بزهکاری و سلامت نوجوانان: نتایج چندگانه از یک فرآیند اجتماعی و ساختاری مشابه

The majority of social scientific research today tends to be discipline specific, concentrating upon distinct phenomena of interest and how individual and structural determinants influence these phenomena. This is evident in research specializations as diverse as criminology and mental health. Each has progressed to theoretical positions that attempt to explain how underlying factors contribute to specific outcomes. Within criminology, one of the dominant paradigms used to explain childhood misconduct, juvenile delinquency, adult crime, and risk-taking behavior — the control perspective — hypothesizes a causal relationship between structural and social processes and the increased likelihood of antisocial consequences.

. Introduction The majority of social scientific research today tends to be discipline specific, concentrating upon distinct phenomena of interest and how individual and structural determinants influence these phenomena. This is evident in research specializations as diverse as criminology and mental health. Each has progressed to theoretical positions that attempt to explain how underlying factors contribute to specific outcomes. Within criminology, one of the dominant paradigms used to explain childhood misconduct, juvenile delinquency, adult crime, and risk-taking behavior — the control perspective — hypothesizes a causal relationship between structural and social processes and the increased likelihood of antisocial consequences. These disciplines have been successful at laying the groundwork for understanding the relationship between social and structural processes and the discipline-specific outcome of interest. However, are these phenomena — mental health problems and delinquency — mutually exclusive or might they be multiple outcomes of similar underlying social and structural processes? Recent epidemiological and clinical evidence reveals a tendency among adolescents to manifest a clustering of delinquency, substance abuse, and mental health problems indicating that they may be associated with one another (Kessler et al., 1996 and Steinhausen et al., 1998; cf. Milin, 1996). This implies that these internalizing and externalizing behaviors may be multiple outcomes of similar structural and social processes suggesting a need to move past a discipline-specific focus to investigate whether adolescents who experience social and structural disadvantage are at risk for a multitude of negative outcomes. 1.1. Juvenile delinquency and social control theories Structural dimensions including the socioeconomic standing and composition of the family as well as social dynamics among parents and children play a crucial role in the development of children (Loeber & Stouthamer-Loeber, 1986). In a review of over 30 studies, they find strong support for the relationship between socioeconomic disadvantage, poor parent–child relations, and children's behavior problems and delinquency. The classic Glueck & Glueck, 1950 and Glueck & Glueck, 1968 study identified differences in the affection and hostility demonstrated by parents of delinquent children compared to nondelinquent children. Others argue that this lack of affection by parents is more likely within nontraditional family compositions such as families with step-parents (Burgess, 1980). Further research has also demonstrated a significant link between family composition, socioeconomic disadvantage, and substance use and delinquency Miller, 1997 and Smart et al., 1994. Hirschi (1969), in his original social bond theory, argued that the developmental correlates of conformity — attachment, commitment, involvement, and belief — which are inculcated within the family domain produces law-abiding citizens in society. The failure of cohesion among children within the family institution due to parental unavailability or poor child–parent relations results in a deficit in the bonds between the child and his/her family and society. Children without these bonds were more likely to engage in risky activities and criminality because they lack a vested interest in their community. In subsequent work, Gottfredson and Hirschi (1990), in their General Theory of Crime, further spell out this personal disposition of self-control among children. They argue that effective parental nurturance, child-rearing, and parental discipline, which are related to structural background factors, are essential for the development of a strong internal self-control among children. The influence of the family is instrumental in the prevention of delinquency and criminality among adolescents and adults. The trait of low self-control has four defining features: it is evident among children as early as grammar school (Wilson, 1993) and largely persistent throughout the life course; behaviors resulting from a deficit in self-control include both criminal and noncriminal acts characterized by an impulsive need for immediate gratification of desires coupled with a preference for risky activities; the various behaviors are not causally predictive of one another, but are multiple outcomes of this underlying trait; and, while an accumulation of these behaviors indicate a low level of self-control, there is an inability to predict which specific behaviors will manifest during the life course. While people possessing this trait have an increased likelihood of engaging in criminal behaviors, its engagement is not a necessary condition of low self-control. People possessing this characteristic are also more likely to engage in other noncriminal risky and impulsive behaviors. Sampson and Laub (1993) extend the work of Gottfredson and Hirschi (1990) by examining additional sources of social capital beyond the family that may temper or contribute to the development of delinquency and criminogenic behaviors. While acknowledging the importance of the family as the central institution, especially in the early years of a child's life, they suggests that other institutions can be a resource to compensate for negative parental relations and bonds. Social capital, in this context, are the personal bonds of emotional interdependencies and role reciprocities outside of the family of origin that provide additional social and psychological resources, which individuals can draw upon as they move through life's major points of transition Coleman, 1988 and Laub & Sampson, 1988. Sampson and Laub (1993) argue that both school and peer groups act as important sources of social capital beyond the familial institution. Gottfredson and Hirschi (1990) suggest that these external sources of social capital are just consequences of the personal disposition of low self-control, for example, poor school performance, early school leaving, etc. because parents have already lost control over these children. However, there has been an accumulation of evidence that sources of social capital outside the family also play an important role in delinquent activities and may moderate the relationship between family and delinquency Braithwaite, 1989 and Wade & Brannigan, 1998. While the differences between these two central control perspectives are important theoretically, the guiding theme directing research in this area is the consistent emphasis on the cascading effect of the underlying structural determinants and the mediating social factors on subsequent delinquent and risk-taking behavior.

5. Results Table 2 and Table 3 provide the unadjusted distribution of the covariates on all eight outcomes for the ARF-OSDUS and the ADD-Health Surveys, respectively. In Table 2, respondents from a nontraditional family had significantly higher rates across all outcomes. In addition, those with lower family income were also more likely to be depressed, to engage in property damage and alcohol use, and to rate their health lower. Income did not influence violence, tobacco, cannabis, and hard drug use. Negative family and school attachment were significantly associated with all seven outcomes. The relationship between CES-D and attachment could not be tested in the ARF-OSDUS. Table 2. Unadjusted distribution of all predictors on outcomes for ARF-OSDUS Survey Predictors CES-D (%) Property (%) Violence (%) Tobacco use (%) Alcohol use (%) Cannabis use (%) Hard drug use (%) Self-rated health (%) Year ns ns ns * § * * ns 1993 17.3 31.7 22.7 23.6 58.2 12.9 18.8 7.0 1995 18.8 33.7 24.2 28.3 61.1 23.2 22.3 7.9 Age * * * * * * * * 11–13 13.2 24.6 27.3 9.5 31.3 2.3 13.7 5.3 14–15 19.5 37.2 27.1 25.9 54.7 14.4 21.2 6.4 16–17 20.8 42.3 25.2 37.5 74.6 31.7 28.3 9.5 18+ 18.7 26.8 14.7 31.5 78.4 25.6 19.5 8.8 Gender * * * ns ns * ns * Male 8.9 39.6 33.3 26.2 60.4 21.4 20.8 5.9 Female 26.8 26.4 14.4 25.9 59.1 15.4 20.5 9.0 Family structure § * * * * * * § Traditional 17.0 31.6 22.0 24.0 58.5 16.5 18.8 7.0 Nontraditional 21.3 36.2 28.3 32.7 63.5 24.0 26.6 8.8 Family income * § ns ns * ns ns * Lowest 32.6 39.0 24.1 27.9 65.8 22.5 21.3 14.7 Middle-low 19.7 33.4 23.2 26.8 60.2 18.3 21.7 8.2 Middle-high 14.0 32.6 24.1 26.8 62.3 18.4 19.4 5.5 High 13.3 29.4 23.5 24.1 53.2 18.6 22.2 5.5 Family attachment * * * * * * * Positive – 19.2 12.8 15.5 48.9 10.6 11.9 6.6 Neutral – 33.4 24.2 23.5 60.8 17.4 18.3 9.1 Negative – 46.9 32.4 40.8 74.3 30.6 31.3 14.5 School attachment * * * * * * * Positive – 23.9 13.8 16.7 56.3 12.3 14.3 6.4 Neutral – 32.1 22.6 23.3 57.7 16.6 16.7 9.3 Negative – 43.2 33.8 40.2 66.6 30.2 31.1 14.6 * P<.001. § P<.01. Table options Table 3. Unadjusted distribution of all predictors on outcomes for ADD-Health Study Predictors CES-D (%) Property (%) Violence (%) Tobacco use (%) Alcohol use (%) Cannabis use (%) Hard drug use (%) Self-rated health (%) Age * * * * * * * * <13.5 15.8 22.9 42.9 10.6 17.9 6.5 2.2 6.1 13.5–15.5 19.1 34.9 45.9 21.6 37.5 19.3 3.8 8.1 15.5–17.5 22.8 38.1 40.9 29.9 54.8 31.0 6.2 5.7 >17.5 24.8 28.7 32.4 33.4 59.9 35.4 5.4 7.1 Gender * * * ns ns ns ns ns Male 17.5 39.0 51.2 27.1 46.8 26.9 5.4 5.7 Female 24.9 28.3 31.1 24.9 46.3 24.5 4.3 7.9 Family structure * * * * ns * ns * Traditional 19.3 30.9 24.7 24.5 45.6 22.5 4.6 5.9 Nontraditional 26.4 39.8 34.9 29.7 48.7 33.3 5.6 9.0 Family income * ns * ns * ns ns * Lowest 26.9 36.2 49.5 27.2 42.8 26.5 5.4 9.9 Middle-low 22.2 33.8 44.1 25.9 45.0 27.1 4.9 7.1 Middle-high 18.8 31.8 36.6 25.9 47.8 23.0 4.4 4.8 High 16.8 31.9 31.5 24.9 51.4 25.6 4.6 5.1 Family attachment * * * * * * * * Positive 15.8 15.8 35.3 18.8 36.5 17.2 2.1 5.2 Neutral 20.2 20.2 40.7 24.2 45.9 23.8 4.1 6.7 Negative 29.3 29.3 48.3 36.7 60.0 38.1 9.0 9.0 School attachment * * * * * * * * Positive 16.4 23.8 30.7 16.8 35.2 2.1 2.1 3.6 Neutral 16.6 30.5 39.2 23.1 43.3 3.0 3.0 5.8 Negative 29.5 43.9 50.5 35.8 58.4 8.7 8.7 10.2 * P<.001. Table options In Table 3, similar patterns emerged in the ADD-Health data. Respondents in nontraditional families were significantly more likely to score positively on all outcomes with the exception of alcohol and hard drug use. The relationship between family income and the eight outcome measures was similar to the ARF-OSDUS data except for the two measures of delinquency. For property damage, the relationship was in the predicted direction but was nonsignificant and the relationship between violence and family income was significant. Negative family and school attachment were significantly associated across all outcomes. Importantly, those with negative family and school attachment were almost twice as likely as those with a positive attachment to have a high score on the CES-D. Table 4 and Table 5 present the adjusted effects (odds ratios) of the covariates on all eight outcomes for the ARF-OSDUS and ADD-Health Surveys, respectively. The results were calculated by including all covariates simultaneously in a logistic regression equation for each outcome. In Table 4, respondents from a nontraditional family were significantly more likely to engage in violence, and all four types of substance use. After controlling for other covariates, family composition no longer had any effect on depression, property damage, and self-rated health. Higher family income significantly reduced the likelihood of depression and low self-rated health but increased the use of tobacco, alcohol, hard drug use and cannabis use. The nonsignificant effect of income on violence use was consistent with the unadjusted effects. Negative family and school attachment once again had a significant influence on all seven outcomes tested. Those respondents with neutral family attachment were also more likely to engage in violence as well as use in all substances but not as likely as those with negative attachment. Negative school attachment was also significantly associated with all seven outcomes. Moreover, students with neutral school attachment were also more likely than those with positive attachment to manifest all outcomes except property and hard drug use but to a weaker extent than those with negative school attachment. Table 4. Adjusted odds ratios of all outcomes by predictors for ARF-OSDUS Survey Predictors CES-D (%) Property (%) Violence (%) Tobacco use (%) Alcohol use (%) Cannabis use (%) Hard drug use (%) Self-rated health (%) Year 1993 – – – – – – – – 1995 1.11 1.16 1.06 1.56* 1.38* 2.54* 1.51* 1.08 Age 11–13 – – – – – – – – 14–15 1.48§ 1.46§ 0.90 2.89* 3.77* 6.96* 1.18 1.08 16–17 1.68* 1.55* 0.62* 4.97* 11.69* 18.50* 1.58§ 1.28 18+ 1.40§ 0.76 0.32* 4.95* 19.74* 21.13* 1.39 1.56 Gender Male – – – – – – – – Female 3.74* 0.60* 0.35* 1.13 0.98 0.82* 1.16 1.72* Family structure Traditional – – – – – – – – Nontraditional 1.17 1.21 1.32§ 1.59* 1.75* 1.73* 1.58* 1.13 Family income Lowest – – – – – – – – Middle-low 0.48* 0.89 0.91 1.24 1.11* 0.93 1.17 0.57§ Middle-high 0.34* 0.97 1.13 1.81* 1.71* 1.06 1.21 0.42* High 0.36* 0.88 0.98 2.08* 1.72* 1.86§ 2.00* 0.40* Family attachment Positive – – – – – – – – Neutral – 1.21 1.32§ 1.84* 2.11* 1.86* 1.47§ 1.33 Negative – 3.97* 3.41* 3.27* 3.69* 3.07* 2.61* 1.85* School attachment Positive – – – – – – – – Neutral – 1.18 1.42* 1.72* 1.37§ 2.10* 1.13 1.55§ Negative – 1.62* 2.19* 3.44* 2.17* 3.91* 2.37* 2.63* 3604 3063 3061 3248 3252 3251 2875 3193 * P<.001. § P<.01. Table options Table 5. Adjusted odds ratios of all outcomes by predictors for ADD-Health Study Predictors CES-D (%) Property (%) Violence (%) Tobacco use (%) Alcohol use (%) Cannabis use (%) Hard drug use (%) Self-rated health (%) Age <13.5 – – – – – – – – 13.5–15.5 1.11 1.54* 0.99 2.03* 2.44* 2.97* 1.37 1.20 15.5–17.5 1.34 1.57* 0.72§ 2.95* 4.69* 5.24* 2.02 0.78 >17.5 1.49§ 1.01 0.48* 3.50* 5.93* 6.60* 1.74 1.00 Gender Male – – – – – – – – Female 1.57* 0.58* 0.39* 0.88 0.98 0.86 0.75 1.40§ Family structure Traditional – – – – – – – – Nontraditional 1.18 1.34* 1.26§ 1.20 1.22§ 1.70* 1.01 1.13 Family income Lowest – – – – – – – – Middle-low 0.83 0.97 0.84 1.01 1.23 1.25 0.94 0.73 Middle-high 0.72§ 0.97 0.64* 1.09 1.50* 1.14 0.89 0.51* High 0.62* 1.03 0.53* 1.05 1.79* 1.37§ 0.96 0.56§ Family attachment Positive – – – – – – – – Neutral 1.21 1.65* 1.16 1.12 1.17 1.18 1.55 1.12 Negative 1.68* 2.28* 1.53* 1.75* 1.79* 1.93* 2.94* 1.31 School attachment Positive – – – – – – – – Neutral 0.91 1.16 1.37* 1.32§ 1.25 1.21 1.14 1.57 Negative 1.68* 1.81* 2.07* 2.14* 2.05* 2.30* 2.89* 2.68* 4600 4568 4564 4600 4600 4600 4600 4600 * P<.001. § P<.01. Table options In Table 5, the structure and social process variables also had a significant influence on the eight outcomes in the ADD-Health Survey. Specifically, respondents from nontraditional families were significantly more likely to engage in property and violent behaviors. They were also more likely to consume alcohol and cannabis. However, in the adjusted effects models, the effect of family composition was no longer significant for CES-D, tobacco use, hard drug use, and self-rated health. Family income maintained its significant influence on CES-D, violence, alcohol use, cannabis use, and self-rated health similar to the unadjusted model (Table 3). Those with higher income were less likely to have high CES-D scores and poor health ratings. They were also less likely to participate in violence but were more likely to participate in alcohol and cannabis use. The increase in substance use among higher income respondents was similar to that identified in the ARF-OSDUS. Family and school attachment both had significant effects on all eight outcomes with the exception of family attachment on self-rated health. The effects of family attachment on the outcome variables were somewhat weaker than those found in the ARF-OSDUS with the exception of hard drug use. Those with neutral family attachment were significantly more likely to engage in property offenses compared to those rated positively but less likely than those with negative attachment. With respect to school attachment, respondents with negative attachment versus those with positive attachment were significantly more likely to have a positive response across all eight outcomes. Moreover, those who had a neutral attachment were also more likely to engage in violence and tobacco use but the odds ratios were smaller than those with negative school attachment. Finally, in a comparison of the overall effects between the two samples, the patterns for the structural and social process variables were very similar. Differences between the two samples were among the structural variables and confounders. With the exception of the relationship between self-rated health and family attachment, the deleterious effects of both negative family and school attachment on the various outcomes were consistently significant for both samples. Since all outcomes were similarly predicted by this process, the second stage of the analysis assessed whether the outcomes were mutually exclusive or whether they were associated with one another. Table 6 presents unadjusted odds ratios between all eight outcomes for both surveys. Inspecting the ARF-OSDUS first, with the exception of the association between CES-D and violence, all associations had significant positive odds ratios. There was variation in the relative strength across the different types of outcomes as they tended to cluster within groups. The clustering between self-rated health and CES-D indicated a psychosomatic dimension of vulnerability (OR=4.03). In addition to the clustering within groups, there were significant comorbid effects between clusters with the exception of violence and CES-D. Table 6. Odds ratios of comorbidity between all outcome measures for both the ARF-OSDUS and ADD-Health Surveys CES-Da Property Violence Tobacco Alcohol Cannabis Hard drugs Self-rated health ARF-OSDUS Survey CES-D 1.65* 1.17 1.98* 1.50* 1.36* 2.06* 4.03* Property 6.58* 3.03* 3.00* 3.25* 3.47* 1.42* Violence 2.63* 2.00* 2.78* 3.10* 1.21§ Tobacco 10.27* 16.27* 6.04* 2.14* Alcohol 19.00* 4.71* 1.53* Cannabis 10.04* 2.08* Hard drugs 2.06* Self-rated health – n 3737 6908 6899 7408 7428 7437 6743 7338 ADD-Health survey CES-D 1.80* 1.62* 1.81* 1.82* 1.71* 1.76* 2.01* Property 3.97* 3.12* 3.17* 4.67* 7.06* 1.75* Violence 1.97* 1.54* 2.23* 2.83* 1.91* Tobacco 7.48* 8.60* 8.12* 2.19* Alcohol 9.22* 7.51* 1.38* Cannabis 17.60* 1.79* Hard drugs 2.33* Self-rated health – n 4605 4573 4569 4605 4605 4605 4605 4605 a In the ARF-OSDUS survey, CES-D was asked in only one schedule reducing its sample size to approximately half of the respondents. * P>.001 § P>.01. Table options In the ADD-Health data, a similar pattern emerged. In all cases, the odds ratio between the eight different outcomes were significant and positive, indicating that they are comorbid. There was a similar clustering within the three distinct groups, delinquent behaviors, substance use, and psychosomatic evaluation. Unlike the ARF-OSDUS, the relationship between CES-D and violence was also significant. In an effort to further examine the issue of comorbidity, an ordinal logistic regression analysis was performed for each data set (not shown) regressing the cumulative frequency of all outcomes on the structural and social predictors simultaneously. For both data sets, negative family attachment (ARF-OSDUS: OR=4.15, P<.0001; ADD-Health: OR=2.42, P<.0001) and negative school attachment (ARF-OSDUS: OR=3.29, P<.0001; ADD-Health: OR=2.75, P<.0001) were significantly associated with increased comorbidity. Moreover, family structure, age, and gender were also significant for both data sets. Family income was the only covariate that was not directly associated with comorbidity for either data set.