Reassessing the family-delinquency association: Do family type, family processes, and economic factors make a difference?

Journal of Criminal Justice 35 (2007) 51 – 67

Reassessing the family-delinquency association: Do family type, family processes, and economic factors make a difference? Kristin Y. Mack a,⁎, Michael J. Leiber b , Richard A. Featherstone a , Maria A. Monserud c a

Department of Sociology, Anthropology, and Criminology, University of Northern Iowa, Baker 356, Cedar Falls, IA, 50614-0513, United States b Wilder School of Government and Public Affairs, Virginia Commonwealth University, 519 Scherer Hall, 923 West Franklin Street, Richmond, VA 23284-2028, United States c Sociology Department, Washington State University, Wilson Hall, Pullman, WA 99164, United States

Abstract The present study drew on four competing theoretical perspectives to examine the relationship between family structure and juvenile delinquency. Using data from the Add Health Study, the authors examined nonserious and serious delinquent behavior across youth from different types of households and also considered how the association between family structure and delinquency might be conditioned by family processes and economic factors. Results from negative binomial regression analyses indicated that, in general, type of household was not a significant predictor of nonserious or serious delinquency. Rather, maternal attachment emerged as the most important determinant of delinquent behavior among youth from all family types. The results are discussed within the context of Hirschi's original interpretation of social control theory and future directions for research are suggested. © 2006 Elsevier Ltd. All rights reserved.

Introduction Various versions of social disorganization theories (e.g., Sampson, 1992; Shaw & McKay, 1932), social control theories (e.g., Hirschi, 1969; Nye, 1958), subcultural theories (e.g., Lewis, 1961; Miller, 1958), and life-course perspectives (e.g., Sampson & Laub, 1993; Thornberry, 1987) cite family as a major factor in the explanation of delinquent behavior. Much of the research literature on the subject indicates that the family generally encourages conformity of youth by monitoring behavior, applying consistent discipline, and developing parent–child attachments (e.g., Gottfredson & Hirschi, 1990; Patterson, 1982). Scholars are not in agreement, however, as to whether single parents are as ⁎ Corresponding author. Tel.: +1 319 273 7630; fax: +1 319 273 7104. E-mail address: [email protected] (K.Y. Mack). 0047-2352/$ - see front matter © 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.jcrimjus.2006.11.015

effective as two parents in their ability to do these things (e.g., Demo, 1992; Hetherington & Kelly, 2002; Popenoe, 1996; Rebellon, 2002; Stacey, 1996; Wilkinson, 1974). There is some evidence that single-mothers place fewer maturity demands on their children, engage in less monitoring, and use less effective disciplinary strategies than families with two parents (e.g., Simons, Simons, & Wallace, 2004). Moreover, some research also lends support to the idea that compared to children from intact families, children living in single-parent families participate in more delinquency (e.g., Dornbusch et al., 1985; Juby & Farrington, 2001; Rodgers & Pryor, 1998; Simons & Chao, 1996), have lower educational achievement, and demonstrate poorer overall adjustment as adults (e.g., Acock & Kiecolt, 1989; Amato, 2000; Amato & Keith, 1991; Loh, 1996; McLanahan & Sandefur, 1994; J. Rankin & Kern, 1994; Wells & Rankin, 1991). The relationship between family structure and

52

K.Y. Mack et al. / Journal of Criminal Justice 35 (2007) 51–67

delinquency appears to be particularly significant when official data is used rather than self-report measures and for certain types of conduct problems, such as status offending (e.g., Free, 1991; Hirschi, 1969; Nye, 1958; J. Rankin & Kern, 1994; Rosen & Neilson, 1982; Van Voorhis, Cullen, Mathers, & Chenoweth Garner, 1988; Wells & Rankin, 1991). Despite a number of studies on the topic, several issues regarding the relationship between family structure and delinquent behavior remain unresolved. One underlying concern is the conceptualization of the single-parent household and the role that family processes and social structure may have on tempering its relationship with delinquency. Prior research, for example, has often characterized the single-parent home in simplistic terms by using the traditional methodological practice of collapsing all single-parent households into a single category (i.e., non-intact or broken homes). Treating all single-parent families as theoretically and empirically equivalent is problematic, however, for several reasons. Most important, it ignores differences that might exist between households that experience divorce, death, or no marriage, especially in terms of family bonds and resources that may condition involvement in delinquency relative to one another and to intact households (Cernkovich & Giordano, 1987; Juby & Farrington, 2001; Sprey, 1967; Wells & Rankin, 1991).1 The present research attempted to address this weakness of the existing literature by using data from the National Longitudinal Study of Adolescent Health (Add Health) to examine in greater detail the extent to which family type, family process variables, and economic factors impact participation in nonserious and serious delinquency. More specific, the study considered whether differences existed in the relationship between family types (i.e., intact, divorce, death, or never married) and delinquency, and if this association was mediated by family processes (i.e., attachment, supervision, and control) and/or economic variables (i.e., membership in the underclass and maternal employment status). Theoretical background A number of theoretical frameworks have been used to explain the relationship between family structure and a variety of youth outcomes, including delinquency.2 Following earlier research, the present article refers to four of these paradigms as social control theory, the social control/parental absence model (see Amato & Keith, 1991; Demuth & Brown, 2004), the family crisis model (see Biblarz & Raftery, 1999; Chen & Kaplan,

1997; Felner, Ginter, Boike, & Cowen, 1981; Wells & Rankin, 1986), and the economic strain model (see Amato & Keith, 1991; Biblarz & Raftery, 1999). Each framework provides insight on how family type, family processes, and structural factors contribute to, or increase the opportunity for, juvenile delinquency. Social control theory Within the delinquency literature, Hirschi's (1969) social control theory is one theoretical perspective used to examine the association between family structure and delinquent behavior (Rebellon, 2002; Simons et al., 2004). Simply stated, his version of social bonding theory contends that individuals participate in delinquency because they lack strong affective attachments to parents, stakes in conformity, involvement in conventional activities, and belief in conventional norms. It is important to note that Hirschi (1969) originally argued attachment between parent and child is paramount and that the strength of this relationship is the most important factor in deterring delinquent behavior. In other words, it is the quality, and not the quantity, of bonds that determines delinquency. Accordingly, a child in a single-parent household who has a strong attachment to his or her custodial parent is expected to evidence less delinquency than a child in a two-parent household who has weak attachments to both parents. If this interpretation of Hirschi's original formulation of social control theory is correct, then whatever impact family composition may have on delinquency should be reduced or even eliminated once quality of attachment between parent and child is taken into account. Social control/parental absence model Hirschi later revised his original position that single parenting may be just as effective in producing positive child outcomes as dual-parenting (see Gottfredson & Hirschi, 1990). Instead, he suggested that parental absence does present significant problems that are rarely resolved by increased attachment. This argument was made because high levels of attachment between a parent and child were thought to be difficult to maintain in the absence of the other parent. Other scholars within the social control perspective echoed this position and suggested that many single-parent households are, by nature, a social setting that hampers the establishment of bonds to conformity because half of the parental unit is absent and unable to provide proper control, supervision, and socialization of the child (e.g., Rebellon, 2002; Sampson & Laub, 1993).

K.Y. Mack et al. / Journal of Criminal Justice 35 (2007) 51–67

The latter version of social control theory is somewhat analogous to the parental absence model in the family literature. For the most part, this approach emphasizes family composition and essentially assumes that, all things being equal, two parents are better able to provide guidance and have more opportunities to monitor, supervise, and respond to the behaviors of their children than single parents (e.g., Amato, 1987; McLanahan & Sandefur, 1994). In addition, a sole parent may have less time to offer support, affection, and counseling to their children, not because they are less caring than two parents, but because they find it more difficult to prioritize their children's needs above other life demands (Amato, 1987). This suggests that children in one-parent families may have more motivation and opportunity to engage in unconventional activities, including delinquency, than those from two-parent households (Rebellon, 2002). Accordingly, the parental absence model maintains that the lack of a parent, regardless of the reason, is a significant predictor of delinquent behavior. Family crisis model While the social control and parental absence models stress the importance of family composition, other theoretical frameworks focus more on the processes surrounding family disruptions as the important determinants of child adjustment and well-being. For example, the family crisis model suggests that psychological distress, emotional resentment, and social tension are more commonly associated with parental divorce and separation than parental death (e.g., Biblarz & Raftery, 1999; Chen & Kaplan, 1997; Wadsworth, 1979; Wells & Rankin, 1986). Furthermore, it has been suggested that children experiencing such crisis events often respond by seeking attention, “acting out,” or engaging in diversionary tactics that involve antisocial behaviors (Felner et al., 1981). In other words, youth who experience parental divorce or separation may feel resentment toward their parents that will, in turn, increase the likelihood of weak family attachment and participation in delinquent activities. Parental death, on the other hand, is a traumatic event that is thought to produce anxiety, emotional distress, and depression in response to the personal loss, but does not usually involve the same degree of emotional resentment present in divorced families (Felner, Farber, Ginter, Boike, & Cowen, 1980). Following this argument, children living with a parent who never married and children in two-parent households are expected to have the least amount of delinquency involvement

53

because a family disruption crisis would not be a precipitating factor for them (McLanahan, 1985). Therefore, identifying the family type, or more specifically, the reason why single-parent households are created, is pivotal in the family crisis model for predicting variation in such things as parental attachment and supervision, and especially, delinquency risk among youth. Economic strain model The economic strain model (e.g., Biblarz & Raftery, 1999), in contrast to these other theoretical perspectives, proposes that economic resources, or lack thereof, influence criminal and deviant behaviors. This approach is similar to the base premise of Merton's (1938) strain theory which highlights the importance of the American cultural emphasis on monetary outcomes and the structural themes that emerge from some versions of social control theories (e.g., Sampson & Laub, 1993). Facing social stigmatism and receiving fewer material and educational advantages than the rest of society, youth living in disadvantaged situations are believed to be more likely than others to engage in both expressive and instrumental deviance. Within the family literature, McLanahan and Sandefur (1994, p. 94) contend that “it is not just low income per se but the loss of economic resources associated with family disruption” that produces negative outcomes among children from single-parent families. They suggest that changes in income following a family disruption often place single-parents in a position where they cannot afford to offer extracurricular activities and other opportunities that might give youth less time to engage in delinquency. Loss of economic resources in single-parent families may also be associated with moves to lower-income neighborhoods which may, in turn, also place youth at higher risk for delinquent behavior. From a methodological perspective, the economic strain model argues that the link between family structure, or family disruption, and delinquency is mediated by the socioeconomic status of the household (Amato & Keith, 1991). So while delinquency may be more common in single-parent households (Proctor, 1998; Wells & Rankin, 1991), this relationship is thought to exist because of the strong connection between economic strain and single-parent households and not because of the structure, or composition, of the family itself. As such, the model proposes that when controlling for socioeconomic factors, the relationship between family composition and delinquency should become statistically insignificant.

54

K.Y. Mack et al. / Journal of Criminal Justice 35 (2007) 51–67

Prior research It is clear that these different theoretical approaches offer distinct positions on the association between family structure and delinquent behavior and the specific social processes that may mediate this relationship. Although research was lacking that focused specifically on family type and each of these family and structural processes, previous studies had offered some support for each theoretical position. Demuth and Brown (2004), for example, found that family structure was much less important than family processes in relation to encouraging delinquency among youth. In fact, their results supported Hirschi's (1969) original assertion that “a parent's physical presence is likely to have a smaller impact on delinquent behavior than a parent's psychological and emotional presence” (Demuth & Brown, 2004, p. 78). These findings paralleled those reported by Cernkovich and Giordano (1987) who discovered that what was important in deterring delinquency participation was the quality of the parent–child relationship, not the structure of the family household itself. Other studies had provided evidence consistent with the expectations posited by the social control/parental absence model. J. Rankin and Kern (1994), for instance, found that strong attachment to two parents had a more significant inhibiting effect against delinquency than strong attachment to only one parent. Similar findings had been reported elsewhere (e.g., Briar & Piliavin, 1965; Nye, 1958). There had also been empirical support for the family crisis model. Research dating back to the 1950s demonstrated that certain single-parent households encouraged more delinquency than others (Amato & Keith, 1991; Glueck & Glueck, 1950; Wadsworth, 1979). For example, Glueck and Glueck (1950) found that boys from homes where a parent had died were slightly more involved in delinquency than children from intact families, but evidenced less delinquent behavior than children living in families created by divorce or separation. Although few researchers had directly compared different types of single-parent families since the Glueck's study, those who had done so found a similar tiered pattern of results. Wadsworth (1979), for instance, used official data on family structure from the 1960s and determined that divorce and parental separation had the greatest relationship with delinquency among White males. Since death of a father had a much smaller effect on delinquent behavior, Wadsworth concluded that parental death was nearly the same as having no family disruption at all.

More recently, Juby and Farrington (2001) made comparisons across a number of family groups to determine similarities and differences in delinquent behavior. Among other things, they found that youth from divorced families were more involved in delinquency than families disrupted by death. This was consistent with the findings of Wells and Rankin (1991) who conducted a meta–analysis of the research on the link between family and delinquent behavior. They concluded that the “type of family break does seem to affect juvenile delinquency. The association with delinquency is slightly (albeit not significantly) stronger for families broken by divorce or separation than by death of a parent” (Wells & Rankin, 1991, p. 88; see also Felner et al., 1980; Nye, 1958). Other researchers suggested that the strongest factor in determining delinquency was economic hardship rather than family composition and/or parental attachment (e.g., Guidubaldi, Cleminshaw, Perry, & McLoughlin, 1983; McNulty & Bellair, 2003a). Much of this research had focused on macro-level contexts, such as community disadvantage, and its association with delinquency among youth (e.g., McNulty & Bellair, 2003b; Peeples & Loeber, 1994; B. Rankin & Quane, 2002; Sampson, 1987; Sampson & Wilson, 1995). While the results from studies of the effects of family and/or household socioeconomic status on delinquent behavior had been mixed (e.g., Wright, Caspi, Moffit, Miech, & Silva, 1999), homes characterized by low socioeconomic status and/or low educational background were shown by some researchers to promote attitudes and tolerate behaviors conducive to participation in delinquency (e.g., Anderson, 1990; Heimer, 1997; McNulty & Bellair, 2003b). Accordingly, the effects of socioeconomic status on delinquency were believed to be more important than family composition or family type or family attachment. Given the relationship between single-mother households and impoverishment (Proctor, 1998), however, the effects of economic indicators on delinquency could be more pronounced among youth from this family type when compared to those in other situations. It is also likely that certain types of family disruptions may lead to more or less negative economic consequences for the individuals who experience them. Divorce, in particular, is commonly thought to have negative effects on the financial situation of women (Faust & McKibben, 1999; Holden & Smock, 1991; Umberson & Slaten, 2000) which means that the relationship between economic strain and delinquency may be greater for youth from divorced homes than those who experienced parental death.

K.Y. Mack et al. / Journal of Criminal Justice 35 (2007) 51–67

Based on competing theoretical explanations and accompanying empirical support, the primary objective of the present study was to examine which of these perspectives and interpretations provided a better understanding of the linkage between family structure and delinquency among youth. The literature on this association had, in general, been void of clarity regarding theoretical linkages (Rebellon, 2002) and the importance that type of family may or may not have in determining delinquent behavior. Very little is also known regarding the extent to which parental attachment, adult supervision, and economic resources may condition the relationships between family composition, family disruption, and delinquency and which factors may provide a better explanation for the association. Method Data and sample Data used for this research came from the National Longitudinal Study of Adolescent Health (Add Health). The study was based on a nationally representative sample of eighty high schools and fifty-two “feeder schools” (i.e., middle or junior high schools) that were stratified by region, urbanicity, school type, ethnic mix, and size. Seventh to twelfth grade students were randomly chosen from the class rosters of the selected schools and were interviewed in their own homes. One parental figure, typically the respondent's mother, completed an interview as well. Adolescent and parental combined data were referred to as the in-home sample (N = 20,745). The data used for the current study were from the first wave of the Add Health in-home sample (1995) and were restricted to reports from White, African American, and Hispanic respondents whose parental data were provided by their biological mothers (n = 9,636).3 Description of variables The coding schemes and distributions of variables in the study are presented in Table 1. The family type variable was created based on information provided by the adolescent respondent's mother regarding her experiences with marital or marriage-like relationships. Based on her responses, subjects were placed in one of the following categories: 1. Intact: current marital status of mother was married and biological father lived in household (n = 8,373);

55

2. Divorce: current marital status of mother was divorced, respondent ever lived with biological father, and number of disruptions experienced by respondent was one (n = 701); 3. Death: current marital status of mother was widowed, respondent ever lived with biological father, and number of disruptions experienced by respondent was one (n = 126); 4. Never: current marital status of mother was single, never married and respondent never lived with biological father (n = 436).4 These categories were then combined to create dichotomous variables that allowed for comparisons between youth from different family types (e.g., intact to divorce). The remaining variables in the analyses were based on measures that were consistent with theory and prior research (Hirschi, 1969; Mack & Leiber, 2005; Simons et al., 2004). Family processes were measured by three variables that attempted to capture the quality of relationship with the mother and the degree of maternal supervision and parental control. Maternal attachment was a scale that consisted of responses to six items that assessed the level of closeness to mother (e.g., “Most of the time, your mother is warm and loving toward you,”). Responses to the items were measured on a five-point scale (1 = not at all, 2 = very little, 3 = somewhat, 4 = quite a bit, 5 = very much). Maternal supervision was measured by responses to a question that asked how often the mother was home when the respondent returned from school (1 = never, 2 = almost never, 3 = some of the time, 4 = most of the time, 5 = always). Parental control was operationalized by responses to a single question that asked if the respondent's parents let him or her make their own decision about the time to be home on weekend nights (0 = no, 1 = yes). Variables that attempted to capture the economic resources of the household were included as well. Education level of the mother was measured by a single question and responses ranged from 1 = some high school or less to 5 = professional training beyond a fouryear college degree. Underclass was a scale comprised of no (0) and yes (1) responses to five questions that asked the respondent's mother if she was currently receiving public assistance such as welfare or other supplemental income (e.g., AFDC, food stamps). To address skewness, this variable was dichotomized so that 0 = no and 1 = yes. In addition, mothers were asked if they were currently working outside the home (0 = no, 1 = yes) and this was used as a measure of maternal employment.

56

K.Y. Mack et al. / Journal of Criminal Justice 35 (2007) 51–67

Table 1 Description of variables (N = 9,636) Variable Independent Family structure Family type a

Family processes Marital attachment b Maternal supervision Parental control

Code

N

%

1 = intact 1 = non-intact 1 = divorce 1 = death 1 = never

8,373 1,263 701 126 436

87 13 7 1 5

Low to high Low to high 0 = no 1 = yes

6,553 3,083

68 32

0 = no 1 = yes 0 = no 1 = yes

8,706 930 1,583 8,053

90 10 16 84

6,476 3,160 4,643 4,993

67 33 48 52

Number of siblings Time of disruption

0 = White 1 = non-White 0 = male 1 = female Low to high Low to high Low to high Low to high Low to high 0 = no 1 = yes Low to high Low to high

Dependent Nonserious delinquency b Serious delinquency b

Low to high Low to high

Economic factors Underclass b Maternal employment

Controls Non-White Female Age Maternal education Risk-taking b Peer attachment Peer deviance b Presence of other adult

a b

9,216 420

Mean

S.D.

Range

26.40 3.31

3.59 1.45

6–30 1–5

15.40 2.64 8.87 4.28 2.49

1.85 1.10 2.50 0.78 2.59

11–21 1–5 4–20 1–5 0–9

1.38 1.41

1.16 4.23

0–12 0–17

0.75 0.79

2.59 1.27

0–6 0–8

96 4

Dummy variable, reference category varies by comparison being made. Individual items and reliability coefficients are presented in Appendix A.

A number of factors that are likely to influence delinquent behavior were also included in each model as control variables. A race variable was constructed based on responses to six questions that asked about the racial origin of the adolescent respondent (i.e., White, Black or African American, Hispanic or Latino, American Indian or Native American, Asian or Pacific Islander, or other). This information was then used to create two dummy variables represented by African Americans and Hispanics with Whites as the reference group. Gender was a dummy variable differentiated by 0 = male and 1 = female. Age was measured as a continuous variable and ranged from eleven to twenty-one years old. To control for other common predictors of delinquency, measures of risk-taking, peer attachment, and

peer deviance were also included. Although the data set did not contain questions that directly related to participation in risk-taking behaviors, responses to four questions that centered on risk-aversion were used as a proxy for this variable (e.g., “When you have problems to solve, one of the first things you do is get as many facts as possible”). Response categories ranged from 1 = strongly agree to 5 = strongly disagree and were reverse coded to create the risk-taking scale. Responses to a single question that asked respondents to assess how much they felt that their friends cared about them (1 = not at all to 5 = very much) was used to measure peer attachment. Peer delinquency was a scale that consisted of responses to three items that asked respondents to indicate how many of their three best

K.Y. Mack et al. / Journal of Criminal Justice 35 (2007) 51–67

friends smoked at least one cigarette a day, drank alcohol, and/or used marijuana at least once a month. The final control variables included in the analyses were single-item indicators that measured the presence of another adult in the household (0 = no, 1 = yes), the number of siblings the respondent had (ranging from zero to twelve), and time since disruption (ranging from zero to seventeen years). These variables were included because they had important implications for the amount of supervision and monitoring provided in singlemother households, as well as being typical control variables in studies of single-parent families (Amato, 2000; Amato & Keith, 1991).5 There were two dependent variables included in the present analyses. Six items that consisted of responses to questions regarding property damage, shoplifting, and interpersonal delinquent behavior were used to measure minor or nonserious delinquency. Individual items were coded 0 if the youth did not report engaging in the act over the past twelve months and 1 if he or she did. These responses were then summed to create a nonserious delinquency scale. More serious delinquent behavior was assessed by a separate set of eight items that referred to participation in activities such as burglary, robbery, and physical violence during the past year. Again responses were coded 0 if the youth did not engage in the act and 1 if he or she did and then summed to create a serious delinquency measure. Individual items and reliability coefficients for the additive scales are presented in Appendix A.6 Analysis procedures To address the complex nature of the Add Health sample design, the analyses for the present study were conducted using Stata. This computer program allows researchers to use estimation commands for complex survey data that account for characteristics of the sample design to ensure that parameter estimates and standard errors are unbiased (Chantala & Tabor, 1999). Multivariate analyses using survey-weighted ordinary least squares regression (for the models predicting maternal attachment and supervision), logistic regression (for parental control, underclass, and maternal employment), and negative binomial regression (for nonserious and serious delinquency) were conducted as appropriate. The latter technique was used because both delinquency measures had a large number of zero values and overdispersion. Under these conditions, negative binomial regression models are recommended because they most closely approximate the distribution of these variables (Long & Freese, 2001).7 To simplify the

57

discussion, only results for the effects of family type, family processes, and economic factors on nonserious and serious delinquency were presented. Each regression model, however, controlled for the variables noted above. The findings for these variables are available upon request from the first author. Findings Comparing family type by family processes and economic resources Recall that three of the perspectives discussed, the social control/parental absence, the family crisis, and the economic strain models, indicated that because of either having only one parent present or experiencing a disruption, differences should be evident among family types in parental/child attachment, supervision, or economic resources. According to the social control/ parental absence model, for example, non-intact households should reveal lower levels of attachment relative to intact households because the parent will be in a more difficult position to meet the needs of their child in terms of love, affection, and supervision (e.g., Rebellon, 2002). The family crisis model acknowledges this possibility, but contends that differences will exist across different types of non-intact households. Most notably, lower levels of attachment, supervision, and control should be present in divorced households relative to those that experienced death or that never married (e.g., McLanahan, 1985). Finally, the economic strain model contends that lack of financial resources, more so than weak parent/child attachment, is of greatest importance and may be more evident among divorced families than other types of single-parent households, especially those created by widowhood (Faust & McKibben, 1999; Holden & Smock, 1991; Umberson & Slaten, 2000). In contrast to these positions, is Hirschi's original interpretation of social control theory that posited quality of attachment between a parent and child is central to determining delinquency involvement rather than the composition of the family per se. In the present study, the authors found support for this latter view and minimal credence for the social control/parental absence model, the family crisis perspective, or the economic strain perspective. In Table 2, for example, contrary to expectations derived from each perspective except Hirschi's original formulation of social control, no differences existed by family type in levels of attachment (column 1) and just a few were present for parental control (column 3). It is

58

K.Y. Mack et al. / Journal of Criminal Justice 35 (2007) 51–67

Table 2 Regression coefficients representing models for family processes and economic factors by family type comparisons Family processes

Comparison Non-intact to intact Divorce to intact Death to intact Never to intact Death to divorce Never to divorce Never to death

Economic factors

Maternal attachment

Maternal supervision

Parental control

Underclass

Maternal employment

(1)

(2)

(3)

(4)

(5)

− 0.05 − 0.03 − 0.07 − 0.12 0.03 − 0.13 − 0.38

− 0.20 ⁎⁎ − 0.40 ⁎⁎ 0.26 0.07 0.63 ⁎⁎ 0.38 ⁎⁎ − 0.08

0.25 ⁎ 0.15 0.43 0.36 ⁎ 0.36 0.19 −0.12

2.18 ⁎⁎ 1.86 ⁎⁎ 1.53 ⁎⁎ 2.93 ⁎⁎ − 0.34 1.36 ⁎⁎ 1.14 ⁎⁎

−0.11 0.35 ⁎ −0.52 −0.61 ⁎⁎ −0.77 −0.75 ⁎ 0.39

Note: Read first family type with sign of each dependent variable; all models include control variables. * p < .05. ⁎⁎ p < .01.

apparent, however, that youth from non-intact families evidenced less maternal supervision than those from intact households and when comparisons were made among non-intact households, those who experienced divorce scored the lowest (column 2). Differences were also evident when households were compared on the economic indicators, in particular, the underclass variable. Youth from non-intact families were poorer than those from intact homes (column 4), and among non-intact households, surprisingly the poorest were those from households where marriage never occurred. In short, there were no significant differences by family types in maternal attachment, and to a lesser extent, this held true for parental control as well. Divorced households, however, scored lower on maternal supervision and households where marriage never occurred were the poorest. Next, negative binominal regression was used to assess the extent to which these findings remained significant once the effects of family type, family processes, and economic factors were considered together to predict both nonserious and serious delinquency. Effects of family type, family processes, and economic factors on delinquency To fully capture the effects of family type, family processes, and economic factors on delinquency, the authors conducted the analyses in several steps. First, the direct effects of family type on nonserious delinquency were examined. Next, to assess the potential mediating effects of the family processes and economic variables on the family type/delinquency relationship, separate models were estimated for each (i.e., family processes and economic factors) that also included family type. The last step involved modeling

for the effects of family type, family processes, and economic factors on nonserious delinquency concurrently. The analysis was then repeated for serious delinquency. The standardized survey-weighted negative binomial regression coefficients for equations predicting nonserious delinquency are presented in Table 3. In Part A, column 1, of Table 3, it was found that family type, for the most part, did not have direct effects on nonserious delinquency. Contrary to expectations, of the seven family type comparisons made, only one was statistically significant. Youth from never married households reported lower levels of nonserious delinquency than their divorced counterparts. As anticipated, however, inverse relationships were evident for maternal attachment and maternal supervision with the dependent variable (column 2). Also, a positive effect existed between being poor and involvement in nonserious delinquency (column 3). Controlling for the effects of both family type and the family process measures revealed similar results (Table 3, Part B). Only the never married family relative to the divorced household comparison produced a statistically significant effect on nonserious delinquency once family processes were considered (column 6). Once again, it was an inverse relationship. The negative relationship between maternal attachment and to a lesser extent maternal supervision, and the dependent variable remained significant. Regardless of non-intact family type, youth who indicated lower attachment with their mother reported more delinquency than their counterparts who reported higher levels of attachment (columns 1 through 7). Recall that the results in Table 2 showed that youth from divorced households also scored lower on maternal supervision relative to youth from other non-intact family types. Despite this difference, no

K.Y. Mack et al. / Journal of Criminal Justice 35 (2007) 51–67

statistically significant differences in maternal supervision among youth from single-mother households emerged to explain nonserious delinquency (columns 5 though 7). Still, youth from each type of non-intact households experienced lower levels of maternal supervision than youth from intact homes (columns 1 through 4).

59

Next, the economic measures were substituted for the three family process variables, and for the most part, this produced parallel findings (Table 3, Part C). None of the family type comparisons produced significant effects on nonserious delinquency. Surprisingly, the positive effect of being poor on participation in nonserious delinquency was reversed once family

Table 3 Negative binominal regression coefficients representing models for nonserious delinquency Part A Direct effects Model 1

Model 2

Model 3

Variable

(1)

(2)

(3)

Family type Non-intact to intact Divorced to intact Death to intact Never to intact Death to divorced Never to divorced Never to death

− 0.96 1.00 − 0.92 − 0.91 − 0.84 − 0.73 ⁎ 1.08

Family processes Maternal attachment Maternal supervision Parental control

−0.81⁎⁎ −0.95 ⁎ 1.05

Economic factors Underclass Maternal employment

0.81⁎⁎ 1.15

Part B Mediating effects Family type and family processes Non-intact to intact

Divorced to intact

Death to intact

Never to intact

Death to divorced

Never to divorced

Never to death

Variable

(1)

(2)

(3)

(4)

(5)

(6)

(7)

Family type Maternal attachment Maternal supervision Parental control

−0.95 −0.81⁎⁎ −0.94⁎⁎ 1.05

− 0.99 − 0.81⁎⁎ 0.94⁎⁎ 1.05

− 0.94 − 0.80⁎⁎ − 0.94⁎⁎ 1.05

− 0.88 − 0.80⁎⁎ − 0.95⁎⁎ 1.06

− 0.89 − 0.89 ⁎ − 0.92 − 0.96

− 0.72 ⁎ − 0.83⁎⁎ − 0.95 1.06

1.04 −0.79⁎⁎ 1.03 1.08

Part C Mediating effects Family type and economic factors Non-intact to intact

Divorced to intact

Death to intact

Never to intact

Death to divorced

Never to divorced

Never to death

Variable

(1)

(2)

(3)

(4)

(5)

(6)

(7)

Family type Underclass Maternal employment

1.03 −0.81 ⁎ 1.15 ⁎

1.03 − 0.79⁎⁎ 1.13 ⁎

− 0.98 − 0.74⁎⁎ 1.10

1.03 − 0.78⁎⁎ 1.11

− 0.89 − 0.84 1.77 ⁎

− 0.76 1.03 1.86⁎⁎

1.08 1.01 1.63⁎⁎

(continued on next page)

60

K.Y. Mack et al. / Journal of Criminal Justice 35 (2007) 51–67

Table 3 (continued) Part D Full effects Family type, family processes, and economic factors Non-intact to intact

Divorced to intact

Death to intact

Never to intact

Death to divorced

Never to divorced

Never to death

Variable

(1)

(2)

(3)

(4)

(5)

(6)

(7)

Family type Maternal attachment Maternal supervision Parental control Underclass Maternal employment

− 1.00 − 0.81 ⁎⁎ − 0.96 1.06 − 0.82 ⁎ 1.09

1.02 − 0.82 ⁎ − 0.95 ⁎ 1.05 − 0.80 ⁎⁎ 1.06

−0.99 −0.80 ⁎⁎ −0.95 ⁎ 1.05 −0.74 ⁎⁎ 1.03

1.00 − 0.80 ⁎⁎ − 0.96 1.07 − 0.78 ⁎⁎ 1.05

−0.91 −0.88 ⁎ −0.97 −0.96 −0.87 1.76

− 0.73 ⁎ − 0.82 ⁎⁎ − 1.00 − 1.07 − 1.06 1.86

1.04 − 0.78 ⁎⁎ 1.08 1.07 1.01 1.64 ⁎

Note: Read independent variable with sign of dependent variable; all models include control variables. ⁎ p < .05. ⁎⁎ p < .01.

type was included in the model. Youth that were classified as underclass or poor reported less rather than more involvement in nonserious delinquency and the effect existed when comparisons were made between each of the non-intact households relative to intact households (columns 1 through 4). The underclass measure had no effect on the dependent variable, however, when each non-intact household type was compared relative to one another (columns 5 through 7). Maternal employment status was also a contributing factor to involvement in nonserious delinquency. A positive effect existed with the dependent variable when comparisons were made between youth from non-intact and intact homes (column 1), between youth from divorced and intact families (column 2), and among comparisons between youth from each type of single-parent household (columns 5 through 7). Up to this point in the analyses, no evidence of mediating effects of family type on nonserious delinquency by family processes or economic resources was found. With one exception, family type had no direct effects on nonserious delinquency. The measures for the family processes and economic resources, however, were shown to be statistically significant predictors of the dependent variable. Although not as pronounced or consistent as anticipated, nor always in the expected direction, these relationships varied to some extent when non-intact households were differentiated and compared. As a whole, however, few differences were observed. Next, full models were estimated that included family type comparisons, the family processes measures, and the economic indicators. These results are presented in Part D of Table 3.

The results are similar to those that have been discussed throughout the analyses. The only family type comparison involving a significant effect on the dependent variable was never married relative to divorced. Once again, youth from the former type of home reported less delinquency than youth from the latter household (column 6). While the effects of maternal supervision on nonserious delinquency, for the most part, disappeared, with the exceptions of comparisons between divorced to intact and death to intact (columns 2 and 3), the significant impact of maternal attachment on nonserious delinquency remained significant (columns 1 through 7). This was also true in regard to the inverse effect of being poor on nonserious delinquent involvement (columns 1 through 4). After controlling for the effects of family type, the family process variables, and the underclass measure, the positive effect maternal employment had on nonserious delinquency, with one exception, disappeared (column 7). In short, family structure, and more specifically, type of single-parent household had little bearing on youth participation in nonserious delinquency. Of the four theoretical perspectives guiding the research, the most support was provided for the social control model as originally posited by Hirschi (1969). Youth that had low maternal attachment, regardless of household type, reported higher levels of involvement in nonserious delinquent behavior than their counterparts who had higher levels of attachment. This finding was consistent with results of previous studies, including those reported by Demuth and Brown (2004). The only evidence to support the family crisis model was discovered for comparisons involving youth from never married

K.Y. Mack et al. / Journal of Criminal Justice 35 (2007) 51–67

households relative to divorced households where the former family type indicated less involvement in nonserious delinquency than the latter. This finding remained fairly constant throughout the analyses. Being a member of the underclass predicted participation in nonserious delinquency when comparisons were made between non-intact and intact households and when

61

each of the non-intact homes was individually compared to their intact counterparts. The effect, however, was contrary to expectations as posited by the economic strain model. Youth categorized as underclass reported less nonserious delinquency than those not viewed as poor. Once family type, the family process variables, and the economic measures were controlled, maternal

Table 4 Negative binominal regression coefficients representing models for serious delinquency Part A Direct effects Family type

Family processes

Economic factors

Variable

(1)

(2)

(3)

Family type Non-intact to intact Divorced to intact Death to intact Never to intact Death to divorced Never to divorced Never to death

1.13 ⁎ 1.12 1.20 1.15 1.05 − 0.96 1.28

Family processes Maternal attachment Maternal supervision Parental control

−0.86 ⁎⁎ 1.01 1.06

Economic factors Underclass Maternal employment

− 1.16 ⁎ 1.02

Part B Mediating effects Family type and family processes Non-intact to intact

Divorced to intact

Death to intact

Never to intact

Death to divorced

Never to divorced

Never to death

Variable

(1)

(2)

(3)

(4)

(5)

(6)

(7)

Family type Maternal attachment Maternal supervision Parental control

− 1.12 ⁎ − 0.86 ⁎⁎ 1.01 1.06

1.11 − 0.87 ⁎⁎ 1.01 1.06

1.18 −0.87 ⁎ 1.02 1.03

1.12 − 0.87 ⁎⁎ 1.02 1.03

1.06 −0.83 ⁎⁎ −0.93 1.27 ⁎⁎

− 0.94 − 0.83 ⁎⁎ − 0.93 1.26 ⁎⁎

1.23 −0.79 ⁎⁎ 1.02 −0.98

Part C Mediating effects Family type and economic factors

Variable Family type Underclass Maternal employment

Non-intact to intact

Divorced to intact

Death to intact

Never to Intact

Death to divorced

Never to divorced

Never to death

(1)

(2)

(3)

(4)

(5)

(6)

(7)

1.09 1.12 1.02

1.09 1.14 1.16

1.19 1.06 −0.99

1.09 1.10 1.02

1.04 1.12 0.95

− 0.92 1.28 1.32 ⁎

1.22 1.19 1.08

(continued on next page)

62

K.Y. Mack et al. / Journal of Criminal Justice 35 (2007) 51–67

Table 4 (continued) Part D Full effects Family type, family processes, and economic factors Non-intact to intact

Divorced to intact

Death to intact

Never to intact

Death to divorced

Never to divorced

Never to death

Variable

(1)

(2)

(3)

(4)

(5)

(6)

(7)

Family type Maternal attachment Maternal supervision Parental control Underclass Maternal employment

1.08 − 0.86 ⁎⁎ 1.01 1.06 1.13 1.02

1.08 − 0.87 ⁎⁎ 1.01 1.06 1.15 ⁎ 1.01

1.16 −0.87 ⁎⁎ 1.02 1.03 1.07 0.99

1.07 − 0.87 ⁎⁎ 1.02 1.03 1.11 1.03

1.07 − 0.82 ⁎⁎ − 0.92 1.27 ⁎ − 1.21 0.93

0.89 − 0.83 ⁎⁎ − 0.92 1.28 ⁎⁎ 1.37 ⁎⁎ 1.27 ⁎

1.16 − 0.79 ⁎⁎ 1.00 − 0.99 1.22 1.08

Note: Read independent variable with sign of dependent variable; all models include control variables. ⁎ p < .05. ⁎⁎ p < .01.

employment had almost a nonexistent effect on the dependent variable. Next, the analyses were repeated to examine the effects that family type, as well as family processes and economic factors, had on serious delinquency. These results are presented in Table 4, and for the most part, paralleled those reported for nonserious delinquency. Similar to the findings reported earlier, family type did not have any statistically significant effects on serious delinquency once the family process variables and economic measures were controlled. Initially, an effect was evident involving non-intact households compared to intact households with the dependent variable (Table 4, Part A, column1, Part B, column 1) and remained until economic factors were considered (Table 4, Part C) and the full model was estimated (Table 4, Part D). Once again, the most consistent effect involved maternal attachment and the dependent variable and this effect was present for all family type comparisons (Table 4, Part A, column 2, Part B, columns 1 through 7, Part D, columns 1 through 7). Isolated effects were evident for parental control and the economic measures but the effects were not consistent. Therefore, similar conclusions were arrived at as those for understanding the relationships between family type, family processes, economic factors and nonserious delinquency. Support was provided for the contention of the original interpretation of social control theory that emphasized the role of attachment as a more important determinant of delinquency than family composition. Consequently, no evidence was found to support the other theoretical perspectives commonly used to explain the family structure and delinquency relationship.

Discussion Social control theory, together with parental absence, family crisis, and economic strain models were used as the theoretical background to examine the relationship between family type and delinquency in the present study. The first two frameworks, social control theory and the parental absence model, make the assumption that two parents are better able to provide affection and supervision of their children than single parents (Gottfredson & Hirschi, 1990; McLanahan & Sandefur, 1994). Consequently, children in oneparent families may have more motivation and opportunity to participate in delinquency than those from two-parent households. The family crisis model is inline with the contention that two parents are better than one but suggests that differences in family processes will exist among singlemother households. That is, this model maintains that psychological distress, emotional resentment, and social tension are more commonly associated with parental divorce and separation than parental death (Amato & Keith, 1991). As a result, children whose parents divorce are expected to evidence more delinquency than those who experience the death of a parent. Children living with a parent who never married and those in two-parent households are expected to have the least amount of involvement in delinquency because there is no family disruption crisis to weaken the attachment bond (Wadsworth, 1979). A third perspective, the economic strain model, focuses on how resources, or the lack thereof, may condition the effects that family type has on delinquent behavior among youth (e.g., McLeod, Kruttschnitt, &

K.Y. Mack et al. / Journal of Criminal Justice 35 (2007) 51–67

Dornfeld, 1994). All three of these frameworks view either family processes or economic resources as mediating the relationship between family type and delinquency. Hirschi's (1969) original interpretation of social control theory, on the other hand, offers a slightly different approach. The quality of the attachment between a parent and child is what is thought to be important, irrespective of family composition, reason for the single-parent status, and/or economic strain. Therefore, the primary goal of the present research was to determine which theoretical framework and corresponding interpretation provides the most useful approach to understanding the relationship between type of family and involvement with delinquency. The findings suggest that, for the most part, differences do not exist in the relationship between family types (i.e., intact, divorce, death, or never married) and juvenile delinquency, nor is the association between family structure and delinquency mediated by family processes or economic factors. Rather, the results overwhelmingly support the premise of Hirschi's original interpretation of social control theory. Maternal attachment was consistently found to be a more important predictor of delinquency than family structure, reason why a family was a single-parent household, or lack of economic resources. Furthermore, it was also found that attachment, for the most part, was a stronger explanation of delinquent behavior than other family processes such as maternal supervision and parental control. While these findings were somewhat unexpected, it should not be surprising that the quality of the parent–child bond may play an important role in the development of delinquent behavior. The results add to a growing body of research that appears to suggest family processes, parent–child attachment in particular, have a stronger influence on delinquency among youth than other factors such as family structure and economic circumstances of the household (see also Cernkovich & Giordano, 1987; Demuth & Brown, 2004; J. Rankin, 1983). Although it was surprising that more empirical support for the parental absence model, the family crisis perspective, or the economic strain position was not found, there were several reasons why the results were relatively modest. First, due to the constraints of using secondary data, several of the measures were weak. The economic variables, in particular, might not capture the full range of financial resources thought to be relevant within the economic strain model. For example, the authors were not able to include household income as an economic indicator because there were too many

63

missing cases on this variable. Demuth and Brown (2004), however, included this measure in their study and found family income to have little effect on delinquency. Instead, it may be more useful to include such economic resource variables as social capital or impoverishment and joblessness, since these are more reflective of long-term and persistent concentrated disadvantage (e.g., McNulty & Bellair, 2003b; Sampson, 1987; Sampson & Wilson, 1995). Furthermore, only micro-level resource measures were included, but research had shown that macro-level economic factors (i.e., percent poverty) might be more relevant when addressing the family type/economic resource/delinquency relationship (Sampson, 1992; Sampson, Raudenbush, & Earls, 1997). The analyses was also restricted to single-mother households because there were not enough single-father families in the Add Health data set to make meaningful comparisons by specific family type subgroups. Similarly, stepfather or stepmother comparison groups were not included because it could not be determined what type of event (i.e., divorce, death, or no marriage) preceded the creation of the stepparent family. It would be productive, however, to determine if the findings or non-findings hold when examining families headed by single fathers, if differences emerge when comparing single-father household types to their single-mother counterparts, or if outcomes are different for children in step-father or step-mother families that were created following different events. There was also evidence that gender and race of the child might be important factors to consider as several studies had found that the effects of living in a singleparent home vary across gender (e.g., Acock & Demo, 1994; Dornbusch et al., 1985; Hetherington, Cox, & Cox, 1982; Miller, 1958; Wallerstein & Kelly, 1980) and racial groups (e.g., Austin, 1978; Matsueda & Heimer, 1987; McLeod et al., 1994). While these variables were controlled for in the present study, it will be important for future research to consider whether race and gender may individually, and/or in combination, influence the relationship between family structure and delinquency. Another reason why the authors might not have found more significant effects was related to the decision to restrict the analyses to youth who had experienced only one family disruption (i.e., divorce or death) or whose mothers were single and never married. While this decision was made to control for the confounding effects of multiple family disruptions (Martinson & Wu, 1992), future studies that include youth who have experienced more than one family disruption may yield greater differences across

64

K.Y. Mack et al. / Journal of Criminal Justice 35 (2007) 51–67

household types. In addition, the examination of the relationship between family structure and delinquency was relatively short-term in nature. With the average time since divorce or death at 1.55 years, there was little consideration in the present study of how youth might be affected by family disruptions over time. It is possible that different effects of these experiences may manifest over longer time periods, and therefore, future research should also examine the long-term influence of single-mother household type on delinquent behavior. Despite the limitations of the present research, and the relatively modest findings, the authors were able to conduct a more thorough examination of the relationship between family type and delinquency than what had previously been conducted. This was accomplished by selecting out for different types of singlemother households, including a never married comparison group, and determining if family processes and economic factors condition the relationship between family type and participation in both nonserious and serious delinquency. The analyses were also based on more recently collected data and included family subgroups that were relatively large in size.

In short, the results lend further support for the effects of family processes (i.e., maternal attachment, and to some degree, maternal supervision) as more significant predictors of delinquent behavior among youth than family structure or family type. The task for future research is to continue the investigation of these relationships to provide even greater clarity to the understanding of how the family and economic resources (or lack thereof) contribute to delinquency. Acknowledgements This research used data from Add Health, a program project designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris, and funded by a grant P01HD31921 from the National Institute of Child Health and Human Development, with cooperative funding from seventeen other agencies. Special acknowledgement is due to Ronald Rindfuss and Barbara Entwisle for assistance in the original design. Persons interested in obtaining data files from Add Health should contact Add Health, Carolina Population Center, 123 West Franklin Street, Chapel Hill, North Carolina 275162524 (www.cpc.unc.edu/projects/addhealth/).

Appendix A. Scale items and alpha coefficients Variable

Individual items

Maternal attachment 1. Most of the time, your mother is warm and loving toward you. 2. You are satisfied with the way your mother and you communicate with each other. 3. Overall, you are satisfied with your relationship with your mother. 4. How close do you feel to your mother? 5. How much do you think she cares about you? 6. When you do something wrong that is important, your mother talks about it with you and helps you understand why it is wrong. Underclass 1. Are you receiving public assistance, such as welfare? Last month, did you or any member of your household receive: 2. Supplemental Security Income (SSI)? 3. Aid to Families with Dependent Children (AFDC)? 4. Food stamps? 5. A housing subsidy or public housing? Risk-taking 1. When you have problems to solve, one of the first things you do is get as many facts about the problem as possible. 2. When you are attempting to find a solution to a problem, you usually try to think of as many different ways to approach the problem as possible. 3. When making decisions, you generally use a systematic method for judging and comparing alternatives. 4. After carrying out a solution to a problem, you usually try to analyze what went right and what went wrong. Peer deviance Of your three best friends, how many: 1. Smoke at least one cigarette a day? 2. Drink alcohol at least once a month? 3. Use marijuana at least once a month? Nonserious In the past twelve months how often did you: delinquency 1. Paint graffiti or signs on someone else's property or in a public place? 2. Deliberately damage property that did not belong to you? 3. Take something from a store without paying for it?

Alpha .85

.77

.74

.75

.67

K.Y. Mack et al. / Journal of Criminal Justice 35 (2007) 51–67

65

Appendix (continued ) Variable Nonserious delinquency

Individual Items 4. Drive a care without the owner's permission? 5. Steal something worth less than $50? 6. Steal something worth more than $50?

Alpha

Serious delinquency In the past twelve months how often did you: .75 1. Get into a serious physical fight? 2. Hurt someone badly enough to need bandages or care from a doctor or nurse? 3. Go into a house or building to steal something? 4. Use or threaten to use a weapon to get something from someone? 5. Take part in a fight where a group of your friends was against another group? During the past twelve months, how often did each of the following things happen: 6. You pulled a knife or gun on someone. 7. You shot or stabbed someone. 8. During the past thirty days, on how many days did you carry a weapon—such as a gun, knife, or club—to school?

Notes 1. In addition to the problematic conceptualization of singleparent households, many of the previous studies in this area used data from the 1960s or earlier, employed family groups that were somewhat small in size, failed to include youth from households where the parents were never married, and focused primarily on Whites and males (e.g., Glueck & Glueck, 1950; Juby & Farrington, 2001; Wadsworth, 1979). The present study attempted to address these important gaps within the extant literature. 2. Comprehensive reviews of these and other theoretical perspectives are found in Amato (2000), Amato and Keith (1991), and Juby and Farrington (2001). 3. Following the recommendations provided by Chantala and Tabor (1999), respondents with missing values on the grand sample weight variable were deleted from the analyses. This reduced the sample size from 20,745 to 18,924. The subpopulation command within Stata was then used to specify the population of interest. This command computed estimates for a single subpopulation while taking into consideration the full data set so that the correct survey weight variables were used in the computations (see Chantala & Tabor, 1999 for more detailed information on these procedures). The decision to restrict the sample to White, Hispanic, and African American respondents whose biological mother provided the parental data was made because there were not enough cases in the other race/ethnicity categories or when someone other than the biological mother provided the parental data to make meaningful group comparisons. This was especially evident once family type subgroups were determined. 4. Although ideally the analyses would have included singlefather comparison groups as well (see Demuth & Brown, 2004), the small number of respondents in these subgroups did not allow for differentiation by family type, and therefore, they were not considered in the present research. The analyses were also restricted to respondents who had experienced only one type of family disruption. This decision was made because it was difficult to isolate the effects of a particular disruption from the effects of multiple disruptions, especially if they occurred for different reasons (Martinson & Wu, 1992). Finally, stepfather or stepmother comparison groups were not included because it was not possible to determine whether the remarriage occurred following a divorce or death, and there was concern about the confounding effects that this might have created.

5. All continuous independent and control variables were standardized before conducting the multivariate analyses to allow for comparisons among regression coefficients within models. 6. A check of the variance inflation factors revealed acceptable levels of shared association, suggesting that there were no problems with multicollinearity. The zero–order correlation results are available upon request from the lead author. The survey estimators in Stata were not true maximum likelihood estimators (they are called pseudo maximum likelihood estimators instead), thus, the computation of −2 log likelihood for each model was inappropriate. 7. The irr option was used when estimating the survey-weighted negative binomial regression models. As a result, each unit increase in an independent variable was interpreted as a percent increase or decrease in the dependent variable.

References Acock, A. C., & Demo, D. (1994). Family diversity and well-being. Thousand Oaks, CA: Sage. Acock, A. C., & Kiecolt, K. (1989). Is it family structure or socioeconomic status? Family structure during adolescence and adult adjustment. Social Forces, 68, 553−571. Amato, P. (1987). Family process in one-parent, stepparent, and intact families: The child's point of view. Journal of Marriage and the Family, 49, 327−337. Amato, P. (2000). The consequences of divorce for adults and children. Journal of Marriage and the Family, 62, 1269−1287. Amato, P., & Keith, B. (1991). Parental divorce and the well-being of children: A meta–analysis. Psychological Bulletin, 110, 26−46. Anderson, E. (1990). Streetwise: Race, class, and change in an urban community. Chicago: University of Chicago Press. Austin, R. L. (1978). Race, father–absence and female delinquency. Criminology, 15, 487−504. Biblarz, T., & Raftery, A. (1999). Family structure, educational attainment, and socioeconomic success: Rethinking the “pathology of matriarchy.” American Journal of Sociology, 105, 321−365. Briar, S., & Piliavin, I. (1965). Delinquency, situational inducements, and commitments to conformity. Social Problems, 13, 35−45. Cernkovich, S., & Giordano, P. (1987). Family relationships and delinquency. Criminology, 25, 295−319.

66

K.Y. Mack et al. / Journal of Criminal Justice 35 (2007) 51–67

Chantala, K., & Tabor, J. (1999). Strategies to perform a design-based analysis using the Add Health data. Retrieved July 7, 2003, from http://www.cpc.unc.edu/projects/addhealth/files/weight1.pdf Chen, Z., & Kaplan, H. (1997). The impact of family structure during adolescence on deviance in early adulthood. Deviant Behavior, 18, 365−391. Demo, D. (1992). Parent–child relations: Assessing recent changes. Journal of Marriage and the Family, 54, 104−117. Demuth, S., & Brown, S. (2004). Family structure, family processes, and adolescent delinquency: The significance of parental absence versus parental gender. Journal of Research in Crime and Delinquency, 41, 58−81. Dornbusch, S., Carlsmith, M., Bushwall, E., Ritter, S., Leiderman, H., Hastorf, A., et al. (1985). Single parents, extended households, and the control of adolescents. Child Development, 56, 326−341. Faust, K. A., & McKibben, J. N. (1999). Marital dissolution: Divorce, separation, annulment, and widowhood. In M. B. Sussman, S. K. Steinmetz, & G. W. Peterson (Eds.), Handbook of marriage and the family (2nd ed., pp. 475–499). New York: Plenum Press. Felner, R., Farber, S., Ginter, M., Boike, M., & Cowen, E. (1980). Family stress and organization following parental divorce or death. Journal of Divorce, 4, 67−76. Felner, R., Ginter, M., Boike, M., & Cowen, E. (1981). Parental death or divorce and the school adjustment of young children. Journal of Consulting and Clinical Psychology, 43, 305−310. Free, M. (1991). Clarifying the relationship between the broken home and juvenile delinquency: A critique of the current literature. Deviant Behavior, 12, 109−167. Glueck, S., & Glueck, E. (1950). Unraveling juvenile delinquency. New York: Commonwealth Fund. Gottfredson, M., & Hirschi, T. (1990). A general theory of crime. Stanford, CA: Stanford University Press. Guidubaldi, J., Cleminshaw, H., Perry, K., & McLoughlin, C. S. (1983). The impact of parental divorce on children: Report of the nationwide NASP study. School Psychology Review, 12, 300−323. Heimer, K. (1997). Socioeconomic status, subcultural definitions, and violent delinquency. Social Forces, 75, 799−833. Hetherington, E. M., Cox, M., & Cox, R. (1982). Effect of divorce on parents and children. In M. E. Lamb (Ed.), Nontraditional families: Parenting and child development (pp. 233−288). Hillsdale, NJ: Lawrence Erlbaum Associates. Hetherington, E. M., & Kelly, J. (2002). For better or for worse: Divorce reconsidered. New York: W.W. Norton. Hirschi, T. (1969). Causes of delinquency. Berkeley: University of California Press. Holden, K. C., & Smock, P. J. (1991). The economic costs of marital dissolution: Why do women bear a disproportionate cost? Annual Review of Sociology, 17, 51−78. Juby, H., & Farrington, D. (2001). Disentangling the link between disrupted families and delinquency. British Journal of Criminology, 41, 22−40. Lewis, O. (1961). The children of Sanchez: Autobiography of a Mexican family. New York: Random House. Loh, E. (1996). Changes in family structure, attained schooling, and adult poverty status. Social Science Quarterly, 77, 145−158. Long, J. S., & Freese, J. (2001). Regression models for categorical dependent variables using Stata. College Station, TX: Stata Press. Mack, K. Y., & Leiber, M. J. (2005). Race, gender, single-mother households, and delinquency: A further test of power–control theory. Youth and Society, 37, 115−144.

Martinson, B., & Wu, L. (1992). Parental histories: Patterns of change in early life. Journal of Family Issues, 13, 351−377. Matsueda, R., & Heimer, K. (1987). Race, family structure, and delinquency: A test of differential association and social control theories. American Sociological Review, 52, 826−840. McLanahan, S. (1985). Family structure and the reproduction of poverty. American Journal of Sociology, 90, 873−901. McLanahan, S., & Sandefur, G. (1994). Growing up with a single parent. What hurts, what helps. Cambridge, MA: Harvard University Press. McLeod, J., Kruttschnitt, C., & Dornfeld, M. (1994). Does parenting explain the effects of structural conditions on children's antisocial behavior? Social Forces, 73, 575−604. McNulty, T., & Bellair, P. E. (2003a). Explaining racial and ethnic differences in adolescent violence: Structural disadvantage, family well-being, and social capital. Justice Quarterly, 20, 1−31. McNulty, T., & Bellair, P. E. (2003b). Explaining racial and ethnic differences in serious adolescent violent behavior. Criminology, 41, 709−748. Merton, R. (1938). Social structure and anomie. American Sociological Review, 3, 672−682. Miller, W. B. (1958). Lower class culture as a generating milieu for gang delinquency. Journal of Social Issues, 14, 5−19. Nye, F. I. (1958). Family relationships and delinquent behavior. New York: Wiley. Patterson, G. (1982). Coercive family process. Eugene, OR: Castalia. Peeples, F., & Loeber, R. (1994). Do individual factors and neighborhood context explain ethnic differences in juvenile delinquency? Journal of Quantitative Criminology, 10, 141−157. Popenoe, D. (1996). Life without father. New York: Free Press. Proctor, B. (1998, September 21). Poverty. In U.S. Census Bureau, the official statistics (p. 40). Washington, DC: U.S. Census Bureau. Rankin, B., & Quane, J. (2002). Social contexts and urban adolescent outcomes: The interrelated effects of neighborhoods, families, and peers on African–American youth. Social Problems, 49, 79−100. Rankin, J. (1983). The family context of delinquency. Social Problems, 30, 466−479. Rankin, J., & Kern, R. (1994). Parental attachments and delinquency. Criminology, 32, 495−515. Rebellon, C. (2002). Reconsidering the broken homes/delinquency relationship and exploring its mediating mechanism(s). Criminology, 40, 103−135. Rodgers, B., & Pryor, J. (1998). Divorce and separation: The outcomes for children. New York: Joseph Rowntree Foundation. Rosen, L., & Neilson, K. (1982). Broken homes. In L. D. Savitz & N. Johnston (Eds.), Contemporary criminology (pp. 126−135). New York: John Wiley. Sampson, R. (1987). Urban Black violence: The effect of male joblessness and family disruption. American Journal of Sociology, 93, 348−382. Sampson, R. (1992). Family management and child development: Insights from social disorganization theory. In J. McCord (Ed.), Fact, frameworks, and forecasts (pp. 63−93). New Brunswick, NJ: Transaction. Sampson, R., & Laub, J. (1993). Crime in the making: Pathways and turning points through life. Cambridge, MA: Harvard University Press. Sampson, R., Raudenbush, S., & Earls, F. (1997). Neighborhoods and violent crime: A multilevel study of collective efficacy. Science, 277, 918−924. Sampson, R., & Wilson, W. (1995). Toward a theory of race, crime, and urban inequality. In J. Hagan & R.D. Peterson (Eds.), Crime

K.Y. Mack et al. / Journal of Criminal Justice 35 (2007) 51–67 and inequality (pp. 37−54). Stanford, CA: Stanford University Press. Shaw, C. R., & McKay, H. D. (1932). Are broken homes a causative factor in juvenile delinquency? Social Forces, 10, 514−524. Simons, R. L., & Chao, W. (1996). Conduct problems. In R. L. Simons & Associates (Ed.), Understanding differences between divorced and intact families. Stress interaction and child outcome (pp. 125−143). Thousand Oaks, CA: Sage. Simons, R. L., Simons, L., & Wallace, E. (2004). Families, delinquency, and crime: Linking society's most basic institution to antisocial behavior. Los Angeles: Roxbury. Sprey, J. (1967). The study of single-parenthood: Some methodological considerations. Family Coordinators, 16, 29−34. Stacey, J. (1996). In the name of the family: Rethinking family values in the postmodern age. Boston: Beacon Press. Thornberry, T. (1987). Toward interactional theory: An examination of reciprocal causal relationships among family, school, and delinquency. Criminology, 25, 863−891. Umberson, D., & Slaten, E. (2000). Gender and intergenerational relationships. In D. H. Demo, K. A. Allen, & M. A. Fine (Eds.), Handbook of family diversity (pp. 105−127). New York: Oxford University Press.

67

Van Voorhis, P., Cullen, F., Mathers, R., & Chenoweth Garner, M. (1988). The impact of family structure and quality on delinquency: A comparative assessment of structural and functional factors. Criminology, 2, 235−261. Wadsworth, M. (1979). Roots of delinquency: Infancy, adolescence and crime. Lanham, MD: Rowman and Littlefield. Wallerstein, J., & Kelly, J. B. (1980). Surviving the breakup: How children and parents cope with divorce. New York: Basic Books. Wells, L. E., & Rankin, J. H. (1986). The broken homes model of delinquency: Analytic issues. Journal of Research in Crime and Delinquency, 23, 68−93. Wells, L. E., & Rankin, J. H. (1991). Families and delinquency: A meta–analysis of the impact of broken homes. Social Problems, 38, 71−93. Wilkinson, K. (1974). The broken family and juvenile delinquency: Scientific explanation or ideology? Social Problems, 21, 726−739. Wright, B., Caspi, A., Moffit, T., Miech, R., & Silva, P. (1999). Reconsidering the relationship between SES and delinquency: Causation but not correlation. Criminology, 37, 175−194.