Does the relationship between family structure and delinquency vary according to circumstances? An investigation of interaction effects.

• Author: Kierkus, Christopher A.

Les recherches empiriques revelent que la dislocation familiale n'est pas etrangere aux comportements delinquants. Selon les dernieres etudes, il se peut que le relachement des liens au sein des familles non traditionnelles soit responsable de cette relation de cause a effet. Cependant, on ne sait pas si les impacts de la dislocation familiale varient en fonction du statut socio-economique (SSE) familial ou du sexe des enfants. Ainsi, certains auteurs pretendent que les impacts criminogenes sont plus marques chez les garcons, alors que d'autres soutiennent que ces impacts negatifs se font sentir davantage chez les filles. On releve les memes types de donnees contradictoires au sujet du SSE. Enfin, un nombre important d'etudes revelent que l'influence de la dislocation familiale varie peu en fonction du sexe et du SSE. La presente etude a pour ojectif de determiner sily a interaction entre la dislocation familiale et les deux variables visees et, pour ce faire, prend ancrage sur une analyse de regression logistique multivariee. Plus precisement, on a analyse un echantillon representatif d'ecoliers ontariens (n = 1891). Or, les resultats de l'etude portent a croire que la structure familiale interagit effectivement avec le SSE mais qu'elle interagit avec un seul type de comportement delinquant, soit la desertion de l'ecole. Ainsi, seuls les enfants provenant de familles non traditionnelles riches sont plus susceptibles de deserter l'ecole. Globalement, la relation entre la structure familiale et la delinquance varie peu, quelles que soient les circonstances.

Empirical research has shown that familial disruption is associated with delinquent behaviour. Recent investigations suggest that reduced levels of attachment in non-traditional families may be responsible for this effect. However, it is not known whether the impact of familial disruption varies according to familial socio-economic status (SES) or the gender of the children. Some authors have argued that the criminogenic influence is greater for boys, while others have maintained that girls are more adversely influenced. Similar contradictory evidence has been reported with respect to SES. Finally, a substantial number of studies have shown that the influence of familial disruption is largely invariant to gender and SES. The goal of this study was to determine whether or not familial disruption interacts with these two variables. Multivariate logistic regression was used in the investigation. A representative sample of Ontario school children was analysed (N = 1,891). The analysis reveals that family structure interacts with SES, but only with respect to one form of delinquent behaviour. This result may represent a chance finding. Overall, the relationship between family structure and delinquency is remarkably similar across circumstances.

Introduction

A popular criminological theory suggests that children who are raised by both of their biological parents are less likely to become involved in delinquency than children raised in alternative family structures. In the past, this has been termed the broken homes hypothesis. Although there is some debate concerning the magnitude of relationship between family structure and delinquency, and about the causal mechanisms that produce it (see Gove and Crutchfield 1982; Rosen 1985; Van Voorhis, Cullen, Mathers, and Garner 1988; Goetting 1994; Kierkus and Baer 2002), there is general consensus that non-traditional family structures are criminogenic (see Free 1991; Wells and Rankin 1991).

This is little consensus, however, concerning how family structure influences delinquency under different circumstances. For instance, a number of studies have reported that familial disruption may have a greater impact on girls than on boys (Datesman and Scarpitti 1975; Andrew 1976; Austin 1978; Biron and LeBlanc 1977; Wilkinson 1980; Offord 1982; Pulkkinen 1983; Steinberg 1987; Anderson, Holmes and Ostresh 1999; Bachman and Peralta 2002). However, a substantial body of literature reached precisely the opposite conclusion (Touliatos and Lindholm 1980; Canter 1982; Gove and Crutchfield 1982; Goldstein 1984; Dornbusch, Carlsmith, Bushwall, Ritter, Leiderman, Hastrof, and Gross 1985; Kandel, Simcha-Fagan, and Davies 1986; Peterson and Zill 1986; Needle, Su, and Doherty 1990). Moreover, several important studies found that family structure and gender do not interact with respect to predicting delinquency (Chilton and Markle 1972; Rankin 1983; Smith and Paternoster 1987; Flewelling and Bauman 1990; Adlaf and Ivis 1997; Sokol-Katz, Dunham, and Zimmerman 1997; Cookston 1999). Similarly, a number of authors have contended that familial disruption is most criminogenic in areas of high socioeconomic status (SES) (A. Berger and Simon 1974; Kraus 1977; Austin 1978; Johnstone 1978; Flewelling and Bauman 1990). However, two studies (Touliatos and Lindholm 1980; Goldstein 1984) reached the opposite conclusion. Competing theoretical explanations have been offered to explain the disparate findings but have not been systematically tested. (2)

The goal of this study was to test a series of hypotheses involving the interaction of family structure with both gender and SES in determining the risk of delinquency. By carefully reviewing the literature and identifying key methodological and conceptual assumptions made by previous investigators, the present study has attempted to resolve these important theoretical debates.

Methodological issues in family structure research

The literature pertaining to methodological issues in family structure research has been reviewed elsewhere (see Kierkus and Baer 2002). Consequently, only key findings pertinent to the present study will be highlighted here. It has been shown that the effect of familial disruption varies according to the type of delinquent behaviour under consideration. Consequently, it is inappropriate to use composite delinquency scales for dependent variables (see Rankin 1983; Van Voorhis et al. 1988). Moreover, there is a general consensus that the use of official statistics substantially overestimates the magnitude of the family structure and delinquency relationship (see Wells and Rankin 1991). Consequently, studies using such measures should be seen as less reliable than those using self-report data. Many early studies have been criticized for operationalizing family structure as a simple dichotomy (i.e., "broken" versus "intact") (see Johnstone 1978; Rankin 1983; Wells and Rankin 1986; Van Voorhis et al. 1988; Flewelling and Bauman 1990; Needle et al. 1990; Free 1991). It has been demonstrated empirically that single-parent homes, reconstituted homes, and homes where neither natural parent is present have unique effects on delinquency (Virkkunen 1976; Wadsworth 1976; Blechman, Berberian, and Thompson 1977; McCarthy, Gersten, and Langner 1982; Rankin 1983, Johnson 1986; Steinberg 1987; Flewelling and Bauman 1990; Needle et al. 1990; Kierkus and Baer 2002). Wells and Rankin (1986) have shown that a four-category operationalization of family structure--intact (if both of the biological parents are present), neither natural parent (if both of the biological parents are lacking), reconstituted (if one biological parent has been replaced by a step-parent), and single-parent (if a biological parent is raising a child alone)--is both valid and reliable.

Review of the literature

Family studies researchers have long suspected that family structure might have different influences on males and females (see Nye 1958; Chilton and Markle 1972; Datesman and Scarpitti 1975; Austin 1978) and on children from wealthy neighbourhoods as opposed to those living in poverty (see A. Berger and Simon 1974; Kraus 1977; Austin 1978; Johnstone 1978). However, there is little consensus regarding precisely how these interactions function, or even whether they are important for understanding why family structure influences delinquency.

There are two possible reasons why reviewers have been unable to reach a consensus on these issues. First, inadequate attention has been paid to the research designs used by studies that have reported contradictory findings. Second, a coherent theoretical understanding of why statistical interactions exist has not developed. Most studies that discovered interaction effects have offered ad hoc theoretical explanations, but no systematic attempt has been made to reconcile these competing theories.

Interactions between family structure and gender

With respect to the issue of gender interactions, the traditional view is that familial disruption is more criminogenic for girls than for boys (Toby 1957; Morris 1964). This assertion is based on the assumption that girls are more family-oriented than boys; consequently, the family represents a more important source of social control for girls (Toby 1957; Morris 1964; Andrew 1976; Austin 1978; Wilkinson 1980; Offord 1982). Since it is known that social control inhibits delinquency (Hirschi 1969; Gottfredson and Hirschi 1990), familial disruption is believed to be more criminogenic for girls.

Ten empirical studies provide support for this hypothesis. However, the results of four should be treated with caution because they are based on official data (Datesman and Scarpitti 1975; Andrew 1976; Offord 1982; Pulkkinen 1983). Official statistics create a serious validity threat when used in gender interaction research because it has been shown that familial disruption is more closely related to official delinquency for girls but more strongly associated with self-report delinquency for boys (Johnson 1986). Three other studies analysed small samples drawn from single cities (Biron and LeBlanc 1977; Steinberg 1987; Anderson et al. 1999). Consequently, their findings cannot be considered generalizable. The remaining three studies based their findings on numerically large probability samples. Therefore, their findings are considered in detail.

Austin...