Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Running to quit? : exploring predictors of attendance in an exercise and smoking cessation intervention Wunderlich, Kelly 2020

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
24-ubc_2020_may_wunderlich_kelly.pdf [ 982.87kB ]
Metadata
JSON: 24-1.0389777.json
JSON-LD: 24-1.0389777-ld.json
RDF/XML (Pretty): 24-1.0389777-rdf.xml
RDF/JSON: 24-1.0389777-rdf.json
Turtle: 24-1.0389777-turtle.txt
N-Triples: 24-1.0389777-rdf-ntriples.txt
Original Record: 24-1.0389777-source.json
Full Text
24-1.0389777-fulltext.txt
Citation
24-1.0389777.ris

Full Text

RUNNING TO QUIT? EXPLORING PREDICTORS OF ATTENDANCE IN AN EXERCISE AND SMOKING CESSATION INTERVENTION  by  Kelly Wunderlich  B.Kin., The University of British Columbia, 2017  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF SCIENCE in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Kinesiology)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  April 2020  © Kelly Wunderlich, 2020 ii  The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, a thesis/dissertation entitled:  Running to quit? Exploring predictors of attendance in an exercise and smoking cessation intervention  submitted by Kelly Wunderlich in partial fulfillment of the requirements for the degree of Master of Science in Kinesiology  Examining Committee: Dr. Guy Faulkner Supervisor  Dr. Carly Priebe Supervisory Committee Member  Dr. Mark Beauchamp Supervisory Committee Member iii  Abstract Run to Quit (RTQ) is a national smoking cessation and learn to run program with promising cessation and physical activity outcomes. However, attrition was high with only 41.1% of participants completing the program. Determining predictors of attendance could help to improve attendance and program effectiveness in future iterations. Given that the program was offered in a group setting, the purposes of this study were to explore predictors of attendance and examine whether including group-related variables added to the prediction of attendance beyond individual variables.  Blocked multiple regression analysis was used, with mean substitution for missing data (n=335). Individual predictors included in block 1 were middle aged adults, older adults, gender, home ownership, quit self-efficacy, run self-efficacy, baseline nicotine dependence (FTND), and baseline moderate-vigorous physical activity (MVPA). Group-related predictors added in block 2 were group cohesion subscales (attraction to group-task, attraction to group-social (ATG-S), group integration-task, group integration-social), transformational leadership (TL) of the coach, belonging, perceived similarity, and group size (control variable).  When only individual predictors were included, the model was statistically significant and explained 4.8% of the variance in attendance (adjusted R2=.048, F(8,326)=3.111, p=.002). Both baseline MVPA (b=-.135, p=.013) and FTND (b=-.135, p=.015) were statistically significant predictors of attendance.  Once group-related predictors were added, the overall model identified additional individual and group-related predictors of attendance. Both individual and group predictors were significant and adding group-related variables explained an additional 4.2% of the variance. Overall, the final model explained 9.0% of the variance (adjusted R2=.090, F(16,318)=3.067, p<.001), with being an older adult (b=.140, p=.023), male iv  (b=-.118, p=.032), having lower FTND (b=-.145, p=.009), lower MVPA (b=-.120, p=.025), higher ATG-S (b=.189, p=.011), higher belonging (b=.183, p=.006), and lower TL (b=-.160, p=.018) significantly predicting higher attendance.  This evaluation identified both individual and group-related predictors of attendance, however, the model explained a modest amount of variance suggesting that there are additional factors that predict attendance, such as logistical and personal reasons, and further exploration is needed. Future RTQ programs may benefit from promoting group-related aspects such as cohesion and sense of belonging. v  Lay Summary Run to Quit is a group-based program to help participants stop smoking and be more physically active. The program seems to meet these goals, but only 4 in 10 participants complete it. This study aimed to understand why some participants attended more sessions than others. For example, whether individual characteristics (e.g., age and gender) or things about the groups (e.g., feeling attracted to the group) affected attendance. We found that both individual characteristics and group-related aspects of the program affected attendance. Participants were more likely to attend when they were older, male, less physically active and less addicted to nicotine at the start of the program, more attracted to the social aspect of the group, and felt like they belonged. More research is needed to explain the role of the coach and to identify what else affects attendance in this program. vi  Preface This thesis is original, unpublished work by the author, Kelly Wunderlich, who was responsible for all major areas of concept formation, data analysis, and preparation of this manuscript. The identification and design of this research project were completed with assistance from the Supervisory Author, Dr. Guy Faulkner, and committee members, Drs. Carly Priebe and Mark Beauchamp.  The data presented in this thesis came from an evaluation of Run to Quit, a collaboration between The University of British Columbia, Running Room, the Canadian Cancer Society, and funded by the Public Health Agency of Canada. This thesis was a secondary analysis of data collected for an evaluation designed by Drs. Faulkner and Priebe. Data collection for RTQ was performed by trained evaluation assistants at each location, and data were entered by the author or Research Assistants in the Population Physical Activity Lab. Data were cleaned, analyzed, and interpreted by the author. No publications have been submitted to date based on this thesis.  Ethical approval for this research was obtained from The University of British Columbia’s Behavioural Research Ethics Board (#H16-00048).      vii  Table of Contents  Abstract ......................................................................................................................................... iii Lay Summary .................................................................................................................................v Preface ........................................................................................................................................... vi Table of Contents ........................................................................................................................ vii List of Tables ..................................................................................................................................x List of Figures ............................................................................................................................... xi List of Abbreviations .................................................................................................................. xii Acknowledgements .................................................................................................................... xiv Dedication .....................................................................................................................................xv Chapter 1: Introduction ................................................................................................................1 1.1 The burden of tobacco smoking ...................................................................................... 1 1.2 The health benefits of smoking cessation ....................................................................... 2 1.3 The challenge of smoking cessation ............................................................................... 3 1.4 Current smoking cessation recommendations and strategies .......................................... 3 1.5 Run to Quit: A group-based smoking cessation and physical activity program ............. 5 1.6 Individual predictors of attendance ................................................................................. 7 1.7 Group smoking cessation programs ................................................................................ 9 1.8 Group dynamics in Run to Quit .................................................................................... 11 Chapter 2: Methods .....................................................................................................................18 2.1 Research design ............................................................................................................ 18 2.2 Participant characteristics ............................................................................................. 18 viii  2.2.1 Sampling procedures ............................................................................................. 19 2.3 Sample size, power, and precision ................................................................................ 19 2.4 Measures and covariates ............................................................................................... 20 2.4.1 Primary outcome: Attendance ............................................................................... 21 2.4.2 Individual predictors ............................................................................................. 22 2.4.2.1 Age .................................................................................................................... 22 2.4.2.2 Week 1 MVPA .................................................................................................. 22 2.4.2.3 Week 1 nicotine dependence ............................................................................ 23 2.4.2.4 Self-efficacy to quit ........................................................................................... 24 2.4.2.5 Self-efficacy to run ........................................................................................... 24 2.4.2.6 Gender ............................................................................................................... 24 2.4.2.7 Home ownership ............................................................................................... 24 2.4.3 Group-related predictors ....................................................................................... 25 2.4.3.1 Transformational leadership (TL) ..................................................................... 25 2.4.3.2 Group cohesion ................................................................................................. 25 2.4.3.3 Perceived similarity .......................................................................................... 26 2.4.3.4 Belonging .......................................................................................................... 27 2.4.4 Group Size ............................................................................................................ 27 2.5 Analysis......................................................................................................................... 28 2.5.1 Data exploration .................................................................................................... 28 2.5.1.1 Data visualization and correlations ................................................................... 28 2.5.1.2 Missing data ...................................................................................................... 28 2.5.2 Intra-class correlation coefficient (ICC) ............................................................... 29 ix  2.5.3 Blocked ordinary least squares (OLS) multiple regression .................................. 29 Chapter 3: Results ........................................................................................................................31 3.1 Participant flow ............................................................................................................. 31 3.2 Statistics and data analysis ............................................................................................ 33 Chapter 4: Discussion ..................................................................................................................44 4.1 Implications ................................................................................................................... 56 4.2 Limitations .................................................................................................................... 59 4.3 Strengths ....................................................................................................................... 61 Chapter 5: Conclusion .................................................................................................................63 References ......................................................................................................................................64  x  List of Tables  Table 1. Power Calculations for Models Using G*Power ............................................................ 20 Table 2. Dummy Coding for Age Variables ................................................................................. 22 Table 3. Comparison of Included and Excluded Participants’ Complete Case Data .................... 32 Table 4. Complete Case and Mean Imputed Pearson Correlations for TTQ Subscales to Investigate Potential Multicollinearity .......................................................................................... 33 Table 5. Missing Data Summary for Complete Case and Mean Imputed Predictors ................... 35 Table 6. Descriptive Statistics for Variables in Complete Case and Mean Imputed Models ....... 37 Table 7. Complete Case (n = 223) Pearson Correlation Matrix for Predictors and Attendance ... 38 Table 8. Mean Imputed Pearson Correlation Matrix for Predictors and Attendance ................... 39 Table 9. Coefficients for Complete Case Model Predicting Attendance ...................................... 41 Table 10. Regression Coefficients for Mean Imputed Model Predicting Attendance .................. 42  xi  List of Figures  Figure 1. Distribution of Attendance Scores for Included Participants (n = 339) ........................ 21 Figure 2. Flowchart of Participants Included in Evaluation ......................................................... 31  xii  List of Abbreviations  Abbreviation Term ATG-S Individual attraction to group - social ATG-T Individual attraction to group - task BMI Body mass index CAN-ADAPTT  Canadian Action Network for the Advancement, Dissemination and Adoption of Practice-Informed Tobacco Treatment FTND Fagerstrom Test for Nicotine Dependence GI-S Group integration – social GI-T Group integration – task IC Individualized Consideration ICC Intra-class Correlation Coefficient II Idealized Influence IM Inspirational Motivation IS Intellectual Stimulation ITT Intent to treat MCAR Missing Completely At Random MVPA Moderate-vigorous physical activity NRT Nicotine Replacement Therapy OLS Ordinary Least Squares PAGEQ Physical Activity Group Environment Questionnaire xiii  RTQ Run to Quit TL Transformational Leadership TTQ Transformational Teaching Questionnaire    xiv  Acknowledgements A sincere thank you to my supervisor, Dr. Faulkner, who has provided abundant opportunities for learning and experiencing research since joining the Population Physical Activity Lab. Your encouragement to get involved in different projects, collaborate, apply for funding and awards, and to just try things have been invaluable. Thank you for sharing your insight and being available for feedback and brainstorming meetings even when you’re busy, and for writing retreats to help finish this thesis. Much appreciation goes to my committee members. Thank you to Dr. Carly Priebe, for always being a phone call or an email away and for providing many hours of feedback on this project. It is always a pleasure to work with you and learn from your experience. Thank you to Dr. Mark Beauchamp, for many helpful suggestions, and for your work that has contributed to the field of group dynamics that I referenced heavily for this thesis. Your insight and ability to push my understanding has been greatly appreciated. A big thank you to my colleagues in the Population Physical Activity Lab: Negin, Krista, Mahabhir, Madelaine, Jackie, Matt, Miya, Mark, Lira, and the Katies. Working with you all has been a pleasure, and you have all provided invaluable support, conversations, and helped to create many fond memories. As always, thank you to my family, particularly my parents for many phone calls, walks, and relaxing trips home, and to my grandparents for our dinner catch ups and games of Rummoli and Uno. Many thanks to Melissa, Cristina, Addy, and Lauren for company during coffee shop sessions, and to my many ultimate teammates and friends who help to keep me healthy. This thesis was supported by the Canadian Institute for Health Research (CIHR). xv  Dedication For CK, whose genuine interest in learning and those around him made for engaging and memorable conversations. Thank you for believing in me and giving me many opportunities to take a break from my thesis while still being productive and contributing to our community. The lessons I learned from you have been invaluable and you are dearly missed as a friend and mentor.1  Chapter 1: Introduction  1.1 The burden of tobacco smoking Smoking cigarettes has been connected to a variety of negative health outcomes, including lung cancer (International Agency for Research on Cancer, 2012; Lee, Forey, & Coombs, 2012; U.S. Department of Health and Human Services, 2014), and cancers of the oral cavity, pharynx, stomach, pancreas, and liver (International Agency for Research on Cancer, 2012). Additionally, smoking has been linked to chronic obstructive pulmonary disease (Forey, Thornton, & Lee, 2011; U.S. Department of Health and Human Services, 2014), chronic bronchitis, and emphysema (Forey et al., 2011). There is a risk not only to smokers, but those who are regularly exposed to second-hand smoke, i.e., “passive” smoking (Johansson, Halling, & Hermansson, 2003; U.S. Department of Health and Human Services, 2014).  Despite these poor health outcomes, smoking continues to be prevalent in Canada. A 2017 survey found that 4.6 million Canadians (15%) smoked cigarettes, a greater number than in 2015 (Government of Canada, 2019). Of those who reported smoking, 3.3 million (11%) were daily smokers, and 1.3 million (4%) were occasional smokers. In 2012, an estimated 18.4% (45,464) of total deaths in Canada were attributed to smoking and second-hand smoke exposure, with 599,390 potential years of life lost (Dobrescu, Bhandari, Sutherland, & Dinh, 2017). It has been estimated that the direct and indirect costs from tobacco use cost Canadians $16.2 billion in 2012. Further reduction of smoking behaviour is necessary in order to curb the burden of smoking, and its far-reaching effects, for all Canadians.  2  1.2 The health benefits of smoking cessation  In light of the detrimental effects of smoking, cessation can offer numerous physical (Godtfredsen & Prescott, 2011) and mental health benefits (Taylor et al., 2014). Physical benefits related to cardiovascular and respiratory health include a reduced risk of mortality for smokers who quit after a myocardial infarction, reduced risk of stroke, abdominal aortic aneurism and risk of rupture, a decreased loss of lung function and improvements in respiratory functions for smokers with mild-moderate chronic obstructive pulmonary disease, and improvements related to asthma (Godtfredsen & Prescott, 2011). Further, a recent study estimated that lowering active tobacco smoking in Canada could lead to the prevention of an estimated 41,191-50,696 cases of cancer by 2042, and 730-3640 cancer cases for second-hand smokers (Poirier et al., 2019). In terms of mental health, a systematic review and meta-analysis that compared mental health outcomes before and after cessation found significant decreases in anxiety scores, depression, stress, and significant increases in psychological quality of life and positive affect after quitting (Taylor et al., 2014). The meta-analysis included studies with smokers from the general population and some clinical populations, with no evidence for different effect sizes between them. These results suggest that there is a wealth of evidence supporting the mental health benefits of smoking cessation for a variety of people.   Following these mental and physical health benefits from smoking cessation, a positive impact on lifespan seems intuitive. This was explored using data from the Cancer Prevention Study II, a prospective cohort study of 1.2 million US adults (Taylor, Hasselblad, Henley, Thun, & Sloan, 2002). The results suggest that quitting smoking at any age can increase longevity, with the biggest increases seen when smoking cessation occurs before 35 years of age. Together, these results emphasize the numerous health benefits of smoking cessation. 3  1.3 The challenge of smoking cessation  Despite the clear benefits of cessation, many people who smoke struggle to stop. Based on data from the National Health Interview Surveys, 68.8% of adult smokers in the USA wanted to quit smoking in 2010, and over half had made a quit attempt in the past year (Centers for Disease Control and Prevention, 2011). Despite this interest in cessation, only about 6% had reported recently quitting. However, nearly 70% of those smokers who tried to quit did not use evidence-based cessation aids such as counselling or medications (Centers for Disease Control & Prevention, 2011). These findings suggest that smokers are interested in cessation, but that it takes many attempts. Better access to, or increased awareness of, evidence-based strategies could lead to greater success for those aiming to stop smoking.  1.4 Current smoking cessation recommendations and strategies  To help combat challenges with cessation, the Canadian Action Network for the Advancement, Dissemination and Adoption of Practice-informed Tobacco Treatment (CAN-ADAPTT) smoking cessation clinical practice guidelines highlighted the importance of counselling and psychosocial approaches, including using both counselling and smoking cessation medication in tandem due to the combination being more effective than either alone (CAN-ADAPTT, 2011). This recommendation was based on high quality evidence and deemed by the authors unlikely to change based on future research.  Similarly, smoking cessation guidelines for health professionals included taking all opportunities to connect behavioural support to smokers through individual or group therapy and promoting the use of Nicotine Replacement Therapy (NRT) when appropriate (West, McNeill, & Raw, 2000). Specifically, smokers who consume ³10 cigarettes each day were identified as the key target for NRT or buproprion. Buproprion was originally an anti-depressant medication that 4  was found to also help reduce the desire to smoke (West, 2003). The combination of pharmacotherapy and behavioural treatment was also supported by a recent Cochrane review of combined pharmacotherapy and behavioural interventions for smoking cessation compared to usual care, brief advice, or less intensive behavioural support (Stead, Koilpillai, Fanshawe, & Lancaster, 2016). Although there is strong evidence for NRT and behavioural approaches compared to unaided cessation (Garcia-Rodriguez et al., 2013; Hughes, Peters, & Naud, 2008), there is also a need for a variety of cessation options as there is no “one size fits all” solution (World Health Organization, 2004). There is a need to provide smoking cessation strategies that appeal to a variety of people who smoke. A more recent approach to smoking cessation has been through physical activity, which has the potential to help manage withdrawal symptoms and cravings (Haasova et al., 2014; Roberts, Maddison, Simpson, Bullen, & Prapavessis, 2012). A Cochrane review examined whether exercise-based programs, or cessation programs combined with exercise, were more effective than a smoking cessation intervention without an exercise component (Ussher, Faulkner, Angus, Hartmann-Boyce, & Taylor, 2019). The authors concluded that there is not enough evidence to recommend exercise alone as a smoking cessation strategy, but that this is of low certainty and future research could change these findings. However, there is a need for further research with larger samples, appropriate intensity exercise interventions, increased exercise adherence, and device assessed physical activity (Ussher et al., 2019).  There is potential for exercise to reduce withdrawal symptoms and cravings, and for it to work in tandem with other cessation strategies. Potential mechanisms include physical activity playing a moderating role between executive function and smoking or by mediating the relationship between smoking and execution function by impacting smoking behaviour (Loprinzi 5  & Walker, 2015). Exercise may also help with cravings by reducing attentional bias to cues that could trigger them (Oh & Taylor, 2014; Van Rensburg, Taylor, & Hodgson, 2009). Additionally, there is no evidence of detrimental effects on smokers who engage in physical activity (Ussher et al., 2019; Ussher, Taylor, & Faulkner, 2014). Integrating physical activity into smoking cessation programs may help to provide additional strategies that appeal to people who are currently smoking. 1.5 Run to Quit: A group-based smoking cessation and physical activity program Run to Quit (RTQ) is a Canada-wide group-based smoking cessation and physical activity intervention (Priebe, Atkinson, & Faulkner, 2016). Over 700 participants took part in group clinics across Canada between 2016 and 2018. Participants registered at a local Running Room store and followed a 10-week learn to run 5km program. Each week the group’s coach led a smoking cessation component, informed by the Canadian Cancer Society’s One Step at a Time booklet (Canadian Cancer Society, 2013), an evidence-based guide for smokers who want to quit, and a walking or running training session (Priebe et al., 2016). This design supported the two health behaviour goals of the program: smoking cessation and increased physical activity levels. The evaluation of RTQ suggested that these goals were achieved. Over the 3 years of evaluation, 55.0% of participants who completed the program (22.1% intent to treat [ITT]) had 7-day continuous abstinence from smoking at the end of program (Priebe, Wunderlich, Atkinson, & Faulkner, 2020). Regardless of whether participants had achieved cessation, significant increases were seen in both self-reported run frequency and moderate-vigorous physical activity compared to baseline in those who completed the program. At six-month follow up 42% of participants were interviewed, and 47.5% of them self-reported no longer smoking (18.8% ITT; 6  Priebe et al., under review). Additionally, ITT measures of smoking level (i.e., carbon monoxide) at week 10 were negatively correlated with attendance (r = -.304, p <.001), suggesting that greater attendance was associated with better smoking cessation outcomes.  These results suggest that RTQ aided participants’ smoking cessation and physical activity goals, and that it promoted long-term abstinence. However, the effectiveness may have been limited by attendance. Only 41.1% of the participants completed the program and there was a relationship between attendance and positive smoking and physical activity outcomes. Lower attendance means less exposure to the curriculum of RTQ that could aid participants with quitting smoking and increasing their physical activity, suggesting that attending fewer sessions would lead to less success for participants. This is supported by previous work where attending more sessions of a smoking cessation or counselling program was related to better quit outcomes (Dorner, Tröstl, Womastek, & Groman, 2011; Iliceto, Fino, Pasquariello, D’Angelo Di Paola, & Enea, 2013; Jacquart et al., 2017; Wenig, Erfurt, Kröger, & Nowak, 2013). There appears to be an important link between attending smoking cessation program sessions and successfully quitting smoking. Despite this connection, a systematic review of factors that influence attrition (defined as withdrawal from a study) in cessation interventions found attrition rates ranged between 10.8% and 77% (Belita & Sidani, 2015). This demonstrates the low completion rates of some studies, and the need for more research on how to retain participants. There have also been calls for more attrition analyses from large, representative community samples, rather than trials with specialized samples (Leeman et al., 2006), and an expansion of focus beyond characteristics of individual participants that predict attrition in smoking cessation interventions (Belita & Sidani, 2015). Using evidence-based strategies as an intervention, such as counseling or a form of NRT, 7  makes smoking cessation more likely (Hartmann-Boyce, Chepkin, Ye, Bullen, & Lancaster, 2019; Stead, Carroll, & Lancaster, 2017; Stead et al., 2016). However, as seen in RTQ, just starting or initially participating in a cessation program does not guarantee success. Determining what encourages participants to attend more RTQ sessions could allow for modifications that would optimize the program. 1.6 Individual predictors of attendance  To date, it appears that most studies investigating factors that affect pharmacological, educational, or behavioural intervention attrition have focused on individual features (Belita & Sidani, 2015). For example, younger participants (Ahluwalia et al., 2002; Geraghty, Torres, Leykin, Perez-Stable, & Munoz, 2012; Leeman et al., 2006; Woods et al., 2002) and those with higher levels of nicotine dependence were less likely to complete interventions (Brouwer & Pomerleau, 2000; Copeland, Martin, Geiselman, Rash, & Kendzor, 2006; Geraghty et al., 2012). These results are similar to a review of factors related to successfully quitting smoking or making a quit attempt (Vangeli, Stapleton, Smit, Borland, & West, 2011). Participants with higher nicotine dependence were less likely to make a quit attempt and less likely to successfully quit once a quit attempt was made. In terms of age, the review found that either no relationship or that older participants were more likely to have a successful quit attempt. This suggests that while age may not be a consistent significant predictor of successfully quitting in the general population, when it is significant older participants are more likely to quit. Together, these findings suggest that both being older and having lower nicotine dependence are related to smoking cessation and attending interventions. In the general population, no association between gender and making a quit attempt, or making a successful quit attempt, were found in the pooled analysis (Vangeli et al., 2011). 8  However, there was some evidence of women being more likely to make a quit attempt (Zhou et al., 2009), and also that women were less likely to have successfully quit (Fidler & West, 2011). In a review of attrition in smoking cessation interventions (Belita & Sidani, 2015), there was some evidence that males were more likely to withdraw before the program began (Ahluwalia et al., 2002; Geraghty et al., 2012; Woods et al., 2002) while women dropped out after the program started (Curtin, Brown, & Sales, 2000). Due to the conflicting results on sex/gender in the smoking cessation literature, a recent review was conducted that included efficacy and effectiveness interventions, as well as prospective observational studies. The authors concluded that women were less likely than men to maintain long-term abstinence during a quit attempt (Smith, Bessette, Weinberger, Sheffer, & McKee, 2016). Overall, there appears to be some evidence that men are more likely to stay in a program after joining (Belita & Sidani, 2015), and that men may be better at maintaining a long-term quit attempt (Smith et al., 2016). Additionally, self-efficacy has been an important predictor of health behaviours such as smoking cessation and physical activity. Participants with higher levels of self-confidence in quitting were more likely to complete an intervention study (Leeman et al., 2006; Nevid, Javier, & Moulton, 1996). This self-confidence is consistent with abstinence self-efficacy, “confidence in one’s ability to abstain from smoking,” (Gwaltney, Metrik, Kahler, & Shiffman, 2009, p. 56) that has been connected to cessation. A meta-analysis on self-efficacy and smoking cessation found that pre-quit self-efficacy was higher for those who were no longer smoking at follow up (Gwaltney et al., 2009), suggesting that confidence in one’s ability to quit is related to successfully quitting. Additionally, timing of self-efficacy with regards to cessation matters: self-efficacy measured before a quit attempt was less strongly related to cessation than when self-efficacy was measured after a quit attempt. This suggests that measuring self-efficacy after a quit 9  attempt could be a better predictor of smoking cessation. Similarly, self-efficacy has been a strong correlate of physical activity behaviour (Bauman et al., 2012), and increasing self-efficacy has been related to increased physical activity behaviour (Williams & French, 2011). Self-efficacy has been a consistent predictor of both smoking cessation and physical activity behaviours.  There has been some support for affluence impacting both smoking cessation, and attrition in interventions. A meta-analysis of studies on the general population found that although there was some evidence for higher affluence predicting a successful quit attempt, overall results were mixed (Vangeli et al., 2011). There was also some limited evidence that lower income was associated with higher attrition in smoking cessation interventions (Belita & Sidani, 2015; Nevid et al., 1996). Although it appears that being less affluent may play a role in smoking cessation outcomes and attrition, further research is needed to further investigate this relationship. Affluence, represented by home ownership, was found to be a significant predictor of attendance during the first three months of a community-based exercise program for older adults (Hawley-Hague et al., 2014). Taken together, age, sex/gender, nicotine dependence, self-efficacy to quit and do physical activity, and affluence appear to be individual factors that could predict program attendance. However, there may be additional factors influencing attendance in RTQ, such as the social environment of group interventions.  1.7 Group smoking cessation programs   Groups have long been used to implement health behaviour change programs and can play a key role in the success of an intervention. A Cochrane review on group behaviour therapy programs (i.e., interventions where participants meet regularly, with a facilitator who is typically trained in smoking cessation counselling) for smoking cessation found that group therapy was 10  more effective than self-help, or other less intense approaches (Stead, Carroll, & Lancaster, 2017). The authors suggested that using self-help material could help 5 of 100 people quit smoking for at least six months, while using group support could help 8-12 of 100 smokers successfully quit for at least 6 months (chance of quitting increased from 50% to 130%). Groups can offer social support, which has been found to help reduce multiple risk behaviours including smoking (Greaney et al., 2018). However, in group smoking cessation research there is further work to be done to understand the role of the group. It appears that most research on social support in smoking cessation literature has primarily focused on “buddies,” usually a partner or friend of some kind who agrees to support the smoker during their cessation attempt (Faseru, Richter, Scheuermann, & Park, 2018). Although these relationships may offer support during everyday life as well as during a program, they do not take into account the group environment that may encourage participants to adhere to or attend an intervention. Further, despite the prevalence of “buddies” in cessation intervention research, there has also been little evidence to support that having a buddy improves quit outcomes (Faseru et al., 2018).  Group smoking cessation programs may be recommended in part due to being “much more cost effective” (Raw, McNeill, & West, 1998, p. S10) rather than recommended for reasons related to the potential of group dynamics to promote success. This view of groups as merely a convenient vehicle for delivering a program misses potential opportunities for both intentionally utilizing the beneficial aspects of groups, as well as teasing apart whether group dynamics play a role in cessation outcomes.   11  1.8 Group dynamics in Run to Quit   The group setting of RTQ allowed for the group itself to help facilitate success. The purpose of the program was to create groups that were united by a goal of smoking cessation and learning to run. The groups were intended to be led by coaches who were encouraging, understanding, and had inspirational stories from their own smoking cessation and/or fitness journey. Further, the program was designed to help participants make new social connections and create a support network with the other group members. Through group conversations and activities, participants were supported as they navigated similar challenges while trying to quit smoking and learn to run a 5 km race. Group cohesion, perceptions of leaders and similarity to other participants have all been related to attendance in previous exercise programs (Beauchamp et al., 2018; Hawley-Hague et al., 2014; Izumi et al., 2015; Smith-Ray, Mama, Reese-Smith, Estabrooks, & Lee, 2012). Although not smoking cessation interventions, these results suggest that these group factors may also play a role in attendance during a smoking cessation and physical activity program.  The behaviour of the coach could be considered from the perspective of transformational leadership. Transformational leaders “stimulate and inspire followers to both achieve  extraordinary outcomes and, in the process, develop their own leadership capacity” (Bass & Riggio, 2006, p. 3).  This style of leadership was originally applied to work and education settings, with a more recent shift towards health promotion (Beauchamp et al., 2010). Transformational leaders may affect participants through inspiring trust and respect by role modeling ethical behaviour and acting in line with their beliefs, sharing expectations and motivating others to achieve their goals, promoting viewing situations from multiple perspectives, and identifying individual needs, caring for others, and celebrating successes (Bass 12  & Riggio, 2006; Beauchamp et al., 2010). It has been applied in a physical activity context, and found to be associated with athlete wellbeing (Stenling & Tafvelin, 2014), and adolescent self-determined motivation, self-efficacy, and intention to engage in leisure time physical activity  (Beauchamp, Barling, & Morton, 2011). Transformational leadership in the RTQ program and its impact on participants’ success has yet to be explored, however, based on the previous finding that the coaches are impactful in RTQ (Glowacki, O’Neill, Priebe, & Faulkner, 2018), this may be a potential predictor of attending more session of the program.  It is also possible that group cohesion promotes attending RTQ, as it has been previously related to attendance in an exercise setting (Dunlop & Beauchamp, 2011; Estabrooks & Carron, 2000). Cohesion, distinct from belonging and similarity, is “a dynamic process which is reflected in the tendency for a group to stick together and remain united in the pursuit of its goals and objectives” (Carron, 1982, p. 124). As outlined by Carron (1982), cohesion has both task and social components, as well as individual and group components. This is reflected in the measurement tool for cohesion in physical activity groups created by Estabrooks and Carron (2000) that includes four dimensions of cohesion: individual attractions to the group – task (ATG-T), individual attractions to the group – social (ATG-S), group integration – task (GI-T), and group integration – social (GI-S). Overall, cohesion is a construct that focuses on what makes a group remain together. Previous literature found that cohesion, specifically the individual attraction to group-task dimension, was predictive of adherence to an exercise program in older adults, and was also related to self-efficacy (Estabrooks & Carron, 2000). However, context can impact cohesion, with task dimensions better predicting adherence for university students, and social dimensions better predicting adherence in community exercise classes for older adults (Spink & Carron, 13  1994). GI-T and GI-S were also found to predict attendance at different time points in a year-long program for older adults, and that intentionally promoting cohesion in a group of older adult exercisers led to better attendance (Estabrooks & Carron, 1999). Overall, cohesion has a positive relationship with exercise adherence (Burke, Carron, & Shapcott, 2008), and the predictive ability of different dimensions of cohesion may depend on the context. The role of group cohesion in smoking cessation has received limited attention. One study that explored the effects found that a cohesion-enhanced group showed better smoking abstinence rates in the short term (Etringer, Gregory, & Lando, 1984). This study used three measures of cohesion to “assess different facets of cohesion by employing sociometric, self-report, and behavioral measures” (p. 1082). Multiple measures of cohesion were used due to the authors claiming that there was no universally accepted measure of cohesion, and all of these measures appeared to focus on the social aspects of cohesion. This is in contrast to Carron’s (1982) definition of cohesion that includes both task and social components. Additionally, Etringer and colleagues’ (1984) results were unable to be replicated when repeated with a larger sample size and biochemical validation of quit outcomes, as well as an analysis that attempted to account for group effects when they were identified (Lando & McGovern, 1991). Specifically, no differences in quit outcomes were observed between the enhanced cohesion at any follow up time point. Given these mixed findings, examining the impact of cohesion from a different theoretical perspective that encompasses task dimensions as well as social could help to better understand the role of cohesion in smoking cessation interventions, including its impact on attendance.  Another group-related variable, perceptual similarity, includes both surface- and deep-level characteristics. Surface-level qualities are more physical in nature and are easily observed, 14  such as age and gender (Dunlop & Beauchamp, 2011), while deep-level qualities require more interactions to discover, such as attitudes, beliefs, and values (Dunlop & Beauchamp, 2011; Harrison, Price, & Bell, 1998). Dunlop and Beauchamp (2011) examined the roles of both cohesion and similarity in group-based exercise programs. They found that task cohesion was better predicted by deep-level similarities, while social cohesion was predicted by surface-level similarities. Additionally, perceived surface-level similarities were a better predictor of attendance during the eight-week programs. These findings support that cohesion and perceived similarity are related and suggest that similarity also impacts attendance in an exercise context. Based on the evidence, similarity and cohesion may be key proponents of attendance in the Run to Quit program.  Belonging, positive interpersonal connections, has also been observed in adolescent activity programs. Belonging is distinct from cohesion and focuses on connectedness rather than participants being drawn to their groupmates and perceiving their groupmates as also being drawn to the social interaction of the group. For example, in a YMCA-based program that promoted social connections, high scores on a belonging measure were found at the end of the sessions (Dowd, Chen, Jung, & Beauchamp, 2015). The authors suggested that sense of belonging may have contributed to the success of the program (i.e., average attendance was 91% of sessions, measures of physical activity and healthy eating behaviour improved). In another adolescent physical activity program that measured belonging at the beginning and end of a summer camp, participants reported feeling like they belonged at both time points, and although the change in the score was not statistically significant; it was still positive and may have not have been sensitive enough to detect changes due to the limited response range (0-4) of the measure (Anderson-Butcher et al., 2013). Belongingness theory proposes that humans have a 15  drive to consistently experience positive affect while interacting with a select group, and that these interactions should be sustainable over time and involve reciprocated concern for each other (Baumeister & Leary, 1995). This theory has been employed in various studies. For example, it was used to decrease feelings of social rejection in a sample of lonely undergraduate students, and framing physical activity as a means to reduce feelings of social rejection may promote the adoption of physical activity behaviours (Dowd et al., 2014). Together, these findings suggest that fostering belonging is a key component of group-based programs, and that feeling a sense of belonging may promote attending programs and changing health behaviours. It could be that feeling connected to other members of a program like RTQ could promote adherence to the program, with those who feel like they belong attending more sessions, and those who do not feel connected to their peers dropping out.  Qualitative results from an evaluation of the first year of RTQ support that group dynamics played an important role in the program. While not focused on investigating group dynamics specifically, an assessment of unprompted open ended responses revealed the importance that participants placed on the role of the group and coach (Glowacki et al., 2018). Some of the strengths of the program identified were group support and accountability, as well as having encouraging and understanding coaches. The variety of running abilities in some groups was seen as a weakness of the program (Glowacki et al., 2018), suggesting that similarity to other participants could be beneficial. These findings suggest that the roles of group-related variables such as cohesion, belonging, participant similarity, and leadership in the program could be further examined. Exploring predictors of attendance could help to optimize the RTQ program for future iterations. Group-related factors such as cohesion, transformational leadership, and perceived 16  similarity, have been related to attendance in group physical activity literature (Beauchamp et al., 2018; Hawley-Hague et al., 2014; Izumi et al., 2015; Smith-Ray et al., 2012), and belonging may also be related (Dowd et al., 2015). However, as far as the author has found their impact on attendance has received limited attention in a smoking cessation literature.  The results of this study could improve future program design by identifying factors that are related to program attendance, such as providing insight into who may benefit the most from a similar program and identifying which participants have lower attendance. This will further our understanding of the effectiveness of RTQ, provide potential information for retaining participants and adapting future versions of the program, and will add to the academic literature on attendance in multi-health behaviour programs. Additionally, there is the opportunity to examine whether including group-related factors such as leadership and cohesion are significant predictors of attendance alongside previously identified individual factors such as nicotine dependence and self-efficacy. Results may identify that enhancing group dynamics should also be emphasized in group-based cessation programs in order to encourage greater success. Purpose The objectives of this study were to: 1. Explore predictors of attendance in RTQ. 2. Determine whether including group-related variables explains additional variance in attendance compared to just individual predictors. Hypotheses 1. Being older, male, owning a home, having greater self-efficacy to quit smoking, greater self-efficacy to run, lower nicotine dependence, higher week 1 moderate-vigorous 17  physical activity, and higher scores on the transformation leadership, belonging, group cohesion, and perceived similarity measures will predict attending more RTQ sessions. 2. Including group-related variables will predict additional variance in attendance compared to only individual predictors.  18  Chapter 2: Methods  2.1 Research design This study was a secondary analysis of data that was collected during a nation-wide evaluation of RTQ. Data collection occurred in spring/summer 2016, 2017, and 2018, as well as fall 2018. Clinics in different locations had staggered start dates, and each clinic was designed to be 10 weeks long. Data were collected using a pre- post design, with surveys completed at weeks 1, 3, and 10.  RTQ included both smoking cessation and physical activity components. Briefly, the Running Room provided a 10 week learn to run 5km program, which was delivered in tandem with the One Step at a Time smoking cessation program from the Canadian Cancer Society (Canadian Cancer Society, 2013). Participants went to a Running Room store for an hour-long clinic once a week. Roughly half of the clinic time was devoted to classroom learning including cessation information and running tips while the other half involved a walk and/or run outdoors led by the coach. Participants were invited to register with a “buddy” who could support them and were encouraged to quit by week 5 of the program if they had not already. A more complete description of the program intervention is provided elsewhere (Priebe et al., 2016).  2.2 Participant characteristics All participants who registered for a RTQ clinic had the option to participate in the evaluation. In order to register for RTQ, participants needed to be of the age of majority in their province, a Canadian citizen, a current smoker or someone who quit within the previous three months, and have smoked at least 100 cigarettes during their life. Details on recruitment methods can be found in a previous paper (Priebe et al., 2016). The entire sample of participants who took 19  part in RTQ from the spring 2016, 2017, 2018, and fall 2018 (n=717) was primarily female (73.1%), with an average BMI of 27.54 kg/m2, and were mainly 35-59 years old (75.8%) with the most common age range being 50-54 years old (19.7%). Participants were generally highly educated, with 69.3% of participants completing college or university. Of the 717 participants who began the program evaluation, 464 (64.7%) completed week 3, and 295 (41.1%) completed week 10. At week 1, participants included current smokers (80.9%), non-smokers (such as smokers who had recently quit; 13%), and buddies who joined to support a friend or partner in the program (11.2%). Buddies could be a current or former smoker or have never smoked. Throughout the program 150 groups were formed, with an average group size of 4.78 ± 3.81 and a range of 1-26 participants. Participants were included in this study’s analyses if they completed surveys at weeks 1 and 3 and were part of a group with at least 3 participants. Any participants that were non-smoking buddies or who did not have attendance data (i.e., the coach did not record attendance in the weekly log) were excluded.  2.2.1 Sampling procedures  All participants who met inclusion criteria were included in the analyses. Data were collected from sites across Canada. No payment was offered for taking part in the evaluation. Ethical approval for this study was obtained from the University of British Columbia’s Ethics Review Board and informed consent was collected from participants prior to completing the first survey.  2.3 Sample size, power, and precision  There was no intended sample size as this was an evaluation of a program using a self-selected population. The sample size (n=339) was determined by how many of the 717 20  participants in RTQ met inclusion criteria. Post-hoc power analyses were conducted using G*Power 3.1 (Faul, Erdfelder, Buchner, & Lang, 2009). Power and effect sizes were determined for each model using the R2 values, a=.05, sample sizes, and number of predictors for both block 1 and the overall model. Results are shown in Table 1. Additional power calculations for whether the proportion of variance in attendance explained by block 1 compared to the overall model were conducted. The analyses calculated power of (1-b) > .9 for all models, apart from power for additional variance in the complete case model. This suggests that the probability of making a Type II error, concluding there is no effect when there actually is one, is low. The mean imputed overall model had a moderate effect size, while the other models had a small-moderate effect size (Cohen, 1988). Table 1. Power Calculations for Models Using G*Power Model Effect size Critical F Numerator df Denominator df Power (1-b error probability) Complete case block 1 .101 1.982 8 214 .940 Complete case overall .143 1.693 16 206 .961 Complete case additional variance .038 1.984 8 206 .483 Mean imputed block 1 .076 1.967 8 326 .968 Mean imputed overall .155 1.675 16 318 .999 Mean imputed additional variance .073 1.968 8 318 .959  2.4 Measures and covariates Each RTQ site had an evaluation assistant (e.g., the coach, a Running Room manager, or a volunteer) who received training on the evaluation, as well as certification to use a coVita piCO+ Smokerlyzer® device and interpret its results. Each group was provided with an evaluation toolkit that contained all necessary materials, and the evaluation assistants administered paper and pen surveys and operated the Smokerlyzer® device. Individual data (e.g., sociodemographic information, moderate-vigorous physical activity (MVPA), smoking history, 21  and nicotine dependence were collected at week 1, while group-related measures and self-efficacies were collected at week 3. Additional measures collected but not included in this evaluation were described previously (Priebe et al., 2016). 2.4.1 Primary outcome: Attendance The primary outcome was attendance, operationalized as the number of sessions that each participant attended during the program (continuous: 1-10, see Figure 1). Attendance data was collected by coaches and reported each week in the coach log. This outcome was chosen for three main reasons: 1) attendance data was objective as it was recorded by coaches and not self-reported, 2) it was available for participants who did not complete week 10 data collection, eliminating the need to address missing data using intent to treat or other methods, and 3) based on the evaluation of the program, participants who completed the program were more likely to have reduced their smoking and increased their physical activity (Priebe, Atkinson, & Faulkner, 2017) – the primary goals of RTQ. Additionally, using a continuous outcome measure provided better insight into whether variance was explained by the predictors (compared to, for example, a dichotomous outcome of whether or not a participant finished the program signified by completing the week 10 survey). Figure 1. Distribution of Attendance Scores for Included Participants (n = 339)  01020304050601 2 3 4 5 6 7 8 9 10Attendance22  2.4.2 Individual predictors  2.4.2.1 Age  Age was reported using a single item with the response options in year categories: 18-24, 25-29, 30-34, 35-39, 40-44, 45-49, 50-54, 55-59, 60-64, and 65 and above.  Due to the categorical nature of the data and the need for dichotomous categories for regression analyses, age was recoded into three categories: “younger adults” (18-39 years), “middle-aged adults” (40-59 years), and “older adults” (60+). Following Field’s (2013) instructions for recoding categorical variables for regressions, younger adults were selected as the base category and two dichotomous dummy variables (one less than the number of categories), that identified middle-aged adults (vs. younger adults), and older adults (vs. younger adults), were created to include in the models. The coding for the dummy variables is shown in Table 2. Table 2. Dummy Coding for Age Variables Age category Dummy variable 1 (middle-aged adults vs. younger adults) Dummy variable 2 (older adults vs. younger adults)  Younger adults 0 0 Middle-aged adults 1 0 Older adults 0 1  2.4.2.2 Week 1 MVPA Week 1 MVPA was defined as the total minutes of recreational and occupational physical activity that was of moderate-vigorous intensity. Total minutes spent doing physical activity in the previous week was collected using the Physical Activity Adult Questionnaire (Garriguet, Tremblay, & Colley, 2015) at week 1. Participants were asked whether they did any active commuting, recreational physical activity, and physical activity while at work, volunteering, and around the home in the past week. If yes, for both recreational and occupational physical activity 23  the item “did any of these recreational physical activities make you sweat at least a little and breathe harder?” was used to determine if the activity was of moderate-vigorous intensity. Participants were then asked to select the days that they did the activities meeting those criteria, and to give the total amount of time spent doing those activities in the past 7 days.  Total minutes of MVPA was calculated by adding the total recreational and occupational physical activity time that was of moderate-vigorous intensity. To attempt to normalize the distribution, data were truncated following the guidelines for the International Physical Activity Questionnaire (The International Physical Activity Group, 2005). Values were truncated so that average activity could not exceed 3 hours per day in each domain (for example, a participant could not do more than 180 minutes of physical activity if they only did one day of activity, and no more than 2160 minutes of activity per week of either recreational or occupational physical activity).  2.4.2.3 Week 1 nicotine dependence The Fagerstrom Test for Nicotine Dependence (FTND; Heatherton, Kozlowski, Frecker, & Fagerstrom, 1991) was used to assess level of nicotine addiction at week 1. This measure contained 6 items, such as “Do you smoke even if you are sick in bed most of the day?” and “On average, how many cigarettes a day do you usually smoke?” One continuous score was calculated by coding and summing the responses (Heatherton et al., 1991). Although the internal reliability was lower than optimal (a = .61) the authors noted that this is not unexpected given that there are few items in the measure, and that it was an improvement over the previous version of the measure (Heatherton et al., 1991). Higher values indicated greater nicotine dependence. 24  2.4.2.4 Self-efficacy to quit Self-efficacy to quit smoking was assessed at weeks 1 and 3 using the item “Using a scale of 1 to 10, where 1 means that you are not at all confident and 10 means you are very confident, please circle how confident you are that you can quit smoking and remain smoke free by the end of this 10-week program?” The week 3 item was used due to participants being more likely to have a realistic idea of their confidence to quit after two weeks of the program compared to on the first night. For the analyses, responses of 1 and 2 were combined into 2 so that were at least 5 responses for the category. This created a range of 2-10. 2.4.2.5 Self-efficacy to run Self-efficacy to run 5 km was assessed at weeks 1 and 3 using the item “Using a scale of 1 to 10, where 1 means that you are not at all confident and 10 means you are very confident, please mark how confident you are that you will be able to run continuously for five kilometers in ten weeks.” The week 3 item was used in the analysis, with responses ranging from 1-10. 2.4.2.6 Gender Participants were asked to identify their gender by the item “I am” with check boxes available for male and female. 2.4.2.7 Home ownership Home ownership was assessed by the item “Now a question about the dwelling in which you live. Is this dwelling?” with a dichotomous response of “Owned by you or a member of this household, even if it is still being paid for”, or “Rented, even if no cash rent is paid”. Home ownership was a proxy for affluence (Hawley-Hague et al., 2014). 25  2.4.3 Group-related predictors 2.4.3.1 Transformational leadership (TL)  The Transformational Teaching Questionnaire (TTQ) to assess transformational leadership in the physical education context was developed by Beauchamp and colleagues (2010), based on the four dimensions of transformational leadership: idealized influence (II), inspirational motivation (IM), intellectual stimulation (IS), and individualized consideration (IC). These dimensions involve inspiring trust and respect through role modeling ethical behaviour by acting in line with one’s beliefs, sharing expectations and motivating others to achieve their goals, promoting viewing situations from multiple perspectives, and identifying individual needs, caring for others, and celebrating successes (Bass & Riggio, 2006; Beauchamp et al., 2010). This measure has been applied in a physical activity context, and found to be associated with athlete wellbeing (Stenling & Tafvelin, 2014), and adolescent self-determined motivation, self-efficacy, and intention to engage in leisure time physical activity  (Beauchamp et al., 2011).  The measure consists of 16 items, 4 items per dimension, on a 5-point Likert-type scale. Items were modified from “my physical education teacher…” to “my Run to Quit coach…” An exemplar question is “My Run to Quit coach shows that he/she cares about me.” Responses could range from 0 (not at all) to 4 (frequently). The items in each dimension were averaged, with higher scores indicating greater perceived transformational leadership. The items were reliable in this sample (16 items; a = .94). 2.4.3.2 Group cohesion  The Physical Activity Group Environment Questionnaire (PAGEQ; Estabrooks & Carron, 2000) was used to measure group cohesion. The PAGEQ has 21-items that are scored on a 9-point scale (1 = very strongly disagree to 9 = very strongly agree). There are four subscales 26  that assess cohesion on both an individual and group level, and in relation to both social activities and tasks. These are individual attraction to group – task (ATG-T), individual attraction to group – social (ATG-S), group integration – task (GI-T), and group integration – social (GI-S). Exemplar questions include “I like the amount of physical activity I get in this program” (ATG-T), “This physical activity group is an important social unit for me” (ATG-S), “Our group is united in its beliefs about the benefits of the physical activities offered in this program” (GI-T), and “Members of our physical activity group often socialize during exercise time” (GI-S; Estabrooks & Carron, 2000, pp. 242–243). The average score for each dimension was calculated.  The PAGEQ was developed for physical activity settings, which may not capture the smoking aspect of RTQ. To help account for this the measure was modified; items in the two task dimensions were added that pertained specifically to smoking but followed the original wording for physical activity. This resulted in an additional 11 items. For example, one modified item was “I like the amount of information about quitting smoking I get in this Run to Quit program.” When Cronbach’s alpha was calculated for the overall measure and for the ATG-T and GI-T subscales, there was little difference when comparing the original items to the original and added smoking-related items (all a > .90). Only the original items were used since they reflect a validated scale and the above results suggest that also examining task cohesion related to the smoking content did not help to assess cohesion beyond the original items.  2.4.3.3 Perceived similarity  Items that assessed perceived similarity to other group members were adapted from previous literature (Beauchamp, Dunlop, Downey, & Estabrooks, 2012; Dunlop & Beauchamp, 2011). Participants were prompted “In my Run to Quit group, I believe that group members are similar to me in terms of…” and asked to respond on a scale of 1 (strongly disagree) to 9 27  (strongly agree) for eight similarities. These items assessed both surface (age, education, physical condition, and ethnicity) and deep (attitudes, personal values, personal beliefs, and life experiences) similarities. A ninth item of “OVERALL, I feel that I am similar to other members of this group” was also included. An average score was calculated for the surface (4 items; a = .75, n = 339) and deep items (4 items; a = .90, n = 339), as well as an overall score that was an average of all items (9 items; a=.90, n = 339).  2.4.3.4 Belonging  Sense of belonging was assessed using 5 items with a 4-point scale anchored by NO! = 1 and YES! = 4 (Anderson-Butcher & Conroy, 2002). Previous data acquired with this measure were found to have acceptable reliability (Anderson-Butcher & Conroy, 2002). Sense of belonging was indicated by the average score of the items, ranging from 1-4. Higher scores indicated a greater sense of belonging. The 5 items were reliable (a = .91, n=339). 2.4.4 Group Size   Group size has been found to affect group dynamics such as cohesion, with being in a smaller group enhancing perceptions of cohesion (Carron & Spink, 1995). In this evaluation, group size was correlated with complete case group-related variables, such as ATG-S (r = -.195, p < .01), average TL (r = -.134, p < .01), and belonging (r = -.193, p < .01). In addition to these variables being statistically significant (p’s < .01) in the mean imputed correlation matrix, GI-T was also significantly correlated (r = -.152, p < .01). Based on these relationships and the literature, group size was included as a control variable in the analyses so that it is held constant when interpreting the results. This helps to prevent group size from confounding the results, helping to better understand the impact of the individual and group-related predictors on attendance. Group size was defined as the total number of participants who completed the week 28  1 survey at the same site, in the same cycle/clinic season. This definition of group size was chosen because regardless of whether all participants attended consistently between weeks 1 and 3, this was the group size that each participant had likely been exposed to in the first two weeks and would be able to reference during the week 3 survey with group measures. 2.5 Analysis Statistical analyses were performed using SPSS versions 24 and 26 (IBM, New York, USA). All statistical tests were two-tailed. A significance level was determined a priori to be a = .05. Descriptive statistics (independent t-tests and chi-square tests) were used to describe the sample. 2.5.1 Data exploration  2.5.1.1 Data visualization and correlations Univariate distributions for each predictor variable vs. the outcome variable were created to become familiar with the data. This also helped to identify any data entry errors or potential outliers. Contingency tables were created to check for sufficient amount of data within each cell. For example, if there were less that 5 cases within a cell, strategies such as combining cells may have been required to have adequate numbers for later statistics. Pearson correlations between each predictor variables and the outcome variable were tested to explore the relationships between the variables.  2.5.1.2 Missing data The data were cleaned to ensure that missing responses were assigned dummy variables when applicable. For example, someone who was not smoking at week 1 may not have filled out the smoking question sections, in which case their missing answers would be not applicable. 29  Patterns of missing data were explored using frequencies and Little’s Missing Completely at Random (MCAR) test.  2.5.2 Intra-class correlation coefficient (ICC) In order to determine if data could be analyzed at the individual level, clustering effects were tested using ICCs. Similar to Beauchamp and colleagues (2018), ICCs were used to calculate design effects (Hox, 2010, as cited in Beauchamp et al., 2018). If design effects were <2 individual-level analyses would be used. If design effects were >2 multilevel modeling would have been needed to take the clustering into account (Muthen & Satorra, 1995).  2.5.3 Blocked ordinary least squares (OLS) multiple regression  If design effects were <2, data would be analyzed at an individual level using blocked multiple linear regression. Blocked OLS multiple regression is an appropriate statistical analysis to assess whether a set of independent (predictor) variables predict a continuous outcome variable, and whether including additional variables explains additional variance in the outcome.  To address the research objectives, predictors were entered into the model in two blocks. The following variables were individual predictors entered in the first block: older adults, middle-aged adults, gender, home ownership, quit self-efficacy, run self-efficacy, baseline nicotine dependence, baseline MVPA. Group-related predictors were entered in the second block: ATG-T, ATG-S, GI-T, GI-S, average transformational leadership score, average perceived similarity, average belonging, and group size (control variable). Group size was included in the second block because it is related to several group-related predictors and does not need to be accounted for in the first block of individual predictors. Including variables in two blocks allowed for exploring which variables were significant predictors in the overall model, and whether including group-related variables explained additional variance in attendance 30  compared to a model of only individual-related predictors. Variables that had a statistically significant contribution to the model were identified in the coefficients table, with standardized beta values showing the amount of change in attendance when all other variables in the model were held constant.  The models were checked for assumptions of multiple regression. The assumption of multicollinearity was addressed if tolerance values were <10 and variance inflation factors were >.2. A plot of ZPRED (the standardized predicted values of the outcome variable based on the model) vs. ZRESID (standardized residuals) was used to check for homogeneity of variance, and histograms and P-P plots were used to assess normality. Durbin-Watson values close to 2 and within 1-3, suggested that the data met the assumption of independence. Influential cases were explored by checking that Cook’s distances were <1, no more than 5% of cases had standardized residual values >2 and no more than 1% of cases had a standardized residual >3, and DFBetas were <1. 31  Chapter 3: Results 3.1 Participant flow Of the 717 participants whose data were collected during Run to Quit, 378 (52.72%) were excluded due to not meeting inclusion criteria. Participants were included in this study’s analyses if they completed surveys at weeks 1 and 3 and were part of a group with at least 3 participants (n=423). Participants who were non-smoking buddies were excluded (n=29), as well as participants who did not have attendance data due to coach logs or attendance being incomplete (n=61). This resulted in a sample of 339 participants who met inclusion criteria (see Figure 2). The included participants were part of 87 groups, with an average group size of 6.34 and range of 3-26 participants. Of the participants who met inclusion criteria for the main analysis, 154 (45.42%) completed the week 10 evaluation.  Figure 2. Flowchart of Participants Included in Evaluation  There were some statistically significant differences between included and excluded participants (see Table 3). Baseline nicotine dependence was higher (4.24 ± 2.57) for included participants than for excluded (3.39 ± 2.71, p <.01). Ethnicity had a weak association with being included in this study (phi = -.081, p = .031; Akoglu, 2018). There were more participants who 32  identified as White included (332; 51.00%) than excluded (319; 49.00%; c2 [1, n = 651] = 4.63, p = .03). There was a weak association (phi = .08, p = .03) between inclusion in the study and education, with fewer participants who had completed secondary school or less (61; 18%) included than were excluded (91; 24.9%; c2 [1, n = 703) = 4.91, p = .03). There was a statistically significant difference in number of sessions attended with the participants included attending more sessions (6.40 ± 2.61) than those excluded (3.81 ± 2.83, p <.01). This difference was likely due to only participants who attended both weeks 1 and 3 of the program being included in this evaluation. Table 3. Comparison of Included and Excluded Participants’ Complete Case Data Variable      t-tests Included Excluded t-value df Significance Attendance 6.40 ± 2.61 n = 339 3.81 ± 2.83 n = 318 -12.19 655 <.01 Week 1 CO (ppm) 16.81 ± 11.71 n = 328 17.30 ± 13.74 n = 325 .49 651 .63 Week 1 BMI 27.31 ± 5.05 n = 332 27.75 ± 8.68 n = 357 .81 687 .42 Age started smoking 15.44 ± 4.36 n = 332 15.48 ± 3.59 n = 338 .12 668 .90 Number of years been smoking 28.27 ± 11.31 n = 307 27.31 ± 11.97 n = 291 -1.00 596 .32 Week 1 FTND 4.24 ± 2.57 n = 326 3.39 ± 2.71 n = 346 -4.24 670 <.01 Week 1 average cigarettes/week 83.84 ± 59.82 n = 313 77.79 ± 61.84 n = 329 -1.26 640 .21 Week 1 MVPA 144.17 ± 203.11 n = 321 145.11 ± 203.96 n = 352 .060 671 .95 Chi-square tests Included Excluded Chi2 Value df Significance Female 246 (46.95%) 278 (53.05%) 1.50 1 .22 Age 18-39 87 93 .03 2 .99 Age 40-59 217 233 .03 2 .99 Age 60+ 35 39 .03 2 .99 Ethnicity (white) 332 (51.00%) 319 (49.00%) 4.63 1 .03 Home owned 244 (49.29%) 251 (50.71%) 1.04 1 .31 Employed full time 222 238 2.76 7 .91 Completed secondary school or less 61 91 4.91 1 .03 Completed post-secondary education 277 274 4.91 1 .03  33  3.2 Statistics and data analysis   Assumptions of OLS regressions were checked using the instructions provided by Field (2013). In the complete case analysis, two cases were below the covariance ratio threshold recommended by Belsey, Kuh, and Welsch (1980), and in the mean imputed model there were four cases below the threshold. The authors suggested that deleting these cases would improve the precision of some of the model’s parameters. Additionally, two cases in the mean imputed model had a centralized leverage value greater than three times the average, which was suggested by Stevens (2002) as a cut off for identifying cases that may have undue influence. However, following Stevens (2002), these points were not removed due to the Cook’s distances all being <1, which suggests that they were not influential data points. To avoid potential multicollinearity, dimensions of the TTQ were combined into one average TTQ score (r’s ≥ .70, complete case – see Table 4 for Pearson correlations). The 16 TTQ items were reliable (a=.94). An overall average of perceived similarity was also created (rdeep, surface) = .732, with 9 reliable items (a=.90). This is similar to previous work by Stevens and colleagues (2018) that used a global score of a measure when subscales were highly correlated. Table 4. Complete Case and Mean Imputed Pearson Correlations for TTQ Subscales to Investigate Potential Multicollinearity Complete Case (n = 223)     II IM IS IC II 1.000 .770** .750** .813** IM  1.000 .728** .810** IS   1.000 .706** IC    1.000 Mean Imputed (n = 335)     II IM IS IC II 1.000 .748** .734** .796** IM  1.000 .712** .801** IS   1.000 .687** IC    1.000 ** p < .01  34  In order to determine if data could be analyzed at the individual level, clustering effects were tested using ICCs. Similar to Beauchamp and colleagues (2018), ICCs were used to calculate design effects (Hox, 2010, as cited in Beauchamp et al., 2018). The ICC for attendance was .054178, with a design effect of 1.154 (1 + [average cluster size – 1]*ICC = (1 + [3.850574713– 1]*.054178) in the mean imputed analysis, while the complete case analysis had an ICC of .031265 with a design effect of 1.054 (1 + [2.719512195– 1]*.031265). This suggests that 5.418% and 3.127% of the variance in attendance was due to groups. Clustering effects do not impact the standard errors of regression parameter estimates when design effects are <2 (Muthen & Satorra, 1995, as cited in Beauchamp et al., 2018), suggesting that the data could be treated at the individual level.  Overall, 2.86% of data were missing (155 responses/[16*339 total responses] = 155/5424 = .028577* 100% = 2.86%; see Table 5). The measures with more than 5% missing data were the average of the TTQ (56; 16.52%), average self-perceived similarity (24; 7.08%), and week 1 MVPA (18; 5.31%). The missingness of the variables in the model was explored using Little’s Missing Completely at Random test, with results suggesting that data were not missing completely at random (c2 [298] = 354.796, p = .013).   35  Table 5. Missing Data Summary for Complete Case and Mean Imputed Predictors   Complete Case Number of Extremes Variable Missing frequency % missing cases for variable Low High Older adults 0 .00% 0 0 Middle aged adults 0 .00% 0 0 Gender 0 .00% 0 0 Home ownership 4 1.18% 0 0 Quit self-efficacy 2 .59% 0 0 Run self-efficacy 2 .59% 0 0 FTND 13 3.83% 0 0 Week 1 MVPA 18 5.31% 0 26 ATG-T 10 2.99% 4 0 ATG-S 2 .59% 2 0 GI-T 9 2.65% 1 0 GI-S 7 2.06% 12 3 TL 56 16.52% 3 0 Belonging 8 2.36% 1 0 Similarity 24 7.08% 1 0 Group size 0 .00% 0 19 Total 155 2.86% 24 48 Three techniques to address missing data were considered: listwise deletion of missing cases, regression imputation, and mean imputation. Deletion could lead to biased estimates because the data were not missing completely at random and would lead to a complete case sample size of n=223, excluding roughly one-third of otherwise eligible participants who are missing a small amount of data. Regression imputation uses existing and missing data to predict what each missing value would be, and requires data to be missing completely at random or missing at random (Sterner, 2011). Mean imputation is when a single value based on the overall mean for an item is used to replace missing values, rather than predicting the value using several other values (Sterner, 2011). Although a form of regression imputation would have been a stronger method to address missing data (Sterner, 2011), mean imputation was used because mean values were easily computed and only a small amount of overall data was missing. Mean imputation (taking the mean of all responses and substituting this value for any missing) was used for quit self-efficacy, run self-efficacy, MVPA, and nicotine dependence. For the multi-item 36  measures, if one item was missing then the score for the measure was missing because it was unable to be calculated. When a score for the cohesion subscales, subscales of the TTQ, perceived similarity, and sense of belonging scores was missing due to a single item, then that item was replaced using mean substitution (the aggregate mean for that item). The score for the measure or subscale was then calculated using the individual’s responses to the other items along with the imputed item. Complete case analyses and mean imputed results are reported. Four participants were missing a response (1.18% of responses) for the dichotomous item regarding home ownership. These missing data were not replaced, resulting in casewise exclusion from the models and a final sample of n=335 for the mean imputed analysis that was missing .074% of the total data (4/5424 total responses = .000737463 = 0.0746% of the data included in the model is missing). For complete case and mean imputed descriptive information for each variable, see Table 6. Overall, attendance had a mean of 6.39 ± 2.61 sessions (n = 335). At week 1, the majority of participants were smoking daily (83.4%), while some were no longer smoking at all (10.7%), and a small amount reported smoking occasionally (5.9%). Fifteen (4.4%) participants were buddies who also smoked. On average, participants started smoking at 15.44 ± 4.36 years old and had been smoking for 28.27 ± 11.31 years. The majority of participants were female (72.6%), identified as white (94.1%), and most participants were 40-59 years old (64%), with other participants being 18-39 (25.7%) and £60 years old (10.3%). Most participants had completed college (41.6%) or university (29.5%), and the majority were employed full time (65.5%). See Tables 7 and 8 for correlation matrices for the complete case and mean imputed models.   37  Table 6. Descriptive Statistics for Variables in Complete Case and Mean Imputed Models Variable Mean ± standard deviation or frequency for complete case (n = 223) Mean ± standard deviation for mean imputed (n = 335) Attendance 6.52 ± 2.60 6.39 ± 2.61 Middle aged adults 144 (64.57%) 216 (64.48%) Older adults 16 (7.17%) 35 (10.45%) Gender (female) 158 (70.85%) 242 (72.24%) Home owned 158 (70.85%) 244 (72.84%) Quit self-efficacy 7.65 ± 2.22 7.74 ± 2.14 Run self-efficacy 7.35 ± 2.70 7.23 ± 2.66 FTND 4.22 ± 2.59 4.23 ± 2.52 Week 1 MVPA 143.41 ± 172.56 144.18 ± 199.00 ATG-T 7.43 ± 1.37 7.44 ± 1.33 ATG-S 6.94 ± 1.42 6.88 ± 1.44 GI-T 6.94 ± 1.26 6.91 ± 1.26 GI-S 5.55 ± 1.29 5.50 ± 1.33 TL 3.42 ± .57 3.44 ± .53 Similarity 6.13 ± 1.30 6.13 ± 1.26 Belonging 3.72 ± .40 3.73 ± .39 Group size 8.23 ± 5.39 8.56 ± 5.75    38  Table 7. Complete Case (n = 223) Pearson Correlation Matrix for Predictors and Attendance Variable 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 1. Attendance 1                 2. Middle aged adults -.008 1                3. Older adults .072 -.375** 1               4. Gender -.108 .123 .025 1              5. Home ownership .116 .206** .025 -.021 1             6. Quit self-efficacy .191** .064 -.034 -.054 .111 1            7. Run self-efficacy .072 -.028 -.095 -.191** .051 .275** 1           8. Week 1 MVPA -.123 .054 -.065 -.006 .007 .018 -.001 1          9. Week 1 FTND -.177** .083 -.011 -.005 .029 -.291** -.021 .076 1         10. ATG-T .111 .055 -.088 -.001 .055 .335** .269** -.048 -.034 1        11. ATG-S .160* .072 -.056 .039 -.035 .269** .142* .038 -.055 .445** 1       12. GI-T .094 -.071 .053 -.046 -.002 .315** .232** -.060 -.009 .582** .564** 1      13. GI-S .072 .038 -.062 -.013 -.136* .272** .137* .103 -.121 .367** .606** .522** 1     14. Average TL .015 .021 -.028 -.011 -.024 .243** .184** .076 -.018 .468** .430** .552** .446** 1    15. Belonging .156* -.007 -.049 -.088 .015 .356** .310** .002 -.048 .446** .442** .465** .316** .509** 1   16. Similarity .085 .159* -.080 .031 .002 .144* .137* -.029 .063 .444** .412** .459** .355** .361** .306** 1  17. Group size -.051 .045 -.083 .130 .090 .071 .042 .093 .089 -.027 -.195** -.075 -.029 -.134* -.193** -.087 1 * p < .05 ** p < .01    39  Table 8. Mean Imputed Pearson Correlation Matrix for Predictors and Attendance Variable 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 1. Attendance 1                 2. Middle aged adults .019 1                3. Older adults .059 -.453** 1               4. Gender -.075 .049 .057 1              5. Home ownership .024 .136* .099 -.019 1             6. Quit self-efficacy .119* .035 .001 -.009 .085 1            7. Run self-efficacy .087 .005 -.121* -.188** .033 .272** 1           8. Week 1 MVPA -.130* .019 -.008 -.103 .037 -.011 .056 1          9. FTND -.159** .018 .026 -.065 .039 -.237** -.044 .076 1         10. ATG-T .109* .034 -.110* .056 .009 .305** .237** -.026 -.046 1        11. ATG-S .162** .055 -.099 .077 -.101 .258** .182** .003 -.067 .484** 1       12. GI-T .088 -.085 -.053 .011 -.013 .281** .188** -.014 -.023 .593** .582** 1      13. GI-S .050 .055 -.144** -.021 -.100 .232** .165** .067 -.085 .389** .579** .532** 1     14. Average TL -.020 .038 -.064 .045 -.002 .250** .169** .079 -.031 .474** .412** .526** .407** 1    15. Similarity .050 .097 -.076 .066 -.004 .132* .088 .015 .083 .410** .393** .414** .328** .305** 1   16. Belonging .177** .058 -.113* -.026 -.002 .333** .266** -.011 -.027 .478** .446** .443** .329** .478** .275** 1  17. Group size -.007 .012 .009 .075 .031 .035 .003 .059 .082 -.074 -.181** -.152** -.021 -.174** -.071 -.189** 1 * p < .05 ** p < .01  40  In the complete case analysis, both block 1 and the overall model were statistically significant and explained 5.8% and 5.7% of the variance in attendance (block 1: adjusted R2 = .058, F(8,214) = 2.715, p = .007; overall: adjusted R2 = .057, F(16,206)=1.841, p = .028). When only individual factors were included in block 1, none were statistically significant predictors of attendance at p <.05, although higher self-efficacy to quit (b = .135, p = .061), nicotine dependence (b = -.133, p = .055), and week 1 MVPA (b = -.113, p = .086) approached statistical significance. When group-related variables were added, the overall model explained 5.7% of the variance, with lower nicotine dependence (b = -.140, p = .049) was a statistically significant predictor of higher attendance. See Tables 9 and 10 for regression results.   41  Table 9. Coefficients for Complete Case Model Predicting Attendance Variable Unstandardized Coefficients Standardized Coefficients t-value p-value 95% Confidence Interval for B B Std. Error b Lower Bound Upper Bound Block 1: Individual predictors  Constant 5.891 .929  6.344 .000 4.060 7.721 Middle aged adults .124 .400 .023 .310 .757 -.664 .911 Older adults  .784 .718 .078 1.092 .276 -.631 2.200 Gender  -.584 .382 -.102 -1.528 .128 -1.337 .170 Home ownership .549 .385 .096 1.425 .156 -.210 1.308 Quit self-efficacy .158 .084 .135 1.886 .061 -.007 .324 Run self-efficacy .015 .067 .016 .224 .823 -.117 .147 Week 1 MVPA -.002 .001 -.113 -1.723 .086 -.004 .000 Week 1 FTND -.134 .069 -.133 -1.927 .055 -.270 .003 Overall Model Constant 3.650 1.813  2.013 .045 .075 7.225 Middle aged adults .043 .409 .008 .104 .917 -.764 .850 Older adults .897 .728 .089 1.233 .219 -.537 2.332 Gender -.638 .391 -.112 -1.633 .104 -1.409 .132 Home ownership .575 .395 .101 1.458 .146 -.203 1.353 Quit self-efficacy .108 .092 .093 1.179 .240 -.073 .290 Run self-efficacy -.008 .070 -.008 -.112 .911 -.145 .129 Week 1 MVPA -.002 .001 -.108 -1.603 .110 -.004 .000 Week 1 FTND -.141 .071 -.140 -1.979 .049 -.281 -.001 ATG-T .046 .167 .024 .274 .784 -.283 .374 ATG-S .289 .172 .158 1.677 .095 -.051 .629 GI-T -.113 .207 -.055 -.545 .586 -.520 .295 GI-S -.039 .182 -.019 -.213 .832 -.398 .321 Average TL -.492 .396 -.108 -1.240 .216 -1.273 .290 Belonging .674 .555 .104 1.214 .226 -.421 1.769 Similarity .104 .158 .052 .658 .511 -.208 .417 Group size .007 .034 .015 .204 .838 -.061 .075     42  Table 10. Regression Coefficients for Mean Imputed Model Predicting Attendance Variables Unstandardized Coefficients Standardized Coefficients t-value p-value 95% Confidence Interval for B B Std. Error b Lower Bound Upper Bound Block 1: Individual predictors Constant 6.101 .796  7.669 .000 4.536 7.666 Middle aged adults .495 .338 .091 1.464 .144 -.170 1.160 Older adults 1.024 .530 .120 1.930 .054 -.020 2.067 Gender -.584 .320 -.100 -1.822 .069 -1.214 .047 Home ownership .000 .323 .000 .001 .999 -.636 .637 Quit self-efficacy .083 .070 .068 1.180 .239 -.055 .221 Run self-efficacy .068 .056 .069 1.207 .228 -.043 .178 Week 1 MVPA -.002 .001 -.135 -2.500 .013 -.003 .000 Week 1 FTND -.140 .058 -.135 -2.439 .015 -.253 -.027 Overall Model Constant 2.652 1.509  1.757 .080 -.317 5.622 Middle-aged adults .482 .338 .088 1.423 .156 -.184 1.147 Older adults  1.196 .525 .140 2.281 .023 .164 2.229 Gender  -.689 .320 -.118 -2.155 .032 -1.318 -.060 Home ownership .102 .320 .017 .319 .750 -.528 .732 Quit self-efficacy .008 .073 .006 .106 .916 -.137 .152 Run self-efficacy .026 .056 .027 .468 .640 -.085 .137 Week 1 MVPA -.002 .001 -.120 -2.251 .025 -.003 .000 Week 1 FTND -.150 .058 -.145 -2.613 .009 -.263 -.037 ATG-T .073 .142 .037 .514 .607 -.206 .351 ATG-S .343 .134 .189 2.556 .011 .079 .607 GI-T .000 .164 .000 .002 .998 -.323 .323 GI-S -.076 .137 -.039 -.555 .579 -.345 .193 Average TL -.788 .331 -.160 -2.384 .018 -1.439 -.138 Similarity -.032 .127 -.016 -.254 .800 -.282 .218 Belonging 1.212 .441 .183 2.746 .006 .344 2.080 Group size .028 .025 .061 1.090 .276 -.022 .077   In the mean imputed analysis, the individual predictors in block 1 explained 4.80% of the variance in attendance (adjusted R2 = .048, F(8,326) = 3.11, p < .01) with having lower nicotine dependence (b = -.135, p = .015), and lower MVPA (b = -.135, p = .013) significantly predicting higher attendance. Non-significant individual predictors were older adults and middle-aged adults, gender, and run and quit self-efficacies. The overall model with group-related variables added was statistically significant (F(16,318) = 3.07, p < .01) and explained 9.0% of the variance in attendance (adjusted R2 = .090). In the overall model, being an older adult (b = .140, p = .023), male (b = -.118, p = .032), having lower nicotine dependence (b = -.145, p = 43  .009), lower MVPA (b = -.120, p = .025), higher ATG-S (b = .189, p = .011), lower average transformation leadership (b = -.160, p = .018), and higher belonging (b = .183, p = .006) significantly predicted attendance. In the overall model, home ownership, quit and run self-efficacies, ATG-T, GI-T, GI-S, and perceived similarity to other group members did not significantly predict attendance.   44  Chapter 4: Discussion The purpose of this thesis was to explore the predictors of attendance in RTQ, and to examine whether including group-related variables added to the prediction of attendance beyond individual variables. Results partially supported the hypotheses. Consistent with the first hypothesis, being older, male, having lower nicotine dependence, higher perceptions of belonging, and higher scores on the individual attraction to group-social cohesion dimension were predictive of attending more RTQ sessions. However, in contrast to the hypothesized directions, lower baseline MVPA and transformation leadership scores were predictive of attending more sessions. Home ownership, quit and run self-efficacies, perceived similarity, and the ATG-T, GI-T, and GI-S dimensions of cohesion were not significant predictors of attendance. The results of the mean imputed model supported the second hypothesis, where including group-related variables predicted additional variance in attendance compared to including only individual predictors. While accounting for a modest amount of variance, these results demonstrated the potential value of including group-related variables alongside more frequently studied individual variables when investigating predictors of attendance in a group-based smoking cessation intervention. The mean imputed model results will be the focus of the discussion, unless otherwise specified, due to having a larger sample size once missing data were accounted for. This offers greater power for the analysis, particularly when looking at additional variance added, and a more complete assessment of the sample by not excluding participants who were missing a small amount of data.  The first hypothesis was partially supported, with two statistically significant predictors of attendance identified when including only individual variables in the model: having lower nicotine dependence and lower MVPA at baseline. Both of these factors were also statistically 45  significant predictors in the overall model. Higher nicotine dependence has been a consistent predictor of attrition in smoking cessation interventions (Belita & Sidani, 2015). Lower nicotine dependence was associated with making a quit attempt and having a successful attempt in the general population (Vangeli et al., 2011). For example, higher levels of nicotine dependence were related to lower likelihoods of completing smoking cessation interventions (Brouwer & Pomerleau, 2000; Copeland et al., 2006; Geraghty et al., 2012), and in another study examining FTND scores and attendance, those attending all sessions had lower FTND scores (Dorner et al., 2011). This study’s finding that lower nicotine scores predicted greater attendance is consistent with nicotine dependence being a consistent predictor of success in smoking cessation interventions. It may be easier for participants with lower baseline nicotine dependence to continue attending. For example, participants who have lower nicotine dependence may be more ready to quit and take on the cessation component (Prokhorov et al., 2001). Nicotine dependence is also related to physical activity levels, as seen with older adults who have higher nicotine dependence doing less physical activity and being more sedentary (Loprinzi & Walker, 2015), and young adult smokers having a lower tolerance for exercise compared to non-smokers (Papathanasiou et al., 2007). Participants with higher nicotine dependence may have been less ready to quit and less inclined to continue doing physical activity once starting the program, resulting in attending fewer sessions.  Week 1 MVPA was a significant predictor of attendance, but the direction was opposite to the hypothesis. Having higher baseline MVPA was hypothesized to lead to higher attendance based on the assumption that those who were already able to run at a certain level would be more likely to meet the challenge of the progressive running program. However, this result could be explained by year 1 participants reporting that having different running abilities in the same 46  group was a weakness of the program (Glowacki et al., 2018) because it could lead to a mismatch between individual and group needs. This mismatch in abilities was particularly salient for those who did not complete the program (Glowacki et al., 2018). Although speculative, this could suggest that participants who were more active at baseline may have felt that the progressive run curriculum in RTQ was not congruent with their abilities and interests, and that this led to them attending fewer sessions. It remains encouraging that those with low levels of MVPA, the target audience of the Run to Quit program, were more likely to attend given the associated health benefits of increasing physical activity.  However, it is worth noting that even after attempting to normalize the MVPA data with truncation, some high values remained, which is reflected in the large standard deviation. These scores may belong to participants who have a physical occupation, and could reflect the limitations of self-reported physical activity where participants may over-report their activity compared to device-measured data (Garriguet et al., 2015). Although no cases were identified as being influential data points in the overall model, further investigation into whether there are outliers in the MVPA data could be conducted. If outliers are detected, the MVPA data could be transformed, or the outliers could have their score changed to a value that is still high, but not so high that it is an outlier (Field, 2013). In the overall model, when group-related predictors were considered alongside the individual factors, additional individual and group-related predictors of attendance were identified. In addition to the individual predictors significant in block 1, both being an older adult (compared to being a younger adult) and male were also statistically significant predictors. However, it should be noted that the introduction of group-related predictors may have led to net suppression effects and statistical artifacts due to multicollinearity between some group-related 47  variables. For example, although age and gender were not significantly correlated with attendance in the correlation matrices, they are statistically significant predictors in the overall model. Additionally, belonging and transformational leadership have small correlations with attendance, yet relatively large beta values in the regression. This may be due to multicollinearity between cohesion subscales that had moderate correlations with each other, as well as with the leadership, belonging, and similarity scores. Although VIF and tolerance values for the model were within acceptable ranges and the correlations were <.7, multicollinearity may have still occurred. The results should therefore be interpreted with caution taking potential net suppression effects and statistical artifacts into account. The finding that being an older adult (compared to a younger adult) was related to having higher attendance is consistent with previous literature. In a review of attrition in smoking intervention studies, younger participants were more likely than older ones to drop out (Belita & Sidani, 2015). These results included studies where younger participants were less likely to return for randomization for a clinical trial (Ahluwalia et al., 2002), and where younger age was predictive of attrition in an e-health intervention (Geraghty et al., 2012). Similarly, being older decreased the likelihood of early dropout in a multi-component smoking intervention study for women, regardless of longest prior quit attempt. The authors suggested that this finding may be due to age being protective against early drop rather than being due to older participants having longer previous quit attempts (Leeman et al., 2006). Older participants in RTQ may have had fewer logistical constraints, such as full-time work and family duties, that could have impacted their ability to attend. However, these results also highlight that those who could benefit the most from smoking cessation may not be attending the program the most. For example, Taylor and colleagues’ (2002) found that those who quit before 35 years of age have the greatest life 48  extensions. Although quitting at any age is beneficial, programs that are effective for younger adults are desirable.  An additional consideration is whether age becoming a significant predictor in the overall model is a statistical artifact. As seen in the correlation matrix, being an older adult had statistically significant negative correlations with group-related variables such as ATG-T, GI-S, and belonging. This could suggest that older adults being a significant predictor in the overall model is not a statistical artifact and that older adults may have reported lower scores on group-related measures. There is a need to further investigate age-related factors that impact participation (Belita & Sidani, 2015), including further attention to the role age plays in group-related aspects of RTQ.  Greater male attendance is consistent with previous literature. Belita and Sidani’s (2015) review of attrition in smoking cessation interventions found that females were more likely to drop out after starting a study, while males were more likely to withdraw before participating (Ahluwalia et al., 2002; Belita & Sidani, 2015). This suggests that males who would otherwise have dropped out may have self-selected out of the intervention earlier, while females in similar positions dropped out during the study. In a recent review of sex/gender differences in efficacy and effectiveness trials, as well as prospective observational smoking cessation studies, the authors concluded that women were less likely than men to maintain long-term abstinence during a quit attempt, and that there may be no difference between those seeking treatment and those in the general population (Smith et al., 2016). This further supports that sex/gender differences exist and that the result in this study is consistent with previous research. Interestingly, women have been reported as being more likely to seek treatment for smoking cessation rather than quitting without aids (Shiffman, Brockwell, Pillitteri, & Gitchell, 2008). This could mean that the men 49  who registered for RTQ were particularly motivated to try the program. Males may also have been more likely to continue attending due to gender differences in attraction to physical activity, such as differences in sport participation (Deaner et al., 2012) and leisure time physical activity (Azevedo et al., 2007). However, these results should be interpreted in the context of this analysis. Gender had a small, non-significant correlation with attendance and was not significantly correlated with any group-related predictors. This suggests that its emergence as a significant predictor in the overall model may be a statistical artifact.  The remaining individual predictors of home ownership, quit self-efficacy, and run self-efficacy were not significant in block 1 or the overall model. Home ownership was a proxy for affluence (Hawley-Hague et al., 2014) and was not a statistically significant predictor of attendance in RTQ. However, consistent with the hypothesis, the direction of the relationship suggested that participants who owned their home were more likely to attend more sessions. In smoking cessation research, a meta-analysis of studies on the general population found that measures of affluence had inconsistent results when predicting making a quit attempt, and that although there was some evidence for higher affluence predicting a successful quit attempt, overall results were mixed (Vangeli et al., 2011). There was also some limited evidence that lower income was associated with higher attrition in smoking cessation interventions (Belita & Sidani, 2015; Nevid et al., 1996). However, in a community group-based exercise program owning a home was associated with poorer attendance than those who rented their home, which may have been due to more affluent participants missing sessions for vacations (Hawley-Hague et al., 2014). The sample in this study appeared to be fairly affluent due to being highly educated, primarily working full-time, and the majority owning their own home. Homogeneity in the sample as well as using a dichotomous measure for affluence may have limited the predictive 50  ability of this variable. Based on the inconsistent results in the smoking cessation literature it is unsurprising that socioeconomic status, represented by home ownership, does not appear to be a significant predictor of attendance in RTQ programs.  Self-efficacy, or self-confidence in some studies, was related to attrition in smoking cessation interventions (Belita & Sidani, 2015). In contrast to this overall trend, quit self-efficacy was not a significant predictor of attendance in this study. Although the direction of the relationship is consistent with previous work on confidence to quit and completing an intervention study (Leeman et al., 2006; Nevid et al., 1996), as well as a meta-analysis on self-efficacy and no longer smoking at follow up (Gwaltney et al., 2009), other factors were stronger predictors of attendance in this model.  Similarly, self-efficacy to run was expected to be a significant predictor of attendance. There is strong evidence that self-efficacy is a consistent correlate of physical activity (Bauman et al., 2012), and that increasing self-efficacy leads to increased physical activity (Williams & French, 2011). Despite the demonstrated relationship with physical activity behaviour, run self-efficacy was not a significant predictor of attendance in the RTQ program. The non-significant findings for both self-efficacies suggest that attendance was more impacted by other factors. For example, even if a participant believed that they would be able to quit smoking and run 5km by the end of the program, they may have been prevented from attending because of the logistic or personal reasons previously described in a qualitative evaluation of the program (Glowacki et al., 2018). Future research could consider whether including an additional measure of self-efficacy would be beneficial. For example, participants could be asked to list barriers that they believe might impact their attendance, and to rate their confidence in overcoming that barrier during the program (cf. Cramp & Bray, 2009). Alternatively, participants with high self-efficacy may not 51  have felt cohesive or a sense of belonging with their group, which could have had a stronger impact on their attendance.  The measurement of self-efficacy may have also impacted the results of this evaluation. Although there is some evidence that single-item measures are comparable to multi-item measures (Hoeppner, Kelly, Urbanoski, & Slaymaker, 2011), multi-item measures that have been assessed for reliability could have been used, such as the Self-Efficacy for Exercise scale (Resnick & Jenkins, 2000) or the Multidimensional Self-efficacy for Exercise Scale (Rodgers, Murray, Wilson, Hall, & Fraser, 2008). As expected based on previous literature demonstrating the positive impacts of groups on smoking cessation and physical activity (Harden et al., 2015; Stead et al., 2017), the inclusion of group-related variables improved the model’s ability to explain variance in attendance. This supports that the group aspect of the program is important and could be further emphasized in future iterations. Including group-related variables identified several more statistically significant predictors of attendance: the ATG-S dimension of cohesion, belonging, and transformational leadership.  Consistent with previous group exercise programs, higher ATG-S was related to higher attendance (Spink & Carron, 1994). This suggests that participants felt motivated to attend when they had higher perceptions of being accepted by the group, and of their interactions with the group. This finding aligns with previous RTQ results where participants reported valuing group discussions and sharing their stories as they made quit attempts, and a feeling of accountability that helped participants to achieve their goals (Glowacki et al., 2018). These sentiments were not affected by completion status in the qualitative evaluation. However, the present study suggests that participants’ social interactions with the group and sense of acceptance may have been a 52  stronger motivator for those who attended more sessions, or those who developed it earlier on in the program may have been more likely to attend consistently. This finding is in contrast to early work on cohesion and exercise class adherence that found that ATG-T was the cohesion subscale that best predicted adherence behaviour (Spink & Carron, 1992, 1993). It was also the dimension found to be a significant predictor for attendance in a clinical setting (Fraser & Spink, 2002). However, a follow up study to Spink and Carron’s earlier work found that when cohesion was measured at week 3, adherence to the last four weeks of the program was predicted by different subscales in university and community settings (Spink & Carron, 1994). For university exercise classes, task dimensions of cohesion were better predictors of adherence while in community exercise groups social dimensions were higher for those with high attendance compared to those who did not attend the last four weeks of classes. Having ATG-S as a significant predictor of attendance in the current study is consistent with Spink and Carron’s (1994) results due to the community-based setting. It is also consistent with more recent research, such as community-based exercise classes for older adults where higher ATG-S scores were associated with attendance at both 3- and 6-months of the program (Hawley-Hague et al., 2014). RTQ was delivered at local Running Room stores and was open to members of the community. The results of this evaluation support that the ATG-S dimensions of cohesion may better predict attendance in a community-based exercise setting. Following belongingness theory (Baumeister & Leary, 1995), participants may have been driven by a need to feel positive affect with their RTQ group that lasts over time. As described by Glowacki and colleagues (2018), feeling supported helped to develop a sense of accountability and was a strong motivator during the program. Strong feelings of belonging appear to have promoted higher attendance in RTQ, aligning with the qualitative responses from 53  year 1 participants. High sense of belonging has been seen previously in youth physical activity programs (Anderson-Butcher et al., 2013; Dowd et al., 2015) and also been used to promote physical activity by framing physical activity as a way to reduce feelings of social rejection (Dowd et al., 2014). However, as previously noted (Anderson-Butcher et al., 2013), the 4-point scale may limit exploring variability and the ability to see statistically significant changes throughout a program. In the RTQ program, feeling comfortable, committed, supported, accepted, and a part of the group appears to have led to participants attending more sessions. This is similar to attraction to group – social that was also a significant predictor. However, belonging is more about feeling a part of the group, while cohesion focuses on what drives the group to “stick” together. Together, the finding that both higher ATG-S and belonging significantly promoted attendance suggests that the social aspects of the group helped motivate participants to attend the program.  Results identified that lower leadership scores predicted higher attendance. Interviews from RTQ’s first year found that coaches played a key role in motivating participants to achieve both smoking cessation and physical activity goals, which was particularly salient for participants who completed the program (Glowacki et al., 2018). The emphasis on the importance of having coaches characterized by being encouraging, supportive, and having a non-judgmental attitude makes the findings of the current study surprising. The year 1 results allude to the potential for transformational leadership to apply in this context, as “Transformational leaders motivate others to do more than they originally intended and often even more than they thought possible. They set more challenging expectations and typically achieve higher performances... Moreover, transformational leaders empower followers and pay attention to their individual needs and personal development, helping followers to develop their own leadership 54  potential” (Bass & Riggio, 2006, p. 4). However, the results in this evaluation should be interpreted with caution. Suppression effects in the model may have impacted the finding. In the correlation matrix, the correlation between attendance and leadership was small and positive, while the standardized regression coefficient was small and negative. Considering these results, along with the statistically significant moderate correlations between leadership and the other group-related measures, it is likely that further work is needed to reduce multicollinearity before this can be interpreted as a robust result. Additionally, mean imputation may have impacted the relationship between leadership and attendance due to the correlation with attendance being small and positive in the complete case, and small and negative in the mean imputed table. Further exploration of this measure is needed. If the finding that lower transformational leadership was related to higher attendance was not due to suppression effects, it may be explained by the measure not capturing what participants described as valuable characteristics of their coaches, or other factors such as the group may have been more important. Characteristics of the coach were not measured in this evaluation but may have helped to assess the impact of the coach on participants’ success. Some of the highlighted characteristics of the coaches, such as being a previous smoker (Glowacki et al., 2018), that participants found helpful were not assessed. As seen in a community-based exercise program for older adults, coach characteristics such as gender, age, experience, and personality can be predictive of attendance (Hawley-Hague et al., 2014). Considering coach characteristics and experience, including smoking history, of the coaches may have helped to further explain variance in attendance.   Using another construct, such as personality, instead of transformational leadership may have better captured what was impactful about the coach. Both Hawley-Hague et al. (2014) and 55  Gainforth and colleagues (2018) used measures based on the “big five” personality traits to assess the relationship between practitioner personality and client quit rates in behavioural support interventions for smoking cessation. Clients with practitioners who had higher extraversion scores were more likely to be abstinent four weeks after a target quit date (Gainforth et al., 2018). The authors suggested extraversion, defined by being lively, enthusiastic, and social (Cheng & Furnham, 2002), may improve success rates because practitioners may be better able to interact with clients, improve client satisfaction, and apply behaviour change techniques confidently. These traits are contrary to the findings of Hawley-Hague and colleagues (2014) in an exercise setting, which found that coaches who had higher scores for “extraversion”, “agreeable”, and “intellectual” had participants with lower class attendance, while those with higher “conscientiousness” traits had higher attendance. Additional experimental research would be needed to examine the mechanisms behind practitioner personality and outcomes in smoking cessation behavioural support interventions. Further exploration into whether a different measure would better capture the impact of the coach on attendance could help to explain the present study’s results. Perceived similarity to other participants was not statistically significant in this study. Previous research found that deep- and surface-level similarity were significant predictors of ATG-S (Dunlop & Beauchamp, 2011), which was a statistically significant predictor of attendance in this study. However, similarity to other group members in this study had a small negative and non-significant beta in the regression. This direction of the regression result is surprising, due to perceived similarity previously being connected to attendance (Dunlop & Beauchamp, 2011). Future research could further explore whether similarity improves attendance directly, or through promoting ATG-S.  56  4.1 Implications Overall, the results of this study suggest that both individual and group-related variables play a role in predicting attendance in a smoking cessation and physical activity intervention. Individual-related factors such as age, gender, nicotine dependence, and baseline MVPA were consistent with previous literature. These factors are non-modifiable but give an indication for whom the program may be currently effective. However, as these characteristics are more often associated with success in the smoking cessation literature, consideration should be given as to how RTQ could be adapted for younger adults, women, and those with higher nicotine dependence. Lower baseline MVPA scores predicting attendance is potentially explained by considering the group context and could suggest that future iterations of the program would benefit from having different groups for different running abilities. There has been some suggestion that logistical factors may result in attrition from cessation programs, such as younger adults being more likely to be constrained by work and women affected by family duties and needing childcare (Belita & Sidani, 2015). Personal and logistical reasons have also been previously reported as reasons for drop out from RTQ (Glowacki et al., 2018), but were not quantitatively measured and therefore unable to be explored in this analysis. Further investigation of these factors is needed to understand their impact on attendance, such as by exploring barrier self-efficacy (cf. Cramp & Bray, 2009).  This study offers new insight for the smoking cessation literature. Although group-based interventions have been promoted (Raw et al., 1998; Stead et al., 2017; West et al., 2000), attrition research has focused on characteristics of individual participants (Belita & Sidani, 2015). The inclusion of group-related factors in addition to individual predictors in this evaluation identified that group-related variables added to the prediction of attendance. In 57  particular, the social aspects of the program, ATG-S and belonging, were statistically significant. Consistent with qualitative results from RTQ, the social experience of being in a group may promote higher attendance and success (Glowacki et al., 2018). However, results should be interpreted with consideration of potential suppression effects in the model. Further research that accounts for potential multicollinearity of the group-related variables could help to explain the impact of the coaches’ transformational leadership on attendance.  The impact of the group provides a focus for potential modifications in future iterations of RTQ, such as providing coach training to enhance group cohesion and sense of belonging through group activities, such as team building (Burke et al., 2008). Promoting perceptions of belonging and ATG-S for all participants could result in them attending more RTQ sessions, offering greater exposure to the smoking cessation curriculum and opportunity for more physical activity. Previous work has used group dynamics principles, such as team building, to facilitate greater group cohesion and adherence (Burke, Carron, Eys, Ntoumanis, & Estabrooks, 2006; Burke et al., 2008). These group dynamics activities have included partner work, group goal setting, and strategies to enhance communication and interaction (Burke et al., 2008). Exercise groups with team building, compared to control groups, had greater ATG-T, as well as fewer drop outs and late arrivals to sessions (Spink & Carron, 1993). A large, community-based walking program, Walk Kansas, that aimed to increase physical activity levels employed strategies such as having groups create unique, self-determined names, having both individual and group goals each week, recruiting pre-existing groups based on location or occupation, weekly newsletters, and group discussions (Estabrooks, Bradshaw, Dzewaltowski, & Smith-Ray, 2008). The program was successful at increasing physical activity for those who were inactive or insufficiently active at the beginning of the program.  58  The RTQ program employed several of these strategies, such as promoting having a buddy register alongside the participant aiming to quit, group discussions each week, and weekly emails from coaches. Further work could be done in RTQ to explore the effectiveness of these strategies, whether they were consistently implemented in different groups, and if some groups were more receptive to them than others. Although these strategies were encouraged, the training provided did not directly emphasize the importance of team building for participants to achieve program outcomes. There is potential to further emphasize the importance of team building in RTQ and to intentionally train coaches to implement strategies to enhance group dynamics. Additionally, evaluating whether implementation of these strategies impacts group dynamics and measures of program success would help to better understand their role in in RTQ. Considering the modest amount of variance that this model explained, there are factors not included in this evaluation that could help to better predict attendance. These may include coach-related factors (Hawley-Hague et al., 2014), including smoking cessation-specific characteristics such as history of smoking (Glowacki et al., 2018). Participants often cited logistical and personal reasons such as work, traffic, and injury as reasons for dropping out of the program (Glowacki et al., 2018). These factors were not measured in this evaluation, but could be considered in future iterations. Belita and Sidani (2015) also suggested that, due to a large focus on individual characteristics, exploring logistical factors such as access to childcare and transportation, as well as contextual factors such as recruitment strategies and characteristics of personnel, should also be explored. Future evaluations could investigate whether these factors impact attendance in RTQ. Future research on RTQ could also explore the mechanisms and relationships between group-related measures in a smoking cessation context. For example, the relationship between 59  similarity and cohesion (Dunlop & Beauchamp, 2011), and between cohesion and the leader’s behaviours (Caperchione, Mummery, & Duncan, 2011; Izumi et al., 2015; Loughead & Carron, 2004). There were moderate, statistically significant correlations between similarity and cohesion subscales (r’s >.33, n = 339, p’s <.01), and average TTQ and cohesion subscales (r’s >.41, n = 339, p’s <.01) that were not explored in this study but that could be avenues for future research. 4.2 Limitations The generalizability of the findings is limited by the homogeneity of the sample and convenience sampling. For example, most participants were women, middle-aged, and educated. This may also impact some measures, such as perceived similarity, which may have low variability or have a ceiling effect. The limitations of the sample characteristics and sampling should be kept in mind when applying the results to other contexts.  In line with the rationale for this study, there was high attrition with 41.1% of participants completing the program and 64.7% completing week 3. Participant drop out prior to week 3 meant that group-related predictors of program completion for those participants were not able to be assessed. The impact of the leader and group on participants who did not return after week 2 was unable to be investigated in this study, and it is worth noting that the majority of participants were excluded from this evaluation because of not completing the week 3 survey. Group-related measures may have captured why these participants dropped out, and the inability to include their data may have diluted the results of this study. Some group-related predictors can be assessed early in a program. For example, surface-level similarity, specifically related to age, was a significant predictor of attendance in a group-based postnatal exercise classes (Beauchamp et al., 2012). This may have been due to surface-level similarity being most important in the earliest sessions before deep-level similarity is more well-known later in the program as the 60  group gets to know each other. Including group-related measures earlier in the RTQ evaluation may have presented different results.  Additionally, the timing of measurement may have impacted results, and changes in group-related measures after week 3 were not assessed. Measurement of group-related constructs at week 3 was consistent with previous literature on group cohesion that assumed some time is needed in order for cohesion to develop (Spink & Carron, 1994). However, previous investigation into whether cohesion changed throughout an exercise program highlighted the potential impact of time on results (Dunlop, Falk, & Beauchamp, 2013). In that study, scores on social dimensions varied between measurements, with ATG-S decreasing then increasing, while GI-S increased overall. In contrast, task cohesion dimensions were relatively constant. These results suggest that if cohesion had been measured at a later time, the social dimensions may have had a different relationship with attendance. Missing data were addressed; however, imputation strategies are not without limitations. Using mean values to replace missing values is a conservative approach that results in less variance, which can result in smaller correlations with other variables (Tabachnick & Fidell, 2013). Although a small amount of data were missing, these limitations should still be recognized when interpreting results. The analysis of transformational leadership and similarity may have also impacted the results. The transformational leadership subscales were collapsed into one average score in this study due to high correlations between the subscales. Following the same reasoning, surface- and deep-level similarity scores were also collapsed into an overall similarity score. This reduced the likelihood of multicollinearity but did not allow for investigation of the different dimensions. 61  Finally, some of the measures used have not yet been validated for the adult setting. For example, the Transformational Teaching Questionnaire was originally created for use in school-based physical education setting and the PAGEQ was originally developed for older adults in physical activity settings, which may not capture the smoking aspect of RTQ (Estabrooks & Carron, 2000). Although transformational leadership has been applied in work settings with adults, a different measure that was not specific to physical activity was used. The measure by Beauchamp and colleagues (2010) has not yet been validated for used with adults, although select items from the TTQ were included in a workplace study of adults (Robertson & Carleton, 2018) and more recently a modified version was used with a sample of adults with spinal cord injury (Martin Ginis, Shaw, Stork, Battalova, & McBride, 2018).  4.3 Strengths  Strengths of this study included that the RTQ program was large scale with over 700 participants included in the data set and over 300 participants meeting inclusion criteria for the present study. Most participants who attended the program took part in the evaluation. This sample size provides sufficient power for the analyses, and can help to assume normality and meet other assumptions for the analyses (Field, 2013). The large-scale nature can also help to offer insight into the scalability of the program since it was implemented to a large number of people at multiple sites over several years and in ‘real-life’ community settings.  The investigation of group-related factors extends the current cessation literature which has focused primarily on individual-level predictors. Investigating the roles of both individual and group-level predictors of attendance in a smoking cessation program has the potential to identify priorities for adapting future interventions, identity who interventions may be most 62  effective for, and may help to explain why participants in more cohesive or similar groups, or who have more transformational leaders, have more success in group-based program.   63  Chapter 5: Conclusion   This study investigated predictors of attendance in a large, multi-year smoking cessation and physical activity intervention. There have been many group-based smoking cessation interventions (Stead et al., 2017), however, the majority of research on attrition in smoking cessation interventions has focused on individual participant characteristics (Belita & Sidani, 2015). Building on group dynamics research from physical activity and exercise programs, this evaluation added to the cessation literature by demonstrating that both individual and group-related variables were significant predictors of attendance, and that including group-related variables helped to better predict attendance. Although the overall prediction of attendance was modest, statistically significant individual predictors of higher attendance were being older, male, having lower week 1 nicotine dependence, and lower week 1 MVPA, while group-related predictors were having higher perceptions of belonging, higher scores on the individual attraction to group-social cohesion dimension, and lower transformational leadership scores. The ‘group’ nature of this physical activity and cessation approach was important to consider and, given the modifiable nature of constructs such as cohesion and belongingness, they should be targets for interventions to increase attendance.   64  References Ahluwalia, J., Richter, K., Mayo, M., Ahluwalia, H., Choi, W., Schmelzle, K., & Resnicow, K. (2002). African American smokers interested and eligible for a smoking cessation trial: Predictors of not returning for randomization. Annals of Epidemiology, 12(3), 206–212. https://doi.org/doi:10.1016/S1047-2797(01)00305- 2 Akoglu, H. (2018). User’s guide to correlation coefficients. Turkish Journal of Emergency Medicine, 18(3), 91–93. https://doi.org/10.1016/j.tjem.2018.08.001 Anderson-Butcher, D., & Conroy, D. E. (2002). Factorial and criterion validity of scores of a measure of belonging in youth development programs. Educational and Psychological Measurement, 62(5), 857–876. https://doi.org/10.1177/001316402236882 Anderson-Butcher, D., Iachini, A., Riley, A., Wade-Mdivanian, R., Davis, J., & Amorose, A. J. (2013). Exploring the impact of a summer sport-based youth development program. Evaluation and Program Planning, 37, 64–69. https://doi.org/10.1016/j.evalprogplan.2013.01.002 Azevedo, M. R., Araújo, C. L. P., Reichert, F. F., Siqueira, F. V., da Silva, M. C., & Hallal, P. C. (2007). Gender differences in leisure-time physical activity. International Journal of Public Health, 52(1), 8–15. https://doi.org/doi:10.1007/s00038-006-5062-1 Bass, B. M., & Riggio, R. E. (2006). Transformational Leadership (Second). Psychology Press. Bauman, A. E., Reis, R. S., Sallis, J. F., Wells, J. C., Loos, R. J. F., Martin, B. W., … Sarmiento, O. L. (2012). Correlates of physical activity: Why are some people physically active and others not? The Lancet, 380(9838), 258–271. https://doi.org/10.1016/S0140-6736(12)60735-1 Baumeister, R. F., & Leary, M. R. (1995). The need To belong: Desire for interpersonal 65  attachments as a fundamental human-motivation. Psychological Bulletin, 117(3), 497–529. https://doi.org/0033-2909 Beauchamp, M. R., Barling, J., & Morton, K. L. (2011). Transformational teaching and adolescent self-determined motivation, self-efficacy, and intentions to engage in leisure time physical activity: A randomised controlled pilot trial. Applied Psychology: Health and Well-Being, 3(2), 127–150. https://doi.org/10.1111/j.1758-0854.2011.01048.x Beauchamp, M. R., Barling, J., Zhen Li, Morton, K. L., Keith, S. E., & Zumbo, B. D. (2010). Development and psychometric properties of the transformational teaching questionnaire. Journal of Health Psychology, 15(8), 1123–1134. https://doi.org/10.1177/1359105310364175 Beauchamp, M. R., Dunlop, W. L., Downey, S. M., & Estabrooks, P. A. (2012). First impressions count: Perceptions of surface-level and deep-level similarity within postnatal exercise classes and implications for program adherence. Journal of Health Psychology, 17(1), 68–76. https://doi.org/10.1177/1359105311408156 Beauchamp, M. R., Ruissen, G. R., Dunlop, W. L., Estabrooks, P. A., Harden, S. M., Wolf, S. A., … Rhodes, R. E. (2018). Group-based physical activity for older adults (GOAL) randomized controlled trial: Exercise adherence outcomes. Health Psychology, 37(5), 451–461. https://doi.org/10.1037/hea0000615 Belita, E., & Sidani, S. (2015). Attrition in Smoking Cessation Intervention Studies: A Systematic Review. The Canadian Journal of Nursing Research, 47(4), 21–40. Belsey, D. A., Kuh, E., & Welsch, R. (1980). Regression diagnostics: Identifying influential data and sources of collinearity. New York: Wiley. Brouwer, R., & Pomerleau, C. (2000). “Prequit attrition” among weight-concerned women 66  smokers. Eating Behaviours, 1(2), 145–151. https://doi.org/10.1016/ S1471-0153(00)00014-3 Burke, S. M., Carron, A., Eys, M., Ntoumanis, N., & Estabrooks, P. (2006). Group versus individual approach? A meta-analysis of the effectiveness of interventions to promote physical activity. Sport and Exercise Psychology Review, 2(1), 19–35. Burke, S. M., Carron, A. V., & Shapcott, K. M. (2008). Cohesion in exercise groups: an overview. International Review of Sport and Exercise Psychology, 1(2), 107–123. https://doi.org/10.1080/17509840802227065 CAN-ADAPTT. (2011). Canadian Smoking Cessation Clinical Practice Guideline: Overview of summary statements. Canadian Cancer Society. (2013). For Smokers Who Want to Quit - One Step at a Time. Caperchione, C., Mummery, W. K., & Duncan, M. (2011). Investigating the relationship between leader behaviours and group cohesion within women’s walking groups. Journal of Science and Medicine in Sport, 14(4), 325–330. https://doi.org/10.1016/j.jsams.2011.03.005 Carron, A. V. (1982). Cohesiveness in Sport Groups: Interpretations and Considerations. Journal of Sport Psychology, 4(2), 123–138. Carron, A. V., & Spink, K. S. (1995). The group size-cohesion relationship in minimal groups. Small Group Research, 26(1), 86–105. Centers for Disease Control & Prevention. (2011). Quitting Smoking Among Adults - United States, 2001-2010. Morbidity and Mortality Weekly Report, 60, 1513–1519. Cheng, H., & Furnham, A. (2002). Personality, peer relations, and self-confidence as predictors of happiness and loneliness. Journal of Adolescence, 25(3), 327–339. https://doi.org/doi:10.1006/jado.2002.0475 67  Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences (2nd ed.). New York, New York: Lawrence Erlbaum Associates. Copeland, A., Martin, P., Geiselman, P., Rash, C., & Kendzor, D. (2006). Predictors of pretreatment attrition from smoking cessation among pre- and postmenopausal, weight-concerned women. Eating Behaviours, 7(3), 243–251. https://doi.org/10.1016/j.eatbeh.2005.10.001 Cramp, A. G., & Bray, S. R. (2009). A prospective examination of exercise and barrier self-efficacy to engage in leisure-time physical activity during pregnancy. Annals of Behavioral Medicine, 37(3), 325–334. https://doi.org/10.1007/s12160-009-9102-y Curtin, L., Brown, R., & Sales, S. (2000). Determinants of attrition from cessation treatment in smokers with a history of major depressive disorder. Psychology of Addictive Behaviours, 14(2), 134–142. https://doi.org/doi:10.1037//0893- 164X.14.2.134 Deaner, R. O., Geary, D. C., Puts, D. A., Ham, S. A., Kruger, J., Fles, E., … Grandis, T. (2012). A Sex Difference in the Predisposition for Physical Competition: Males Play Sports Much More than Females Even in the Contemporary U.S. PLoS ONE, 7(11), e49168. https://doi.org/10.1371/journal.pone.0049168 Dobrescu, A., Bhandari, A., Sutherland, G., & Dinh, T. (2017). The Costs of Tobacco Use in Canada, 2012. Ottawa. Dorner, T. E., Tröstl, A., Womastek, I., & Groman, E. (2011). Predictors of short-term success in smoking cessation in relation to attendance at a smoking cessation program. Nicotine & Tobacco Research, 13(11), 1068–1075. https://doi.org/10.1093/ntr/ntr179 Dowd, A. J., Chen, M. Y., Jung, M. E., & Beauchamp, M. R. (2015). “Go Girls!”: psychological and behavioral outcomes associated with a group-based healthy lifestyle program for 68  adolescent girls. Translational Behavioral Medicine, 5(1), 77–86. https://doi.org/10.1007/s13142-014-0285-9 Dowd, A. J., Schmader, T., Sylvester, B. D., Jung, M. E., Zumbo, B. D., Martin, L. J., & Beauchamp, M. R. (2014). Effects of Social Belonging and Task Framing on Exercise Cognitions and Behavior. Journal of Sport & Exercise Psychology, 36, 80–92. https://doi.org/http://dx/doi.org/10.1123/jsep.2013-0114 Dunlop, W. L., & Beauchamp, M. R. (2011). Does similarity make a difference? Predicting cohesion and attendance behaviors within exercise group settings. Group Dynamics, 15(3), 258–266. https://doi.org/10.1002/anie.198104611 Dunlop, W. L., Falk, C. F., & Beauchamp, M. R. (2013). How dynamic are exercise group dynamics? Examining changes in cohesion within class-based exercise programs. Health Psychology, 32(12), 1240–1243. https://doi.org/10.1037/a0030412 Estabrooks, P. A., Bradshaw, M., Dzewaltowski, D. A., & Smith-Ray, R. L. (2008). Determining the impact of walk Kansas: Applying a team-building approach to community physical activity promotion. Annals of Behavioral Medicine, 36(1), 1–12. https://doi.org/10.1007/s12160-008-9040-0 Estabrooks, P. A., & Carron, A. V. (1999). Group Cohension in Older Exercises: Prediction and Intervention Effects. Journal of Behavioral Medicine, 22(6), 575–588. https://doi.org/10.1023/A Estabrooks, P. A., & Carron, A. V. (2000). The Physical Activity Group Environment Questionnaire: An instrument for the assessment of cohesion in exercise classes. Group Dynamics, 4(3), 230–243. https://doi.org/10.1037/1089-2699.4.3.230 Etringer, B. D., Gregory, V. R., & Lando, H. A. (1984). Influence of group cohesion on the 69  behavioral treatment of smoking. Journal of Consulting and Clinical Psychology, 52(6), 1080–1086. https://doi.org/10.1037/0022-006X.52.6.1080 Faseru, B., Richter, K. P., Scheuermann, T. S., & Park, E. W. (2018). Enhancing partner support to improve smoking cessation. Cochrane Database of Systematic Reviews (Online), (8), CD002928. https://doi.org/10.1002/14651858.CD002928.pub4.www.cochranelibrary.com Faul, F., Erdfelder, E., Buchner, A., & Lang, A.-G. (2009). Statistical power analyses using G*Power 3.1: Tests for correlation and regression analyses. Behavior Research Methods, 41(4), 1149–1160. https://doi.org/10.3758/BRM.41.4.1149 Fidler, J., & West, R. (2011). Enjoyment of smoking and urges to smoke as predictors of attempts and success of attempts to stop smoking: a longitudinal study. Drug and Alcohol Dependence, 115(1–2), 30–34. Field, A. (2013). Discovering Statistics Using IBM SPSS Statistics. (M. Carmichael, Ed.) (4th ed.). London, England: SAGE Publications. Forey, B. A., Thornton, A. J., & Lee, P. N. (2011). Systematic review with meta-analysis of the epidemiological evidence relating smoking to COPD, chronic bronchitis and emphysema. BMC Pulmonary Medicine, 11, 36. https://doi.org/10.1186/1471-2466-11-36 Fraser, S. N., & Spink, K. S. (2002). Examining the Role of Social Support and Group Cohesion in Exercise Compliance. Journal of Behavioral Medicine, 25(3), 233–249. https://doi.org/10.1023/A:1015328627304 Gainforth, H. L., Aujla, S. Y., Beard, E., Croghan, E., & West, R. (2018). Associations between Practitioner Personality and Client Quit Rates in Smoking Cessation Behavioural Support Interventions. Journal of Smoking Cessation, 13(2), 103–109. https://doi.org/10.1017/jsc.2017.10 70  Garcia-Rodriguez, O., Secades-Villa, R., Florez-Salamanca, L., Okuda, M., Liu, S.-M., & Blanco, C. (2013). Probability and predictors of relapse to smoking: Results of the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC). Drug and Alcohol Dependence, 132(3), 479–485. https://doi.org/https://doi.org/10.1016/j.drugalcdep.2013.03.008 Garriguet, D., Tremblay, S., & Colley, R. C. (2015). Comparison of Physical Activity Adult Questionnaire results with accelerometer data. Statistics Canada, Catalogue No. 82-003-X • Health Reports, 26(7), 11–17. Geraghty, A., Torres, L., Leykin, Y., Perez-Stable, E., & Munoz, R. (2012). Understanding attrition from international Internet health interventions: A step towards global eHealth. Health Promotion International, 28(3), 442–452. Glowacki, K., O’Neill, M., Priebe, C. S., & Faulkner, G. (2018). ‘When you put the Group and the Running Together. . .’: A Qualitative Examination of Participant Experiences of the Canadian Run to Quit program. Journal of Smoking Cessation, (May), 1–7. https://doi.org/10.1017/jsc.2018.13 Godtfredsen, N. S., & Prescott, E. (2011). Benefits of smoking cessation with focus on cardiovascular and respiratory comorbidities. Clinical Respiratory Journal, 5(4), 187–194. https://doi.org/10.1111/j.1752-699X.2011.00262.x Government of Canada. (2019). Canadian Tobacco, Alcohol and Drugs Survey (CTADS): summary of results for 2017. Retrieved May 16, 2019, from https://www.canada.ca/en/health-canada/services/canadian-tobacco-alcohol-drugs-survey/2017-summary.html Greaney, M. L., Puleo, E., Sprunck-Harrild, K., Haines, J., Houghton, S. C., & Emmons, K. M. 71  (2018). Social Support for Changing Multiple Behaviors: Factors Associated With Seeking Support and the Impact of Offered Support. Health Education & Behavior, 45(2), 198–206. https://doi.org/10.1177/1090198117712333 Gwaltney, C. J., Metrik, J., Kahler, C. W., & Shiffman, S. (2009). Self-Efficacy and Smoking Cessation: A Meta-Analysis. Psychology of Addictive Behaviors, 23(1), 56–66. https://doi.org/10.1037/a0013529 Haasova, M., Warren, F. C., Ussher, M. H., Janse Van Rensburg, K., Faulkner, G., Cropley, M., … Taylor, A. H. (2014). The acute effects of physical activity on cigarette cravings: Exploration of potential moderators, mediators and physical activity attributes using individual participant data (IPD) meta-analyses. Psychopharmacology, 231(7), 1267–1275. https://doi.org/10.1007/s00213-014-3450-4 Harden, S. M., McEwan, D., Sylvester, B. D., Kaulius, M., Ruissen, G., Burke, S. M., … Beauchamp, M. R. (2015). Understanding for whom, under what conditions, and how group-based physical activity interventions are successful: a realist review. BMC Public Health. https://doi.org/10.1186/s12889-015-2270-8 Harrison, D. A., Price, K. H., & Bell, M. P. (1998). Beyond Relational Demography : Time and the Effects of Surface- and Deep-Level Diversity on Work Group Cohesion. The Academy of Management Journal, 41(1), 96–107. Hartmann-Boyce, J., Chepkin, S. C., Ye, W., Bullen, C., & Lancaster, T. (2019). Nicotine replacement therapy versus control for smoking cessation ( Review ), (5). https://doi.org/10.1002/14651858.CD000146.pub5.www.cochranelibrary.com Hawley-Hague, H., Horne, M., Campbell, M., Demack, S., Skelton, D. A., & Todd, C. (2014). Multiple levels of influence on older adults’ attendance and adherence to community 72  exercise classes. Gerontologist, 54(4), 599–610. https://doi.org/10.1093/geront/gnt075 Heatherton, T. F., Kozlowski, L. T., Frecker, R. C., & Fagerstrom, K.-O. (1991). The Fagerstrom Test for Nicotine Dependence: a revision of the Fagerstrom Tolerance Questionnaire. British Journal of Addiction, 86(9), 1119–1127. https://doi.org/10.1111/j.1360-0443.1991.tb01879.x Hoeppner, B. B., Kelly, J. F., Urbanoski, K. A., & Slaymaker, V. (2011). Comparative utility of a single-item versus multiple-item measure of self-efficacy in predicting relapse among young adults. Journal of Substance Abuse Treatment, 41(3), 305–312. https://doi.org/10.1016/j.jsat.2011.04.005 Hox, J. J. (2010). Multilevel Analysis: Techniques and Applications (2nd ed.). New York, NY: Routledge. Hughes, J. R., Peters, E. N., & Naud, S. (2008). Relapse to smoking after 1 year of abstinence: A meta-analysis. Addictive Behaviors, 33(12), 1516–1520. https://doi.org/https://doi.org/10.1016/j.addbeh.2008.05.012 Iliceto, P., Fino, E., Pasquariello, S., D’Angelo Di Paola, M. E., & Enea, D. (2013). Predictors of success in smoking cessation among Italian adults motivated to quit. Journal of Substance Abuse Treatment, 44(5), 534–540. https://doi.org/10.1016/j.jsat.2012.12.004 International Agency for Research on Cancer. (2012). Personal Habits and Indoor Combustions: A review of human carcinogens. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans, 100E, 46–167. Izumi, B. T., Schulz, A. J., Mentz, G., Israel, B. A., Sand, S. L., Reyes, A. G., … Diaz, G. (2015). Leader Behaviors, Group Cohesion, and Participation in a Walking Group Program. American Journal of Preventive Medicine, 49(1), 41–49. 73  https://doi.org/https://doi.org/10.1016/j.amepre.2015.01.019 Jacquart, J., Papini, S., Davis, M. L., Rosenfield, D., Powers, M. B., Frierson, G. A., … Smits, J. A. (2017). Identifying attendance patterns in a smoking cessation treatment and their relationships with quit success. Drug and Alcohol Dependence, 174, 65–69. https://doi.org/10.1016/j.drugalcdep.2017.01.007 Johansson, A., Halling, A., & Hermansson, G. (2003). Indoor and outdoor smoking: Impact on children’s health. European Journal of Public Health, 13, 61–66. Lando, H. A., & McGovern, P. G. (1991). The influence of group cohesion on the behavioral treatment of smoking: A failure to replicate. Addictive Behaviors, 16(3–4), 111–121. https://doi.org/10.1016/0306-4603(91)90003-Z Lee, P. N., Forey, B. A., & Coombs, K. J. (2012). Systematic review with meta-analysis of the epidemiological evidence in the 1900s relating smoking to lung cancer. BMC Cancer, 12, 385. https://doi.org/10.1186/1471-2407-12-385 Leeman, R., Quiles, Z., Molinelli, L., Terwal, D., Nordstrom, B., Garvey, A., & Kinnunen, T. (2006). Attrition in a multi-component smoking cessation study for females. Tobacco Induced Disease, 3(2), 59–71. https://doi.org/10.1186/1617- 9625-3-2-59 Loprinzi, P. D., & Walker, J. F. (2015). Nicotine Dependence, Physical Activity, and Sedentary Behavior among Adult Smokers. North American Journal of Medical Sciences, 7(3), 94–99. Loughead, T. M., & Carron, A. V. (2004). The mediating role of cohesion in the leader behavior - Satisfaction relationship. Psychology of Sport and Exercise, 5(3), 355–371. https://doi.org/10.1016/S1469-0292(03)00033-5 Martin Ginis, K. A., Shaw, R. B., Stork, M. J., Battalova, A., & McBride, C. B. (2018). Pilot study of a training program to enhance transformational leadership in spinal cord injury 74  peer mentors. Spinal Cord Series and Cases, 4(1), 1–10. https://doi.org/doi:10.1038/s41394-018-0065-8 Muthen, B. O., & Satorra, A. (1995). Complex sample data in structural equation modeling. Sociological Methodology, 25, 267–316. https://doi.org/http://dx.doi.org/10.2307/271070 Nevid, J. S., Javier, R. A., & Moulton, J. L. (1996). Factors predicting participant attrition in a community-based, culturally specific smoking-cessation program for Hispanic smokers. Health Psychology, 15(3), 226–229. https://doi.org/10.1037/0278-6133.15.3.226 Oh, H., & Taylor, A. H. (2014). Self-regulating smoking and snacking through physical activity. Health Psychology, 33(4), 349–359. https://doi.org/10.1037/a0032423 Papathanasiou, G., Georgakopoulos, D., George Georgoudis, G., Spyropoulos, P., Perrea, D., & Evangelou, A. (2007). Effects of chronic smoking on exercise tolerance and on heart rate-systolic blood pressure product in young healthy adults. European Journal of Cardiovascular Prevention and Rehabilitation, 14(5), 646–652. Poirier, A. E., Ruan, Y., Grevers, X., Walter, S. D., Villeneuve, P. J., Friedenreich, C. M., & Brenner, D. R. (2019). Estimates of the current and future burden of cancer attributable to active and passive tobacco smoking in Canada. Preventive Medicine, 122, 9–19. https://doi.org/10.1016/j.ypmed.2019.03.015 Priebe, C. S., Atkinson, J., & Faulkner, G. (2016). Run to Quit: Program design and evaluation protocol. Mental Health and Physical Activity, 11, 38–45. https://doi.org/10.1016/j.mhpa.2016.09.003 Priebe, C. S., Atkinson, J., & Faulkner, G. (2017). Run to Quit: An evaluation of a scalable physical activity-based smoking cessation intervention. Mental Health and Physical Activity, 13, 15–21. https://doi.org/10.1016/j.mhpa.2017.08.001 75  Priebe, C. S., Wunderlich, K. B., Atkinson, J., & Faulkner, G. E. J. (2020). Multi-year outcomes of the physical activity-based smoking cessation intervention Walk or Run to Quit: A three-year evaluation throughout the program’s “scaling up” phase [Unpublished Manuscript]. Prokhorov, A. V., Hudmon, K. S., Moor, C. A. de, Kelder, S. H., Conroy, J. L., & Ordway, N. (2001). Nicotine dependence, withdrawal symptoms, and adolescents’ readiness to quit smoking. Nicotine & Tobacco Research, 3, 151–155. Raw, M., McNeill, A., & West, R. (1998). Smoking Cessation Guidelines for Health Professionals: A guide to effective smoking cessation interventions for the health care system. Thorax, 53(Supplement 5), S1–S18. https://doi.org/10.1136/thx.53.2008.s1 Resnick, B., & Jenkins, L. S. (2000). Testing the Reliability and Validity of the Self-Efficacy for Exercise Scale. Nursing Research, 49(3), 154–159. Roberts, V., Maddison, R., Simpson, C., Bullen, C., & Prapavessis, H. (2012). The acute effects of exercise on cigarette cravings, withdrawal symptoms, affect, and smoking behaviour: Systematic review update and meta-analysis. Psychopharmacology, 222(1), 1–15. https://doi.org/10.1007/s00213-012-2731-z Robertson, J. L., & Carleton, E. (2018). Uncovering How and When Environmental Leadership Affects Employees’ Voluntary Pro-environmental Behavior. Journal of Leadership & Organizational Studies, 25(2), 197–210. https://doi.org/https://doi.org/10.1177/1548051817738940 Rodgers, W. M., Murray, T. C., Wilson, P. M., Hall, C. R., & Fraser, S. N. (2008). Evidence for a multidimensional self-efficacy for exercise scale. Research Quarterly for Exercise and Sport, 79(2), 222–234. https://doi.org/10.1080/02701367.2008.10599485 Shiffman, S., Brockwell, S. E., Pillitteri, J. L., & Gitchell, J. G. (2008). Individual differences in 76  adoption of treatment for smoking cessation: Demographic and smoking history characteristics. Drug and Alcohol Dependence, 93, 121–131. https://doi.org/10.1016/j.drugalcdep.2007.09.005 Smith-Ray, R. L., Mama, S., Reese-Smith, J. Y., Estabrooks, P. A., & Lee, R. E. (2012). Improving Participation Rates for Women of Color in Health Research: The Role of Group Cohesion. Prevention Science, 13(1), 27–35. https://doi.org/10.1007/s11121-011-0241-6 Smith, P. H., Bessette, A. J., Weinberger, A. H., Sheffer, C. E., & McKee, S. A. (2016). Sex/gender differences in smoking cessation: A review. Preventive Medicine, 92, 135–140. https://doi.org/10.1016/j.ypmed.2016.07.013 Spink, K. S., & Carron, A. V. (1992). Group cohesion and adherence in exercise classes. Journal of Sport & Exercise Psychology, 14, 78–86. Spink, K. S., & Carron, A. V. (1993). The effects of team building on the adherence patterns of female exercise participants. Journal of Sport & Exercise Psychology, 15(1), 39–49. Spink, K. S., & Carron, A. V. (1994). Group cohesion effects in exercise classes. Small Group Research, 25(1), 26–42. Stead, L. F., Carroll, A., & Lancaster, T. (2017). Group behaviour therapy programmes for smoking cessation. Cochrane Database of Systematic Reviews, (3), CD001007. https://doi.org/10.1002/14651858.CD001007.pub3.www.cochranelibrary.com Stead, L. F., Koilpillai, P., Fanshawe, T., & Lancaster, T. (2016). Combined pharmacotherapy and behavioural interventions for smoking cessation (Review), (3). https://doi.org/10.1002/14651858.CD008286.pub3.www.cochranelibrary.com Stenling, A., & Tafvelin, S. (2014). Transformational Leadership and Well-Being in Sports: The Mediating Role of Need Satisfaction. Journal of Applied Sport Psychology, 26(2), 182–196. 77  https://doi.org/10.1080/10413200.2013.819392 Sterner, W. (2011). What is missing in counseling research? Reporting missing data. Journal of Counseling and Development, 89(1), 56–62. https://doi.org/10.1002/j.1556-6678.2011.tb00060.x Stevens, J. P. (2002). Applied multivariate statistics for the social sciences (4th ed.). Hillsdale, NJ: Erlbaum. Stevens, M., Rees, T., Coffee, P., Haslam, S. A., Steffens, N. K., & Polman, R. (2018). Leaders promote attendance in sport and exercise sessions by fostering social identity. Scandinavian Journal of Medicine and Science in Sports, 28(9), 2100–2108. https://doi.org/10.1111/sms.13217 Tabachnick, B. G., & Fidell, L. S. (2013). Using Multivariate Statistics (6th ed.). Boston, MA: Pearson Education. Taylor, D. H., Hasselblad, V., Henley, S. J., Thun, M. J., & Sloan, F. A. (2002). Benefits of smoking cessation for longevity. American Journal of Public Health, 92(6), 990–996. https://doi.org/10.2105/AJPH.92.6.990 Taylor, G., McNeill, A., Girling, A., Farley, A., Lindson-Hawley, N., & Aveyard, P. (2014). Change in mental health after smoking cessation: Systematic review and meta-analysis. BMJ (Online), 348, 1–22. https://doi.org/10.1136/bmj.g1151 The International Physical Activity Group. (2005). Guidelines for Data Processing and Analysis of the International Physical Activity Questionnaire (IPAQ) – Short and Long Forms. Retrieved from https://sites.google.com/site/theipaq/scoring-protocol U.S. Department of Health and Human Services. (2014). The Health Consequences of Smoking: 50 Years of Progress: A Report of the Surgeon General. Atlanta, GA. Retrieved from 78  https://www.ncbi.nlm.nih.gov/books/NBK179276/pdf/Bookshelf_NBK179276.pdf Ussher, M. H., Faulkner, G. E. J., Angus, K., Hartmann-Boyce, J., & Taylor, A. H. (2019). Exercise interventions for smoking cessation. Cochrane Database of Systematic Reviews, 2019(10). https://doi.org/10.1002/14651858.CD002295.pub6 Ussher, M. H., Taylor, A., & Faulkner, G. (2014). Exercise interventions for smoking cessation. Cochrane Database of Systematic Reviews, (8). https://doi.org/10.1002/14651858.cd002295.pub5. Van Rensburg, K. J., Taylor, A., & Hodgson, T. (2009). The effects of acute exercise on attentional bias towards smoking-related stimuli during temporary abstinence from smoking. Addiction, 104(11), 1910–1917. https://doi.org/10.1111/j.1360-0443.2009.02692.x Vangeli, E., Stapleton, J., Smit, E. S., Borland, R., & West, R. (2011). Predictors of attempts to stop smoking and their success in adult general population samples: A systematic review. Addiction, 106(12), 2110–2121. https://doi.org/10.1111/j.1360-0443.2011.03565.x Wenig, J. R., Erfurt, L., Kröger, C. B., & Nowak, D. (2013). Smoking cessation in groups-who benefits in the long term? Health Education Research, 28(5), 869–878. https://doi.org/10.1093/her/cyt086 West, R. (2003). Bupropion SR for smoking cessation. Expert Opinion on Pharmacotherapy, 4(4), 533–540. https://doi.org/10.1517/14656566.4.4.533 West, R., McNeill, A., & Raw, M. (2000). Smoking cessation guidelines for health professionals: An update. Thorax, 55(12), 987–999. Retrieved from http://discovery.ucl.ac.uk/110749/ Williams, S. L., & French, D. P. (2011). What are the most effective intervention techniques for changing physical activity self-efficacy and physical activity behaviour - And are they the 79  same? Health Education Research, 26(2), 308–322. https://doi.org/10.1093/her/cyr005 Woods, M., Harris, K., Mayo, M., Catley, D., Scheibmeir, M., & Ahluwalia, J. (2002). Participation of African Americans in a smoking cessation trial: A quantitative and qualitative study. Journal of the National Medical Association, 94(7), 609–618. World Health Organization. (2004). Policy Recommendations for Smoking Cessation and Treatment of Tobacco Dependence. Retrieved from https://www.who.int/tobacco/resources/publications/en/intro_chapter3.pdf?ua=1 Zhou, X., Nonnemaker, J., Sherrill, B., Gilsenan, A. W., Coste, F., & West, R. (2009). Attempts to quit smoking and relapse: factors associated with success or failure from the ATTEMPT cohort study. Addictive Behaviors, 34, 365–373.  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0389777/manifest

Comment

Related Items