Social control and strenuous exercise among late adolescent college students: Parents versus peers as influence agents

Journal of Adolescence 37 (2014) 543e554

Contents lists available at ScienceDirect

Journal of Adolescence journal homepage: www.elsevier.com/locate/jado

Social control and strenuous exercise among late adolescent college students: Parents versus peers as influence agents John A. Pugliese a, *, Morris A. Okun b a b

Department of Social & Behavioral Sciences, Dixie State University, St. George, UT 84770, USA Department of Psychology, Arizona State University, Tempe, AZ 85287-1104, USA

a b s t r a c t Keywords: Social control Physical activity Exercise Late adolescence Peers Parents

In the context of a model of health-related social control, we compared the associations among social control strategies, affective and behavioral reactions, and exercise for parental and peer influence agents. Late adolescent college students (n ¼ 227) completed questionnaires that focused on social control from a parent or a peer who attempted to increase their exercising. Results from this cross-sectional study revealed that most relationships in the model were similar for parent and peer influence agents, however, (a) negative social control was a stronger predictor of reactance among parents than peers; (b) positive affect was a stronger predictor of attempts to change among peers than parents; and (c) positive affect predicted frequency of strenuous exercise only among parents. Decreasing parents’ use of negative social control strategies and increasing adolescents’ positive affective reactions to parental social control agents may be keys to promoting positive lifestyle changes in late adolescence. Ó 2014 The Foundation for Professionals in Services for Adolescents. Published by Elsevier Ltd. All rights reserved.

Regular physical activity is essential for healthy development as youth transition from adolescence to adulthood. Nevertheless, according to the U.S. Department of Health and Human Services (HHS, 2008, 2011), 39% of young adults between 18 and 24 years old do not meet national recommended physical activity levels for adults. The evidence suggests that young adults in college may experience increased weight gain, and reduced physical activity and exercise (Racette, Deusinger, Strube, Highstein, & Deusinger, 2005). Moreover, research has found considerable continuity between physical activity levels in adolescence, and later adulthood (Hallal, Victoria, Azevedo, & Wells, 2006; Perkins, Jacobs, Barber, & Eccles, 2004). Thus, a sedentary and physically inactive lifestyle in college is a risk factor for many chronic diseases during adulthood, including heart disease, stroke, colon cancer, diabetes, and osteoporosis (U.S. Department of Health and Human Services, 2000). Socialization and exercise in late adolescence Socialization primarily occurs within the context of close relationships, and developmental researchers have traditionally focused on the influence of the family (Brewer & Caporael, 1990; Cooper, 1994). Although parents remain salient as socializers, peers become increasingly important agents of social influence throughout adolescence (Collins & Steinberg, 1998). The shift toward peers yielding more social influence is evident in the increasing amounts of time adolescents spend with peers than

* Corresponding author. Tel.: þ1 435 879 4634. E-mail address: [email protected] (J.A. Pugliese). http://dx.doi.org/10.1016/j.adolescence.2014.04.008 0140-1971/Ó 2014 The Foundation for Professionals in Services for Adolescents. Published by Elsevier Ltd. All rights reserved.

544

J.A. Pugliese, M.A. Okun / Journal of Adolescence 37 (2014) 543e554

with parents (Larson & Verma, 2000). Some of the domains in which peers exert influence include personal appearance, interpersonal behavior, entertainment, and leisure activities (Hartup, 1996). Within the exercise and physical activity domain, a few studies have compared the relative influence of peers versus parents. For example, research focusing on social support has consistently found that peer support is more predictive of adolescent physical activity than parental support (Beets, Vogel, Forlaw, Pitetti, & Cardinal, 2006; Prochaska, Rodgers, & Sallis, 2002). Within college-age samples, research has tended to focus solely on the support of peers when examining influences on exercise (Gruber, 2008; Okun et al., 2003). Despite the power of peer norms on youth physical activity (Okun, Karoly, & Lutz, 2002), gaps still remain in our understanding of parental and peer influence on physical activity among college age adolescents. In the context of a health-related model of social control, the present study examines type of influence agent (parent versus peer) as a moderator of the direct and indirect relations between influence strategies and strenuous exercise among late adolescents. The decision to focus only on strenuous exercise was based on the finding that social support and social negativity contribute much more to the prediction of strenuous exercise than to the prediction of moderate or mild exercise (Okun et al., 2003). The model of health-related social control Social control refers to deliberate efforts to influence and regulate the behavior of another person (Rook, 1995; Rook & Pietromonaco, 1987). Social control is conceptually different from social support; the latter focusing on the psychological, physical, and financial help provided by others in an individual’s social network (Lin, Simeone, Ensel, & Kuo, 1979; Sarason, Sarason, & Pierce, 1994). Explicit acts of health-related social control occur when network members attempt to persuade an individual to increase health-enhancing behaviors or reduce health-compromising behaviors. Positive social control strategies involve providing information, positively reinforcing healthy behavior, and showing concern for the recipient. Negative social control strategies, in contrast, involve behaviors such as ridiculing, pressuring, and making an unflattering comparison between the recipient and someone else (Lewis & Rook, 1999). Tucker and Anders (2001) proposed that positive social control is positively related to both positive affect and engaging in positive health behaviors whereas negative social control is associated with both negative affect and negative behavioral reactions such as ignoring the social influence attempts, doing the opposite of what the social influence agent wants, and hiding unhealthy behavior. Furthermore, they proposed that positive affect mediates the association between positive social control and engaging in healthy behaviors and that negative affect mediates the associations between negative social control and taking no action, hiding unhealthy behavior, and increasing the frequency of unhealthy behavior. Several tests of components of this meditational model have demonstrated its utility (Logic, Okun, & Pugliese, 2009; Okun, Huff, August, & Rook, 2007; Tucker, 2002; Tucker & Anders, 2001; Tucker, Elliott, & Klein, 2006; Tucker, Orlando, Elliott, & Klein, 2006). In studies of the consequences of health-related social control, researchers have taken one of two approaches. One approach has been to examine a single social control agent such as spouses or romantic partners (Logic et al., 2009; Okun et al., 2007; Tucker & Anders, 2001). The second approach has been to allow participants to choose their most important social control agent and then to ignore type of social control agent in the analyses (Lewis & Rook, 1999; Tucker, 2002; Tucker, Elliott, et al., 2006; Tucker, Orlando, et al., 2006). In contrast, the present study focused on participants who indicated that either parents or peers were currently exerting the most influence on their exercise behavior. The type of social control agent was treated as a moderator in a model examining the relationships among social control strategies, affect, behavioral reactions, and strenuous exercise. Similar to prior research, both the direct and indirect relations between social control strategies and affective reactions on strenuous exercise were examined. The hypothesized mediational model of the association between social control strategies and frequency of strenuous exercise Fig. 1 depicts the hypothesized meditational model that we tested in the present study. Parentechild relationships undergo a developmental transition from a high degree of parental control in childhood, to increasing independence and autonomy exerted by offspring throughout adolescence and emergent adulthood (Collins, 1990). The transition may be characterized by variability in the degree of conflict and tension due to parental control

Positive Social Control

+

Positive Affect

+

Attempts to Change

+ -

Negative Social Control

+

Reactance

-

+

Strenuous Exercise

-

Ignoring Influence Agent

-

Fig. 1. Hypothesized meditational model of the relation between type of influence strategy and frequency of strenuous exercise.

J.A. Pugliese, M.A. Okun / Journal of Adolescence 37 (2014) 543e554

545

(Shanahan, McHale, Osgood, & Crouter, 2007). The interaction between adolescents’ greater sense of autonomy and the use of control by parents may explain why many youth discount parental authority over personal matters throughout adolescence (Fuligni, 1998). Parents’ perceived lack of influence may be magnified in an environment where there are competing influences and consequently, they may utilize increasingly negative strategies if attempts at control are perceived as unsuccessful. Moreover, adolescents are more likely to engage in recreational activities with peers than with parents (Larson & Verma, 1999). Peer recreational experiences provide adolescents with opportunities to share, compare, and judge their own experiences and represent opportunities for social comparison and influence (Paxton, Schutz, Wertheim, & Muir, 1999). Therefore, we predicted that college students would report more negative social control strategies when the influence agent was a parent than a peer. Furthermore, we predicted that college students would report more positive social control strategies when the influence agent was a peer rather than a parent. Numerous studies have demonstrated a positive association between positive social control and positive affect (Logic et al., 2009; Okun et al., 2007; Tucker & Anders, 2001; Tucker, Elliott, et al., 2006; Tucker, Orlando, et al., 2006). Therefore, it was hypothesized that positive social control will be a significant predictor of positive affect. These studies have also shown a positive association between negative social control and negative affect. Negative social control tactics from agents also have been linked to reactance from the targets (Butterfield & Lewis, 2002). Reactance refers to “an impulse to restore behavioral freedoms that are perceived to have been threatened or lost” (Brehm, 2000, p. 10). Logic et al. (2009) observed that negative social control was strongly correlated with reactance (r ¼ .60). Consequently, we predicted that negative social control would be a significant predictor of reactance. According to Carver and Scheier (1998), positive emotions increase motivation to pursue goals. Consistent with this notion, several researchers have found both domain-specific and cross-domain relationships between affective reactions to social control attempts and the target’s behavioral response to such attempts (Tucker, 2002; Tucker, Elliott, et al., 2006). For example, in a path analysis, Tucker, Orlando, et al. (2006) reported a significant, positive direct effect between positive affect and the target’s attempts to change behavior and a negative direct effect with ignoring the agent’s social control attempts. We expected to replicate these findings and hypothesized that positive affect would be a significant, positive predictor of the target’s attempts to change their behavior in response to parent’s [peer’s] social control attempts, and a significant, negative predictor of ignoring a parent’s [peer’s] social control attempt. According to Brehm (1999) individuals in a state of reactance seek to reduce it by engaging in behaviors that will reinstate their freedom. Ignoring or doing nothing in the face of a parent’s [peer’s] social control attempts may represent one strategy for reactance reduction. In a study of dating partners, Logic et al. (2009) found that reactance (but not negative affect) fully mediated the association between negative social control and hiding unhealthy behavior (another potential reactance reduction strategy). Thus, we expected reactance to be a significant, positive predictor of ignoring the social control agent’s attempt to change their exercise behavior and a significant negative predictor of participant’s attempts to change their behavior by increasing the amount they exercise. As originally investigated by Tucker (2002), and in other work with colleagues (Tucker, Elliott, et al., 2006; Tucker, Orlando, et al., 2006), behavioral responses to social control attempts have been the final endogenous variable. Within each study, behavioral responses that are specific to control attempts were investigated, and these responses tend to correlate consistently. For example, ignoring the social control agent’s attempts to change a health behavior has exhibited an inverse relation with attempts to change the health behavior that the social agent is trying to influence, with correlations ranging from .30 to .61 (Tucker, 2002; Tucker, Elliott, et al., 2006; Tucker, Orlando, et al., 2006). In the present study, we examined whether these specific behavioral responses are reflected in a global measure of strenuous exercise. If attempts at behavioral change are successful, we expected to find that (a) attempt to change is a significant, positive predictor of a global measure of strenuous exercise and (b) ignoring the social control agent is a significant negative predictor of a global measure of strenuous exercise. To summarize, with respect to the final endogenous variabledstrenuous exercisedwe hypothesized that two associations would be mediated: (a) positive social control / positive affect / attempts to change behavior / strenuous exercise, and (b) negative social control / reactance / ignoring the influence agent / strenuous exercise. With respect to type of influence agent, we expected the model to differ in terms of the magnitude of the mediated relationship between social control and strenuous exercise. Given the central developmental goal of achieving independence from parents during adolescence, it was hypothesized that (a) the mediated relationship between positive social control and strenuous exercise would be significantly stronger for peers than parents, and (b) the mediated relationship between negative social control and strenuous exercise would significantly stronger for parents than peers. We expected these differences in mediated associations to emerge primarily through how youth respond to parental versus peer social control attempts. Therefore, we expected a stronger relationship to emerge between negative social control and reactance when the influence agent was a parent than a peer and conversely, a stronger relationship between positive social control and positive affect when the influence agent was a peer than a parent. In testing the model depicted in Fig. 1, we were aware that results have varied with respect to whether affective reactions fully mediate the relation between social control attempts and behavioral reactions to social control (Okun et al., 2007; Tucker & Anders, 2001) and that cross-domain relations have been observed in several studies (e.g., negative social control exerting a direct effect on positive affect) (Logic et al., 2009; Tucker, Orlando, et al., 2006). Nevertheless, we decided to begin our inferential analyses of frequency of strenuous exercise in late adolescence with a parsimonious model that posited full mediation and no cross-domain associations.

546

J.A. Pugliese, M.A. Okun / Journal of Adolescence 37 (2014) 543e554

Finally, the present model represents a sequential mediation model, whereby it is hypothesized that two mediators intervene between social control attempts and the final endogenous variable. The sequence is largely based on the idea that (a) affect serves as a primary mediator between social control and specific behavioral responses to attempts to control behavior, and (b) specific behavioral responses serve as a primary mediator between affective reactions and global behavior. The latter represents how this process is linked to the occurrence of typical or global behavior. To determine whether our hypothesized sequence is supported by the data, we compared the sequence in Fig. 1 to an alternative sequence of the proposed mediators. Specifically, we examined a model in which specific behavioral responses (i.e., attempts to change & ignoring the influence agent) preceded affective reactions. Both our hypothesized model (Fig. 1) and the alternative sequence model were compared in terms of model fit. Method Sample We recruited participants for our study from students enrolled in an introduction to psychology course offered at a large southwestern state university. To fulfill a course requirement, students have an option of participating in research. At the beginning of each semester, students complete a battery of measures. We included a screening item on the batteryd“During the past 3 months, has anyone attempted to get you to exercise more frequently?” Students who reported being the target of an exercise-related social attempt were contacted via e-mail regarding participating in a study of social influence. When students arrived in the laboratory, they read and signed an informed consent form. The sample for the current study consisted of the 227 students who indicated that either a friend or a parent was the most influential agent with respect to exercise. The gender composition of the sample was 56% female and 44% male. Seventy-three percent of the participants were Caucasian, non-Hispanic 11% were Hispanic, 4% were Asian, 3% were Native American, 1% were African American, and the remaining 8% of the participants were of other ethnicities or were of mixed heritage. Over half of the participants were younger than 19 years old (57%), the remaining 43% were between 19 and 22 years old. Measures Social control agents Participants were asked to indicate whether individuals with whom they have various types of relationships (e.g., parent, friend, sibling) tried to get them to exercise more frequently during the past 3 months. Participants then indicated which type of person tried to influence them the most to exercise more frequently during the past 3 months. Relational characteristics Participants were asked three questions about their relationship with the most influential social control agent. Duration of relationship with this person was assessed on an 8-point scale with anchor points ranging from three months (0) to two years or more (7). Frequency of contact with this person in the last 30 days was assessed on a 10-point scale with anchor points ranging from 1e3 days (0) to 28e30 days (9), and duration of contact with this person was assessed on a 10-point scale with anchor points ranging from 1 h a day or less (0) to 22e24 h per day (9). Situational characteristics Participants were asked three questions about exercise-related social control pertaining to the duration, the frequency, and time elapsed since most recent attempt by the most influential social control agent. Duration of social control influence attempts by this person was assessed on a 9-point scale with anchor points ranging from a week or less (0) to one year or longer (8). Recency of influence attempt was assessed on a 7-point scale with anchor points ranging from a week or less 1 (0) to three months (6) and frequency of influence attempt was assessed on a 7-point scale with anchor points ranging from one time (1) to at least once a day (6). Demographic characteristics Participants were asked questions about their sex, ethnicity, and the age and the sex of the most influential agent. Social control strategies Positive and negative social control strategies were measured with 10 items developed by Lewis and Rook (1999). Participants were told to think about how their parent or friend had tried to influence them to change their frequency of exercising during the past three months. Participants rated how true or untrue it was that parents or friends used each strategy to get them to change the exercise behavior. The items were rated on a 5-point scale with anchor points of false (0) and true (4). Six of the items tapped into positive social control (e.g., the influence agent offered to help me change the health behavior) and four of the items tapped into negative social control (e.g., the influence agent tried to make me feel guilty). Scores on these scales were computed by averaging the responses to the appropriate items. In previous studies, estimates of the internal consistency reliability for the positive social control strategies scale were .65 and .66 and estimates of the internal consistency reliability for the negative social control strategies scale were .72 and .73 (Logic et al., 2009; Okun et al., 2007). In

J.A. Pugliese, M.A. Okun / Journal of Adolescence 37 (2014) 543e554

547

the present study, the coefficient alphas for the positive and negative social control scales were .62 and .70, respectively, for participants reporting on parental social control and .60 and .65, respectively, for participants reporting on peer social control. Positive affect aroused by social control attempts The positive affect aroused by the most influential agent’s social control attempts were measured using items developed by Lewis and Rook (1999). Participants were instructed to think about how they felt at the time that their friend [parent] was trying to get them to exercise more frequently. Participants rated how strongly they experienced eight different positive emotions (e.g., appreciative) on a 5-point scale with anchor points of very weakly (0) and very strongly (4). Scores on this scale were computed by averaging the responses to the eight items. In previous studies, the estimates of the internal consistency reliability for the positive affect scale were .89 and .91 (Logic et al., 2009; Okun et al., 2007). In the current study, the coefficient alphas for the positive affect scale were .90 for participants reporting on parent social control and .89 for participants reporting on peer social control. Reactance aroused by social control attempts The reactance aroused by the most influential agent’s social control attempts were measured using three items developed by Lewis (1992). Participants were instructed to think about how they reacted to their friend’s [parent’s] attempts to influence them to increase the frequency of exercising. The three items on this scale assessed the extent to which it was true or false that the influence attempts made them feel (a) like they had less control over increasing the frequency of exercising, (b) like they were pretending to do something about the situation so that the influence agent would leave them alone, and (c) angry. Participants made their ratings on a 5-point scale with anchor points of false (0) and true (4). Scores on this scale were computed by averaging the responses to the three items. In a previous study, the internal consistency reliability for this scale was estimated to be .72, and scores on the reactance scale correlated .62 with scores on a measure of negative affect (Logic et al., 2009). In the current study, the coefficient alphas for the reactance scale were .83 for participants reporting on parent social control and .80 for participants reporting on peer social control. Behavioral responses to social control We utilized two items developed by Tucker (2002) to assess the extent to which participants attempt to change their behavior or ignore the social influence agent. Participants’ attempts to change were assess by asking the following question, “When the parent [peer] has attempted to get you to increase the frequency of exercising, how often have done what the influence agent asked you to do?” Higher scores indicate greater attempts to change their current levels of exercise. Similarly, the extent to which participants ignored the social influence agent were assessed by asking, “When the parent [peer] has attempted to get you to increase the frequency of exercising, how often have you ignored the influence agent or done nothing?” The response scale for both items was based on a 4-point scale with anchor points of never (0) and often (3). Tucker and Anders (2001) reported a coefficient alpha of .54 for ratings of spousal attempts to influence three different health behaviors. Attempt to change has been found to be correlated .30 with ignoring social control attempts in previous research examining behavioral responses to social control (Tucker, 2002). Ignoring social control attempts has been found to be correlated .33 and .47 with doing the opposite of what the social control agent wants (Tucker, 2002; Tucker, Elliott, et al., 2006). The correlation between ignoring social control attempts and hiding an unhealthy behavior has ranged from .06 to .24 (Tucker, 2002; Tucker, Elliott, et al., 2006; Tucker, Orlando, et al., 2006). The present study focuses on the extent to which adolescents attempt to change (i.e., engage) and ignore the influence agent.

Table 1 Differences between participants identifying parents and peers as influence agents on relational characteristics, situational characteristics, and the main study variables. Variable

Relationship characteristics Duration of contact Days in contact Time spent in contact Situational characteristics Duration of influence attempts Recency of influence attempts Frequency of influence attempts Main study variables Frequency of strenuous exercise Attempt to change Ignore influence agent Reactance Positive affect Negative social control Positive social control

Parents (n ¼ 107)

Peers (n ¼ 120)

t

p

2.76 3.11 2.61

7.45 0.46 2.90

<0.001 0.65 <0.01

3.49 1.02 3.05

2.30 0.72 1.66

7.27 2.75 0.78

<0.001 0.01 0.43

2.26 1.97 1.22 0.48 2.44 1.13 1.90

1.77 0.86 1.04 0.77 0.96 0.90 0.78

2.27 1.89 2.83 5.19 6.04 1.77 3.70

0.02 0.06 0.01 <0.001 <0.001 0.08 <0.01

M

SD

M

SD

6.93 6.50 2.22

0.68 3.06 0.24

4.97 6.32 2.97

5.84 1.59 2.89

2.44 1.30 1.41

1.75 1.77 1.62 1.17 1.68 1.35 2.28

1.61 0.72 1.09 1.15 0.93 1.00 0.79

548

J.A. Pugliese, M.A. Okun / Journal of Adolescence 37 (2014) 543e554

Frequency of strenuous exercise behavior Frequency of strenuous exercise was assessed using the Leisure Time Exercise Questionnaire (Godin & Shephard, 1985). The instructions for the questionnaire were as follows: “Considering a typical 7-day period (a week), how many times on the average do you do the following kinds of exercise for more than 15 min during your free time?” Participants answered this question with respect to (a) strenuous, (b) moderate, and (c) mild exercise. The description that accompanied strenuous exercise was “heart beats rapidly (i.e., running, jogging, hockey, football, soccer, squash, basketball, x-country skiing, swimming, vigorous long distance bicycling).” The description that accompanied moderate exercise was “not exhausting (i.e., fast walking, baseball, tennis, easy bicycling, volleyball, badminton, easy swimming, alpine skiing, popular and folk dancing).” The description that accompanied mild exercise was “minimal effort (i.e., yoga, archery, fishing from river bend, bowling, horseshoes, golf, snow-mobiling, easy walking).” Participants indicated their weekly frequencies for each form of exercise on a 9-point scale ranging from 0 (never) to 8 (8 time a week or more). Okun et al. (2002) reported a testeretest reliability coefficient for total scores over a period between two and six weeks of .60. Concurrent validity coefficients obtained by correlating total scores with other measures of exercise participation have ranged from .32 to .61 (Duncan & McAuley, 1993; Miller, Freedson, & Kline, 1994; Sallis, Buono, Roby, Carlson, & Nelson, 1990). Results Descriptive analyses Fifty-three percent of the late adolescents nominated a peer and 47 percent nominated a parent as the most influential agent with respect to exercise. Sex of participant was significantly related to selecting a parental versus peer influence agent, c2 (226, 1) ¼ 5.70, p < .05. On the one hand, females were more likely to select parental (54%) than peer (46%) influence agents. On the other hand, males were more likely to select peer (62%) than parental (38%) influence agents. Sex of participant was not significantly related to sex of the parental influence agent, c2 (106, 1) ¼ 1.02, but sex of participant was significantly related to sex of the peer influence agent, c2 (119, 1) ¼ 66.49, p < .001. Eighty-nine percent of the males and 86% of the females chose same-sex peer influence agents. The percentage of parental influence agents who were female (72%) was significantly greater than the percentage of peer influence agents who were female (47%), c2 (226, 1) ¼ 14.43, p < .001. Table 1 examines the relations between type of influence agent and (a) relational characteristics, (b) situational characteristics, and (c) the main study variables. With respect to relational characteristics, as expected, participants knew parental influence agents (M ¼ 6.93) longer than peer influence agents (M ¼ 4.97), t (225) ¼ 7.45, p < .001. Participants spent more time in contact with peers (M ¼ 2.97) than parents (M ¼ 2.22), t (225) ¼ 2.90, p < .01. Participants who selected parents and peers did not differ in terms of number of days of contact during the past month. With respect to situational characteristics, parents (M ¼ 5.84) had been trying to influence participants to increase the frequency of exercising for a longer period of time than peers (M ¼ 3.49), t (225) ¼ 7.27, p < .001. In contrast, peers (M ¼ 1.02) made more recent influence attempts than parents (M ¼ 1.59), t (225) ¼ 2.75, p ¼ .01. Frequency of influence attempts did not differ by influence agent. With respect to the main study variables, participants engage in strenuous exercise less frequently when the social influence agent was a parent (M ¼ 1.75) as opposed to a peer (M ¼ 2.26), t (225) ¼ 2.27, p < .05. Parents parent (M ¼ 1.35) tended to be rated higher than peers (M ¼ 1.13) on negative social control strategies, t (225) ¼ 1.77, p ¼ .08, and surprisingly, parents strategies (M ¼ 2.28) were rated higher than peers (M ¼ 1.90) on positive social control strategies, t (225) ¼ 3.70, p < .001. As compared to peer influence agents (M ¼ 1.22), parental influence agents (M ¼ 1.62) were associated with more ignoring of influence attempts, t (225) ¼ 2.83, p ¼ .01. On the one hand, participants reported more reactance when the influence agent was a parent (M ¼ 1.17) as compared to a peer (M ¼ 0.48), t (225) ¼ 5.19, p < .001, and on the other hand, participants reported more positive affect when the influence agent was a peer (M ¼ 2.44) rather than a parent (M ¼ 1.68), t (225) ¼ 6.04, p < .001. Table 2 presents correlations for the main study variables by type of influence agent. Positive social control was modestly correlated with both positive affect and negative social control, but, somewhat surprisingly, not with frequency of strenuous exercise for either influence agent (Wilson & Spink, 2012). Negative social control was strongly associated with reactance,

Table 2 Interrelations among the main study variables. Variables

1

1. 2. 3. 4. 5. 6. 7.

e

Frequency of strenuous exercise Attempt to change Ignore influence agent Reactance Positive affect Negative social control Positive social control

0.35*** 0.35*** 0.19* 0.41*** 0.05 0.11

2 0.32*** e 0.50*** 0.22* 0.35*** 0.04 0.25**

Note: Parents are in the lower diagonal; Peers are in the upper diagonal. *p < .05, **p < .01, ***p < .001.

3

4

5

6

7

0.31*** 0.55*** e 0.45*** 0.57*** 0.26** 0.16

0.26** 0.38*** 0.39*** e 0.41*** 0.60*** 0.00

0.22* 0.58*** 0.54*** 0.32*** e 0.20* 0.34***

0.07 0.21* 0.23** 0.41*** 0.15 e 0.08

0.14 0.14 0.13 0.03 0.23* 0.18 e

J.A. Pugliese, M.A. Okun / Journal of Adolescence 37 (2014) 543e554

549

inversely associated with positive affect, and positively associated with ignoring influence attempts. Consistent with previous research, negative social control was not related to frequency of strenuous exercise either for parents or peers (Wilson & Spink, 2012). Positive affect was strongly, inversely related to ignoring influence attempts and reactance, and was positively related to frequency of strenuous exercise. Reactance was strongly positively correlated with ignoring influence attempts and inversely related to frequency of strenuous exercise. As expected, ignoring influence attempts was inversely correlated and attempt to change was positively correlated with frequency of strenuous exercise for both types of influence agents. It is noteworthy that the relationship between negative social control and reactance was larger when the influence agent was a parent (r ¼ .60) than a peer (r ¼ .41) as was the relationship between positive affect and frequency of strenuous exercise (.41 versus .22). In contrast, the association between positive affect and attempt to change was larger when the influence agent was a peer (r ¼ .58) than a parent (r ¼ .35). Inferential analyses Model testing We used path analysis to estimate versions of the model depicted in Fig. 1. The models were stacked, which means the model fit was examined simultaneously for groups of youth who had nominated either (a) a parent or (b) a peer as the primary influence agent. The model for each group tests the extent to which the association between type of social control tactic and strenuous exercise is mediated by affective and behavioral reactions to social control. The analysis began with a fully constrained model. The model did not provide a reasonable fit to the data, c2 (227, 38) ¼ 66.99, p ¼ .003, RMSEA ¼ .08. To improve the fit of the model to the data, additional pathways were allowed to vary in a series of steps (see Table 3). Pathway selection was based on theoretical considerations and an examination of modification indices and standardized residuals. Modification indices provide the expected change in model Chi-Square due to the inclusion of an unconstrained pathway. Decreases in model Chi-Square, RMSEA and AIC, typically indicate better fit to the underlying observed data. The first two pathways that were unconstrained in the model in Fig. 1 were from negative social control to reactance (Table 3, Model 2), and positive affect to attempt to change (Table 3, Model 3). Across both models, a significant decrease in the model Chi-Square values were observed, as well as notable decreases in RMSEA and AIC fit indices. Only the latter pathway, between positive affect and attempt to change, resulted in an adequate fit to the data, c2 (227, 36) ¼ 48.59, p ¼ .08, RMSEA ¼ .07. The next pathway to be unconstrained in the model was from negative social control to positive affect (Table 3, Model 4). The inclusion of the pathway resulted in a significant decrease in model Chi-Square, and decreases in both RMSEA and AIC, c2 (227, 35) ¼ 37.31, p ¼ .36, RMSEA ¼ .06. Finally, the inclusion of an additional pathway between positive affect and strenuous exercise (Table 3, Model 5) within the parent model only, provided improved fit over the prior model (Table 3, Model 4). Figs. 2 and 3 present the final models with path coefficients for the parent and peer groups, respectively. The final model1 provided optimal fit to the data, c2 (227, 34) ¼ 31.71, p ¼ .58, RMSEA ¼ .00, AIC ¼ 75.71, CFI ¼ 1.00. Given that the degrees of freedom for this model were greater than the Chi-Square value, RMSEA was effectively zero. As expected, the pathway between positive social control and positive affect was equivalent across both groups (g ¼ .34, p < .001). By contrast, there was a statistical difference in regard to the pathway between negative social control and reactance for parents and peers (g ¼ .65 p < .001 and g ¼ .35, p < .001, respectively). As hypothesized, the pathway from positive affect to attempts to change was significant, and this pathway differed for parents and peers (g ¼ .21, p < .01 and g ¼ .48, p < .001, respectively). There was no difference between groups with regard to reactance as a significant predictor of ignoring the influence agent (g ¼ .28, p < .001). Contrary to our predictions, positive affect was a significant predictor of strenuous exercise only in the parent group (g ¼ .42, p < .05). Finally, ignoring the influence agent was a marginally significant negative predictor of strenuous exercise (g ¼ .23, p < .08). The model accounted for 20% and 12% of the variation in strenuous exercise for parent and peer groups, respectively. Alternative model There are often multiple alternative path models that may account for the observed relationships among the main study variables, especially when the data are cross-sectional (Loehlin, 2004; MacKinnon, Fairchild, & Fritz, 2007). The hypothesized model in Fig. 1 depicts a specific sequence of mediators whereby specific behavioral reactions (attempt to change and ignoring) follow affective reactions to social control attempts. As previously mentioned, we decided to examine an alternative model whereby affective reactions follow specific behavioral reactions to social control attempts. There was no theoretical basis to guide model fitting, so we relied on modification indices when determining which pathways should be unconstrained. The alternative model did not fit the data as well as our hypothesized model, c2 (227, 42) ¼ 47.52, p ¼ .06, RMSEA ¼ .06, AIC ¼ 91.52, CFI ¼ .96. The model explained 13% of the variation in strenuous exercise within both parent and

1 A model including gender of the participant and frequency of influence attempts as covariates of strenuous exercise was also examined. The covariate model fit the data well, c2 (227, 57) ¼ 58.8, p ¼ .59, RMSEA ¼ 0.00, AIC ¼ 119.83, CFI ¼ 1.00, albeit with a greater number of parameters to estimate. Only the relationship between gender and frequency of strenuous exercise in the peer model remained statistically significant. With peer influence agents, males exercised strenuously more frequently than females. Moreover, a comparison of parameters estimates between the covariate model and the final model revealed only small changes in the endogenous pathways to strenuous exercise. Given no substantive advantage to including the covariates, we decided to drop them from the model in order to present a more parsimonious model.

550

J.A. Pugliese, M.A. Okun / Journal of Adolescence 37 (2014) 543e554

Table 3 Model fit statistics and nested model tests for mediational model of social control. Model

c2

df

p

RMSEA

AIC

CFI

1. 2. 3. 4. 5.

66.99 57.33 48.59 37.31 31.71

38 37 36 35 34

0.003 0.017 0.079 0.363 0.580

0.08 0.07 0.06 0.02 0.00

102.99 95.33 88.58 79.31 75.71

0.92 0.94 0.96 0.99 1.00

Hypothesized model Free path from NSC to REACT Free path from PA to ATC þPath from NSC to PA þPath from PA to EX e parents only

Model difference tests

Dc2

Ddf

p

1 2 3 4

9.66 8.74 11.28 5.60

1 1 1 1

0.002 0.003 <0.001 0.017

vs vs vs vs

2 3 4 5

Note: PA ¼ positive affect, REACT ¼ reactance, NSC ¼ negative social control, ATC ¼ attempt to change, EX ¼ frequency of strenuous exercise.

.42(.28)*

Positive Social Control

.34 (.29)***

.21(.27)**

Positive Affect

Attempts to Change .47(.20)** -.12(.17)*

E -.21(-.22)***

-.23(.23)*** E

Negative Social Control

Reactance .69(.63)***

.28(.27)***

-.51(-.46)***

E

Strenuous Exercise

-.20(.25)*** E

Ignoring Influence Agent

-.23(-.14

Fig. 2. Final version of the mediational model of the relation between type of influence strategy agent and frequency of strenuous exercise when the influence agent is identified as a parent. Dashed lines represent paths added to the hypothesized model. (Standardized estimates are inside the parentheses.) y p<.08, *p<.05, **p<.01, ***p<.001.

peer groups. An examination of the pathways in the alternative model revealed a number of direct effects that were not present in the hypothesized model. For example, wherein reactance fully mediates the relationship between negative social control strategies and ignoring the influence agent in the hypothesized model, ignoring the influence agent did not fully mediate the relationship between negative social control and reactance in the alternative model. Similarly, positive affect did not fully mediate the relationship between attempt to change and strenuous exercise in the alternative model. The alternative model does provide an adequate fit to the data, but given the added pathways due to a lack of full mediation, it is not as parsimonious as the hypothesized model in Fig. 1. Therefore, we examined the indirect effects observed in the hypothesized model presented in Table 3, Model 5. Indirect effects Bias-corrected bootstrap procedures (McKinnon, 2008) were used to examine the indirect effects of social control strategies within each group (see Table 4). The model-generated indirect effects, bootstrap estimated bias, and confidence intervals are summarized in Table 4. Statistically significant indirect effects were observed between positive and negative social control strategies and strenuous exercise. As hypothesized, in both parent and peer groups a significant positive indirect effect .00(.00)

Positive Social Control

.34(.29)***

Positive Affect

.48(.54)***

Attempts to Change .47(.24)** -.12(-.13)*

E

-.51(-.47)***

E Negative Social Control

Reactance .35(.37)***

E

-.23(-.28)***

-.21(-.22)***

.28(.23)***

-.20(.22)***

Strenuous Exercise

E

Ignoring Influence Agent

-.23(-.15

Fig. 3. Final version of the mediational model of the relation between type of influence strategy agent and frequency of strenuous exercise when the influence agent is identified as a peer. Dashed lines represent paths added to the hypothesized model. (Standardized estimates are inside the parentheses.) y p<.08, *p<.05, **p<.01, ***p<.001.

J.A. Pugliese, M.A. Okun / Journal of Adolescence 37 (2014) 543e554

551

Table 4 Bias-corrected bootstrap estimates of the indirect effects of positive and negative strategies. Indirect path

Parents PSC / PA / EX PSC / PA / ATC / EX PSC / PA / IIA / EX NSC / PA / EX NSC / PA / IIA / EX NSC / PA / ATC / EX NSC / REACT / IIA / EX NSC / REACT / ATC / EX Peers PSC / PA / ATC / EX PSC / PA / IIA / EX NSC / PA / IIA / EX NSC / PA / ATC / EX NSC / REACT / IIA / EX NSC / REACT / ATC / EX

Model

Bootstrap estimates

ab

Bias

LL

UL

0.14 0.03 0.04 0.09 0.03 0.02 0.04 0.02

0.002 0.000 0.001 0.001 0.000 0.000 0.001 0.000

0.040 0.006 0.003 0.153 0.056 0.042 0.092 0.001

0.245 0.064 0.086 0.025 0.002 0.004 0.004 0.042

0.08 0.04 0.03 0.05 0.02 0.01

0.001 0.001 0.000 0.000 0.001 0.000

0.025 0.003 0.056 0.088 0.049 0.001

0.127 0.086 0.002 0.016 0.002 0.022

Note: PA ¼ positive affect, REACT ¼ reactance, PSC ¼ positive social control, NSC ¼ negative social control, IIA ¼ ignore influence agent, ATC ¼ attempt to change, EX ¼ frequency of strenuous exercise.

between positive social control and strenuous exercise via positive affect and attempt to change was observed, ab ¼ 0.03, 95% CI [0.01, 0.06] and ab ¼ 0.08, 95% CI [0.03, 0.13], respectively. The bias-corrected bootstrap estimate of the difference between the both parent and peer indirect effects (see Table 5) was also statistically significant, Dab ¼ 0.04, 95% CI [0.08, 0.01]. Similarly, in both the parent and peer groups a significant negative indirect effect between negative social control and strenuous exercise via reactance and ignoring the influence agent was observed, ab ¼ 0.04, 95% CI [0.092, 0.004] and ab ¼ 0.02, 95% CI [0.049, 0.002], respectively. The difference between parents and peers in terms of this pathway was also statistically significant, Dab ¼ 0.03, 95% CI [0.01, 0.05]. Unexpectedly, when the parent was the primary influence agent a positive indirect effect between positive social control and strenuous exercise via positive affect was found to be statistically significant, ab ¼ 0.14, 95% CI [0.04, 0.25]. One cross-domain effect emerged; a statistically significant negative indirect effect between negative social control and strenuous exercise via positive affect was found, ab ¼ 0.09, 95% CI [0.15, 0.03]. Discussion The present study compared two primary socializing agentseparents and peerseand examined how type of influence agent influences adolescent youth physical activity within a framework of health-related social control. Research demonstrates that peers become the primary socializing agents during adolescence (Collins & Laursen, 2004). Interestingly, college students were just as likely to nominate parents (47%) as opposed to peers (53%) as the type of person who exerted the most influence on trying to increase their frequency of exercise, c2 (227, 1) ¼ 0.74. This finding suggests that at least in some domains, parents continue to be an important socialization agent of college-age youth (Collins & Steinberg, 1998). Contrary to our prediction, positive social control scores were higher for parents than peers. In addition, there was a trend, consistent with our prediction, for participants to report higher negative social control scores for parents than for peers. It is noteworthy that participants reported higher scores for positive social control than negative social control for parents, suggesting that on average, parents utilize a positive approach overall. Due to investment in their offspring, parents may be more motivated than peers to employ social control strategies targeting health behaviors (Tinsley, 2003). Despite the

Table 5 Bias-corrected bootstrap estimates of the difference in indirect effects between parents and peers. Indirect path

PSC / PA / ATC / EX Parents Peers Difference NSC / PA / ATC / EX Parents Peers Difference

Model

Bootstrap estimates

ab

Bias

LL

UL

0.03 0.08 0.04

0.001

0.076

0.009

0.02 0.05 0.03

0.000

0.005

0.053

Note: All model values are rounded to nearest hundredth. PA ¼ positive affect, REACT ¼ reactance, PSC ¼ positive social control, NSC ¼ negative social control, IIA ¼ ignore influence agent, ATC ¼ attempt to change, EX ¼ frequency of strenuous exercise.

552

J.A. Pugliese, M.A. Okun / Journal of Adolescence 37 (2014) 543e554

difference in positive social control scores between parents and peers, parents evoked less positive affect than peers. One explanation for this effect is that the issue of control, whether positive or negative, is perceived differently by college-age adolescents when the agent is a parent as compared to a peer. Occurring against a backdrop of the transition to independence, parental social control becomes more salient in late adolescence. For example, when parents (but not peers) use a positive social control strategy such as, offering to help with exercising more frequently, it may be appraised by a college-age adolescent as an effort to control his or her behavior. Consistent with our hypothesis, parents also evoked more reactance than peers. Successful development of peer relationships is a second developmental goal of adolescence (Brown, 2004; Hartup, 1996). Thus, late adolescents may place a greater priority on maintaining a relatively positive climate with respect to peer than parental relationships. In the context of a more positive relationship climate, the need for college-age youth to restore behavioral freedom in response to social control attempts would be lower with peers than with parents. The results of Okun et al. (2002) would therefore dovetail with the current study, suggesting that adolescent’s responsiveness to peer norms may partially operate through peer social control. Direct and indirect effects of positive social control As hypothesized, positive social control strategies were related to positive affect and this relationship was the same whether the influence agent was a parent or a peer. Other researchers have also observed a positive association between positive social control strategies and positive affective reactions (Tucker & Mueller, 2000). Consistent with previous research with romantic dating partners (Logic et al., 2009), the final model also demonstrates that positive affect in response to social control fully mediates the relation between positive social control and subsequent endogenous variables. As predicted, the indirect effect between positive social control and strenuous exercise, via positive affect and attempts to change, was lower for parents than for peers. As expected, positive affect was a significant, negative predictor of ignoring the influence agent. Other researchers have also observed that more positive affective reactions to social control are associated with less negative behavioral reactions (Okun et al., 2007). Unexpectedly, positive affect was also a significant predictor of strenuous exercise, when the influence agent is a parent. One explanation is that frequent positive affect may facilitate goal striving in the health behavior domain by conferring resources such as energy and optimism that are linked with effective self-improvement (Lyubomirsky, King, & Diener, 2005). However, why this effect would emerge only for parental social influence agents remains unclear. Direct and indirect effects of negative social control As predicted, negative social control was a strong predictor of reactance. Consistent with previous research, negative social control strategies is associated with more psychological reactance (Logic et al., 2009). Also consistent with our predictions, the relationship between negative parental social control and reactance was almost twice as large when the influence agent was a parent than a peer. Negative social control was also directly related to positive affect irrespective of the type of influence agent. Other researchers have also observed cross-domain effects (Tucker, Elliott, et al., 2006; Tucker, Orlando, et al., 2006). Overall, the extent that negative social control strategies, such as making an unflattering social comparison on progress made on health goals, represent interpersonal stressors, they may shrink an individual’s affective space (Zautra, Berkhof, & Nicolson, 2002) and exert spillover effects on affective outcomes (Giacobbi, Hausenblas, & Frye, 2005). When compounded by the developmental goal of seeking autonomy, as is the case when a parent is the influence agent, negative social control attempts may be particularly stressful. The final model also demonstrates an indirect effect between negative social control and ignoring the influence agent by increasing reactance and by decreasing positive affect. Due to the large difference in the direct effect on reactance, it followed that the indirect effect between negative social control and strenuous exercise, via reactance and ignoring, was larger when the influence agent was a parent than peer. In accord with previous research, reactance was a significant, positive predictor of ignoring the influence agent and exhibited a negative indirect effect on strenuous exercise via ignoring the influence agent (Logic et al., 2009). As individuals experience more reactance in response to social control attempts, they seek to restore their freedom. One strategy for accomplishing this goal is to ignore the efforts of the influence agent. Finally, the differences between parents and peers should not obscure the similarities between each type of influence agent. Recall that parents and peers were similar in their frequency of influence attempts. Moreover, the majority of the pathways in the parent and peer models were equivalent. For example, all three cross-domain pathways were found to be similar. For arguably the most important socialization agents in adolescence, the results provide further evidence for the ‘dual pathways’ model of social control (Okun et al., 2007). Limitations The present study has several limitations. First, we used a cross-sectional design which does not permit causal inferences regarding the relations among the set of variables (see Spink, Strachan, & Odnokon, 2008). Although mediational modeling implies causation, we explicitly compared the final model to an alternative model in order to lend greater support to the hypothesized sequence of mediators. Ultimately, any questions regarding causation will require a true experimental design, in

J.A. Pugliese, M.A. Okun / Journal of Adolescence 37 (2014) 543e554

553

which both the exogenous variables and mediating variables are manipulated. Second, we collected data at an urban university where many students live in close proximity to their parents. Students who live in close proximity with parents may be more susceptible to parental attempts at control. Students who do not live at home, or are away at school, may be experience far less parental positive social control attempts; in other words, the differences observed in the present study between parents and peers as agents of positive social control may be reversed for these youth. Third, we relied solely on data collected from the target of the social control attempts. If youth are more sensitive to attempts at control, parental attempts may be more salient in a milieu of multiple social influences. Second, all variables, including strenuous exercise, were assessed via questionnaires. Despite the advantage of practicality, the precision of selfreport measures of exercise may suffer compared to the use of measures obtained from mechanical and electronic devices such as pedometers. In general researchers are recommended to employ at least two different measurement techniques when assessing any type of physical activity (Bar-Or & Rowland, 2004). Finally, participants reported on only one type of social control agent. Youth strenuous exercise may be influenced by parents and peers simultaneously, and any number of other social influence agents such as trainers, coaches, siblings, etc. Implications The findings of the present study suggest two pathways for increasing strenuous exercise and other health-enhancing behaviors in college-age adolescents. One target for interventions involves the social control strategies used by parents and peers. We have shown that the use of negative social control strategies is associated with a reduction in positive affect and an increase in reactance. Decreases in positive affect and increases in reactance, in turn, appear to foster doing nothing in response to social control attempts. Due to the reactance experienced by youth in response to negative social control, interventions should focus on decreasing parents’ use of negative social control strategies. A second target of interventions involves increasing the positive affective reactions of college-age adolescents to parental social control agents. In the current study, positive affective reactions exerted a direct effect on frequency of strenuous exercise. Furthermore, relative to peer social control agents, we showed that parental social control agents evoked less positive affective reactions from college-age adolescents. Perhaps, college-age adolescents have a mindset to interpret parental social control as an attempt to stifle their growing autonomy when, in fact, it is often motivated by parental concern for their health and well-being. By teaching college-age adolescents to embed their parents’ social control attempts in a more positive motivational context, they may experience more positive affective reactions when their parents use positive social control strategies. References Bar-Or, O., & Rowland, T. W. (2004). Pediatric exercise medicine. Champaign, IL: Human Kinetics Publishers. Beets, M. W., Vogel, R., Forlaw, L., Pitetti, K. H., & Cardinal, B. J. (2006). Social support and youth physical activity: the role of provider and type. American Journal of Health Behavior, 30, 278e289. Brehm, J. (1999). The intensity of emotion. Personality & Social Psychology Review, 3, 2e22. Brehm, J. (2000). Reactance. In A. E. Kazdin (Ed.), Encyclopedia of psychology (Vol. 7); (pp. 10e12). New York, NY, US: Oxford University Press. Brewer, M. B., & Caporael, L. R. (1990). Selfish genes vs. selfish people: sociobiology as origin myth. Motivation & Emotion, 14(4), 237e243. Brown, B. (2004). Adolescents’ relationships with peers. In R. Lerner, & L. Steinberg (Eds.), Handbook of adolescent psychology (2nd ed) (pp. 363e394). New York: Wiley. Butterfield, R. M., & Lewis, M. (2002). Health-related social influence: a social ecological perspective on tactic use. Journal of Social and Personal Relationships, 19, 505e526. Carver, C. S., & Scheier, M. F. (1998). On the self-regulation of behavior. New York: Cambridge University Press. Collins, W. A. (1990). Parentechild relationships in the transition to adolescence: continuity and change in interaction, affect, and cognition. In R. Montemayor, G. R. Adams, & T. P. Gullotta (Eds.), From childhood to adolescence: A transitional period? (pp. 85e106). Newbury Park, CA: Sage. Collins, W. A., & Laursen, B. (2004). Parenteadolescent relationships and influences. In R. Lerner, & L. Steinberg (Eds.), Handbook of adolescent psychology (2nd ed) (pp. 331e361). New York: John Wiley & Sons. Collins, W. A., & Steinberg, L. (1998). Adolescent development in interpersonal context. In D. William, & N. Eisenberg (Eds.), Handbook of child psychology: Social, emotional, and personality development. Hoboken, NJ: John Wiley & Sons, Inc. Cooper, C. R. (1994). Cultural perspectives on continuity and change in adolescents’ relationships. In R. Montemayor, G. R. Adams, & T. P. Gullotta (Eds.), Personal relationships during adolescence (pp. 78e100). Thousand Oaks, CA: Sage. Duncan, T. E., & McAuley, E. (1993). Social support and efficacy cognitions in exercise adherence: a latent growth curve analysis. Journal of Behavioral Medicine, 16, 199e218. Fuligni, A. J. (1998). Authority, autonomy, and parenteadolescent conflict and cohesion: a study of adolescents from Mexican, Chinese, Filipino, and European backgrounds. Developmental Psychology, 34, 782e792. Giacobbi, P. R., Hausenblas, H. A., & Frye, N. (2005). A naturalistic assessment of the relationship between personality, daily events, leisure-time exercise and mood. Psychology of Sport & Exercise, 6, 67e81. Godin, G., & Shephard, R. J. (1985). A simple method to assess exercise behavior in the community. Canadian Journal of Applied Sport Sciences, 10, 141e146. Gruber, K. J. (2008). Social support for exercise and dietary habits among college students. Adolescence, 43, 557e575. Hallal, P. C., Victoria, C. G., Azevedo, M. R., & Wells, J. C. K. (2006). Adolescent physical activity and health. Sports Medicine, 36, 1019e1030. Hartup, W. W. (1996). The company they keep: friendships and their developmental significance. Child Development, 67, 1e13. Larson, R., & Verma, S. (1999). How children and adolescents spend time across cultural settings of the world: work, play and developmental opportunities. Psychological Bulletin, 125, 701e736. Lewis, M. A. (1992). The influence of social control on health behaviors and psychological distress (Unpublished doctoral dissertation). Irvine: University of California. Lewis, M. A., & Rook, K. S. (1999). Social control in personal relationships: Impact on health behaviors and psychological distress. Health Psychology, 18, 63e 71. Lin, N., Simeone, R. S., Ensel, W. M., & Kuo, W. (1979). Social support, stressful life events, and illness: a model and an empirical test. Journal of Health and Social Behavior, 20, 108e119. Loehlin, J. C. (2004). Latent variable models: An introduction to factor, path, and structural equation analysis. Mahwah, NJ: Lawrence Erlbaum Associates.

554

J.A. Pugliese, M.A. Okun / Journal of Adolescence 37 (2014) 543e554

Logic, M., Okun, M. A., & Pugliese, J. A. (2009). Expanding the mediational model of the effects of health-related social control. Journal of Applied Social Psychology, 39, 1373e1396. Lyubomirsky, S., King, L., & Diener, E. (2005). The benefits of frequent positive affect: does happiness lead to success? Psychological Bulletin, 131, 803e855. MacKinnon, D. P. (2008). Introduction to statistical mediation analysis. New York, New York: Lawrence Erlbaum & Associates. MacKinnon, D. P., Fairchild, A. J., & Fritz, M. S. (2007). Mediation analysis. Annual Review of Psychology, 58, 593e614. Miller, D. J., Freedson, P. S., & Kline, G. M. (1994). Comparison of activity levels using the Caltrac accelerometer and five questionnaires. Medicine and Science in Sport and Exercise, 26, 376e382. Okun, M. A., Huff, B. P., August, K. J., & Rook, K. (2007). Testing hypotheses distilled from four models of the effects of health-related social control. Basic & Applied Social Psychology, 29, 185e193. Okun, M. A., Karoly, P., & Lutz, R. (2002). Clarifying the contribution of subjective norm to predicting leisure time exercise. American Journal of Health Behavior, 26, 296e305. Okun, M. A., Ruehlman, L., Karoly, P., Lutz, R., Fairholme, C., & Schaub, R. (2003). Social support and social norms: do both contribute to predicting leisuretime exercise? American Journal of Health Behavior, 27, 493e507. Paxton, S. J., Schutz, H. K., Wertheim, E. H., & Muir, S. L. (1999). Friendship clique and peer influences on body image concerns, dietary restraint, extreme weight-loss behaviors, and binge eating in adolescent girls. Journal of Abnormal Psychology, 108, 255e266. Perkins, D. F., Jacobs, J. E., Barber, B. L., & Eccles, J. S. (2004). Childhood and adolescent sports participation as predictors of participation in sports and physical fitness activities during young adulthood. Youth & Society, 35, 495e520. Prochaska, J. J., Rodgers, M. W., & Sallis, J. F. (2002). Association of parent and peer support with adolescent physical activity. Research Quarterly for Exercise and Sport, 73, 206e210. Racette, S. B., Deusinger, Si S., Strube, M. J., Highstein, G. R., & Deusinger, R. H. (2005). Weight changes, exercise, and dietary patterns during freshman and sophomore years of college. Journal of American College Health, 53, 245e251. Rook, K. S. (1995). Relationship research at the crossroads. Journal of Social and Personal Relationships, 12, 601e606. Rook, K. S., & Pietromonaco, P. (1987). Close relationships: ties that heal or ties that bind?. In W. H. Jones, & D. Perlman (Eds.), Advances in personal relationships (Vol. 1); (pp. 1e35) Greenwich, CT: JAI Press. Sallis, J. F., Buono, M. J., Roby, J. J., Carlson, D., & Nelson, J. A. (1990). The Caltrac accelerometer as a physical activity monitor for school-age children. Medicine and Science in Sport and Exercise, 22, 698e703. Sarason, I. G., Sarason, B. R., & Pierce, G. R. (1994). Social support: global and relationship-based levels of analysis. Journal of Social and Personal Relationships, 11, 295e312. Shanahan, L., McHale, S. M., Osgood, D. W., & Crouter, A. (2007). Conflict frequency with mothers and fathers from middle childhood to late adolescence: within- and between-families comparisons. Developmental Psychology, 43, 539e550. Spink, K. S., Strachan, S. M., & Odnokon, P. (2008). Parental physical activity as a moderator of the parental social influence-child physical activity relationship: a social control approach. Social Influence, 3, 189e201. Tinsley, B. J. (2003). How children learn to be healthy. Cambridge, U.K.: Cambridge University Press. Tucker, J. S. (2002). Health-related social control within older adults’ relationships. Journal of Gerontology Series B: Psychological Sciences and Social Sciences, 57B, 387e395. Tucker, J. S., & Anders, S. L. (2001). Social control of health behaviors in marriage. Journal of Applied Social Psychology, 31, 467e485. Tucker, J. S., Elliott, M., & Klein, D. (2006). Social control of health behavior: associations with conscientiousness and neuroticism. Personality and Social Psychology Bulletin, 32, 1143e1152. Tucker, J. S., & Mueller, J. S. (2000). Spouses’ social control of health behaviors: use and effectiveness of specific strategies. Personality & Social Psychology Bulletin, 26, 1120e1130. Tucker, J. S., Orlando, M., Elliot, M. N., & Klein, D. J. (2006). Affective and behavioral responses to health-related social control. Health Psychology, 25, 715e 722. United States Department of Health and Human Services. (2011). Healthy People 2020. Physical activity: National data. Retrieved from http://www. healthypeople.gov/2020/topicsobjectives2020/nationaldata.aspx?topicId¼33. U.S. Department of Health and Human Services. (2000). Healthy people 2010: Understanding and improving health. Washington, DC: U.S. Government Printing Office. US Department of Health and Human Services (HHS), Office of Disease Prevention and Health Promotion. (2008). 2008 Physical activity guidelines for Americans. Washington, D.C. Wilson, K. S., & Spink, K. S. (2012). Predicting parental social influences: the role of physical activity variability. Psychology of Sport & Exercise, 13, 1e9. Zautra, A. J., Berkhof, J., & Nicolson, N. A. (2002). Changes in affect interrelations as a function of stressful events. Cognition and Emotion, 16, 309e318.