IA Copilot a noté :
Dafotec Récupération De Données

Le 2025-04-26 03:59:32

Using the social ecological model to build a path analysis model of physical activity in a sample of active US college students Using the social ecological model to build a path analysis model of physical activity in a sample of active US college students Jonathan J. Stewart West Virginia University, jjstewart@mix.wvu.edu Follow this and additional works at: https://researchrepository.wvu.edu/etd Part of the Health Psychology Commons, and the Sports Studies Commons Recommended Citation Stewart, Jonathan J., "Using the social ecological model to build a path analysis model of physical activity in a sample of active US college students" (2020). Graduate Theses, Dissertations, and Problem Reports. 7652. https://researchrepository.wvu.edu/etd/7652 This Dissertation is protected by copyright and/or related rights. It has been brought to you by the The Research Repository @ WVU with permission from the rights-holder(s). You are free to use this Dissertation in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you must obtain permission from the rights-holder(s) directly, unless additional rights are indicated by a Creative Commons license in the record and/ or on the work itself. This Dissertation has been accepted for inclusion in WVU Graduate Theses, Dissertations, and Problem Reports collection by an authorized administrator of The Research Repository @ WVU. For more information, please contact researchrepository@mail.wvu.edu. Using the social ecological model to build a path analysis model of physical activity in a sample of active US college students Jonathan Stewart, M.S. Dissertation submitted to the College of Physical Activity and Sport Sciences at West Virginia University in partial fulfillment of the requirements for the degree of Doctorate of Philosophy in Kinesiology With an emphasis in Sport and Exercise Psychology Sam Zizzi, Ed.D. (Committee Chair) Christa Lilly, Ph.D. Scott Barnicle, Ph.D. Jack C. Watson II, Ph.D. Department of Sport Sciences Morgantown, West Virginia 2019 Keywords: physical activity, alcohol use, college, social ecological model, achievement goal Copyright 2019 Jonathan Stewart Abstract Using the social ecological model to build a path analysis model of physical activity in a sample of active US college students Jonathan Stewart, M.S. Objective: To examine how achievement goal orientation, perceived barriers and benefits, self efficacy, on-campus residence, transportation, and binge drinking impact physical activity. Participants: Five hundred and twenty (70.23% female) college students participated in the study during Fall 2014. Methods: Students completed an online questionnaire that measured environmental and psychosocial factors, and physical activity behaviors. Results: A path analysis revealed that self-efficacy, episodes of binge drinking, use of active transportation, and use of public transportation all had significant direct effects on physical activity. Meanwhile, perceived barriers had a significant negative direct effect on physical activity. Conclusion: Results indicate that both environmental and psychological factors influence engagement in physical activity. iii Acknowledgements I would like to start by thanking my family for being a consistent source of support and motivation. Clay and Logan, you have both helped keep me motivated and grounded throughout this journey. I am incredibly proud of you both. Mom and Dad, you have both taught me so much. To my whole family, you all have given me so much over the years and I appreciate you all more than you know. Next I would like to thank my WVU family for making this adventure a memorable one. Thank you for listening to me when I needed to vent, for teaching me to dress in the cold weather, for the early morning office chats, and the company during those late night writing sessions. You have all left a permanent imprint on my life in one way or another. I am a better person for the time that we all had together. Aaron and Zenzi. Thank you for being consistent sources of guidance, motivation, and laughs no matter the physical distance. Carra, thanks for challenging me and providing wisdom I didn’t even know I needed. I would also like to thank all of my Public Health colleagues and peers. My experiences in Public Health have shaped my view of the world in ways I never could have imagined. To the WVU and SEP faculty, thank you for giving me the opportunity to pursue my dream. Your trust and faith in me helped me grow into the person that I am today. Dr. Zizzi, thank you for your unwavering support, especially near the end. Your trust in me helped me learn to trust myself. Thank you for meeting me at the base of the mountain and guiding me to the top. Dr. Lilly, thank you for your guidance, support, and (perhaps most of all) your patience. Your willingness to go above and beyond is greatly appreciated. Dr. Watson, thank you for being there throughout my time in Morgantown and not letting me disappear once I left town. Dr. Barnicle, thank you for your support and for challenging me take a different perspective. Dr. Etzel, your wise words continue to resonate with me both personally and professionally. Ashley, you helped me find my passion in applied work and reminded me why I wanted to study SEP in the first place. For that I will be forever grateful. Dr. Giacobbi, thank you helping me to develop my research skills as I was first embarking on this PhD journey and for helping introduce me to the world of Public Health. I would also like to thank all of colleagues at the SRC. Nancy, thank you for always providing a break from academia, your mentorship, and professional and general life guidance. You always said what I needed to here, regardless of whether or not I wanted to hear it. I’ll always be grateful for that and for everything you have done for me. Last, but certainly not least, thank you to my R2 friends and family. Hayley, James, Brittney, Matt, Will, and Patrick. Thank you for being a steady source of support, motivation, a sounding board, and for putting up with me as I finished this journey. iv Table of Contents Introduction ..................................................................................................................................... 1 Socioecological Model ................................................................................................................ 4 Environmental Factors ................................................................................................................ 5 Individual Factors ........................................................................................................................ 6 Methods........................................................................................................................................... 9 Research Design .......................................................................................................................... 9 Instruments ................................................................................................................................ 10 Data Analysis ............................................................................................................................ 13 Results ........................................................................................................................................... 14 Discussion ..................................................................................................................................... 16 Limitations ................................................................................................................................ 21 Future Directions ....................................................................................................................... 22 Application ................................................................................................................................ 24 Table 1 – Sample Characteristics .................................................................................................. 39 Table 2 – Descriptive Statistics ..................................................................................................... 41 Table 3 - Correlations ................................................................................................................... 43 Table 4 – Fit Indices ..................................................................................................................... 45 Table 5 – Indirect Effects .............................................................................................................. 46 Figure 1 – Hypothesized Model .................................................................................................... 47 Figure 2 – SAS Revised Model .................................................................................................... 48 Figure 3 – Final Model ................................................................................................................. 49 Appendix A. Extended Review of the Literature .......................................................................... 50 References ..................................................................................................................................... 77 Running Head: A MODEL OF PHYSICAL ACTIVITY 1 Introduction Physical inactivity and alcohol consumption are two risk factors commonly associated with the development of a number of chronic diseases and premature death (Lee et al., 2012; Warburton et al., 2006; World Health Organization, 2018). Alcohol consumption, including an increased number of daily drinks, drinking frequency, and heavy episodic drinking, has been associated with academic problems, injuries, and risky behavior in college students and emerging adults (18-24 years of age) (Hingson, 2017; Kuperberg & Padgett, 2017; Rinker et al., 2016). In contrast, physical activity can serve as a protective factor and reduce the risk of developing chronic diseases such as diabetes, heart disease, and colon cancer (Kyu et al., 2016; Lee, Sesso, Oguma, & Paffenbarger, 2003). Individuals who are physically active in early adulthood (18-22 years old) are more likely to be physically active later in life (Nogueira et al., 2009). Thus, physical activity behaviors and increased frequency of drinking during the college years (i.e. generally 18-24 years of age) have the potential to have lasting impacts later in life. The American College Health Association (2019) defined recommended levels of physical activity as a minimum of 30 minutes of moderate intensity cardio or aerobic exercise on 5 or more days per week, or at least 20 minutes of vigorous cardio or aerobic exercise on 3 or more days per week. The American College Health Association (2019) reported data from 54,497 students in the Spring 2019 National College Health Assessment II (ACHA-NCHA II). In this large sample, 46.2% of undergraduate students met recommended levels of exercise based on self-reported data (American College Health Association, 2019). The ACHA semi-annual survey includes items assessing a wide variety of health- and academic-related variables using a comprehensive self-report survey. Colleges and universities use this data to compare self reported behaviors on their campus to national norms. A MODEL OF PHYSICAL ACTIVITY 2 One of the most common health behaviors studied along with physical activity in US college students is alcohol consumption. The American College Health Association (2019) recently stated that 55.8% of undergraduate college students reported consuming an alcoholic beverage within the previous 30 days. Among the students who consumed alcohol, 33.3% reported doing something they regretted while drinking in the past 12 months. Furthermore, Soedamah-Muthu, De Neve, Shelton, Tielemans, and Stamatakis (2013) have reported a joint association between alcohol consumption, level of physical activity, and risk of cardiovascular mortality and all-cause mortality. Physical activity was measured in MET hours/week. METs (or metabolic equivalents) represent energy expenditure at different intensities (i.e. 1 MET represents sitting quietly). Alcohol was measured in units (1 unite = 8g of ethanol or approximately 4 oz. of wine or 8 oz. of beer). When physical activity was low (.1 to 5 MET-hours/week for males and .1 to 4 MET-hours/week for females) and alcohol consumption was high (>35 units/week for males and >21 units/week for females) there is an increased risk of cardiovascular mortality (HR 1.95) and all-cause mortality (HR 1.64). However, the researchers found that when physical activity was high (>5 MET-hours/week for males and >4 MET hours/week for females), high alcohol intake was not linked to increased risk of cardiovascular mortality. Thus, both alcohol intake and physical activity are important for reducing risk of all cause mortality and cardiovascular mortality. However, given that high alcohol intake was not associated with increased risk of cardiovascular mortality in the presence of high levels of physical activity, physical inactivity may be a larger contributing factor to the risk of cardiovascular mortality than alcohol consumption. Contrary to what may be expected, level of physical activity is commonly associated with alcohol consumption in college students. Students who self-reported consumption of alcohol within the past 30 days were 40% more likely to have used the campus recreation facility A MODEL OF PHYSICAL ACTIVITY 3 compared to those who had not consumed alcohol (Miller, Noland, Rayens, & Staten, 2008). The amount of alcohol consumed while binge drinking (r=.13) and self-reported level of drinking (r=.08) have been found to be positively associated (though weakly) with leisure time physical activity (Stuntz, Smith, & Vensel, 2017). Graupensperger, Wilson, Bopp, and Blair Evans (2018) found that alcohol consumption was associated with vigorous, but not moderate, physical activity across a six month study. Despite these associations, the underlying mechanism for the relationship between alcohol consumption and physical activity is unclear. Some hypothesize that students may engage in physical activity to compensate for the alcohol consumed while drinking (Abrantes et al., 2017; Graupensperger et al., 2018). A variety of other factors have been linked to alcohol consumption in college students, including location of residence, social influence, and alcohol related norms (Abrantes et al., 2017; Arterberry, Smith, Martens, Cadigan, & Murphy, 2014; Graupensperger et al., 2018; Weitzman, Nelson, & Wechsler, 2003; Yoon, Kim, & Lee, 2014). For example, researchers have investigated the relationship between protective behavioral strategies, alcohol related norms, and alcohol behavior in a sample of college students (Arterberry et al., 2014). The researchers noted that social norms, such as perceived alcohol consumption among other students, were positively associated with alcohol use. These studies did not examine important environmental factors in relation to physical activity, such as transportation. Given the positive association between alcohol consumption and levels of physical activity, it is important to have a better understanding of this relationship, and the factors that impact both behaviors to reduce alcohol consumption without reducing physical activity. Therefore, there is a need to approach these behaviors through a framework such as the social ecological model. This will allow for the inclusion of different factors, at multiple levels, that influence engagement in health behaviors. A MODEL OF PHYSICAL ACTIVITY 4 Socioecological Model The simple provision of physical activity recommendations by organizations and the government has been ineffective at increasing physical activity behaviors at the population level (Guthold et al., 2018; Pratt et al., 2015; Sallis et al., 2016; Schwartz et al., 2019). Despite increased awareness of these recommendations, from 2001 to 2016, the level of insufficient physical activity in high income countries increased from 31.6% to 36.8% (Guthold et al., 2018). Thus, researchers have emphasized the application of theoretical frameworks to the study of physical activity (R. E. Rhodes et al., 2019). For example, in a recent review, researchers summarized randomized (RCT) and non-randomized (NRCT) control trials that promoted physical activity in university students (Maselli et al., 2018). Researchers utilized more than one theory to inform intervention design in multiple trials. All but one of the effective interventions addressed multiple components of physical activity behavior, however, the majority of these studies focused on individual or interpersonal factors and excluded environmental factors, which are often the most dynamic and complex in nature. The central theory of a social ecological model is that behavior is the result of various nested levels of interpersonal, intrapersonal, and environmental influences (Bronfenbrenner, 1977; Sallis et al., 2008; Spence & Lee, 2003). Over time, physical activity researchers have begun to favor the incorporation of multiple levels of influence. Bauman et al. (2012) examined reviews of physical activity with a focus on individual, interpersonal, environmental, regional or national policies, and global factors across a wide array of age groups and cultures. The authors of this review note that both environmental and personal factors may influence physical activity behavior. Thus, the inclusion of multiple environmental, psychosocial, and behavioral factors may provide additional insight into physical activity behavior beyond a more singular focus (Bauman et al., 2012; Sallis et al., 2006; Spence & Lee, 2003). To accomplish this goal, A MODEL OF PHYSICAL ACTIVITY 5 researchers may need to use more advanced statistical methods that allow several factors across levels to be evaluated simultaneously. Environmental Factors Researchers have examined environmental factors associated with physical activity by including the relationship between residence (on/off campus), distance of residence from campus recreation centers, and level of physical activity (Allen & Ross, 2013; Castle, Alman II, Kostelnik, & Smith, 2015; Essiet, Baharom, Shahar, & Uzochukwu, 2017; Miller et al., 2008; Reed & Phillips, 2005; Staten, Miller, Noland, & Rayens, 2005; Watson, Ayers, Zizzi, & Naoi, 2006; Yoon et al., 2014). Students who lived on-campus or within one mile of the campus recreation facility were more likely to use the facility compared to those who lived off-campus or over one mile away (Castle et al., 2015; Watson et al., 2006) and typically report higher levels of physical activity compared to those who live off campus or further away (Miller et al., 2008; Staten et al., 2005; Yoon et al., 2014). In a random sample of 899 undergraduate students, those who lived on campus were 44% (OR=1.44) more likely to use the facility compared to those who lived off campus (Miller et al., 2008). In short, students living on campus tend to engage in higher levels of physical activity compared to those living off-campus. Environmental factors, such as location of residence (e.g. on-campus), are not only positively associated with physical activity, but have also been associated with increased alcohol consumption in college students (Castle et al., 2015; Staten et al., 2005; Yoon et al., 2014). Environmental factors can contribute to first year students beginning to binge drink (e.g. 5 or more drinks for males) in college. Weitzman and colleagues (2003) used national data to determine factors associated with binge drinking in freshman students. They found that first year students who lived in coed on-campus (OR=1.90) or Greek housing (OR=2.85) were significantly more likely to begin binge drinking compared to students who lived off-campus A MODEL OF PHYSICAL ACTIVITY 6 with a roommate (OR=.82) or with parents (OR=.40). Similarly, in another study, the authors reported female students who lived on-campus consumed more alcoholic beverages than female students who lived off-campus (Yoon et al., 2014). In line with the socioecological framework, community and individual level factors such as transportation, self-efficacy and motivation may also play an important role in determining physical activity behavior. For example, in a study with college students in Ireland, Murphy and colleagues (2019) found that students who had a longer travel time (lived 10 minutes further from their university) were less likely to be classified as active commuters (OR=.59), participate in physical activity only at the university (OR=.80), or fall in the high physically active cluster (OR=.58). At the same time, an increase in motivation (e.g. feeling motivated to be physically active) increased the likelihood that students would participate in physical activity only at the university (OR=1.13) or be placed in the high active cluster (OR=1.27). In other words, for every one unit increase on the Likert scale in motivation, students were 27% more likely to be classified in the high active cluster. This may be especially important given the negative relationship between living further from campus and physical activity. Therefore, in addition to modes of transportation, motivation may be an important intrapersonal factor that can influence behavior in conjunction with environmental factors. Individual Factors Researchers have reported that motivation may facilitate beneficial beliefs about physical activity and lead to sustained behavior in college students (Kilpatrick et al., 2003; Watson et al., 2006; Zizzi et al., 2006). Achievement goal theory describes how individuals define success in specific achievement contexts, such as exercise, and thus how they are motivated to reach their goals (Nicholls, 1989; Roberts et al., 1998). In achievement goal theory, achievement goals aren’t just targets, rather, they represent an orientation toward tasks that include associated views A MODEL OF PHYSICAL ACTIVITY 7 about success, effort, ability, and purpose (Pintrich, 2000). Specifically, task-oriented motivation (success occurs with learning and mastery) was positively associated with exercise intensity, years exercising, and exercise enjoyment. On the other hand, ego orientation (success is defined through comparison to others or some standard) was not significantly correlated to these constructs (Kilpatrick et al., 2003). Moreover, research shows that as college students transition toward maintenance (i.e., sustain recommended guidelines of 150 minutes of physical activity for more than 6 months) of physical activity, level of task focus continues or increases, while reliance on ego reference cues decreases (Zizzi et al., 2006). In a study involving 569 college students, Zizzi et al. (2006) found that, students who exercised regularly were more likely to be in the high task/high ego group than the low task/low ego group. Additionally, task orientation has been associated with the belief that success was related to effort, interest, and adaptive achievement strategies (Biddle et al., 2003; Duda & Nicholls, 1992). Thus, a task involvement may yield sustained effort, more adaptive behaviors, and persistence in physical activity engagement compared to ego involvement, which has been a consistent finding in the literature (Biddle et al., 2003; Duda, 1989; Duda & Nicholls, 1992; Kilpatrick et al., 2003). In addition to motivation, self-efficacy is often found to be related to level of physical activity (Maselli et al., 2018; Young et al., 2014). Self-efficacy refers to an individual’s belief in their ability to take the actions necessary to cope with a situation or achieve a desired outcome (Bandura, 1982) and has been associated with increased physical activity in college students (Shaikh et al., 2018). Specifically, these researchers observed that exercise self-efficacy was positively associated with days of strenuous physical activity. Two factors that may influence an individual’s self-efficacy and level of physical activity are perceived benefits and barriers (Bandura, 1982; Grubbs & Carter, 2002; Horacek et al., 2018; A MODEL OF PHYSICAL ACTIVITY 8 King et al., 2014). Common benefits of physical activity reported by college students include improved physical appearance, physical fitness, and health (Grubbs & Carter, 2002; King et al., 2014). College students also face a number of barriers to physical activity such as lack of knowledge, self-efficacy, time, and resources (Sukys et al., 2019). In a sample of 480 college students, King and colleagues (2014) noted that perceived benefits were positively associated with vigorous physical activity, while perceived barriers were negatively associated with vigorous physical activity. However, other researchers have suggested that the impact of perceived barriers outweighs the impact of perceived benefits on physical activity (Hurley et al., 2018). Thus, further investigation is needed to clarify the role of perceived benefits and barriers in determining physical activity behavior. Although previous research has established the relationship between psychological and environmental factors and physical activity, they have traditionally focused on these factors separately from each other and/or have not used path analysis to develop a model of these factors. Thus, the purpose of the current study is to utilize a socioecological framework to investigate the relationship between environmental and psychological correlates of physical activity. More specifically, we will examine how achievement goal orientation, perceived barriers, perceived benefits, self-efficacy, on-campus residence, transportation, and binge drinking are related to physical activity. A secondary purpose is to explore the interactions between achievement goal orientation and barrier self-efficacy, as well as the interactions between on-campus residence, transportation, and binge drinking. Figure 1 represents a diagram of the proposed model. Perceived benefits, barriers, and self-efficacy are associated with physical activity behavior (Grubbs & Carter, 2002; King et al., 2014; Maselli et al., 2018; Shaikh et al., 2018). Individuals with a high task orientation may persist in the face of challenges and barriers to A MODEL OF PHYSICAL ACTIVITY 9 physical activity (Biddle et al., 2003; Kilpatrick et al., 2003). Thus, the relationship between achievement orientation and physical activity may be mediated by perceived barriers and benefits. Previous researchers have established a positive association between living on-campus, alcohol consumption, and physical activity, as well as between alcohol consumption and physical activity (Castle et al., 2015; Graupensperger et al., 2018). Thus, it was hypothesized that alcohol consumption and transportation would mediate the relationship between living on-campus and physical activity. In summary:  Task and ego motivations were proposed to be negatively associated with perceived barriers and positively associated with perceived benefits  Task and ego motivations were proposed to be positively associated with physical activity  Location of residence was proposed to be positively associated with alcohol consumption, use of public and active transportation, and physical activity  Self-efficacy, alcohol consumption, and use of active and public transportation were proposed to be positively associated with physical activity engagement  Self-efficacy was expected to mediate the relationship between perceived barriers, perceived benefits, and physical activity  Alcohol consumption and transportation were expected to mediate the relationship between location of residence and physical activity Methods Research Design The present study completed a secondary analysis of data collected via a cross sectional survey. Self-reported data were collected from 629 university students. The research design was quantitative and correlational in nature. A MODEL OF PHYSICAL ACTIVITY 10 Instruments The present survey was a modified version of a previously published survey that was used with similar populations (Zizzi et al., 2004, 2006). The final survey included a total of 96 questions that assessed residence (on or off-campus), transportation, physical activity, forms of exercise, barriers to exercise, confidence, support, primary reason for campus recreation facility use, desired facility improvements, goal orientation, alcohol use, and demographic information. During survey development, input was sought from experts in the field as well as staff from the university’s Student Recreation Center’s Wellness staff on several items. Achievement goal orientation. Achievement goal orientation refers to how an individual defines personal success in specific achievement contexts and thus their motivation to reach their goal success (Nicholls, 1989; Roberts et al., 1998). The Perception of Success Questionnaire for Exercise (POSQ-E; Zizzi et al., 2006) was used to measure goal orientation. The POSQ-E consists of 11-items (6 task orientation and 5 ego orientation) measured on a 4-point Likert type scale. The items are averaged to determine task and ego orientation. The higher the score on the task orientation subscale, the more the individual defines their success by personal mastery and improvement. The higher the score on the ego orientation subscale, the more the individual defines their success as outperforming others or some standard. A sentence stem of “When exercising, I feel most successful when…” was used for each item in the POSQ-E (Zizzi et al., 2006). In previous research the two subscales of the POSQ-E combined to explain approximately 65% of variance and had good internal reliability with alpha values of .87 (task) and .88 (ego). The questionnaire has also demonstrated convergent validity with stages of change for exercise participation and factor validity. Average item response was used for each subscale, task and ego. A MODEL OF PHYSICAL ACTIVITY 11 Binge drinking. Binge drinking is defined as the consumption of 5 or more alcoholic beverages for males, (4 or more for females) in one sitting (ACHA, 2014). Similarly, the Substance Abuse and Mental Health Services Administration states that binge drinking occurs on one occasion or over the course of a couple of hours (National Institute on Alcohol Abuse and Alcoholism, 2017). An alcoholic drink was defined for the participants as 12 ounces of beer, 5 ounces of wine, or 1.25 ounces of hard alcohol. The questionnaire contained four items modeled after the ACHA (2014) questions (last 30 days) and the definition of binge drinking. A binary (yes/no) question was used to assess if the participants consumed alcohol in the past 30 days. For the purpose of this study, binge drinking was assessed by asking how many times they consumed 5 or more drinks (4 drinks for females) in one sitting over the past two weeks. The number of hours for “one sitting” was not defined for participants. Physical activity. Physical activity can be defined as any physical movement that leads to an increase in energy output (R. E. Rhodes et al., 2017). Examples and definitions for moderate (brisk walking, gardening, activities that cause small increases in breathing or heart rate) and vigorous (running, aerobics, activities that cause large increases in your breathing or heart rate) physical activity were provided for participants. Physical activity was measured with modified questions from the Behavioral Risk Factor Surveillance System Survey Questionnaire (CDC, 2014). These questions included the number of days they engage in physical activity (moderate or vigorous) for at least 10 minutes and on those days, the time in minutes spent per day engaging in moderate or vigorous activity. Number of days of physical activity was multiplied by minutes to determine weekly minutes of physical activity. Weekly minutes of physical activity was rescaled (divided by 100) prior to running path analysis. Barriers to physical activity. Participants were asked how often different barriers interfere with or prevent them from exercising. Perceived barriers to physical activity were A MODEL OF PHYSICAL ACTIVITY 12 assessed using a 4-point Likert type scale (1-Never to 4-Frequently). Barriers included not having time, feeling self-conscious, and fear of injury. Similar barriers and approaches to the measurement of physical activity have been cited in previous research involving college students (Ball et al., 2018; Bray, 2007; Gyurcsik et al., 2004; Sukys et al., 2019). Item responses were summed for total perceived barriers to physical activity. Benefits of physical activity. Perceived benefits for physical activity refer to potential improvements or gains that will occur through engagement in physical activity (Brown, 2005). Participants responded to a question that asked them to rate how important different factors were in their decision to engage or not engage in physical activity. Potential benefits included, more energy, feeling less stressed, increased confidence, and improved sleep among others. Students responded on a scale from “Not at all important (1)” to “Extremely important (4)”. Item responses were summed to create total perceived benefits of physical activity. Self-efficacy. Self-efficacy, or an individual’s confidence in their ability to begin or maintain physical activity was assessed with a single question. Respondents were asked “How sure or confident are you that you can start or continue to exercise for at least 30 minutes per day at least 5 days per week?”. They responded using a 4-point Likert type scale of very unsure to very sure. Responses of 1, 2, or 3 were coded as a ‘0’ for lower self-efficacy and responses of 4 were coded as a ‘1’ for high self-efficacy for physical activity. Mode of transportation. Transportation was assessed with a single question. Students were asked “What method of transportation do you use the most to get around town?”. Response options included: walk, bike, my car, various forms of public transportation, and other. The other option included space to fill in an unlisted mode of transportation. The response options were dummy coded. Walk and bike were combined to form the ‘Active Transportation’ group. The “my car” response served as the reference group. A MODEL OF PHYSICAL ACTIVITY 13 On-campus residence. A single binary question was used to assess whether the respondent lived on or off-campus. Greek life affiliation. Membership in a fraternity or sorority was assessed with a single yes or no question. Gender. Gender was assessed with a single question. Students were asked their gender and to select either “male” or “female”. Class standing. Class standing refers to academic class level. Participants were asked to select their class standing. Options included first year student, sophomore, junior, senior (including 5th year), and graduate or professional. Data Analysis Descriptive statistics, including mean and standard deviation, and bivariate correlations were calculated. Path analysis was used to test the hypothesized model (Figure 1) in SAS v. 9.4 (Cary, NC, 2015). Due to missing data, Full Information Maximum Likelihood (FIML) was used for model estimation. The data set was checked for multicollinearity, outliers, and normality. Path analysis was chosen to examine the directionality of the relationships between the variables. Root mean square error of approximation (RMSEA), standardized root mean square residual (SRMR), comparative fit index (CFI), and χ2 were all used to assess model fit. RMSEA values less than .08, SRMR values less than .05, CFI values approaching or exceeding .95, and a nonsignificant χ2 were used to identify acceptable fit (Hooper et al., 2008; Hu & Bentler, 1999; Weston & Gore, 2006). A model with acceptable fit means the proposed model was supported. The model controlled for multiple variables, including membership in a Greek organization, gender, and class standing. Data Cleaning A MODEL OF PHYSICAL ACTIVITY 14 A total of 629 students responded to the study. However, there were 59 respondents who opened the survey and didn’t respond to any questions. There were 10 respondents found to be outliers for weekly minutes of physical activity with values greater than 871 minutes (3 standard deviations plus the mean). These cases were removed. Analysis were run to check for the assumptions of homoscedasticity, multicollinearity, and normality. Multicollinearity was not present with Tolerance less than 1 (.62 to .85) and VIF values below 10 (1.17 to 1.61). Results Participants were enrolled as full- or part-time students at a midsize mid-Atlantic university. According to university records, the institution the sample was drawn from was 48.96% female and 80.34% Caucasian/White. The majority of respondents in the current sample were female (n=330), Caucasian/White (n=423), and lived off-campus (n=365). The participants were distributed across undergraduate (Freshman, Sophomore, Junior, Senior) and graduate class standings. A small percentage of the sample were involved in club (n=40) or intramural sports (n=52). The majority of respondents participated in sports while attending high school for at least one season (n=405). Many students reported free access to a fitness facility in their residence hall or housing complex (n= 295). The sample was relatively active as 48.86% met or exceeded 180 minutes of physical activity per week. On average, respondents lived nearly 14 minutes from the campus recreation center (M=13.73, SD=9.41). Table 1 contains additional sample frequencies. Descriptive statistics and correlations were calculated for the weekly minutes of physical activity (MVPA; M=233.75, SD=162.23), perceived barriers (barriers; M=25.44, SD=6.18), binge drinking behavior (binge; M=1.84, SD=2.18), and average responses for each achievement goal orientation subscale (task: M=3.45, SD=.54; ego: M=2.24, SD=.77), among other variables. Means and standard deviations are presented in Table 2. Physical activity was positively correlated with binge drinking (r=.168, p<.01), active transport (r=.123, p=.016), self-efficacy A MODEL OF PHYSICAL ACTIVITY 15 for physical activity (r=.368, p<.0001) and being male (r=.149, p<.01). Weekly minutes of physical activity was negatively correlated with perceived barriers (r=-.372, p<.0001). Additional correlations are presented in Table 3. Prior to model testing, the physical activity variable was rescaled in order to reduce difference in scale of standard deviations among the variables and prevent analysis errors (O’Rouke & Hatcher, 2013). Physical activity values were divided by 100 and the standard deviation was reduced from 162.23 to 1.62. The initial hypothesized path model, figure 1, which depicted relationships between achievement goal orientation, perceived barriers, perceived benefits, binge drinking, living on campus, self-efficacy, and primary form of transportation was tested. Car as primary form of transportation was used as a referent category for active and public transportation. Lower self-efficacy was used as the referent group for self-efficacy. Additionally, the effects of gender, membership in a sorority or fraternity, and class standing were controlled for in the model. Female was used as the referent group for gender. The reference group for membership in a sorority or fraternity was not being a member of a fraternity or sorority. The hypothesized model did not have good model fit (χ2 = 300.58, df = 43, χ2, p<.0001, SRMR=.0789, RMSEA=.098, CFI=.685). Thus the hypothesized model was not supported. The next model tested (Figure 2) added covariances between task orientation and ego orientation, task orientation and perceived benefits, ego orientation and Greek life membership, ego orientation and gender, and perceived benefits and gender. This model showed improved fit but still did not meet “good” fit criteria (χ2 = 276.55, df = 51, χ2 p<.0001, SRMR=.075, RMSEA=.084, CFI=.724). Through further model development, constraints were placed on covariances and additional paths were drawn based on modification indices. The final model, Figure 3, included a path from public to active transit and dropped several non-meaningful paths. A MODEL OF PHYSICAL ACTIVITY 16 This resulted in a more parsimonious model with acceptable fit (χ2 = 100.57, df = 54, χ2 p<.0001, SRMR=.050, RMSEA=.037, CFI=.943). The fit indices for the models can be found in Table 4. The final model explained 24.51% of variance in physical activity. Perceived barriers had a negative statistically significant direct effect on physical activity (β = -.252, p < .001). High self efficacy (β = .286, p < .001; in reference to lower self-efficacy), binge drinking (β = .137, p = .026), active transportation (β = .158, p = .002; in reference to car), and public transportation (β = .105, p = .047; in reference to car) all had statistically significant positive direct effects on physical activity. Standardized indirect effects can be found in table 5. Perceived barriers, on campus residence, public transportation, Greek life, and task orientation had statistically significant indirect effects on physical activity. Discussion Although the original model was only partially supported, the present study did find support for a multi-path approach to predicting physical activity. Both environmental and psychological factors were found to have significant effects on physical activity behavior. This finding is consistent with the central theory of social ecologic models (Bronfenbrenner, 1977; Sallis et al., 2008). In the present sample, the effects of individual level factors were stronger in predicting physical activity than the environmental factors. This finding is consistent with previous literature in which interpersonal factors had a larger direct effect on behavior than behavior settings or perceived environment (R. E. Rhodes et al., 2019; Yen & Li, 2019). Socioecological frameworks propose that interpersonal and environmental factors are interconnected. Thus, the strength of social cognitive factors may be due in part to unobserved environmental factors such as modeled behavior and verbal persuasion (Bandura, 1982; Ickes, McMullen, Pflug, & Westgate, 2016). The data from the present study can be used to lend support to these assertions. A MODEL OF PHYSICAL ACTIVITY 17 Public transportation use had a direct effect on active transportation, and both forms of transportation had direct effects on weekly minutes of physical activity. The positive relationship between active transportation and minutes of physical activity is supported by previous research (Murphy et al., 2019). Living on-campus had a positive effect on the use of active or public transportation. It’s possible that, by living on-campus, students had shorter distances to travel and thus were more likely to utilize these forms of transportation. It’s also possible that on campus students had more convenient and regular access to public transportation. For instance, Simons et al. (2014) noted that travel time was a critical factor that influenced young adults’ decision to travel by walking or biking. The importance of travel time may, at least partially explain both the negative effect of public transportation use on active transportation, but, positive effect of public transportation on physical activity. Students in the present sample may have chosen public transportation over active transportation to get to their destination for a few reasons. The built environment for the current sample is generally not very walkable or bike friendly and the public transportation options are relatively consistent and generally accessible. The opportunity to reduce the amount of walking in unfavorable conditions may have also influenced reliance on public transportation. Despite the decision to use public transit, they likely had to rely on walking to get to their bus stop, or to their destination once they exited public transit. Typically, the choice to use public transportation begins and ends with at least a few minutes of walking. This finding adds to the college student literature because most studies have focused exclusively on psychosocial factors and ignored important contextual variables in students’ immediate environments (Maselli et al., 2018; R. E. Rhodes et al., 2019). Researchers have primarily utilized theoretical approaches that focus on the individual such as social cognitive theory, dual-process theories, and self-determination theory (Rhodes et A MODEL OF PHYSICAL ACTIVITY 18 al., 2019). While these approaches have shown some effectiveness (Rebar et al., 2016; Teixeira et al., 2012; Young et al., 2014) they fail to incorporate important environmental variables such as transportation (Bauman et al., 2012). The social ecological approach allows for the individual level factors, such as self-efficacy, to be investigated alongside more broad factors like primary mode of transportation. This is important as these findings may be used to inform future interventions. Researchers have found that incorporating multiple factors can lead to successful attempts to change physical activity behavior (Maselli et al., 2018). Thus, examining the role of these important contextual variables may better inform future interventions. In line with previous literature (Castle et al., 2015; Graupensperger et al., 2018), binge drinking was positively associated with physical activity behavior. Researchers have previously stated that physical activity may increase with alcohol consumption as a way to compensate for unhealthy behavior (Graupensperger et al., 2018). Although this may be true, similar to previous research, affiliation with a fraternity or sorority was found to have a positive direct effect on binge drinking and indirect effect on physical activity via binge drinking (Buscemi et al., 2011). This interactive effect suggests that there may be other cultural norms that can help explain the relationship between binge drinking and physical activity beyond the purging of calories. For instance, social norms, alcohol expectancies, and a lack protective behavioral strategies can all influence alcohol consumption (Barry et al., 2016; N. Rhodes et al., 2019; Tyler et al., 2017). Barry and colleagues (2016) reported members of fraternities and sororities used fewer protective behavioral strategies (i.e. alternating between alcoholic and non-alcoholic beverages) compared students who were not involved with fraternities or sororities. The belief that one should engage in drinking behavior and intent to drink has been positively associated with Greek life (N. Rhodes et al., 2019). If these injunctive norms are influencing identity development, A MODEL OF PHYSICAL ACTIVITY 19 membership in a fraternity or sorority may further impact drinking behavior (Thompson & Romo, 2016). Individual level factors also impacted engagement in physical activity. For example, self efficacy had a direct, positive effect on physical activity behavior. Belief in ability has been commonly linked to physical activity behavior (R. E. Rhodes et al., 2019), and this finding is congruent with multiple theories in which self-efficacy, or similar constructs can be found (Bandura, 2004; Hagger & Chatzisarantis, 2014; Ryan & Deci, 2017). For instance competence, similar to self-efficacy, describes the need to feel proficient and effectively interact with one’s environment (Ryan & Deci, 2017). Both competence and self-efficacy are positively associated with physical activity behavior across a number of studies, with self-efficacy usually one of the strongest psychosocial predictors of physical activity behavior (Bauman et al., 2012; Chu et al., 2019; Farren et al., 2017; Ng et al., 2012; Shaikh et al., 2018). The strength of self-efficacy as a predictor for physical activity was also highlighted in the present study’s findings. Perceived benefits had a positive direct effect on perceived barriers but not on physical activity. This finding is consistent with previous literature (King et al., 2014; Simons et al., 2014). Simons and colleagues (2014) recommended that some benefits, specifically ecological and health, should not be emphasized when attempting to promote active transportation to young adults. Perceived barriers, however, did have a significant negative effect on physical activity. Perceived barriers have been associated with reduced resistance training (Hurley et al., 2018), vigorous physical activity (King et al., 2014), and overall levels of physical activity (Horacek et al., 2018; Sukys et al., 2019). According to the health belief model, individuals will engage in a behavior based on their perceptions of benefits of, and barriers to, behavior (Tran et al., 2017). It is possible that some of the health benefits of physical activity may not have been salient for the present sample, and thus did not translate into increased intentions to be active. Perceived A MODEL OF PHYSICAL ACTIVITY 20 barriers to physical activity may have been more relevant than the perceived benefits of physical activity to the present sample. This could have led to barriers being more salient than perceived benefits. Thus perceived barriers may play a more critical role in determining physical activity engagement in college students. In contrast to the hypothesized model, there was not a significant effect by either goal orientation subscale on physical activity. However, task orientation did have a significant direct effect on perceived barriers. This indirect effect may help explain why previous researchers have reported an association between high task orientation and perseverance (Kilpatrick et al., 2005). Individuals who focus on self-improvement and mastery may see overcoming barriers as part of the process. For instance, hard work, self-improvement, and overcoming difficulties are features of task orientation. This attitude may facilitate adaptive behaviors such as time management and learning how to exercise and thus directly impact the perceived severity of barriers without directly impacting physical activity itself. Additionally, the sample was highly active which may have contributed to task and ego goal orientations not having a direct impact on physical activity. For the current sample, physical activity engagement may be more reflective of automatic processes such as implicit attitude or habit. Habits are developed over time as behaviors are repeatedly performed (Gardner, 2015; Lally et al., 2010; Wood & Neal, 2009). A large portion of the present sample participated in high school sports prior to attending college. These past behaviors may have become routinized and habitual. Rebar and colleagues (2016) suggested that when behaviors become routine, they are regulated by more automatic habitual processes beyond more conscious processes. The sample was also made up largely of female college students. Female college students typical engage in lower levels of physical activity and consume less alcohol than their male counterparts (Abrantes et al., 2017; Graupensperger et al., 2018; Miller et al., 2008; Stuntz et al., A MODEL OF PHYSICAL ACTIVITY 21 2017; Towne et al., 2017). Being a male was positively correlated with physical activity and binge drinking in the present sample. However, gender did not have a significant effect on physical activity in the final model. Despite the highly active sample, it is plausible that the large proportion of female students may have impacted the findings. Researchers have reported that a variety of factors, including alcohol consumption, can influence female college students’ physical activity differently compared to males (Davis et al., 2017; Kilpatrick et al., 2005; Shaffer et al., 2017). Thus, a more heterogeneous sample could yield different findings. In summary, the relationships between variables in the tested models supports the utilization of social ecological frameworks to investigate factors that impact physical activity in college students (Bauman et al., 2012; Sallis, 2018; Sallis et al., 2008). Individual and environmental factors can influence college students’ engagement in physical activity. As supported by the literature, self-efficacy can influence level of physical activity in college students (Farren et al., 2017). Broader environmental factors can also influence physical activity behavior (Sallis et al., 2008). The use of a social ecological framework to guide the investigation of factors that influence a more diverse sample of college students’ physical activity is needed. Limitations Multiple limitations should be considered when interpreting the results of this study. The participants were university students and were not randomly selected from the campus population. The sample was over representative of female students (70.77% of the sample compared to 48.96% of the institution’s population). The recruitment strategy did reach a broad audience and was made accessible to nearly all students, however the sample does not accurately reflect the institutional makeup. This may have impacted the findings of the current study. Furthermore, the current sample was made up of active students. The factors investigated in the current study may have different impacts on the physical activity behaviors of sedentary college A MODEL OF PHYSICAL ACTIVITY 22 students and active students. Thus these findings may not be applicable to college students who do not exercise regularly. There is, however, value in understanding how factors impact those who are more active. Similar to how individuals may watch elite athletes in order to improve performance, enhanced understanding of factors associated with increased activity in some students help sedentary students increase their physical activity engagement. The present study also used self-report measures which may be subject to social desirability biases. Both physical activity and alcohol behaviors could be over-reported, or the behaviors in this sample could be unique to those that responded. Future Directions Future studies should aim to include more diverse sample, both in terms of activity level and demographic characteristics. The presented model should be tested in both active and sedentary populations. Comparisons between the models could help researchers further determine similarities and differences between the two samples. A more heterogeneous sample will also allow for researchers to control for factors such as gender. Males tend to have higher levels of physical activity when compared to females and can have different preferences and motives for engaging in physical activity. Further examination of the roles of gender and psychosocial and environmental variables can help advance our understanding of colleges students’ physical activity behavior (Davis et al., 2017). Gathering data from multiple institutions in different regions may help to create a more diverse sample with increased generalizability (Graupensperger et al., 2018). In addition to demographic makeup, data should also be gathered from students across academic disciplines. Researchers have found that academic disciplines can influence health related behaviors (Gathman et al., 2017; Shaikh et al., 2018). More diverse sample populations can help clarify the roles of different factors in determining college students’ physical activity behaviors. A MODEL OF PHYSICAL ACTIVITY 23 The exact mechanism underlying the positive relationship between alcohol consumption and physical activity in unclear (Davis et al., 2017). Researchers have suggested that college students may engage in physical activity as a way to compensate for consuming alcohol (Davis et al., 2017; Graupensperger et al., 2018). Psychosocial and environmental factors that influence physical activity should be investigated in conjecture with their role in impacting alcohol consumption. For instance, future researchers should investigate how social norms interact with motivation and perceived barriers to impact both alcohol consumption and physical activity in college students (Horacek et al., 2018; N. Rhodes et al., 2019). This may shed light on factors that directly influence behavior or operate indirectly through self-efficacy or decisions to utilize public or active transportation. Researchers should consider investigating these variables in specific populations where established norms and behaviors may contribute to greater alcohol use, such as fraternities and sororities (Barry et al., 2016; N. Rhodes et al., 2019; Thompson & Romo, 2016). It may also be worth exploring, in these cases, the potential side-effects of these mechanisms. This could include additional behaviors such as compensatory eating (Abrantes et al., 2017). Researchers can utilize more advanced statistical techniques, beyond correlations, to better explain the underlying relationships between these variables. Techniques such as path analysis and multiple regression may help advance understanding of the relationships between environmental and individual factors and their effects on behavior. Lastly, the present findings support a social ecologic framework for understanding physical activity in college students. Nevertheless, more objective measures and longitudinal study designs can strengthen our understanding of factors that influence physical activity. Future studies should consider using more objective measurements where available (i.e. accelerometers for physical activity) (Shaikh et al., 2018; Towne et al., 2017). Objective measures of physical A MODEL OF PHYSICAL ACTIVITY 24 activity allow for more accurate measurements of frequency, duration, and intensity (Murphy et al., 2019; Towne et al., 2017). . Future researchers should also utilize more longitudinal approaches. This will allow for more investigation into causal relationships among factors associated with physical activity (Towne et al., 2017). These findings can enhance our understanding of the relationship between individual and environmental factors. Application The use of more longitudinal approaches can be used to examine the relationship between environmental changes, psychosocial factors, and behavior. For instance, researchers may investigate the relationship of different environmental factors such as availability of various forms of transportation, neighborhood walkability, and individual factors such as self-efficacy, perceived barriers, and motivation across time. These findings may be particularly useful if data collection occurs before and after campus development changes that impact students’ ability to use active or public transportation. Interventions may be designed to target specific predictors of physical activity behavior in college students. College wellness programs, administrators, and recreation and student-life coordinators may seek to identify which factors associated with physical activity can be modified. Furthermore, they may attempt to find intervention approaches that can help increase or maintain physical activity while decreasing unhealthy behaviors, such as binge drinking. Institutions may want to target various barriers faced by college students in order to help facilitate their engagement in physical activity. Barriers at the environmental, policy, and individual or interpersonal levels of the social ecological model can be addressed. For instance, colleges and universities could enact policies that require students to enroll in mandatory physical education (PE) or lifetime activity classes. This could help reduce some of the barriers students commonly report such as not having someone to exercise with and a lack of knowledge A MODEL OF PHYSICAL ACTIVITY 25 about how to exercise. Furthermore, these PE courses may help students develop more self efficacy in the ability to become active through vicarious and personal experiences. Another policy change that may increase activity is to limit the number of cars on campus. Active students may be more active because of an increased reliance on active (such as biking or walking) and public transportation over the use of cars. Moving parking to the edge of campus would likely require students to utilize active transportation to get to more central locations on campus. Symbolically this may also serve to alter the norms of transportation on campus as cars would be restricted to the outer edge. Active transportation would then become a regular behavior by while on campus, while driving would be limited to off campus activities. Colleges and universities can also leverage school pride and identity to challenge sedentary norms by seeking to establish a culture in which students choose opportunities to be active over sedentary activities. Other barriers reported by college students are not having a safe place to be active and/or a lack of time. Institutions can alter the built environment to improve safety for active transportation by increasing the number of sidewalks, adding street lights, safety patrols, and/or safety call boxes to existing and new sidewalk. The current institution has multiple campuses. Adding campus recreation centers on each campus will increase access and convenience for students. In addition to providing more gym space, institutions can provide additional well-lit recreation fields and basketball or tennis courts to provide accessible alternatives for those that aren’t on main campus. This would increase the number of safe spaces and potentially reduce transportation time. A reduction in all, or even some, of the aforementioned barriers could make engaging in physical activity a more desirable and less costly behavior, thus increasing physical activity levels across the student body. Stakeholders should take these changes, and others, into A MODEL OF PHYSICAL ACTIVITY 26 consideration in future attempts to improve levels of physical activity on college campuses. These changes should include larger environmental and policy adjustments, as well as, target individual level factors. The application of social ecological models can facilitate these efforts. A MODEL OF PHYSICAL ACTIVITY References 27 Abrantes, A. M., Scalco, M. D., O’Donnell, S., Minami, H., & Read, J. P. (2017). Drinking and exercise behaviors among college students: Between and within-person associations. Journal of Behavioral Medicine, 40(6), 964–977. https://doi.org/10.1007/s10865-017 9863-x Allen, R., & Ross, C. M. (2013). An assessment of proximity of fitness facilities and equipment and actual perceived usage by undergraduate university students: A pilot study. Recreational Sports Journal, 37(2), 123–135. https://doi.org/10.1123/rsj.37.2.123 American College Health Association. (2014). American College Health Association-National College Health Assessment II: ACHA-NCHA II Sample Survey. American College Health Association. (2019). American College Health Association-National College Health Assessment II: Undergraduate Student Executive Summary Spring 2019. American College Health Association. Arterberry, B. J., Smith, A. E., Martens, M. P., Cadigan, J. M., & Murphy, J. G. (2014). Protective behavioral strategies, social norms, and alcohol-related outcomes. Addiction Research & Theory, 22(4), 279–285. https://doi.org/10.3109/16066359.2013.838226 Ball, J. W., Bice, M. R., & Maljak, K. A. (2018). Exploring the Relationship Between College Students’ Barriers to Exercise and Motivation. American Journal of Health Studies,