عم رهایی از بزهکاری: اثر ازدواج بازبینی و توسعه یافته

Abstract Desistance from criminal offending has become the source of a considerable amount of research attention. Much of this literature has examined how environmental factors, such as marriage, employment, and delinquent peers contribute to the desistance process. A relatively unexplored possibility, however, is that desistance from criminal behavior is partially due to genetic factors. To test this possibility, data from the National Longitudinal Study of Adolescent Health (Add Health) were used to examine the effects that five different genetic polymorphisms (DAT1, DRD2, DRD4, 5HTT, and MAOA) have on desistance from delinquent involvement. Three broad findings emerged. First, marriage significantly increased desistance. Second, some of the genetic polymorphisms had significant independent effects on desistance. Third, for males, the genetic polymorphisms interacted with marital status to predict variation in desistance. The findings underscore the importance of using a biosocial perspective to examine factors related to criminal desistance.

. Introduction Involvement in delinquency rises markedly in the beginning of adolescence, peaks around the ages of eighteen and nineteen, and begins to decline sharply thereafter (Hirschi and Gottfredson, 1983). By early adulthood most people have “aged out” of delinquent involvement and adult criminal activity is confined to a relatively small pool of chronic offenders (DeLisi, 2005). Although the age-crime curve is one of the most firmly established empirical regularities within criminology, the reasons that account for desistance remain the source of debate (Collins, 2004, Gottfredson and Hirschi, 1990, Sampson and Laub, 1993, Sampson and Laub, 2005, Laub and Sampson, 2003, Moffitt, 1993 and Warr, 1998). One of the more prominent explanations of desistance, and one that has achieved a considerable amount of empirical support, is Sampson and Laub’s (1993) age-graded theory of informal social control. The theory posits that involvement in conventional social institutions, such as marriage, employment, and military, contributes to de-escalation in offending frequency among habitual offenders, even those with a seemingly high criminal propensity. One of the main findings to emerge out of their analysis—and one that paralleled the findings garnered in other studies—was that criminal behavior was relatively stable over long periods of time. Unlike most other extant theorists, Sampson and Laub also noted that some offenders—even those with lengthy criminal records—eventually desisted from engaging in unlawful acts. The question thus became: What factors account for criminal desistance? To answer this question, Sampson and Laub extended the logic of Hirschi’s (1969) social bonding theory into adulthood. Through extensive face-to-face interviews with participants of the Glueck sample and through quantitative analysis of the data, Sampson and Laub discovered that desistance from offending was related to three adult social bonds: employment, marriage, and military service. Offenders who had married, who had gained lawful employment, or who had a history of military service were much more likely to desist from crime than those offenders who lacked these bonds. The relationship between adult social bonds and desistance from crime was relatively straightforward: once an individual begins to accumulate social capital, such as being married or obtaining a steady job, they have a stake in conformity. Any type of criminal action jeopardizes their social standing in conventional society. An arrest, for example, may cause a spouse to file for a marital dissolution. On the other hand, individuals who fail to develop adult social bonds will have much less to lose by engaging in crime and therefore they will be at-risk for persisting with their antisocial behavior throughout adulthood. Although Sampson and Laub (1993) identified three different types of adult social bonds, research analyzing the predictors of criminal desistance has centered primarily on marriage. As a result, we follow the lead of other scholars and examine the effect that marriage has on desistance from delinquent involvement. We also explore the potential reasons why some married people desist from offending, whereas other married people persist with their antisocial conduct. Specifically, we examine whether five genetic polymorphisms (DAT1, DRD2, DRD4, 5HTT, and MAOA) condition the effect that marriage has on criminal desistance using data from the National Longitudinal Study of Adolescent Health (Add Health).

. Results We begin our analysis by predicting involvement in delinquency at wave 1 (0 = non-delinquent, 1 = delinquent) for the full sample and separately by gender. As shown in the first column of Table 1, the serotonin transporter gene (5HTT) has a significant negative effect on delinquent involvement, whereas the other three genetic polymorphisms fail to reach statistical significance. In addition, and consistent with predictions, family risk and low self-control have significant positive effects on the wave 1 delinquency scale. The middle column of Table 1 contains the result for the male sample. Similar to the findings generated using the full sample, 5HTT maintains a significant inverse association with delinquency. The low self-control scale maintains a positive relationship with the delinquency scale. The last model in Table 1 depicts the results for females. In this equation, 5HTT has a negative effect on delinquency, while family risk, low self-control, and race all have positive effects on delinquent involvement. Table 1. Logistic regression models predicting delinquent involvement at wave 1 (N = 1994) Full sample Male sample Female sample b SE b SE b SE Genetic polymorphisms Dopamine transporter gene .13 .25 .00 .37 .21 .36 Dopamine D2 receptor gene .10 .12 .15 .18 .04 .17 Dopamine D4 receptor gene .08 .12 .01 .18 .03 .17 Serotonin transporter gene −.33b .13 −.32a .19 −.32a .18 Monoamine oxidase A gene .22 .18 −.01 .17 Control variables Family risk .19b .07 −.01 .10 .28b .10 Criminal father −.08 .17 .07 .25 −.23 .23 Low self-control .26b .02 .21b .03 .32b .04 Age −.01 .04 .01 .05 −.03 .05 Race .16 .13 −.20 .19 .45b .19 Gender .03 .12 Cox and Snell R2 .09 .06 .13 a Significant at the .10 level, two-tailed. b Significant at the .05 level, two-tailed. Table options Next, we examine the genetic and environmental effects on desistance. As Table 2 reveals, the dopamine D4 receptor gene (DRD4) has a direct and independent effect on desistance. The effect is negative, meaning that as the number of risk alleles increases, the odds of desistance increase as well. Remember, however, that no research has ever examined whether genetic polymorphisms are related to desistance, so the negative coefficient for this polymorphism is not unexpected. Consistent with prior research (Laub et al., 1998 and Sampson and Laub, 1993), married persons are more likely to desist from delinquency. In addition, respondents with a criminal father and respondents with low self-control are at risk for persisting with their aberrant behavior into adulthood. Table 2. Logistic regression models predicting desistance for the full sample (N = 1555) Model 1 Model 2 Model 3 Model 4 b SE b SE b SE b SE Genetic polymorphisms Dopamine transporter gene −.40 .29 −.40 .29 −.40 .29 −.40 .29 Dopamine D2 receptor gene .10 .12 .05 .13 .10 .12 .10 .12 Dopamine D4 receptor gene .22a .12 .22a .13 .22a .13 .22a .12 Serotonin transporter gene −.16 .13 −.16 .13 −.16 .13 −.17 .13 Life-course transition Marriage 1.26b .22 1.03b .27 1.25b .27 1.11b .40 Control variables Family risk −.01 .06 −.01 .13 −.01 .06 −.01 .06 Criminal father −.28a .16 −.27a .16 −.28a .16 −.28a .16 Low self-control −.08b .02 −.08b .02 −.08b .02 −.08b .02 Age .16b .04 .16b .04 .16 .04 .16 .04 Race −.14 .13 −.14 .13 −.14 .13 −.14 .13 G × E interactions DRD2 × marriage .59 .45 DRD4 × marriage .04 .44 5HTT × marriage .21 .47 Cox and Snell R2 .10 .10 .10 .10 a Significant at the .10 level, two-tailed. b Significant at the .05 level, two-tailed. Table options Models 2, 3, and 4 in Table 2 introduce the gene × environment interactions. For all of these equations, DRD4 and marriage have significant independent effects on desistance, but none of the G × E interaction terms are significant for the full sample. In this case, the effect that marriage has on desistance is not conditioned by the genetic polymorphisms. Table 3 presents the results of the logistic regression models predicting desistance for males. Model 1 includes all five of the genetic polymorphisms, marital status, and the control variables as covariates. For this model, the dopamine transporter gene (DAT1) has a statistically significant negative effect on desistance, while the dopamine D2 receptor gene (DRD2), DRD4, and monoamine oxidase A (MAOA) have statistically significant positive effects on desistance. Similar to the models estimated for the full sample, being married increased the odds of desisting. Table 3. Logistic regression models predicting desistance for males (N = 745) Model 1 Model 2 Model 3 Model 4 Model 5 b SE b SE b SE b SE b SE Genetic polymorphisms Dopamine transporter gene −.75b .38 −.74a .38 −.77b .38 −.75b .38 −.75b .38 Dopamine D2 receptor gene .27a .16 .18 .17 .26 .16 .27a .16 .27 .16 Dopamine D4 receptor gene .32a .17 .30a .17 .24 .17 .32a .17 .32a .17 Serotonin transporter gene .07 .17 .07 .17 .07 .17 .08 .17 .07 .17 Monoamine oxidase A .28a .16 .28a .16 .28a .16 .28a .16 .18 .17 Life-course transition Marriage 1.58b .32 1.09b .37 1.23b .34 1.74b .64 1.16b .35 Control variables Family risk −.04 .09 −.05 .10 −.04 .09 −.04 .09 −.05 .10 Criminal father −.15 .22 −.14 .22 −.15 .22 −.15 .22 −.16 .22 Low self-control −.07b .03 −.07b .03 −.07b .03 −.07b .03 −.08b .02 Age .13b .05 .13b .05 .13b .05 .12b .05 .13b .05 Race .04 .18 .03 .18 .04 .18 .04 .18 .05 .18 G × E interactions DRD2 × marriage 1.60a .82 DRD4 × marriage 1.81a 1.08 5HTT × marriage −.21 .73 MAOA × marriage 2.03a 1.08 Cox and Snell R2 .08 .09 .09 .08 .09 a Significant at the .10 level, two-tailed. b Significant at the .05 level, two-tailed. Table options Models 2, 3, 4, and 5, introduce the gene × environment interaction terms. In model 2, DRD2 interacts with marriage to predict desistance. Substantively, this means that respondents who possessed the A1 allele and who were married had a greater likelihood of desisting compared to respondents with just the A1 allele or with respondents who were only married. The inclusion of this G × E reduces the main effect of DRD2 to statistical insignificance; however, marriage continues to exert an independent effect on desistance. Model 3 shows that the interaction between DRD4 and marriage is statistically significant and that the independent effect of DRD4 falls from statistical significance. Marital status remains statistically significant. The interaction between DRD4 and marriage can be interpreted to mean that those respondents who possessed the 7R allele and who were married had at greater odds of desisting in comparison with respondents who were only married or who only possessed the 7R allele. As revealed in Model 4, the interaction between 5HTT and marriage is not significant. Finally, in Model 5, the interaction between MAOA and marriage is statistically significant and entering this G × E term into the equation reduces the main effect of MAOA to insignificance. As with all of the other equations, the independent effect of marriage on desistance remains statistically significant. Table 4 presents the findings of the multivariate models for females. Across all five of the models in the table, three findings emerge. First, only one genetic polymorphism—5HTT—has a statistically significant and negative independent effect on desistance. Second, marriage increases the odds of desistance in models 1, 2, and 3; however, when interaction terms are introduced for 5HTT × marriage and for MAOA × marriage (models 4 and 5), the effect that marriage has on desistance falls from significance. Third, none of the G × E interaction terms are significant predictors of desistance. Table 4. Logistic regression models predicting desistance for females (N = 810) Model 1 Model 2 Model 3 Model 4 Model 5 B SE b SE b SE b SE b SE Genetic polymorphisms Dopamine transporter gene .18 .43 .18 .43 .19 .43 .18 .43 .16 .43 Dopamine D2 receptor gene −.13 .19 −.13 .20 −.13 .19 −.13 .19 −.14 .19 Dopamine D4 receptor gene .09 .19 .09 .19 .17 .21 .09 .19 .09 .19 Serotonin transporter gene −.45b .20 −.45b .20 −.45b .20 −.50b .22 −.46b .21 Monoamine oxidase A −.11 .20 −.11 .20 −.11 .20 −.11 .20 −.22 .22 Life-course transition Marriage .89b .30 .91b .41 1.22b .45 .52 .52 .39 .42 Control variables Family risk .02 .09 .02 .09 .01 .09 .02 .09 .02 .09 Criminal father −.40a .24 −.40a .24 −.38 .24 −.39a .24 −.42a .24 Low self-control −.09b .03 −.09b .03 −.08b .03 −.09b .03 −.09b .03 Age .22b .06 .22b .06 .21b .06 .22b .06 .22b .06 Race −.42b .20 −.42b .20 −.41b .20 −.42b .20 −.41b .20 G × E interactions DRD2 × marriage −.05 .59 DRD4 × marriage −.65 .60 5HTT × marriage .53 .62 MAOA × marriage .88 .59 Cox and Snell R2 .06 .06 .06 .06 .06 a Significant at the .10 level, two-tailed. b Significant at the .05 level, two-tailed. Table options Last, and as shown in Table 5, we examine whether the genetic polymorphisms have an effect on propensity to marriage. Once the effects of the control variables are held constant, none of the genetic polymorphisms are associated with marital status. These insignificant findings are observed for the full sample and for the gender-specific models. Table 5. Logistic regression models predicting marital status Full sample Male sample Female sample b SE b SE b SE Genetic polymorphisms Dopamine transporter gene .33 .36 −.05 .47 .75 .55 Dopamine D2 receptor gene .01 .14 .23 .22 −.13 .19 Dopamine D4 receptor gene .05 .14 −.15 .23 .18 .18 Serotonin transporter gene .21 .15 .32 .24 .11 .20 Monoamine oxidase A gene −.12 .23 −.19 .19 Control variables Family risk −.06 .07 −.02 .13 −.09 .09 Criminal father .43a .18 .55a .27 .32 .24 Low self-control .05a .02 .03 .04 .06a .03 Age .41a .05 .44a .07 .40a .06 Race −.54a .16 −.40 .26 −.57a .21 Gender −.62a .14 Cox and Snell R2 .07 .07 .07 a Significant at the .05 level, two-tailed.