, moderate to vigorous exercise. These results support the hypothesis tha t awareness of personal M/P is related to greater engagement in PA Furthermore, i t is hypothesized that behaviors and goals (i.e., PA and healthy lifestyles) that are explicitly integrated with one's life meaning and purpose are more likely to be maintain ed. Indeed, some interventions have been designed to explicitly connect what is valuable or meaningful to a person with their behavior. Notably, acceptance and commitment therapy ( ACT; Hayes, Strosahl, & Wilson, 1999) proposes that commitment to value dir ected activities is important for behavior change. In one pilot study an ACT intervention designed

PAGE 19

5 to in crease value directed behaviors (as well as other acceptance based strategies such as willingness to experience distress) and PA was more effective tha n an education based control at increasing visits to a fitness center over 8 weeks (Butryn, Forman, Hoffman, Shaw, & Juarascio, 2011). However, it is not clear how much the results were driven by value d irected behaviors versus other acceptance based appro aches. Another approach, taken from the Disconnected Values Model (Anshel, 2010), suggests that individuals are more likely to engage in healthy behaviors when they realize there is a discrepancy (or disconnect) between what they value (e.g., health family ) and how they behave (e.g ., lack of exercise). Anshel and colleagues ( Anshel, Brinthaupt, & Kang, 2010; Anshel, Kang, & Brinthaupt, 2011) have tested their 10 week intervention (including an orientation based on the Disconnected Values Model, an ex ercise prescription, and weekly personal training) in university employees and have shown that the intervention increases physical fitness and well being. However, both tests of the intervention did not have a control group, and the extent to which the int ervention results are due to the theoretical model rather than the structured personal training components of the program is not clear Whereas these interventions provide evidence that connecting what is meaningful or valuable to a person to their health behaviors may improve engagement in those behaviors none of these interventions have explicitly examined the proposed theoretical mediators (e.g., value d irected behavior, values connection) to their outcomes. Moreover, none of these interventions have ex amined personal M/P in relation to PA adoption or maintenance. The primary limitation of the current literature is that all current studies of M/P and PA are cross sectional in nature; t o my knowledge, no studies have specifically incorporated life M/P int o a longitudinal observational study of PA adoption. Therefore, this study will

PAGE 20

6 address a large gap in the research by studying the relationship between M/P and PA over time in a group of previously sedentary new exercisers Additionally, no studies have t ested the hypothesis that greater salience of M/P on a daily basis is related to PA on that particular day. This investigation will address that hypothesis in a daily diary study of PA and M/P. Moreover, no studies have incorporated a theoretical model tha t attempts to explain how M/P may influence PA over time. Thus, this study will attempt to integrate M/P theory with an existing model of health behavior change i.e., Self D etermination Theory (SDT) Self D etermination Theory It is often cited that M/P provides motivation to confront life's problems (Sagy, Antonovsky, & Adler, 1990) and to develop well formed organized goals (McKnight & Kashdan, 2009). This is c onsistent with the assertion that behaviors integrated with M/P are mo re likely to be maintain ed SDT ( Ryan & Deci, 2000) is a relatively young and particularly promising theory of motivation that posits that behaviors that are internalized or intrinsically motivated are more likely to be maintained than behaviors that are externalized or extrinsic ally motivated. Deci and Ryan (2000) define intrinsic motivation as the "active engagement with tasks that people find interesting and that, in turn, promote growth" (p. 233). SDT also explains why people may engage in behaviors that are not interesting an d enjoyable The SDT sub theory called organismic integration theory postulates that motivation exist s on a continuum of behavioral regulations : (a) amotivation (lacking intention to act); (b) external regulation (driven by external rewards and punishments ), (c) introjected regulation (motivated by guilt or need to comply); (d) identified regulation (driven by personal importance or conscious valuing); (e) integrated regulation (motivated by

PAGE 21

7 congruence or synthesis with the self); and (f) intrinsic motivati on (driven by interest or enjoyment) (see Figure 1). Figure 1. The SDT Internalization Continuum. In this sense, SDT is a theory of behavioral self regulation. At times, individuals engage in behavior because of external factors, such as rewards or punishments, and at times, they eng age in behavior because of internal factors, such as enjoyment. B ecause motivation exists on a continuum, individuals can also be motivated to engage in behavior for many reasons (e.g., "I exercise because it is a part of who I am [integrated regulation] b ut I would also feel guilty if I did not exercise [introjected regulation] ."). A large literature based on SDT exists and several studies have demonstrated that self determined or intrinsic motivations are positively related to psychological and physical well being (cf., Deci & Ryan, 2000; Ryan & Deci, 2000). The idea that meaning may foster intrinsic motivation is an implicit assumption of SDT that has not been explicitly tested though a few recent author s (Weinstein, Ryan, & Deci, 2012 ) have begun to in tegrate the M/P and SDT literature s suggesting that this is a timely area of research. SDT also hypothesizes that people have a natural inclination toward health and wellness (Ryan & Deci, 2000), but life is full of distractions that deter individuals fro m Amotivation External Regulation Introjected Regulation Identified Regulation Integrated Regulation Intrinsic Motivation Extrinsic Motivation Intrinsic Regulation More integrated and autonomously motivated

PAGE 22

8 making healthy choices. In a system atic review of 66 studies examining the relations between SDT constructs and exercise and PA 91% of the studies found a positive association between autonomous regulations and exercise behavior (Teixeira, Carraa, Mark land, Silva, & Ryan, 2012). The vast majority of the evidence comes from cross sectional studies demonstrat ing that people who report more self determined or internally regulated motivations also report greater participation in PA and exercise and more pos itive psychological outcomes of exercise participation (Chatzirantis & Hagger, 2007; Hall, Rodgers, Wilson, & Norman, 2010; Ingledew & Markland, 2004; Ingledew & Ma rkland, 2008; Landry & Soloman, 2004 ; Markland & Ingled ew, 2007; Mullan & Markland, 1997; Se bire, Standage, & Vansteenkiste, 2009; Standage, Sebire, & Loney, 2008; Wilson & Rodgers, 2002; Wilson, Rodgers, Fraser, & Murray, 2004). There are several prospective longitudinal studies (range of 4 24 weeks) that demonstrate that basic psychological needs and autonomous motivation at time 1 are predictive of PA behavior at time 2 ( Barbeau, Sweet, & Fortier, 2009 ; Gunnell, Crocker, Mack, W ilson, & Zumbo, 2014; Hagger, Chatzirantis, & Harris, 2006a,b; Kwan, Caldwell Hooper, Magnan, & Bryan, 2009). However, none of these prospective studies examine d SDT constructs in a behavior change context; participants were either already active or were n ot explicitly recruited because they intended to change their activity levels. One study of middle aged women recruited from the community specifically examined autonomous regulation in relation to behavioral intentions to engage in PA and change in PA fro m time 1 to time 2 (6 months later; Fortier, Kowal, Lemyre, & Orpana, 2009). Participants were eligible if they stated they intended to increase their PA over the next 6 months. Autonomous regulation at baseline was positively related to intentions to incr ease PA which were

PAGE 23

9 positively related to PA behavior change. However, contrary to the hypotheses, autonomous motivation was not significantly related to behavior change. Fortier and colleagues (2009) hypothesized this was because there was limited variabi lity in autonomous motivation, and the sample was highly internally regulated already. Another possible hypothesis is that the majority of participants were meeting PA guidelines at baseline which may limit the ability to predict increases in PA behavior Fortier et al. (2009) suggest that work examining previously sedentary individuals desiring to increase their PA would add to the SDT literature. R esearch suggests that new (previously sedentary) exercisers, in non SDT based exercise interventions, tend t o naturally decrease in external regulations and increase in more internalized motivations over time (Rodgers, Hall, Duncan, Pearson, & Milne, 2010) and more self determined motivations are associated with greater exercise persistence ( Gorin et al., 2008; Hagger & Chatzirantis, 2008; Teixeira Silva, Mata, Palmeira, & Mark l a nd, 2012) In exercise initiates starting new exercise programs, identified and integrated forms of regulation tend to increase more quickly over time than intrinsic motivation (Rodgers et al., 2010; Teixeira, Carraa et al., 2012). This suggests that new exercisers may be more likely to continue engaging in exercise because exercise becomes increasingly consistent with their identities or values, rather than because they experience increased enjoyment of exercise Based on cross sectional, prospective, and intervention research, t here is ample evidence to sup port the hypothesis that when exercise is more internalized, the behavior is more likely to be maintained. However, little is known about the natural progression of the SDT motivation internalization process in previously sedentary exercise initiates not p articipating in formalized exercise interventions.

PAGE 24

10 How does a behavior become self determined or internally motivated? The SDT Process Model of Behavior Change (Fortier, Duda, Guerin, & Teixeira, 2012) and a SDT sub theory called the Basic Needs Theory (Ry an & Deci, 2000) suggest that social environments that support the three basic psychological needs of autonomy (feeling behavior is self organized, accompanied by a sense of volition), relatedness (feeling connected to others), and competence (feeling capa ble of achieving goals) foster the internalization of motivation and facilitate behavior change Evidence suggests that increases in psychological needs satisfaction for PA over three years are positively related to PA behavior changes in adolescents (Gunn ell, BÂŽlanger, & Brunet, 2016). More over, in exercise initiates in non SDT based interventions, basic psychological needs satisfaction (autonomy and competence) is predictive of increases in autonomous regulation (e.g., Edmunds, Ntoumanis, & Duda, 2007). S upport for this model has also been observed in studies of PA interventions incorporating SDT in that they have shown success in increasing internalized regulation and increasing PA by targeting these three basic needs (Fortier et al., 2007; Fortier et al., 2011; Jolly et al., 2009; Rouse, Duda, Ntoumanis, Jolly, & Williams, 2010; Silva et al., 2008 ; Silva et al., 2010 ; Silva et al., 2009 ) Despite the promise of the SDT approach, none of these studies have explicitly examined M/P in the conte xt of SDT as an added value to the model. Given previous hypotheses that M/P helps individuals maintain behaviors that are integrated with their most important values and life goals (Scheier et al., 2006), it is likely that those who live with awareness of their personal M/P also have more internalized or integrated forms of behavioral regulation. Figure 2 illustrates the SDT Process Model of Behavior Change (Fortier et al., 2012) with the important addition of M/P. Fortier and colleagues (2012) model inclu des psychological needs satisfaction, behavioral regulations

PAGE 25

11 motivation, PA behavior, and psychological well being. However, I have modified the model to include M/P. Figure 2. Model Integrating SDT and M/P to I ncrease PA and W ell being. The SDT Process Model of Behavior Change suggests that behavior change occurs in social environments where individuals perceive their three basic psychological needs are being met Individuals must feel (1) more competent in their abilities to perform certain behaviors (competence) ; (2) supported by t heir social group (relatedness); and (3) that they have the opportunity to choose how and when the behaviors will be performed (autonomy). When all three needs are satisfied, individuals tend to engage in more autonomously regulated or internalized (e.g., identified, integr ated, and intrinsic) behaviors. More internalized behaviors are more likely to be maintained; thus, PA that is more autonomously regulated will be more likely to be maintained. Autonomously regulated behavior leads to improved psychological well being. I suggest that M/P will add to this mode l in three specific ways : Psychological Well Being Physical Activity Acquisition and Maintenance Psycholog ical Needs Satisfaction (Autonomy, Relatedness, Competence) Behavioral Regulations Motivation Meaning and Purpose

PAGE 26

12 1. W hen the three basic psychological needs are met, the environment supports individuals in developing their own personal M/P Thus, there will be a positive relationship between psychological n eeds satisfaction and M/P. 2. When individuals have greater awareness of their person al M/P they will likely act in ways that are more consistent with their M/P, i. e., more internally regulated; therefore, individuals who live with greater awareness of thei r M/P will also be more internally regulated to engage in PA An implicit assumption is that the participants in this study will value PA and PA is consistent with their overall M/P. Because these individuals will be voluntarily starting an exercise progr am, it is likely that this assumption is true. 3. Finally, previous evidence suggests that M/P is directly and positively related to psychological well being (e.g., Scheier et al., 2006; Steger et al., 2006). Meaning and Mood A possible rival hypothesis to the M/P as motivation theory is that M/P is a consequence of improved mood as a result of activity. Evidence suggests that increased PA is related to reduced depressive symptoms (Goodwin, 2003; Hooker, MacGregor, Funderburk, & Maisto, 2013; Taliaferro, Rienzo, Pigg, Miller, & Dodd, 2008) and increased positive affect (Carels Coit, Young, & Berger, 2007) The "Feelings as Information" theory proposed by Schwarz (2001) suggests that individuals use positive affect as informatio n to make judgments about other areas of their lives. For instance, sedentary individuals who start exercising more will experience improved mood (Annesi, 2004) Because they are experiencing more positive affect, they will use their mood to make judgments about how

PAGE 27

13 meaningful their lives are. If they are feeling more positive, then they will rate their meaning in life as being greater as well. Some evidence suggests that positive mood is related to global evaluation of M/P (Hicks & King, 2008; 2009; King et al., 2006). However, when experimentally primed with images of religious commitment (Hicks & King, 2008) or social relatedness (Hicks & King, 2009), the relationship between M/P judgments and positive affect decline s This suggests that positive moods may predispose individuals to feel that their lives are meaningful; however, individuals use additional factors to judge how meaningful their lives actually are. Moreover, in a study of M/P and objectively assessed PA (Hooker & Masters, 2014 ), positive aff ect and depressive symptoms were not significantly related to PA over the course of three consecutive days and controlling for both positive affect and depressive symptoms did not significantly reduce the relationship between M/P and PA Although positive affect seems to have important relations with M/P and PA it is not clear that it is a confounder of the relationship between M/P and PA However, in the present study, mood and depressive symptoms are measured to account for this potential confound Self monitoring of M/P, Mood, and PA as a Possible Intervention One goal of this study is to examine M/P salience in relation to PA controlling for daily mood. It is possible that having participants record their daily M/P and PA could, in fact, be an in tervention in itself Indeed, a recent meta analysis found that self monitoring of PA is an effective technique to increase PA self efficacy and PA behavior (Olander et al., 2013). However, given the limited literature on the association between M/P and PA it is not clear whether monitoring M/P, mood, and PA all together is in fact, an intervention. Thus, in

PAGE 28

14 the current study participants we re randomized to the daily self monitoring of M/P, mood, and PA condition or to a random survey assessment control. Purpose of the Study Interventions that enable individuals to maintain PA have proven difficult for behavioral scientists, and novel approaches to behavior change are needed. Given that p revious research has only examined this relationship between M/P and PA in cross sectional studies a longitudinal study of these associations as adults begin PA programs may help us better understand the relationships of M/P with PA Moreover, no studies have explicitly examined whether salience of M/P is r elated to PA on a daily basis. To fill the gap s in the research, t his study addresse s three aims: ( 1) Determine whether self monitoring of daily M/P mood, and PA results in increased PA compared to a n assessment control random survey condition; (2) Examine the relationship between daily ratings of M/P and PA over time adjusting for daily mood; and ( 3 ) Examine whether M/P predicts PA and psychological functioning beyond an existing and supported model of health behavior change, the SDT Process Model of Behavior Change. The primary hypotheses of this study are that : (1) self monitoring will not increase PA compared to the control condition; (2) daily awareness of M/P will be positively related to greater engagement in PA during that same day ; and ( 3 ) M/P will predict increased engagement of PA over time (12 weeks) beyond the SDT Process M odel of Behavior C hange.

PAGE 29

15 CHAPTER II METHOD Overview This study used a randomized controlled trial design (RCT) and was registered with clinicaltrials.gov ( NCT02538068 ). Participants were randomized to a daily self monitoring of M/P, mood, and PA condition for 4 weeks or to a random survey assessment control. The relationship between daily M/P and PA was examined for the self monitoring group only. A conceptu al model of the relationship of M/P with PA and psychological well being was also tested. The model, presented in Figure 2, wa s adapted from the work of Fortier and colleagues (2012) and was enhanced by adding the crucial M/P component. The study examine d whether the basic psy chological needs posited by SDT we re related to M/P and internalization of PA motivation (behavioral regulations motivation). In turn, i nternalized motivation and M/P we re hypothesized to be related to greater engagement in PA and greater psychological well being. Additionally, it wa s hypothesized that M/P would be directly related to psychological well being as studies have shown that people who engage in intrinsically meaningful life activities that are consistent with their innermost values and aspirations experience greater life satisfaction and well being (Steger, Kashdan, & Oishi, 2008). Participants and Recruitment Participants were recruited from the Anschutz Medical Campus and the greater Denver area in a variety of wa ys. First, some individuals were informed about the study when they began a new membership at the Anschutz Health and Wellness Center (AHWC) Center staff asked potentially eligible individuals if they were interested in having a research assistant contact them about the study. Second, one recruitment email was sent to new

PAGE 30

16 members within one week of joining the center to determine if they were interested in participating in the study. Other methods of recruitment, including flyers on the campus and in the c ommunity, email announcements to University of Colorado Denver (UCD) faculty and staff, posting of the study on the University Clinical Trials website and the AHWC website, and word of mouth were also used. Participants were screened for eligibility to pa rtic ipate. The eligibility criteria were chosen to form a sample of sedentary adults who were joining a fitness center. To be included, participants had to be (1) at least 3 0 years of age; (2) able to read and understand English; (3) sedentary (engaging in < 60 minutes of moderate to vigorous exercise per week) for the last 3 months; and (4) joining the AHWC Individuals were excluded from the study if they (1) ha d medical or physical contraindications to participate in PA ; (2) ha d an existing diagnosis of cardiovascular disease; or (3) we re pregnant. The rationale for including adults age 3 0 and older wa s that M/P becomes more salient for adults who enter middle and older age (Steger, Oishi, & Kashdan, 2009) and PA declines for both gender s and most ethnic groups at middle age (Hawkins et al., 2009) Therefore, middle to older age is a particularly important time to understand the processes that predict long term maintenance of PA and is also particularly apropos for studying M/P in this context. The rationa le for enrolling adults who read and underst ood English wa s that t he majority of the measures had only been validated in English. Participants had to be previously sedentary because the study wa s designed to examine the process of PA adoption. Finally, par ticipants had to be joining the AHWC because a secondary outcome was fitness center attendance. Moreover, because the study was observational and did not involve a formalized intervention, it was thought that participants joining a fitness center would be more likely to attempt to become regularly

PAGE 31

17 physically active Participants who ha d medical or physical contraindications to participate in PA (defined as a positive Physical Activity Readiness Questionnaire [ PAR Q ; Thomas, Reading, & Shephard, 1992] score) were excluded. Participants who were unable to engage in certain forms of PA due to illness or injury would artificially skew the averages of PA in the sample. Additionally, this study wa s designed as a study of primary prevention of cardiovascular diseas e; therefore those with CVD were excluded. Finally, it may not have been safe for pregnant women who were not already active to start an exercise program without supervision from a physician; thus, they were also excluded from this study. Randomization To assess the effect of self monitoring daily M/P and PA p articipants were randomized 1:1 to the self monitoring or the control condition. A blocked randomization sequence was created using a random number generator in Microsoft Excel. The randomization i ncluded eight blocks so that the number of participants randomized to each group would be even for every 20 participants. Participants in the self monitoring condition completed daily surveys for the first four weeks (28 days total). In addition, at the en d of the daily survey, a quote appeared to help prime participants to think about their personal M/P (see Appendix A ). Participants randomized to the control condition completed eight surveys delivered at blocked random intervals over the first 4 weeks. Th e surveys asked participants to recall their activities for the last 24 hours but explicitly did not ask participants to record their PA or rate their M/P. Participants were randomized to get one survey on a weekend day (Saturday or Sunday) and one survey on a weekday (Monday Friday) for each of the four weeks.

PAGE 32

18 Procedure The study procedures were reviewed and approved by the Colorado Multiple Institutional Review Board ( COM IRB). Interested individuals complete d a phone or in person screening prior to enrollment. Research assistants acquired verbal consent for screening from potential participants, and ask ed questions in the screening questionnaire (see Appendix B ) to determine if the individual wa s eligible to p articipate. If eligible, research assistants scheduled an in person meeting to complete consent and baseline measurements. Participants were offered $20 for each month they participated in the study, for a total of $60 for the three months. This was availa ble either as a discounted rate off their AHWC membership s or in the form of a gift card, which they received at the end of the study Additionally, participants were entered into a drawing for a variety of gift cards every time they completed a survey for the study. Participants were observed as they began new PA program s for 12 weeks They completed self report measures at baseline (withi n 2 week s of joining the AHWC), 4 weeks and 12 weeks (see Figure 3 for study timeline ; see Appendices C D and E for questions ) During the baseline visit, research assistants reviewed the full COMIRB approved consent form with participants. Participants completed the baseline questionnaire, and were randomized to either the self monitoring or control condition. Res earch assistants showed participants how to complete the daily measures that were to be sent to them via email. After participants were given an overview of what to expect for the next 12 weeks, research assistants conducted a brief fitness assessment to gather measures of body size (height, weight, body mass index [BMI], body fat percent, and waist circumference) and physical fitness (cardiovascular, strength, and flexibility) The baseline visit took, on average, between

PAGE 33

19 45 6 0 minutes to complete. Participants randomized to the self monitoring condition were asked to complete daily measures (at 8:00 PM each day) of M/P, mood, and PA for the four weeks after baseline (see Appendix E) Participants in the control condition compl eted a survey that included a 24 hour recall of daily activities (e.g., sleeping, eating, working, etc.) on eight random days over the first four weeks Participants in the self monitoring condition also received the 24 hour recall of daily activities on e ight random days. For the purposes of this study, the control condition was considered an assessment control and was not designed to encourage self monitoring. Both the self monitoring and control surveys were designed to take less than 10 minutes to complete. Figure 3 Study T imeline. Study data were collected and managed using REDCap electronic data capture tools hosted at the University of Colorado Denver (Harris et al., 2009). REDCap (Research Electronic Data Capture) is a secure, web based application designed to support data capture for research studies, providing: 1) an intuitive interface for validated data entry; 2) audit trails for tracking data manipulation and export procedures; 3) automated export procedures for seamless da ta downloads to common statistical packages; and 4) procedures for importing data from external sources. Screening Baseline & Randomization 4 week Follow up 12 week Follow up Daily measures via email for 4 weeks and 8 random surveys OR 8 random surveys via em ail over 4 weeks

PAGE 34

20 Participants received email notifications sent directly from REDCap prompting them to complete the surveys at the specified times. At the end of the 12 weeks, participants completed a second in person fitness assessment. Measures for each time point are described in Table 1 Measures As part of their participation in this study, participants complete d a variety of self report measur es. These surveys w ere all completed online through the survey software REDCap. All measures were chosen to satisfy the aims of this project. At baseline, demographics (age, race, ethnicity, education, religious affiliation living situation, family status, and income) and heal th histories were collected to use as covariates in analyses. Additionally, measures of body size (weight, height, BMI waist circumference, body fat percent) and physical fitness were collected at the baseline and 12 week follow up as objective measures o f cardiovascular health risk. An overview of the measures collected is represe nted in Table 1 To examine Aim 2 participants in the self monitoring condition completed daily measures of mood, M/P, and self reported PA via email for four weeks. Participants in both conditions completed 24 hour recalls of daily activities on eight random days over the first four weeks. Daily m ood. Daily positive and negative mood were measured using a positive and negative affect scale p reviously used in a study of daily M/P and daily mood (Steger et al., 2008). Eight items measured positive affect (relaxed, proud, excited, appreciative, enthusiastic, happ y, satisfied, and curious) and five items measured negative affect (sluggish, afraid sad, anxious, and angry). Participants rated their mood using a 5 point Likert type

PAGE 35

21 scale ranging from 1 ( very slightly or not at all ) to 5 ( extremely ). Sums of the positive and negative affect scales have been shown to be positively correlated with the ir respective scales on the Positive and Negative Affect Schedule (PANAS; Watson, Clark, & Tellegen, 1988) ( r Table 1 Measures C ollected in this Study Aim Primary Outcome Secondary Outcomes Primary Predictor Secondary Predictors/ Mediators Covariates Aim 1 PA ( I PAQ SF at 4 week and 12 weeks ) Fitness center attendance M/P SDT mediator, Psychological well being (Subjecti ve Vitality, Life Satisfaction) Self monitoring vs. Control N/A N/A Aim 2 Daily self reported minutes of PA Fitness center attendance Self reported PA intensity Daily M/P N/A Daily mood Aim 3 PA (I PAQ SF at 4 weeks and 12 weeks ) Psychological well being (Subjective Vitality, Life Satisfaction), Body size, Physical Fitness Meaning in Life (MILQ) Purpose in life ( LET; Secondary), Basic Psychological Needs in Exercise, Behavioral Regulations in Exercise Demographics, health history, depressive symptoms Note. IPAQ SF = International PA Questionnaire Short Form; LET = Life Engagement Test; M/P = Meaning and purpose; MILQ = Meaning in Life Questionnaire; PA = Physical Activity; S DT = Self Determination Theory

PAGE 36

22 = .55 .57) and to have very high internal consistency (" = .97 .98). In this study, the internal consistency was very high for positive affect (" s = 86 .94 ) and acceptable to high for negative affect (" s = .7 0 .86 ) over the 28 days Daily M/P s alience Daily M/P salience was measured using the Thoughts of Meaning Scale (T O MS), which was developed for this study. The T O MS included the 2 item Daily Meaning Scale (DMS ; Steger et al., 2008) Participants rate d the extent to which the two statements we re true for them at the moment ("How meaningful does your life feel?" and "How much do you feel your life has a purpose?"). To increase the potency of the measurement self monitoring as a possible intervention and to more directly assess M/P salience, eight additional items were added to this scale. Participants rate d the extent to which they thought about M/P that day (e.g., "How much have yo u thought about your purpose in life today?"). Participants complete d these items using Likert type rating scales ranging from 1 ( not at all ) to 7 ( absolutely [DMS] or quite a bit [T O MS]). The DMS has been shown to have very strong reliability ( = .98) and to be positively related to daily engagement in eudaimonic behaviors (i.e., behaviors that are consistent with values; Ryan & Deci, 2001) ( = .13) in daily studies of M/P (Steger et al., 2008) In this study, the internal consistency of all 10 items over the 28 days was very high ( s = .88 .96 ). PA Both self report and objective measures of PA were gathered. Participants report ed the type of exercise activities they engaged in (e.g., jogging, swimming, yoga etc. ), the duration of the act ivity (min/day), and intensity of engagement (using Borg's Category Ratio exertion scale also known as the Ratings of Perceived Exertion Scale ; Borg, 1998 ). To rate the average intensity of their activities, participants were shown a scale ranging f rom 6 to 20, where 6 corresponded to no exertion at all, 11 corresponded to light activity, 15

PAGE 37

23 corresponded to hard (heavy) activity, and 20 corresponded to maximal exertion This method was used in a previous daily diary study of exercise (Lutz, Stults Koehmain en, & Bartholomew, 2010) and allow ed for analysis of different components of PA (duration, frequency [ days/week ] and intensity of PA ). In addition to daily self report of PA the number of days that participants attended the fitness center was obtained fr om fitness center records 24 hour activity recall. A 24 hour recall of activities was delivered on 8 random days over the first 4 weeks to participants in both conditions. Participants were asked to recall their activities from a list of 46 common activi ties (e.g., sleeping, driving, working, eating, etc.) for every half hour period over the previous 24 hours (from 8:00 p.m. the night before to 7:30 p.m. the night the survey was delivered). Participants could report up to two activities for each half hour period. There were also two "other" categories that participants could specify if they were engaging in an activity that was not on the list. To examine Aim 3 participants complete d measures of the model in Figure 2 at baseline, 4 and 12 weeks, and those were as follows: Psychological n eeds s atisfaction. Satisfaction of the three basic needs in SDT (autonomy, relatedness, and competence) was measured using the Basic Psychological Needs in Exercise Scale (BPNES) (Vlachopoulos & Michailidou, 2006) The BPNES ha s 1 1 items with four items corresponding to autonomy ( e.g., "The way I exercise is in agreement with my choices and interests." ), four items corresponding to competence ( e.g., "I am able to meet the requirements of my exercise program." ), and thr ee items corresponding to relatedness ( e.g., "My relationships with the people I exercise with are close." ). Participants rate d their needs satisfaction on a 5 point Likert type scale ranging from 1 ( I don't agree at

PAGE 38

24 all ) to 5 ( I completely agree ). The BPN ES has demonstrated strong internal consistency ("s ranging from .81 to .92) high test retest reliability over four weeks (ICC = .97), and construct validity (Vlachopoulos & Michailidou, 2006) Perceived competence has been shown to be positively related to exercise frequency ( r = .21; Vlachopoulos & Michailidou, 2006). The internal consistency of the whole scale was high at baseline ( = .85), 4 weeks ( = .90), and 12 weeks ( = 92 ). Behavior al r egulations m otivation. Motivation internalization for exercise was measured using the Behavioral Regulation in Exercise Questionnaire 2 ( Markland & Tobin, 2004; Mullan, Markland, & Ingledew, 1997; BREQ 2). The 19 item BREQ 2 measured motivations for ex ercise on the SDT continuum (see Figure 1). The 5 subscales were : amotivation ( e.g., I don't see why I should have to exercise ."); external regulation ( e.g., I exercise because other people say I should .") ; introjected regulation ( e.g., I feel guilty when I don't exercise ."); identified regulation ( e.g., It's important to me to exercise regularly ."); and intrinsic motivation ( e.g., I exercise because it's fun ."). Since the development of the BREQ 2, four items used to assess integrated regulation for exercise were developed (e.g., "I exercise because it is consistent with my life goals" ; Wilson, Rodgers, Loitz, & Scime, 2006) These items were added to the existing BREQ 2 Participants rate d the extent to which they engage d in exercise (or d id not eng age in exercise) for each of the reasons on a 5 point Likert type scale ranging from 0 ( not true for me ) to 4 ( very true for me ). The six subscales we re combined using the bifurcation approach outlined by Wilson and colleagues (Wilson, Sabiston, Mack, & Bl anchard, 2012) into two scaled scores: autonomous and controlled regulation Autonomous regulation was the average of the intrinsic, integrated, and identified scales whereas controlled regulation was the average of the external and introjected scales.

PAGE 39

25 The BREQ 2 has been shown to have good psychometric properties, including good internal consistency ("s range from .78 .93; Markland & Tobin, 2004; Wilson et al., 2006). Additionally, m ore internalized forms of regulation (identified, integrated, and int rinsic) have been shown to be positively related to psychological needs satisfaction and to self reported PA (Wilson et al., 2006). In this study, the internal consistency of the autonomous regulation score was very high at all three time points ( "s = .92 .94). The controlled regulation score was acceptable at all three time points ( "s = .78 .82). M/P. M/P was measured using The Mean ing in Life Questionnaire (MILQ; Steger et al., 2006) and the Life Engagement Test (LET; Scheier et al., 2006) Meaning in l ife. The 10 item MILQ measure d the extent to which a person perceives his or her life as meaningful and searches for meaning in life. This measure included two subscales: presence measure d the sense that one's life is meaningful (e.g., "My life has a clear sense of purpose" ) whereas search measured the drive and orientation towards finding a sense of meaning in life (e.g., "I am always looking to find my life's purpose" ). Participants rate d the extent ea ch statement wa s true for them on a 7 poin t Likert type scale ranging from 1 ( absolutely untrue ) to 7 ( absolutely true ). Responses were summed for each subscale. Steger and colleagues (2006) found that both scales demonstrate good internal consistency ( "s = .86 and .92 for presence and search, res pectively) and have moderate test retest reliability over one month ( r s = .70 and .73 for presence and search, respectively). Evidence for convergent and discriminant validity indicate d that the presence subscale wa s positively associated with health and well being indicators (i.e., life satisfaction, joy) and negatively associated with depressive symptoms and negative emotionality. Conversely, the search subscale wa s not associated with health and well being indicat ors but wa s positively

PAGE 40

26 associated with depressive symptoms and negative emotionality. In this study, the internal consistency was very high at all three time points ( "s = .89 .94). Purpose in l ife. The 6 item LET measured purpose in life and the extent to which individuals believed their activities were valuable and important (e.g., "To me, the things I do are all worthwhile." ) Participants rate d the extent to which they agree d with each item on a 5 point Likert type scale ranging from 1 ( strongly disag ree ) to 5 ( strongly agree ). Odd items were reverse scored and the six items were summed for a total score. Scheier and colleagues (2006) examined the test retest reliability of this measure in four different samples and found that the LET was moderately st able ( r s ranged from .61 .76) over 4 months. Additionally, they tested the convergent and discriminant validity of the measure and found that it was positively associated with many health and well being indicators (i.e., life satisfaction, general physical heal th) and negatively associated with depressive symptoms. The internal consistency for this measure was high in a previous sample of adult community members (" = .85; Hooker & Masters, 2014 ). In this study, the internal consistency was high to very high at a ll three time points ( "s = .85 .93). PA In addition to the PA measures described above at baseline 4 weeks, and 12 weeks, participants complete d the International Physical Activity Questionnaire Short Form (IPAQ SF) a self report measure of activit y in the last 7 days ( Craig et al., 2003 ) The short form consists of seven items and ask s participants to estimate the amount of moderate and vigorous exercise, walking, and sedentary activity they engaged in during the past 7 days. Items we re intended to provide separate scores on walking, moderate intensity, and vigorous intensity activity and computation of a total score requires the multiplication of the duration (in minutes) times the frequency (days) of walking, moderate intensity, and

PAGE 41

27 vigorous intensity activities). Categorical (Low, Medium, and High activity) and continuous (minutes, metabolic equivalent of task [MET] minutes of PA ) scores were calculate d for each participant. Evidence suggests that t he IPAQ SF has demonstrated accepta ble measurement properties, on par with other self report measures of PA Test retest reliability over 8 days has been shown to be very good, with r s ranging from .71 .91, and the self report version is moderately correlated with objectively measured (acce lerometer) PA over the same 7 day period ( r s ranging from .26 to .47; Craig et al., 2003). Psychological w ell b eing. Psychological well being was measured using two measures: The Subjective Vitality Scale (SVS ; Ryan & Frederick, 1997 ) and The Satisf action with Life Scale (SWLS; Diener, Emmons, Larsen, & Griffin, 1985). Subjective v itality. The 7 item SVS measure d feeling active, alive, enthusiastic, and energetic (e.g., "I feel alive and vital" ) Participants rate d the extent they generally fel t thi s way on a 7 point Likert type scale ranging from 1 ( not at all true ) to 7 ( very true ). Item 2 ( "I don't feel very energetic" ) wa s dropped (per author instructions and confirmatory factor analysis results, see Bostic, Rubio, & Hood, 2000 ) and the rest of t he items we re summed for a total SVS score. Previous psychometric studies demonstrated that the SVS correlate d positively with other positive measures of well being (e.g., self esteem, self actuali zation, satisfaction with life) and negatively with poor ps ychological well being (e.g., depression, anxiety, psychopathology), and wa s int ernally consistent (" = .84 .86) (Ryan & Frederick, 19 97 ) The internal consistency of the SVS in this study was very high at all three time points ( "s = .88 .91). Life satisfaction. On th e SWLS participants rate d their agreement with five statements on a 7 point Likert scale ranging from 1 ( strongly disagree ) to 7 ( strongly agree ).

PAGE 42

28 Responses we re summed so that higher scores correspond ed with greater satisfaction with l ife. Diener and colleagues (1985) conducted a thorough assessment of the SWLS's reliability and validity. Previous assessments of reliability indicate d that this measure is internally consistent (" = .87) and ha d good two month temporal stability ( r = .82) Evidence for construct validity indicate d that the SWLS correlate d positively with other measures of subjective well being and negatively correlate d with measures of personality psychopathology. Evidence for criterion validity indicate d that the measure correlate d highly with interviewer's rating of the individual's satisfaction with life (Diener et al., 1985) In the current study, t he scale demonstrated high internal consistency at all three time points ( "s = .86 .90). Depressive s ymptoms. Depressive symptoms were used as a covariate in the path analysis model. The 8 item Patient Health Questionnaire 8 (PHQ 8; Kroenke & Spitzer, 2002) was used to measure depressive symptoms. Participants rate d the extent to which they were bothered by a series of eigh t problems (e.g., "little interest or pleasure in doing things" ) over the past 2 weeks on a scale ranging from 0 ( not at all ) to 3 ( nearly every day ). If they experienced any of the eight problems, they were asked to rate how difficult those problems ha d b een for them ( not difficult at all, somewhat difficult, very difficult, or extremely difficult). The PHQ 8 wa s derived from the PHQ 9, but the ninth item assessing suicidal ideation was omitted as per recommendations from the aut hors because this study (1) used self administered survey s ; and (2) depression wa s assessed as a secondary measure (Kroenke & Spitzer, 2002). Evidence for validity for this scale indicate d that higher scores on the PHQ were related to greater likelihood of being diagnosed with any d epressive disorder

PAGE 43

29 (Kroenke & Spitzer, 2002). The internal consistency of the PHQ 8 was high at all three time points in this study ( "s = .82 .85). Physical f itness. Physical fitness was assessed using the standardized fitness test available at AHWC. Participants completed three physical fitness tests, including one for aerobic fitness ( Young Men's Christian Association [ YMCA ] 3 minute step test; Golding, 2000 ) one f or strength ( grip strength as assessed by Takei Hand Grip Dynamometer A5401; Takei Scientific Instruments, Tokyo, Japan ) and one for flexibility ( sit and reach test; Holt, Pelham, & Burke, 1999) Additionally, participants' resting heart rates were gather ed at the beginning of the fitness test. Raw scores for each test (resting heart rate, step test, grip strength, and sit and reach) were recoded into categorical scores based on fitness targets by age and gender, with scores ranging from 1 (poor fitness) t o 5 (excellent fitness). The fitness targets were based on the guidelines from the American C ollege of Sports Medicine (ACSM; Percia, Davis, & Dwyer, 2012) The four scores were summed for a total score (ranging from 4 to 20) and then recoded into a fitnes s percent score ranging from 0 100%. M/P and PA c onnection. As an exploratory measure, two items were embedded in the BREQ 2 to assess whether participants made an active connection between M/P and reasons to do PA Participants rated the two items on the same scale as the BREQ 2, or as the extent to which the two statements were true for them (" Exercise gives me more energy to do the things that really matter to me in life. and Engaging in regular exercise helps me reach my life goals. "). The two items were summed for a total score, and had acceptable internal consistency at each of the three time points ( = .69, .73, and .70 at baseline, 4 weeks, and 12 weeks, respectively).

PAGE 44

30 Power Analysis The sample size was determined based on power analyses as well as feasibility. GPower 3.1 (Faul, 2007) was used to estimate power for the three aims. Aim 1. Effect of M/P, m ood, and PA s elf m onitoring on PA Power for Aim 1 was estimated for a two group, r epeated measures design with three time points. The group by time interaction was of primary interest. Assuming a two tailed test with = .05, a moderate correlation between the three time points ( r = .50), and power = .80, a sample size of 160 was needed to detect a small difference ( f = .10) between the groups. Aim 2. Relationship of d aily M/P s alience to d aily PA. P ower for a within subjects repeated measures design with 25 measurements per subject, allowing for three days of missed data collection per subject (out of 28 total days) was estimated Additionally, a small moderate correlation within the individual of PA duration across days ( r = .30) was used to estimat e the correlation of the repeated measures over time. A sample size of 42 would have 95 % power to detect a small to moderate relationship ( f = .15) between daily M/P and PA given an alpha of .05. Because participants were randomly assigned to this conditio n in a 1 :1 fashion and the sample size requirements for Aim 1 and Aim 3 were 160 (see below), 80 subjects wer e randomized to this condition, allowing for 39 subjects to dropout while still maintaining this level of power. Aim 3 Examining the SDT process m odel of PA adoption with and without M/P. Because path analysis is essentially a series of multiple regressions, power for the path analysis was calculated using the largest regression in the model ( 6 predictors in the model plus 7 covariates). A sample size of 131 was needed to detect a small moderate relationship ( f = .15) between the 13 predictors and the outcome ( PA ) with 80% power. However, given

PAGE 45

31 the longitudinal nature of the study, a 20% attrition rate wa s expected. Therefore, a target sample size of 1 60 was recruited for the project. Data Analysis Analyses were conducted in SAS 9. 4 (SAS Institute, Inc., 2015 ) and M plus 7. 4 (MuthÂŽn & MuthÂŽn, 201 5 ) SAS was used to manage the data and calculate descriptive statistics. Descriptiv e statistics were used to examine distributions and assess potential violations of assumptions. Variables that did not meet assumptions were transformed prior to analysis. Specifically, for PA square root transformations were used to normalize the data. B aseline differences between the self monitoring and control group were examined using the appropriate statistical tests ( 2 for categorical data and t tests for continuous data). Longitudinal PA patterns were graphed using a spaghetti plot to determine if there were any prototypical patterns of change. Additionally, patterns of missing data were examined to determine whether missing data we re missing at random, missing completely at random, or mi ssing not at random. Individuals who missed the 4 week or 12 week follow up ass essments were compared to those who did not miss those assessments on baseline variables using the appropriate tests ( 2 for categorical data and t tests for continuous data). T he proportion of completed daily surveys was calculated and correlated with baseline variables. Non random missing data patterns were taken into account when interpreting the results. Specifically, if the missing data analysis suggest ed that missing data w ere related to the relationship between the primary predictor and the outcome (e.g., participants who were missing more follow up data also had lower M/P and PA at the beginning of the study), then these patterns were noted in the limitations of the study. Finally, bivariate correlations among all study predictors, covariates, and outcomes were calculated to examine patterns in the data.

PAGE 46

32 Aim 1. Effect of M/P, m ood, and PA s elf m onitoring on PA Mixed repeated measures models were used to examine the differences between treatment groups in PA and M/P over time (at baseline, 4 weeks and 12 weeks) Each model included three predictors: treatment group (between factor), time (within factor), and the group by time interaction. Models were estimated using u nstructured covariance matrices and restricted maximum likelihood (REML) estimation which is recommended for estimating the covariance structure of the data (Fitzmaurice, Laird, & Ware, 2011). PA measured by the IPAQ SF was the primary outcome. To exami ne secondary outcomes, the model was repeated with fitness center attendance (number of visits per week at each time point ) and different self reported PA intensities (vigorous, moderate, walking) as the dependent variables Additionally, the model was repeated with hypothesized mediators, including M/P (measured by the MILQ and LET) basic psychological needs satisfaction (BPNES), and autonomous regulation as dependent variables. Finally, the models were repeated with psychological well being and depres sion variables as exploratory outcomes. The group by time interactions were of primary interest. Group means over time were graphed using time plots. Aim 2 : Relationship of d aily M/P s alience to d aily PA Multi level models were used to examine the relatio nship between daily M/P and daily self reported PA (frequency, intensity, duration) controlling for mood and clustering at the time level. Multi level models contro l l ed for repeated measures within subjects to ensure that standard errors we re appropriately estimated and Type 1 error rates we re not inflated. Models were run in M plus which control led for clustering of data and use d cases that have partially complete data through full information maximum likelihood estimation. As a secondary outcome, daily M/P was examined in relation to fitness center visits as a measure of objective PA to examine

PAGE 47

33 consistency with the results of the self report A multilevel model with the same predictors (meaning, positive affect, and negative affect) was used to predict whether or not participants visited the fitness center that day (coded as a binary outcome : 1=visit; 0=no visit). Aim 3 : Examining the SDT process model of PA adoption with and without M/P. All SDT and M/P process variables we re assessed at the three main assessments. P ath analysis exami ned variations of the model in Figure 2. M plus was used for analysis so that full information maximum likelihood account ed for any missing data. Models were tested cross sectionally at each assessment (baseline, 4 week, and 12 week) and then longitudinally to examine the change from baseline to 12 weeks. The initial model (base SDT Pr ocess M odel of Behavior C hange without M/P) was tested for goodness of fit u sing several fit indices, including Akaike's Information Criterion (AIC), Bayesian Information Criterion (BIC), # 2 test of model fit, comparative fit index (CFI), root mean square error approximation (RMSEA), and standardized root mean residual (SRMR) Hu and Bentler's (1999) recommended cut points for each fit index were used to evaluate the fit of each model including a non significant # 2 RMSEA < .06, CFI > .95, and SRMR < .08 All models controlled for six covariates on PA (gender, age, employment [ful l time vs. not full time], income [! $40,000 annually v. < $40,000 annually], ethnicity [white v. non white], and marital status [married v. not married]). In the cross sectional models, basic psychological needs satisfaction predicted autonomous regulation, which predicted self reported PA (assessed by the IPAQ SF) which in turn predicted psych ological well being (combined vitality and life satisfaction). Basic psychological needs satisf action then directly predicted psychological well being. As a second step, meaning in life (the presence subscale of the MILQ) was added to the models.

PAGE 48

34 Specifically, basic psychological needs satisfaction predicted meaning in life, which then predicted PA and psychological well being. The original SDT Process Model of Behavior Change was compared to the modified model using comparative fit statistics ( AIC and BIC ) and change in variance in PA accounted for by the model ( R 2 ). To simplify the longitudinal mod els, psychological well being was not included. Two types of longitudinal models were examined. The first was designed to predict absolute levels of PA at 12 weeks. T he original SDT Process Model of Behavior Change was modeled first, with baseline basic ps ychological needs satisfaction predicting 4 week autonomous regulation and 4 week autonomous motivation predicting 12 week PA To modify the model, baseline meaning in life was added to the model. Basic psychological needs satisfaction at baseline was use d to predict baseline meaning in life, which was then used to predict 12 week PA A second type of longitudinal model was used to predict residualized change in PA from baseline to 4 weeks and from 4 weeks to 12 weeks In the base SDT Process Model of Behavior Change psychological needs satisfaction at time 1 predict ed change in autonomous regulation (between baseline and 4 weeks ) which predict ed change in self reported PA (assessed by the IPAQ SF from baseline to 4 weeks and from 4 weeks to 12 weeks ) The modified SDT model with meaning in life added was then assessed for goodness of fit. Because meaning in life was considered relatively stable, baseline meaning in life was used to predict 4 week and 12 week changes in PA The modified SD T model with the addition of meaning in life was compared to the model without meaning in life to determine whether the addition of meaning in life improved the prediction of PA Comparative model fit statistics

PAGE 49

35 ( AIC and BIC ) were used to compare the overall models and a change in R 2 determine d the extent to which prediction of PA wa s improved. Exploratory a nalyses. To examine the effects of meaning salience on other 12 week outcomes (e.g., PA as measured by the IPAQ SF, body size and composition, physical fitness, and psychological well being), meaning salience over the 28 days was averaged for each person in the self monitoring arm. Additionally, the standard deviation in meaning salience was calculated for each person. Pearson correlations were used to examine the relationships between the average and standard deviation of meaning salience in relation to these outcomes. Subsequently a series of linear regression models were used to examine t he relationships between meaning salienc e (average and variability) and fitness center visits total PA and M/P (MILQ and LET) at 12 weeks All models controlled for gender, age, race, employment status, marital status, and income. Pearson correlations w ere also used to assess the relationship of global M/P (MILQ and LET, respectively) in relation to body size and composition, physical fitness, PA intensity, and the number of visits at the fitness center. Finally, Pearson correlations were used to assess the relationships between the exploratory M/P and PA connection variable to the same outcomes.

PAGE 50

36 CHAPTER III RESULTS A total of 160 adults consented to participate in the study and completed basel ine questionnaires. Demographic and health characteristics of the entire sample and by arm are presen ted in Table 2 On average, participants w ere 43.3 years old ( SD = 11.4, range = 30 72) The sample was predominantly fem ale (76.9%) and highly educated; nearly one half having completed a graduate or professional degree (48.7%) and an additional third had completed a 4 year college degree (34.8%). Additionally, the vast majority of the sample was employed (80.6% employed full time) and reported high annual household income (57.5% reported an annual income of $80,000 or more). The sample was not very religious; nearly one third of the sample (31.3%) reported that their religious affiliation was "I consider myself spiritual, but not religious." Nearly one third (31.3%) of the sample reported that it had been 6 months or less since they were regularly physically active, and only a small percentage (1.9%) reported that they had never been regularly physically active. The majority of participants (9 0.0%) reported that there had been times in their lives when they were regularly physically active for at least 2 months and subsequently not physically active for at least 3 months. When asked how many times in their adult lives that had happened, the med ian response was 4 ( range = 1 50 times). The most endorsed reason for stopping regular PA was lack of time due to work (51 .9%), followed by lack of interest in PA (28.8%) and personal stress (28.8%). More than two thirds (68.8%) reported that they had pa rticipated in athletics in their lifetimes.

PAGE 51

37 Table 2 Baseline D emographics for the Entire S ample and B etween G roups Variable All ( N = 160) n (%) Self Monitoring ( n = 80) n (%) Control ( n = 80) n (%) # 2 p Femal e 123 (76.9) 62 (77.5) 61 (76.3) 0.04, .85 Race/Ethnicity African American Asian Hispanic or Latino/a White, non Hispanic Mixed or other 12 (7.5) 9 (5.6) 10 (6.3) 118 (73.8) 11 (6.9) 7 (8.8) 6 (7.5) 5 (6.3) 57 (71.3) 5 (6.3) 5 (6.3) 3 (3.8) 5 (6.3) 61 (76.3) 6 (7.5) 1.55, .82 Marital Status Single, never married Currently married Cohabiting or in a long term relationship Divorced Widowed ( n = 159) 41 (25.8) 82 (51.6) 13 (8.2) 18 (11.3) 6 (3.1) ( n = 79) 24 (30.4) 32 (40.5) 10 (12.7) 11 (13.9) 2 (2.5) 17 (21.3) 50 (62.5) 3 (3.8) 7 (8.8) 4 (5.0) 10.00, .04 Employment Status Employed full time Employed part time Retired Partially disabled Unemployed Student Homemaker 129 (80.6) 9 (5.6) 6 (3.8) 2 (1.3) 2 (1.3) 10 (6.3) 2 (1.3) 62 (77.5) 5 ( 6.3 ) 2 (2.5) 1 (1.3) 2 (2.5) 6 (7.5) 2 (2.5) 67 (83.8) 4 (5.0) 4 (5.0) 1 (1.3) 0 (0.0) 4 (5.0) 0 (0.0) 7.37, .39 Education High school or equivalent Some college 2 year college degree 4 year college degree Graduate or professional degree ( n = 158) 5 (3.2) 13 (8.2) 8 (5.1) 55 (34.8) 77 (48.7) 2 (2.5) 7 (8.8) 4 (5.0) 34 (42.5) 22 (41.3) ( n = 78) 3 (3.9) 6 (7.7) 4 (5.1) 21 (26.9) 44 (56.4) 4.90, .30 Annual household income Less than $20,000 $20,000 $39,999 $40,000 $59,999 $60,000 $79,999 $80,000 $99,999 $100,000 or more 7 (4.4) 18 (11.3) 25 (15.6) 18 (11.3) 31 (19.4) 61 (38.1) 7 (8.8) 12 (15.0) 12 (15.0) 9 (11.3) 15 (18.8) 25 (31.3) 0 (0.0) 6 (7.5) 13 (16.3) 9 (11.3) 16 (20.0) 36 (45.0) 11.06, .05

PAGE 52

38 Table 2 cont. Variable All ( N = 160) n (%) Self Monitoring ( n = 80) n (%) Control ( n = 80) n (%) # 2 p Religious Affiliation Catholic Protestant Christian Latter Day Saint Jewish 25 (15.6) 36 (22.5) 1 (0.6) 9 (5.6) 17 (21.3) 17 (21.3) 0 (0.0) 1 (1.3) 8 (10.0) 19 (23.8) 1 (1.3) 8 (10.0) 3.84, .05 0.14, .71 1.00, .32 5.77, .02 Muslim Hindu Buddhist Atheist I consider myself spiritual but not religious None Other 2 (1.3) 1 (0.6) 1 (0.6) 4 (2.5) 50 (31.3) 25 (15.6) 12 (7.5) 2 (2.5) 0 (0.0) 0 (0.0) 2 (2.5) 23 (28.8) 15 (18.8) 7 (8.8) 0 (0.0) 1 (1.3) 1 (1.3) 2 (2.5) 27 (33.8) 10 (12.5) 5 (6.3) 2.03, .15 1.00, .32 1.01, .32 0.00, 1.00 0.47, .50 1.19, .28 0.36, .55 Family History of Coronary Heart Diseas e 78 (48.8) 45 (56.3) 43 (53.8) 1.60, .21 Family History of Stroke ( n = 159) 67 (42.1) ( n = 79) 34 (43.0) 33 (41.3) 0.05, .82 Family History of Hypertension 96 (60.0) 45 (56.3) 51 (63.8) 0.94, .33 Family History of Diabetes 76 (47.5) 38 (47.5) 38 (47.5) 0.00, 1.0 Family History of Obesity 74 (46.3) 31 (38.8) 43 (53.8) 3.62, .06 Smoking Status Non smoker Former smoker (quit in the last 6 months) Current smoker 151 (94.4) 3 (1.9) 6 (3.8) 74 (92.5) 3 (3.8) 3 (3.8) 77 (96.3) 0 (0.0) 3 (3.8) 3.06, .22 Time Since Last Regular PA 4.01, .68 Less than 6 months 6 months 1 year 1 2 years 2 5 years 5 10 years More than 10 years Never regularly active 50 (31.3) 39 (24.4) 24 (15.0) 30 (18.8) 8 (5.0) 6 (3.8) 3 (1.9) 27 (33.8) 20 (25.0) 13 (16.3) 15 (18.8) 3 (3.8) 1 (1.3) 1 (1.3) 23 (28.8) 19 (23.8) 11 (13.8) 15 (18.8) 5 (6.3) 5 (6.3) 2 (2.5) Why stopped most recent activity Lack of time because of work Lack of time because of household duties 83 (51.9) 22 (13.8) 37 (46.3) 12 (15.0) 46 (57.5) 10 (12.5) 2.03, .15 0.21, .65

PAGE 53

39 Table 2 cont. Variable All ( N = 160) n (%) Self Monitoring ( n = 80) n (%) Control ( n = 80) n (%) # 2 p Why stopped most recent activity Lack of time because of children Lack of time because of social activities Lack of time because of spouse Lack of money Lack of facilities Lack of PA partner Lack of interest in PA Health problems Injury Season or weather change Personal stress Other 29 (18.1) 12 (7.5) 12 (7.5) 26 (16.3) 15 (9.4) 26 (16.3) 48 (30.0) 17 (10.6) 25 (15.6) 34 (21.3) 46 (28.8) 18 (11.3) 15 (18.8) 6 (7.5) 5 (6.3) 15 (18.8) 10 (12.5) 12 (15.0) 19 (23.8) 8 (10.0) 14 (17.5) 17 (21.3) 22 (27.5) 5 (6.3) 14 (17.5) 6 (7.5) 7 (8.8) 11 (13.8) 5 (6.3) 14 (17.5) 29 (36.3) 9 (11.3) 11 (13.8) 17 (21.3) 24 (30.0) 13 (16.3) 0.04, .84 0.00, 1.0 0.36, .55 0.73, .39 1.84, .18 0.18, .67 2.98, .08 0.07, .80 0.43, .51 0.00, 1.0 0.12, .73 4.01, .04 Participated in athletics 110 (68.8) 59 (73.8) 51 (63.8) 1.86, .17 Note. Sample size is only reported in a cell if there are missing data. There were no differences in age between the self monitoring ( M = 43.4, SD = 10.7) and control ( M = 43.1, SD = 12.0) groups at baseline, t (158) = 0.20, p = .84. However, the control group was significantly more likely to be married ( p = .04) and report a Jewish religious affiliation ( p = .02) and less likely to report a Catholic religious affiliation ( p = .05) than the self monitoring group. There was also a trend towards the control group reporting a greater annual income than the sel f monitoring group ( p = .05). There were no other significant differences between the self monitoring and control groups at baseline.

PAGE 54

40 Study Flow & Missing Data Analysis Of the 160 participants that consented and completed baseline, 14 0 ( 87.5 %) participants completed the 4 week follow up survey and 14 0 ( 87.5 %) participants completed the 12 week follow up survey and visit (see Figure 4 for CONSORT diagram). Seven participants (4.3%) withdrew from the study. Two stated that they did not have enough time, one did not provide a reason, and four stated other reasons for withdrawal (e.g., stressful life events, disliking the online survey questions or format). Eleven participants (6.8%) were lost to follow up and could not be reached to complete the stu dy after multiple contact attempts Participants in the self monitoring group were not significantly more likely to miss the 4 week assessment, 2 (1) = 1.11 p = 29, or the 12 week assessment 2 (1) = 0.25 p = 62 compared to the control group. Aims 1 & 3: Baseline, 4 week, and 12 week m issing d ata. Participants who missed the 4 week assessment were compared to participants who completed the 4 week survey. Those who did not complete the survey reported significantly greater perceived competence i n PA at baseline ( M = 11.7, SD = 3.5) than those who did complete the survey ( M = 9.6, SD = 3.3, t [154] = 2.84, p = .005). Moreover, those who did not complete the survey were significantly less satisfied with their lives at baseline ( M = 19.6, SD = 8.0) than those who completed the survey ( M = 23.8, SD = 6.5, t [157] = 2.35, p = .019). Those who did not complete the survey reported marginally more total PA at baseline ( square rooted METs, M = 40.0, SD = 25.2) compared to those who completed the survey ( M = 31.3, SD = 17.7, t [158] = 1.78, p = .07). There was also a trend that those who did not complete the 4 week survey were more likely to make less than $40,000 per year than those who completed the survey, 2 (1) = 3.23, p = .07.

PAGE 55

41 The patterns for missing 12 week surveys were also examined. Those who did not complete the 12 week survey were more likely to be older ( M = 49.8 years, SD = 11.4 years) than those who completed the survey ( M = 42.4, SD = 11.1, t [158 ] = 2.66 p = .00 9 ). They were also rep orted significantly less vitality ( M = 20.7, SD = 7.1) and life satisfaction ( M = 19.5, SD = 8.0) at baseline than those who completed the survey (Vitality: M = 24.3, SD = 6.6, t [15 6 ] = 2. 18 p = .0 3; Life Satisfaction: M = 23.9, SD = 6.5, t [157] = 2.61 p = .01). There was a trend that those who did not complete the survey were less likely to be married than those who completed the survey, 2 (1) = 3.53, p = .06. Aim 2: Daily m issing d ata. With 80 participants randomized to receive the daily surveys, there were 2240 possible surveys distributed. Of these, 1813 (80.9%) surveys were accessed (some questions were answered ) and 1691 (75.5%) were complete d (all questions were answered) Each par ticipant completed a median of 24 of the 28 daily surveys (85.7%), with a range of 2 to 28 surveys completed. Study Variables at Baseline, 4 weeks, and 12 weeks Study variables were examined at baseline, 4 weeks, and 12 weeks to determine if there were lo ngitudinal patterns or trends over time. Descriptive statistics on these variables are presented in Table 3. Of note, PA variables were all significantly positively skewed and leptokurtic; thus, a square root transformation was used to normalize the data. Trajectories of individual participants' PA over the 12 weeks is plotted in Figure 5. The spaghetti plot suggest ed that that average person increases in PA from baseline to 4 weeks slightly, and then PA levels flatten from 4 weeks to 12 weeks. However, so me individuals show drastic increases in PA from baseline to 4 weeks and drastic decreases from 4 weeks to 12 weeks. The plot demonstrated that the greatest variability in PA was at 4 weeks.

PAGE 56

42 Figure 4. CONSORT Diagram. Undeliverable ( n = 60) Not eligible ( n = 52) Baseline and Randomization ( n = 160) Self Monitoring Condition ( n = 80 ) Random survey control ( n = 80 ) Refused ( n = 5) No contact ( n = 29) Refused ( n = 14) No contact ( n = 3) Dropped out ( n = 3 ) Lost to follow up ( n = 4) Missed survey ( n = 3) Dropped out ( n = 2 ) Lost to follow up ( n = 3) Missed survey ( n = 1) Partial Response ( n = 4) Dropped out ( n = 0 ) Lost to follow up ( n = 1) Partial Response ( n = 2) Dropped out ( n = 2 ) Lost to follow up ( n = 3) Responded ( n = 292) Emails sent to new members ( n = 948) Contacts from other recruitment sources ( n = 59) Screened for eligibility ( n = 248) Refused ( n = 3) Not interested ( n = 1 ) Not enough compensation ( n = 2 ) Not eligible ( n = 7 3 ) Too active ( n = 64 ) Too young ( n = 5 ) Medical reasons ( n = 3 ) Not planning to become regularly active ( n = 1) Completed 4 week ( n = 70 ) Completed 4 week ( n = 7 0 ) Completed 12 week ( n = 7 0 ) Completed 12 week ( n = 70 ) Eligible, no baseline ( n = 12)

PAGE 57

43 Table 3 Descriptive S tatistics of S tudy V ariables at B aseline, 4 weeks, and 12 weeks Baseline Variable n M SD Min Max Skew Kurt Total PA 1 -160 32.2 18.6 0.1 97.2 0.95 1.35 Vigorous PA 1 -160 15.2 17.1 0.1 69.3 0.87 0.02 Moderate PA 1 -160 10.2 11.8 0.1 60.0 1.38 2.31 Walking Activity 1 -160 20.3 14.3 0.1 64.5 1.11 1.34 Fitness Center Attendance -159 2.4 1.7 0 7 0.49 0. 32 Physical Fitness -153 48.1 21.0 0 93.8 0.12 0.61 Meaning in Life .92 159 26.6 5.8 5 35 0.86 0.71 Purpose in Life .93 157 21.5 8.1 12 30 0.28 0.82 Basic Psychological Needs Satisfaction .85 154 29.8 8.1 11 48 0.00 0.29 Autonomy .80 160 11.9 3.6 4 20 0.02 0.45 Competence .75 156 9.5 3.4 4 20 0.35 0.32 Relatedness .88 158 8.5 3.5 3 15 0.01 0.97 Autonomous Regulation .92 157 9.2 3.2 0 16 0.17 0.11 Controlled Regulation .79 159 4.4 2.7 0 11 0.41 0.47 Psychological Well being 2 .92 157 47.2 12.3 11 72 0.34 0.35 Subjective Vitality .88 158 23.9 6.7 6 41 0.03 0.05 Life Satisfaction .90 159 23.4 6.8 5 35 0.58 0.46 Depressive Symptoms .85 154 5. 6 4.4 0 24 1.29 2.40 Body Mass Index -156 29.9 7.6 17.9 68.1 1.63 4.74 Body Fat % -151 33.4 8.9 9.8 50.0 0.28 0.56 Physical Fitness -153 48.1 21.0 0 93.8 0.12 0.61 4 Week s Variable n M SD Min Max Skew Kurt Total PA 1 -135 42.7 22.2 0.1 129.5 0.45 1.35 Vigorous PA 1 -140 23.7 20.4 0.1 91.7 0.47 0.09 Moderate PA 1 -138 16.7 14.5 0.1 71.0 1.25 2.86 Walking Activity 1 -137 24.5 16.9 0.1 64.5 0.85 0.31 Fitness Center Attendance -159 1.4 1.5 0 7 1.2 2 1. 57 Meaning in Life .94 139 26.6 6.2 5 35 0.34 0.63 Purpose in Life .85 138 24.4 4.0 12 30 1.09 1.17 Basic Psychological Needs Satisfaction .90 138 33.8 8.91 17 53 0.03 0.82 Autonomy .80 143 13.0 3.4 4 20 0.07 0.48 Competence .87 140 11.6 3.8 4 20 0.08 0.56 Relatedness .86 142 9.2 3.3 3 15 0.25 0.75 Autonomous Regulation .94 129 9.9 3.3 1.67 16 0.02 0.71 Controlled Regulation .78 135 4.1 2.4 0 11 0.31 0.46 Psychological Well being 2 .93 138 51.6 12.6 18 77 0.39 0.58

PAGE 58

44 Table 3 cont. Variable n M SD Min Max Skew Kurt Subjective Vitality .91 136 27.3 7.1 9 42 0.12 0.63 Life Satisfaction .90 138 24.3 6.4 9 35 0.62 0.30 Depressive Symptoms .85 138 4.0 3.9 0 18 1.38 1.79 12 Weeks Variable n M SD Min Max Skew Kurt Total PA 1 -140 42.1 21.4 0.1 116.0 0.52 0.60 Vigorous PA 1 -141 22.8 19.9 0.1 100.4 0.70 0.96 Moderate PA 1 -141 15.9 14.2 0.1 64.8 0.98 1.22 Walking Activity 1 -140 24.2 16.4 0.1 64.5 0.62 0.01 Fitness Center Attendance -159 1.1 1.5 0 7 1.44 1.69 Physical Fitness -127 52.7 20.2 12.5 100 0.07 0.55 Meaning in Life .89 141 27.4 5.3 9 35 0.94 0.96 Purpose in Life .87 141 25.1 4.1 12 30 1.21 1.42 Basic Psychological Needs Satisfaction .92 137 34.7 9.7 13 55 0.41 0.54 Autonomy .87 139 13.4 3.8 4 20 0.39 0.63 Competence .87 142 12.0 3.8 4 20 0.20 0.03 Relatedness .88 140 9.3 3.4 3 15 0.26 0.89 Autonomous Regulation .92 136 10.4 2.9 0.67 16 0.37 0.53 Controlled Regulation .82 139 4.2 2.6 0 12 0.47 0.29 Psychological Well being 2 .91 139 54.5 11.2 24 74 0.53 0.18 Subjective Vitality .89 141 28.4 6.7 7 42 0.29 0.07 Life Satisfaction .86 139 26.1 5.5 10 35 0.75 0.04 Depressive Symptoms .82 132 3.8 3.5 0 20 1.50 3.30 Body Mass Index -135 29.7 7.2 19.0 64.9 1.55 4.45 Body Fat % -131 33.1 8.5 8.7 49.2 0.29 0.47 Physical Fitness -127 52.7 20.2 12.5 100 0.07 0.55 1 Square root transformed metabolic units. 2 Combined subjective vitality and life satisfaction scales. Note : n = number of complete; M = Mean; SD = Standard Deviation; Min = Minimum; Max = Maximum; Skew = Skewness; Kurt = Kurtosis.

PAGE 59

45 Figure 5. Trajectories of PA O ver T ime. Aim 1. Effect of a Self monitoring of M/P, Mood, and PA Self Monitoring on PA PA A mixed model examining differences between the self monitoring and control groups over the 12 weeks on PA as measured by the I PAQ SF, revealed that the group by time interaction was not significant, F (1, 13 6 ) = 0.0 3 p = 97. Similarly, there were no significant differences in total PA by study group, F (1, 143) = 0.00 p = 99 There was a significant increase in PA in both groups over the first four weeks, F (1, 1 34 ) = 14.55 p = .000 2 On average, the sample increased by 128.1 minutes of PA per week from baseline to 4 weeks. From four weeks to 12 weeks, there was no significant difference in PA F (1, 135) =

PAGE 60

46 0.12, p = .69. Thus, the gains in PA from baseline to four weeks were, on average, maintained by the sample (see Figure 6 and Table 4 ) Figure 6. Mean PA by C ondition from B aseline to 12 weeks. Secondary analyses revealed there were no significant group differences on type of activity, including vigorous activity F (1, 148) = 0.13, p = .72, moderate activity, F (1, 148) = 1.67, p = 20 and walking F (1, 148) = 0.23, p = .63 Similar patterns to total PA were observed for vigorous activity, moderate activity, and walking over time. There were significant increases from baseline to 4 weeks ( p s < .01) and those gains were maintained from 4 weeks to 12 weeks ( p s > .32).

PAGE 61

47 Table 4 Mean s and S tandard D eviations of S tudy V ariables by C ondition Self Monitoring Baseline 4 weeks 12 weeks Total PA 32.4 (19.6) 41.9 (42.3) 42.3 (21.5) Vigorous PA 15.4 (18.4) 23.0 (20.1) 24.1 (22.2) Moderate PA 9.2 (11.3) 16.3 (14.2) 13.9 (12.6) Walking Activity* 20.5 (15.1) 23.0 (15.8) 24.0 (15.8) Fitness Center Attendance 2.5 (1.7) 1. 6 ( 1.6 ) 1.3 (1. 7 ) Physical Fitness 50.2 (20.5) -53.9 (19.8) Meaning in Life 26.2 (5.9) 26.2 (6.2) 26.9 (5.8) Purpose in Life 24.6 (3.9) 24.2 (4.3) 25.1 (4.3) Basic Psychological Needs Satisfaction 31.1 (7.6) 34.5 (9.8) 35.5 (10.1) Autonomous Regulation 9.4 (3.2) 9.8 (3.4) 10.2 (3.2) Subjective Vitality 24.3 (6.8) 27.4 (7.5) 28.4 (6.7) Life Satisfaction 23.6 (6.5) 24.0 (6.4) 25.5 (6.1) Depressive Symptoms 5.1 (3.7) 3.9 (4.0) 3.7 (3.4) Control Baseline 4 weeks 12 weeks Total PA 32.0 (17.7) 43.5 (23.9) 41.9 (21.5) Vigorous PA 15.0 (15.9) 24.4 (20.7) 21.4 (17.2) Moderate PA 11.1 (12.3) 17.0 (14.8) 17.8 (15.5) Walking Activity* 20.1 (13.6) 26.0 (17.6) 24.5 (17.2) Fitness Center Attendance 2. 3 (1.8) 1. 2 (1. 3 ) 0.9 (1.2) Physical Fitness 46.0 (21.5) -51.6 (20.7) Meaning in Life 27.0 (5.8) 26.9 (6.2) 28.0 (4.8) Purpose in Life 24.2 (3.8) 24.5 (3.8) 25.1 (3.8) Basic Psychological Needs Satisfaction 28.5 (8.5) 33.2 (8.1) 33.8 (9.2) Autonomous Regulation 9.0 (3.2) 9.9 (3.2) 10.6 (2.6) Subjective Vitality 23.5 (6.7) 27.2 (6.9) 28.5 (6.4) Life Satisfaction 23.1 (6.1) 24.6 (6.4) 26.6 (4.9) Depressive Symptoms 6.1 (5.1) 4.1 (3.9) 3.9 (3.5) *Square root transformed variable. Note. PA = Physical activity. The group by time interaction was not significant for fitness center attendance, F (2, 156) = 0.79, p = .46. However, there was a significant reduction in fitness center visits from the first week ( M = 2.4, SD = 1.7) to the fourth week ( M = 1.4, SD = 1.5), F (1, 157) = 39.58, p < .0001, and again from the fourth week to the twelfth week ( M = 1.1, SD = 1.5), F (1, 157) = 4.05, p = .046 (see Figure 7). Similarly, there was not a significant group by time

PAGE 62

48 interaction on physical fitness, F (1, 132) = 0.44, p = .51. There was a significant increase in physical fitness over the 12 weeks, F (1, 134) = 7.20, p = .008. On average, participants increased their physical fitness wellness scores by 4.6%. Figure 7. Mean Fitness C enter V isits by C ondition from B aseline to 12 weeks. M/P. Meaning in life did not differ by group at any of the three time points F (1, 146) = 2.08, p = .15, or over time, F (2, 137) = 1.52, p = .22. Similarly, there were no differences between the groups in purpose in life, F (1, 152) = 0.0 3, p = .87. However, there was a significant increase in purpose in life from baseline ( M = 24.4, SD = 3.8) to 12 weeks ( M = 25.1, SD = 4.1) for both groups F (2, 139) = 3.38, p = .04 (see Figure 8 )

PAGE 63

49 Figure 8. Mean Purpose in L ife by C ondition from B aseline to 12 weeks. SDT m ediators. There was no difference between groups in basic psychological needs satisfaction over time, F ( 2, 142) = 0.20, p = .82. However, basic psychological needs satisfaction significantly increased in both groups over the fi rst four weeks, F (1, 145) = 14.44, p = .0002. There was no significant difference in basic psychological needs satisfaction from 4 weeks to 12 weeks, F (1, 132) = 0.58, p = 0.57. Thus, gains made in the first four weeks were, on average, maintained from 4 weeks to 12 weeks (see Figure 9 ). A similar pattern was observed with changes in autonomous regulation. There were no significant differences between groups, F (1, 159) = 0.01, p = .91; however, autonomous

PAGE 64

50 regulation significantly increased from baseline to 4 weeks, F (1, 142) = 16.49, p < .0001 and marginally increased from 4 weeks to 12 weeks, F (1, 131) = 3.76, p = .05 (see Figure 10 ). Figure 9. Mean Basic P sychological Needs S atisfaction by C ondition from B aseline to 12 weeks.

PAGE 65

51 Figure 10. Mean Autonomous R egulation by C ondition from B aseline to 12 weeks. Psychological w ell b eing. There were no significant differences between groups in vitality, F (1, 154) = 0.08, p =.77 life satisfaction, F (1, 152) = 0.18, p = .67, or depressive symptoms, F (1, 155) = 0.31, p = .58 over the 12 weeks However, vitality significantly increased over the first four weeks, F (1, 137) = 28.78, p < .0001, and marginally increased from 4 weeks to 12 weeks, F (1, 1 36) = 3.33, p = .07 (see Figure 11 ). Life satisfaction marginally increased from baseline to 4 weeks, F (1, 140) = 3.68, p = .06, and significantly increased from 4 weeks to 12 weeks, F (1, 136) = 15.0, p = .0002 (see Figure 12 ). Depressive symptoms signif icantly decreased from baseline to 4 weeks, F (1, 147) = 16.23, p

PAGE 66

52 < .0001. There was no difference in depressive symptoms from 4 weeks to 12 weeks, F (1, 136) = 0.52, p = .47 (see Figure 13 ). Figure 11 Mean V itality by C ondition from B aseline to 12 weeks.

PAGE 67

53 Figure 12. Mean L ife S atisfaction by C ondition from B aseline to 12 weeks.

PAGE 68

54 Figure 13. Mean D epressi ve Symptoms by C ondition from B aseline to 12 weeks Aim 2. Relationship of Daily M/P Salience to Daily PA For Aim 2 analyses included the 80 participants who were randomized to record daily meaning salience mood, and PA for 4 weeks (28 days). Descriptive statistics for variables in Aim 2 are presented in Table 5 and correlations among the variables are presented in Table 6

PAGE 69

55 Table 5 Descr iptive S tatistics of Variables in Aim 2 Variable n M (SD) Min Max Thoughts of Meaning Scale 176 2 45.2 ( 12.5 ) 10 70 .9 4 Positive Affect 1752 24.9 (6.7) 8 40 .91 Negative Affect 1764 8.5 (3.6) 5 23 .80 PA Minutes 1795 41.2 (51.6) 0 480 -PA Intensity 12 80 12.1 (2.6) 6 19 -Table 6 Pearson Correlatio ns a mong the Variables in Aim 2 p < .001 In a mixed model with days nested within participants and controlling for daily mood, meaning salience was significantly and positively associated with daily minutes of PA ( # = .2 1 p < .0001). Participants who reported greater daily meaning salience also reported more minutes of activity on the same days. Positive affect was not significantly associated with daily PA when controlling for both meaning salience and negative affect ( # = 0 1 p = .94). Controlling for positive affect and meaning salience n egative affect was negatively associated with PA ( # = .1 4 p < .0001); on days when participants experienced more negative affect, they also engaged in fewer minutes of PA The mixed mod el was repeated with PA intensity as the dependent variable. On days that participants reported engaging in 1 2 3 4 1. Thoughts of Meaning Scale -2. Positive Affect .66* -3. Negative Affect .15* .36* -4. PA Minutes .13* .11* .14* -5. PA Intensity .05 .03 .10* .33*

PAGE 70

56 PA and controlling for positive and negative affect meaning salience was associated with increased intensity of PA ( # = 21 p < .000 1 ). Thus, greater salience of meaning was associated with more intense levels of PA In a multilevel logistic regression model predicting fitness center attendance and controlling for positive and negative mood daily meaning salience was related to signif icantly greater likelihood of visiting the fitness center, Odds Ratio (OR) (1608) = 1.48 (95% CI = 1.18, 1.86), p = .0008. For every standard deviation increase in daily meaning salience participants were 48% more likely to visit the fitness center on tha t day. Neither positive affect, OR = 1.004 (95% CI = 0.81, 1.24), p = .97, or negative affect, OR = 0.86 (95% CI = 0.72, 1.02), p = .08, were significantly related to same day fitness center visits. Aim 3 Examining the SDT P rocess M odel of PA A doption w ith and without M/P. Prior to examining the path analytic models, bivariate Pearson correlations were used to examine the associations among Aim 3 variables. Correlations are presented in Table 7 Model 1a: Cross sectional b aseline m odel. At baseline, the SDT Process Model of Behavior Change had acceptable fit (see Table 8 for model fit statistics) The model and standardized path coefficients are presented in Figure 14 Basic psychological needs satisfaction was significantly and positively related to aut onomous regulation for PA which was significantly and positively related to PA PA was not significantly related to psychological well being, but psychological needs satisfaction was significantly and positively related to PA The base SDT Process Model a ccounted for 16% of PA and 18% of psychological well being at baseline.

PAGE 71

57 Table 7 Pearson Correlations a mong Aim 3 Model Variables Note. BPNES = Basic Psychological Needs in Exercise Scale; AR = Autonomous Regulation; MIL = Meaning in Life; PA = PA ; PWB = Psychological Well Being Significant ( p < .05) correlations are in bold 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Baseline 1. BPNES -2. AR .51 -3. MIL .17 .22 -4. PA .27 .33 .02 -5. PWB .31 .33 .63 .14 -4 week 6. BPNES .23 .13 .18 .12 .07 -7. AR .28 .74 .15 .22 .13 41 -8. MIL .03 .19 .73 .12 .54 .23 .28 -9. PA .03 .03 .1 9 .27 02 .23 .12 .07 -10. PWB .15 .28 60 .17 .73 .30 .36 .73 .12 -12 week 11. BPNES .33 .25 .03 .22 .03 .63 .36 .03 .17 21 -12. AR .36 .70 .06 .34 .09 .39 .81 .06 .18 .24 .55 -13. MIL .02 .13 .57 .15 .45 .20 .16 69 20 .63 .17 .10 -14. PA .06 .09 .00 26 .02 .20 .15 .11 .42 .17 .30 .25 .06 -15. PWB .13 .16 43 .21 .62 .28 .22 54 .1 1 .77 .40 .26 .64 .19

PAGE 72

58 Table 8 Model Fit S tatistics Model # 2 df p RMSEA (90% CI) CFI SRMR AIC BIC Cross sectional Models 1a. Baseline 16.44 9 .06 .07 (.00 .13) .92 .03 3242.25 3315.14 1b. Mod. Baseline 20.98 16 .18 .05 (.00 .09) .97 .04 4153.63 4244.74 2a. 4 week 19.13 9 .02 .09 (.03 .15) .79 .04 2814.33 2884.05 2b. Mod. 4 week 25.20 16 .07 .07 (.00 .11) .93 .04 3591.68 3678.83 3a. 12 week 19.32 9 .02 .09 (.03 .15) .89 .04 2835.89 2905.97 3b. Mod. 12 week 31.31 16 .01 .08 (.04 .13) .91 .05 3613.32 3700.92 Longitudinal Models 4a. Absolute PA 9.56 8 .30 .04 (.00 .11) .94 .03 1848.56 1886.80 4b. Mod. absolute PA 15.66 16 .48 .00 (.00 .07) 1.00 .04 2825.68 2877.31 5a. Change in PA 30.24 21 .09 .05 (.00 .09) .96 .04 4958.78 5077.22 5b. Mod. change in PA 37.13 30 .17 .04 (.00 .08) .97 .04 5932.20 6068.87 Note. # 2 = chi square test of model fit; df = degrees of freedom; p = p value; RMSEA = root mean square error of approximation; CFI = comparative fit index; SRMR = standardized root mean square residual; AIC = Akaike's Information Criterion; BIC = Bayesian Information Criterion; Mod. = Modified; PA = Physical Activity *** p < .001; ** p < .01; p < .05 Figure 14. Cross sectional SDT Process Model of Behavior Change at Baseline. Model 1b: Modified c ross sectional b aseline m odel. Meaning was added to the model, and the model fit the data very well (see Table 8) The modified model is presented in Figure 15 The associations found in the base SDT Process Model were maintained. Basic Bas ic Psychological Needs Satisfaction Autonomous Regulation R 2 = .26 Physical Activity R 2 = .16 Psychological Well being R 2 = .18 .51*** .33** .08 .28**

PAGE 73

59 psychological needs satisfaction was positively related to experience of meaning in life. Meaning was not significantly related to PA at baseline, but it was significantly and po sitively associated with psychological well being. There was no change in R 2 from the base model in PA but the variance in psychological well being accounted for by the model significantly increased to 46%. Thus, meaning improved the prediction of psychol ogical well being but not of PA at baseline. *** p < .001; ** p < .01; p < .05 Figure 15. Cross sectional M odified SDT Process Model of Behavior Change at Baseline. Model 2a: Cross sectional 4 w eek m odel. The base SDT Process Model of Behavior Change at week four did not fit the data well (see Table 8) The path model is presented in Figure 16 Similar to the model at baseline, basic psychological needs satisfaction was positively and significantly related to both autonomous regulation and to psychological well being. However, autonomous regulation was not significantly related to Basic Psychological Needs Satisfaction Autonomous Regulation R 2 = .26 Physical Activity R 2 = .16 Psychological Well being R 2 = .46 .51*** .34** .11 .28** .17* .13 .03 .58*** Meaning in Life R 2 = .03

PAGE 74

60 PA at 4 weeks and PA was not significantly related to psychological well being. One possible reason that the model did not fit the data well is that basic psychological needs satisfaction was significantly correlated with PA at 4 weeks (see Table 7), but this associat ion was not fully accounted for by the mediation through autonomous regulation. The model accounted for 12% of the variance in PA (with no significant predictors of PA ) and 13% of the variance in psychological well being. *** p < .001; ** p < .01; p < .05 Figure 16. Cross sectional SDT Process Model of Behavior Change at 4 weeks Model 2b: Modified c ross sectional 4 week m odel. The modified model is presented in Figure 17 and the model fit the data well. Similar to baseline, basic psychological needs satisfaction was positively associated with meaning, but meaning was not signific antly related to PA Thus, the model did not significantly account for more variance in PA Meaning was strongly and positively associated with psychological well being, and the variance in psychological well being accounted for by the model increased subs tantially to 58%. Basic Psychological Needs Sat isfaction Autonomous Regulation R 2 = .16 Physical Activity R 2 = .12 Psychological Well being R 2 = .13 .40** .13 .11 .27**

PAGE 75

61 *** p < .001; ** p < .01; p < .05 Figure 17. Cross sectional Modified SDT Process Model of Behavior Change at 4 weeks Model 3a: Cross sectional 12 week m odel. The base SDT Process Model of Behavior Change at 12 weeks was a poor fit to the data. The path model i s presented in Figure 18 Basic psychological needs satisfaction was significantly and positively related to autonomous regulation, which was significantly and positively related to PA PA was not significantly related to psychological well being. Similar to the cross sectional 4 week model, one possible reason that the model did not fit the data well is that basic psychological needs satisfaction was significantly correlated with PA at 12 weeks (see Table 7), but again this association was not fully accoun ted for by the mediation through autonomous regulation. The model accounted for 31% of the variance in autonomous regulation, 18% of the variance in PA and 20% of the variance in psychological well being. .14* Basic Psychological Needs Satisfaction Autonomous Regulation R 2 = .26 Physical Activity R 2 = .12 Psychological Well being R 2 = .58 .40** .12 .02 Meaning in Life R 2 = .06 .24** .19* .06 .72***

PAGE 76

62 *** p < .001; ** p < .01; p < .05 Figure 18. Cross sectional SDT Process Model of Behavior Change at 12 weeks Model 3b: Modified c ross sectional 12 week m odel. Meaning in life was added to a modified 12 week model (see Figure 19 ), and the model fit was acceptable. Basic psychological needs satisfaction was significantly and positively associated with presence of meaning, and presence of meaning was significantly and positively associated with psychological well being. However, me aning was not significantly related to PA The modified model accounted for 31% of the variance in autonomous regulation, 21% of the variance in PA and 54% of the variance in psychological well being. Thus, the modified model increased the proportion of v ariance in PA accoun ted for by the predictors by 3%; however, this change is not significant because the association between meaning in life and PA was not significant. Basic Psychological Needs Satisfaction Autonomous Regulation R 2 = .31 Physical Activity R 2 = .18 Psychological Well being R 2 = .20 .55*** .30** .10 .37***

PAGE 77

63 *** p < .001; ** p < .01; p < .05 Figure 19. Cross sectional Modified SDT Process Model of Behavior Change at 12 weeks Model 4a: Longitudinal m odel p redicting a bsolute l evels of PA at 12 weeks Because PA was not significantly associated with psychological well being in the cross sectional models, psychological well being was not included in the longit udinal model. The base SDT Process Model was an excellent fit to the data (see Table 8 and Figure 20 ) Basic psychological needs satisfaction at baseline was significantly and positively related to autonomous regulation at 4 weeks Those who reported great er autonomous regulation at 4 weeks also reported significantly more PA at 12 weeks. The model accounted for 18% of the variance in PA at 12 weeks. .30*** Autonomous Regulation R 2 = .26 Physical Activity R 2 = .21 Psychological Well being R 2 = .58 .55*** .28*** .03 Basic Psychological Needs Satisfaction Meaning in Life R 2 = .06 .17* .04 .13 .61***

PAGE 78

64 *** p < .001; ** p < .01 Figure 20. Longitudinal SDT Process Model of Behavior Change Predicting Absolute PA at 12 weeks. Model 4b : Modified l ongitudinal m odel p redicting a bsolute l evels of PA a t 12 weeks. The modified longitudinal SDT Process Model is presented in Figure 21 and was an excellent fit to the data. The relationships found in the base SDT longitudinal model were maintained. Meaning in life at baseline was significantly related to basic psychological needs satisfaction at baseline However, baseline meaning in life was not significantly related to PA at 12 weeks Adding meaning in life to the model did not significantly increase the amount of variance in 12 week PA accounted for the model. Model 5a : Longitudinal m odel p redicting c hange in PA at 4 weeks and 12 weeks. A longitudinal path model was used to assess the SDT Process Model of Behavior Change and predict residualized change in PA from baseline to 4 weeks and from 4 weeks to 1 2 weeks (see Figure 22). The model was a good fit to the data (Table 8). Basic psychological needs satisfaction at baseline was significantly and positively related to autonomous regulation at baseline, which was significantly and positively related to PA at baseline. Basic psychological needs satisfaction was significantly and negatively related to the change in autonomous regulation from baseline to 4 weeks. Thus, those with lower basic psychological needs satisfaction at baseline experienced greater cha nge in autonomous regulation over the first 4 weeks. Autonomous regulation at 4 weeks was not significantly Basic Psychological Needs Satisfaction (Baseline) Autonomous Regulation (Week 4) R 2 = .08 Physical Activity (Week 12) R 2 = .18 .28*** .25***

PAGE 79

65 related to change in PA from baseline to 4 weeks. However, autonomous regulation at 4 weeks was positively and significantly related to change in PA from 4 weeks to 12 weeks. Those with higher autonomous regulation in the first 4 weeks also experienced a greater increase in PA from 4 weeks to 12 weeks. *** p < .001; ** p < .01; p < .05 Figure 21. Modified Longitudinal SDT Process Model of Behavior Change Predicting Absolute PA at 12 weeks. Meaning in Life (Baseline) R 2 = .03 .17* .02 Basic Psychological Needs Satisfaction (Baseline) Autonomous Regulation (Week 4) R 2 = .08 Physical Activity (Week 12) R 2 = .18 .28*** .24***

PAGE 80

66 *** p < .001; ** p < .01 Figure 22. Longitudinal SDT Process Model of Behavior Change Predicting Change in PA at 4 weeks and 12 weeks. Model 5b: Modified l ongitudinal m odel predicting c hange in PA at 4 weeks and 12 weeks. The model was modified to include meaning in life at baseline as a predictor of change in PA at 4 weeks and 12 weeks (see Figure 23) The modified model was an excellent fit to the data. The relationships modeled in the base SDT model remained the same in the modified model. Basic psychological needs satisfaction at baseline was significantly and positively related to presence of meani ng in life at baseline. Baseline meaning in life was significantly and positively related to the change in PA from baseline to 4 weeks. Those who Autonomous Regulation (Week 4) R 2 = .55 Physical Activity (Week 4) R 2 = .29 Physical Activity (Week 12) R 2 = .32 Basic Psychological Needs Satisfaction (Baseline) Autonomous Regulation (Baseline) R 2 = .26 Physical Activity (Baseline) R 2 = .15 .51*** .32*** .17* .81*** .36*** .16* .12 .39***

PAGE 81

67 had a greater sense of meaning reported a greater change in PA behavior from baseline to 4 weeks. Adding meani ng in life increased the amount of variance in change in PA accounted for by the model from 29% to 32%. However, baseline meaning in life was not significantly related to change in PA from 4 weeks to 12 weeks. Exploratory Outcomes Average and v ariability of m eaning s alience in r elation to PA and f itness o utcomes. Pearson correlations were used to examine the relationship of the average and standard deviation of the daily meaning salience score in relation to PA physical fitness, body composition and size and psychological well being outcomes. In addition, the relationship between global meaning and purpose and the meaning salience was also assessed. Pearson correlations are presented in Table 9. Those with greater average meaning salience reported higher global meaning in life and purpose in life, greater subjective vitality, better life satisfaction, and fewer depressive symptoms. They also reported more minutes of moderate PA at 4 weeks. Conversely, they reported fewer minutes of vigorous PA at 4 weeks.

PAGE 82

68 *** p < .001; ** p < .01 Figure 23. Modified L ongitudinal SDT Process Model of Behavior Change Predicting Change in PA at 4 weeks and 12 weeks. Those with greater variability in meaning salience during the first 28 days had fewer visits to the fitness center during the first week and during the twelf th week They also reported fewer minutes of vigorous activity at 12 weeks and more depressive symptoms at baseline, 4 weeks, and 12 weeks. There were no significant relationships between meaning salience and objective outcomes, including BMI, body fat per cent, and physical fitness. Meaning in Life (Baseline) R 2 = .03 .17* .20* .08 .12 Autonomous Regulation (Week 4) R 2 = .55 Physical Activity (Week 4) R 2 = .29 Physical Activity (Week 12) R 2 = .32 Basic Psychological Needs Satisfaction (Baseline) Autonomous Regulation (Baseline) R 2 = .26 Physical Activity (Baseline) R 2 = .15 .51*** .32*** .17* .81*** .36*** .16* .12 .39***

PAGE 83

69 Table 9 Pearson Correlations b etween the TOMS and O ther S tudy O utcomes Thoughts of Meaning Scale (TOMS) Baseline Week 4 Week 12 Measure M SD M SD M SD Number of Fitness Center Visits .14 .26 .01 .18 .0 1 33 Total PA .00 .07 .04 .04 .06 .13 Vigorous PA .15 .08 .26 .14 .04 .23 Moderate PA .14 .03 .25 .14 .02 .11 Walking Activity .11 .04 .21 .20 .22 .13 Body Mass Index .02 .02 --.01 .03 Body Fat Percent .13 .03 --.11 .05 Physical Fitness .01 .03 --.00 .05 Meaning in Life .43 .10 .61 .10 .53 .04 Purpose in Life .54 .00 .58 .09 .51 .18 Subjective Vitality .44 .11 .54 .09 .3 8 .08 Satisfaction with Life .44 .04 .53 .09 .47 .03 Depressive Symptoms .19 .25 .25 .36 .35 .30 Note M = Average TOMS over 28 days; SD = Standard Deviation of TOMS over 28 days ; PA = Physical Activity Significant correlations are in bold Four multiple regression models were used to examine the relationships between meaning salience and 12 week o utcomes, including fitness center visits, total PA, and M/P. Results of these models are presented in Table 10. After controlling for demographics, the variability of meaning salience in the first 28 days w as a significant predictor of visits to the fitnes s center at 12 weeks Those who had, on average, lower variability of meaning salience had more visits to the fitness center during the 12 th week Conversely, meaning salience was not a significant predictor of total PA at 12 weeks. After controlling for relevant demographics, average meaning salience in the first 28 days was a significant predictor of both global ratings of meaning in life (MILQ) and purpose in life (LET) at 12 weeks. Interestingly, those who reported an annual income of $40,000 or greate r reported significantly less meaning in life than those who reported a lower income.

PAGE 84

70 Table 10 Regression M odels with M eaning Salience P redicting 12 week Outcomes Dependent Variable ( # ) Predictor Fitness Center Visits Total PA MILQ LET Female .10 .04 .02 .06 Age .1 1 .10 .06 .03 White 07 .09 .07 .04 Work full time 33 .10 .15 .01 Income ( $40,000) 1 9 .17 .42 .16 Married .17 .15 .19 .04 Meaning Salience Average .03 .03 .53 .47 Meaning Salience Variability 40 .12 .10 .06 F 2. 51 0.67 5.22 3.24 p .0 19 .716 <.0001 .0038 R 2 .2 2 .08 .40 .29 Note MILQ = Meaning in Life Questionnaire; LET = Life Engagement Test; PA = Physical Activity. Significant correlations are in bold Relationships between global M/P, PA and o bjective o utcomes. Pearson correlations were used to examine the relationships between global meaning and pur pose and objective outcomes (see Table 10). There was some evidence that greater meaning and purpose was related to greater PA prospectiv ely. As previously noted, baseline meaning in life was significantly related to total PA at four weeks, and this was primarily in more minutes of moderate activity. Meaning in life at 4 weeks was also significantly and positively related to fitness center visits during the fourth week, and meaning in life at 12 weeks was positively related to moderate PA at 12 weeks. Those who reported a greater sense of purpose in life at 4 weeks also reported more minutes of walking activity at 12 weeks.

PAGE 85

71 Table 11 Pear son Correlations between G lobal M/P PA and O bjective O utcomes Baseline Week 4 Week 12 Measure MILQ LET MILQ LET MILQ LET Baseline Number of Fitness Center Visits .04 .06 .00 .08 .02 .07 Total PA .02 .08 .12 .14 .15 .22 Vigorous PA .01 .03 .09 .08 .07 .15 Moderate PA .02 .07 .13 .11 .08 .18 Walking Activity .10 .08 .13 .17 .21 .15 Body Mass Index .21 .26 .1 7 .13 .18 .09 Body Fat Percent .11 .08 .05 .02 .11 .01 Physical Fitness .10 .10 .05 .00 .01 .09 Week 4 Number of Fitness Center Visits .15 .09 .18 .11 .12 .13 Total PA .19 .07 .07 .12 .20 .11 Vigorous PA .15 .05 .03 .02 .08 .04 Moderate PA .17 .06 .15 .10 .15 .08 Walking Activity .08 .03 .01 .14 .21 .12 Week 12 Number of Fitness Center Visits .07 .00 .01 .08 .04 .10 Total PA .00 .01 .11 .09 .06 .15 Vigorous PA .11 .01 .08 .07 .02 .12 Moderate PA .03 .07 .14 .06 .17 .07 Walking Activity .03 .02 .15 .18 .04 .11 Body Mass Index .30 .32 .21 .17 .21 .09 Body Fat Percent .21 .16 .12 .05 .16 .00 Physical Fitness .12 .14 .09 .14 .13 .06 Note M ILQ = Meaning in Life Questionnaire ; LET = Life Engagement Test ; PA = Physical Activity Significant correlations are in bold There was also evidence that engaging in PA may lead to increased meaning and purpose in life. Those who reported more total PA and moderate activity at baseline reported greater purpose in life at 12 weeks, and those who reported more walking at baseline reported greater purpose in life at 4 weeks and greater meaning in life at 12 weeks. Walking at 4 weeks was also significantly and positively related to meaning in life at 12 weeks.

PAGE 86

72 Finally, there was evidence that those with greater BMI reported less meaning and purpose in life at all three time points Meaning in life at baseline was also significantly and negatively related to body fat percent at 12 weeks. Those with a greater sense of meaning in life when they started their PA programs had a lower body fat percentage 12 weeks later. Relationshi ps between M/P and PA c onnection, M/P, PA and o bjective o utcomes. The exploratory variable examining the strength of participants' experienced connections between M/P and PA was assessed in relation to other study variables. Correlations between these me asures are presented in Table 1 2 There were small positive associations between stronger M/P and PA connection and global evaluations of meaning and purpose cross sectionally but not prospectively. Similarly, M/P and PA connection was positively associate d with several PA variables cross sectionally, including total PA vigorous PA and moderate PA Baseline PA was prospectively and positively associated with a greater M/P connection. The more activity individuals were doing at the beginning of the study, the stronger they rated the connection between M/P and PA Baseline ratings of M/P and PA connection were ne gatively and significantly associated with BMI and body fat percent at baseline and at the end of the study.

PAGE 87

73 Table 1 2 Pearson Correlations between M/P and PA C onnection, M/P, PA and O bjective O utcomes Measure Baseline Week 4 Week 12 Meaning Salience Average .24 .21 .18 Meaning Salience Variability .04 .09 .06 Baseline Meaning in Life .16 .11 .02 Purpose in Life .24 .15 .08 Number of Fitness Center Visits .14 .14 .08 Total PA .22 .22 .23 Vigorous PA .19 .14 .06 Moderate PA .17 .18 .15 Walking Activity .13 .09 .22 Body Mass Index .14 .11 .02 Body Fat Percent .17 .10 .01 Physical Fitness .03 .00 .03 Week 4 Meaning in Life .18 .26 .04 Purpose in Life .11 .15 .11 Number of Fitness Center Visits .10 .08 .16 Total PA .00 .06 .09 Vigorous PA .05 .07 .05 Moderate PA .08 .10 .12 Walking Activity .01 .00 .08 Week 12 Meaning in Life .11 .10 .11 Purpose in Life .15 .09 .18 Number of Fitness Center Visits .0 1 .10 .26 Total PA .00 .17 .28 Vigorous PA .12 .13 .33 Moderate PA .04 .11 .17 Walking Activity .10 .07 .11 Body Mass Index .19 .10 .04 Body Fat Percent .22 .08 .04 Physical Fitness .02 .02 .04 Note Significant correlations are in bold

PAGE 88

74 CHAPTER IV DISCUSSION The purpose of the study was to examine the associations between M/P and PA in the context of SDT, in previously sedentary adults beginning new exercise programs. The first aim was to test whether self monitoring of M/P, mood, and PA daily for the first four weeks increased exercise more than a random survey control. Results sug gest that the self monitoring condition had no effect on PA or secondary outcomes of M/P, SDT mediators, psychological well being, body size/composition, or physical fitness. The second aim examined whether daily salience of M/P was related to daily PA in the first four weeks of starting an exercise program. Results revealed that controlling for daily mood, on days when participants reported greater salience of meaning, they also reported more minutes of PA were more likely to visit the fitness center, and reported greater intensity of PA when they did exercise. The final aim of the study was to examine whether adding M/P to the SDT Process Model of Behavior Change predicted PA above and beyond the base SDT model. Results of this aim were mixed, with M/P at baseline predicting greater change in PA from baseline to 4 weeks. However, M/P was not related to PA cross sectionally at any of the three time points or to PA at 12 weeks. Thus, it appears that M/P salience is important during the behavior change proces s, but further research needs to examine the role of global evaluations of M/P during PA adoption and maintenance. Aim 1. Effect of M/P, Mood, and PA Self Monitoring on PA A major question of this study was whether just asking people to report on their d aily M/P and also their daily mood and PA would be an intervention in itself. Previous PA interventions have shown that self monitoring is an effective tool for PA behavior change

PAGE 89

75 (Olander et al., 2013). Consistent with the hypothesis, r esults suggest that just asking people to report on M/P salience, mood, and PA did not improve PA adoption or maintenance in the first 12 weeks of making a behavior change. The daily surveys, were in fact, framed as such, and were not labeled an "intervention" to participant s. Perhaps giving participants the rationale for self monitoring would increase the likelihood that the self monitoring would have had an effect on PA or other outcomes (e.g., calling the daily surveys an intervention and noting that self monitoring of behavior is one strategy to improve behavior change) Another possible reason the daily surveys may not have impacted PA is that participants reported at the end of the da y. Self monitoring may work best when it is an ongoing process that occurs throughout the day because it can remind individuals to engage in the behavior that they are monitoring. For instance, participants could go about their day, while not actively thi nk ing about engaging in PA or what is meaningful to them, and then complete the self monitoring task in the evening. By the time they completed the self monitoring task, it may have been too late to engage in PA for the day and any potential impact of the monitoring on PA dissipated overnight On the other hand, it may take a stronger intervention to increase the pairing of M/P salience to PA in order to increase PA Indeed, it was hypothesized that those who connect their personal sense of M/P to the rea sons they want to make a PA behavior change, keep those reasons salient or in the forefront of their minds as they go about their days, and then make decisions based on that salience would be most successful. It is likely that just asking people to report on M/P salience and PA at the end of the day did not make a strong enough connection for them. Instead, asking people to actively connect what is meaningful to them or consistent with their sense of purpose in life to the reasons why they should make the

PAGE 90

76 b ehavior change would likely be a more effective intervention. Once participants make the connection, then providing timely reminders to engage in PA and think about M/P could increase their activity. For example, individuals could engage in a task to incre ase their connection between their what gives their lives meaning and their goals to engage in healthy behaviors. Individuals could first identify what areas give their lives the most meaning (e.g., work, family, relationships, spirituality, etc.). After t hey identify these areas, they would be asked to make connections to how being healthier (e.g., by engaging in PA) supports their values and goals in the previously identified areas. For instance, one person might state that engaging in PA gives them more energy at work, which is a key area that gives their lives meaning. After identifying several points of connection, an interventionist can design tailored messages to be delivered at key choice points for PA (e.g., a person plans to walk at lunch time and a reminder would be delivered a few minutes prior to lunch) that remind the person of their previously made connection between meaning and PA. This message could bring the salience of meaning to the forefront of their minds, and thus, the choice to engage in PA may be more likely. Not surprisingly, one of the biggest implications of these findings is that PA and M/P are not easily changed by measurement alone. Thus, th e purpose of this aim was more of a methodological question than it was an intervention e fficacy question. As researchers, we are highly concerned about measurement reactivity, and this is especially true when PA is the primary outcome. The results of this study suggest that repeated daily measurement of M/P, mood, and PA does not directly inf luence several outcomes, including PA M/P, psychological well being, and SDT mediators.

PAGE 91

77 Aim 2. Relationship of Daily M/P Salience to Daily PA As hypothesized, on days when participants thought more about what is meaningful or valuable to them, they also engaged in more minutes of PA and were more likely to visit the fitness center. This relationship existed even after controlling for positive an d negative mood, suggesting that meaning salience is an important predictor of PA Previous research has highlighted the importance of mood for predicting PA (Carels et al., 2004), but this is the first study to show that after controlling for mood, meanin g salience is a predictor of PA Somewhat surprisingly, meaning salience was also associated with greater PA intensity on days on which participants reported engaging in PA It may be that individuals who frequently think about what makes their liv e s meani ngful and make decisions based on their personal sense of M/P also more consciously engaged in exercise versus PA Exercise is a generally more intentional and intense behavior than PA and thus individuals who make more intentional decisions to engage in exercise may be making those decisions because they are consistent with their personal M/P. M eaning salience seems to be a more robust predictor of PA than global ratings of M/P (see below for further discussion). Several studies have shown that global ratings of M/P are positively associated with PA ( Holahan et al. 2008; Holahan & Suzuki, 2006; Homan & Boyatzis, 2010; Hooker & Masters 2014 ; Ruuskanen & Ruoppila, 1995; Takkinen et al. 2001 ). M/P seems important for PA but only to the extent to which M/P is salient for an individual on a daily basis. Global rating s of M/P are quite stable (e.g., Steger et al., 2006), and most of our empirical base of the relations between M/P and PA have been from stati c glimpses of individuals' reported M/P and PA behavior Spring and colleagues (Spring, Gotsis, Paiva, & Spruijt Metz, 2013) note that the majority of health behavior change

PAGE 92

78 theories involve "causal constructs [that] are intrapsychic, conscious, somewhat v aguely specified processes whose quantitative relationship to health behavioral change is only imprecisely specified" (p. 35). Global ratings of M/P fall in this category; it is difficult to understand how this global sense of M/P would be related to engag ing in PA on a day to day basis especially if individuals do not think about this other than when they are asked about it on a survey In contrast, meaning salience is a dynamic process and could be used as motivation to engage in healthy behavior. Indeed Frankl states this nicely, For the meaning of life differs from man to man, from day to day and from hour to hour. What matters, therefore, is not the meaning of life in general but rather the specific meaning of a pe rson's life at a given moment" (Fran kl, 1985, p. 1 08 ). The conscious process of thinking about what is meaningful to an individual when faced with everyday decisions, such as the decision about whether or not to engage in PA that day, could be uniquely applied to just in time behavioral inte rvention s to increase engagement in PA. Indeed, the advent of mobile technology allows researchers to design investigations that can tap into these dynamic processes to (1) increase knowledge of proximal predictors of behavior in context and (2) intervene at apropos times to improve health behavior adoption and maintenance (Spruijt Metz, et al., 2015). A better understanding of these proximal predictors of behavior could change health behavior theory, and subsequently health behavior interventions, dramatic ally. Meaning salience is one such predictor. Aim 3. Examining the SDT P rocess M odel of PA A doption with and without M/P. For the third aim, the models largely supported the SDT Process Model of Behavior Change. Individuals that reported greater psychological needs satisfaction also reported greater autonomous regulation, PA and psychological well being. Those who reported more

PAGE 93

79 a utonomous regulation also engaged in more PA at baseline and at 12 weeks In contrast to SDT the ory, autonomous regulation was not associated with PA at 4 weeks. Longitudinal models also supported the SDT Process Model, suggesting that the SDT significantly explained variance in absolute PA at 12 weeks and the change in PA from baseline to 4 weeks an d from 4 weeks to 12 weeks. Of note, the 4 week and 12 week longitudinal models did not adequately represent the data One postulate of the SDT Process Model of Behavior Change is that basic psychological needs satisfaction is associated with greater auton omous regulation, and in turn, autonomous regulation is positively associated with PA. Moreover, the relationship between basic psychological needs satisfaction and PA is fully mediated by autonomous regulation However, th e full mediation was not observed at 4 weeks and 12 weeks after starting a PA program. This could be because changes in basic psychological needs satisfactio n early in a new PA program do not quickly translate into changes in autonomous motivation. For example, one could imagine that a ne w previously sedentary exercise initiate may experience rapid increases in competence. However, just because individuals are more competent in engaging in PA (e.g., knowing what exercise s to do or how to do the exercises), does not mean they quic kly becom e more integrated within their values or become more enjoyable. SDT principles do seem to account for a significant proportion of the variance in PA behavior during PA adoption; however, it does seem that adjustments to the SDT Process Model may be needed to more fully understand PA adoption in prev iously sedentary exercise initiates (e.g., accounting for meaning salience or other health behavior theory constructs ). These findings significantly advance the SDT literature in the area of PA adoption and maintenance. Of the longitudinal studies of these processes, only two have had more than

PAGE 94

80 two time points (Edmunds et al., 2007; Gunnell et al., 2016), only two have examined the processes in previously sedentary individuals (Edmunds et al., 2 007; Rodgers et al., 2010), and only one has observed these processes in individuals intending to become more active outside the context of an exercise intervention (including both non SDT interventions and SDT interventions; Fortier et al., 2009). This st udy adds to the current literature by providing an observational study of SDT hypothesized behavior change constructs in previously sedentary adults attempting to increase their PA Many of the previous studies have examined SDT processes in samples that included non exercisers, regular exercisers, and everyone in between. Because t he motivation continuum ranges from amotivation to intrinsic motivation, SDT may do a better job distinguishing regular exercisers from non exercisers than predicting PA adoptio n in exercise initiates For example, in this study, the controlled regulation composite score of the BREQ 2 demonstrated very low means and a restricted range of scores across all three time points. This suggests that participants did not strongly agree w ith the controlled regulation statements. Conversely, participants may not have endorse d intrinsic motivation, or engaging in exercise because of enjoyment, as high as regular exercisers. Previous research supports this hypothesis A cross stages of change, those in the pre preparation stages endorse the highest levels of controlled regulation (external, introjected) and the lowest levels of autonomous regulation (identified, intrinsic) compared to those in more advanced stages; individuals in the maintenanc e stage endorse the highest levels of autonomous regulation and the lowest levels of controlled regulation (Mullan & Markland, 1997). A group of individuals in the midst of a behavior change may be somewhere in the middle. Exercise initiates do generally b ecome more self determined in their exercise regulations during the first 6 months

PAGE 95

81 of behavior change; however, they may never fully reach the levels of autonomous motivation of regular exercisers or those who have exercised at least three times per week for a minimum of 6 months (Rodgers et al., 2010). Moreover, their endorsement of self determined regulation is more variable than that of regular exercisers, suggesting that their levels of motivation, as least in the first 6 months, are not stable (Rodger s et al., 2010). Exercise initiates likely see the value of exercise and it may be consistent with their goals, but still need the extrinsic forms of motivation to encourage the behavior. Indeed, previously sedentary exercise initiates may never experience sustained intrinsic motivation to engage in PA and the goal of behavior change may need to be integrated regulation of the behavior, or engaging in the behavior because it is consistent with one's innermost values, identity, and self concept ("I am an ex erciser"). Others have started to call for this approach (e.g., Stevens & Bryan, 2015) and suggest that capitalizing on the more integrated forms of external regulation may be the best inroad to long term maintenance of behavior change in this group. One of the interesting findings in this study was that the cross sectional SDT model s did not accurately predict PA behavior at 4 weeks, but it was more predictive of behavior at baseline and 12 weeks. This suggests that engaging in PA at 4 weeks after starting a new exercise program may require more strategies than just the motivation regulation suggested by SDT, and broader models of health behavior change may be necessary. Specifically, time dependent models of health behavior change may help predict behavior adoption and maintenance. The Transtheoretical Model ( TTM; also known as Stages of Change; Prochaska & Velicer, 1997 ) one such time dependent model of health behavior change, has been created in order to match strategies of inte rvention to the stage of change of the participant. However, the TTM lumps all behavior between preparation and maintenance into the

PAGE 96

82 action stage T he acti on stage may be the most important stage for understanding how individuals go from starting a behav ior to maintaining a behavior Understanding the psychological factors that occur during the action stage, including the variability of self determined regulations and meaning in life, may help us better explain behavior during this time. It is unclear wha t psychological processes are changing on a day to day basis in the midst of behavior change, likely because these factors are transient and difficult to capture with our current methods of measurement. Interestingly, those who reported greater levels of m eaning in life at baseline also reported more PA at 4 weeks, and meaning in life at baseline was one of the only predictors of change in PA from baseline to 4 weeks. This suggests that those who felt their lives were more meaningful at the beginning of a b ehavior change were engaging in more PA 4 weeks later. Further analyses revealed this was primarily through increased moderate activity at 4 weeks. This suggests that a strong sense of meaning in life may be an important factor that helps individuals in th e early stages of behavior change to adopt a new healthy behavior, such as PA As Ryff and Singer (1998) noted, believing that one's life is meaningful is a prerequisite to taking care of oneself with healthy behaviors. Individuals with a stronger sense of meaning in life also endorsed a greater connection between M/P and exercise at baseline and 4 weeks, and that connection was associated with more PA at baseline and 4 weeks. Thus, exercise initiates with a stronger sense of M/P tend to also believe that e xercise supports their achievement of life goals. In contrast to previous research ( Holahan et al. 2008; Holahan & Suzuki, 2006; Homan & Boyatzis, 2010; Hooker & Masters 2014 ; Ruuskanen & Ruoppila, 1995; Takkinen et al. 2001 ), M/P was largely not cross sectionally related to PA with a couple exceptions

PAGE 97

83 (greater meaning in life at 4 weeks was related to more visits to the fitness center, meaning in life at 12 weeks was related to more moderate PA at 12 weeks). This may be for a number of reasons. First, this is the first study to examine this relationship in a group of previously sedentary exercise initiates. The variability in PA behavior was likely restricted at baseline (individuals were sedentary) and then became much more variable over time. In cont rast, global evaluations of meaning in life were relatively stable over the 12 weeks of starting an exercise program. A second reason is that the relationship between a global sense of meaning in life and PA may not be perfectly linear. For example, Ryff and Singer's (1998) supposition that engaging in health behaviors requires a necessary level of M/P in life may be true, but does having much more meaning above and beyond the necessary level translate to inc reased engagement in healthy behaviors? Meaning salience, or the extent to which individuals think about and are aware of that meaning, seems to have a stronger relationship with PA than global ratings of M/P as was observed in this study. Thus, meaning i n life at baseline may predict PA at 4 weeks above and beyond the SDT Process M odel of Behavior Change because of the sense that engaging in PA is important for supporting health sense of M/P, and life goals. This hypothesis was supported in the explorato ry analyses, which revealed that the extent to which participants endorsed a meaning and exercise connection was related to global meaning and purpose as well as PA at baseline. However, these associations seemed to weaken over the course of the study. Per haps individuals are most motivated to engage in PA behavior at the beginning of starting an exercise program; they view the behavior as consistent with their goals and sense of purpose and they imagine that their future selves are happier after they achi eve their goals

PAGE 98

84 Then the real work of having to engage in a regular PA routine starts and motivation likely wanes, especially for those who do not maintain the activity. Another interesting finding was that PA at earlier time points was longitudinally related to greater M/P at later time points. This raises the question which comes first? The hypotheses in this study revolve around the idea that meaning is motivation for engaging in healthy behaviors, but per haps there is a bi directional relationship. For some, engaging in PA could increase a sense of meaning, which may be related to the findings that PA improves mood (Annesi, 2004) and executive functioning ( Etnier & Chang, 2009 ). Meaning is a combination of affective, motivational, and cognitive factors ( Reker, 2000; Wong, 1989 ), and the sense of coherence that stems from meaning (e.g., life makes sense) could be related to better executive functioning. Other evidence suggests that models of change in both p sychological factors and behavior are relatively equivalent in terms of which comes first behavior or psychological constructs ( Gunnell et al., 2016). Perhaps in some cases, the behavior comes first, and then individuals attempt to rationally explain the behavior (e.g., I exercise because it is consistent with my goals or because my life must be meaningful ). Unfortunately, the type of analysis required for this hypothesis (latent growth curve modeling) requires larger samples than that was recruited for t his project. The main findings of this study suggest that modifications to health behavior change theories are needed. Indeed, the process of health behavior change seems to be a dynamic process, and drilling down to the daily decisions of individuals in the process of making a change may be necessary to fully understand the psychological constructs predicting changes. Moreover, meaning in life and meaning salience need to be considered in the modification of health behavior change theori es. A sense of meaning is closely tied to an

PAGE 99

85 individual's identity, which is likely related to long term maintenance of behavior change. For a behavior to become a habit, it must be fully integrated with the self, and meaning in life and autonomous regulat ion may be key factors to make PA behavior more habitual. Thus, theoretical modifications, and resulting interventions, are needed to fully understand and improve engagement in health behavior. Strengths This study has several strengths. First, this was th e first study of previously sedentary exercise initiates that examine M/P in the context of behavior change. Second, objective assessments of PA (fitness center visits), body size (BMI, body fat, and waist circumference), and physical fitness (cardiovascul ar, strength, and flexibility) were gathered to support the primary outcome of self reported PA Third the study used a RCT design to examine the effects of a measurement intervention on primary and secondary outcomes. Finally, the study's emphasis on me aning salience, rather than global ratings of M/P, is an innovative contribution to the field. Limitations The primary limitation of this study is reliance on self report measures of PA Self report PA measures are known to have poor associations with obje ctive PA measures (Troiano, Pettee Gabriel, Welk, Owen, & Sternfeld, 2011), but are considered reliable for rank ordering PA behavior (M ‰ sse & de Niet, 2011). Thus, the self reported PA measures should be reliable in distinguishing those who do more activi ty from those who do less activity, but the absolute levels of PA may not be representative of actual behavior. A second and important limitation is that the sample was predominantly female, highly educated, and able to afford a membership to a fitness cen ter. Thus, these results may not generalize to less

PAGE 100

8 6 educated and less affluent samples although it may be representative of those who join private fitness centers Also, as with all l ongitudinal studies, missing data are of concern. It is likely that part icipants who were lost to follow up or dropped out were less likely to continue to engage in PA However, the proportion of participants lost in this study (12.5%) was much less than what was expected (20%). Further, missing data analyses suggested that th ere were very few differences between those who completed the follow up assessments at 4 weeks and 12 weeks and those who did not complete those assessments. Of note, participants retrospectively reported their daily meaning salience and PA at the end of t he day. Thus, this study cannot determine the within day time order of meaning salience and PA It could be that participants engaged in PA which increased their meaning salience, rather than meaning salience increasing the likelihood of engaging in PA A different ecological momentary assessment (EMA) design may be able to assess the question of time order more effectively. A final limitation to note is that although this study is able to establish time order relationships across the 12 weeks, it is not able to establish causality. A sense of meaning in life may not be fully subjected to experimental manipulation, i.e., we c annot assign individuals to have meaning or to not have meaning but interventions to incr ease meaning salience may be an appropriate strategy to test these ideas Future Directions There are several possible future directions for this research. One would be to recruit a larger sample from the general community, without the requirement that the participants be joining a fitness center, or to recruit from several different fitness centers, to increase the generalizability of the findings and replicate the r esults. Secondly, examining these processes over longer time periods (more than three months) and in larger samples would allow for

PAGE 101

87 more complex linear growth curve modeling to examine the changes in the predictors (meaning in life, SDT mediators) and PA b ehavior. An EMA design with multiple within day measurements and objective monitoring of PA would improve the assessment of time ordering of meaning salience and PA behavior. For example, the objective PA monitor could capture bouts of PA and then examine whether there was a stronger relationship between meaning salience prior to a bout of PA behavior or between PA and meaning salience after a bout of activity. Finally, interventions incorporating meaning salience to increase PA in sedentary adults with int entions to be more active would be a logical experimental step. It would be hypothesized that individuals who were primed to think more about meaning salience and to integrate their sense of M/P with their reasons to be more active would be more likely to maintain PA over time. Conclusions PA is an important health behavior for preventing many c hronic illnesses including CVD, as well as for increasing mental health, vitality, and longevity ( Physical Activity Guidelines Advisory Committee, 2008). Most adu lts do not engage in regular PA and determining factors related to long term PA maintenance in previously sedentary exercise initiates is vital to our understanding of behavior change. This study provides evidence that basic psychological needs satisfacti on and autonomous regulation from the SDT Process Model of Behavior Change and meaning salience are related to PA adoption. Future research should continue to explore these factors in exercise initiates in combination with other health behavior change the ories, in order to advance health behavior theory and design better interventions to improve PA adoption and maintenance.

PAGE 102

88 REFERENCES Adabonyan, I., Loustalot, F., Kruger, J., Carlson, S. A., & Fulton, J. E. (2007). Prevalence of highly active adults Beh avioral risk factor surveillance system, 2007. Preventive Medicine, 51, 139 143. Annesi, J. J. (2004). Mood states of formerly sedentary younger and older women at weeks 1 and 10 of a moderate exercise program. Psychological Reports, 94, 1337 1342. Anshel M. H. (2010). The disconnected values (intervention) model for promoting healthy habits in religious institutions. Journal of Religion and Health, 49, 32 49. Anshel, M. H., Brinthaupt, T. M., & Kang, M. (2010). The disconnected values model improves ment al well being and fitness in an employee wellness program. Behavioral Medicine, 36, 113 122. Anshel, M. H., Kang, M., & Brinthaupt, T. M. (2011). A values based approach for changing exercise and dietary habits: An action study. International Journal of Sp ort and Exercise Psychology, 8, 413 432. Barbeau, A., Sweet, S N., & Fortier, M. (2009). A path analytic model of self determination theory in a physical activity context. Journal of Applied Biobehavioral Research, 14, 103 118. Borg, G. (1998). Borg's perc eived exertion and pain scales. Champaign, IL: Human Kinetics. Bostic, T. J., Rubio, D. M, & Hood, M. (2000). A validation of the subjective vitality scale using structural equation modeling. Social Indicators Research, 52, 313 324. Boyle, P. A., Barnes, L. L., Buchman, A. S., & Bennett, D. A. (2009). Purpose in life is associated with mortality among community dwelling older persons. Psychosomatic Medicine, 71, 574 579. Butryn, M. L., Forman, E., Hoffman, K., Shaw, J., & Juarascio, A (2011). A pilot study of acceptance and commitment therapy for promotion of physical activity. Journal of Physical Activity and Health, 8, 516 522. Carels, R., 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, 706 722. Chatzsirantis, N. L. D., & Hagger, M. S. (2007). Intrinsic motivation and self determination in exercise and sport: Reflecting on the past and sketchin g the future. In M. S. Hagger & N. L. D. Chatzirantis (eds.), Intrinsic motivation and self determination in exercise and sport (pp. 281 296) Champaign, IL; Human Kinetics.

PAGE 103

89 Cohen, R., Bavishi, C., & Rozanski, A. (2016). Purpose in life and its relationsh ip to all cause mortality and cardiovascular events: A meta analysis. Psychosomatic Medicine, 78, 122 133. Craig, C. L., Marshall, A. L., Sjšstršm, M., Bauman, A. E., Booth, M. L., Ainsworth, B. E. Oja, P. (2003). International Physical Activity Questionn aire: 12 Country reliability and validity. Medicine & Science in Sports & Exercise, 35, 1381 1395. Deci, E. L., & Ryan, R. M. (2000). The "what" and "why" of goal pursuits: Human needs and the self determination of behavior. Psychological Inquiry, 11, 227 268. Diener, E., Emmons, R., Larsen, R., & Griffin, S. (1985). The Satisfaction with Life Scale. Journal of Personality Assessment, 49, 71 75. Edmunds, J., Ntoumanis, N., & Duda, J. L. (2007). Adherence and well being in overweight and obese patients referred to an exercise on prescription scheme: A self determination theory perspective. Psychology of Sport and Exercise, 8, 722 740. Etneir, J. L., & Chang, Y. (2009). The effect of physical activity on executive function: A brief commentary on definitio ns, measurement issues, and the current state of the literature. Journal of Sport and Exercise Psychology, 31, 469 483. Faul, F. (2007). G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav i ora l Research Methods, 39 (2), 175 191. Fitzmaurice, G. M., Laird, N. M., & Ware, J. M. (2011). Applied longitudinal analysis (2 nd ed.). Hoboken, NJ: John Wiley & Sons, Inc. Fortier, M. S., Duda, J. L., Guerin, E., & Teixeira, P. (2012). Promoting physical act ivity: Development and testing of self determination theory based interventions. International Journal of Behavioral Nutrition and Physical Activity, 9, 20. Fortier, M. S., Hogg, W., O'Sullivan, T. L., Blanchard, C., Reid, R. D., Sigal, R. J., Beaulac J. (2007). The physical activity counseling (PAC) randomized controlled trial: Rationale, methods, and interventions. Applied Physiology Nutrition and Metabolism, 32, 1170 1185. Fortier, M. S., Hogg, W., O'Sullivan, T., Blanchard, C., Sigal, R. J., Reid, R. D., Culver, D. (2011). Impact of integrating a physical activity counsellor into the primary health care team: physical activity and health outcomes of the physical activity counselling (PAC) randomized control trial. Applied Physiology Nutrition and Metabolism, 36, 503 514. Fortier, M. S., Kowal, J., Lemyre, L., & Orpana, H. M. (2009). Intentions and actual physical activity behavior change in a community based sample of middle aged women: Contributions from the theory of planned behavior and self de termination theory. International Journal of Sport and Exercise Psychology, 7, 46 67.

PAGE 104

90 Frankl, V. E. (1985). Man's search for meaning. New York, NY: Simon and Schuster. Glazer, N. L., Lyass, A., Esliger, D. W., Blease, S. J., Freedson, P. S., Massaro, J. M. Vasan, R. S. ( 2013). Sustained and shorter bouts of physical activity are related to cardiovascular health. Medicine and Science in Sports & Exercise, 45 (1), 109 115. Golding, L. A. (2000). YMCA fitness testing and assessment manual. Champaign, IL: Huma n Kinetics. Goodwin, R. D. (2003). Association between physical activity and mental disorders among adults in the United States. Preventive Medicine, 36, 698 703. Gorin, A. A., Pinto, A., West, D., Niemeier, H., Fava, J., & Wing, R. R. (2008). Losing weight because you want to rather than because you feel you have to: Motivational predictors of weight loss outcomes. Obesity, 16, S214. Gunnell, K. E., BÂŽlanger, M., & Brunet, J. (2016). A tale of two models: Changes in psychol ogical needs satisfaction and physical activity over 3 years. Health Psychology, 35, 167 177. Gunnell, K. E., Crocker, P. R. E., Mack, D. E., Wilson, P. M., & Zumbo, B. D. (2014). Goal contents, motivation, psychological need satisfaction, well being and p hysical activity: A test of self determination theory over 6 months. Psychology of Sport and Exercise, 15, 19 29. Hagger, M. S. & Chatzi sa rantis, N. L. D. (2008). Self determination theory and the psychology of exercise. International Review of Sport and Exercise Psychology, 11, 79 103. Hagger, M. S., Chatzisarantis, N. L. D., & Harris, J. (2006a). The process by which relative autonomous motivation affects intentional behavior: Comparing effects across dieting and exercise behaviors. Motivation & Emotion, 30, 307 321. Hagger, M. S., Chatzisarantis, N. L. D., & Harris, J. (2006b). From psychological need satisfaction to intentional behavior: Testing a motivational sequence in two behavioral contexts. Personality and Social Psychology Bulletin, 32, 131 148. Hall, C. R., Rodgers, W. M., Wilson, P. M., & Norman, P. (2010). Imagery use and self determined motivation in a community sample of exercisers and non exercisers. Journal of Applied Social Psychology, 40, 135 152. Harris, P. A., Taylor, R., Thiel ke, R., Payne, J., Gonzalez, N., & Conde, J. G. (2009). Research electronic data capture (REDCap) A metadata driven methodology and workflow process for providing translational research informatics support. Journal of Biomedical Informatics, 42 (2), 377 3 81. Hawkins, M. S., Storti, K. L., Richardson, C. R, King, W. C., Strath, S. J., Holleman, R. G., & Kriska, A. M. (2009). Objectively measured physical activity of USA adults by

PAGE 105

91 sex, age, and racial/ethnic groups: A cross sectional study. International Journal of Behavioral Nutrition and Physical Activity, 6 31. Hayes, S. C., Strosahl, K., & Wilson, K. G. (1999). Acceptance and commitment therapy: An experimental approach to behavior change. New York: Guilford Press. Hicks, J. A., & King, L. A. (2008). Religious commitment and positive mood as information about meaning in life. Journal of Research in Personality, 42, 43 57. Hicks, J. A., & King, L. A. (2009). Positive mood and social relatedness as information about meaning in life. The Journal of Positi ve Psychology: Dedicated to furthering research and promoting good practice, 4, 471 482. Holahan, C. K., Holahan, C. J., & Suzuki, R. (2008). Purposiveness, physical activity, and perceived health in cardiac patients. Disability and Rehabilitation, 30 177 2 1778. Holahan, C. K., & Suzuki, R. (2006). Motivational factors in health promoting behavior in later aging. Activities, Adaptation, & Aging, 30 47 60. Holt, L. E., Pelham, T. W., & Burke, D. G. (1999). Modifications to the standard sit and reach flexib ility protocol. Journal of Athletic Training, 34, 43 47. Homan, K. J., & Boyatzis, C. J. (2010). Religiosity, sense of meaning, and health behavior in older adults. The International Journal for the Psychology of Religion, 20 173 186. Hooker, S. A., MacGr egor, K. L., Funderburk, J. S., & Maisto, S. A. (2013). Body mass index and depressive symptoms in primary care settings: Examining the moderating roles of smoking status, alcohol consumption and vigorous exercise. Clinical Obesity, 4, 21 29. Hooker, S. A. & Masters, K. S. (2014). Purpose in life is associated with physical activity measured by accelerometer. Journal of Health Psychology. Advance online publication. Retrieved from http://hpq.sagepub.com/content/early/2014/07/28/1359105314542822 Hu, L., & B entler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1 55. Ingledew, D. K., & Markland, D. (2008). The role of motives in exercise participat ion. Psychology and Health, 23 807 28. Jolly, K., Duda, J. L., Daley, A., Eves, F. F., Mutrie, N., Ntoumanis, N., Williams, G. C. (2013). Evaluation of a standard provision versus an autonomy promotive exercise referral programme: Rationale and study des ign. BMC Public Health, 9, 1 9.

PAGE 106

92 King, L. A., Hicks, J. A., Krull, J. L., & Del Gaiso, A. K. (2006). Positive affect and the experience of meaning in life. Journal of Personality and Social Psychology, 90, 179 196. Krause, N. (2009). Meaning in life and mor tality. The Journals of Gerontology: Series B: Psychological Sciences and Social Sciences, 64B, 517 527. Krause, N., & Shaw, B. A. (2003). Role specific control, personal meaning, and health in late life. Research on Aging, 25, 559 586. Kroenke, K., & Spitzer, R. (2002). The PHQ 9: A new depression diagnostic and severity measure. Psychiatric Annals, 32, 509 515. Kwan, B., Caldwell Hooper, A., Magnan, R. E., & Bryan, A. D. (2011). A longitudinal diary study of the effects of causality orientations on e xercise related affect. Self and Identity, 10, 363 374. Landry, J. B., & Solomon, M. A. (2004). African American women's self determination across the stages of change for exercise. Journal of Sport and Exercise Psychology, 26, 457 469. Lloyd Jones, D., Adams, R., Carnethon, M., De Simone, G., Ferguson, B., Ford, E. Hong, Y. (2009). Heart disease and stroke statistics 2009 update: a report from the American Heart Association Statistics Committee and Stroke Statistics Subcommittee. Circul ation, 119, 480 486. Lloyd Jones, D. M., Hong, Y., Labarthe, D., Mozaffarian, D., Appel, L. J., Van Horn, L. Rosamond, W. D. (2010). Defining and setting national goals for cardiovascular health promotion and disease reduction: The American Heart Associa tion's strategic impact goal through 2020 and beyond. Circulation, 121, 586 613. Lutz, R. S., Stults Kolehmainen, M. A., & Bartholomew, J. B. (2010). Exercise caution when stressed: Stages of change and the stress exercise participation relationship. Psych ology of Sport and Exercise, 11, 560 567. Macera, C. A., Ham, S. A., Yore, M. M., Jones, D. A., Kimsey, C. D., Kohl, H. W., & Ainsworth, B. E. (2005). Prevalence of physical activity in the United States: Behavioral Risk Factor Surveillance System, 2001. P reventing Chronic Disease, 2 (2), A17. M ‰ sse, L. C. & de Niet, J. E. (2011). Sources of validity evidence needed with self report measures of physical activity. Journal of Physical Activity and Health, 9, S44 S55. Marcus, B. H., Dubbert, P. M., Forsyth, L. H., McKenzie, T. L., Stone, E. J., Dunn, A. L., & Blair, S. N. (2000). Physical activity behavior change: Issues in adoption and maintenance. Health Psychology, 19, 32 41.

PAGE 107

93 Marcus, B. H., Owen, N., Forsyth, L. H., Cavill, N. A., & Fridinger, F. (1998). Physical activity interventions using mass media, print media, and information technology. American Journal of Preventive Medicine, 15, 362 378. Markland, D., & Ingledew, D. K. (2007). Exercise participation motives. In M. S. Hagger & N. L. D. Chatzirantis (eds.), Intrinsic motivation and self determination in exercise and sport (pp. 23 34) Champaign, IL; Human Kinetics. Markland, D., & Tobin, V. (2004). A modification to the Behavioural Regulation in Exercise Questi onnaire to include an assessment of amotivation. Journal of Sport Exercise Psychology, 26, 191 196. McKnight, P. E., & Kashdan, T. B. (2009). Purpose in life as a system that creates and sustains health and well being: An integrative, testable theory. Revi ew of General Psychology, 13, 242 251. Mullan, E., & Markland, D. (1997). Variations in self determination across stages of change for exercise in adults. Motivation and Emotion, 21 349 362. Mullan, E., Markland, D., & Ingledew, D. K. (1997). A graded con ceptualization of self determination in the regulation of exercise behavior: Development of a measure using confirmatory factor analytic procedures. Personality and Individual Differences, 23, 745 752. MuthŽn B & MuthŽn L. (2010). Mplus version 6 [Comp uter software]. O'Connor, B. P., & Vallerand, R. J. (1998). Psychological adjustment variables as predictors of mortality among nursing home residents. Psychology and Aging, 13, 368 374. Olander, E. K., Fletcher, H., Williams, S., Atkinson, L., Turner, A., & French, D. P. (2013). What are the most effective techniques in changing obese individuals' physical activity self efficacy and behaviour: A systematic review and meta analysis. Intern ational Journal of Behavioral Nutrition and Physical Activity, 10, 29. Percia, M., Davis, S., & Dwyer, G. (2012 January 10 ). Getting a professional fitness assessment. ACSM Fit Society. Retrieved from https://www.acsm.org/public information/articles/2012/ 01/10/getting a professional fitness assessment Physical Activity Guidelines Advisory Committee. (2008). Physical activity guidelines advisory committee report, 2008. Washington, DC: U.S. Department of Health and Human Services. Pinquart, M. (2002). Creati ng and maintaining purpose in life in old age: A meta analysis. Ageing International, 27, 90 114. Probst Hensch, N., Tanner, M., Kessler, C., Burri, C., & KŸnzli, N. (2011). Prevention: A cost effective way to fight the non communicable disease epidemic. S wiss Medical Weekly, 141, 1 8.

PAGE 108

94 Prochaska, J. O., & Velicer, W. F. (1997). The transtheoretical model of health behavior change. American Journal of Health Promotion, 12, 38 48. Reker, G. T. (2000). Theoretical perspective, dimensions, and measurement of ex istential meaning. In G. T. Reker, Exploring existential meaning: Optimizing human development across the life span (pp. 39 55). Thousand Oaks, CA: Sage Publications. Reker, G. T., Peacock, E. J., & Wong, P.T. (1987). Meaning and purpose in life and well being: A life span perspective. Journal of Gerontology, 42, 44 49. Rodgers, W. M., Hall, C. R., Duncan, L. R., Pearson, E., & Milne, M. I. (2010). Becoming a regular exerciser: Examining change in b ehavioural regulations among exercise initiates. Psychology of Sport and Exercise, 11, 378 386. Rouse, P. C., Duda, J. L., Ntoumanis, N., Jolly, K. B., & Williams, G. C. (2010). A longitudinal examination of self determination theory in participants of a U K based exercise on referral program. Journal of Sport and Exercise Psychology, 32 215. Ruuskanen, J. M., & Ruoppila, I. (1995). Physical activity and psychological well being among people aged 65 to 84 years. Age and Ageing, 24 292 296. Ryan, R., & Deci, E. (2000). Self determination theory and the facilitation of intrinsic motivation, social development, and well being. American Psychologist, 55 68 78. Ryan, R. M., & Deci, E. L. (2001). On happiness and human potentials: A review of research on he donic and eudaimonic well being. Annual Review of Psychology, 52, 141 166. Ryan, R. M., & Frederick, C. M. (1997). On energy, personality, and health: Subjective vitality as a dynamic reflection of well being. Journal of Personality, 65 529 565. Ryff, C. & Singer, B. (1998). The contours of positive human health. Psychological Inquiry, 9, 1 28. Sagy, S., Antonovsky, A., & Adler, I. (1990). Explaining life satisfaction in later life: The sense of coherence model and activity theory. Behavior, Health, & Agi ng, 1, 11 25. SAS Institute, Inc. (20 15 ). The SAS System for Windows (Version 9. 4 ) [Computer software]. Cary, NC: SAS Institute, Inc. Sebire, S. J., Standage, M., & Vansteenkiste M. (2009). Examining intrinsic versus extrinsic exercise goals: Cognitive, affective, and behavioural outcomes. Journal of Sport and Exercise Psychology, 31 189 210. Scheier, M. F., Wrosch, C., Baum, A., Cohen, S., Martire, L. M., Matthews, K. A., Zdan iuk, B. (2006). The Life Engagement Test: Assessing purpose in life. Journal of Behavioral Medicine, 29, 291 298.

PAGE 109

95 Schwarz, N. (2001). Feelings as information: Implications for affective influences on information processing. In L. L. Martin & G. L. Clove ( Eds.), Theories of mood and cognition: A user's guidebook (pp. 159 176). Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Shortreed, S. M., Peeters, A., & Forbes, A. B. (2013). Estimating the effect of long term physical activity on cardiovascular disease and mortality: Evidence from the Framingham Heart Study. Heart, 99, 649 654. Silva, M. N., Markland, D. A., Minderico, C. S., Vieira, P. N., Castro, P. N., Coutinho, S. R., Teixeira, P. (2008). A randomized controlled trial to evaluate self determination th eory for exercise adherence and weight control: Rationale and intervention description. BMC Public Health, 8, 234 237. Silva, M. N., Markland, D., Vieira, P. N., Coutinho, S. R., Carraa, E. V., Palmeira, A. L., Teixeira, P. J. (2010). Helping overweight women become more active: Need support and motivational regulations for different forms of physical activity. Psychology of Sport and Exercise, 11 592 601. Silva, M. N., Vieira, P. N., Coutinho, S. R., Matos, M. G., Sardinha L. B., Teixeira, P. J. (2009). Using self determination theory to promote physical activity and weight control: A randomized controlled trial in women. Journal of Behavioral Medicine, 33 110 122. Spring, B., Gotsis, M., Paiva, A., & Spruijt Metz, D. (2 013). Healthy apps. IEEE Pulse, 4, 34 40. Spruijt Metz, D., Hekler, E., Saranummi, N., Intille, S., Korhonen, I., Nilsen, W., Pavel, M. (2015). Building new computational models to support health behavior change and maintenance: New opportunities in behav ioral research. Translational Behavioral Medicine, 5, 335 346. Standage, M., Sebire, S. J., & Loney, T. (2008). Does exercise motivation predict engagement in objectively assessed bouts of moderate intensity exercise? A self determination theory perspectiv e. Journal of Sport and Exercise Psychology, 30 337 352. Steger, M. F., Frazier, P., Oishi, S., & Kaler, M. (2006). The Meaning in Life Questionnaire: Assessing the presence of and search for meaning in life. Journal of Counseling Psychology, 53 80 93. Steger, M. F., Kashdan, B., & Oishi, S. (2008). Being good by doing good: Daily eudaimonic activity and well being. Journal of Research in Personality, 42, 22 42. Steger, M. F., Oishi, S., & Kashdan, T. B. (2009). Meaning in life across the life span: Leve ls and correlates of meaning in life from emerging adulthood to older adulthood. The Journal of Positive Psychology, 4 (1), 43 52.

PAGE 110

96 Stevens, C. J., & Bryan, A. D. (2015). A case for leveraging integrated regulation strategies to optimize health benefits from self determined exercise behavior. Annals of Behavioral Medicine, 49, 783 784. Takei Hand Grip Dynamometer A5401 [Apparatus]. (n.d.). Tokyo, Japan: Takei Scientific Instruments. Takkinen, S., Suutama, T., & Ruoppila I. (2001). More meaning by exercising? Physical activity as a predictor of a sense of meaning in life and self rated health and functioning in old age. Journal of Aging and Physical Activity, 9 128 141. Taliaferro, L. A., Rienzo, B. A., Pigg, J. R. M., Miller, M. D., & Dodd, V. J. (2008). Associations between physical activity and reduced rates of hopelessness, depression, and suicidal behavior among college students. Journal of American College Health, 57, 427 435. Teixeira, P. J., Carraa E. V., Markland, D., Silva, M. N. & Ryan, R. M. (2012). Exercise, physical activity, and self determination theory: A systematic review. International Journal of Behavioral Nutrition and Physical Activity, 9, 78. Teixeira, P. J., Silva, M. N., Mata, J., Palmeira, A. L., & Markland, D. (2012). Motivation, self determination, and long term weight control. International Journal of Behavioral Nutrition and Physical Activity, 9 22. Thomas, S., Reading, J., & Shephard, R. I. (1992). Revision of the Physical Ac tivity Readiness Questionnaire. Canadian Journal of Sport Sciences, 17 338 345. Troiano, R. P., Berrigan, D., Dodd, K. W., Masse, L. C., Tilert, T., & McDowell, M. (2008). Physical activity in the United States measured by accelerometer. Medicine and Scie nce in Sport and Exercise, 40, 181 188. Troiano, R. P., Pettee Gabriel, K. K., Welk, G. J., Owen, N. & Sternfeld, B. (2011). Reported physical activity and sedentary behavior: Why do you ask? Journal of Physical Activity and Health, 9, S68 S75. Vlachopoulos, S. P., & Michailidou, S. (2006). Development and initial validation of a measure of autonomy, competence, and relatedness in exercise: The Basic Psychological Needs in Exercise Scale. Measurement in Physical Education and Exercise Science 10 (3), 179 201. Watson, D., Clark, L., & Tellegen, A. (1988). Development and validation of brief measures of positive and negative affect: The PANAS Scales. Journal of Personality and Social Psychology, 54 (6), 1063 1070. Weinstein, N., Ryan, R. M., & Deci, E. L. (2012). Motivation, meaning, and wellness: A self determination perspective on the creation and internalization of personal meanings and life goals. In P. T. P. Wong (ed.), The human quest for meaning:

PAGE 111

97 Theories, research, and applications (pp. 81 106 ). New York, NY: Routledge, Taylor & Francis Group. Wilson, P. M., & Rodgers, W. M. (2002). Examining the psychometric properties of the behavioural regulation in exercise questionnaire. Measurement in Physical Education & Exercise Science 6 1 21. Wilso n, P., Rodgers, W. M, Fraser, S. N., & Murray, T. C. (2004). The relationship between exercise regulations and motivational consequences in men and women. Research Quarterly for Exercise and Sport, 75 81 91. Wilson, P. M., Rodgers, W. M., Loitz, C. C., & Scime, G. (2006). "It's who I amreally!" The importance of integrated regulation in exercise contexts. Journal of Applied Biobehavioral Research, 11, 79 104. Wilson, P. M., Sabiston, C. M., Mack, D. E., & Blanchard, C. M. (2012). On the nature and functio n of scoring protocols used in exercise motivation research: An empirical study of the behavioral regulation in exercise questionnaire. Psychology of Sport and Exercise, 13, 614 622. Wong, P. T. (1989). Personal meaning and successful aging. Canadian Psychology/Psychologie Canadienne, 30, 516 525.

PAGE 112

98 A PPENDIX A Daily Quotes for the Self Monitoring Condition Day Quote 1 "There is great difficulty for a person to have a purpose in his future if he has no sense of accomplishment in his past." 2 "A life that is not self examined, self criticized is not a life worth living." Plato 3 "My principle is: for heaven's sake, do not be perfect but by all means try to be complete, whatever that means." Carl Jung 4 "Every good habit that is worth possessing mus t be paid for in strokes of daily effort." Charles Darwin 5 "The greatest discovery of our generation is that human beings, by changing the inner attitudes of their minds, can change their lives." 6 "True enjoyment comes from activity of the mind & exe rcise of the body; the two are ever united." Humboldt 7 "Goals that are not written down are just wishes." Unknown 8 "You will never find time for anything. You must make it." Charles Burton 9 "It is never too late to be what you might have been." George Elliot 10 "The tragedy of life doesn't lie in not reaching your goal. The tragedy lies in having no goal to reach. It isn't a calamity to die with dreams unfulfilled, but it is a calamity not to dream." Benjamin Elijah Mays 11 "Consult not yo ur fears, but your hopes and dreams. Think not about your frustrations, but about your unfulfilled potential. Concern yourself not with what you have tried and failed in, but with what it is still possible for you to do." Pope John XXIII 12 "You've got to think about big things' while you're doing small things, so that all the small things go in the right direction." Alvin Toffler 13 "Love is our true destiny. We do not find the meaning of life by ourselves alone we find it with another." Thomas Merton 14 "There is not one big cosmic meaning for it all; there is only the meaning we each give to our life, an individual meaning, an individual plot, like an individual novel, a book for each person." Ana•s Nin 15 "Human life is purely a matter of de ciding what's important to you." Anonymous

PAGE 113

99 16 "When we are motivated by goals that have deep meaning, by dreams that need completion, by pure love that needs expressing then we truly live life." Greg Anderson 17 "If you can't figure out your purpose, figure out your passion. For your passion will lead you right into your purpose." Bishop T.D. Jakes 18 "When you discover your mission, you will feel its demand. It will fill you with enthusiasm and a burning desire to get to work on it." W. Clement Stone 19 "For the meaning of life differs from man to man, from day to day and from hour to hour. What matters, therefore, is not the meaning of life in general but rather the specific meaning of a person's life at a given moment." Victor E. Frankl 20 "He who has a why' to live for can bear almost any how.'" Frederick Nietzsche 21 "The mystery of human existence lies not in just staying alive, but in finding something to live for." Fyodor Dostoyevsky 22 "You were put on this earth to achieve your greatest self, to live out your purpose, and to do it courageously." Steve Maraboli 23 "The purpose of life is a life of purpose." Robert Byrne 24 "The best way to insure you achieve the greatest satisfaction out of life is to behave in tentionally." Deborah Day 25 "Purpose expresses most deeply what makes you a unique individual. Your purpose defines who you are, how you live your life, and how you lead. Your purpose provides you with inner strength and a drive to live and lead each day. It equips you with what you need to face the challenges of the day and of life. Your purpose provides context and meaning to your life." Thomas Narofsky 26 "You are here for a special mission. There is a purpose of your life. Find that purpose of your life, work on it and live you r life happily." Raaz Ojha 27 "Finding your purpose is a lifelong adventure. Enjoy the journey." Todd Stocker 28 "I believe a purpose is something for which one is responsible; it's not just divinely assigned." Michael J. Fox

PAGE 114

100 APPENDIX B Screenin g and Eligibility Form Inclusion Yes No 1. Are you between the ages of 3 0 and 89 years old? 2. Are you able to read and understand English? 3. Are you planning to join (or have you joined in the last week) the Anschutz Health and Wellness Fitness Center? If a participant answers NO to any of the above questions, he or she is not eligible. Physical Activity Staging Regular Physical Activity : For physical activity to be considered "regular" it must be done for 30 minutes at a time (or more) per day and be done at least four days per week. The intensity of activity does not have to be vigorous but should be enough to increase your heart rate and/or breathing level somewhat. Examples of activities could include brisk walking, leisure biking, swimming line dancing, and aerobics classes or any other activities with a similar intensity level. According to the above definition: 4. Do you currently engage in regular physical activity? Yes No 5. Do you intend to engage in regular physical activity in the next 6 months? Yes No 6. Do you intend to engage in regular physical activity in the next 30 days? Yes No 7. Have you been regularly physically active for the past 3 months? Yes No Godin Leisure Time Exercise Questionnaire 8. During a typical 7 day period (a week), how many times on the average do you do the following kinds of exercise for more than 15 minutes during your free time (write on each line the appropriate number)? Times per Week Minutes per Week (t otal) a. Strenuous Exercise (in which your heart beats rapidly) ___________ ___________

PAGE 115

101 (For example: running, jogging, hockey, football, soccer, squash, basketball, cross country skiing, judo, roller skating, vigorous swimming, vigorous long distance bicycling) b. Moderate Exercise (which is not exhausting) (For example: fast walking, baseball, tennis, easy bicycling, volleyball, badminton, easy swimming, alpine skiing, dancing) ___________ ___________ c. Mild Exercise (minimal effort) (For example: yoga, archery, fishing, bowling, horseshoes, golf, snowmobiling, easy walking) ___________ ___________ 9. During a typical 7 day period, in your leisure time, how often do you engage in any regular activity long enough to work up a sweat and you r heart beats rapidly? Often Sometimes Never/Rarely Exclusion Yes No 10. Has your doctor ever told you that you have cardiovascular disease, heart disease, or heart failure (Hypertension is OK) ? 11. (For women only) Are you currently pregnant? 12. Has your doctor ever said that you have a heart condition and that you should only do physical activity recommended by a doctor? 13. Do you feel pain in your chest when you do physical activity? 14. In the past month, have you had chest pa in when you are not doing physical activity? 15. Do you lose your balance because of dizziness or do you ever lose consciousness? 16. Do you have a bone or joint problem (for example, back, knee or hip) that could be made worse by a change in your physical activity?

PAGE 116

102 17. Is your doctor currently prescribing drugs (for example, water pills) for your blood pressure or heart condition? 18. Do you know of any other reason why you should not do physical activity? If participant answers YES to any of t he above questions, he/she is not eligible.

PAGE 117

103 APPENDIX C Baseline Only Measures Demographics 1) Sex: _________ Male _________ Female 2) Age: ___________ 3) Choose one racial group that best describes you: ____ White ____ Native Hawaiian or other Pacific Islander ____ Black or African American ____ Mixed Race (more than 1) ___ Asian ____ Other (please specify) ____ American Indian/Alaskan Native 4) Choose one ethnic category that best describes you: ______ Hispa nic or Latino ______ Not Hispanic or Latino 5) Marital Status: ______ Single, never married ______ Currently married ______ Currently separated ______ Currently divorced ______ Widowed 6) Employment status: ______ Employed full time ______ Total disabled temporary ______ Employed part time ______ Total disabled permanent ______ Retired ______ Unemployed ______ Partially disabled temporary ______ Student ______ Partially disabled p ermanent ______ Homemaker 7) Highest level of education completed: ______ Less than high school ______ 2 year college degree (Associates)

PAGE 118

104 ______ High school/GED ______ 4 year college degree (BA, BS) ______ Some college ______ Graduat e or professional degree 8) What is the approximate annual income for your household? ______ Less than $20,000 ______ $20,000 $39,999 ______ $40,000 $59,999 ______ $60,000 $79,999 ______ $80,000 $99,999 ______ More than $100,000 9. What is your current religious affiliation? ______ Catholic ______ Protestant Christian (if so, which denomination? ______________________) ______ Latter Day Saint (Mormon) ______ Jewish ______ Muslim ______ Hindu ______ Buddhist ______ Atheist ______ None ______ Other (Please specify: _____________________________) Medical History 1. Do you have a family history (father, mother, grandparents or siblings) of coronary heart disease? YES NO 2. Do you h ave a family history (father, mother, grandparents or siblings) of stroke? YES NO 3. Do you have a family history (father, mother, grandparents or siblings) of hypertension? YES NO 4. Do you have a family history (father, mother, grandparents or siblings) of diabetes? YES NO

PAGE 119

105 5. Do you have a family history (father, mother, grandparents or siblings) of obesity? YES NO 6. Are you currently a smoker? a. No b. I quit less than 6 months ago c. Yes

PAGE 120

106 APPENDIX D Baseline, 4 week, and 12 week Measures Basic Psychological Needs in Exercise (BPNES) The following sentences refer to your overall experiences in exercise as opposed to any particular situation. Using the 1 5 scale below, please indicate the extent to which you agree with these statements by choosing one number f or each statement. I don't agree at all I agree a little bit I somewhat agree I agree a lot I completely agree 1 2 3 4 5 1. I feel I have made a lot of progress in relation to the goal I want to achieve. 1 2 3 4 5 2. The way I exercise is in agreement with my choices and interests. 1 2 3 4 5 3. I feel I perform successfully the activities of my exercise program. 1 2 3 4 5 4. My relationships with the people I exercise with are very friendly. 1 2 3 4 5 5. I feel that the way I exercise is the way I want to. 1 2 3 4 5 6. I feel exercise is an activity which I do very well. 1 2 3 4 5 7. I feel I have excellent communication with the people I exercise with. 1 2 3 4 5 8. I feel that the way I exercise is a true expression of who I am. 1 2 3 4 5 9. I am able to meet the requirements of my exercise program. 1 2 3 4 5 10. My relationships with the people I exercise with are close. 1 2 3 4 5 11. I feel that I have the opportunity to make choices with regard to the way I exercise. 1 2 3 4 5

PAGE 121

107 Behavioral Regulations in Exercise Questionnaire 2 nd Version ( BREQ 2 ) W hy do you engage in exercise? We are interested in the reasons underlying peoples' decisions to engage, or not engage in physical exercise. Using the scale below, please indicate to what extent each of the following it ems is true for you. Please note that there are no right or wrong answers and no trick questions. We simply want to know how you personally feel about exercise. Your responses will be held in confidence and only used for our research purposes. Not true f or me Sometimes true for me Very true for me 0 1 2 3 4 1. I exercise because other people say I should. 0 1 2 3 4 2. I feel guilty when I don't exercise. 0 1 2 3 4 3. I value the benefits of exercise. 0 1 2 3 4 4. I exercise because it's fun. 0 1 2 3 4 5. Exercise gives me more energy to do the things that really matter to me in life. 0 1 2 3 4 6. I exercise because it is consistent with my life goals. 0 1 2 3 4 7. I don't see why I should have to exercise. 0 1 2 3 4 8. I take part in exercise because my friends/family/partner say I should. 0 1 2 3 4 9. I feel ashamed when I miss an exercise session. 0 1 2 3 4 10. It's important to me to exercise regularly. 0 1 2 3 4 11. I can't see why I should bother exercising. 0 1 2 3 4 12. I consider exercise to be part of my identity. 0 1 2 3 4 13. I enjoy my exercise sessions. 0 1 2 3 4 14. Engaging in regular exercise helps me reach my life goals. 0 1 2 3 4 15. I exercise because others will not be pleased with me if I don't. 0 1 2 3 4 16. I don't see the point in exercising. 0 1 2 3 4 17. I feel like a failure when I haven't exercised in a while. 0 1 2 3 4 18. I consider exercise to be a fundamental part of who I am. 0 1 2 3 4 19. I think it is important to make the effort to exercise regularly. 0 1 2 3 4 20. I find exercise a pleasurable activity. 0 1 2 3 4 21. I feel under pressure from my friends/family to exercise. 0 1 2 3 4 22. I get restless if I don't exercise regularly. 0 1 2 3 4 23. I get pleasure and satisfaction from participating in exercise. 0 1 2 3 4

PAGE 122

108 24. I think exercise is a waste of time. 0 1 2 3 4 25. I consider exercise to be consistent with my values. 0 1 2 3 4 *These questions were added to assess the connection between M/P and physical activity goals. Meaning in Life Questionnaire ( MILQ ) Please take a moment to think about what makes your life feel important to you. Please respond to the following statements as truthfully and accurately as you can, and also please remember that these are very subjective questions and that there are no right or wrong answers. Please answer acco rding to the scale below: Absolutely untrue Mostly untrue Somewhat untrue Can't Say True or False Somewhat true Mostly true Absolutely true 1 2 3 4 5 6 7 1. I understand my life's meaning. 1 2 3 4 5 6 7 2. I am looking for something that makes my life feel meaningful. 1 2 3 4 5 6 7 3. I am always looking to find my life's purpose. 1 2 3 4 5 6 7 4. My life has a clear sense of purpose. 1 2 3 4 5 6 7 5. I have a good sense of what makes my life meaningful. 1 2 3 4 5 6 7 6. I have discovered a satisfying life purpose. 1 2 3 4 5 6 7 7. I am always searching for something that makes my life feel significant. 1 2 3 4 5 6 7 8. I am seeking a purpose or mission for my life. 1 2 3 4 5 6 7 9. My life has no clear purpose. 1 2 3 4 5 6 7 10. I am searching for meaning in life. 1 2 3 4 5 6 7

PAGE 123

109 L ife Engagement Test (LET) Please answer the following questions about yourself by indicating the extent of your agreement using the following scale: 1 = strongly disagree; 2 = disagree; 3 = neutral; 4 = agree; 5 = strongly agree. Be as honest a s you can throughout, and try not to let your response to one question influence your response to other questions. There are no right or wrong answers. Strongly disagree Disagree Neutral Agree Strongly agree 1. There is not enough purpose in my life. 1 2 3 4 5 2. To me, the things I do are all worthwhile. 1 2 3 4 5 3. Most of what I do seems trivial and unimportant to me. 1 2 3 4 5 4. I value my activities a lot. 1 2 3 4 5 5. I don't care very much about the things I do. 1 2 3 4 5 6. I have lots of reasons for living. 1 2 3 4 5 Physical Activity Program Participation Why are you starting a physical activity program? (Baseline) Why are you continuing to engage in a physical activity program? (4 weeks & 12 weeks) International Physical Activity Questionnaire 7 Day Recall We are interested in finding out about the kinds of physical activities that people do as part of their everyday lives. The questions will ask you about the time you spent being physically active in the last 7 days Please answer each ques tion even if you do not consider yourself to be an active person. Please think about the activities you do at work, as part of your house and yard work, to get from place to place, and in your spare time for recreation, exercise or sport. Think about all the vigorous activities that you did in the last 7 days Vigorous physical activities refer to activities that take hard physical effort and make you breathe much harder than normal. Think only about those physical activities that you did for at least 1 0 minutes at a time.

PAGE 124

110 1. During the last 7 days on how many days did you do vigorous physical activities like heavy lifting, digging, aerobics, or fast bicycling? _____ days per week No vigorous physical activities Skip to question 3 2. How much time did you usually spend doing vigorous physical activities on one of those days? _____ hours per day _____ minutes per day Don't know/Not sure Think about all the moderate activities that you did in the last 7 days Moderate activities refer to activities that take moderate physical effort and make you breathe somewhat harder than normal. Think only about those physical activities that you did for at least 10 minutes at a time. 3. During the last 7 days on how many days did yo u do moderate physical activities like carrying light loads, bicycling at a regular pace, or doubles tennis? Do not include walking. _____ days per week No moderate physical activities Skip to question 5 4. How much time did you usually spend doing m oderate physical activities on one of those days? _____ hours per day _____ minutes per day Don't know/Not sure

PAGE 125

111 Think about the time you spent walking in the last 7 days This includes at work and at home, walking to travel from place to place, and any other walking that you have done solely for recreation, sport, exercise, or leisure. 5. During the last 7 days on how many days did you walk for at least 10 minutes at a time? _____ days per week No walking Skip to question 7 6. How much time did you usually spend walking on one of those days? _____ hours per day _____ minutes per day Don't know/Not sure The last question is about the time you spent sitting on weekdays during the last 7 days Include time spent at work, at home, while doing course work and during leisure time. This may include time spent sitting at a desk, visiting friends, reading, or sitting or lying down to watch television. 7. During the last 7 days how much time did you spend sitting on a week day ? _____ hours per day _____ minutes per day Don't know/Not sure

PAGE 126

112 Satisfaction with Life Scale (SWLS) Below are five statements with which you may agree or disagree. Using the 1 7 scale below, indicate your agreement with each item by placing the appropriate number in the line preceding that item. Please be open and honest in your responding. Strongly disagree Disagree Slightly disagree Neither agree or disagree Slightly agree Agree Strongly agree 1 2 3 4 5 6 7 1. In most ways my life is close to my ideal. 1 2 3 4 5 6 7 2. The conditions of my life are excellent. 1 2 3 4 5 6 7 3. I am satisfied with life. 1 2 3 4 5 6 7 4. So far I have gotten the important things I want in life. 1 2 3 4 5 6 7 5. If I could live my life over, I would change almost nothing. 1 2 3 4 5 6 7 Subjective Vitality Scale ( SV S ) Please respond to each of the following statements by indicating the degree to which the statement is true for you in general in your life. Use the following scale: Not at All True Somewhat True Very True 1 2 3 4 5 6 7 1. I feel alive and vital. 1 2 3 4 5 6 7 2. I don't feel very energetic. 1 2 3 4 5 6 7 3. Sometimes I feel so alive I just want to burst. 1 2 3 4 5 6 7 4. I have energy and spirit. 1 2 3 4 5 6 7 5. I look forward to each new day. 1 2 3 4 5 6 7 6. I nearly always feel alert and awake. 1 2 3 4 5 6 7 7. I feel energized. 1 2 3 4 5 6 7

PAGE 127

113 Patient Health Questionnaire 8 ( PHQ 8 ) Over the past two weeks, how often have you been bothered by any of the following problems? Not at all Several days More than half the days Nearly every day 1. Little interest or pleasure in doing things 0 1 2 3 2. Feeling down, depressed, or hopeless 0 1 2 3 3. Trouble falling or staying asleep, or sleeping too much 0 1 2 3 4. Feeling tired or having little energy 0 1 2 3 5. Poor appetite or overeating 0 1 2 3 6. Feeling bad about yourself or that you are a failure or have let yourself or your family down 0 1 2 3 7. Trouble concentrating on things, such as reading the newspaper or watching television 0 1 2 3 8. Moving or speaking so slowly that other people could have noticed? Or the opposite being so fidgety that you have been moving around a lot more than usual? 0 1 2 3 If you checked of any problems, how difficult have these problems made it for you to do your work, take care of things at home, or get along with other people? Not difficult at all Somewhat difficult Very difficult Extremely difficult

PAGE 128

114 APPENDIX E Baseline and 12 week Measures Fitness Exam Weight _______lbs Height _______inches Body Mass Index _______kg/m 2 Waist Circumference _______inches Body Fat Percent _______% Resting Heart Rate _______bpm YMCA 3 min Step Test Recovery Heart Rate _______bpm Sit and reach _______cm Grip strength (Hand Dynamometer) _______lbs

PAGE 129

115 APPENDIX F Daily Measures Thoughts of Meaning Scale (TOMS) How meaningful does your life feel right now? Not at All Quite a bit 1 2 3 4 5 6 7 How much do you feel your life has purpose right now? Not at All Quite a bit 1 2 3 4 5 6 7 How much have you thought about what makes your life meaningful today? Not at All Quite a bit 1 2 3 4 5 6 7 How much have you thought about your purpose in life today? Not at All Quite a bit 1 2 3 4 5 6 7 My activities today were consistent with my life goals and purpose in life. Not at All Quite a bit 1 2 3 4 5 6 7 Today, I have thought about my reasons for living. Not at All Quite a bit 1 2 3 4 5 6 7 The activities I did today were valuable and worthwhile. Not at All Absolutely 1 2 3 4 5 6 7

PAGE 130

116 When choosing my activities for today, I did so with my purpose in life in mind. Not at All Absolutely 1 2 3 4 5 6 7 My activities today supported what is meaningful in my life. Not at All Absolutely 1 2 3 4 5 6 7 The decisions I made today were based on what is valuable and meaningful to me. Not at All Absolutely 1 2 3 4 5 6 7 Affect Very Slightly or Not at All Extremely 1 2 3 4 5 1. Relaxed 2. Proud 3. Excited 4. Appreciative 5. Enthusiastic 6. Happy 7. Satisfied 8. Curious 9. Sluggish 10. Afraid 11. Sad 12. Anxious 13. Angry Physical Activity How many minutes of physical activity did you do today? _____

PAGE 131

117 What activities did you do? Check all that apply. Walking Jogging or Running Biking or cycling Elliptical Group fitness class Step aerobics Stair climbing Yoga Weight Lifting Strength based exercises (e.g., push ups, abdominal exercises, pull ups) Dancing Hiking/Backpacking/Mountain Climbing Canoein g or Rowing Swimming Sport (Basketball, Volleyball, Soccer, Tennis, etc.) Gardening/Lawn work/House work Other__________________ What was your average level of exertion during your activities? 9 corresponds to "very light" exercise. For a healthy person, it is like walking slowly at his or her own place from some minutes. 13 on the scale is "somewhat hard" exercise, but it still feels OK to continue. 17 "very hard" is very strenuous. A healthy person can still go on, but he or she really has to push him or herself. It feels very heavy, and the person is very tired. 19 on the scale is an extremely strenuous exercise level. For most people this is the most strenuous exercise they have ever experienced. 6 No exertion at all 7 Extremely light (7.5) 8 9 Very light 10 11 Light 12 13 Somewhat hard

PAGE 132

118 14 15 Hard (heavy) 16 17 Very hard 18 19 Extremely hard 20 Maximal exertion