1278 Journal of Marriage and the Family children (effect sizes)were small,however,rang- that divorce also has positive consequences for ing from .08 of a standard deviation for psycho- some children.For example,a qualitative study logical adjustment to .23 of a standard deviation by Arditti (1999)found that many offspring from for conduct.For some outcomes,studies conduct- divorced families,especially daughters,reported ed in the 1980s yielded smaller effect sizes than developing especially close relationships with earlier studies conducted in the 1960s and 1970s their custodial mothers-a finding that is consis- Amato and Keith (1991)speculated that the gap tent with some quantitative work (Amato in well-being between children with divorced and Booth,1997).In addition,Amato,Loomis,and nondivorced parents might have narrowed either Booth (1995).Amato and Booth(1997),Hanson because divorce became more socially accepted or (1999),and Jekielek (1998)found that offspring because parents were making greater efforts to re- were better off on a variety of outcomes if parents duce the potentially disruptive impact of divorce in high-conflict marriages divorced than if they on their children. remained married.When conflict between parents During the 1990s,the number of people is intense,chronic,and overt,divorce represents touched by divorce increased,school-based pro- an escape from an aversive home environment for grams for children of divorce became common, children.Only a minority of divorces,however and mediation and education courses for divorcing appear to be preceded by a high level of chronic parents became mandatory in many states(Emery, marital conflict (Amato Booth.1997).For this Kitzmann,Waldron,1999).Given these trends. reason,divorce probably helps fewer children one might expect studies conducted in the 1990s than it hurts to reveal a continued closing of the gap in well- being between children with divorced parents and CAUSATION OR SELECTION? children with married parents.An examination of studies conducted in the 1990s.however.does not The selection perspective holds that differences support this hypothesis. between children from divorced and nondivorced A large number of studies in the 1990s contin- families are due to factors other than marital dis- ued to find that children with divorced parents ruption,including parents'personality character- score lower than children with continuously mar- istics,inept parenting,predivorce marital discord, ried parents on measures of academic success or genetic influence.Consistent with a selection (Astone McLanahan.1991:Teachman.Paasch. perspective,Capaldi and Patterson (1991)found Carver.1996).conduct (Doherty Needle. that mothers'antisocial personalities accounted 1991;Simons and Associates,1996),psycholog- for the association between mothers'marital tran- ical adjustment(Forehand,Neighbors,Devine, sitions and boys'adjustment problems.In con- Armistead.1994:Kurdek.Fine.Sinclair.1994). trast.other studies found significant estimated ef- self-concept (Wenk,Hardesty,Morgan,Blair, fects of divorce even after controlling for aspects 1994).social competence (Beaty.1995:Brodzin- of parents'personalities,including depression sky,Hitt,&Smith,1993),and long-term health (Demo Acock.1996a)and antisocial personal- (Tucker et al.,1997).Furthermore,effect sizes in ity traits (Simons and Associates.1996) the 1990s appear comparable to those of earlier Longitudinal studies provide another type of decades.For example,across 32 studies of chil- evidence.Cherlin and colleagues (1991)found dren's conduct published in the 1990s,the mean that children from maritally disrupted families had effect size was-.19,which is not appreciably dif- more postdivorce behavior problems than children ferent from the mean value of -.18 for studies from nondisrupted families.These differences conducted in the 1980s,as reported in Amato and however,were apparent several years prior to di Keith(1991).Similarly,across 29 studies of psy- vorce,especially for boys.Amato and Booth chological adjustment published in the 1990s,the (1996)found that problems in parent-child rela- mean effect size was-.17,which is slightly larger tionships (including parents'reports that their than the mean value of-.10 for studies conducted children had given them more than the usual num- in the 1980s,also as reported in Amato and Keith. ber of problems)were present as early as 8 to 12 In general,the small but consistent gap in well- years before divorce.Aseltine (1996)and Heth- being between children from divorced and two- erington(1999)obtained comparable results with parent families observed in earlier decades per- regard to children's internalizing behavior,exter- sisted into the 1990s. nalizing behavior,social competence,and self-es- As with studies of adults.a few studies suggest teem,and Doherty and Needle (1991)found com-1278 Journal of Marriage and the Family children (effect sizes) were small, however, rang￾ing from .08 of a standard deviation for psycho￾logical adjustment to .23 of a standard deviation for conduct. For some outcomes, studies conduct￾ed in the 1980s yielded smaller effect sizes than earlier studies conducted in the 1960s and 1970s. Amato and Keith (1991) speculated that the gap in well-being between children with divorced and nondivorced parents might have narrowed either because divorce became more socially accepted or because parents were making greater efforts to re￾duce the potentially disruptive impact of divorce on their children. During the 1990s, the number of people touched by divorce increased, school-based pro￾grams for children of divorce became common, and mediation and education courses for divorcing parents became mandatory in many states (Emery, Kitzmann, & Waldron, 1999). Given these trends, one might expect studies conducted in the 1990s to reveal a continued closing of the gap in well￾being between children with divorced parents and children with married parents. An examination of studies conducted in the 1990s, however, does not support this hypothesis. A large number of studies in the 1990s contin￾ued to find that children with divorced parents score lower than children with continuously mar￾ried parents on measures of academic success (Astone & McLanahan, 1991; Teachman, Paasch, & Carver, 1996), conduct (Doherty & Needle, 1991; Simons and Associates, 1996), psycholog￾ical adjustment (Forehand, Neighbors, Devine, & Armistead, 1994; Kurdek, Fine, & Sinclair, 1994), self-concept (Wenk, Hardesty, Morgan, & Blair, 1994), social competence (Beaty, 1995; Brodzin￾sky, Hitt, & Smith, 1993), and long-term health (Tucker et al., 1997). Furthermore, effect sizes in the 1990s appear comparable to those of earlier decades. For example, across 32 studies of chil￾dren’s conduct published in the 1990s, the mean effect size was 2.19, which is not appreciably dif￾ferent from the mean value of 2.18 for studies conducted in the 1980s, as reported in Amato and Keith (1991). Similarly, across 29 studies of psy￾chological adjustment published in the 1990s, the mean effect size was 2.17, which is slightly larger than the mean value of 2.10 for studies conducted in the 1980s, also as reported in Amato and Keith. In general, the small but consistent gap in well￾being between children from divorced and two￾parent families observed in earlier decades per￾sisted into the 1990s. As with studies of adults, a few studies suggest that divorce also has positive consequences for some children. For example, a qualitative study by Arditti (1999) found that many offspring from divorced families, especially daughters, reported developing especially close relationships with their custodial mothers—a finding that is consis￾tent with some quantitative work (Amato & Booth, 1997). In addition, Amato, Loomis, and Booth (1995), Amato and Booth (1997), Hanson (1999), and Jekielek (1998) found that offspring were better off on a variety of outcomes if parents in high-conflict marriages divorced than if they remained married. When conflict between parents is intense, chronic, and overt, divorce represents an escape from an aversive home environment for children. Only a minority of divorces, however, appear to be preceded by a high level of chronic marital conflict (Amato & Booth, 1997). For this reason, divorce probably helps fewer children than it hurts. CAUSATION OR SELECTION? The selection perspective holds that differences between children from divorced and nondivorced families are due to factors other than marital dis￾ruption, including parents’ personality character￾istics, inept parenting, predivorce marital discord, or genetic influence. Consistent with a selection perspective, Capaldi and Patterson (1991) found that mothers’ antisocial personalities accounted for the association between mothers’ marital tran￾sitions and boys’ adjustment problems. In con￾trast, other studies found significant estimated ef￾fects of divorce even after controlling for aspects of parents’ personalities, including depression (Demo & Acock, 1996a) and antisocial personal￾ity traits (Simons and Associates, 1996). Longitudinal studies provide another type of evidence. Cherlin and colleagues (1991) found that children from maritally disrupted families had more postdivorce behavior problems than children from nondisrupted families. These differences, however, were apparent several years prior to di￾vorce, especially for boys. Amato and Booth (1996) found that problems in parent–child rela￾tionships (including parents’ reports that their children had given them more than the usual num￾ber of problems) were present as early as 8 to 12 years before divorce. Aseltine (1996) and Heth￾erington (1999) obtained comparable results with regard to children’s internalizing behavior, exter￾nalizing behavior, social competence, and self-es￾teem, and Doherty and Needle (1991) found com-