StudyRe_PositiveYouthDev_Lee_Vandell — File 25-0857

Original article Out-of-School Time and Adolescent Substance Use Kenneth T. H. Lee, M.A. *, and Deborah Lowe Vandell, Ph.D. School of Education, University of California, Irvine, California Article history: Received April 8, 2015; Accepted July 16, 2015 Keywords: Adolescence; Substance use; Out-of-school time; Adolescent development A B S T R A C T Purpose: High levels of adolescent substance use are linked to lower academic achievement, reduced schooling, and delinquency. We assess four types of out-of-school time (OST) contextsd unsupervised time with peers, sports, organized activities, and paid employmentdin relation to tobacco, alcohol, and marijuana use at the end of high school. Other research has examined these OST contexts in isolation, limiting efforts to disentangle potentially confounded relations. Methods: Longitudinal data from the National Institute of Child Health and Human Development Study of Early Child Care and Youth Development (N ¼ 766) examined associations between different OST contexts during high school and substance use at the end of high school. Results: Unsupervised time with peers increased the odds of tobacco, alcohol, and marijuana use, whereas sports increased the odds of alcohol use and decreased the odds of marijuana use. Paid employment increased the odds of tobacco and alcohol use. Unsupervised time with peers pre- dicted increased amounts of tobacco, alcohol, and marijuana use, whereas sports predicted decreased amounts of tobacco and marijuana use and increased amounts of alcohol use at the end of high school. Conclusions: Although unsupervised time with peers, sports, and paid employment were differ- entially linked to the odds of substance use, only unsupervised time with peers and sports were significantly associated with the amounts of tobacco, alcohol, and marijuana use at the end of high school. These findings underscore the value of considering OST contexts in relation to strategies to promote adolescent health. Reducing unsupervised time with peers and increasing sports participation may have positive impacts on reducing substance use. 2015 Society for Adolescent Health and Medicine. All rights reserved. IMPLICATIONS AND CONTRIBUTION This study examines re- lations between four different out-of-school time contexts and adoles- cent substance use. When considered in the same analytic model, unsuper- vised time with peers and sports were the best pre- dictors for substance use, underscoring the potential importance of out-of- school contexts as sour- ces of risk and protection for substance abuse. For adolescents, drug and alcohol use is related to decreases in motivation and academic achievement, reductions in cognitive processes, and increases in school misbehavior [1]. Furthermore, substance use in adolescence is a strong predictor for subsequent substance abuse, health problems, educational failure, mental health services, and needs for drug and alcohol treatment [2]. The purpose of this article is to examine links between ado- lescents’ out-of-school time (OST) contexts and substance use at the end of high school. Four common OST contexts are considered as follows: (1) unsupervised time with peers; (2) sports; (3) other organized activities such as band, speech, and student government; and (4) paid employment. These OST con- texts constitute much of adolescents’ discretionary time outside the school day [3]. For the most part, the effects of these contexts on adolescent developmental outcomes have been studied in separate research literature [3]. Unsupervised time with peers has been viewed as a prob- lematic setting that promotes youth deviance [4] including substance use [5e7]. Osgood’s extension of Routine Activity Theory [6] posits that unsupervised time with peers places youth at risk for misbehavior and deviant behaviors because of a convergence of three factors, such as the lack of adult * Address correspondence to: Kenneth T. H. Lee, M.A., School of Education, University of California, 3200 Education, Irvine, CA 92697. E-mail address: [email removed] (K.T.H. Lee). www.jahonline.org 1054-139X/ 2015 Society for Adolescent Health and Medicine. All rights reserved. http://dx.doi.org/10.1016/j.jadohealth.2015.07.003 Journal of Adolescent Health 57 (2015) 523e529

supervision, the lack of structure, and the presence of peers who may encourage the risky acts [6]. Consistent with Routine Ac- tivity Theory, prior empirical research has found unsupervised time with peers to be linked to increased drug and alcohol use [5e7]. This research did not, however, take into account other OST contexts, such as organized activities and paid employment. Perhaps, it is not unsupervised time with peers, per se, but the lack of organized activities that is linked to substance use. Organized activities, in contrast, is an OST context that theorists [8,9] have identified as promoting positive youth development. Critical aspects of organized activities such as sports, arts, and community service clubs are opportunities for enrichment and challenge, supportive relationships with adult leaders, positive peer networks, and a chance for choice and voice [3]. Empirical research has found partici- pation in community service clubs and sports to be related to higher graduation rates and less alcohol and marijuana use, although effects of sports participation vary in response to peer cultures in the high school [10]. The positive re- lationships from adults and peers gained in these organized activities may provide protection from the societal pressures of adolescent substance use. Paid employment is a third out-of-school context that has been posited to have both negative and positive implications [11]. Paid employment has been linked to increased substance use for youth with high work intensity [12e16] but at the same time has also been linked to lower rates of substance use when work quality is high [17]. Paid work may expose adolescents to more adult-like situations for which they are unprepared. For example, adolescents may spend time with older coworkers, increasing the chances of engaging in different substances. Because prior research has examined OST contexts in separate studies, it has not been possible to disentangle potentially con- founding relations. It is not clear, for example, if positive effects of organized activities are an artifact of less unsupervised time with peers or vice versa. Another limitation is that much of the prior research linking OST contexts to substance use has used a simple (yes/no) indicator of substance use instead of looking at amount of substance use [18]. High levels of substance use represent greater risk [19], so both are considered in this article. Finally, prior research has typically measured participation in OST contexts at a single point in time rather than cumulative participation over time [4,6,7,10,12,14,20]. We expect OST participation across the high school years to be a more robust predictor. In summary, the present article examines the four different OST activities (unsupervised time with peers, sports, organized activities, and paid employment) measured early and late in the high school career in relation to both the odds and amounts of three different types of substance use (tobacco, alcohol, and marijuana), while controlling family and child factors and for prior substance use. Methods Participants were part of the National Institute of Child Health and Human Development Study of Early Child Care and Youth Development (NICHD SECCYD), a prospective longitudinal study conducted at 10 research sites (Pittsburg, PA; Seattle, WA; Phil- adelphia, PA; Little Rock, AR; Boston, MA; Lawrence, KS; Chapel Hill, NC; Charlottesville, VA; Madison, WI; and Irvine, CA) across the United States. Children (n ¼ 1,352) were studied from birth until the end of high school. At birth, 26% of the mothers in the recruited sample had no more than a high school education at recruitment, 20% had incomes not greater than 200% of the poverty level, and 22% were of color [21]. The present study focuses on four OSTcontexts and substance use. These were measured at age 15 and the end of high school and substance use at age 15 (n ¼ 955) and at the end of high school (n ¼ 766). Because the analyses are based on secondary analysis of deidentified data, it was considered exempt from human subjects’ consideration from the institutional review board. Measures Measures of the OST contexts at 15 years and end of high school are presented first, followed by measures of substance use collected at 15 years and end of high school. Variables used as covariates are then described. Out-of-school time contexts Unsupervised time with peers. At age 15 and at the end of high school, adolescents reported how many weekdays and how many weekend hours they spend at least 30 minutes in the af- ternoon or evening after school with other kids such as friends or neighbors, not including brothers or sisters, without an adult. The scores for weekdays ranged from 0 to 5 weekdays and weekend hours ranged from 0 to 8 hours. At the two ages, the unsupervised time with peers variable was constructed by averaging the standardized value of weekdays and weekend hours at age 15 and end of high school. Values at these two time points were then averaged to create the average intensity of unsupervised time with peers during high school. Higher values indicate more unsupervised time with peers. Sports participation. At age 15 and end of high school, adoles- cents reported the number of days of sports-related activities during a typical week, ranging from 0 to 7 days. Participation at age 15 and at the end of high school was averaged. Other organized activities. Adolescents also reported participa- tion in each of the five other forms of organized activities during the past year as follows: (1) arts (music, dance, drama, or art); (2) academic clubs (Spanish, computer, etc); (3) nonacademic clubs or groups; (4) religious groups; and (5) volunteer or community service work. For these activities, adolescents indicated the number of days of participation during a typical week, ranging 0 to 7 days. Participation was calculated by taking the sum of all nonsports activities participated by subjects, ranging from 0 to a possible 7 days of participation. Participation at age 15 and at the end of high school was averaged to create a composite of struc- tured activity participation throughout adolescence. Paid employment. Adolescents reported whether they had a paying job at age 15 and at the end of high school. If an adoles- cent reported having a paying job during the school year, he or she also indicated the number of hours per week typically worked using five categories (more than 20 hours, 16e20 hours, 11e15 hours, 6e10 hours, and 1e5 hours). If adolescents reported that they were not employed, work hours per week were coded as 0 hours. If participants indicated that they worked during the school year, the work hours per week was coded as the midpoint of each categorical variable. Values at age 15 and at the end of K.T.H. Lee and D.L. Vandell / Journal of Adolescent Health 57 (2015) 523e529 524

high school were averaged to create a variable: intensity of paid employment hours during high school. Substance use Substance use at age 15. At age 15, adolescents were asked whether they ever used marijuana, drank alcohol, or smoked cigarettes. Substance use at the end of high school. Students completed an online survey that asked (1) if they had ever used marijuana; (2) if they had ever drank alcohol; or (3) if they had ever smoked cigarettes at the end of high school. To indicate the amounts of substance use, adolescents re- ported how often they used (1) marijuana; (2) alcohol; and (3) cigarettes in the past 30 days using six categorical responses: more than once a day; once a day; more than once a week; once a week; once every two weeks; or none over the past 30 days. Responses to these categorical variables were converted to a continuous scale ranging from 0 to 45 times over the past 30 days. If adolescents indicated that they never used marijuana, drank alcohol, or smoked cigarettes, number of times a partici- pant used each substance in the past 30 days was coded as zero times, respectively. Covariates. Measures of family and adolescent characteristics were collected and used as covariates. Demographic character- istics reported by mothers at the child’s birth were the study child’s gender, race (white, black, Hispanic, other), maternal age, and maternal education in years. At age 15, mothers reported family income and family size, which were used to calculate income-to-needs ratio. Site-fixed effects were also included to account for time invariant characteristics shared between sub- jects who were recruited at the same research site. Two measures of the quality of the home environment were collected at age 15. The home observation for measurement of the environment [22] assesses physical and emotional aspects of the home environment. Maternal support and sensitivity were assessed during a semistructured mothere child interaction [23]. Adolescents self-reported their impulsivity at age 15 and end of high school, using the impulsivity subscale of the Weinberger Adjustment Inventory [24]. Impulsivity has been found to be an individual characteristic related to substance use [25e27]. Analysis To examine effects of the unique contribution of the four OST contexts on substance use at the end of high school, multivariate logistic and ordinary least squares (OLS) regressions were con- ducted using robust standard errors [28], controlling for the full list of aforementioned covariates. Three forms of substance use (cigarette use, alcohol use, and marijuana use) were tested in separate models. Following recommended analytic practices [29,30], multiple imputation techniques were used to account for missing values in our main predictors and various covariates. For each analysis, 50 data sets were created using chained equations using pre- dictive mean matching to impute continuous variables, multi- nomial logistic regression for categorical variables, and logistic regression for binary variables. Cases with a missing dependent variable were used during the imputation process but deleted before the analysis following the multiple impute and then delete method [30]. Results Descriptive statistics Table 1 provides descriptive statistics for the four OST con- texts averaged between age 15 and end of high school and the three types of substance use, and the covariates used in the an- alyses. The sample was 51% female, and in terms of race/ ethnicity, the sample was 81% white, 8% black, 6% Hispanic, and 5% Asian/other at age 15. Mothers reported, on average, 14.69 years of schooling (standard deviation ¼ 2.42) at childbirth. Prior cigarette, alcohol, and marijuana use at age 15 occurred in 10%, 24%, and 9% of the target sample, respectively. Table 2 provides the correlations among the four OSTcontexts and among the three types of substance use. More unsupervised time with peers during high school was related to higher amounts of paid employment (r ¼ .16) and less involvement in Table 1 Summary statistics (N ¼ 766) Mean or % Standard deviation Minimum Maximum Substance use Age 15 (yes/no) % Yes Cigarette 9.93% Alcohol 24.33% Marijuana 8.59% End of high school (yes/no) % Yes Cigarette 26.68% Alcohol 59.97% Marijuana 25.07% Amount at end of high school (times per month) Cigarette 5.61 13.67 .00 45.00 Alcohol 4.00 7.21 .00 45.00 Marijuana 3.94 10.77 .00 45.00 Predictor variables Out-of-school contexts Unsupervised time with peers .00 1.00 2.11 1.99 Sports .00 1.00 .94 1.73 Organized activities .00 1.00 1.33 4.46 Paid employment .00 1.00 .96 3.50 Covariates Female 51.27% Race/ethnicity White 80.86% Black 8.43% Hispanic 5.62% Asian/other 5.09% Birth Maternal age 29.24 5.34 18.00 46.00 Maternal education 14.69 2.42 7.00 21.00 Age 15 Income-to-needs ratio 5.40 5.39 .08 42.92 Parenting composite (standardized) .00 .81 4.10 1.78 Adolescent impulsivity 2.38 .75 1.00 4.71 N 766 Unsupervised time with peers is the standardized value of the average of stan- dardized days per week and standardized hours per weekend. Paid employment is the standardized value of hours per week. Sports and organized activities is the standardized value of days per week. Parenting composite is the average of age 15 standardized HOME and maternal sensitivity scores. HOME ¼ home observation for measurement of the environment. K.T.H. Lee and D.L. Vandell / Journal of Adolescent Health 57 (2015) 523e529 525

organized activities in high school (r ¼ .10). More time in paid employment was related to less time in sports (r ¼ .13). The three types of substance use at the end of high school were significantly correlated, whether looking at dichotomous (yes/ no) usage or amount of substance use. Correlations ranged from r ¼ .13 to .31. OST contexts and substance use were also correlated. More unsupervised time was related to higher amounts of cigarette (r ¼ .19), alcohol (r ¼ .23), and marijuana use (r ¼ .24). More sports were related to lower amounts of cigarette (r ¼ .17) and marijuana use (r ¼ .08) but more alcohol use (r ¼ .10). More involvement in organized activities was related to lesser amounts of cigarette (r ¼ .14), alcohol (r ¼ .11), and marijuana use (r ¼ .15). More time in paid employment was related to higher amounts of cigarette (r ¼ .10) and alcohol use (r ¼ .08). Predicting the odds of substance use. Table 3 provides the results for three logistic regressions using robust standard errors to test the unique associations between four OST contexts on ever using cigarette, alcohol, or marijuana at the end of high school, while controlling for family and child factors and for prior substance use at age 15. These logistic regressions predict the increase or decrease in the odds of substance use at the end of high school. Robust standard errors were used to deal with issues concerning heterogeneity and lack of normality. The point estimates using these estimators are the same as an ordinary OLS but change the standard errors to deal with minor concerns such as those regarding normality, heteroskedasticity, or some observations with large residuals [28]. As reported in Table 3, more unsupervised time with peers during high school significantly increased the odds of using cig- arettes, using alcohol, and using marijuana at the end of high school. More time in paid employment significantly increased the odds of cigarette and alcohol use but not marijuana use at the end of high school. More time in sports increased the odds of alcohol use but decreased the odds of marijuana use. More time in other organized activities during high school was not related tothe odds of cigarette, alcohol, or marijuana use at the end of high school. Predicting amount of substance use (number of times per month) at the end of high school. Table 4 provides the results of the OLS regressions predicting the effects of OST contexts on (1) amount of tobacco use; (2) amount of alcohol use; and (3) amount of marijuana use at the end of high school, controlling for family and child factors and prior substance use at age 15. More unsupervised time with peers during high school pre- dicted higher amounts of tobacco, alcohol, and marijuana use at the end of high school (effect sizes of .11, .13, and .16, respec- tively). More time in sports activities during high school pre- dicted lower amounts of cigarette and marijuana use at the end of high school but higher amounts of alcohol use (effect sizes of .12, .09, and .09, respectively). Intensity of paid employ- ment and intensity of organized activities were not related to amount of substance use at the end of high school. Follow-up analyses. In a series of follow-up analyses, we exam- ined alternative explanations for the relations found between OST contexts and substance use. First, we tested the links be- tween substance use at age 15 and amount of participation in the four OST contexts at the end of high school OST, controlling for family and child characteristics and amount of OST participation at age 15 (Table 5). This analysis tested the alternative hypothesis that adolescents’ substance use resulted in differential selection of subsequent out-of-school contexts. Cigarette use, alcohol use, and marijuana use at age 15 were not related to intensity of participation in the four OST contexts at the end of high school. Next, we tested associations between participation in the four OST contexts at age 15 and amount of substance use at the end of high school, controlling for family and child factors, and age 15 substance use. This model differs from our primary models in that we focus on the OST contexts at age 15. The amount of Table 2 Correlations between substance use at the end of high school and main predictor variables (N ¼ 747) 1 2 3 4 5 6 7 Amount of substance use Cigarette 1 Alcohol .13*** 1 Marijuana .31*** .27*** 1 OST contexts Unsupervised time with peers .19*** .23*** .24*** 1 Sports .17*** .10** .08* .00 1 Organized activities .14*** .11** .15*** .10** .04 1 Paid employment .10** .08* .02 .16*** .13*** .00 1 OST ¼ out-of-school time. *p < .05; **p < .01; ***p < .001. Table 3 Links between out-of-school time contexts and substance use at the end of high school (N ¼ 766)dodds ratio Substance use (yes/no) at the end of high school Cigarette use Alcohol use Marijuana use Predictor variables (standardized) Unsupervised time with peers 1.39** (.15) 1.47*** (.14) 1.71*** (.18) Sports .89 (.09) 1.19* (.10) .75** (.08) Organized activities .82 (.09) .93 (.08) .82 (.10) Paid employment 1.46*** (.15) 1.28** (.12) 1.10 (.11) Covariates Prior substance use at age 15 (yes/no)a 3.89*** (1.26) 3.38*** (.83) 3.26*** (1.06) Female .59** (.11) .69* (.12) .32*** (.07) Race/ethnicity (reference category: white) Black .57 (.25) .52 (.19) 1.04 (.44) Hispanic .61 (.26) 1.25 (.48) 1.14 (.48) Asian/other 1.17 (.50) .72 (.29) .43 (.27) Birth Maternal age .99 (.02) 1.03 (.02) 1.04* (.02) Maternal education 1.00 (.05) 1.08 (.05) 1.07 (.06) Age 15 Income-to-needs ratio .97 (.02) 1.04 (.02) 1.02 (.02) Parenting composite 1.07 (.15) 1.13 (.15) 1.22 (.19) Adolescent impulsivity 1.71*** (.18) 1.45*** (.14) 1.74*** (.19) Observations 765 766 764 Exponentiated coefficients and robust standard errors are in parentheses. Site- fixed effects included. *p < .05; **p < .01; ***p < .001. a Prior use differed for each analytic model. Prior use in cigarette use model was an indicator if the adolescent ever smoked cigarettes at age 15. Prior use in alcohol use model was an indicator if the adolescent ever drank alcohol at age 15. Prior use in marijuana use model was an indicator if the adolescent ever smoked marijuana at age 15. K.T.H. Lee and D.L. Vandell / Journal of Adolescent Health 57 (2015) 523e529 526

unsupervised time with peers at age 15 predicted amount of cigarette use (b ¼ .08, p < .05) and amount of marijuana use (b ¼ .07, p < .05) at the end of high school. Intensity of organized activities at age 15 predicted amount of alcohol use at the end of high school (b ¼ .09, p < .05). However, the size of the coefficients is smaller than those in Table 4dsuggesting that looking only at age 15 was less informative. A potential concern in our analysis was attrition. It is possible that cases who were lost between age 15 and end of high school are most likely to be substance users. We ran a frequency table between substance use at age 15 and if they were in the sample at the end of high school. We find the attrition between age 15 and end of high school to be random, as the percentage of any substance use at age 15 was effectively consistent if they were in the sample or if they were a function of attrition. A total of 219 cases were lost between age 15 and end of high school, and 66 of these cases or 30% of the sample reported substance use at age 15. This is not statistically different from the 194 (26%) of the 736 who reported substance use at age 15 and are used in the analysis. We were also concerned about the use of a fixed effects model to account for research site instead of a random effects model. The direction and magnitude of our coefficients from a random- effects model matched those of our fixed effects models. How- ever, the random-effects models provide a smaller standard error and a less-conservative estimate. Furthermore, the intraclass correlations of the dependent variables range between .01 and .02 and indicate little between-site variance. Another possible threat to the general conclusion is collin- earity between the predictor variables. We calculated the centered variance inflation factors for the independent variables in our preferred OLS models. The largest variance inflation factor was 1.68 across all the three OLS models, which is below the cutoff value regarded as high or indicative of collinearity [31]. Discussion The results from our analyses both support and extend pre- vious research examining relations between adolescent OST contexts and substance use. Consistent with predictions of Routine Activity Theory [6] and with prior empirical research [4], more unsupervised time with peers was found to increase both the odds and amount of substance use reported by adolescents at the end of high school. These effects were found for all the three forms of substance use examined in the present studydtobacco, alcohol, and marijuana. That these relations were found, even when time in other OST contexts were controlled, suggests that lack of adult supervision, the presence of peers, and minimal structure are important processes influencing substance use in adolescence. Other forms of OST also appeared to be linked to substance use in adolescence, suggesting that these relations were not simply artifacts of a confounding with unsupervised time. Participation in sports appeared to be a protective factor for some forms of substance use; it was associated with reduced odds of Table 5 Links between age 15 OST participation and substance use at the end of high school controlling for age 15 substance use (N ¼ 766) Amount of substance use at the end of high school (standardized) Cigarette use Alcohol use Marijuana use Age 15 predictor variables (standardized) Unsupervised time with peers .08* (.04) .03 (.04) .07* (.04) Sports .06 (.04) .09* (.03) .04 (.03) Organized activities .01 (.04) .04 (.04) .04 (.03) Paid employment .03 (.04) .00 (.04) .05 (.04) Covariates Prior substance use at age 15 (yes/no)a .95*** (.17) .41*** (.11) .78*** (.20) Female .13 (.07) .18** (.07) .38*** (.07) Race/ethnicity (reference category: white) Black .19 (.15) .12 (.18) .02 (.16) Hispanic -.27 (.15) .19 (.21) .16 (.17) Asian/other .03 (.19) .15 (.17) .35** (.11) Birth Maternal age .01 (.01) .01 (.01) .00 (.01) Maternal education .01 (.02) .04** (.02) .02 (.02) Age 15 Income-to-needs ratio .01* (.01) .01 (.01) .00 (.01) Parenting composite .03 (.06) .06 (.06) .02 (.05) Adolescent impulsivity .13** (.04) .20*** (.05) .18*** (.04) Constant .63* (.30) .65* (.28) .04 (.29) Observations 765 766 764 Robust standard errors are in parentheses. Site-fixed effects included. OST ¼ out-of-school time. *p < .05; **p < .01; ***p < .001. a Prior use differed for each analytic model. Prior use in cigarette use model was an indicator if the adolescent ever smoked cigarettes at age 15. Prior use in alcohol use model was an indicator if the adolescent ever drank alcohol at age 15. Prior use in marijuana use model was an indicator if the adolescent ever smoked marijuana at age 15. Table 4 Links between out-of-school time contexts and amount of substance use at the end of high school (N ¼ 766) Amount of substance use at the end of high school (standardized) Cigarette use Alcohol use Marijuana use Predictor variables (standardized) Unsupervised time with peers .11** (.04) .13*** (.04) .16*** (.04) Sports .12*** (.03) .09** (.03) .09** (.03) Organized activities .04 (.03) .05 (.03) .05 (.03) Paid employment .04 (.04) .07 (.04) .02 (.04) Covariates Prior substance use at age 15 (yes/no)a .89*** (.17) .37*** (.10) .71*** (.20) Female .13* (.07) .17** (.06) .38*** (.07) Race/ethnicity (reference category: white) Black .15 (.15) .14 (.18) .00 (.17) Hispanic .29 (.15) .20 (.20) .14 (.17) Asian/other .07 (.19) .10 (.17) .32** (.11) Birth Maternal age .01 (.01) .01 (.01) .00 (.01) Maternal education .01 (.02) .05** (.02) .02 (.02) Age 15 Income-to-needs ratio .01* (.01) .01* (.01) .00 (.01) Parenting composite .02 (.06) .07 (.06) .04 (.05) Adolescent impulsivity .12** (.04) .18*** (.04) .17*** (.04) Constant .50 (.30) .70** (.27) .04 (.30) Observations 765 766 764 Robust standard errors are in parentheses. Site-fixed effects included. *p < .05; **p < .01; ***p < .001. a Prior use differed for each analytic model. Prior use in cigarette use model was an indicator if the adolescent ever smoked cigarettes at age 15. Prior use in alcohol use model was an indicator if the adolescent ever drank alcohol at age 15. Prior use in marijuana use model was an indicator if the adolescent ever smoked marijuana at age 15. K.T.H. Lee and D.L. Vandell / Journal of Adolescent Health 57 (2015) 523e529 527

marijuana use and lower amounts of tobacco and marijuana use at the end of high school. At the same time, consistent with some prior research [10,32], participation in sports was also linked to more alcohol use. Consistent with the prior literature, the results suggest adult supervision and potential peer effects in sports activities are important mechanisms that influence adolescent substance use. Paid employment in high school, in contrast, was associated with increased odds of tobacco and alcohol use. Others have hypothesized that the workplace may expose adolescents to older coworkers who may influence substance use, but the developmental consequences of paid employment depend on the individual [11]. The results show adolescent employment was linked to the odds of cigarette and alcohol use but not to the amount of marijuana use, controlling for other OST contexts. The findings suggest that older coworkers may have introduced ad- olescents to these substances, but other contextual or individual factors are predictive of continued substance use. A surprising finding, or lack of findings, in the present study pertained to organized activities. Prior research has found spe- cific organized activities in high school to serve a protective role with respect to substance use [10,20]. No significant relations were detected in the present study, although these relations “approached” significance, perhaps because of aggregation across activities. Limitations The biggest limitation lies with the inability to make causal claims. Future work regarding adolescent substance use can examine causal links between unsupervised time with peers on substance use through interventions designed to reduce unsu- pervised time with peers in adolescents. Reducing the amount of unsupervised time that adolescents spend with peers may be an effective strategy for preventing adolescent substance use and abuse. Another limitation of this study lies with adolescent partici- pation in various types of organized activities and paid employment. In this study, we were able to separate sports from other types of organized activities but grouped the other types of organized activities because of an insufficient number of ado- lescents participating in these activities to study them separately. We also looked at amount of any type of paid employment rather than looking at amount of different types of paid employment. Future research should look into amount of participation in different types of OST organized activities and paid employment. Acknowledgments The age 15 data collection of the National Institute of Child Health and Human Development (NICHD) Study of Early Child Care and Youth Development was funded by a cooperative agreement (5 U10 HD027040), which calls for scientific collab- oration between the grantees and NICHD staff. The end-of-high school data collection was funded by the Charles Stewart Mott Foundation. The content of this article is solely the responsibility of the named authors and does not necessarily represent the official views of the Eunice Kennedy Shriver, NICHD or the Na- tional Institutes of Health. The authors express their appreciation to the Early Child Care Research Network for its tireless work on this research collaboration over a 20-year period. We are also grateful for helpful comments from Weilin Li, Anamarie Auger, Sabrina Kataoka, and Kim Pierce for this article. References [1] Bryant AL, Schulenberg JE, O’Malley PM, et al. How academic achievement, attitudes, and behaviors relate to the course of substance use during adolescence: A 6-year, multiwave national longitudinal study. J Res Adolescence 2003;13:361e97. [2] Hawkins JD, Catalano RF, Miller JY. Risk and protective factors for alcohol and other drug problems in adolescence and early adulthood: Implications for substance abuse prevention. Psychol Bull 1992;112:64e105. [3] Vandell DL, Larson RW, Mahoney JL, Watts TW. Children’s organized ac- tivities. In: Lerner RM, ed. Handbook of Child Psychology and Develop- mental Science. Ecological Settings and Processes, Vol. 4. New York: John Wiley & Sons, Inc; 2015. [4] Osgood DW, Anderson AL, Shaffer JN. Unstructured leisure in the after- school hours. In: Mahoney JL, Larson RW, Eccles JS, eds. Organized Activ- ities as Contexts of Development: Extracurricular Activities, After-school and Community Programs. New Jersey: Psychology Press; 2005:45e64. [5] Flannery DJ, Williams LL, Vazsonyi AT. Who are they with and what are they doing? Delinquent behavior, substance use, and early adolescents’ after-school time. Am J Orthopsychiatry 1999;69:247e53. [6] Osgood DW, Wilson JK, O’Malley PM, et al. Routine activities and individual deviant behavior. Am Sociol Rev 1996;61:635e55. [7] Richardson JL, Radziszewska B, Dent CW, Flay BR. Relationship between after-school care of adolescents and substance use, risk-taking, depressed mood, and academic-achievement. Pediatrics 1993;92:32e8. [8] Larson RW. Toward a psychology of positive youth development. Am Psychol 2000;55:170e83. [9] Durlak JA, Weissberg RP, Pachan M. A meta-analysis of after-school pro- grams that seek to promote personal and social skills in children and ad- olescents. Am J Community Psycholol 2010;45:294e309. [10] Eccles JS, Barber BL. Student council, volunteering, basketball, or marching band: What kind of extracurricular involvement matters? J Adolesc Res 1999;14:10e43. [11] Staff J, Mont’Alvao A, Mortimer JT. Children at work. In: Lerner RM, ed. Handbook of Child Psychology and Developmental Science. Ecological Settings and Processes, Vol. 4. New York: John Wiley & Sons, Inc; 2015. [12] Monahan KC, Lee JM, Steinberg L. Revisiting the impact of part-time work on adolescent adjustment: Distinguishing between selection and socializ- ation using propensity score matching. Child Development 2011;82: 96e112. [13] Resnick MD, Bearman PS, Blum RW, et al. Protecting adolescents from harm. Findings from the National Longitudinal Study on Adolescent Health. JAMA 1997;278:823e32. [14] Bachman JG, Staff J, O’Malley PM, et al. Twelfth-grade student work in- tensity linked to later educational attainment and substance use: New longitudinal evidence. Developmental Psychol 2011;47:344e63. [15] Paschall MJ, Flewelling RL, Russell T. Why is work intensity associated with heavy alcohol use among adolescents? J Adolesc Health 2004;34:79e87. [16] Mortimer JT, Finch MD, Ryu S, et al. The effects of work intensity on adolescent mental health, achievement, and behavioral adjustment: New evidence from a prospective study. Child Development 1996;67:1243e61. [17] Staff J, Uggen C. The fruits of good work: Early work experiences and adolescent deviance. J Res Crime Delinq 2003;40:263e90. [18] Moore MJ, Werch CE. Sport and physical activity participation and sub- stance use among adolescents. J Adolesc Health 2005;36:486e93. [19] Grant JD, Scherrer JF, Lynskey MT, et al. Adolescent alcohol use is a risk factor for adult alcohol and drug dependence: Evidence from a twin design. Psychol Med 2006;36:109e18. [20] Fredricks JA, Eccles JS. Is extracurricular participation associated with beneficial outcomes? Concurrent and longitudinal relations. Develop- mental Psychol 2006;42:698. [21] NICHD Early Child Care Research Network. Nonmaternal care and family factors in early development: An overview of the NICHD Study of Early Child Care. J Appl Dev Psychol 2001;22:457e92. [22] Caldwell BM, Bradley RH. Home observation for measurement of the environment. Little Rock, AR: University of Arkansas at Little Rock; 1984. [23] Vandell DL, Belsky J, Burchinal M, et al, Network NECCR. Do effects of early child care extend to age 15 years? Results from the NICHD Study of Early Child Care and Youth Development. Child Development 2010;81:737e56. [24] Weinberger DA, Schwartz GE. Distress and restraint as superordinate di- mensions of self-reported adjustmentdA typological perspective. J Pers 1990;58:381e417. [25] Verdejo-Garcia A, Lawrence AJ, Clark L. Impulsivity as a vulnerability marker for substance-use disorders: Review of findings from high-risk research, problem gamblers and genetic association studies. Neurosci Biobehav Rev 2008;32:777e810. K.T.H. Lee and D.L. Vandell / Journal of Adolescent Health 57 (2015) 523e529 528

[26] Dawe S, Loxton NJ. The role of impulsivity in the development of substance use and eating disorders. Neurosci Biobehav Rev 2004;28: 343e51. [27] Vangsness L, Bry BH, LaBouvie EW. Impulsivity, negative expectancies, and marijuana use: A test of the acquired preparedness model. Addict Behav 2005;30:1071e6. [28] White H. A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica 1980;48:817e38. [29] Allison PD. Missing data. Thousand Oaks, CA: Sage; 2001. [30] von Hippel PT. Regression with missing Ys: An improved strategy for analyzing multiply imputed data. Sociol Methodol 2007;37:83e117. [31] Gordon RA. Regression analysis for the social sciences. New York, NY: Taylor & Francis; 2010. [32] Mays D, Thompson NJ. Alcohol-related risk behaviors and sports partici- pation among adolescents: An analysis of 2005 youth risk behavior survey data. J Adolesc Health 2009;44:87e9. K.T.H. Lee and D.L. Vandell / Journal of Adolescent Health 57 (2015) 523e529 529