بدرفتاری با کودک و بزهکاری نوجوانان: بررسی نقش تعیین سطح و بی ثباتی جایگذاری

Abstract Children who experience maltreatment are at increased risk of engaging in delinquent behavior. Although little is known about the mechanisms responsible for this increased risk, the use of substitute care placement and placement instability are often identified as correlates. It is not clear from prior studies, however, whether delinquency precedes or follows placement instability. The current study adds significantly to the literature by identifying selected factors related to child maltreatment and delinquency and disentangling the timing of delinquency petitions relative to movements within the child welfare system. The results indicate that substantiated victims of maltreatment average 47% higher delinquency rates relative to children not indicated for abuse or neglect. In addition, approximately 16% of children placed into substitute care experience at least one delinquency petition compared to 7% of all maltreatment victims who are not removed from their family. Placement instability further increases the risk of delinquency for male foster children, but not for female foster children. Other characteristics related to delinquency include race, age, and recurrence of maltreatment.

1. Introduction Children who experience maltreatment are at increased risk of delinquency. Prior research indicates that anywhere from 9% to 29% of maltreated children engage in delinquent behavior (Kelly et al., 1997, Smith & Thornberry, 1995, Stewert et al., 2002, Widom, 1989, Widom, 2003 and Zingraff et al., 1993). Despite the consensus that maltreatment increases the risk of delinquency, there is little agreement about the mechanisms responsible for this increase. This is problematic for delinquency theorists, child welfare practitioners, and policy makers interested in the development of effective social interventions. In the current study, we (1) compare delinquency rates for maltreated and non-maltreated children in the city of Chicago and surrounding Cook County suburbs, (2) identify which victims of abuse and neglect are more likely to engage in delinquency, and (3) determine whether or not substitute care placement and placement instability mediates the experience of maltreatment and delinquency. 1.1. Child maltreatment In 1999, approximately 2.9 million children in the United States were the focus of a child protection investigation (U.S. Department of Health and Human Services, 2001). Of these 2.9 million children, 826,162 (28%) were associated with a substantiated report of maltreatment. Specifically, 58.4% were indicated as victims of neglect, 21.3% victims of physical abuse, and 11.3% victims of sexual abuse. Children aged 3 and under had the highest rate of substantiated maltreatment of any age group (13.9 children per thousand). African Americans had the highest rate of substantiated maltreatment of any racial group (25.2 children per thousand). Approximately 7.5% of indicated victims experienced a subsequent incident of abuse and/or neglect within 6 months of their initial substantiated report. 1.2. Social ecology of maltreatment Although child maltreatment, by definition, is an event occurring within the family or substitute care setting such as a foster home, day care center, or group home, the physical abuse and neglect of children is best understood as the manifestation of an unfolding sequence of underlying problems that often are initiated prior to the family's formation and could be located as well in community and cultural conditions (Pecora, Whittaker, Maluccio, & Barth, 2000). This broader ecological perspective shifts attention away from a narrow focus on parental psychopathology and family dysfunction and highlights the way that community and cultural conditions insinuate themselves in the development of the child, both inside the family and later on as the child moves into school, forms peer relationships, and matures into young adulthood (Testa & Furstenberg, 2002). Belsky's (1980) framework for understanding child maltreatment is one such perspective. Belsky's model is an adaptation of Bronfenbrenner's (1979) theory of human development and considers the psychological and sociological factors associated with parenting practices. Such parenting practices are believed to be multiply determined by parent, child and contextual factors. As many of the families involved within the child welfare system experience a range of psycho-social problems, understanding child maltreatment from an ecological perspective is important for both practitioners and policy makers because it enables one to identify the conditions that foster and maintain maladaptive behaviors. Once identified, one can tailor services and policies to meet the individual needs of families. At times, the provision of in-home services sufficiently meets the needs of families. Other times however, the provision of such services does not reduce the risk of harm and it's deemed necessary to place children in substitute care settings. Out-of-home placements are generally considered temporary with the eventual goal of family reunification. Despite the provision of both in-home and out-of-home services, maltreated children are at risk of experiencing a wide range of negative outcomes including delinquency. 1.3. Juvenile delinquency In 2000, approximately 2.4 million juveniles were arrested (Synder, 2002). Approximately 99,000 arrests were associated with a violent crime (e.g. murder, forcible rape, robbery, aggravated assault). Approximately 518,000 arrests were associated with a property crime (burglary, larceny-theft, motor vehicle theft, arson). The remaining arrests were associated with a “nonindex” crime (e.g. drug abuse violation, liquor laws, vandalism). Although declining in recent years, delinquency continues to be a major social problem. Thus, it is important to understand the etiology of these behaviors. The current study investigates whether substitute care placement and placement instability help to explain the initiation of delinquency for children in the child welfare system. We hypothesize that substitute care placement and especially the instability of such placements increases the risk of delinquency for maltreated children because it disrupts the social bonds that tie youth to significant caregivers and conventional institutions. Our hypotheses are informed by both theories of social capital and social control. 1.4. Social capital and social control Social capital refers to the durable relationships of commitment, trust, and obligation that bind people together and facilitate the giving and exchanging of resources (Bourdieu, 1985, Coleman, 1988 and Portes, 1998). The premise of social control theory is that greater social capital results in greater compliance and commitment from youth that helps prevent them from engaging in delinquency (Furstenberg & Hughes, 1995). Building on the work of Hirschi (1969), social control theorists posit that everyone shares a pre-socialized propensity to deviate from community norms. Healthy upbringing depends on parents' and other socializing agents' making consistent investments in the care, education, and supervision of children, which instill in them a sense of attachment, commitment, and obligation that tie them to the family and conventional role models. When confronted with opportunities to engage in nonconforming or undesirable behaviors, children with extensive and strong social ties will have a greater stake in conformity and will be disinclined to engage in delinquent behavior that might jeopardize those relationships. Child abuse and neglect are abnormal forms of investment and disinvestment in children's social capital, which undermine mutual relationships of commitment, trust, and obligation. Weakened relationships and ambivalent commitments to conventional behaviors help explain the variable levels of control that children exercise over anti-social impulses. When the bond with society in general is weak (Durkheim, 1951) and attachment to the family more specifically is insecure (Bowlby, 1988), individuals feel less bound by moral restraints and are less sensitive to the feelings and expectations of conventional role models. The greater the degree of social disconnection, the higher is the risk of participating in delinquent behavior (Hirschi, 1969 and Thornberry et al., 1991). 1.5. Instability and delinquency Prior research has established a relationship between disruptions and transitions within the family system and delinquency. Using data from three longitudinal studies (Rochester Youth Development Study, Denver Youth Survey, and Pittsburg Youth Study), researchers modeled the relationship between family transition and delinquency (Thornberry, Smith, Rivera, Huizinga, & Stouthamer-Loeber, 1999). The sample was comprised of 4000 youth and interviews were completed every 6 months with children between 13 and 18 years of age. Family disruption and transition was measured by comparing the structure of the family unit between interviews. The authors report that youth in urban areas were likely to experience at least one disruption or transition during adolescence. Approximately 45% of the sample experienced two or more changes in family units. These changes had a significant effect on the probability of delinquency. In Pittsburg, 90% of the youth with five or more disruptions or transitions reported engaging in delinquent behavior. The study of family instability and delinquency is not a recent development, nor is the conceptualization of disruption and transition limited to the residential or familial context. Prior research indicates that instability within the educational system is predictive of a variety of negative outcomes including delinquency (Reynolds, Chang, & Temple, 1998) and failure to complete high school (Rumberger & Larson, 1998). We anticipate finding similar problems associated with placement instability in foster care. The instability of foster care is often associated with a range of negative outcomes including child behavior problems, feelings of insecurity, and overall dissatisfaction with the foster care experience (Festinger, 1983, Kurtz et al., 1993 and Redding et al., 2000). In part, the problems associated with placement instability inspired the development of federal initiatives intended to increase family permanence for children who might otherwise would languish in long-term foster care. More relevant to the current study, the provision of child welfare services, the use of substitute care placement, and placement instability are often, but not always identified as predictors of involvement with juvenile corrections (English et al., 2000, Jonson-Reid & Barth, 2000a, Jonson-Reid & Barth, 2000b, Jonson-Reid & Barth, 2003 and Widom, 1991). These past studies have greatly advanced the understanding of the connections between maltreatment, substitute care, and delinquency. Yet, there are several notable limitations within this body of work. The current study builds on prior research and makes a new contribution by addressing the following limitations: (1) the samples used in prior research have not represented the full array of maltreated children, (2) the delinquency measures often fail to capture the full range of offenses, and (3) it is often not clear whether placement instability contributes to, or is a consequence of delinquency. 1.6. Sampling Prior studies of maltreatment and delinquency have not followed birth cohorts through 18 years of age. In general, prior samples have either excluded very young children (Jonson-Reid, 2002, Jonson-Reid & Barth, 2000a and Jonson-Reid & Barth, 2000b) or children in early adolescence (Widom, 1991; English, Widom, & Branford, 1999). Sampling strategies that exclude children under 5 years of age are problematic because such strategies exclude a substantial proportion of the maltreated population. Twenty-eight percent of the maltreated children are aged 3 and under. Another 25% of the victims are between the ages of twelve and seventeen. Thus, sampling strategies that only explore the maltreatment–delinquency connection for children under the age of 12 is also problematic. If the goal is to understand the overlap between the child welfare and juvenile justice populations, it is important that the samples are representative of all child welfare cases. The current study uses birth cohorts to address this issue. The cohorts represent all children with substantiated reports of maltreatment in the large and diverse area of Chicago and its surrounding Cook County suburbs. In addition to ensuring a representative sample, the advantage of analyzing birth cohorts is that the individual cases are exposed to the risk of both maltreatment and delinquency for the same period of time. If birth cohorts are selected far back enough so that the full 18 years of exposure to potential delinquency is observable, it is not necessary to censor cases for incomplete observations. Another challenge associated with sampling in maltreatment–delinquency research is the accurate representation of cases receiving in-home and out-of-home service. Only about one-third of the children with a substantiated report of maltreatment enter out-of-home care. The samples used to understand maltreatment and delinquency should reflect this distribution. In general, the samples used in prior research either do not reflect the contemporary trends of placement utilization (Grogan-Kaylor & Otis, 2003 and Widom, 1991) or are limited to only those children in substitute care (Jonson-Reid & Barth, 2000a, Jonson-Reid & Barth, 2000b and Runyan & Gould, 1985). These limitations introduce sample-selection effects that can bias statistical estimation. We attempt to address part of this issue by including all children with substantiated reports—regardless of whether they enter a substitute care placement. 1.7. Delinquency measure Juvenile delinquency encompasses a wide range of offenses. Yet some of the most recent and rigorous work in this area has focused solely on either the most serious (Jonson-Reid, 2002, Jonson-Reid & Barth, 2000a and Jonson-Reid & Barth, 2000b) or perhaps least serious (Jonson-Reid & Barth, 2003) offender. To capture the breadth of possible offending, the delinquency measure used in the current study includes the entire range of official petitions in the juvenile court. 1.8. Disentangling delinquency and placement It is critical to disentangle the timing of the delinquent offending relative to the timing of foster home placements. In part, the limitations associated with prior sampling strategies result from an inability to accurately identify the timing of delinquent offending. Limiting samples to children 11 years of age and younger reduced the likelihood that a delinquency arrest has yet to occur (Runyan & Gould, 1985 and Widom, 1991). The petition data used in the current study include the offense date. Thus, we are able to pinpoint whether delinquent offending precedes or follows placement instability. In summary, the goals of the present study are to (1) compare the delinquency rates for all maltreated and non-maltreated children, and (2) investigate the factors associated with delinquency for victims of child abuse and neglect. We examine the role of substitute care placement and the instability of such placements. We hypothesize that the risk of delinquency increases for children removed from the family home and for children repeatedly moved within the child welfare system.

Results The results are organized in four sections. First, delinquency rates for IDCFS involved and non-involved youth are displayed. Second, bivariate relationships are displayed between child demographics, maltreatment history, placement information, and delinquency. Third, we display the specific types of delinquency offenses. Finally, we present the results from several logistic regression analyses. The first set of regressions model the probability of delinquency for all maltreated children. The second set of regressions is limited to children in placement. These regression analyses focus particular attention on the relationship between placement, placement instability, and juvenile delinquency. Significant gender interactions emerged throughout the analyses. Thus, we present separate models for males and females. 3.1. Delinquency rates The delinquency rates for IDCFS involved and non-involved youth are displayed in Fig. 1. The trend for both groups indicates a steady decline between 1995 and 2000. The delinquency petition rates are consistently higher for IDCFS youth. These differences are greatest in 1998 and 1999 when the rate associated with IDCFS youth is 56% (4330 vs. 2780) and 57% (3739 vs. 2389) greater than the rate associated with non-IDCFS youth. Cook County delinquency petition rate, 1995–2000. Fig. 1. Cook County delinquency petition rate, 1995–2000. Figure options 3.2. Bivariate analysis The cross tabulations and chi-square statistics are presented for both males (see Table 2) and females (see Table 3). Consistent with risk in the general population, maltreated males are far more likely to engage in delinquency relative to maltreated females (14% vs. 4%). African American and Hispanic males are more likely to have a delinquency petition [16% and 11%, respectively, as compared with white males (7%)]. The delinquency rate for African American females (5%) is greater than the rate for both Hispanic (3%) and white (2%) females. Significant differences also emerge between the recurrence of maltreatment and delinquency. Males and females with three or more substantiated reports of maltreatment exhibit the highest risk of delinquency. The risk of delinquency does not appear related to the type of maltreatment. Children with at least one placement in substitute care were significantly more likely to have a delinquency petition compared with the children who had never entered a substitute care placement (23% vs. 11% for males, and 8% vs. 3% for females). Finally, placement instability appears to increase the risk of delinquency for males, but not females. Table 2. Cross-tabulations: child characteristics and delinquency, males (n=8910) Not delinquent, n (%) Delinquent, n (%) Overall 7666 (86) 1244 (14) Race African American 5149 (84) 1010 (16) Hispanic 901 (89) 109 (11) White 1616 (93) 125 (7) χ2=105.56, p<0.001 Type of substantiated maltreatment Physical abuse 3664 (85) 625 (15) Neglect 5146 (85) 886 (15) Emotional abuse 10 (71) 4 (29) Substance exposure 161 (92) 14 (8) Sexual abuse 521 (89) 65 (11) Maltreatment recurrence One substantiated report 5215 (88) 684 (12) Two substantiated reports 1390 (82) 297 (18) Three or more substantiated reports 1061 (80) 263 (20) χ2=84.54, p<0.001 Placement No placement 5875 (89) 697 (11) At least one placement 1791 (77) 547 (23) χ2=234.86, p<0.001 Placement instability (n=1994) No change in placement 574 (88) 77 (12) Two placement 323 (89) 42 (11) Three placements 202 (84) 39 (16) At least four placement 581 (79) 156 (21) χ2=28.84, p<0.001 Table options Table 3. Cross-tabulations: child characteristics and delinquency, females (n=9766) Not delinquent, n (%) Delinquent, n (%) Overall 9339 (96) 427 (4) Race African American 6279 (95) 348 (5) Hispanic 1287 (97) 36 (3) White 1773 (98) 43 (2) χ2=38.32, p<0.001 Type of substantiated maltreatment Physical abuse 4151 (95) 205 (5) Neglect 5443 (95) 275 (5) Emotional abuse 15 (88) 2 (12) Substance exposure 173 (99) 1 (1) Sexual abuse 2112 (96) 93 (4) Maltreatment recurrence One substantiated report 6415 (96) 242 (4) Two substantiated reports 1676 (96) 72 (4) Three or more substantiated reports 1248 (92) 113 (8) χ2=59.20, p<0.001 Placement No placement 7185 (97) 233 (3) At least one placement 2154 (92) 194 (8) χ2=111.88, p<0.001 Placement instability (n=2091) No change in placement 650 (94) 39 (6) Two placement 400 (94) 26 (6) Three placements 281 (97) 10 (3) At least four placement 640 (93) 45 (7) χ2=3.84, p>0.05 Table options 3.3. Type of delinquent offending There were 2515 individual charges associated with the 1671 offenders. The most common delinquency petitions were associated with a property (32%) and violent index (31%) offense. Property index offenses include arson, burglary, motor vehicle theft, vandalism, and larceny theft. Violent index offenses include robbery, sexual assault, aggravated assault and homicide. Only 11 (0.06%) of the 1671 offenders were charged with homicide. Other delinquency petitions include the unlawful use of a weapon (5%), possession and/or distribution of drugs (13%), and a variety of court order violations (19%). The court order violations include liquor violation, traffic violation, public order offense, disorderly conduct, and obstruction of justice. 3.4. Predicting delinquency for all maltreated children The full sample (n=18,676) was used to model the probability of engaging in delinquent behavior. The initial model included a variety of demographic (e.g. age at first maltreatment, gender, race) and social history (e.g. maltreatment, placement) variables. We also tested a variety of interaction terms. Significant interactions emerged between race and gender, and the history of maltreatment and gender. To better understand these interactions, we developed separate models for males (see Table 4) and females (see Table 5). The third column in each table, Exp(b), is the estimated odds ratio. This ratio estimates the change in the odds of membership in the target group (those with a delinquency petition) for a one unit increase in the predictor variable. White is the reference group for both African American and Hispanic. Table 4. Logistic regression: predicting delinquency for males (n=8910) Independent variables Model 1 Model 2 Model 3 b S.E. Exp(b) b S.E. Exp(b) b S.E. Exp(b) Child demographics Age at maltreatment 0.077*** 0.007 1.08 0.096*** 0.008 1.10 0.085*** 0.008 1.09 African American 0.931*** 0.099 2.54 0.869*** 0.101 2.38 0.776*** 0.102 2.17 Hispanic 0.378** 0.138 1.46 0.404** 0.139 1.49 0.410** 0.139 1.51 Maltreatment history Physical abuse −0.043 0.079 0.96 −0.111 0.080 0.89 Neglect 0.050 0.093 1.05 −0.058 0.094 0.94 Two substantiated reports 0.524*** 0.083 1.69 0.401*** 0.084 1.49 Three substantiated reports 0.814*** 0.094 2.58 0.576*** 0.099 1.78 Placement information Child placed 0.635*** 0.071 1.89 Model Chi-square (df) 234.69 (3)*** 345.38 (7)*** 424.92 (8)*** ** p<0.01. *** p<0.001. Table options Table 5. Logistic regression: predicting delinquency for all maltreated females (n=9766) Independent variables Model 1 Model 2 Model 3 b S.E. Exp(b) b S.E. Exp(b) b S.E. Exp(b) Child demographics Age at maltreatment 0.044*** 0.011 1.05 0.066*** 0.012 1.07 0.056*** 0.012 1.06 African American 0.828*** 0.164 2.29 0.783*** 0.165 2.19 0.676*** 0.167 1.97 Hispanic 0.064 0.230 1.07 0.103 0.230 1.11 0.110 0.231 1.12 Maltreatment history Physical abuse 0.006 0.116 1.01 −0.097 0.118 0.90 Neglect 0.104 0.132 1.11 −0.014 0.133 0.99 Two substantiated reports 0.123 0.144 1.13 −0.030 0.147 0.97 Three substantiated reports 0.950*** 0.141 2.59 0.663*** 0.149 1.94 Placement information Child placed 0.747*** 0.114 2.11 Model Chi-square (df) 57.98 (3)*** 116.91 (7)*** 158.81 (8)*** *** p<0.001. Table options The age of the child at maltreatment is a significant predictor for both males and females. Older children are more likely to engage in delinquency. For every additional year, the odds of delinquency increase by 1.09 for males and 1.06 for females. Both African Americans and Hispanics males are more likely to have a delinquency petition as compared with white males. Specifically, the odds of delinquency are 2.17 times greater for African American males than white males and 1.51 times greater for Hispanic males. Similarly, the odds of delinquency are 1.97 times greater for African American females as compared with white females. However, the risk of delinquency is not significantly different when comparing Hispanic and white females. The recurrence of maltreatment is associated with subsequent delinquency for both males and females. For males, two or more substantiated reports of maltreatment increase the risk of later delinquency. Males with two incidents of maltreatment are 1.49 times more likely, and males with three or more incidents of maltreatment are 1.78 times more likely to engage in delinquency as compared with males with only one substantiated report. For females, recurrence does not increase the risk of delinquency until it reaches a threshold of three or more substantiated incidents. The odds of delinquency are 1.94 times greater for females with three or more substantiated reports of maltreatment as compared with females with only one substantiated report. The odds of delinquency increase for children in substitute care placement. Males in placement are 1.89 times more likely to be delinquent as compared with the males who remain in the family home. Similarly, females in placement are 2.11 times more likely to be delinquent as compared with females who remain in the family home. 3.5. Predicting delinquency for children in placement The sample includes 4085 children with at least one placement in out-of-home care. Significant gender interactions emerged. Again, we developed separate regression models for males (see Table 6) and females (see Table 7). Table 6. Logistic regression: predicting delinquency for males in placement (n=1994) Independent variables Model 1 Model 2 Model 3 b S.E. Exp(b) b S.E. Exp(b) b S.E. Exp(b) Child demographics Age at placement 0.064** 0.021 1.07 0.069** 0.022 1.07 0.074*** 0.022 1.07 African American 0.696** 0.239 2.00 0.686** 0.240 1.98 0.681** 0.241 1.98 Hispanic 0.158 0.373 1.17 0.150 0.374 1.16 0.093 0.376 1.10 Maltreatment history Physical abuse −0.223 0.149 0.800 −0.323* 0.152 0.724 Neglect −0.172 0.199 0.842 −0.234 0.201 0.792 Two substantiated reports 0.219 0.161 1.24 0.150 0.164 1.16 Three substantiated reports 0.181 0.170 1.19 0.063 0.174 1.07 Placement instability Two movements −0.014 0.206 0.987 Three movements 0.428* 0.216 1.53 At least four movements 0.754*** 0.157 2.13 Model Chi-square (df) 22.39 (3)*** 25.64 (7)*** 56.44 (10)*** * p<0.05. ** p<0.01. *** p<0.001. Table options Table 7. Logistic regression: predicting delinquency for females in placement (n=2091) Independent variables Model 1 Model 2 Model 3 b S.E. Exp(b) b S.E. Exp(b) b S.E. Exp(b) Child demographics Age at placement −0.017 0.032 0.98 0.016 0.034 1.02 0.018 0.034 1.02 African American 0.760 0.397 2.14 0.766 0.399 2.15 0.782 0.400 2.19 Hispanic 0.269 0.566 1.30 0.406 0.569 1.50 0.420 0.570 1.52 Maltreatment history Physical abuse −0.224 0.217 0.799 −0.238 0.219 0.788 Neglect 0.344 0.313 1.41 0.340 0.312 1.40 Two substantiated reports 0.006 0.278 1.00 0.002 0.279 1.00 Three substantiated reports 0.697** 0.250 2.01 0.694** 0.253 2.00 Placement instability Two movements 0.072 0.267 1.07 Three movements −0.559 0.367 0.572 At least four movements 0.098 0.236 1.10 Model Chi-square (df) 5.89 (3) 22.39 (7)** 26.45 (10)** ** p<0.01. Table options The logistic regression model for females is noticeably different from the model for males. Only one independent variable is statistically different from zero for females in placement. Specifically, the odds of delinquency are two times greater for females with three or more substantiated reports of maltreatment as compared with females with only one substantiated report. Age at first placement is not associated with subsequent delinquency and type of maltreatment and placement instability do not exceed the threshold of statistical significance. Race (specifically African American) is marginally significant (p value <0.052). In contrast, five of the independent variables for males are significantly related to delinquency. Age at first placement increases the risk of delinquency. Children placed at an older age are more likely to engage in delinquency. African Americans in placement are also more likely to have a delinquency petition. The odds of delinquency are 1.98 times greater for African American males as compared with white males. The risk of delinquency is not different when comparing Hispanic and white youth. The type of maltreatment is also related to delinquency. Children who experience physical abuse are less likely to have a delinquency petition. Finally, placement instability increases the risk of delinquency for males in substitute care placement. The odds of delinquency are 1.54 times greater for males with three placements, and 2.13 times greater for males with four or more placements as compared with males with only one placement. The recurrence of maltreatment is not related to subsequent delinquency for males. As separate logistic models were developed for males and females, it can be difficult to grasp the differences in risk across gender. To illustrate these differences, as well as the effects of race and placement instability, we translate the logits into probabilities. We calculate and compare the probability of delinquency for white and African American children, with a history of neglect, and who have entered placement at 14 years of age. The placement instability variable is manipulated to illustrate its effect on delinquency. We compare those with one placement to those with four placements. Using the coefficients in Table 7, the equation for females is logit(delinquency)=0.018(age)+0.782(race)+0.340 (neglect)+0.098 (instability)−4.09. Using the coefficients in Table 6, the equation for males is logit(delinquency)=0.074(age)+0.681(race)−0.234(neglect)+0.754(instability)−2.95. Race is equivalent to zero for whites. Instability is equivalent to zero for cases with only one placement. To calculate the probability, we use elogit(delinquent)/(1+e−logit(delinquent)). The probabilities are displayed in Table 8. Three findings are important to note. Females have much lower probabilities of delinquency, and these probabilities do not change when foster placements are unstable. In contrast, the probability of delinquency for males increases significantly when foster care placements are unstable. Moreover, the probability of delinquency for African American males is especially high. The relative frequency of delinquency is 33% for African American males with four placements. Table 8. Comparing probability of delinquency by gender, race, and placement stability Case characteristics Probability White female, neglect, 14 years old, 1 out-of-home placement 0.03 White female, neglect, 14 years old, 4 out-of-home placements 0.03 African American female, neglect, 14 years old, 1 out-of-home placement 0.06 African American female, neglect, 14 years old, 4 out-of-home placements 0.07 White male, neglect, 14 years old, 1 out-of-home placement 0.10 White male, neglect, 14 years old, 4 out-of-home placements 0.20 African American male, neglect, 14 years old, 1 out-of-home placement 0.19 African American male, neglect, 14 years old, 4 out-of-home placements 0.33