The Social Science Journal 42 (2005) 241–256
Termination of parental rights: Which foster care children are affected? Kelly Noonan a,∗ , Kathleen Burke b a
Department of Economics, Rider University, 2083 Lawrenceville Road, Lawrenceville, NJ 08648, USA b SUNY Cortland, Cortland, NY, USA
Abstract In 1997, the Adoption and Safe Families Act (ASFA) was passed with a primary goal of expediting the process of placing foster children with permanent or adoptive families. In order to meet this goal, ASFA requires states to terminate parental rights if a child has been in foster care for 15 of the most recent 22 months. Prior empirical research on foster care dependence supports the provision in ASFA to expedite the discharge process because over time children are progressively less likely to be discharged from foster care. However, very little research has examined what impact terminating parental rights will have on this goal. One of the first steps is to examine which children are most likely to see the rights of their parents terminated and how these children differ from those children who are returned home. Using a competing risks hazard model we find many differences between the children who are sent home and those children whose parents have their rights terminated. © 2005 Elsevier Inc. All rights reserved.
1. Introduction Approximately 270,000 children were in the foster care system at the end of 1985. By the end of 1996, the number of children in foster care had nearly doubled to approximately 502,000 (U.S. House of Representatives, Committee on Ways and Means, 1998). This dramatic growth prompted the federal government to enact the Adoption and Safe Families Act of 1997 (ASFA). One of the primary goals of ASFA is to expedite the process of permanently placing foster ∗ Corresponding author. Tel.: +1 609 895 5539. E-mail address:
[email protected] (K. Noonan).
0362-3319/$ – see front matter © 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.soscij.2005.03.012
242
K. Noonan, K. Burke / The Social Science Journal 42 (2005) 241–256
children. One provision in ASFA that supports this goal requires states to terminate parental rights if a child has been in care for 15 of the most recent 22 months (Christian, 1999). The provisions in ASFA to expedite the discharge process are supported by the results of empirical research on foster care dependence which established that the probability of leaving the system decreases as the time spent in foster care increases (Benedict, White, & Stallings, 1987; Courtney, 1994; Fanshel, 1971; Goerge, 1990; Jenkins, 1967; Noonan, 2000). However, this research did not specifically address those children who were adopted from the foster care system or those children whose parents’ rights were terminated. Courtney and Wong (1996) analyze children who are adopted from the foster care system. They find that the probability of leaving foster care increases over time for these children. While they did not explore the probability of parental rights termination, they hypothesize that this probability will mimic the probability of adoption. Our paper tests this hypothesis and simultaneously examines the probability that a child will be returned home to his or her biological parents. Using a data set from the New Jersey foster care system, we employ a competing risks hazard model that allows children to either be reunified with their parents or have the rights of their parents terminated. We find many differences between the children who are reunified and the children who experience parental rights termination. In particular, the characteristics of the child are very influential in determining the amount of time spent in care before parental rights are terminated. It appears that parental rights are only terminated for those children who are most likely to be adopted which supports the hypothesis of Courtney and Wong (1996). If parental rights are not terminated until a potential adoptive family has been identified, terminating parental rights will not expedite the discharge process.
2. Literature review The policy of expediting the discharge process is corroborated by the research on a child’s initial length of stay in the foster care system. Thus, we begin by highlighting some of the literature on the initial length of stay in the foster care. We then discuss the limited literature regarding the termination of parental rights. 2.1. Initial length of stay Many studies on the initial length of stay find that children are less likely to exit the foster care system as their length of stay in care increases (Benedict et al., 1987; Courtney, 1994; Fanshel, 1971; Goerge, 1990; Jenkins, 1967). This phenomenon is known as negative duration dependence and it means that children are progressively less likely to leave care as their time in care increases. In addition, these studies find that children are very likely to become childhood dependents on the foster care system if they remain in care for more than two consecutive years. That is, a child who is not discharged quickly enough from the system is likely to spend his or her entire childhood in the foster care system. While all of the studies mentioned on the initial length of stay find a declining probability of exiting care, the methodology used has changed dramatically over time. One of the earliest
K. Noonan, K. Burke / The Social Science Journal 42 (2005) 241–256
243
studies, Jenkins (1967), simply calculates the rates at which children exit care. While she finds that the rate at which children exit care declines over time, she does not account for any observable characteristics. All of the subsequent studies extend the work of Jenkins (1967) by using multivariate analysis to control for observable characteristics. Beginning with Fanshel (1971), who uses multiple correlation analysis, the probability of discharge and the initial length of stay are explored in more detail. Fanshel’s (1971) modeling techniques, however, do not account for children whose spells in foster care are censored. Benedict et al. (1987) and Courtney (1994) both make use of a Cox proportional hazard model to examine the initial length of stay. This model is an excellent choice because it allows for both multivariate analysis and censoring of the initial length of stay. Finally, Goerge (1990) also uses a hazard model and allows for unobserved heterogeneity. Regardless of the methodology used, all of the studies discussed consistently find that foster care spells exhibit negative duration dependence. Courtney and Wong (1996) find an exception to this phenomenon. They extend the prior research by allowing for multiple routes of exit from foster care. While they do observe that children are less likely to be returned to their family or guardian over time, they also find that the probability of adoption increases over time. Since parental rights must be terminated prior to adoption, they hypothesize that the timing of parental rights termination will be very similar to the timing of adoption. As stated earlier, this will be the case if parental rights are not terminated until potential adoptive parents are available for a child. In addition to the findings on the how the probability of discharge changes over time, prior research has also examined the influence of other factors on the initial length of stay. For example, the initial length of stay in foster care is significantly longer for children who enter the foster care system because they have a drug abuse problem, a disability, or a behavioral problem (Benedict & White, 1991; Jenkins, 1967; Lawder, Poulin, & Andrews, 1986; Seaburg & Tolley, 1986). The length of stay is also longer for children who are placed in foster care with a relative (Benedict & White, 1991; Goerge, 1990; Wulczyn & Goerge, 1992). While long-term dependence on the foster care system is often seen as a negative outcome, for those children placed in stable homes with relatives long-term dependence is not a bad outcome. 2.2. Termination of parental rights The Adoption Assistance and Child Welfare Act of 1980 was established to provide services in order to reduce the probability that a child would enter the foster care system. Once a child was placed in foster care, the primary goal was to reunify children with their biological parents (Glisson, Bailey, & Post, 2000). Unfortunately, as discussed earlier, the number of children in foster care dramatically increased between 1985 and 1996. In part, this was due to many children spending years in the foster care system. Prior to the passage of the Adoption and Safe Families Act of 1997 (ASFA), there was much debate over the rights of parents versus the rights of the children. ASFA was enacted in order to help protect the rights of children by expediting the discharge process. As stated earlier, one way the law does this is by requiring parental rights to be terminated if the child has spent 15
244
K. Noonan, K. Burke / The Social Science Journal 42 (2005) 241–256
of the most recent 22 months in care (Christian, 1999). It also includes certain guidelines that allow for the immediate termination of parental rights. These guidelines include cases where the parent subjects the child to “aggravated circumstances” including sexual abuse, physical abuse or abandonment (Freundlich, 1999). To remain eligible for federal funding, states must adhere to these new guidelines set for the termination of parental rights. However, they may set even stricter guidelines. For instance, in the state of New Jersey parental rights can be terminated if a child spends 12 months in foster care. The termination of parental rights places a large economic burden on the government by turning the child into a “legal orphan” who must then be supported by the foster care system and/or other governmental transfer payments until the child is either adopted or emancipated from the system. Surprisingly, to the best of our knowledge, very few analyses on the termination of parental rights have been completed. Guggenheim (1995) conducted one study on the termination of parental rights using multiple years of data to determine how the number of “legal orphans” changed through the years. He finds the number of cases where parental rights are terminated increased over time. However, these results should be viewed very cautiously because he does not control for the rising number of children in the foster care system. A number of studies have also analyzed which children are most likely to experience termination of parental rights. Schetky, Angell, Morrison, and Sack (1979) and Jellinek, Murphy, Poitrast, Quinn, and Bishop (1992) find that in most cases parental rights are terminated due to parental neglect rather than abuse. In contrast, Wattenberg, Kelley, and Kim (2001) find that the mother’s substance abuse led to the termination of rights in over half of the cases they examine. All three of these studies show that the children who experience parental rights termination come from poor families where parents are involved in high-risk behavior and have a history of problems, both physical and mental. Furthermore, the parents are poorly educated and not regularly employed (Borgman, 1981). In the studies discussed, multivariate analysis was not used to examine the reasons behind the termination of parental rights, which is one of the contributions of this paper. Furthermore, the majority of these studies only examined the children and parents who have already experienced the termination of parental rights (Borgman, 1981; Schetky et al., 1979; Wattenberg et al., 2001). Therefore, it cannot be determined if these findings are unique to children who experience parental rights termination or if foster children are more likely to have these characteristics.
3. Purpose The main goal of this paper is to examine two alternative outcomes: termination of parental rights and reunification. We examine how the probability of terminating parental rights changes over time and compare the probability of parental rights termination to the probability of reunification. In order to better inform policy makers, we aim to achieve a better understanding of the timing of these events. Following the hypothesis of Courtney and Wong (1996), we expect to find that the probability of terminating parental rights will increase over time. In contrast, the probability of discharge is expected to decrease over time.
K. Noonan, K. Burke / The Social Science Journal 42 (2005) 241–256
245
If parental rights are not terminated until an adoptive family is identified and if adoptability factors into the termination of parental rights decision, we expect to find the characteristics of the child to be very influential in determining the amount of time spent in care until termination of parental rights occurs. More specifically, we expect African American children, children of Hispanic decent, older children, and disabled children to have a lower probability of parental rights termination because they have less chance of being adopted.
4. Data The data that we use for our analysis stem from a federally mandated data collection program that began in 1994 requiring states to report on all children in their child welfare agencies. This resulted in a national data set known as the Adoption and Foster Care Analysis and Reporting System (AFCARS). AFCARS is one of the most extensive data sets that has ever been available for the analysis of foster care. For the purpose of this paper we use the AFCARS data set for New Jersey from the fiscal year 1998. With these data, we examine which children in their first spell of foster care are most likely to see the rights of their parents terminated and which children are most likely to be discharged to their biological parents. 4.1. Descriptive statistics As shown by Table 1, children enter the foster care system for a variety of reasons. The first column of the table lists all of the possible reasons for entry into the foster care system. For Table 1 Frequency of reasons for entry Percentagea
Standard deviation
Child Reasons Child disability Child behavior Alcohol abuse Drug abuse
16.40 13.60 0.38 51.55
0.3703 0.3428 0.0612 0.4998
Parent Reasons Physical abuse Sexual abuse Neglect Alcohol abuse Drug abuse Parents incarcerated Can’t cope Abandonment Relinquishment
47.73 44.65 58.72 5.82 27.64 1.81 76.26 2.66 11.86
0.4995 0.4971 0.4924 0.2341 0.4472 0.1332 0.4255 0.1610 0.3223
Other Reasons Parent death Inadequate housing
1.83 20.54
0.1340 0.4040
N
9571 a
Since children can enter care for multiple reasons, the column does not sum to 100.
246
K. Noonan, K. Burke / The Social Science Journal 42 (2005) 241–256
each child who enters the care system, there is at least one reported reason for entry. However, multiple entry reasons may be reported. For ease of exposition, these reasons are categorized into three groups: Child Reasons, Parent Reasons and Other Reasons. Included in the Child Reasons for entry are children who enter because of a disability, a child behavior problem, alcohol abuse or drug abuse. It is important to note that those children who enter care because they are born addicted to an illegal drug are considered drug abusers. The Parent Reasons for entry into foster care include: physical abuse, sexual abuse, neglect, alcohol abuse, drug abuse, a parent’s incarceration, can’t cope, abandonment and relinquishment. The Other Reasons for entry are parental death and inadequate housing. The most common reasons for entry, as shown by Table 1, are Can’t Cope, Neglect, Child Drug Abuse, Physical Abuse and Sexual Abuse. With the exception of Child Drug Abuse, the most common reasons for entry are all parental reasons. This indicates that children tend to enter the care system because of “bad” parenting not because they are “bad” children. Inclusion of the reasons for entry in our estimation allows for children who are placed in care for different reasons to experience different outcomes. We expect the Child Reasons for entry to decrease the probability that parental rights will be terminated because these children will have less chance of adoption. Alternately, for those parents who relinquish their rights, we expect that parental rights will be terminated very quickly. Finally, for those children who are physically or sexually abused the probability of being discharged to the parents should be lower. The parent(s) would have to undergo treatment and/or incarceration before the child would be returned. Once a child is in the foster care system there are many possible ways that a child can leave or be discharged from the system. However, many of these discharge alternatives, such as adoption, are not relevant until after the termination of parental rights. Therefore, in this paper we only consider two possible outcomes for the child: (1) the child can be discharged to the biological parents or (2) the parental rights to the child can be terminated. Table 2 shows these two potential outcomes, the percentage of children realizing each outcome and the average time a child spends in care until one of these outcomes is realized.1 As expected, children who are sent home to their parents have shorter spells in care (1 year) than those who experience parental rights termination (2.8 years). It is important to consider a few details that may not be evident from this table. First, while children who are discharged to their biological parents have the shortest spells in care, they are also the group that is most likely to re-enter care (Noonan, 2000). Second, the time until the termination of parental rights is not the end of the child’s time in care. Rather, it is the point in time that may open the door to many alternative discharge options, such as adoption. Finally, many children remain in care, 50.07%, who have neither had the rights of their parents terminated nor been reunified with their parents. Table 2 Average time until each child outcome Discharged to parents Mothers rights terminated
N
Percentage
Average time (years)
2132 2647
22.28% (0.4161) 27.65% (0.4473)
1.03 (1.6010) 2.80 (2.0510)
Standard errors are in parenthesis. Source: New Jersey AFCARS fiscal year 1998.
K. Noonan, K. Burke / The Social Science Journal 42 (2005) 241–256
247
Table 3 Means and standard deviations of other variables Mean
S.D.
Child Characteristics Male African American Hispa decent Diagnosed disability Number of placements Age at first entry
0.5306 0.6170 0.0942 0.2039 2.3003 5.7075
0.4991 0.4861 0.2922 0.4030 2.1690 5.5315
Case Goal Reunify Kin Adoption Foster care Emancipation Goal not yet established
0.5529 0.0701 0.3709 0.0370 0.0145 0.0176
0.4972 0.2553 0.4617 0.1887 0.1196 0.1313
Placement Setting Pre-adoptive home Kin Foster family home Group home Institution Supervised independent living
0.0276 0.0249 0.0712 0.0692 0.1668 0.0105
0.1638 0.1557 0.4577 0.2538 0.3728 0.1017
Foster Family Structure Married couple Unmarried couple Single female Single male Not applicable
0.3189 0.0648 0.3341 0.0092 0.2730
0.4661 0.2461 0.4717 0.0955 0.4455
Caretaker Family Structure Married couple Unmarried couple Single female Single male Unknown
0.1615 0.0109 0.6341 0.0320 0.1622
0.3675 0.1037 0.4817 0.1759 0.3686
0.2978 0.1504 0.7492 0.0381 0.2258 1161.22
0.4573 0.3574 0.4335 0.1915 0.4181 2166.61
Economic Variables Foster Care System Funding Child Support Funding Medicaid Funding SSI Funding Other Funding Source Monthly Foster Care Payment N
9571
Source: New Jersey AFCARS fiscal year 1998.
Means and standard deviations for the remaining variables included in our estimation are presented in Table 3. We break the remaining variables into six categories: Child Characteristics, Case Goal, Placement Setting, Foster Family Structure, Caretaker Family Structure and Economic Variables.
248
K. Noonan, K. Burke / The Social Science Journal 42 (2005) 241–256
Included among the Child Characteristics are the child’s gender, ethnicity, the presence of a diagnosed disability, the number of placements and the age at first placement (age when first entering foster care). When examining the Child Characteristics, the over-representation of African Americans is very evident. While the United States Census Bureau (2001) estimates that only 14.6% of the population in New Jersey is African American, approximately 61.70% of the study sample is African American. This signifies a correlation between foster care participation and lower socio-economic status. We consider children in their first spells of care, however, during any given spell a child may endure many different foster care placements. As indicated by Table 3, the average child has experienced more than two different placements. The number of placements is included to act as a proxy for a behavioral problem that is evident after entry into the foster care system. An important clarification needs to be made between the two variables “child disability” (listed in Table 1) and “diagnosed disability” (listed in Table 3). Child disability indicates the child enters the care system due to a disability. Diagnosed disability indicates the child is clinically diagnosed as suffering from a disability. Since the diagnosis of a disability can occur at any time, before, during or after entry into foster care it is more common for a child to be clinically diagnosed with a disability (20.39%) than it is for a child to enter care due to a disability (16.40%). As discussed above, we expect these child characteristics to significantly impact the timing of parental rights termination. More specifically, if adoptability is a criterion for termination of parental rights, we hypothesize that African American children, children of Hispanic decent, disabled children and older children are less likely to experience parental rights termination. Similarly, we expect the probability that parental rights are terminated to decrease with the number of foster care placements. The variables in the Case Goal category include reunification, placement with kin, adoption, foster care, emancipation, and goal not yet established. The two most common goals set by social workers are reunification (55.29%) and adoption (30.79%). The results may seem surprising given the current policy environment, however, most social workers initially attempt to reunify the child with his or her biological parent(s) unless the child is in imminent danger. For those children with a case goal of reunification we expect to see a lower probability of parental rights termination and a higher probability of discharge since the social workers will be working toward discharging the child to his or her parents. Furthermore, for those children with the case goal of adoption we expect a higher probability of parental rights termination. The options for a child’s Placement Setting include pre-adoptive home, kin, foster family home, group home, institution, or supervised independent living. The most common placement setting for a child in foster care is a foster home (70.17%). We expect to find that children placed in pre-adoptive homes are less likely to be discharged to their parent(s) and are more likely to experience the termination of parental rights so the adoption process may begin. Moreover, if a relative’s home is a stable environment, we expect children placed with relatives are less likely to experience the termination of parental rights and less likely to be discharged to their parents. For those children who are placed with a foster family there are many different living situations listed under Foster Family Structure. These family structures include a married couple, an unmarried couple, a single female, a single male, or not applicable for those children
K. Noonan, K. Burke / The Social Science Journal 42 (2005) 241–256
249
not living with a foster family. Most children placed with foster care families are with a married couple or with a single female. Nationally, according to the U.S. Census Bureau (2000), households headed by single females are the most likely to be in poverty (27.8%) and therefore, these households may not have the necessary resources to maintain custody. As such, we expect the probability of discharge to be higher for children placed in the foster home of a single female. The Caretaker Family Structure refers to the living arrangement of the child prior to entry into foster care. These family structures include a married couple, an unmarried couple, a single female, a single male, or not applicable. Typically, this family is the biological family, however, this is not always the case. The majority of children placed in the foster care system come from families with only a single female adult (63.41%). As discussed above, it is much more likely that female-headed households are in poverty. Therefore, these households may not have the necessary resources to regain custody, which will lower the probability of discharge for these children. The Economic Variables considered in this paper include the sources of funding for the foster children (Foster Care System Funding, Child Support Funding, Medicaid Funding, SSI Funding, and other funding sources) and the total amount of funding received on a monthly basis. Many children who are in the foster care system are eligible for state and/or federal aid. This aid comes from various sources such as child Support and Medicaid. The average child receives $1,161.22 per month in funding. Since children who receive any financial resources should be easier to place in the foster care system, we expect the probability of discharge to be lower for these children.
5. Competing risks hazard model As discussed earlier, each child in the foster care system is at “risk” for two possible outcomes: reunification or the termination of parental rights. These two outcomes are “competing” because while in the foster care system a child may experience either one of the two outcomes. However, once one of the outcomes has been realized the other outcome can no longer occur. Therefore, in order to determine the “risk” of each outcome, we use a competing risks hazard model. While the name of this statistical model may have some negative connotation, it is purely semantic and does not imply any negativity about a child’s outcome. This type of model is an excellent choice because it allows for competing outcomes and it offers many additional benefits. First, for those children who are currently in a spell of foster care, the length of stay is a right-censored variable. Due to the censoring, ordinary least squares estimation will result in biased predictions. Hazard models or spell level analysis easily incorporate censored variables. Second, the probability of leaving the foster care system changes over time. Other models that examine the probability of an occurrence, such as logit and probit models, do not allow the probability to change over time. Before a hazard model can be utilized, some notation must be defined. While all of the variables discussed are for a particular child, for notational simplicity the child subscripts will be suppressed until they are needed. Let X be a vector of observable characteristics. The hazard rate for exiting care because of reunification, λe (s|X) is the rate at which children exit a spell of
250
K. Noonan, K. Burke / The Social Science Journal 42 (2005) 241–256
foster care to be reunified with their parents on the sth day conditional on being in care on the sth day. Alternately, the hazard rate for the termination of parental rights is given by λt (s|X). Thus, the overall probability of either reunification or the termination of parental rights is given by λcr (s|X) = λe (s|X) + λt (s|X) where it is assumed that these two alternatives are independent. In order to estimate this hazard model, a functional form must be assumed. When estimating a hazard model, choosing the appropriate functional form is critical because the estimation results may depend on the specification choice. For the purpose of this paper, five different distributions were considered; exponential, logit, normal, gamma, and Weibull. Based on the Akaike Information Criteria (AIC) the Weibull distribution best fit the data (Stata Reference Manual, 1999). Therefore, only the results from the Weibull model are discussed. The hazard function for the Weibull model can be written as λm (s|X) = exp{Xβm }
1 (1/σm )−1 s σm
where σ m is a parameter to be estimated. The shape of the hazard function will depend on the value of σ m . If σ m < 1 the hazard increases with time in care, if σ m > 1 the hazard decreases with time in care, and if σ m = 1 the hazard is constant.
6. Empirical results Before analyzing the coefficient estimates of the competing risks model, we present a graph of the estimated hazard functions in Section 6.1. Section 6.2 discusses the coefficient estimates in order to compare those children who experience either reunification or parental rights termination. 6.1. Graphical hazard analysis Using the coefficient estimates in Table 4, the two hazard functions estimated in this competing risks model are graphed in Fig. 1. These hazard functions are for an African American male child who entered care at the mean age due to a drug abuse problem because his mother could not cope. His case goal is reunification and he is currently placed with a single female in a foster home and has experienced the mean number of foster care placements. He receives Medicaid and the mean amount of foster care funding. As clearly shown, the hazard functions for reunification and the termination of parental rights are very different. The hazard for reunification decreases over time because the parameter σ e is greater than 1. Since the probability of reunification decreases over time, children are less likely to be reunified with their parents as time in care increases. This finding is consistent with the previous studies
K. Noonan, K. Burke / The Social Science Journal 42 (2005) 241–256
251
Table 4 Results for competing risks hazard model Discharge
Parental rights terminated
Coefficients
Hazard ratio
Coefficients
Hazard ratio
Child Characteristics Male African American Hispanic decent Diagnosed disability Age at first entry Age at first entry squared Number of placements
0.0342 (0.0442) −0.0684 (0.0518) −0.1414b (0.0800) −0.3241a (0.1595) −0.0100 (0.0167) 0.0022a (0.0010) −0.1421a (0.0155)
1.0348 0.9339 0.8681 0.7232 0.9900 1.0022 0.8675
0.0198 (0.0377) −0.1418a (0.0546) −0.2210a (0.0780) 0.3590a (0.0753) −0.0932a (0.0166) 0.0062a (0.0013) −0.0689a (0.0094)
1.0200 0.8678 0.8017 1.4319 0.9110 1.0062 0.9334
Child Reasons Child disability Child behavior Alcohol abuse Drug abuse
0.2585 (0.1743) 0.1711a (0.0792) −0.1376 (0.2887) −0.0097 (0.0790)
1.2950 1.1866 0.8715 0.9903
−0.3444a (0.0859) −0.0269 (0.0769) −0.4184 (0.6837) −0.2483a (0.0819)
0.7087 0.9735 0.6581 0.7801
Parental Reasons Physical abuse Sexual abuse Neglect Alcohol abuse Drug abuse Parents incarcerated Can’t cope Abandonment Relinquishment
−0.1297a (0.0622) −0.0652 (0.0776) 0.1974a (0.0664) 0.1869a (0.0792) 0.0979b (0.0548) −0.2660 (0.1929) 0.1421a (0.0614) 0.2198 (0.1537) −2.5237a (0.3580)
0.8784 0.9369 1.2183 1.2056 1.1029 0.7664 1.1527 1.2459 0.0802
−0.0543 (0.0533) −0.0374 (0.0800) 0.1060 (0.0746) 0.1472 (0.1029) 0.2496a (0.0461) −0.0269 (0.1032) 0.0758 (0.0577) 0.0842 (0.0963) 0.3664a (0.0487)
0.9472 0.9633 1.1119 1.1586 1.2835 0.9735 1.0788 1.0879 1.4425
Other Reasons Parental death Inadequate housing
−0.2247 (0.2100) −0.1317a (0.0557)
0.7988 0.8766
0.6177a (0.1156) −0.1476a (0.0478)
1.8546 0.8628
Case Goal Reunify Adoption Foster care Kin Goal not yet established
0.6072a (0.1630) −1.6596a (0.02298) −1.0392a (0.2261) 0.2261 (0.1743) 0.4481a (0.1824)
1.8353 0.1902 0.3537 1.2537 1.5654
0.3923a (0.1936) 1.0488a (0.1945) −0.2908 (0.2132) −1.0161a (0.2486) −1.1019a (0.5334)
1.4804 2.8542 0.7477 0.3620 0.3322
0.0000 0.2893 1.0531 0.4442
0.6902a (0.0780) −0.3025 (0.2429) 0.2357b (0.1313) −0.3784 (0.3343)
1.9941 0.7390 1.2658 0.6849
0.3955 (0.0797) 0.0629 (0.1218) 0.0559 (0.0823 0.0115 (0.2732)
1.4851 1.0649 1.0575 1.0116
0.0328 (0.1002) 0.0471 (0.1127) −0.1158 (0.0992) −0.3143b (0.1629)
1.0333 1.0482 0.8906 0.7303
0.1012 (0.0756) 0.3962 (0.2947) 0.0778 (0.0613) −0.0076 (0.1317)
1.1065 1.4862 1.0809 0.9924
0.1731a (0.0741) 0.1431 (0.1737) 0.1823a (0.0609) 0.1528 (0.1166)
1.1890 1.1538 1.2000 1.1651
Placement Setting Pre-adoption Kin Group home Supervised independent living Foster Family Structure Married couple Unmarried couple Single female Single male Caretaker Family Structure Married couple Unmarried couple Single female Single male
−10.7447a (0.2243) −1.2403a (0.1449) 0.0518 (0.0996) −0.8116 (0.1968)
252
K. Noonan, K. Burke / The Social Science Journal 42 (2005) 241–256
Table 4 (Continued) Discharge
Economic Variables Foster Care System Support Child Support Funding Medicaid Funding SSI Funding Other Source of Funding Monthly Funding Duration Dependence σ Constant Log likelihood N a b
Parental rights terminated
Coefficients
Hazard ratio
Coefficients
Hazard ratio
−3.4743a (0.2923) −0.0697 (0.0911) −1.5419a (0.0986) −0.6095a (0.1901) 0.1970b (0.1095) 0.0000 (0.0000)
0.0310 0.9326 0.2140 0.5436 1.2177 1.0000
−0.0250 (0.0500) 0.2768a (0.0581) 0.7463a (0.2200) −0.0577 (0.1002) 0.5757 (0.2252) −0.0000a (0.0000)
0.9753 1.3189 2.1092 0.9440 1.7784 1.0000
1.3068 (0.0234) −5.2643 (0.2572) a
0.6361 (0.0118) −13.1193a (0.4050) −10336.74 9571
|t| > 1.96. |t| > 1.65.
on the length of stay in foster care, that find children are progressively less likely to leave care as time in care increases. In contrast, the hazard for the termination of parental rights increases over time. In this case, the parameter σ t is significantly less than 1. As the time in care increases, children are more likely to see the rights of their parents terminated. This finding is consistent with a law that mandates the termination of parental rights for children who are in care for extended periods of time. In addition, it shows that the risk associated with the termination of parental rights is similar to the risk associated with adoption as hypothesized by Courtney and Wong (1996).
Fig. 1. Hazard rates for the Competing Hazard Model.
K. Noonan, K. Burke / The Social Science Journal 42 (2005) 241–256
253
6.2. Competing risks As revealed by the graph, the hazard of reunification is very different from the hazard for the termination of parental rights. Table 4 presents the coefficient estimates and the hazard ratios from these estimations in order to discuss these differences in more detail. The hazard ratios are reported because the coefficient estimates are difficult to interpret. If the hazard ratio is greater than 1 the risk of the event occurring will be increasing and conversely, if the hazard ratio is less than 1 the risk will be decreasing. More specifically, if the hazard ratio is equal to 2, the risk doubles. If the hazard ratio is equal to .5, the risk is cut in half. 6.2.1. Child characteristics As expected, every child characteristic with the exception of the child’s gender significantly impacts the amount of time the child spends in care until the termination of parental rights. However, these variables are not all significant in predicting the probability of reunification. Consistent with Courtney and Wong (1996), we find that African American children and children of Hispanic decent have a significantly lower risk of the termination of parental rights. Moreover, these children are also less likely to be reunified with their parents. Consistent with the findings of Benedict and White (1991), Jenkins (1967), Lawder et al. (1986) and Seaburg and Tolley (1986) we find children who are diagnosed with a disability have a lower risk of reunification. Surprisingly, these children are also more likely to experience the termination of parental rights. Since a child diagnosed with a disability may be harder to care for, parents may be more apprehensive about having their children return home and social workers may try to expedite a permanent placement by quickly terminating parental rights. As expected, the probability of parental rights termination is lower for older children. In addition, as the number of placements a child has in foster care increases, both the risk of reunification and the risk of termination of parental rights are lower. 6.2.2. Reasons for entry The reasons the child enters foster care have very different effects on the two hazard rates. For instance, children who enter care because of neglect, parental alcohol abuse, or can’t cope are significantly more likely to be discharged to their parents. Alternately, as expected, children who enter care due to physical abuse are significantly less likely to be reunified with their parents. Under current law, children who are in imminent danger are not returned to their parents. Thus, the results indicate that children are not considered to be in imminent danger until there has been physical abuse. Surprisingly, none of these reasons for entry, not even physical abuse, significantly impact the probability of parental rights termination. This supports the hypothesis that parental rights are not being terminated until there are potential adoptive parents. As expected, all of the Child Reasons for entry into care reduce the probability that parental rights will be terminated. Since these children may be the most difficult to place in adoptive homes, social workers may be more apprehensive about terminating parental rights. These findings indicate a need for greater support for children with emotional, behavioral, drug and health problems while in the foster care system.
254
K. Noonan, K. Burke / The Social Science Journal 42 (2005) 241–256
Surprisingly, two of the Child Reasons for entry, child disability and child behavior, increase the probability that a child will be discharged to their parents. Perhaps these parents feel more guilt for having placed their children in the foster care system. As expected, both the relinquishment of parental rights and the death of a parent have significantly higher risk of the termination of parental rights, while the risk of reunification is lower. Clearly, in both of these cases, the social worker should be able to terminate parental rights more quickly. 6.2.3. Remaining variables The Case Goal set by the social worker significantly impacts both of the hazard rates. As expected, if the goal of the social worker is to place the child up for adoption the likelihood of the termination of parental rights is significantly increased. Moreover, children with this same case goal have a significantly lower probability of being discharged to their parents. Children with the case goal of reunification, have an increased probability of discharge, as expected. Surprisingly, however, these children also have a higher probability of parental rights termination. After long spells in care, social workers will work towards placing these children in adoptive homes even though their initial goal was for reunification. As expected within in the Placement Setting category, we find children who are placed with kin have a lower likelihood of being reunified with their parents, as well as a lower likelihood for the termination of parental rights. For children who are placed with a relative, social workers may have less of an incentive to endorse an exit from care. Unlike a typical foster home, these children are in long-term stable settings. Furthermore, as hypothesized, children placed in a pre-adoptive home are significantly more likely to experience the termination of parental rights and less likely to be discharged. The Foster Family Structure plays much less of a role than the other variables discussed. Surprisingly, children placed with a single male have a significantly lower probability of parental rights termination. As hypothesized, children whose Caretaker Family Structure was single female are significantly more likely to experience the termination of parental rights. Surprisingly, we find that children who enter from a home with a married couple are also significantly more likely to experience the parental rights termination. Many of the Economic Variables such as Foster Care System Support, Medicaid Funding, and SSI Funding reduce the likelihood that children will be discharged to their parents. Since only the children from low-income homes are likely to receive these sources of funding, this indicates as hypothesized that children in poverty are less likely to be reunited with their parents.
7. Conclusions and future research In this paper, we use a two-way competing risks hazard model to determine which children are at greatest risk for the termination of parental rights. Based on our estimation results, there are two major findings. First, our results emphasize the importance for future researchers to allow for more than one outcome or different discharge alternatives when examining foster
K. Noonan, K. Burke / The Social Science Journal 42 (2005) 241–256
255
care dependence. The hazard of reunification and the hazard of terminating parental rights are very different. While the hazard of reunification increases over time, the hazard of terminating parental rights decreases over time. Second, adoptability does appear to factor into the parental rights termination decision. We find that the characteristics of the child are very influential in determining how much time a child will spend in care before the parental rights are terminated. More specifically, as hypothesized, African American children, children of Hispanic decent, children who experience multiple placements and older children will spend more time in care until the termination of parental rights. Furthermore, African American children, children of Hispanic decent and children with multiple placements are less likely to be reunited with their parents. Since these results indicate that these children will spend significantly more time in the foster care system, policies should be established to improve the outcomes for these children. For example, there should be incentives in place for families who adopt either African American children or children of Hispanic decent. In addition, all of the Child Reasons for entry into care decrease the probability of parental rights termination. As discussed, social workers may be less inclined to terminate parental rights if adoptive parents have not been identified. Extensive counseling should be available for these children to assist them in overcoming any emotional problems in order to increase their chances of adoption. While this paper indicates a distinct difference between children who are discharged to their parents and the children whose parents have their rights terminated, it is just a beginning in analyzing the impact of the termination of parental rights. To fully assess the current policy, it is imperative to consider all policy implications. That is, policy makers need to be aware of both the expected and the unexpected impacts that a policy may have. For example, given the distinct differences we found, specifically in the child’s characteristics, the termination of parental rights may only occur when an adoptive home is already present. If so, mandating the termination of parental rights for all children will do nothing more than create more legal orphans. Only future research will help determine this. In addition, our results are specific to New Jersey, thus future research should extend our model by including AFCARS data from additional states and additional years.
Note 1. We use the termination of the mother’s rights as the termination date. In all but one case, the father’s rights are terminated at the same time or earlier than the mother’s rights.
Acknowledgements The data utilized in this paper were made available by the National Data Archive on Child Abuse and Neglect, Cornell University, Ithaca NY; and have been used by permission. Data from the study AFCARS were originally collected by the Children’s Bureau, Department of Health and Human Services. Neither the collector of the original data, the funder, the Archive, Cornell
256
K. Noonan, K. Burke / The Social Science Journal 42 (2005) 241–256
University, or its agents or employees bear any responsibility for the analyses or interpretations presented here. We have benefited from comments by Mark Prus and Mechtild Nagle as well as the research assistance of Sara Tewksbury, Roxana Garcia and Jennifer Marogi. An earlier draft of this paper was presented at the 2000 New York State Economic Association Meetings. References Benedict, M. I., & White, R. B. (1991). Factors associated with foster care length of stay. Child Welfare, 70(1), 45–58. Benedict, M. I., White, R. B., & Stallings, R. (1987). Race and length of stay in foster care. Social Work Research and Abstracts, 23(4), 23–26. Borgman, R. (1981). Antecedents and consequences of parental rights termination for abused and neglected children. Child Welfare, 60, 391–404. Christian, S. (1999). 1998 State Legislative Responses to the Adoption of Safe Families Act of 1997. State Legislative Report, 24(5). Courtney, M. E. (1994). Factors associated with the reunification of foster children with their families. Social Service Review, 68, 81–108. Courtney, M. E., & Wong, Y. I. (1996). Comparing the timing of exits from substitute care. Children and Youth Services Review, 18(4/5), 307–334. Fanshel, D. (1971). The exit of children from foster care: An interim research report. Child Welfare, 50(2), 65–81. Freundlich, M. (1999). Expediting termination of parental rights: Solving a problem or sowing the seeds of a new predicament? Capital University Law Review, 28(1), 97–110. Glisson, C., Bailey, J. W., & Post, J. A. (2000). Predicting the time children spend in state custody. Social Service Review, 74, 253–280. Goerge, R. M. (1990). The reunification process in substitute care. Social Service Review, 64, 422–457. Guggenheim, M. (1995). The effects of recent trends to accelerate the termination of parental rights of children in foster care—an empirical analysis in two states. Family Law Quarterly, 29, 121–140. Jellinek, M. S., Murphy, J. M., Poitrast, F., Quinn, D., & Bishop, S. (1992). Serious child mistreatment in Massachusetts: The course of 206 children through the courts. Child Abuse & Neglect, 16, 179–185. Jenkins, S. (1967, October). Duration of foster care: Some relevant antecedent variables. Child Welfare, 46, 450–456. Lawder, E. A., Poulin, J. E., & Andrews, R. G. (1986). A study of 185 foster children 5 years after placement. Child Welfare, 65(3), 241–251. Noonan, K. (2000). An investigation of dependence on the foster care system. Unpublished doctoral dissertation, State University of New York at Stony Brook. Seaburg, R. J., & Tolley, E. S. (1986, Fall). Predictors of the length of stay in foster care. Social Work Research and Abstracts, 22, 11–17. Schetky, D. H., Angell, R., Morrison, C. V., & Sack, W. H. (1979). Parents who fail: 51 cases of termination of parental rights. Journal of the American Academy of Child Psychiatry, 18, 366–393. Stata Corp. (1999). Stata reference manuals: Release 6.0. College Station, TX: Stata Corporation. U.S. Census Bureau. (2000). Poverty estimates by selected characteristics 1998 and 1999. Retrieved March 15, 2000 from http://www.census.gov/hhes/poverty/poverty99/pv99est1.html. U.S. Census Bureau. (2001). New Jersey state and county quickfacts. Retrieved May 15, 2001 from http://quickfacts. census.gov/qfd/states/34000.html. U.S. House of Representatives, Committee on Ways and Means. (1998). Background material and data on programs within the jurisdiction of the committee on ways and means: 1998 Green Book. Washington, DC: US Government Printing Office. Wattenberg, E., Kelley, M., & Kim, H. (2001). When the rehabilitation ideal fails: A study of parental rights termination. Child Welfare, 80, 405–432. Wulczyn, F., & Goerge, R. M. (1992). Foster care in New York and Illinois: The challenge of rapid change. Social Service Review, 66, 278–294.