Public Health Nursing Vol. 32 No. 5, pp. 430–439 0737-1209/© 2015 Wiley Periodicals, Inc. doi: 10.1111/phn.12176
LIFESPAN: POPULATION STUDIES
Dyadic Effects of Individual and Friend on Physical Activity in College Students Gwang Suk Kim, PhD, RN,1 Chung Yul Lee, PhD, RN,1 In Sook Kim, PhD, RN,1 Tae Hwa Lee, PhD, RN,1 Eunhee Cho, PhD, RN,1 Hyeonkyeong Lee, PhD, RN,1 Linda L. McCreary, PhD, RN,2 and Su Hee Kim, MSN, RN3 1
College of Nursing, Nursing Policy Research Institute, Yonsei University, Seoul, Korea; 2College of Nursing, Health Systems Science, University of Illinois at Chicago, Chicago, Illinois; and 3Department of Nursing, Graduate School of Yonsei University, Seoul, Korea Correspondence to: Su Hee Kim, Department of Nursing, Graduate School of Yonsei University, 50-1 Yonsei-ro, Seodaemun-gu, Seoul 120-752, Korea. E-mail: [email protected]
ABSTRACT Objective: Physical activity (PA) tends to decline throughout the college years, and close friends’ influence is known to be an important factor in maintaining PA. This study examined the actor effect and partner effect between an individual and his/her friend regarding the influence of self-efficacy and social support on PA among Korean college students. Design and Sample: Cross-sectional survey data from 108 pairs of individual students and friends were analyzed. Measures: The survey questionnaire measured PA, self-efficacy toward exercise, social support for PA, anxiety and depression, community environments, and perceived health status. Structural equation modeling with path analysis was conducted to test Actor-Partner Interdependence Model (APIM) explaining close relationships on PA. Results: One-sided partner effect that friends' perceived friend support was directly related to individual's PA (b = 0.20, p < .05) was revealed. Regarding actor effects, self-efficacy was directly related to higher levels of PA for individual and friend. Perceived health status was related to higher level of individuals’ PA. Conclusions: These results suggest a role for public health nurses in developing interventions for college-aged young adults that promotes friend support for PA as well as individual self-efficacy toward PA, to engage young adults in establishing lifelong health-promoting PA. Key words: college student, health status, Korean, physical activity, self-efficacy, social support.
Physical inactivity has been identified as a primary risk factor of chronic diseases, while regular physical activity (PA) reduces negative health conditions such as cardiovascular disease, respiratory disease, type 2 diabetes, and depression across the life span and in all populations (U. S. Department of Health and Human Service, 2008). In both the United States and Korea, college students’ level of PA tends to decrease throughout their college years (Huang et al., 2003; Park & Kim, 2013; Racette, Deusinger, Strube, Highstein, & Deusinger, 2008), putting them at increased risk for these negative health outcomes. College students are in the transition
period between late adolescence and adulthood developmentally, with enough unscheduled time and cognitive maturity to control their own lifestyle behaviors and form healthy habits, including PA. However, the unhealthy behavior patterns established in this period may impact on the rest of their lives (Racette et al., 2008). Compared to college students in the United States, Korean college students are less physically active. The rate of regular moderate-intensity exercise among college students in Korea was reported as 20.8% (Kim, Park, Kim, & Han, 2010), whereas the rate of regular moderate and vigorous-intensity
Kim et al.: Dyadic Effect on Physical Activity exercise in American college students was 45.5% (American College Health Association, 2009). Seo et al. (2012) also reported that Korean college students have a much higher rate of physical inactivity than other East Asian college students: 7.2% for Singapore, 13.5% for Taiwan, 16.8% for Hong Kong, and 28.5% for Korea. Hence, increasing college students’ PA has become one of Korea’s major public health priorities. Research is needed that explores the determinants of college students’ PA to help public health nurses and other professionals to develop effective interventions to promote PA. Previous studies have examined the correlates of adolescents’ and/or young adults’ PA. Those correlates can be categorized as intrapersonal factors: self-efficacy, perceived health status, mood, and body mass index (BMI); interpersonal factors: family support and friend support; and community environmental factors (Huang et al., 2003; Keating, Guan, Pinero, & Bridges, 2005; Kim et al., 2013; Thome & Espelage, 2004). Self-efficacy has been identified as an important determinant of PA, with high self-efficacy related to a high level of PA (Dunn, 2011; Kim et al., 2013). In a systematic review of the studies on determinants of college students’ PA, Keating et al. (2005) found that self-efficacy was one of the most significant contributors to PA or exercise behavior. However, they recommended further study regarding college students’ PA, because research focusing on this population has been quite limited. Social support has been considered not only as a correlate of PA, but a primary source of self-efficacy as well. Researchers reported that social support was a significant contributor to college students’ PA (Gruber, 2008; Kim et al., 2013; Wallace, Buckworth, Kirby, & Sherman, 2000). When young people participate with a friend, their motivation to be physically active was improved, and they exercised together for a longer time than when alone (Salvy et al., 2009). Gruber (2008) suggested that close friends’ values placed on PA are more powerful motivators than those of family in college students. In addition, Kim et al. (2013) showed that friends’ social support correlated positively with female college students’ PA and with their self-efficacy as well. The Actor-Partner Interdependence Model (APIM) is a model to explain the level of interde-
pendence in two-person relationships. This model hypothesizes how each person’s emotion, cognition, or health behavior affects those of the partner (Cook & Kenny, 2005; Kelly et al., 2003). Regarding health behavior, the attributes and behaviors of one person in a dyad can affect the outcomes of the other person (Campbell & Kashy, 2002). The main components of APIM are the actor effects and the partner effects. Actor effects are defined as how much a person’s current behavior is predicted by his or her own correlates. In contrast, partner effects are defined as how much a person’s current behavior is predicted by the partner’s correlates (Cook & Kenny, 2005). Thus, we can use APIM to explore the actor and partner effects in the context of dyadic intimate relationships such as romantic couple, friends, and parent-child. The most frequent application of APIM has been to describe the dyadic effects of husbands and wives on each other in health-related areas such as social support and quality of life (Thomson, Molloy, & Chung, 2012) and dementia knowledge/self-efficacy/depression, and dementia preventive behavior (Ko & Shin, 2013). We identified only one study that applied APIM to PA: Ayotte, Margrett, and Patrick (2013) reported bidirectional actor effects and partner effects of husbands’ and wives’ characteristics on both their own and each other’s PA. Although many studies have asserted the association between friends’ relationships and PA (Kim et al., 2013; Salvy et al., 2009; Wallace et al., 2000), those studies have not examined dyadic effects, including self-efficacy and social support. In addition, APIM has not been used in studies to examine the effects of individuals’ and friends’ characteristics on PA. Therefore, the purpose of this study was to examine the influence of actor effects and partner effects of Korean college students and their friends, specifically self-efficacy and social support, on their level of PA.
Methods Design and sample The ethical committee of the institutional review board approved all of the study procedures. A cross-sectional survey design was used for this study. A committee of six nursing faculty evaluated the content validity and appropriateness of ques-
Public Health Nursing
tionnaire, which was pilot tested with five students to ensure comprehensibility of the questionnaire. Minor revision was needed to improve understanding and clarity of the items; the time needed to complete the survey was 15–20 min. After receiving permission from university administration, we recruited survey participants in the student activity center, where students frequently travel in and out and most students’ clubs were located. Posters encouraged participation in the study, including the purpose and duration of the survey, and the requirement to participate together with a friend on campus. Four senior college students, including two nursing students, were trained and participated in the data collection method and process under the supervision of the investigators. Participants were asked to sign an informed consent document and complete a pen and paper survey containing items measuring PA, self-efficacy, social support, and related variables. Over a 10-day survey period, 374 questionnaires were distributed; 364 completed questionnaires were received (response rate of 97.3%). The total sample included participants categorized as singles, pairs, and groups of three or more friends. Among them, pairs were selected from those who identified each other as friends. Only 108 pairs had appropriate data to examine the dyadic effect between individuals and friends. For each pair, the first participant who appeared in the database was designated as actor and the second as partner. After the participants completed the survey questionnaire, they received the services of blood pressure check and individual health counseling for their blood pressure levels and regarding their own questions. In addition, a stationery package was provided as compensation for their time.
Measures The level of participants’ PA was assessed using the nine-item International Physical Activity Questionnaire short form (Craig et al., 2003), which is publicly available for use without permission (https:// sites.google.com/site/theipaq/). Oh, Yang, Kim, and Kang (2007) tested the validity and reliability of the Korean version of IPAQ-SF and reported moderate reliability, with a range of Spearman Rho values from .427 to .646 (median: .542). Oh et al. reported the criterion validity by showing a moderate correlation with measurements using acceler-
ometers (p = .039). To arrive at a continuous variable, the authors calculated the score of Metabolic Equivalent Task (MET)-minutes using the IPAQ Research Committee’s (2005) formula: Walking MET-minutes/week ¼ 3:3 walking minutes walking days Moderate MET-minutes/week ¼ 4:0 moderate-intensity activity minutes moderate days Vigorous MET-minutes/week ¼ 8:0 vigorous-intensity activity minutes vigorous days Total physical activity MET-minutes/week ¼ sum of Walking þ Moderate þ Vigorous MET-minutes/week scores Self-efficacy toward exercise was measured by using a modified version of Marcus, Selby, Niaura, and Rossi’s (1992) five-item scale, translated into Korean and modified by Lee and Chang (2001). Each item was rated on a 5-point scale from “not at all confident = 1” to “very confident = 5.” Higher scores indicated higher self-efficacy toward exercise. Marcus et al. reported test-retest reliability of this scale as .90, and Lee and Chang reported Cronbach’s a as .75. Cronbach’s a in this study was .79. Social support for PA was measured by influences from family and friends to promote participants’ PA. Saunders et al. (1997) originally developed the Social Influences Scale as eight dichotomous yes/no items (four items for family and four for friends) indicating whether family or friends provided direct support for PA in the following four ways: (1) expressing concern about physical activities, (2) encouraging PA, (3) proposing to exercise together, and (4) doing exercise together. The total score of each yes response was computed separately for the family and friends subscales. Possible scores ranged from 0 to 4, and higher scores were indicative of higher level of social support. Saunders et al. (1997) reported the internal consistency of this test as .75 and test-retest reliability as .78. Cronbach’s alpha coefficient for the family support was .63, while for the friend support was .73.
Kim et al.: Dyadic Effect on Physical Activity Participants rated their perceived health status by answering the question: “In general, would say your health status is excellent, good, moderate, fair, or poor” to measure general health status. This question consistently has been shown to be a valid measure of perceived health status (Sibthorpe, Anderson, & Cunningham, 2001). An additional question, “Compared with others your same age, would say your health status is excellent, good, moderate, fair, or poor?” was used to measure relative health status. Anxiety and depression were measured by the Hospital Anxiety and Depression Scale (HADS), developed by Zigmond and Snaith (1983). There are 14 items, seven measuring anxiety and seven for depression, rated on a 4-point scale from 0 to 3; higher scores indicated higher levels of anxiety and depression. Oh, Min, and Park (1999) translated and adapted the HADS to be culturally appropriate for use with Korean samples. Cronbach’s a for the anxiety subscale was .85, while for the depression subscale a was .75. Community environments for PA were measured by the Perceived Neighborhood Environments scale. This scale was originally developed by De Bourdeaudhuij, Sallis, and Saelens (2003) as a brief nine-item assessment tool. Roh, Lee, Lee, and Kim (2012) translated it into Korean, and tested scale validity and reliability. This scale was composed of items regarding accessibility to stores or transportation, connectivity of street networks, sidewalks for walking and cycling, safety for walking, existence of neighborhood space for PA, aesthetics of the streets, and recreation facilities. Each item was rated on a 4-point scale from “strongly disagree = 1” to “strongly agree = 4.” Higher scores indicated a more activity-friendly neighborhood. Cronbach’s a in this study was .63.
Analytic strategy Demographic and other variables’ distributions were analyzed using frequencies, means and ranges as appropriate. The associations between individual’s PA-related factors and friend’s PA-related factors were analyzed using Pearson’s correlations. Structural equation modeling (SEM) with path analysis was used to test actor effect and partner effect of self-efficacy, social support, and correlates on PA between individual and friend. SEM with path analysis was an appropriate test of whether
the APIM fit the data to explain the effect of the close friend relationships on health behavior. To evaluate the model fit, the v2, root mean square error of approximation (RMSEA), and comparative fit index (CFI) were analyzed. Adequate model fit is indicated by the p-value of v2 > .05, a RMSEA < .08, and CFI ≥ .90 (Hu & Bentler, 1999; Tabachnick & Fidell, 2000). In all analyses, a pvalue of