Gardner, Roth, Brooks-Gunn

Developmental Psychology 2008, Vol. 44, No. 3, 814 – 830

Copyright 2008 by the American Psychological Association 0012-1649/08/$12.00 DOI: 10.1037/0012-1649.44.3.814

Adolescents’ Participation in Organized Activities and Developmental Success 2 and 8 Years After High School: Do Sponsorship, Duration, and Intensity Matter?
Margo Gardner, Jodie Roth, and Jeanne Brooks-Gunn
Teachers College, Columbia University
Using data from the National Education Longitudinal Study, the authors examined relations between educational, civic, and occupational success in young adulthood and the duration and intensity of participation in organized activities during high school. They also examined these relations as a function of sponsorship (i.e., school- vs. community-sponsored organized activities). They found that youths who participated in organized activities for 2 years demonstrated more favorable educational and civic outcomes in young adulthood than those who participated for 1 year. More intensive participation was also associated with greater educational, civic, and occupational success in young adulthood— particularly among youths who participated in activities for 2 years. Educational attainment often mediated the relations between temporal measures of participation and young adult civic and occupational outcomes. With the exception of analyses examining occupational success, findings varied little as a function of sponsorship. Of note, analyses revealed that both temporal measures of participation were positively associated with young adult outcomes as many as 8 years after high school. Keywords: organized activities, extracurricular activities, after-school activities, positive youth development, young adulthood

Growing evidence demonstrates positive associations between participation in organized activities outside of school hours and positive youth development (Feldman & Matjasko, 2005; Mahoney, Larson, & Eccles, 2005).1 Studies suggest that participation in organized activities is associated with fewer problem behaviors, better academic performance, and better psychosocial adjustment (Mahoney et al., 2005). Recent findings also suggest that relations between participation in organized youth activities and positive outcomes may extend into young adulthood (e.g., Zaff, Moore, Papillo, & Williams, 2003). Accordingly, interest in organized youth activities has surged among families, policymakers, and researchers during the last decade. More than half of American teenagers participate in organized activities (Mahoney, Harris, & Eccles, 2006), and recent federal, state, and city initiatives aim to increase access to these activities (e.g., federal 21st Century Community Learning Centers, California’s Proposition 49, Chicago’s After School Matters initiative). Despite consensus on the overall value of organized activities, questions remain regarding the extent to which specific dimensions of participation (e.g., duration and intensity) are associated with short- and long-term success. Additionally, there is little consensus on the pathways linking participation in organized ac-

tivities to later developmental success. In this study, we investigate whether the duration and intensity of participation in organized school- and community-sponsored activities during adolescence are associated with educational, civic, and occupational success in young adulthood. We also explore one of the possible pathways linking participation in organized activities and positive young adult outcomes: We examine the extent to which educational successes associated with participation in organized youth activities may facilitate opportunities for greater civic and occupational success during young adulthood.

Organized Activities and Positive Youth Development
The literature on positive youth development provides a useful framework for conceptualizing how youths’ experiences during the after-school hours may contribute to development. Theory and research on positive youth development emphasize the plasticity of human development and suggest that cultivating positive, supportive relationships with people and social institutions encourages healthy development (Lerner et al., 2005; Roth, Brooks-Gunn, Murray, & Foster, 1998). According to this perspective, organized activities encourage healthy development because they provide more developmental supports and opportunities than other common after-school pursuits, like watching television or hanging out with friends. These supports and opportunities may include (a) physical and psychological safety, (b) appropriate structure, (c) supportive relationships, (d) opportunities to belong, (e) positive
1 Organized activities is a catchall phrase that includes any activity or program supervised by an adult that involves other youths and meets outside of school hours on a regular basis (Mahoney et al., 2005).

Margo Gardner, Jodie Roth, and Jeanne Brooks-Gunn, National Center for Children and Families, Teachers College, Columbia University. This research was supported by Grant 2501 from the William T. Grant Foundation. Correspondence concerning this article should be addressed to Margo Gardner, National Center for Children and Families, Teachers College, Columbia University, Box 39, 525 West 120th Street, New York, NY 10027. E-mail: [email protected] 814

ORGANIZED ACTIVITIES AND DEVELOPMENTAL SUCCESS

815

social norms, (f) support for efficacy and mattering, (g) opportunities for skill building, and (h) the integration of family, school, and community efforts (Eccles & Gootman, 2002). Various nomenclatures have been used to describe the healthy outcomes associated with participation in activities that offer these kinds of supports and opportunities. One influential taxonomy, termed “the five Cs,” assigns these outcomes to five categories: (a) competence in academic, social, and vocational areas; (b) confidence or a positive self-identity; (c) connections to community, family, and peers; (d) character or positive values, integrity, and moral commitment; and (e) caring and compassion (Lerner, Fisher, & Weinberg, 2000). Preliminary evidence supports the theoretical contention that these five Cs load onto a single higher order latent construct and that all five Cs therefore make essential contributions to individuals’ abilities to thrive in adolescence and adulthood (Lerner et al., 2005). Although research on this higher order thriving construct is still limited, empirical evidence supports positive relations between participation in organized after-school activities and many of the individual five Cs (see Roth & BrooksGunn, 2003a, 2003b). The issue of self-selection, however, remains an obstacle to definitive conclusions about the causal role of participation in organized activities. Because participation in organized activities is voluntary and constrained by personal factors (e.g., motivation, skills, financial resources) and institutional factors (e.g., availability, transportation, costs), selection undoubtedly plays a role in the link between participation in organized activities and developmental success. Indeed, modest and mixed findings from randomized studies of after-school programs— one of the many types of organized activities that youths participate in—suggest that the robust findings observed in many nonexperimental studies may capitalize on the influence of self-selection (see Harvard Family Research Project, 2006, for a review of experimental studies of after-school programs). One approach that reduces, but does not eliminate, the influence of self-selection in nonexperimental research is to control for factors known to predict both participation in organized activities and positive outcomes. The studies reviewed below, which informed our research questions, included such statistical controls unless otherwise noted.

Dimensions of Participation in Organized Activities
Different organized activities offer different opportunities and supports for development and may therefore encourage different “Cs” (Larson, Hansen, & Moneta, 2006; Roth & Brooks-Gunn, 2003a, 2003b). Youths also differ with respect to the length (i.e., duration) and frequency (i.e., intensity) of their participation in organized activities (Simpkins, Little, & Weiss, 2004) and may thus have varied opportunities for developmental gains. Yet, with relatively few recent exceptions, extant research typically compares the outcomes of nonparticipants with the outcomes of those who participate in any type of organized activity for any length of time (see Feldman & Matjasko, 2005, for a review). In the present study, we examine associations between three dimensions of participation in organized activities during adolescence—sponsorship (i.e., school vs. community sponsored), duration, and intensity— and the development of educational, civic, and occupational successes in young adulthood.

Some investigations of sponsorship suggest that participation in school- versus community-sponsored activities should be differentially associated with specific outcomes. In particular, some argue that participation in school-sponsored activities may be more strongly associated with educational outcomes (e.g., Kaufman & Gabler, 2004). A corollary to this view is that participation in community-sponsored activities should be more strongly associated with civic outcomes. The reality, however, is that schoolsponsored activities do not always focus on academics (e.g., service clubs, team sports), and community-sponsored activities do not always focus on promoting civic involvement (e.g., community arts programs, sports leagues). Accordingly, findings on the relations between sponsorship and outcomes are mixed. Analyses of data from the National Education Longitudinal Study (NELS) indicate that participation in schoolsponsored, but not non-school-sponsored, extracurricular activities during high school predicts higher grades in 12th grade and higher rates of college attendance 2 years later, even after controlling for prior grades (Marsh & Kleitman, 2002). Similar analyses of NELS data indicate that more intensive participation in school-sponsored, but not non-school-sponsored, activities is positively associated with attending a competitive postsecondary institution (Swanson, 2002). In contrast, other longitudinal studies have found that youths who participate in school- or community-sponsored activities demonstrate similar educational outcomes (e.g., Casey, Ripke, & Huston, 2005), and at least one cross-sectional study found that participation in community-sponsored activities is associated with better educational outcomes than participation in school-sponsored activities (Brown & Evans, 2002). There is also little support for the contention that participation in community-sponsored, versus school-sponsored, activities leads to greater civic involvement. Analyses of data from NELS and the National Longitudinal Study of Adolescent Health indicate that political involvement during adulthood is more strongly tied to participation in schoolsponsored activities than to participation in community-sponsored activities during adolescence (McFarland & Thomas, 2006). The extent to which sponsorship is associated with differences in youth and young adult outcomes therefore remains unclear. Researchers have just begun to consider whether differences among youths in temporal dimensions of participation are associated with differences in positive outcomes (see Roth, Malone, & Brooks-Gunn, 2007, for a complete review). There are at least two reasons to expect such associations. First, research suggests that relationships with adult leaders, peers, and institutions may at least partly explain the association between participation in organized activities and positive developmental outcomes (Eccles & Gootman, 2002; Roth et al., 1998). These relationships take time to develop, and those who participate over longer periods of time may be more likely to benefit from them. Second, those who participate frequently, over an extended period of time, may reap the rewards of lengthier and more frequent exposure to the opportunities and supports that organized activities provide. Reviews of a variety of types of after-school activities and programs (e.g., Eccles & Gootman, 2002; Feldman & Matjasko, 2005; Roth et al., 2007) reveal a dearth of studies on participation duration. A few recent studies, however, have identified associations between participation in organized activities over multiple years and positive adolescent and young adult outcomes— particularly educational outcomes. Longitudinal analyses of a large

816

GARDNER, ROTH, AND BROOKS-GUNN

sample of high school students indicate that although there are no differences in grades or attitudes toward school between nonparticipants and 1-year participants in school-sponsored activities, youths who participate for 2 or 3 years earn higher grades, demonstrate more positive attitudes toward school, and have greater academic aspirations than nonparticipants (Darling, 2005; Darling, Caldwell, & Smith, 2005). Analyses of NELS data also reveal positive relations between participation in organized activities over three waves of data collection during middle and high school (vs. two waves or fewer) and multiple young adult outcomes (i.e., college attendance, voting, and volunteering at age 20; Zaff et al., 2003). Finally, analyses of data from the Carolina Longitudinal Study demonstrate positive associations between participation in high-school-sponsored activities over 2 years (vs. 1) and college attendance at age 20, particularly for youths with lower interpersonal competence (Mahoney, Cairns, & Farmer, 2003). Findings on the relations between participation intensity and youth development are also limited and somewhat mixed. For instance, some have found significant positive associations between the number of hours that adolescents devote to schoolsponsored extracurricular activities and varied academic outcomes (e.g., high school grades, academic attitudes and aspirations, postsecondary attendance; Darling, 2005; Marsh & Kleitman, 2002), whereas others have identified only marginally significant relations between time spent in school-sponsored extracurricular activities and adolescents’ grades (Cooper, Valentine, Nye, & Lindsay, 1999). The link between academic outcomes and the intensity of youths’ participation in community-sponsored activities is equally unclear. Although some evidence suggests that the intensity of adolescents’ participation in community-sponsored activities is positively associated with achievement test scores (Cooper et al., 1999), studies have also found negative (Marsh & Kleitman, 2002) and nonsignificant relations (Cooper et al., 1999) between more intensive participation in non-school-sponsored activities and grades during adolescence. Finally, recent studies of participation intensity measured without regard to sponsorship suggest that intensity may be positively associated with concurrent (RoseKrasnor, Busseri, Willoughby, & Chalmers, 2006) but not longitudinal (Busseri, Rose-Krasnor, Willoughby, & Chalmers, 2006) measures of high school academic functioning. Further complicating the picture, findings from a few studies indicate that at high levels, intensive participation may be associated with adverse adolescent and young adult outcomes, particularly in the academic domain (Cooper et al., 1999; Marsh & Kleitman, 2002). These findings have been used to make the argument that too much time in organized activities distracts students from other important experiences (e.g., family time, schoolwork) and hinders development. This position, termed the overscheduling hypothesis (Mahoney, Harris, & Eccles, 2006), draws support from a tempered version of a zero-sum theoretical model, which suggests that at high levels of intensity, commitments to extracurricular activities compete with commitments to academic activities and adversely affect academic performance (see discussion of the zero-sum and threshold model in Marsh & Kleitman, 2002). Adverse consequences are often observed only among youths who participate at extremely high levels, however (Cooper et al., 1999; Marsh & Kleitman, 2002). For instance, recent analyses of adolescents’ time diaries from the Panel Study of Income Dynamics—Child Development Supplement II found

that among some youths (often African American), participating in organized activities for more than 20 hr per week predicted slightly lower scores on a few measures of well-being (e.g., self-esteem, frequency of parent–adolescent discussions) as compared with less frequent participation (Mahoney et al., 2006). Participating at this high level generally predicted more positive outcomes than nonparticipation, however. Moreover, among those participating less than 20 hr per week, more frequent participation generally predicted better well-being (e.g., reading achievement, psychological adjustment, lack of substance use, and good parent–adolescent relationships). These findings support the contention that within the normal range, more intensive participation is often associated with more positive outcomes.

The Present Study
Extant research suggests that relations between participation in organized activities and positive youth and young adult development may vary as a function of sponsorship, duration, and intensity. Several unanswered questions remain, however. First, investigations of these three dimensions rarely overlap. It is therefore unclear whether and to what extent the implications of sponsorship, duration, and intensity vary as a function of each other. For instance, we know very little about differences in the associations between temporal participation measures and youth development across sponsorship categories. Additionally, it is unclear whether the positive associations between participation intensity and youth development vary as a function of duration. For instance, participation duration may be so strongly associated with positive outcomes that among, youths who score high on duration, more intensive participation offers few additional benefits. The benefits related to high scores on duration may represent the ceiling of benefits attributable to participation in organized activities. Second, although some researchers have examined associations between participation in organized activities and young adult outcomes immediately following high school (i.e., age 20, or 2 years after high school), the lack of available follow-up data has, until recently, prevented the examination of relations between participation and outcomes later in young adulthood. Thus, it is unclear whether the positive outcomes associated with participation in organized activities are maintained throughout young adulthood. We address this question, examining positive young adult outcomes both 2 and 8 years after high school. We focus on outcomes in three domains that are developmentally relevant for young adults: educational attainment, civic engagement, and occupational success. These three domains tap into four of the five Cs: Educational attainment reflects positive development in the academic area of the first C; civic engagement (e.g., volunteering, voting) reflects positive development in select aspects of the third, fourth, and fifth Cs; and occupational success reflects positive development in the vocational area of the first C. Because our data do not contain a measure of the second C (i.e., confidence and positive identity), it was not possible to examine relations between participation measures and the higher order “thriving” construct that arises from the five Cs. Finally, it is important to acknowledge that young adults who are successful in one domain are often successful in others. For instance, U.S. Census data indicate that a greater percentage of college-educated young adults (67%) than non-college-educated

ORGANIZED ACTIVITIES AND DEVELOPMENTAL SUCCESS

817

young adults (38%) voted in a recent election (Holder, 2006). College-educated young adults also earn more income than noncollege-educated young adults (U.S. Department of Education, 2005). One plausible explanation for this pattern of findings is that postsecondary education heightens civic awareness and opens the door to career opportunities that are unavailable to those who do not continue their education beyond high school. Consequently, we examine the extent to which educational attainment mediates positive relations between participation duration and intensity in adolescence and civic and occupational success in young adulthood. To our knowledge, prior studies have not explicitly examined these pathways. In summary, using a national longitudinal data set, we attempt to address three sets of unanswered questions: 1. Is the duration of youths’ participation in organized activities during high school positively associated with educational, civic, and occupational success as few as 2, and as many as 8, years after high school? Do these relations vary as a function of sponsorship? 2. Among those who participate in organized activities during high school, is more intensive participation linearly associated with better educational, civic, and occupational outcomes 2 and 8 years after high school? Or, are there diminishing returns in young adult outcomes at very high levels of intensity? Do associations between intensity and young adult outcomes vary as a function of duration? Do relations between intensity and young adult outcomes vary as a function of sponsorship? 3. Are associations between temporal participation measures and positive civic and occupational outcomes mediated by educational attainment? In an effort to reduce the influence of selection biases, our analyses control for demographic characteristics (gender, race/ ethnicity, and socioeconomic status) that have been linked to participation in organized activities (Marsh & Kleitman, 2002; Wimer et al., 2006) and to developmental success in young adulthood (Holder, 2006; Lopez & Kirby, 2005; Planty, Regnier, & Owings, 2003; U.S. Department of Education, 2005). Although it was impossible to control for performance on prior measures of outcome variables (i.e., our models predict young adult outcomes that have no adolescent equivalents), we control for a variety of academic, behavioral, and familial constructs that capture important dimensions of adolescents’ overall adjustment. Prior findings from NELS suggest that constructs in these domains have significant implications for participation in organized activities and positive young adult outcomes (e.g., Marsh & Kleitman, 2002; Zaff et al., 2003). We acknowledge that many other unmeasured characteristics may influence selection into activities, and that the power to draw causal inference is therefore limited.

school (N 14,915; 56.2% of 8th graders plus 1,093 new responders from 10th- and 12th-grade follow-ups), and 8 years after high school (N 12,144; 46.3% of 8th graders plus 760 new responders from 10th- and 12th-grade follow-ups).2 The sample for the present study includes those who completed assessments in 10th grade, 12th grade, and 2 and/or 8 years after high school (N 14,038). From this sample, we excluded those who had dropped out of high school because they lacked data on school-based extracurricular activities (n 2,009). Additionally, students who had moved to a new high school between 10th and 12th grades (n 848) or who had missing data on this variable (n 152) were excluded owing to potential substantive differences in their high school experiences. The analytic sample consists of the remaining 11,029 students (56.9% of the 10th-grade sample).3 Using the entire sample of individuals who had completed at least one of the surveys administered at 10th grade, 12th grade, 2 years after high school, or 8 years after high school (N 20,557), we compared our analytic sample (N 11,029) with excluded youths (N 9,528) on three demographic dimensions: gender, race/ethnicity, and parent education. There were no significant differences between the analytic and excluded samples with respect to gender composition. There were, however, differences among the two samples on race/ethnicity and parent education. Relative to the analytic sample, excluded youths were less likely to be Asian/Pacific Islander (3.9% excluded vs. 7.6% analytic), 2 (1) 101.249, p .001, and more likely to be Hispanic (13.5% excluded vs. 12.1% analytic), 2(1) 7.190, p .01, or Native American (1.4% excluded vs. 0.9% analytic), 2(1) 9.292, p .01.4 Parents of excluded youths, relative to parents of youths in the analytic sample, were also less likely to have obtained at least some postsecondary education (68.4% excluded vs. 71.1% analytic), 2(1) 14.962, p .001. Table 1 presents demographic information for the analytic sample.

Method Sample
Data for the present study come from NELS, a national, longitudinal study that began with an assessment of 24,599 8th graders during the spring of 1988. Follow-up assessments were administered during 10th grade (N 19,394; 70.8% of 8th graders plus 1,970 new responders added at 10th-grade follow-up), 12th grade (N 19,220; 69.6% of 8th graders plus 2,093 new responders added at 10th- and 12th-grade follow-ups), 2 years after high

The number of new responders at 10th and 12th grades that we report (i.e., those who did not complete a questionnaire at 8th grade but did complete a questionnaire at 10th or 12th grade) differs from a similar set of numbers provided in official NELS documentation. The NELS documentation reports the number of individuals added to the 8th-grade sample at the 10th- and 12th-grade follow-ups (the resulting 10th- and 12th-grade samples are referred to as “freshened” samples). However, in NELS documentation, the Ns for youths added to the freshened samples at 10th and 12th grades reflect only those added to the number of sampled 8th-grade youths (N 26,432). Not all sampled youths completed an 8th-grade questionnaire, however. The Ns for new responders reported in this article—which also include nonresponders sampled at 8th grade who completed 10th- and/or 12th-grade questionnaires—are larger than the Ns reported for youths in the freshened 10th- and 12th-grade samples in NELS documentation. 3 Sample sizes for each analysis vary from this figure owing to missing data on the dependent and/or independent variables. The tables contain the exact sample size for each analysis. 4 Frequencies for ethnicity and parent education reported for the analytic sample in the text differ slightly from those reported for the analytic sample in Table 1. Percentages reported in the text are based on 2 2 crosstabulations that use listwise deletion. In the text, missing data for individuals in the excluded sample affect percentages reported for the analytic sample.

2

818

GARDNER, ROTH, AND BROOKS-GUNN

Table 1 Frequencies and Descriptive Statistics for Participation, Outcome, and Control Variables
Frequencies Variable N % Participation variables Duration School-sponsored activities Nonparticipants 1-year participants 2-year participants Community-sponsored activities Nonparticipants 1-year participants 2-year participants Intensity School-sponsored activitiesa 1-year participantsb 2-year participantsc Community-sponsored activitiesd 1-year participantsb 2-year participantsc M Descriptive statistics SD Min Max

771 1,844 7,708 1,641 2,832 5,830

7.5 17.9 74.7 15.9 27.5 56.6 1.06 2.19 0.55 0.78 Young adult outcomes 1.28 1.24 0.38 0.37 0.00 0.00 0.20 0.20 5.00 5.00 3.00 2.60

Two years after high school Postsecondary attendance Yes No Volunteering Yes No Voting Yes No Eight years after high school Postsecondary completion Yes No Volunteering Yes No Voting Yes No Full-time employment Yes No Prior year income

8,175 2,753 4,589 5,416 6,067 4,871 3,727 5,484 3,114 6,029 4,111 5,019 7,123 2,138

74.8 25.2 45.9 54.1 55.5 44.5 40.5 59.5 34.1 65.9 45.0 55.0 76.9 23.1 $26,086 Control variables $20,955 $0 $500,000

Gender Male Female Ethnicity Black Hispanic Asian/Pacific Islander Native American White Parent education Greater than high school High school or less 10th/12th test scores 10th/12th problem behavior 10th/12th parental monitoring
a b

5,439 5,590 1,125 1,356 842 105 7,576 7,627 3,139

49.3 50.7 10.2 12.3 7.6 1.0 68.8 70.8 29.2 51.60 0.55 2.21 9.67 0.74 0.58 27.86 0.00 1.00 71.82 4.00 3.00

School-sponsored intensity scores reflect the total number of hours spent in an average week on all combined school-sponsored extracurricular activities. For 1-year participants intensity scores reflect the frequency of participation during the single year in which youth participated. c For 2-year participants intensity scores reflect the frequency of participation averaged across 10th and 12th grades. d Community-sponsored intensity scores were computed by averaging frequency scores across five activities (religious activities, community youth groups, music/art/dance lessons, sports lessons, and community service).

ORGANIZED ACTIVITIES AND DEVELOPMENTAL SUCCESS

819

Measures Activity Participation Variables
Participation duration. At the 10th- and 12th-grade assessments, students reported whether (i.e., yes or no) and at what level (i.e., participant or leader) they participated in each of the following school-sponsored activities during the past year: sports, cheerleading, pom-pom/drill teams, music groups, plays/musicals, student government, academic honor societies, newspaper/yearbook, service clubs, academic clubs, hobby clubs, and FFA/FHA/FTA (Future Farmers, Homemakers, and Teachers of America) organizations. Students also indicated how frequently they participated in each of following five organized community-sponsored activities during the past year: religious activities, community youth groups, music/art/dance lessons, sports lessons, or community service.5 We used responses to these items to construct four dichotomous measures of participation: participation in at least one schoolsponsored activity during 10th grade, participation in at least one school-sponsored activity during 12th grade, participation in at least one community-sponsored activity during 10th grade, and participation in at least one community-sponsored activity during 12th grade.6 We constructed separate measures of the duration of youths’ participation in school- and community-sponsored activities using these dichotomous participation variables. For each of these two types of activities, youths who did not participate during 10th or 12th grade were assigned a score of 0 (nonparticipants), youths who participated in 10th or 12th grade but not both grades were assigned a score of 1 (1-year participants), and youths who participated in both 10th and 12th grades were assigned a score of 2 (2-year participants). See Table 1 for frequencies. Participation intensity. We used data on participation frequency to construct intensity variables. Students reported the total number of hours that they spent in an average week on all combined school-sponsored extracurricular activities during 10th and 12th grades. We recoded these response scales for consistency across 10th and 12th grades such that 0 no participation in a typical week, 1 less than 1 hr per week, 2 1– 4 hr per week, 3 5–9 hr per week, 4 10 –19 hr per week, and 5 20 hr or more per week. Students also indicated how often they participated in each of the five aforementioned organized communitysponsored activities at each assessment (0 never or rarely, 1 less than once per week, 2 at least once/twice per week, and 3 everyday/almost everyday). Frequency scores were averaged across these five items within each assessment period to yield separate 10th- and 12th-grade measures of community-sponsored participation intensity. Among 2-year participants, frequency scores were averaged across 10th and 12th grades within sponsorship categories, yielding separate measures of school- and community-sponsored participation intensity. For 1-year participants, school- and community-sponsored intensity scores reflect only the intensity of participation during the year that youths participated. We present descriptive statistics for all four intensity variables in Table 1.

ment, civic engagement, and occupational success. In analyses of outcomes 2 years after high school we examined only educational attainment and civic engagement. We did not examine occupational measures 2 years after high school because many individuals at this age are enrolled in school full time. We examined successes in all three domains 8 years after high school. All outcome variables are described below, and descriptive statistics are presented in Table 1. Educational attainment. We used postsecondary attendance (i.e., whether participants attended a postsecondary institution at any time after completing high school) as a measure of educational attainment 2 years after high school. Eight years after high school, attainment was operationalized as completion of a 4-year postsecondary degree. We used self-reports of young adults’ highest level of education to construct these dichotomous variables. Civic engagement. We examined two dimensions of civic engagement: volunteering and voting. Dichotomous measures of both outcomes were constructed 2 and 8 years after high school. Two years after high school, we defined volunteering as participation in at least one service activity sponsored by a postsecondary institution (i.e., volunteering services to fellow students, volunteering services to community groups) or a community-based agency (e.g., organized volunteer work such as at a hospital, volunteer work for trade or professional organizations). Eight years after high school, volunteering was defined as participation in one or more community-based service activities during the past year (i.e., volunteering for youth organizations, civic/community volunteering). We used single items to measure voting 2 and 8 years after high school. Two years after high school individuals indicated whether they voted in a state, local, or national election during the past year. Eight years after high school participants responded to the same question but reflected on voting behavior over the prior 2 years. Occupational success. We examined two measures of occupational success: full-time employment and income. Eight years after high school, a dichotomous measure of full-time employment (i.e., 35 hr or more per week) was constructed using self-reports of hours worked per week. Self-reports of annual income, measured continuously, were also obtained for the year prior to the 8-year post-high-school assessment (i.e., 1999). The income distribution was positively skewed. Thus, we computed and used the natural log of income in analyses.
Students also indicated whether they participated in organized activities at the 8th-grade assessment. Because data on participation intensity were not collected during 8th grade, we used only 10th- and 12th-grade activity data to construct participation variables. 6 The response scale for questions about community-sponsored participation ranged from 0 to 3, where 0 never or rarely participate, 1 participate less than once per week, 2 participate at least once or twice per week, 3 participate every day or almost every day. These items were recoded into dichotomous variables, where responses of never or rarely were reassigned a score of 0, and responses ranging from less than once per week to every day or almost every day were reassigned a score of 1. Thus, a score of 0 may occasionally indicate very rare participation rather than nonparticipation. Previous studies using NELS data have recoded and dichotomized responses to these items in the same manner (e.g., Zaff et al., 2003).
5

Young Adult Outcome Variables
We used data collected 2 and 8 years after high school to construct outcome variables in three domains: educational attain-

820 Control Variables

GARDNER, ROTH, AND BROOKS-GUNN

We controlled for gender (1 female), race/ethnicity (four dummy variables: Black, Hispanic, Asian/Pacific Islander, Native American vs. White), and socioeconomic status (SES) in all models. We used a binary parent education measure as a proxy for SES (1 some education beyond high school). Additionally, we controlled for three measures of individual and family adjustment. We averaged (a) 10th- and 12th-grade scores on standardized tests (i.e., composite of math and reading scores) to yield a composite indicator of high school achievement; (b) 10th- and 12th-grade scores on a measure of getting into trouble at school (0 never and 4 10 or more times per week) to yield a composite indicator of high school behavior problems; and (c) 10th- and 12th-grade scores on an item indicating parents’ knowledge of their children’s friends’ parents (1 knows none of friends’ parents and 3 knows many of friends’ parents) to yield a composite indicator of parental monitoring (see Table 1 for descriptive statistics). All three adjustment variables were centered at their means.

Results
We followed a two-part analysis plan. First, we examined direct and indirect relations between the duration of youths’ participation in school- and community-sponsored activities during high school and positive young adult outcomes 2 and 8 years after high school. Second, we examined direct and indirect relations between participation intensity and young adult outcomes in two subsamples of youths, 1- and 2-year participants. We conducted analyses in Stata 8.0 (Stata Corporation, 2003) using robust standard errors to account for the clustering of participants within high schools. Because of the large sample size and the number of models tested, we set the statistical significance criterion at p .01 in all analyses.

and the outcome (i.e., a significant c path) and a significant relation between duration and our mediating variable, postsecondary attendance or completion (i.e., a significant a path). When these criteria were met, we included postsecondary attendance or completion as an additional predictor in models predicting civic and occupational outcomes (see Model 2 results in Tables 3, 5, and 6). We used Sobel tests to determine the statistical significance of indirect relations through postsecondary education when the above described models revealed significant c, a, and b paths. The Sobel test assesses the significance of the product a b, which is approximately equivalent to the quantity c – c when using linear regression (MacKinnon, Warsi, & Dwyer, 1995). A standard error and z score for the term a b can be calculated using the following formulas: z score ab/SEab, where SEab (a2SEb2 b2SEa2)1/2 (Baron & Kenny, 1986; Sobel, 1982). Because the Sobel test was developed for use with linear models, a and b coefficients and their standard errors were appropriately transformed using procedures for logistic models outlined in MacKinnon and Dwyer (1993). The coefficients for the indirect paths computed using the Sobel method are thus reported as transformed indirect effects (TIEs). Results support complete mediation when (a) the TIE is significant and (b) introducing postsecondary attendance or completion as a predictor reduces the relation between duration and the outcome to nonsignificance. Partial mediation, on the other hand, is supported when the TIE is significant but introducing postsecondary attendance or completion as a predictor does not completely eliminate the significant relation between duration and the outcome.

Direct Relations Between Participation Duration and Educational Attainment
The duration of participation in school-sponsored activities was positively associated with educational attainment (see Table 2). Although 1-year participants and nonparticipants did not differ with respect to postsecondary attendance 2 years after high school, the odds of attending a postsecondary institution were 54% higher among 2-year participants than among 1-year participants (z 6.15, p .001). Eight years after high school, the odds of completing a postsecondary degree were 36% lower among nonparticipants than
7 The goal of this study was to examine the relations between participation and intensity and young adult outcomes net of controls (rather than to examine the attenuating effects of specific controls). We therefore do not present results for models without controls. Tables for models without controls are available from Margo Gardner upon request. 8 The primary goal of the analyses of participation duration was not to determine whether 2 years of participation was associated with more favorable outcomes than nonparticipation. As previously stated, numerous studies have compared youths who participate in organized activities with youths who do not participate. Our primary interest was in determining whether 2 years of participation was associated with more favorable outcomes than 1 year of participation. Thus, 1-year participants, rather than nonparticipants, were assigned to the reference category. We did, however, conduct preliminary bivariate analyses to check the assumption that 2-year participants demonstrate better young adult outcomes than nonparticipants. We found that, with one exception, 2-year participants showed more favorable outcomes than both 1-year participants and nonparticipants. The only deviation from this pattern was the absence of any differences among the three participation categories with respect to full-time employment 8 years after high school.

Participation Duration and Outcomes in Young Adulthood
Analyses examining participation duration proceeded in two stages. First, we used linear regression (in analyses of income) and logistic regression (in all other analyses) to examine the relations between the duration of participation in school- and communitysponsored activities and young adult outcomes, net of all controls.7 Duration variables were converted to dummy variables within each sponsorship category, and 1-year participants were assigned to the reference category (see Table 2 and Model 1 analyses in Table 3).8 Second, we tested the extent to which postsecondary attendance (for 2-year outcomes) and completion (for 8-year outcomes) mediated relations between participation duration and young adult civic and occupational outcomes. We followed Baron and Kenny’s (1986) steps for testing mediation, which require (a) a significant relation, c, between the independent and dependent variables; (b) a significant relation, a, between the independent variable and the mediator; (c) a significant relation, b, between the mediator and the dependent variable while controlling for the independent variable; and (d) a significant reduction in the relation between the independent and dependent variables when the mediator is included as a predictor in the model (in this step, the relation between the independent and dependent variables is denoted c ). In keeping with these guidelines, we tested for mediation only where prior analyses identified a significant direct relation between duration

ORGANIZED ACTIVITIES AND DEVELOPMENTAL SUCCESS

821

Table 2 Direct Relations (Odds Ratios) Between Participation Duration and Educational Attainment
Postsecondary attendance 2 years after high school 1.30** 1.41** 1.54** 2.11** 0.91 2.09** 1.10** 0.81** 1.19* 0.78 1.54** 0.82* 1.44** 1,683.12** 13 9,679 Postsecondary completion 8 years after high school 1.21* 1.32 0.79 1.93** 0.42 2.13** 1.13** 0.83** 1.30** 0.64* 1.78** 0.76* 1.49** 1,719.45** 13 8,289

Variable Control variables Female Black Hispanic Asian/Pacific Islander Native American Parent education high school Test scores Behavior problems Parental monitoring Duration variablesa School-sponsored activities 0 years 2 years Community-sponsored activities 0 years 2 years Wald 2 df N

Note. Each column is a separate regression model that predicts postsecondary outcomes. a One-year participants are the reference group within each sponsorship category. * p .01. ** p .001.

among those who participated in school-sponsored activities for 1 year (z –3.01, p .01) and 78% higher among those who participated in school-sponsored activities for 2 years than among those who participated for only 1 year (z 7.04, p .001). There were differences across all three community-sponsored participation categories with respect to educational attainment (see Table 2). Two years after high school, the odds of attending a postsecondary institution were 18% lower among nonparticipants than among 1-year participants (z –2.62, p .01) and 44% higher among 2-year participants than among 1-year participants (z 5.43, p .001). Eight years after high school, the odds of completing a postsecondary degree were 24% lower among nonparticipants than among those who participated in communitysponsored activities for 1 year (z –2.77, p .01) and 49% higher among those who participated in community-sponsored activities for 2 years than among those who participated for only 1 year (z 6.19, p .001).

Direct and Indirect Relations Between Participation Duration and Civic Engagement
Volunteering. Participation in school-sponsored activities was positively associated with volunteering. Although 1-year participants and nonparticipants did not differ 2 years after high school, Model 1 analyses (see Table 3) indicate that the odds of volunteering 2 years after high school were 56% greater among those who participated in school-sponsored activities for 2 years than among those who participated for 1 year (z 6.45, p .001). Postsecondary attendance partially mediated this relation (see Model 2, Table 3; TIE .016, z 5.54, p .001). Eight years after high school, Model 1 analyses (see Table 3) suggest that the

odds of volunteering were 46% lower among nonparticipants than among 1-year participants in school-sponsored activities (z – 4.69, p .001) and 21% higher among 2-year participants than among 1-year participants (z 2.70, p .01). The inverse relation between nonparticipation (vs. 1 year of participation in schoolsponsored activities) and volunteering 8 years after high school was partially mediated by postsecondary completion (see Model 2, Table 3; TIE –.004, z –2.62, p .01), and the positive relation between 2 years (vs. 1 year) of participation in schoolsponsored activities and 8-year post-high-school volunteering was completely mediated by postsecondary completion (see Model 2, Table 3; TIE .008, z 4.23, p .001). Participation in community-sponsored activities was also positively associated with volunteering. Two years after high school, Model 1 analyses (see Table 3) indicate that the odds of volunteering were 33% lower among nonparticipants than among 1-year participants (z – 4.94, p .001) and 91% higher among 2-year participants than among 1-year participants (z 11.92, p .001). Postsecondary attendance partially mediated both of these relations (see Model 2 in Table 3; TIE n o n p a r t i c i p a n t s – 0.006, z –2.57, p .01; TIE2-year participants 0.015, z 5.00, p .001). Eight years after high school, Model 1 analyses (see Table 3) indicate that although youths who participated in community-sponsored activities for 1 year did not differ from nonparticipants, the odds of volunteering were 69% greater among those who participated in community-sponsored activities for 2 years than among those who participated for only 1 year (z 8.97, p .001). The positive association between 2 years of participation in community-sponsored activities and volunteering 8 years after

822

GARDNER, ROTH, AND BROOKS-GUNN

Table 3 Regression Coefficients for Relations Between Participation Duration and Civic and Occupational Outcomes
Two years after high school Volunteering Variable Controls Female Black Hispanic Asian/Pacific Islander Native American Parent high school Test scores Behavior problems Parental monitoring Durationb School-sponsored 0 years 2 years Community-sponsored 0 years 2 years Postsecondary education Wald 2 or Fc df N M1a 0.99 1.33* 1.15 1.07 1.52 1.35** 1.04** 0.90* 1.32** 0.82 1.56** M2 0.97 1.26* 1.07 1.00 1.53 1.21** 1.02** 0.93 1.28** 0.88 1.47** M1 0.93 0.75** 0.66** 0.35** 0.74 1.41** 1.02** 0.90* 1.16** 0.85 1.25** Voting M2 0.92 0.74** 0.65** 0.35** 0.76 1.36** 1.02** 0.91* 1.15* 0.86 1.21* Volunteering M1 1.29** 1.68** 1.19 1.11 1.79 1.21* 1.02** 1.00 1.36** 0.54** 1.21* M2 1.27** 1.65** 1.20 1.08 1.84 1.16* 1.01* 1.02 1.34** 0.54** 1.18 M1 0.98 1.34* 0.91 0.52** 1.03 1.11 1.01** 0.89* 1.26** 0.91 1.21* Eight years after high school Voting M2 0.97 1.33* 0.91 0.51** 1.04 1.10 1.01** 0.90* 1.25** 0.91 1.19* Employment M1 0.44** 1.06 0.79* 0.66** 0.64 0.97 0.99 0.91 1.10 1.11 1.11 M2 — — — — — — — — — — — — — — — — — Income M1 0.33** 0.03 0.05 0.00 0.36* 0.02 0.01** 0.01 0.07** 0.00 0.03 0.05 0.02 — 33.16 13, 1224 7,370 M2 — — — — — — — — — — — — — — — — —

0.67** 0.69** 0.85 0.86 0.87 0.88 0.76* 0.77* 0.99 1.91** 1.85** 1.24** 1.21** 1.69** 1.64** 1.29** 1.27** 1.01 ** ** ** — 2.35 — 1.41 — 1.36 — 1.13 — 938.33 1,069.61 580.30 616.65 498.34 514.04 334.69 333.94 255.22 13 14 13 14 13 14 13 14 13 8,887 8,857 9,687 9,648 8,232 8,173 8,221 8,162 8,338

Note. Each column is a separate regression model that predicts civic or occupational outcomes. All coefficients are odds ratios from logistic models, except for coefficients in income models, which are unstandardized betas from a linear model. a M1 Model 1; M2 Model 2. Dash indicates term or model was not tested. M2 was tested only when previous analyses (see Table 2) found direct relations between duration and postsecondary attendance/completion and M1 analyses found significant direct relations between at least one of the four duration variables in a given model and the outcome. b One-year participants are the reference group in each sponsorship category. c Multivariate Wald chi-square values are reported for logistic models, and multivariate F statistics are reported for the linear income model (all are significant at p .001). * p .01. ** p .001.

high school was partially mediated by postsecondary completion (see Model 2, Table 3; TIE .006, z 4.02, p .001). Voting. School-sponsored participation was associated with voting only when youths participated for 2 years. There were no differences between nonparticipants and 1-year participants on voting 2 years after high school. Model 1 analyses (see Table 3) indicate that the odds of voting 2 years after high school were 25% higher among 2-year participants than among 1-year participants (z 3.69, p .001), however. Postsecondary attendance partly mediated this association (see Model 2, Table 3; TIE .007, z 4.45, p .001). Eight years after high school, although there were no differences in voting between nonparticipants and those who participated in school-sponsored activities for 1 year, Model 1 (see Table 3) analyses indicate that the odds of voting 8 years after high school were 21% higher among those who participated in schoolsponsored activities for 2 years compared with those who participated for 1 year (z 2.82, p .01). This relation was not mediated by postsecondary completion. Community-sponsored participation was also associated with voting. Although there were no differences between nonparticipants and 1-year participants 2 years after high school, Model 1 analyses (see Table 3) indicate that the odds of voting 2 years after high school were 24% higher among 2-year participants than among 1-year participants (z 4.10, p .001). Postsecondary attendance partially mediated this association (see Model 2 in

Table 3; TIE .006, z 4.15, p .001). Eight years after high school, Model 1 analyses (see Table 3) indicate that the odds of voting were 24% lower among nonparticipants than among those who participated in community-sponsored activities for 1 year (z –3.39, p .01) and 29% higher among those who participated in community-sponsored activities for 2 years than among those who participated for only 1 year (z 4.62, p .001). These relations were entirely direct and not mediated by postsecondary completion.

Direct and Indirect Relations Between Participation Duration and Occupational Outcomes
Full-time employment. Model 1 analyses (see Table 3) revealed no significant relations between the duration of participation in school- and community-sponsored activities and the odds of full-time employment 8 years after high school. Income. Model 1 analyses (see Table 3) revealed no significant relations between the duration of participation in school- or community-sponsored activities and income 8 years after high school.

Participation Intensity and Outcomes in Young Adulthood
Intensity was measured differently for 1- and 2-year participants. For 2-year participants, intensity scores reflect the average

ORGANIZED ACTIVITIES AND DEVELOPMENTAL SUCCESS

823

Table 4 Direct Relations (Odds Ratios) Between Participation Intensity and Educational Attainment
Postsecondary attendance 2 years after high school Variable M1a M2 Postsecondary completion 8 years after high school M1 M2

Sample: 1-year participants Control variables Female Black Hispanic Asian/Pacific Islander Native American Parent education high school Test scores Behavior problems Parental monitoring Intensity School-sponsored Community-sponsored Wald 2 df N 1.38* 1.06 1.49 1.81 0.96 1.84** 1.08** 0.84 1.01 1.22** — 251.94** 10 1,680 1.23 1.47 1.64** 4.93** 1.08 1.97** 1.10** 0.72** 1.11 — 1.49* 453.83** 10 2,642 1.21 1.50 0.83 2.27* 0.23 1.95** 1.13** 0.79 0.96 1.19* — 217.75** 10 1,373 1.23 1.20 0.82 2.72** 0.66 1.67** 1.14** 0.78* 1.44** — 1.42 420.45** 10 2,218

Sample: 2-year participants Control variables Female Black Hispanic Asian/Pacific Islander Native American Parent education high school Test scores Behavior problems Parental monitoring Intensity School-sponsored Community-sponsored Wald 2 df N
a

1.34** 1.72** 1.57** 2.01* 0.70 2.13** 1.10** 0.82** 1.26** 1.31** — 1,122.36** 10 7,356

1.34** 1.32 1.38 1.26 0.81 2.09** 1.11** 0.87 1.33** — 1.33* 718.46** 10 5,557

1.29** 1.33 0.79 1.81** 0.42 2.20** 1.13** 0.84* 1.32** 1.26** — 1,346.93** 10 6,394

1.14 1.23 0.75 1.62** 0.40 2.24** 1.13** 0.83* 1.28** — 1.54** 1,012.12** 10 4,855

Note. Each column is a separate model that predicts postsecondary outcomes. M1 Model 1; M2 Model 2. Dash indicates term was not tested. * p .01. **p .001.

frequency of participation across 2 years. For 1-year participants, intensity scores reflect the frequency of participation during a 1-year period. These differences necessitated separate analyses for 1- and 2-year participants.9 Analyses for both subsamples proceeded in three stages. First, we used logistic and linear regression to examine direct relations between participation intensity and young adult outcomes after including controls (see Table 4 and Model 1 analyses in Tables 5– 6). School- and communitysponsored participation intensity variables were entered as predictors in separate regression models in order to prevent limiting the sample to those who participated in both kinds of activities (i.e., because analyses use listwise deletion, cases with intensity data for only one type of activity would be excluded from models examining both types of activities simultaneously). Second, we added quadratic intensity terms to these models to test for nonlinear relations. Quadratic models were considered an improvement over

linear models if the quadratic term was significant at p .01 and the model R2 or pseudo-R2 value increased significantly upon including the quadratic term. Findings are reported only for quadratic models that meet these criteria. Third, following the procedure described for analyses of duration, we examined indirect relations between participation intensity and young adult civic and occupational outcomes via postsecondary attendance and completion (see Model 2, Tables 5– 6).

9 As noted above, intensity values were computed differently for 1versus 2-year participants. Consequently, it was not possible to construct a single intensity variable to be used in analyses examining interactions between duration and intensity. We conducted separate intensity analyses for 1- and 2-year participants as an alternative to interaction analyses.

824

GARDNER, ROTH, AND BROOKS-GUNN

Direct Relations Between Participation Intensity and Educational Attainment
More intensive participation in school-sponsored activities was associated with greater educational attainment (see Model 1 analyses, Table 4). Two years after high school, a one-unit increase in school-sponsored participation intensity was associated with a 22% increase in the odds of postsecondary attendance among 1-year participants (z 4.39, p .001) and a 31% increase in the odds of postsecondary attendance among 2-year participants (z 9.22, p .001). Eight years after high school, a one-unit increase in school-sponsored participation intensity was associated with a 19% increase in the odds of postsecondary completion among 1-year participants (z 2.99, p .01) and a 26% increase in the odds of postsecondary completion among 2-year participants (z 9.51, p .001). The addition of a quadratic school-sponsored intensity term significantly increased the pseudo-R2 value for the model predicting postsecondary completion among 2-year participants (R2linear model .2325, R2quadratic model .2333), F(1, 6382) 6.66, p .01 (quadratic model intensity coefficients: ORlinear 1.58, z 5.27, p .001; ORquadratic 0.95, z –2.66, p .01). The positive association between school-sponsored participation intensity and the odds of postsecondary completion reached a plateau at an intensity score of 4.60. More intensive participation in community-sponsored activities was linearly associated with greater educational attainment (see Model 2 analyses, Table 4). Two years after high school, a oneunit increase in community-sponsored participation intensity was associated with a 49% increase in the odds of postsecondary attendance among 1-year participants (z 3.27, p .01) and a 33% increase in the odds of postsecondary attendance among 2-year participants (z 2.65, p .01). Eight years after high school, more intensive community-sponsored participation was not associated with postsecondary completion among 1-year participants. A one-unit increase in community-sponsored participation intensity was, however, associated with a 54% increase in the odds of postsecondary completion among 2-year participants (z 4.59, p .001).

Direct and Indirect Relations Between Participation Intensity and Civic Engagement
Volunteering. More intensive participation in schoolsponsored activities was generally associated with better odds of volunteering. Two years after high school, Model 1 results (see Table 5) indicate that a one-unit increase in school-sponsored participation intensity was associated with an 18% increase in the odds of volunteering among 1-year participants (z 3.85, p .001). The addition of a quadratic school-sponsored intensity term, however, significantly increased the pseudo-R2 value for this model (R2linear model .0418, R2quadratic model .0507), F(1, 1467) 13.75, p .01 (quadratic model intensity coefficients: ORlinear 1.88, z 5.06, p .001; ORquadratic 0.88, z –3.95, p .001). Among 1-year participants, the odds of volunteering 2 years after high school increased as intensity increased from 0 (no participation during a typical week) to 2.41 (between 1– 4 and 5–9 hr per week) but decreased thereafter. Moreover, the odds of volunteering were slightly lower above an intensity score of 4.83 (nearly 20 hr per week) than at an intensity score of 0. This

curvilinear relation was not mediated by postsecondary attendance (i.e., the quadratic term for school-sponsored participation intensity among 1-year participants did not predict postsecondary attendance). More intensive school-sponsored participation among 2-year participants, however, was linearly related to volunteering 2 years after high school (OR 1.23, z 9.57, p .001). This linear relation was partially mediated by postsecondary attendance (see Model 2, Table 5; TIE .025, z 7.04, p .001). Eight years after high school, school-sponsored participation intensity was not associated with volunteering among 1-year participants. Among 2-year participants, however, Model 1 results (see Table 5) suggest that a one-unit increase in school-sponsored participation intensity was associated with a 14% increase in the odds of volunteering 8 years after high school (z 6.15, p .001). Postsecondary completion partly mediated this relation (see Model 2, Table 5; TIE .010, z 4.15, p .001). More intensive participation in community-sponsored activities was also associated with better odds of volunteering. Two years after high school, Model 1 analyses (see Table 6) indicate that a one-unit increase in community-sponsored participation intensity was associated with a 46% increase in the odds of volunteering among 1-year participants (z 3.43, p .01) and a dramatic 196% increase in the odds of volunteering among 2-year participants (z 12.91, p .001). For 1-year participants, this relation was partially mediated by postsecondary attendance (see Model 2, Table 6; TIE .017, z 3.06, p .01). The TIE through postsecondary attendance was not significant for 2-year participants. Eight years after high school, Model 1 results (see Table 6) suggest that a one-unit increase in community-sponsored participation intensity, though not associated with volunteering among 1-year participants, was associated with a 132% increase in the odds of volunteering among 2-year participants (z 9.90, p .001). Postsecondary completion partially mediated this association (see Model 2, Table 6; TIE .005, z 3.03, p .01). Voting. We found limited relations between schoolsponsored participation intensity and voting. Two years after high school, Model 1 analyses (see Table 5) reveal no direct relations between school-sponsored intensity and voting among 1- or 2-year participants. Eight years after high school, Model 1 results (see Table 5) indicate that a one-unit increase in school-sponsored participation intensity was directly associated with a 14% increase in the odds of voting among 1-year participants (z 2.92, p .01). This relation was not mediated by postsecondary completion. Among 2-year participants, school-sponsored participation intensity was not associated with voting 8 years after high school. We also found few positive associations between communitysponsored participation intensity and voting. Two years after high school, more intensive participation in community-sponsored activities was not associated with voting among 1-year participants. Among 2-year participants, however, Model 1 results (see Table 6) indicate that a one-unit increase in participation intensity was directly associated with a 26% increase in the odds of voting 2 years after high school (z 2.88, p .01). This relation was not mediated by postsecondary attendance (i.e., the TIE was not significant). Similarly, 8 years after high school, more intensive participation in community-sponsored activities was not associated with voting among 1-year participants. Among 2-year participants,

ORGANIZED ACTIVITIES AND DEVELOPMENTAL SUCCESS

825

Table 5 Regression Coefficients for Relations Between School-Sponsored Participation Intensity and Civic and Occupational Outcomes
Two years after high school Volunteering Variable M1a M2 Voting M1 M2 Volunteering M1 M2 Eight years after high school Voting M1 M2 Employment M1 M2 M1 Income M2

Sample: 1-year participants in school-sponsored activities Controls Female Black Hispanic Asian/Pacific Islander Native American Parent high school Test scores Behavior problems Parental monitoring Intensity Postsecondary education Wald 2 or Fb df N 0.99 1.09 1.34 0.95 1.94 1.45* 1.03** 0.88 1.38* 1.18* — 65.73 10 1,479 — — — — — — — — — — — — — — 0.85 0.67 0.51** 0.29** 0.51 1.23 1.01 0.92 1.11 1.07 — 69.35 10 1,690 — — — — — — — — — — — — — — 1.45* 1.75* 1.41 0.95 1.98 1.32 1.01 0.96 1.32 1.10 — 38.21 10 1,357 — — — — — — — — — — — — — — 0.90 0.94 0.59* 0.51* 0.57 1.00 1.01 0.94 1.50** 1.14* — 52.51 10 1,355 0.89 0.93 0.60* 0.48* 0.59 0.98 1.00 0.94 1.52** 1.12 1.35 56.31 11 1,346 0.38** 1.20 0.77 0.86 0.41 0.85 1.00 0.89 1.00 0.99 — 55.73 10 1,380 — — — — — — — — — — — — — — 0.37** 0.06 0.12 0.09 0.28 0.02 0.01** 0.00 0.06 0.04 — 11.55 10, 682 1,188 — — — — — — — — — — — — — —

Sample: 2-year participants in school-sponsored activities Controls Female Black Hispanic Asian/Pacific Islander Native American Parent high school Test scores Behavior problems Parental monitoring Intensity Postsecondary education Wald 2 or Fb df N 1.08 1.43** 1.04 1.04 1.37 1.40** 1.04** 0.92 1.33** 1.23** — 437.64 10 6,866 1.05 1.34* 0.98 0.99 1.40 1.27** 1.03** 0.94 1.29** 1.20** 2.26** 531.95 11 6,847 0.99 0.80 0.69** 0.38** 0.86 1.49** 1.03** 0.89* 1.18** 1.04 — 336.88 10 7,355 — — — — — — — — — — — — — — 1.28** 1.79** 1.11 1.16 1.60 1.26** 1.02** 0.98 1.40** 1.14** — 227.10 10 6,355 1.26** 1.76** 1.12 1.14 1.65 1.20* 1.01* 1.00 1.38** 1.13** 1.33** 244.25 11 6,311 1.01 1.52** 0.97 0.55** 1.08 1.22* 1.02** 0.87* 1.26** 1.04 — 157.25 10 6,346 — — — — — — — — — — — — — — 0.46** 1.08 0.84 0.67** 0.77 0.99 0.99* 0.91 1.13 1.08* — 200.17 10 6,431 0.45** 1.04 0.85 0.64** 0.78 0.93 0.98** 0.93 1.11 1.05 1.61** 226.58 11 6,381 0.31** 0.02 0.04 0.01 0.40 0.03 0.01** 0.01 0.07** 0.04** — 34.03 10, 1149 5,715 0.32** 0.03 0.03 0.02 0.38 0.00 0.00 0.02 0.06* 0.03** 0.23** 40.59 11, 1147 5,676

Note. Each column is a separate regression model that predicts civic or occupational outcomes. All coefficients are odds ratios from logistic models, except for coefficients in income models, which are unstandardized betas from linear models. M1 Model 1; M2 Model 2. Dash indicates term or model was not tested. M2 was tested only when previous analyses (see Table 4) revealed direct relations between intensity and postsecondary attendance/completion and M1 analyses revealed significant direct relations between intensity and outcomes. b Multivariate Wald chi-square values are reported for logistic models, and multivariate F statistics are reported for linear income models (all are significant at p .001). * p .01. ** p .001.
a

however, a one-unit increase in community-sponsored participation intensity was directly associated with a 34% increase in the odds of voting (z 3.67, p .001; see Model 1 in Table 6). Postsecondary completion did not mediate this relation.

Direct and Indirect Relations Between Participation Intensity and Occupational Outcomes
Full-time employment. More intensive participation in schoolsponsored activities was not associated with employment among 1-year participants. Among 2-year participants, however, Model 1 results (see Table 5) reveal that a one-unit increase in schoolsponsored participation intensity was associated with an 8% increase in the odds of full-time employment (z 2.80, p .01).

Postsecondary completion completely mediated this relation (see Model 2, Table 5; TIE .016, z 5.31, p .001). The intensity of youths’ participation in community-sponsored activities was not associated with full-time employment. Income. Although more intensive participation in schoolsponsored activities was not associated with income among 1-year participants, Model 1 results (see Table 5) reveal a positive association between school-sponsored participation intensity and income among 2-year participants ( .04, t 5.50, p .001). Postsecondary completion partially mediated this association (see Model 2, Table 5; TIE .029, z 7.00, p .001). The intensity of youths’ participation in community-sponsored activities was not associated with income.

826

GARDNER, ROTH, AND BROOKS-GUNN

Table 6 Regression Coefficients for Relations Between Community-Sponsored Participation Intensity and Civic and Occupational Outcomes
Two years after high school Volunteering Variable M1a M2 M1 Voting M2 Volunteering M1 M2 M1 Eight years after high school Voting M2 Employment M1 M2 Income M1 M2

Sample: 1-year participants in community-sponsored activities Controls Female Black Hispanic Asian/Pacific Islander Native American Parent HS Test scores Behavior problems Parental monitoring Intensity Postsecondary education Wald 2 or Fb df N 0.99 1.26 1.16 1.18 1.84 1.42* 1.03** 0.88 1.40** 1.46* — 96.01 10 2,387 0.95 1.16 1.05 1.01 1.81 1.27 1.01 0.93 1.36** 1.35* 2.73** 161.55 11 2,376 1.00 0.71 0.60** 0.32** 0.54 1.31* 1.02* 0.96 1.11 1.22 — 103.07 10 2,648 — — — — — — — — — — — — — — 1.39* 1.53* 1.20 0.91 2.52* 1.21 1.02* 0.94 1.19 1.27 — 49.22 10 2,206 — — — — — — — — — — — — — — 0.91 1.30 0.91 0.40** 1.41 1.11 1.01 0.93 1.41** 1.02 — 53.72 10 2,201 — — — — — — — — — — — — — — 0.38** 0.94 0.66* 0.65 0.62 0.87 1.01 0.91 1.09 0.90 — 97.86 10 2,242 — — — — — — — — — — — — — — 0.35** 0.08 0.10 0.02 0.28 0.05 0.01** 0.01 0.08 0.01 — 13.76 10, 878 1,983 — — — — — — — — — — — — — —

Sample: 2-year participants in community-sponsored activities Controls Female Black Hispanic Asian/Pacific Islander Native American Parent high school Test scores Behavior problems Parental monitoring Intensity Postsecondary education Wald 2 or Fb df N 0.98 1.34* 1.12 1.01 1.23 1.29* 1.04** 0.87* 1.29** 2.96** — 401.27 10 5,202 0.97 1.29 1.08 0.98 1.24 1.18 1.03** 0.89 1.25** 2.93** 2.08** 482.05 11 5,184 0.89 0.77 0.64** 0.39** 0.72 1.45** 1.03** 0.83** 1.25** 1.26* — 276.95 10 5,561 0.88 0.76* 0.63** 0.39** 0.72 1.40** 1.02** 0.83** 1.23** 1.25* 1.37** 290.11 11 5,542 1.16 1.61** 1.15 1.15 1.59 1.18 1.02** 1.01 1.32** 2.32** — 222.91 10 4,817 1.15 1.61** 1.17 1.14 1.66 1.13 1.02** 1.03 1.30** 2.25** 1.32** 321.24 11 4,791 1.03 1.36* 0.96 0.56** 0.77 1.06 1.02** 0.86* 1.22* 1.34** — 126.10 10 4,813 1.03 1.36* 0.96 0.56** 0.79 1.05 1.02** 0.87* 1.20* 1.33** 1.11 124.93 10 4,787 0.46** 1.16 0.89 0.70* 0.59 1.06 0.99* 0.88 1.16 0.88 — 143.57 10 4,872 — — — — — — — — — — — — — — 0.30** 0.04 0.05 0.04 0.72 0.06 0.01** 0.00 0.07* 0.00 — 23.17 10; 1,088 4,314 — — — — — — — — — — — — — —

Note. Each column is a separate regression model that predicts civic or occupational outcomes. All coefficients are odds ratios from logistic models, except for coefficients in income models, which are unstandardized betas from linear models. a M1 Model 1; M2 Model 2. Dash indicates term or model was not tested. M2 was tested only when previous analyses (see Table 4) revealed direct relations between intensity and postsecondary attendance/completion and M1 analyses revealed significant direct relations between intensity and civic or occupational outcomes. b Multivariate Wald chi-square values are reported for logistic models, and multivariate F statistics are reported for linear income models (all are significant at p .001). * p .01. ** p .001.

Discussion
Consistent with extant theory and research, our findings demonstrate that participation in organized activities during high school is positively associated with educational, civic, and, to some extent, occupational success in young adulthood. Our findings extend prior research and suggest that adolescents who participate in organized activities intensively and over a period of 2 years (vs. 1) are particularly likely to achieve these successes in young adulthood. Perhaps most impressive, we found that the associations between participation intensity and duration and positive development extend well into young adulthood, as many as 8 years after high school. The relations between temporal dimensions of participation and young adult outcomes are not always straightforward, however. Our findings build on prior studies that

have simply documented relations between participation duration and intensity and young adult outcomes (Marsh & Kleitman, 2002; Swanson, 2002; Zaff et al., 2003), and suggest that we must consider the complex interplay between duration, intensity, and sponsorship, as well as the paths that link participation to young adult outcomes (see results summary in Table 7). We highlight these new findings as we discuss each of our research questions. We first asked whether adolescents who participate in organized activities over 2 years achieve greater young adult success than those who participate for a single year. In keeping with limited extant research (e.g., Mahoney et al., 2003; Zaff et al., 2003), we found that although 1 year of participation in school- or community-sponsored activities during adolescence was associated with some positive young adult outcomes, 2 years of partic-

ORGANIZED ACTIVITIES AND DEVELOPMENTAL SUCCESS

827

Table 7 Summary of Direct and Mediated Relations Between Participation Variables and Young Adult Outcomes
Two years after high school Variable Duration analyses School-sponsored 0 years 2 years Community-sponsored 0 years 2 years Intensity analyses School-sponsored 1-year sample 2-year sample Community-sponsored 1-year sample 2-year sample Postsecondary attendance Volunteering Voting Postsecondary completion Eight years after high school Volunteering Voting Employment Income

NS

NS /P /P /P
q

NS /P NS /P NS NS NS

/P /C NS /P NS /P NS /P

NS

NS NS NS NS NS /C NS NS

NS NS NS NS NS /P NS NS

/P /P

q

NS NS

NS

Note. NS nonsignificant relation; significant direct relation; P partial mediation (direct relation remains significant after adding postsecondary predictor, but indirect path through postsecondary education is significant); C complete mediation (direct relation is no longer significant after adding postsecondary predictor, and indirect path through postsecondary education is significant); superscript q quadratic relation.

ipation was associated with even greater odds of postsecondary attendance and completion and greater odds of civic engagement 2 and 8 years after high school (see Table 7). This was true regardless of sponsorship. Prior analyses of NELS data have identified positive relations between multiple years of participation in organized activities (measured without regard to sponsorship) and educational and civic outcomes 2 years after high school (Zaff et al., 2003), but our results are novel in that they suggest that (a) the relations between participation duration and positive educational and civic outcomes are maintained throughout much of young adulthood (i.e., 8 years after high school) and (b) the relations between participation duration and positive young adult outcomes vary little as a function of sponsorship. Although our analyses do not permit definitive conclusions about causality, one explanation for the positive associations between participation duration and educational and civic outcomes is that youths who participate over 2 years (as opposed to 1 year) have more exposure to the developmental supports and opportunities afforded by organized activities and therefore derive greater benefits from participation. Another explanation is that longer periods of participation facilitate the kind of positive interpersonal relationships that are thought to encourage positive youth development (Eccles & Gootman, 2002; Roth et al., 1998). Because our measure of duration does not address changes in specific activities between 10th and 12th grade (e.g., a switch from school orchestra in 10th grade to school service clubs in 12th grade), the present findings cannot be interpreted as clear support for the second explanation. Both explanations should be investigated in future research. We did not find positive associations between participation duration and occupational success in young adulthood (see Table 7). This was true regardless of sponsorship. This is not completely surprising given results from recent analyses of NELS data suggesting that participation in varsity and elite school sports (e.g., as a team captain), but not intramural or junior varsity sports, is

positively associated with full-time employment and income in young adulthood (Carlson, Scott, Planty, & Thompson, 2005). These findings might be interpreted as evidence that intensity plays an important role in predicting occupational outcomes, as elite/varsity athletes likely invest significantly more time in sports than intramural/junior varsity athletes. Consistent with this hypothesis, our analyses revealed that more intensive participation in school-sponsored activities among 2-year participants was, in fact, associated with greater occupational success in young adulthood. Participation in organized activities over longer periods of time at lower levels of intensity, however, may not be sufficient to facilitate greater occupational success in young adulthood. It is unclear why the same does not appear to be true for educational or civic outcomes (i.e., why duration alone is sufficient to predict greater educational and civic success). One plausible explanation stems from the fact that entry into the workforce often occurs long after exiting high school (i.e., after completion of postsecondary education), and occupational success, unlike civic involvement, requires a daily commitment. Associations between activity participation and later occupational success may therefore occur only after youths exceed a uniquely high participation threshold. In our second research question, we asked whether more intensive participation in school- or community-sponsored activities during high school is associated with more favorable outcomes in young adulthood. Consistent with studies that find positive relations between participation intensity and youth and young adult outcomes (e.g., Darling, 2005; Marsh & Kleitman, 2002), we found many positive associations between participation intensity during high school and educational and civic outcomes in young adulthood. As discussed previously, we also found some positive associations between participation intensity and occupational outcomes in young adulthood (see Table 7). Moreover, analyses largely failed to support the overscheduling hypothesis—the notion that very intensive participation limits opportunities for other enriching experiences and hinders development (Cooper et al.,

828

GARDNER, ROTH, AND BROOKS-GUNN

1999; Mahoney et al., 2006; Marsh & Kleitman, 2002). We detected only two curvilinear associations between participation intensity and young adult outcomes. First, among 2-year participants, the positive relation between more intensive participation in school-sponsored activities and the odds of postsecondary completion reached a plateau between 10 –19 and 20 hr per week. Counter to the overscheduling hypothesis, however, the odds of postsecondary completion did not decline between 10 –19 and 20 hr per week. Second, up to 4 hr per week, more intensive participation in school-sponsored activities among 1-year participants was positively associated with volunteering 2 years after high school, but as participation increased to 5–9 hr per week, the odds of volunteering declined. It is noteworthy, however, that for the vast majority of 1-year participants (the 87% who participated 4 hr per week or less), more intensive participation in school-sponsored activities predicted greater odds of volunteering 2 years after high school. In the context of all other results obtained in the present study, these findings provide little evidence for the adverse consequences of very intensive participation. The positive associations between participation intensity and young adult successes were not unconditional, however, and were often contingent on the length of time that youths participated (see Table 7). Among those who participated for 2 years, more intensive participation in school- and/or community-sponsored activities was associated with more favorable educational, civic, and occupational outcomes both 2 and 8 years after high school. This suggests that more intensive participation in organized activities is associated with gains in young adult outcomes beyond those associated with participation over an extended period of time (i.e., 2 years). On the other hand, among those who participated in organized activities for only 1 year, there were very limited positive associations between participation intensity and young adult outcomes. More intensive participation in school-sponsored activities among 1-year participants was ultimately (i.e., 8 years after high school) associated with only postsecondary completion and voting, and more intensive participation in community-sponsored activities among 1-year participants was not associated with any of the outcomes measured 8 years after high school. Prior studies of NELS data have documented positive relations between participation intensity and postsecondary attendance (e.g., Marsh & Kleitman, 2002; Swanson, 2002), but to our knowledge, this is the first study to explicitly consider the intersection of intensity and duration. The present findings do not explain why more intensive participation should be associated with fewer positive outcomes when participation is limited to 1 year. One contributing factor may be that adolescents who participate in organized activities for 1 year tend to participate less frequently than youths who participate over a period of 2 years (see Table 1). If the links between participation intensity and positive development are at least partly explained by greater exposure to the developmental supports and opportunities provided by organized activities, one plausible explanation is that youths must exceed a minimum intensity threshold (i.e., a minimum amount of exposure to supports and opportunities) before positive young adult outcomes are observed. The failure to find consistently positive associations between intensity and young adult outcomes among those who participated in organized activities for only 1 year might stem from the fact that too few of these individuals exceeded the minimum intensity threshold. Our anal-

yses were not designed to address this question, however, and future research is needed to determine how much time youths must spend in activities before positive outcomes are observed (Riggs & Greenberg, 2004). Differences in the relations between participation intensity and young adult outcomes as a function of sponsorship are also difficult to explain. Ultimately (i.e., 8 years after high school), we found fewer associations between community-sponsored participation intensity and young adult outcomes than between schoolsponsored participation intensity and young adult outcomes. Differences in the average frequency of youths’ participation in school- versus community-sponsored activities may, for reasons discussed in the preceding paragraph, partly explain this discrepancy (i.e., on average, youths may spend less time in communitysponsored activities than in school-sponsored activities). We cannot make this determination on the basis of the present findings, however, because our school- and community-sponsored intensity scales are not equivalent. Future studies using more comparable measures of school- and community-sponsored participation intensity are needed in order to determine whether these discrepant findings are a measurement artifacts, and if not, whether they may be attributed to differences in youths’ participation intensity across sponsorship categories. Sponsorship also had noteworthy implications for the relation between participation intensity and occupational outcomes. We found that among 2-year participants, more intensive participation in school-sponsored, but not community-sponsored, activities was positively associated with full-time employment and income. One plausible explanation, worthy of further research, is that schoolsponsored activities, relative to community-sponsored activities, offer developmental supports and opportunities that are more relevant for later occupational success. It is also possible that specific types of school-sponsored activities are driving these findings. As previously noted, recent analyses of NELS data show that participation in elite school-sponsored athletics is positively associated with full-time employment and income (Carlson et al., 2005). If relations between participation intensity and occupational outcomes are specific to school-sponsored sports, this would explain why we did not find associations between communitysponsored participation intensity and occupational success. To test this hypothesis, future investigations should examine relations between participation intensity and occupational outcomes across different types of activities (varsity sports vs. band/orchestra, academic clubs, etc.). In our final research question, we asked whether participation duration and intensity are indirectly associated with positive young adult civic and occupational outcomes via educational attainment. Except for the fully direct associations between participation and voting 8 years after high school, we found that all significant relations between the duration of participation in organized activities during adolescence and young adult civic and occupational outcomes were at least partly mediated by postsecondary attendance or completion (see Table 7). Similar findings emerged in analyses of participation intensity—particularly with respect to outcomes measured 8 years after high school. With the exception of relations between intensity and voting, all significant relations between participation intensity and civic and occupational outcomes 8 years after high school were at least partly mediated by postsecondary completion. In other words, youths who partici-

ORGANIZED ACTIVITIES AND DEVELOPMENTAL SUCCESS

829

pated in organized activities more frequently and/or over a longer period of time were more likely to attend and graduate from a postsecondary institution and, in turn, were more likely to volunteer and achieve occupational success 8 years after high school. These findings are consistent with the notion that completion of a 4-year postsecondary degree encourages civic awareness and opens the door to career opportunities that may be unavailable to those who do not continue their education beyond high school. To our knowledge, these findings are a novel contribution to the literature on temporal participation measures and civic and occupational success in young adulthood. Although our findings make a relatively compelling case for the value of intensive involvement in organized activities over a period of more than 1 year, it is important to acknowledge several limitations. First, although we have controlled for a number of selection factors, it is never possible to control for all factors that predict both participation and outcomes. Unmeasured selection factors may at least partly account for the apparent relations between temporal participation measures and positive young adult outcomes. Likewise, we cannot rule out the possibility that selection at least partly explains the observed relations between postsecondary education and civic and occupational outcomes. Recent findings suggest that many of the cognitive and socioemotional skills that predict educational attainment also predict occupational outcomes (Heckman, Stixrud, & Urzua, 2006). We therefore caution against attributing a causal role to participation in organized activities or to postsecondary attendance and completion. Moreover, we suggest that future studies include analyses that more explicitly consider the role of selection (e.g., analyses examining interactions between temporal measures of participation and youth characteristics related to selection). Second, our measures of participation are general and do not assess several more specific dimensions of participation that have probable implications for young adult outcomes. Although we found that lengthier periods of involvement in organized activities were associated with better young adult outcomes, our measure of duration did not assess the continuity of youths’ participation in specific activities (e.g., whether youths participated in band in 10th and 12th grades or switched to an academic club in 12th grade). Additionally, our measure of duration did not distinguish between 1-year participants who dropped out of activities after 10th grade versus those who joined activities during 12th grade. Analyses examining these detailed aspects of duration were beyond the scope of the present study, but they are important steps for future research. Conclusions about the value of more intensive participation must also be tempered by recognition of the imprecise method of measuring participation frequency. Participation in some activities fluctuates markedly across seasons (e.g., marching bands practice more intensively in the fall than in the spring) and may not be adequately captured by measures of the frequency of participation in an “average” week. Intensity measures that do not account for these fluctuations (such as those available in NELS) may result in inaccurately estimated intensity coefficients for youth in seasonal activities (McHale, Crouter, & Tucker, 2001). Third, we did not consider the implications of simultaneous participation in more than one kind of activity. Our findings suggest that school- and community-sponsored activities are similarly associated with young adult outcomes, but it is impossible to determine whether this is because school- and community-

sponsored activities are truly associated with similar young adult outcomes or because adolescents who participate in one kind of activity also participate in the other. This is a question we plan to address in future research. Finally, though the NELS data set lacks measures of activity quality, it is important to point out that the positive relations between participation duration and intensity and young adult outcomes likely depend on the quality of the activities that youths participate in. Several studies highlight the developmental risks associated with participation in poor quality activities (e.g., Mahoney & Stattin, 2000; Smith & Smoll, 1997; Vandell, Shumow, & Posner, 2005). Thus, we strongly caution against the conclusion that intensive participation in any organized activity over a period of more than 1 year, regardless of quality, should necessarily predict positive young adult outcomes. Despite these qualifications, the results of the present study clearly suggest that greater involvement in organized activities during high school, whether measured in terms of duration or intensity, is associated with better odds of educational, civic, and, to some extent, occupational success in young adulthood. Our findings further indicate that adolescents who score high on both temporal dimensions of participation (i.e., duration and intensity) may be particularly likely to demonstrate advantageous outcomes as young adults. Finally, our results show that, with few exceptions, the developmental implications of participation duration and intensity are similar across sponsorship categories. Although longer term, intensive participation in organized activities was associated with occupational success only among those who participated in school-sponsored activities, longer term, intensive participation in school- or community-sponsored activities was often associated with better educational and civic (particularly volunteering) outcomes. These findings provide a nuanced view of the associations between participation in organized activities and young adult development and extend prior studies that have measured participation without regard to duration, intensity, or sponsorship. The mechanisms that link participation duration and intensity to positive young adult outcomes remain poorly understood, however. We find that educational success partly explains the link between participation in organized activities and young adult civic and occupational success, but the pathways that link participation duration and intensity to educational success remain less clear. This is a critically important question for future research, and the answer may be the key to developing youth activities that prepare adolescents to meet the challenges of young adulthood.

References
Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psychological research. Journal of Personality and Social Psychology, 51, 1173–1182. Brown, R., & Evans, W. P. (2002). Extracurricular activity and ethnicity: Creating greater school connection among diverse student populations. Urban Education, 37, 41–58. Busseri, M. A., Rose-Krasnor, L., Willoughby, T., & Chalmers, H. (2006). A longitudinal investigation of breadth and intensity of youth activity involvement and successful development. Developmental Psychology, 42, 1313–1326. Carlson, D., Scott, L., Planty, M., & Thompson, J. (2005). What is the status of high school athletes 8 years after their senior year? (National Center for Education Statistics Rep. No. 2005303). Washington, DC: U.S. Department of Education, Institute of Education Sciences.

830

GARDNER, ROTH, AND BROOKS-GUNN Marsh, H. W., & Kleitman, S. (2002). Extracurricular school activities: The good, the bad, and the nonlinear. Harvard Educational Review, 72, 464 –511. McFarland, D. A., & Thomas, R. J. (2006). Bowling young: How youth voluntary associations influence adult political participation. American Sociological Review, 71, 401– 425. McHale, S. M., Crouter, A. C., & Tucker, C. J. (2001). Free-time activities in middle childhood: Links with adjustment in early adolescence. Child Development, 72, 1764 –1778. Planty, M., Regnier, M., & Owings, J. (2003). Volunteer service by young people from high school through early adulthood (NCES 2004 –365). Washington, DC: U.S. Department of Education, National Center for Education Statistics. Riggs, N. R., & Greenberg, M. T. (2004). After-school youth development programs: A developmental– ecological model of current research. Clinical Child and Family Psychology Review, 7, 177–190. Rose-Krasnor, L., Busseri, M. A., Willoughby, T., & Chalmers, H. (2006). Breadth and intensity of youth activity involvement as contexts for positive development. Journal of Youth and Adolescence, 35, 385– 499. Roth, J. L., & Brooks-Gunn, J. (2003a). What is a youth development program? Identifying defining principles. In R. M. Lerner, F. Jacobs, & D. Wertlieb (Eds.), Promoting positive child, adolescent, and family development: A handbook of program and policy innovations (Vol. 2, pp. 197–224). Thousand Oaks, CA: Sage. Roth, J. L., & Brooks-Gunn, J. (2003b). Youth development programs: Risk, prevention, and policy. Journal of Adolescent Health, 32, 170 –182. Roth, J., Brooks-Gunn, J., Murray, L., & Foster, W. (1998). Promoting healthy adolescents: Synthesis of youth development program evaluations. Journal of Research on Adolescence, 8, 423– 459. Roth, J., Malone, L. M., & Brooks-Gunn, J. (2007). Series on time matters in organized activities. New York: National Center for Children and Families, Teachers College, Columbia University. Simpkins, S. C., Little, P. M. D., & Weiss, H. D. (2004). Understanding and measuring attendance in out-of-school time programs. Cambridge, MA: Harvard Family Research Project. Smith, R. E., & Smoll, F. L. (1997). Coaching the coaches: Youth sports as a scientific and applied behavioral setting. Current Directions in Psychological Science, 6, 16 –21. Sobel, M. E. (1982). Asymptotic confidence intervals for indirect effects in structural equation models. Social Methodology, 13, 290 –312. Stata Corporation (2003). Intercooled Stata for Windows (Version 8.0) [Computer software]. College Station, TX: Author. Swanson, C. B. (2002). Spending time or investing time? Involvement in high school curricular and extracurricular activities as strategic action. Rationality and Society, 14, 431– 471. U.S. Department of Education. (2005). Digest of education. Washington, DC: National Center for Educational Statistics. Retrieved June 24, 2006, from http://nces.ed.gov/programs/digest/d05/foreword.asp Vandell, D. L., Shumow, L., & Posner, J. (2005). After-school programs for low-income children: Differences in program quality. In J. L. Mahoney, R. W. Larson, & J. S. Eccles (Eds.), Organized activities as contexts of development: Extracurricular activities, after-school and community programs (pp. 437– 456). Mahwah, NJ: Erlbaum. Wimer, C., Bouffard, S. M., Caronongan, P., Dearing, E., Simpkins, S., Little, P. M. D., et al. (2006). What are kids getting into these days? Demographic differences in youth out-of-school time participation. Cambridge, MA: Harvard Family Research Project. Zaff, J. F., Moore, K. A., Papillo, A. R., & Williams, S. (2003). Implications of extracurricular activity participation during adolescence on positive outcomes. Journal of Adolescent Research, 18, 599 – 630.

Casey, D. M., Ripke, M. N., & Huston, A. C. (2005). Activity participation and the well-being of children and adolescents in the context of welfare reform. In J. L. Mahoney, R. W. Larson, & J. S. Eccles (Eds.), Organized activities as contexts of development: Extracurricular activities, after-school and community programs (pp. 65– 84). Mahwah, NJ: Erlbaum. Cooper, H., Valentine, J. C., Nye, B., & Lindsay, J. J. (1999). Relationships between five after-school activities and academic achievement. Journal of Educational Psychology, 91, 369 –378. Darling, N. (2005). Participation in extracurricular activities and adolescent adjustment: Cross-sectional and longitudinal findings. Journal of Youth and Adolescence, 34, 493–505. Darling, N., Caldwell, L. L., & Smith, R. (2005). Participation in schoolbased extracurricular activities and adolescent adjustment. Journal of Leisure Research, 37, 51–76. Eccles, J. S., & Gootman, J. A. (Eds.). (2002). Community programs to promote youth development. Washington, DC: National Academy Press. Feldman, A., & Matjasko, J. L. (2005). The role of school-based extracurricular activities in adolescent development: A comprehensive review and future directions. Review of Educational Research, 75, 159 –210. Harvard Family Research Project. (2006). Learning from small-scale experimental evaluations of after school programs (Out of School Time Evaluation Snapshot No. 8). Cambridge, MA: Harvard Graduate School of Education. Heckman, J. J., Stixrud, J., & Urzua, S. (2006). The effects of cognitive and noncognitive abilities on labor market outcomes and social behavior. Journal of Labor Economics, 24, 411– 482. Holder, K. (2006). Voting and registration in the election of November 2004 [Electronic version]. U.S. Census Bureau Current Population Reports (March). Retrieved June 24, 2006, from http://www.census.gov/ prod/2006pubs/p20 –556.pdf Kaufman, J., & Gabler, J. (2004). Cultural capital and the extracurricular activities of girls and boys in the college attainment process. Poetics, 32, 145–168. Larson, R. W., Hansen, D. M., & Moneta, G. (2006). Differing profiles of developmental experiences across types of organized youth activities. Developmental Psychology, 42, 849 – 863. Lerner, R. M., Fisher, C. B., & Weinberg, R. A. (2000). Toward a science for and of the people: Promoting civil society through the application of developmental science. Child Development, 71, 11–20. Lerner, R. M., Lerner, J. V., Almerigi, J. B., Theokas, C., Phelps, E., Gestsdottir, S., et al. (2005). Positive youth development, participation in community youth development programs, and community contributions of fifth-grade adolescents: Findings from the first wave of the 4-H study of positive youth development. Journal of Early Adolescence, 25, 17–71. Lopez, M. H., & Kirby, E. (2005). Electoral engagement among minority youth. College Park: University of Maryland, Center for Information and Research on Civic Learning and Engagement. MacKinnon, D. P., & Dwyer, J. H. (1993). Estimating mediated effects in prevention studies. Evaluation Review, 17, 144 –158. MacKinnon, D. P., Warsi, G., & Dwyer, J. H. (1995). A simulation study of mediated effect measures. Multivariate Behavioral Research, 30, 41– 62. Mahoney, J. L., Cairns, B. D., & Farmer, T. W. (2003). Promoting interpersonal competence and educational success through extracurricular activity participation. Journal of Educational Psychology, 95, 409 – 418. Mahoney, J. L., Harris, A. L., & Eccles, J. S. (2006). Organized activity participation, positive youth development, and the over-scheduling hypothesis. Society for Research in Child Development Social Policy Report, 20(4), 1–30. Mahoney, J. L., Larson, R. W., & Eccles, J. S. (2005). Organized activities as contexts of development: Extracurricular activities, after-school and community programs. Mahwah, NJ: Erlbaum. Mahoney, J. L., & Stattin, H. (2000). Leisure activities and adolescent antisocial behavior: The role of structure and social context. Journal of Adolescence, 23, 113–127.

Received January 5, 2007 Revision received October 17, 2007 Accepted November 29, 2007