Linked lives: the intergenerational transmission of antisocial behavior

Terence P. Thornberry

INTRODUCTION

There is a strong commonsense notion of intergenerational continuity in behavior. Children are often presumed to take after their parents, a notion reflected in many adages: "The apple never falls far from the tree," "Cut from the same cloth," and "Like father, like son' to mention a few. The assumption of intergenerational continuity in behavior, including antisocial behavior, also pervades the scientific literature. Intergenerational continuities have been suggested and observed for a variety of behaviors, including aggression (Doumas, Margolin, & John, 1994; Huesmann, Eron, Lefkowitz, & Walder, 1984), alcohol and drug use (Velleman, 1992), family violence (Fagan, Hansen, & Stewart, 1983), criminal history (Farrington, Jolliffe, Loeber, Stouthamer-Loeber, & Kalb, 2001; Farrington, Lambert, & West, 1998), and child abuse (Widom, 1989).

Despite the plausibility of intergenerational continuities in behavior, there is surprisingly little research that prospectively traces the life course of adjacent generations to see if parents who are antisocial also have children who are antisocial. There is even less information about the causal processes that might account for any observed level of intergenerational continuity. In this paper we examine whether a parent's involvement in antisocial behavior during adolescence increases the risk that his or her children will also become involved in antisocial behavior. If it does, we examine potential mediators that may account for how their lives become bound together in this fashion. We also examine whether there are differences in the mediating pathways for mothers and fathers, because much less is known about the impact of fathers on their children's development.

To examine these issues, we use data from the Rochester Youth Development Study (RYDS), a multigenerational, prospective investigation of child development and antisocial behavior. The study began in 1988 with an initial sample of 1,000 adolescents (mean age = 13.5) who were interviewed for a total of 12 assessments between then and 1997. These participants are referred to as Generation 2 (G2). In 11 of the 12 assessments, a parent was also interviewed; they are referred to as Generation 1 (G1). In 1999 we began following G2's oldest biological child, Generation 3 (G3), with annual assessments. As a result, the Rochester project has long-term prospective data on three generations of the same families that allow us to examine a variety of hypotheses about intergenerational processes.

CONCEPTUAL FRAMEWORK

Our conceptual framework for examining these issues is derived from a general life-course perspective on human development (Elder, 1985, 1997) and, in particular, from interactional theory's explanation for antisocial behavior (Thornberry, 1987; Thornberry & Krohn, 2001). A central theme of the life-course perspective is that of interdependent or linked lives. Human development takes place in the context of intertwined social relationships and the shape of one's life course is influenced by the shape of the life courses of others. 'Actors do not behave or decide as atoms outside a social context... Their attempts at purposive action are instead embedded in concrete, ongoing systems of social relations" (Granovetter, 1985, p. 487).

One of the most intimate and influential of these relationships is that between parent and child. As Glenn Elder has noted, "Each generation is bound to fateful decisions and events in the other's life course" (1985, p. 40). Events like catastrophic illness, drug addiction, and divorce not only affect the individual and his or her spouse, but ripple out to affect both younger and older generations. For any focal individual, for example, heavy involvement in antisocial behavior and drug use potentially impacts both the person's parents and children.

Although the concept of linked lives refers to reciprocal influences between individuals, for example, between parent and child, in this paper we are particularly interested in one of these directions, the impact of parental behavior on the child's development. We investigate whether and how "fateful decisions and events" that have shaped the parent's life course ripple out to affect the development of their children.

The full explanation of how G2's involvement in antisocial behavior influences G3's involvement is obviously complex. Explanations range from genetic models in which there is direct transmission of risk from parent to child to shared-environment models in which the behavioral similarity is spurious, produced by each generation responding to similar environmental stimuli. Testing the full range of these possible influences is virtually impossible at the current time, and beyond the scope of the present analysis. Our approach, therefore, is more focused. We start with a core part of interactional theory's model of antisocial behavior that has been shown to be predictive of delinquency and other problem behaviors within a generation. We then develop an intergenerational extension of that model and test it, using data from the Rochester study. The core model we examine is presented in Fig. 1.

Intragenerational Influences

The causal paths on the left-hand side of Fig. 1 refer to the influence of G 1 on G2 and reflect a core set of causal processes related to antisocial behavior. The most proximal influence modeled here is that of parenting styles. Jang and Smith (1997) identified two central dimensions of parenting for the explanation of adolescent antisocial behavior: positive affective relations between the parent and child, and direct parental control over the child's behavior via clear supervision and consistent discipline (see also Loeber & Stouthamer-Loeber, 1986). These dimensions are similar to what Baumrind (1991) identified as authoritative parenting, the combination of high parental responsiveness and high control.

The basic premise is that G1's effective parenting of G2 reduces the chances of G2 becoming involved in antisocial behavior. The child's behavioral freedom is constrained to the extent to which there is a strong affective bond between the parent and child. Fearful of losing his or her attachment to the parent, the child is constrained by parental values and norms that, by and large, are prosocial. Similarly, parents who monitor their children's whereabouts and behaviors, and who use consistent and fair disciplinary practices, are more likely to prevent the onset and development of antisocial behavior. These basic assertions are embedded in most theories of delinquency and antisocial behavior (e.g., Catalano & Hawkins, 1996; Conger et al., 1992; Hirschi, 1969; Patterson, Reid, & Dishion, 1992). Also, there is an impressive body of empirical data that links effective parenting to reduced levels of antisocial behavior (e.g., Conger et al., 1992; Krohn, Stern, Thornberry, & Jang, 1992; Loeber & Stouthamer-Loeber , 1986; Patterson et al., 1992).

Interactional theory also posits that parenting styles are influenced by the family's position in the social structure. In particular, family poverty reduces effective family management skills and positive parent-child relations (Belsky, Woodworth, & Crnic, 1996; Conger, Ge, Elder, Lorenz, & Simons, 1994; Patterson et al., 1992; Stern & Smith, 1995). (6)

In brief, one pathway to antisocial behavior is from poverty to ineffective parenting and then to delinquency. These pathways are intragenerational, because they attempt to explain G2's antisocial behavior by focusing on contemporaneous behaviors and statuses of the parents. If they are accurate, and much empirical evidence suggests they are (see Loeber & Stouthamer-Loeber, 1986), these pathways can be generalized and applied to all generations. Thus, the basic model is repeated in the right-hand portion of Fig. 1, this time with respect to the influence of 02 on G3. For the reasons just presented, G2's financial stress7 influences their parenting and that, in turn, influences G3's antisocial behavior.

To this point, these two sections of the model represent standard, intragenerational examinations of antisocial behavior. That is, our understanding of how the parent influences the child is limited to portions of the parent's life course that are unfolding contemporaneously with the child's. These contemporaneous influences are certainly important and are one of the most fundamental ways in which the parent and child have linked lives. However, the notion of linked lives for parents and children is not limited to concurrent linkages; it includes developmental ones as well, and that leads us to intergenerational influences. Can our understanding of a focal generation's antisocial behavior be enhanced by an understanding of the previous generation's own development and their own earlier antisocial behavior? (8) Linking the two sections of Fig. 1 begins to address this.

Intergenerational Influences

We posit three paths by which G2's adolescent antisocial behavior influences G3's antisocial behavior (Fig. 1). The first is a direct path; parents who are antisocial are likely to have children who are antisocial (e.g., Farrington et al., 1998; Huesmann et al., 1984). This is, in many ways, the least informative path because it sheds little, if any, light on the causal processes involved. Its inclusion is important, however, as the model at this time only incorporates a few of the possible mediating paths. (9)

The other pathways from G2 antisocial behavior to G3 antisocial behavior are indirect and focus attention on the life-course consequences of G2's involvement in adolescent antisocial behavior. Interactional theory emphasizes that antisocial behavior is embedded in reciprocal causal loops. Life-course development influences antisocial behavior, and antisocial behavior also influences life-course development:

early, persistent involvement in antisocial behavior and delinquency generates cumulative and cascading consequences in the person's life course. It reduces the formation of social bonds and social capital and increases embeddedness in deviant networks and belief systems, all of which serve to foreclose conventional lifestyles and entrap the individual in deviant lifestyles. (Thornberry & Krohn, 2001, p.296)

Because of these consequences, adolescent antisocial behavior is likely to reduce the chances that the person will become an effective parent when he or she has children. Parents who are aggressive, who have limited human and social capital, and who remain embedded in deviant social networks are less likely to have the personal and social resources necessary to effectively discharge the responsibilities of parenthood (Belsky, 1984). Thus, one mediator of intergenerational continuity in antisocial behavior from G2 to G3 is likely to be the effectiveness of G2's parenting of G3.

Another path by which G2's adolescent antisocial behavior can ultimately influence G3's antisocial behavior is via its impact on G2's position in the social structure. Adolescent antisocial behavior is likely to create disorder in the transition to adulthood and increase the likelihood of financial stress. In turn, stress is likely to lead to poor family management skills and brittle parent--child relationships (Belsky et al., 1996; McLoyd, 1990; Patterson et al., 1992; Rutter & Giller, 1983; Stern & Smith, 1995). As just noted, these parenting deficits are the most proximate and powerful influences on the development of G3's childhood antisocial behavior.

Finally, our conceptual model posits direct effects from G1's poverty to G2's financial stress, recognizing that there is some level of intergenerational stability in social class position (Rodgers, 1995). It also contains a direct path from G1 parenting behavior to G2 parenting behavior. There are a variety of processes by which children adopt the parenting styles of their parents (Putallaz, Costanzo, Grimes, & Sherman, 1998; Rutter, 1989). Chen and Kaplan (2001) have shown intergenerational stability for positive aspects of parenting, and Simons, Whitbeck, Conger, and Chyi-In (1991) have shown similar effects for harsh/negative aspects of parenting. Although G1 influences G2 in these ways, we anticipate that G1's influence on G3 is indirect, mediated by its impact on G2's life course.

Gender Differences

We also examine whether intergenerational continuities in antisocial behavior, and the role of parenting as a mediator, vary for G2 mothers and fathers. Relative to what is known about the impact of mothers, there is surprisingly little information about the impact of fathers on the development of their children's antisocial behavior (Doherty, Kouneski, & Erickson, 1998; Parke & Stearns, 1993), and even less comparing mothers and fathers. Understanding how fathers influence their children's development, and whether these processes differ from maternal influences, is an important topic in its own right. It has become even more important in light of recent policy recommendations to promote marriage and to reunite parents to improve child outcomes (Popenoe, 1996; Tamis-LeMonda & Cabrera, 1999), combined with evidence that the presence of antisocial fathers may actually increase their children's risk for problem behaviors (Furstenberg & Harris, 1993; Jaffee, Moffitt, Caspi, & Taylor, in press).

Our general expectation in examining possible gender differences is twofold. First, we expect that the basic processes presented in Fig. I apply to both G2 mothers and fathers. That is, we expect that there is intergenerational continuity in antisocial behavior from G2 mothers and fathers to their G3 children (Cohen, Kasen, Brook, & Hartmark, 1998; Elder, Caspi, & Downey, 1986; Huesmann et al., 1984), and that parenting behaviors have a strong proximal and mediating influence on G3's antisocial behavior for both mothers and fathers (Buehler & Gerard, 2002; Caspi & Elder, 1988; Dunn, Davies, O'Connor, & Sturgess, 2000; Sorenson & Brownfield, 1991). Second, however, we anticipate that the strength of some of the relationships presented in Fig. 1 varies by G2 gender. Three relationships seem particularly important.

First, we hypothesize that the level of intergenerational continuity in antisocial behavior is stronger for G2 fathers than for G2 mothers. Antisocial behavior is more prevalent, frequent, and serious, and exhibits greater stability across the life course for males than for females (Elliott, Huizinga, & Menard, 1989; Moffitt, Caspi, Rutter, & Silva, 2001). Given the greater centrality of antisocial behavior for males, we anticipate that it is more consequential for their own development and that of their children. Our expectation that the level of continuity in antisocial behavior is greater from G2 fathers than from G2 mothers is consistent with results presented by Farrington et al. (2001).

Second, we hypothesize that parenting styles more fully mediate the impact of adolescent antisocial behavior for G2 mothers than for G2 fathers. In American culture, mothers are cast as the primary parent and ..... norms are stricter on the centrality and endurance of the mother--child dyad" (Doherty et al., 1998). Fathers typically play a secondary role in parenting (LaRossa, 1986; Parke, 1981). The father--child relationship is much less enduring and much more strongly shaped by contextual influences, especially relations with the other parent, than is the mother---child relationship (Doherty et al., 1998). Consistent with this difference in centrality, the impact of earlier development on parenting style has been found to be stronger for females than for males. "Difficult, ill-tempered boys may experience problems in the adult domain most central to men's lives: work. Problem girls, on the other hand, may adjust poorly to the role traditionally most salient to women: parenting" (Elder et al., 1986, p.329) . As a result, we hypothesize that parenting behavior is a more powerful mediator of the impact of G2's earlier antisocial behavior for mothers than for fathers. There is very little empirical data on this issue in the intergenerational literature. However, Wu and Kandel (1995) did find that the effect of father's criminality on delinquency was not mediated by socialization variables, whereas mother's criminality had an indirect effect on daughter's delinquency via parental rule setting (see also Buehler & Gerard, 2002).

Third, we hypothesize that the direct link from G1 parenting style to G2 parenting style is stronger for G2 mothers than for fathers. In the Rochester data, virtually all the GI respondents are mothers (see below); therefore, we are hypothesizing that same-sex continuity in parenting is stronger than cross-sex continuity. Part of this, we expect, is due to the direct modeling of gender roles by the daughters. As Simons et al. (1991) have noted, given the centrality of the parenting role to the life course of women, adolescent girls may be more likely than adolescent boys to engage in anticipatory socialization and to pay particular attention to their mother's style of parenting. Consistent with this notion, Elder et al. (1986) and Simons et al. (1991) found somewhat stronger continuity in parenting style for 01 mothers and daughters than for G1 mothers and sons, and Snarey (1993) found that mother--son relationships during childhood were not predictive of the son's parenting style. (10)

PREVIOUS STUDIES

A number of previous studies have empirically examined the intergenerational transmission of antisocial behavior and mediating processes like those contained in our model. Unfortunately, many of them are methodologically weak. One of the most fundamental flaws is reliance on the G2 parent as the sole source of data, reporting on their own behavior as well as the behavior of their parents and their children (e.g., Zuravin, McMillen, DePanfilis, & Risley-Curtiss, 1996). As a result, important concepts about G1's behavior are measured retrospectively and estimates of the magnitude of the relationships across generations may be distorted by reliance on a common reporter. Other problems, especially with earlier studies, include the use of small samples and samples selected on the basis of one generation's involvement in antisocial behavior. Good methodological critiques are presented by Kaufman and Zigler (1992), MacEwen (l994), and Velleman (1992). Because of these well-known problems, our review of the literatu re focuses on studies that have collected prospective data from multiple generations of the same families. We are concerned with two core questions: first, are adjacent generations similar with respect to involvement in antisocial behavior and, second, to what extent does G2's parenting style mediate this association?

Several studies have examined the extent to which children follow in the footsteps of their parents with respect to antisocial behavior. There is a well-established literature that indicates an association between lifetime parental criminality and offspring criminality (e.g., Farrington et al., 1998; Hagan & Palloni, 1990; Osborn & West, 1979). Cairns et al. (1998), Patterson (1998), and others have argued, however, that for the examination of intergenerational continuity our focus should be on similar forms of antisocial behavior measured at similar developmental stages for both generations. When this is done, the evidence for behavioral similarity across the generations weakens considerably. Indeed, the existing literature presents inconsistent and somewhat contradictory results. Some studies report robust intergenerational relationships (Elder et al., 1986; Farrington, 1993; Huesmann et al., 1984; Kaplan & Liu, 1999; McCord, 1991; Tapscott, Frick, Wootton, & Kruh, 1996), whereas other studies find no sign ificant correlation between G2 and G3 antisocial behavior (Cairns et al., 1998; Cohen et al., 1998; Dobkin, Tremblay, & Sacchitelle, 1997; Serbin et al., 1998). Yet other studies find significant relationships under some conditions (e.g., father's criminality on daughter's antisocial behavior; Wu & Kandel, 1995), but not in general. Thus, the commonsense notion that adjacent generations are likely to behave in similar fashion with respect to antisocial behavior is not universally supported in the scientific literature. Good longitudinal studies with prospective designs (e.g., Cairns et al., 1998; Elder et al., 1986) differ considerably in their conclusions.

Somewhat fewer studies have examined the extent to which G2's parenting behaviors mediate the intergenerational transmission of antisocial behavior. Of those that have, Brook, Tseng, Whiteman, and Cohen (1998), and Elder et al. (1986) (11) find that parenting practices contribute to the transmission of risk from one generation to the next. Cairns et al. (1998) and Tapscott et al. (1996) do not find the effect to be mediated by parenting behaviors or paternal contact, respectively. And other studies provide more ambiguous results. Kaplan and Liu (1999) report that parenting behavior is important when considered as the only mediator, but not when the parent's current psychological distress is considered simultaneously. Wu and Kandel (1995) found a mediating effect for G2 mothers (via parental norm-setting), but not for G2 fathers.

In the end, it is hard to draw firm conclusions about the intergenerational transmission of antisocial behavior from the existing literature. There is evidence that G2's involvement in antisocial behavior increases the risk that G3 will also become involved in antisocial behavior, and that at least part of the association is transmitted via G2's parenting practices. So too, though, is there empirical evidence that is inconsistent with these claims. At this point, it seems that only additional prospective research will shed light on these issues. The basic purpose of this analysis is to use data from the RYDS to examine this topic further.

METHODS

Sampling and Retention

The study began in 1988 when an initial sample of 1,000 adolescents was selected from the population of seventh and eighth graders in the public schools of Rochester, New York. Given the project's initial interest in serious, chronic delinquents, a stratified sample was selected so that (a) the entire public school population is represented in the sample, but (b) high-risk youth are overrepresented, and (c) the findings can be appropriately weighted to represent the target population. Overrepresentation was accomplished by disproportionately stratifying on gender, males were oversampled 75% versus 25%, and proportionately stratifying on residence in high-crime areas of the city (see Krohn & Thornberry, 1999, for details). Because the true probability of selection is known, the panel can be weighted to represent the cohort of all seventh and eighth graders in the Rochester public schools in 1988. In the analyses that follow, the sample is weighted.

Phase 1 of the RYDS covered the adolescent years, ages 13.5-17.5. During this phase, we interviewed the G2 adolescents nine times and the G1 parents eight times, each at 6-month intervals. In 85% of the cases, G1 is G2's biological mother, and in 10%, G1 is G2's stepmother. In Phase 2 we interviewed the G2 participants and their parents at three annual intervals, at ages 20-22. At Wave 12, 85% (846) of the initial 1,000 adolescent participants were reinterviewed; parent interviews were completed for 83% of the families. Krohn and Thornberry (1999) compared those retained and not retained at Wave 10 on gender, social class, family structure, early drug use, delinquency, property crime, and violent crime for the total panel as well as for each racial/ethnic group. Of the 28 significance tests, none attained statistical significance (p < .05). Nevertheless, we also use poststratification weighting to account for the minor level of differential attrition experienced. That is, at Year 1 of the intergenerational s tudy, eligible participants were weighted to make the sample representative of the initial weighted Phase 1 sample.

The eligible G3 participants for the intergenerational study consisted of the first-born biological children (age 2 and older) of the 865 G2 participants who were interviewed at Wave 11 or 12. Because of age differences in some key measures, the current analysis is limited to children 4 years of age and older in Year 1, and so our discussion of sampling focuses only on them.(12) Given this criterion, there were 357 eligible G3 participants and we successfully recruited 83% (n = 296) for the intergenerational study. Cooperation varies by G2 gender. RYDS G2 mothers are almost always the child's primary caregiver, and 97% of them (128 of 132) enrolled in the intergenerational study in Year 1. Despite the fact that only 30% of the RYDS G2 fathers lived with the eligible child at the beginning of the study and many of the nonresident fathers had only sporadic contact with their child, 75% of their children (168 of 225) enrolled in the study in Year 1. For the RYDS fathers, major reasons why the G3 child is not in the study are that the child's mother refused, the father has lost contact with the G3's mother, or the G2 father refused to participate. We compared the 168 G2 fathers who participated with the 57 who did not in terms of race/ethnicity, age at the birth of G3, adolescent delinquency, high school graduation, history of maltreatment, and changes in family structure during adolescence. Only one of the differences was statistically significant (p < .05); high school dropouts were more likely to participate than graduates (82% vs. 70%). Attrition from Years 1 to 2 was quite small; in Year 2, 99% of those interviewed in Year 1 (294 of 296) were reinterviewed.

Procedures

To understand G3's development, we collect data annually from three key respondents: the G3 child (if age 8 or older), the RYDS G2 parent, and the child's other caregiver (OCG). The OCG is the person who, in addition to the RYDS parent, is primarily responsible for raising the child. In the case of RYDS fathers, this is overwhelmingly (93.5%) the child's biological mother. In the case of the RYDS mothers, however, the OCG varies. In a substantial portion of the cases (35.4%), it is the child's biological father. Because many are single-parent families, grandmothers (39.6%), partners of the 02 parent (12.5%), and other relatives (12.5%) are also included.

Although there are a total of 296 G3 participants at Year 1, because of missing data 220 are included in this analysis: 109 G2 fathers and Ill G2 mothers. Many of the measures about G3's behavior and about G2's parenting assume that the parent respondent is involved, beyond some minimal level, in the child's life. For example, assessments of the child's behavior on the Achenbach CBCL or disciplinary styles toward G3 are relatively meaningless for parents who have little or no actual contact with G3. Because of that, these questions were skipped if the parent respondents did not supervise the child for at least an hour or more during the month prior to the interview. There were 43 RYDS fathers and 8 RYDS mothers who did not meet this criterion. An additional 16 RYDS fathers and 9 RYDS mothers had missing data on one or more of the measures from the earlier phase of the study. We compared those included and excluded from this analysis on the same six variables used to assess noncooperation. Only one, race/ethn icity, was statistically significant (p < .05); White respondents were less likely to be included than African American or Hispanic respondents.

The age, gender, and race/ethnicity distribution of the sample of G3 participants at Year 1 is as follows. The 03 sample is 53% male. Age ranges from 4 to 13 and is fairly normally distributed with the median age being slightly older than 6. There are seven cases skewed above age 10. (13) The sample is largely African American (73%), with 21% Hispanic, 5% White, and 2% other races.

Measurement

Data from participant interviews in Phase 1 (Waves 1-9) of the original study and from Stage 1 (Years 1 and 2) of the intergenerational study were used to operationalize the concepts in our theoretical model. The temporal order of variables in the analysis is maintained to the extent possible; the only cross-sectional relationship in Fig. 1 is that between G2 Financial Stress and G2's Parenting of G3, both of which are assessed at Year I of the intergenerational study. Also, different reporters are used for measures of different concepts wherever possible. This is particularly important for assessing antisocial behavior: G2 delinquency is based on self-reports and G3's early antisocial behavior on the other caregiver's report. Descriptive data and the reporter for all measures used in this analysis are presented in Table I. A brief description of the specific measures used in the analysis follows.

GI Family Poverty. G1's household income and family size were used to construct a dichotomous variable indicating whether or not the family lived below the federally defined poverty level during G2's early adolescence (02 age 13.5).

G1 Affective Ties to G2. G1's affective ties to G2 are measured by an 11-item scale derived from Hudson's Index of Parental Attitudes (Hudson, 1996). This scale assesses the degree of warmth and lack of hostility in the parent-child relationship. Sample items are as follows: ''You think your child is terrific," ''You really enjoy your child' and "You feel very angry toward your child" (reverse-coded). The scale uses a four-category response set ranging from Never to Often. The current analysis averages values for Waves 2-5 (G2 ages 14-15.5) to produce a smoothed estimate of Gi 's attachment to G2 during early adolescence. Higher scores indicate that G 1 has more positive feelings toward 02. On the basis of Cronbach's alpha, reliabilities for this scale at the three waves are .83 or .84.

G1 Consistency of Discipline. A six-item scale is used to measure 01's consistency in disciplining G2. Sample items are as follows: "How often do you follow through with a punishment after (adolescent) is warned to stop doing something, but doesn't stop?" and "Once a punishment has been decided, how often can (adolescent) get out of it?" The same four-category response set is used. The current analysis averages values for Waves 2-5 (02 ages 14-15.5). Higher scores indicate more consistency in discipline. Cronbach's reliabilities for this scale in the three waves range from .70 to .73. Previous publications provide evidence of the validity of both parenting scales (e.g., Krohn et al., 1992; Stern & Smith, 1995).

G2 Adolescent Delinquency. G2's adolescent antisocial behavior is measured by their cumulative frequency of self-reported delinquency from Waves 3 to 9 (G2 ages 14.5-17.5). The general delinquency index used in this analysis includes 24 nonoverlapping items covering a range of delinquent behaviors from minor offenses like status offenses, vandalism, and petty theft, to serious crimes like burglary, aggravated assault, and robbery. Respondents were asked if they had committed each offense during the past 6 months and, if so, how often. Prior to analysis, coders screened the data to ensure that responses fit the category of delinquency being measured, and that the reported actions were not so trivial that they would be ignored by law enforcement officials. Because the frequency measures are positively skewed, a log transformation was used.

The next three measures are based on the Year 1 interviews in the intergenerational study. The respondent is the RYDS G2 parent.

G2 Financial Stress. G2's level of financial stress was operationalized as the number of five signs of financial hardship that G2 had experienced in the year prior to the interview. These include skipping or cutting the size of meals, getting behind on rent, having trouble paying utility bills, having utilities cut off, and receiving public assistance.

G2 Affective Ties to G3. G2's affective ties to G3 are measured by a 10-item scale derived from Hudson's Index of Parental Attitudes (Hudson, 1996). Nine of the items are identical to those asked of G1 with respect to G2. The reliability for this scale is .74. Higher scores indicate that G2 has more positive feelings toward G3 .

G2 Consistency of Discipline. A four-item scale is used to measure G2's consistency in disciplining G3. Three items are identical to the ones used in the G1 measure. The differences in the items on these scales for the two generations reflect our general measurement strategy of trying to maintain as many common items across the generations as possible, while at the same time ensuring that the measures are appropriate to the participants' developmental stage. The reliability for this scale is .68, and higher scores indicate more consistency in discipline.

G3 Early Antisocial Behavior. The final measure is based on Year 2 interviews with the child's OCG. Externalizing behaviors is a 33-item scale derived from the Child Behavior Checklist (CBCL) that includes reports of the child's hostility, aggression, and noncompliance with rules (Achenbach, 1991). For each item, there is a 3-point response choice indicating whether the behavior occurs never, sometimes, or often. By using the Year 2 measure, we ensure that the assessment of G3 's antisocial behavior occurs after G2's report of parenting behavior. By using the OCG report, we avoid the problem of common method variance distorting the relationships between G2's and G3's antisocial behaviors and between G2's parenting style and G3's antisocial behavior. The reliability for this scale is .91.

RESULTS

We estimated our conceptual model of the intergenerational transmission of antisocial behavior (Fig. 1) in a set of structural equation models using LISREL. Separate models were estimated for G2 fathers and G2 mothers.

Measures of the parent's affective ties to the child and consistency of discipline were used as indicators for two latent constructs measuring parenting, one for G1 and one for G2. For the G2 fathers, factor loadings indicate strong associations between these indicators for both G1 (.82 and .69, respectively) and G2 (.57 and .52). The factor loadings for the G2 mothers are also strong for both GI (.70 and .87, respectively) and G2 (.64 and .58).

G2 Fathers

Table II presents the correlations among the latent constructs for the G2 fathers. There is evidence of intergenerational continuity in antisocial behavior between G2 and G3 . The correlation between G2's adolescent delinquency and G3's early antisocial behavior, as reported by the OCG, is .31. In addition, effective parenting behavior is negatively correlated with antisocial behavior in both generations (-.28 and -.47, respectively). G1's family poverty is significantly correlated with G2 financial stress, but there is no significant correlation between G1's parenting style and G2's parenting style.

The results for the structural equation model for G2 fathers are presented in Fig. 2. The overall fit of the model is adequate. The ratio of chi-square to the degrees of freedom is 1.2, well below the benchmark of 5.0 (Bollen, 1989). Also, the Goodness of Fit Index (GFI) is above .90 (GFI = .96).

There appears to be some transfer of risk with respect to antisocial behavior from the G2 fathers to their oldest biological child. G2's antisocial behavior during adolescence is significantly and positively related to the level of G3's antisocial behavior during childhood (fi = .19). Recall that the measure of G3's antisocial behavior is based on the report by the child's other major caregiver, and so this result is not the product of a common reporter.

Although this direct effect is evident, a large part of the continuity in antisocial behavior across the generations is mediated by G2's parenting style as hypothesized in our model. Adolescent involvement in delinquency is related to a less effective parenting style by G2 ([beta] = - .32). In turn, G2's parenting at Year 1 is negatively associated with G3's antisocial behavior at Year 2 ([beta] = -.41). The indirect effect of G2 adolescent delinquency through G2 parenting of G3 is .13 (-.32 x -.41 = .13). Because the correlation between G2 antisocial behavior and G3 antisocial behavior is .31, about 40% of this effect is mediated by this indirect path. G2 parenting styles characterized by weak affective ties and less consistent control strategies are associated with higher levels of antisocial behavior during G3's childhood.

As anticipated, G2's parenting of G3 impacts G3's antisocial behavior and we have already seen one source of this parenting: G2's adolescent delinquency. Are there others? Interestingly, for males there is no statistically significant direct path from G1's parenting to G2's parenting. There is, however, a substantial indirect path (.09) operating through G2's antisocial behavior (- .28 x - .32 = .09). Less warm and more inconsistent parenting by G1 parents is associated with higher levels of adolescent delinquency and, as we saw, that is associated with less effective styles of parenting by G2 fathers.

Finally, there is another intergenerational path leading to G2's parenting style. G1 family poverty increases the chances that G2 will experience higher levels of financial stress during their early 20s and that, in turn, decreases effective parenting styles (.24 x - .26 = -.06). The path reflecting economic disadvantage operates independently of the path that operates via G2's antisocial behavior for the G2 fathers.

G2 Mothers

The correlations among the latent constructs for G2 mothers are presented in Table III. There is a significant positive correlation (.22) between G2's delinquency and G3's antisocial behavior. There is also evidence of intergenerational similarity between G1 and G2 with respect to parenting styles (.47) and economic position (.18). Economic disadvantage and effective parenting are negatively correlated in both generations. There is a smaller than anticipated correlation between G1's parenting and G2's adolescent delinquency (r = -.15; p < .06), but G2's parenting has a sizable relationship with G3's antisocial behavior (- .37). We turn now to the structural equation model to see if these bivariate relationships are maintained in a multivariate environment.

The results for the structural equation model for G2 mothers are presented in Fig. 3. The overall fit of the model is again acceptable; the ratio of chi-square to degrees of freedom is 2.6 and the GFI is .92.

The pattern of results for the structural equations is somewhat different for the G2 mothers, as compared to the G2 fathers. For the G2 mothers, adolescent involvement in delinquency does not exert a significant direct effect on the level of G3's antisocial behavior during childhood. Its effect is entirely indirect, operating via G2's parenting style.

Young mothers who had been involved in delinquency exhibit less effective parenting styles toward G3 ([beta] = - .22), and, in turn, G2's parenting style is significantly related to G3's antisocial behavior ([beta] = -.33). Weak affective ties and inconsistent control strategies are positively related to G3's antisocial behavior. The indirect effect for mother's early antisocial behavior, through G2 parenting, on G3's antisocial behavior is .07 (- .22 x - .33 = .07).

G1's parenting of G2 has a significant and sizable direct effect ([beta] = .42) on G2's parenting of G3, an effect that was absent for the males. If the affective bond between GI and G2 was less warm and efforts at parental control were less consistent, so too is the parent-child relationship between G2 and G3. This effect is direct and does not operate indirectly via G2's adolescent delinquency as it does for the males. Indeed, GI's parenting of 02 does not have a significant association with G2's adolescent antisocial behavior, contrary to the model's predictions.

Economic disadvantage plays a more central role for the G2 mothers. Growing up in impoverished families is positively related to the level of financial hardship experienced by the young 02 mother; in turn, financial stress is negatively related to the quality of her parenting ([beta] = -.35). The indirect effect is -.06 (.17 x -.35 = -.06). Poverty in the family of origin is also associated with less effective parenting styles by 01 and, as just mentioned, there is a noticeable level of intergenerational stability in parenting style from G1 to G2. That parenting style, of course, has a major impact on G3's antisocial behavior. So, unlike the model for fathers, for mothers there is a significant indirect effect of family poverty through 01's parenting style on G2's parenting style (-.19 x .42 = - .08).

DISCUSSION

This paper investigated the phenomenon of linked lives. In particular, we examined the level of intergenerational transmission of antisocial behavior and the extent to which financial stress and the quality of parenting mediate this intergenerational linkage. We also examined the extent to which these linkages vary for G2 mothers and G2 fathers.

The major point of similarity for G2 mothers and fathers is the immediate and powerful association of their parenting style on G3's early antisocial behavior. Parenting behaviors by 02 that are characterized by warmth and consistency are associated with lower levels of antisocial behavior by G3. Aside from this strong commonality, the results vary by G2's gender. As hypothesized earlier, antisocial behavior plays a more central role in the model for fathers and parenting style plays a more powerful mediating role in the model for mothers.

For the G2 fathers, adolescent involvement in delinquency had both direct and indirect effects (via parenting) on G3's antisocial behavior. Also, G1's parenting style has no direct impact on G2's parenting style; it does, however, have a sizable indirect effect via G2's involvement in adolescent antisocial behavior. Thus, G2's earlier antisocial behavior appears to play a major role in linking the behavioral processes that lead to intergenerational similarity in antisocial behavior for these young fathers and their children.

In contrast, adolescent antisocial behavior plays a much more muted role in the model for the G2 mothers. It is a contributor to G2's ineffective parenting, but that is all. It exerts no direct effect, nor does it mediate the role of G1's parenting style.

The relationship between G1's parenting of G2 and G2's adolescent delinquency also differed by G2 gender. For the males, the relationship was negative and statistically significant; for the females, it was not significant in the structural equation model. The result for the G2 females is inconsistent with some previous literature that demonstrates an impact of parenting on antisocial behavior for both males and females (see Loeber & StouthamerLoeber, 1986). Interestingly though, our finding is consistent with some of those reported in the intergenerational literature. Cairns et al. report that "the links between parenting measures and assessments of early school aggressive patterns were consistent and reliable [only] for the total sample and for boys" (1998, p. 1171). Wu and Kandel report correlations between mother's parenting (closeness, love withdrawal, and punitive discipline) and adolescent delinquency for boys and girls. For the boys, the correlations (-.20, .18, and .16, respectively) are significant, but for the girls only the first correlation (- .26) is significant, the other two (.02, .08) are not (1995, pp. 76-77). More generally, some studies indicate that family aggression has a larger impact on disrnptive behavior for boys than for girls (see, for example, Carlson, 1990; Moffitt et al., 2001). These findings, and ours, suggest that it may be wise to reexamine gender differences in the impact of parenting on antisocial behavior, as well as their role in intergenerational processes.

There is considerable similarity in parenting styles from G1 to G2 for the G2 mothers, but not for the 02 fathers. As we hypothesized earlier, it appears that the intergenerational transfer of parenting styles is gender-specific. Recall that 95% of the G1 respondents were mothers. In our data, therefore, young G2 mothers seem to take on the parenting styles of their mothers, but young G2 fathers do not take on the parenting styles of their mothers, unless it influenced their own development by increasing their delinquency. These results are consistent with those of Elder et al. (1986) and Simons et al. (1991), who also found that parenting styles across the generations were more similar for mother--daughter than mother--son dyads. Understanding the mechanisms that link the generations in this fashion is an important question for future research.

Economic disadvantage, both for G1 and for G2, plays a somewhat more integral role in the model for G2 mothers. Perhaps this is because, in both generations, the mother almost always lives with and is the primary caregiver for the child. Poverty and financial hardship may, therefore, have a larger, more immediate impact on their relationship with their children than is the case with the fathers, many of whom are removed from day-to-day involvement with the child.

So, is there intergenerational continuity in antisocial behavior, and do parenting styles mediate this linkage? The answer appears to be a qualified "yes" and to be somewhat gender-specific. Intergenerational continuity is more evident for G2 fathers than for G2 mothers. And parenting styles appear to be a more potent mediator for G2 mothers than for G2 fathers.

Although these results add to our understanding of a number of central issues about the intergenerational transmission of antisocial behavior, they are not complete and they raise a number of important questions for future research. Among them are the following.

This study was not able to investigate all of the possible pathways that are likely to play a role in accounting for behavioral similarities across the generations. Indeed, our current model is quite constrained. It is possible that part of the observed similarity in antisocial behavior may reflect underlying genetic similarities in such domains as poor impulse control, negative emotionality, and poorer emotion regulation skills (Mednick, Gabrielli, & Hutchings, 1984; Raine, Brennan, & Farrington, 1997). If genetic influences play a role in linking the generations with respect to antisocial behavior, our results suggest that they may be more prominent for the G2 fathers. The association between mother and child was smaller and more fully mediated by parenting style, whereas for fathers it was larger and much of it remained even when financial stress and parenting were considered.

There are also other social and psychological factors that need to be considered. For example, there are aspects of G2's adolescent development other than antisocial behavior, for example, educational attainment and the formation of social capital, that can influence their parenting style and G3's development. There are also pathways other than parenting that can mediate the impact of G2's antisocial behavior on G3's antisocial behavior. For example, we did not consider family structure and partner relationships, the role of G3's other biological parent, and that parent's own history and resources. Thus, the present findings are incomplete and future analyses, with both these and other data, are needed to place them in a broader developmental context.

Another issue for future consideration concerns intergenerational discontinuity. Although we observed intergenerational continuity in antisocial behavior, especially for the fathers, we also observed substantial levels of discontinuity. Many children whose parents were involved in adolescent delinquency are not themselves involved in antisocial behavior, at least at early ages, and vice versa. This finding is consistent with the intergenerational literature and requires explanation. Identifying the protective factors that prevent some children from following in the footsteps of their previously delinquent parent, and identifying the risk factors that generate delinquency in a new generation, are both challenging issues. Future research needs to address issues of intergenerational discontinuity as well as issues of continuity. The role of the child's other biological parent, and that parent's extended family, may be particularly important for understanding discontinuity. Unlike the G2 parent, the other parent has not been followed prospectively, however, and incorporating their influence in intergenerational investigations is methodologically challenging.

The results reported here suggest sizable differences in intergenerational processes by G2's gender. We are not yet able to investigate simultaneously differences by G3's gender, given our sample size. Recent research (Moffitt et al., 2001), however, suggests that the risk processes for childhood and adolescent antisocial behavior are quite similar for boys and girls. Nevertheless, as we continue to enroll new G3 participants and reinterview the participants annually, the sample size will expand and we will address intergenerational processes by G3 gender.

Although the current analysis is limited in these ways, it is not without its strengths. Compared to most intergenerational studies, it avoids many of the major problems identified by earlier critics (Kaufman & Zigler, 1992; MacEwen, 1994; Tedeschi & Felson, 1994; Velleman, 1992). The sample is relatively large for an intergenerational study and it is based on an initially representative community sample. Members of three adjacent generations have been followed prospectively, minimizing reliance on retrospective data and helping to ensure proper temporal order. Also, by interviewing both the RYDS G2 parent and another caregiver for each child, the problem of common method variance, especially with respect to the measures of G3's antisocial behavior, is avoided.

On the basis of these and other positive design features, we think the results presented here contribute to our understanding of the intergenerational transmission of antisocial behavior. Lives do appear to be linked in this respect and family risks, including economic disadvantage and parenting, are clearly among the links that advance our understanding of behavioral continuities and point to directions for reducing them.

Table I

Coding and Descriptive Statistics for Variables



          Variable                        Description

G1 family poverty             0-1 indicator of G1's poverty
                              status, Wave 1
G1 affective ties to G2       Average of 11-item scales,
                              Waves 2-5
G1 consistency of discipline  Average of 6-item scales, Waves 2-5
G2 adolescent delinquency     Log of sum of the frequency of
                              24 acts, Waves 3-9
G2 financial stress           Sum of five signs of financial
                              hardship, Year 1
G2 affective ties to G3       10-item scale, Year 1
G2 consistency of discipline  4-item scale, Year 1
G3 early antisocial behavior  Standardized extenalizing subscale
                              of the CBCL, Year 2

                                           G2 fathers (n = 109)

          Variable            Reporter     Mean          SD

G1 family poverty                G1        0.38         0.49

G1 affective ties to G2          G1        3.44         0.36

G1 consistency of discipline     G1        2.75         0.36
G2 adolescent delinquency        G2        2.84         2.24

G2 financial stress              G2        1.10         1.16

G2 affective ties to G3          G2        4.43         0.40
G2 consistency of discipline     G2        3.69         0.81
G3 early antisocial behavior    OCG        0.10         1.01


                                 G2 mothers (n = 111)

          Variable              Mean          SD

G1 family poverty               0.27         0.45

G1 affective ties to G2         3.54         0.30

G1 consistency of discipline    2.89         0.45
G2 adolescent delinquency       2.15         2.09

G2 financial stress             0.94         1.17

G2 affective ties to G3         4.40         0.44
G2 consistency of discipline    3.38         0.79
G3 early antisocial behavior   -0.17         0.95

Table II

Correlations Among Latent Constructs, G2 Fathers

                              G1 family  G1 parenting  G2 adolescent
                               poverty      of G2       delinquency

G1 parenting of G2              -.06
G2 adolescent delinquency        .02         -.28 *
G2 financial stress              .24 *       -.02           .04
G2 parenting of G3              -.06          .00          -.30 *
G3 early antisocial behavior     .03         -.05           .31 *

                              G2 financial  G2 parenting
                                 stress        of G3

G1 parenting of G2
G2 adolescent delinquency
G2 financial stress
G2 parenting of G3               -.27 *
G3 early antisocial behavior      .12          -.47 *

*p <. 05 (one-tailed)

Table III.

Correlations Among Latent Constructs, G2 Mothers

                              G1 family  G1 parenting  G2 adolescent
                               poverty      of G2       delinquency

G1 parenting of G2              -.19 *
G2 adolescent delinquency        .03       -.15
G2 financial stress              .18 *     -.05            .10
G2 parenting of G3              -.15        .47 *         -.32 *
G3 early antisocial behavior     .05       -.17            .22 *

                              G2 financial  G2 parenting
                                 stress        of G3

G1 parenting of G2
G2 adolescent delinquency
G2 financial stress
G2 parenting of G3               -.40 *
G3 early antisocial behavior      .14          -.37 *

*p < .05 (one-tailed).

ACKNOWLEDGMENTS

Support for this work was provided by the National Institute of Mental Health (5 RO1 MH56486), the U.S. Department of Justice's Office of Juvenile Justice and Delinquency Prevention (96-MU-FX-0014), the National Institute on Drug Abuse (5 ROI DA055 12-07), and the National Science Foundation (SES-8912274), and was aided by grants to the Center for Social and Demographic Analysis at the University at Albany from NICHD (P30 HD32041) and NSF (SBR-95 12290). Points of view or opinions in this document are those of the authors and do not necessarily represent the official position or policies of the funding agencies.

Received May 15, 2002; revision received October 9, 2002; accepted October 14, 2002

(6.) In interactional theory, family poverty is one indicator of a broader concept of structural adversity, which also includes welfare dependence. residence in areas of concentrated poverty, social isolation, and, especially, their co-occurrence. The present analysis utilizes the most central indicator, family poverty.

(7.) A measure of G2's family poverty is not available.

(8.) We are explicitly concerned with earlier antisocial behavior by the parent because the notion of intergenerational continuity in behavior ultimately refers to similar behavior at the same developmental stage (Cairns, Cairns, Xie, Leung, & Hearne, 1998; Huesmann et al., 1984). Concurrent antisocial behavior by the parent is better viewed as a stressor that interferes with effective parenting and adds to our understanding of contemporaneous, or intragenerational, models.

(9.) The model we are examining here is only one part, albeit a very central part, of the general conceptual model that guides this study (Thornberry, Smith, Krohn, Lizotte, & Rodriquex, 1997). Other aspects of the model (e.g., school attainment, social support, etc.) will be examined in the future as the G3 sample increases in size and as additional waves of data are collected,

(10.) The Snarey study did not have data on daughters.

(11.) Caspi and Elder (1988) examine similar issues with the same data set and report similar relationships.

(12.) There were 75 children ages 2-3 in Year 1, not enough to support separate analysis. New 2-year-olds are recruited in subsequent years.

(13.) The results do not change if these children are removed.

REFERENCES

Achenbach, T. M. (1991). Manual for the Child Behavior Checklist/4-18 and 1991 Profile. Burlington: University of Vcrmont.

Baumrind, D. (1991). The influence of parenting style on adolescent competence and substance use. Journal of Early Adolescence, 11, 56-95.

Beisky, J. (1984). The determinants of parenting: A process model. Child Development, 55, 83-96.

Belsky, J., Woodworth, S., & Crnic, K. (1996). Troubled family interaction during toddlerhood. Development and Psychopathology. 8, 477-495.

Bollen, K. A. (1989). Structural equations with latent variables. New York: Wiley.

Brook, J. S., Tseng, L.-J., Whiteman, M., & Cohen, P. (1998). A three generation study: Intergenerational continuities and discontinuities and their impact on the toddler's anger. Genetic, Social and General Psychology Monographs, 24, 335-351.

Buehler, C., & Gerard, J. M. (2002). Marital conflict, ineffective parenting, and children's and adolescents' maladjustment. Journal of Marriage and the Family, 64, 78-92.

Cairns, R. B., Cairns, B. D., Xie, H., Leung, M.-C., & Hearne, S. (1998). Paths across generations: Academic competence and aggressive behaviors in young mothers and their children. Developmental Psychology, 34, 1162-1174.

Carlson, B. E. (1990). Adolescent observers of marital violence. Journal of Family Violence, 5, 285-299.

Caspi, A., & Elder, G. H., Jr. (1988). Emergent family patterns: The intergenerational construction of problem behavior and relationships. In R. A. Hinde & J. Stevenson-Hinde (Eds.), Relationships within families: Mutual influences (pp. 218-240). Oxford: Oxford University Press.

Catalano, R. F., & Hawkins, J. D. (1996). A social development model: A theory of antisocial behavior. In J. D. Hawkins (Ed,), Delinquency and crime (pp. 149-197). New York: Cambridge University Press.

Chen, Z.-Y., & Kaplan, H. B. (2001). Intergenerational transmission of constructive parenting. Journal of Marriage and the Family, 63, 17-31.

Cohen, P., Kasen, S., Brook, J. S., & Hartmark, C. (1998). Behavior patterns of young children and their offspring: A two-generation study. Developmental Psychology, 34, 1202-1208.

Conger, R. D., Conger, K. J., Elder, 0. H., Jr., Lorenz, F. 0., Simons, R. L., & Whitbeck, L. B. (1992). A family process model of economic hardship and adjustment of early adolescent boys. Child Development, 63, 526-541.

Conger, R. D., Ge, X., Elder, 0. H., Jr., Lorenz, F. 0., & Simons, R. L. (1994). Economic stress, coercive family process and developmental problems of adolescents. Child Development, 6S, 541-561.

Dobkin, P. L., Tremblay, R. E., & Sacchitelle, C. (1997). Predicting boys' early-onset substance abuse from father's alcoholism, son's disruptiveness, and mother's parenting behavior. Journal of Consulting and Clinical Psychology, 6S, 86-92.

Doherty, W. J., Kouneski, E. F., & Erickson, M. F. (1998). Responsible fathering: An overview and conceptual framework. Journal of Marriage and the Family, 60, 277-292.

Doumas, D., Margolin, G., & John, R. S. (1994). The intergenerational transmission of aggression across three generations. Journal of Family Violence, 9, 157-175.

Dunn, J., Davies, L. C., O'Connor, T. G., & Sturgess, W. (2000). Parents' and partners' life course and family experiences: Links with parent--child relationships in different family settings. Journal of Child Psychology and Psychiatry, 41, 955-968.

Elder, G. H., Jr. (1985). Perspectives on the life course. In 0. H. Elder Jr. (Ed.), Life course dynamics (pp. 23-49). Ithaca, NY: Cornell University Press.

Elder, G. H,, Jr. (1997). The life course and human development. In R. M. Lerner (Ed.), Handbook of child psychology: Theoretical models of human development (pp. 939-991). New York: Wiley.

Elder, G. H., Jr., Caspi, A., & Downey, G. (1986). Problem behavior and family relationships: Life course and intergenerational themes. In A. Sorensen, F. Weinert, & L. Sherrod (Eds.), Human development and the life course: Multidisciplinary perspectives (pp. 293-340). Hillsdale, NJ: Erlbaum.

Elliott, D. S.. Huizinga, D., & Menard, S. (1989). Multiple problem youth: Delinquency, substance use, and mental health problems. New York: Springer-Verlag.

Fagan, J., Hansen, K. V., & Stewart, D. (1983). Violent men or violent husbands: Background factors and situational correlates of spousal violence. In D. Finkelhor, R. J. Gelles, G. T. Hotaling, & M. A. Strauss (Eds.), The dark side of families (pp. 49-67). Beverly Hills, CA: Sage.

Farrington, D. P. (1993). Understanding and preventing bullying. In M. Tonry (Ed.), Crime & justice: A review of research (Vol. 17, pp. 381-458). Chicago: University of Chicago Press.

Farrington, D. P., Jolliffe, D., Loeber, R., Stouthamer-Loeber, M., & Kalb, L. M. (2001). The concentration of offenders in families, and family criminality in the prediction of boys' delinquency. Journal of Adolescence, 24, 579-596.

Farrington, D. P., Lambert, S., & West, D. J. (1998). Criminal careers of two generations of family members in the Cambridge Study in Delinquent Development. Studies on Crime and Crime Prevention, 7, 85-106.

Furstenberg, F. F., & Harris, K. M. (1993). When and why fathers matter: Impacts of father involvement on the children of adolescent mothers. In R. Lerman & T. Ooms (Eds.), Young unwed fathers: Changing roles and emerging policies (pp. 117-138). Philadelphia: Temple University Press.

Granovetter, M. S. (1985). Economic action and social structure: The problem of embeddedness. American Journal of Sociology, 91, 481-510.

Hagan, J., & Palloni, A. (1990). The social reproduction of a criminal class in working-class London, circa 1950-1980. American Journal of Sociology, 96, 265-299.

Hirschi, T. (1969). Causes of delinquency. Berkeley: University of California Press.

Hudson, W. H. (1996). WALMYR assessment scales scoring manual. Tempe, AZ: WALMYR.

Huesmann, L. R., Eron, L. D., Lefkowitz, M. M., & Walder, L.O. (1984). The stability of aggression over time and generations. Developmental Psychology, 20, 1120-1134.

Jaffee, S. R., Moffitt, T. E., Caspi, A., & Taylor, A. (in press). Life with (or without) father: The benefits of living with two biological parents depend on the father's antisocial behavior. Child Development.

Jang, S. J., & Smith, C. A. (1997). A test of reciprocal causal relationships among parental supervision, affective ties, and delinquency. Journal of Research in Crime and Delinquency, 34, 307-336.

Kaplan, H. B., & Liu, X. (1999). Explaining transgenerational continuity in antisocial behavior during early adolescence. In P. Cohen, C. Slomkowski, & L. N. Robins (Eds.), Historical and geographical influences on psychopathology (pp. 163-191). Mahwah, NJ: Erlbaum.

Kaufman, J., & Zigler, E. (1992). The intergenerational transmission of abuse is overstated. In R. J. Gelles & D. R. Loseke (Eds.), Current controversies on family violence (pp. 209-221). Newbury Park, CA: Sage.

Krohn, M. D., Stern, S. B., Thornberry, T. P., & Jang, S. J. (1992). The measurement of family process variables: An examination of adolescent and parent perceptions of family life on delinquent behavior. Journal of Quantitative Criminology. 8, 287-315.

Krohn, M.D., & Thornberry, T. P. (1999). Retention of minority populations in panel studies of drug use. Drugs and Society, 14, 185-207.

LaRossa, R. (1986). Becoming a parent. Newbury Park, CA: Sage.

Loeber, R., & Stouthamer-Loeber, M. (1986). Family factors as correlates and predictors of juvenile conduct problems and delinquency. In M. Tonry & N. Morris (Eds.), Crime and justice: An annual review of research (Vol. 7, pp. 29-149). Chicago: University of Chicago Press.

MacEwen, K. E. (1994). Refining the intergenerational transmission hypothesis. Journal of Interpersonal Violence, 9, 350-365.

McCord, J. (1991). The cycle of crime and socialization practices. The Journal of Criminal Law and Criminology, 82, 211-228.

McLoyd, V. (1990). The impact of economic hardship on black families and children: Psychological distress, parenting, and socioemotional development. Child Development, 61, 311-346.

Mednick, S. A., Gabrielli, W. F., & Hutchings, B. (1984). Genetic influence in criminal convictions: Evidence from an adoption cohort. Science, 224, 891-894.

Moffitt, T. E., Caspi, A., Rutter, M., & Silva, P. (2001). Sex differences in antisocial behavior: Conduct disorder delinquency, and violence in the Dunedin Longitudinal Study. New York: Cambridge University Press.

Osborn, S. G., & West, D. J. (1979). Conviction records of fathers and sons compared. British Journal of Criminology, 19, 120-133.

Parke, R. (1981). Fathers, Cambridge, MA: Harvard University Press.

Parke, R. D., & Stearns, P. N. (1993). Fathers and child rearing. In G. H. Elder Jr., J. Modell, & R. D. Parke (Eds.), Children in time and place. Developmental and historical insights (pp. 147-170). New York: Cambridge University Press.

Patterson, G. R. (1998). Continuities-A search for causal mechanisms (Comment on the special section). Developmental Psychology, 34, 1263-1268.

Patterson, G. R., Reid, J. B., & Dishion, T. J. (1992). Antisocial boys. Eugene, OR: Castalia Press.

Popenoe, D. (1996). Life without father. New York: Martin Kessler Books.

Putallaz, M., Costanzo, P. R., Grimes, C. L., & Sherman, D. M. (1998). Intergenerational continuities and their influences on children's social development. Social Development, 7, 389-427.

Raine, A., Brennan, P. A., & Farrington, D. P. (1997). Biosocial bases of violence. In A. Raine, P. A. Brennan, D. P. Farrington, & S. A. Mednick (Eds.), Biosocial bases of violence (pp. 1-20). New York: Plenum.

Rodgers, J. R. (1995). An empirical study of intergenerational transmission of poverty in the United States. Social Science Quarterly, 76(1), 178-194.

Rutter, M. (1989). Intergenerational continuities and discontinuities in serious parenting difficulties. In D. Cicchetti & V. Carlson (Eds.), Child maltreatment: Theory and research on the causes and consequences of child abuse and neglect (pp. 317-348). Cambridge: Cambridge University Press.

Rutter, M., & Giller, H. (1983). Juvenile delinquency: Trends and perspectives. New York: Guilford Press.

Serbin, L. A., Cooperman, J. M., Peters, P. L., Lehoux, P. M., Stack, D. M., & Schwartzman, A. E. (1998). Intergenerational transfer of psychosocial risk in women with childhood histories of aggression, withdrawal, or aggression and withdrawal. Developmental Psychology, 34, 1246-1262.

Simons, R. L., Whitbeck, L. B., Conger, R. D., & Chyi-In, W. (1991). Intergenerational transmission of harsh parenting. Developmental Psychology, 27, 159-171.

Snarey, J. (1993). How fathers care for the next generation: A four decade study. Cambridge: Harvard University Press.

Sorenson, A. M., & Brownfield, D. (1991). The measurement of parental influence: Assessing the relative effects of father and mother. Sociological Methods and Research, 19, 511-535.

Stern, S. B., & Smith, C. A. (1995). Family processes and delinquency in an ecological context. Social Service Review, 69, 703-731.

Tamis-LeMonda, C. S., & Cabrera, N. (1999). Perspectives on father involvement: Research and policy. Social Policy Report, 13, 1-25.

Tapscott, M., Prick, P. J., Wootton, J., & Kruh, I. (1996). The intergenerational link to antisocial behavior: Effects of paternal contact. Journal of Child and Family Studies, 5, 229-240.

Tedesehi, J. T., & Felson, R. B. (1994). Violence, aggression, amid coercive actions. Washington, DC: American Psychological Association.

Thornberry, T. P. (1987). Toward an interactional theory of delinquency. Criminology, 25, 863-891.

Thornberry, T. P., & Krohn, M. D. (2001). The development of delinquency: An interactional perspective. In S. O. White (Ed.), Handbook of youth and justice (pp. 289-305). New York: Plenum.

Thornberry, T. P., Smith, C. A., Krohn, M.D., Lizotte, A. J., & Rodriguez, M. (1997). Intergenerational transmission of antisocial behavior, Grant proposal submitted to the National Institute of Mental Health, Rockville, MD.

Velleman, R. (1992). Intergenerational effects--A review of environmentally oriented studies concerning the relationship between parental alcohol problems and family disharmony in the genesis of alcohol and other problems. I: The intergenerational effects of alcohol problems. The International Journal of the Addictions, 27, 253-280.

Widom, C. S. (1989). The intergenerational transmission of violence, In N. A. Weiner & M. E. Wolfgang (Eds.), Pathways to criminal violence (pp. 137-201). Newbury Park, CA: Sage.

Wu, P., & Kandel, D. B. (1995). The roles of mothers and fathers in intergenerational behavioral transmission: The case of smoking and delinquency. In H. B. Kaplan (Ed.), Drugs, crime, and other deviant adaptations: Longitudinal studies (pp. 49-81). New York: Plenum.

Zuravin, S., McMillen, C., DePanfilis, D., & Risley-Curtiss, C, (1996). The intergenerational cycle of child maltreatment. Journal of Interpersonal Violence, 11, 315-335.

Terence P. Thornberry, (1,5)

Adrienne Freeman-Gallant, (2)

Alan J. Lizotte, (1)

Marvin D. Krohn, (3)

Carolyn A. Smith, (4)

(1.) School of Criminal Justice, University at Albany, Albany, New York.

(2.) Hindelang Criminal Justice Research Center, University at Albany, Albany, New York.

(3.) Department of Sociology, University at Albany, Albany, New York.

(4.) School of Social Welfare, University at Albany, Albany, New York.

(5.) Address all correspondence to Terence P. Thornberry, School of Criminal Justice, University at Albany, 135 Western Avenue, Albany, New York; e-mail: t.thornberry@albany.edu.

COPYRIGHT 2003 Springer
COPYRIGHT 2008 Gale, Cengage Learning