Measurement Error in the Reported Reasons for Entry into the Foster Care System

To date, much of the research on foster dependence hinges on the validity of the reasons for entry into the foster care system. Yet, no one has tested these data. Since these reasons for entry help to assess individual differences in foster care children, the purpose of this study is to more closely examine these reasons. Using data from the Adoption and Foster Care Analysis Reporting System, we begin with exploratory factor analysis on the reported reasons for entry. Next, we specify and test a structural measurement error model of reasons for entry. The reported reasons for entry are not mutually exclusive. Rather, there are five significant commonalities across these various indicators. The commonalities are combined across the reported reasons for entry into the foster care system to create a set of mutually exclusive factors that represent reasons. We apply these factors to a model of dependence on the foster care system. Compared to a model that includes all of the individual indicators, we are able to get a better idea of the kinds of children that are at risk for delayed exits from foster care. ∗ 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 University, or its agents or employees bear any responsibility for the analyses or interpretations presented here. ∗∗ Debra S. Dwyer is an Assistant Professor at SUNY Stony Brook ([email protected]). Kelly Noonan is an Assistant Professor at Rider University ([email protected]). Any questions or comments are greatly appreciated. The federally funded US foster care system is available to children who are orphaned, neglected or delinquent in out-of-home care. While the system is designed to provide insurance against inadequate living arrangements for children, it is a social assistance program available to any children in need of a home. Once in the foster care program children are in the care of social workers. The objective of the social workers is to maximize the well-being of individual children in their care. In order to do this, decisions need to be made based on the child's specific circumstances that have placed them in care initially. The purpose of this study is to more closely examine the data available for assessing individual differences in foster care children. Indicators of reasons for entry into the system may provide us with a lot of information about unobserved differences in these children. Therefore, much of the research to date hinges on these data. To date, nobody has tested the validity of the data. Because of the loose eligibility requirements, the children who participate in the program are heterogeneous. Their reasons for entry into the program differ and consequently, their experiences within the program differ. There are a number of reasons for entry that we place into one of three categories. A child enters the foster care system when a parent is no longer able to care for him or her adequately. This may occur if a child is more challenging to care for than the average child. Consequently, we would say the reason for entry into the system is a "Child Reason". Children also end up on the system because their parents abuse them or their parents are not suited to care for them, and we say these children enter for "Parent Reasons". "Other Reasons" for entering the foster care system may include the absence of adequate care due to death of a parent or financial inability to care for the child. The kind of care these children need, and specifically the amount of time they spend on the system, should vary based on their situation upon entry into the system. Therefore, many of the studies to date hinge on these "reasons for entry" as explanatory variables in determining the length of stay in the system or other outcomes (Benedict et.al, 1987; Benedict & White, 1991; Dwyer & Noonan, 2000; Fanshel, 1971; Goerge, 1990; Jenkins, 1967; Lawder et.al, 1986; Noonan, 2000; Seaburg & Tolley, 1986). This is because a key source of heterogeneity across children entering the foster care system can be identified through their reasons for entry. A child who enters because of a drug addiction problem at age 13 is very different from a newborn whose parents died in an automobile accident. Clearly treatments must vary across each of these cases in order to maximize the child’s well-being. Studies to date primarily focus on the length of stay and dependence on the foster care system and some of the main determinants in these models are the "reasons for entry". This is because foster care policies are aimed at reducing the length of stay. We rely on these survey measures or entry reasons to capture unobserved child and parent characteristics. Theoretically this makes sense. Yet, these measures have never been tested. Since conclusions and policy implications hinge on these measures, the purpose of this paper is to test their validity as indicators of heterogeneity across foster care children. There is no reason to believe that the reasons are mutually exclusive. Our priors would suggest otherwise. For example, we argue that you cannot have child reasons for entry without parent reasons since the parents are responsible for making decisions on 1 Since studies find that dependence is correlated with negative outcomes later in life and dependence is also costlier, policies focus on minimizing dependence on the system behalf of the children. It is difficult to assess outcomes by reasons in the presence of collinearity. We will be using a new national data set, Adoption and Foster Care Analysis and Reporting System (AFCARS) which contains very detailed information about reasons for entry. We begin with exploratory factor analysis using the reasons for entry reported in AFCARS. Next, we specify and test a structural measurement error model of reasons for entry. We learn that there are six significant commonalities across these various indicators as opposed to the three broader categories we had predicted. As a consequence we have a better understanding of what the indicators are measuring. We use the results in a model of dependence on the foster care system. Section I provides some background. We describe the data in Section II. The Conceptual Model is presented in Section III followed by the empirical methodologies in Section IV. Results and conclusions follow in Sections V and VI. I. Background Based on a literature that finds that duration on the program is positively correlated with negative outcomes later in life (Barth, 1990; Blome, 1997; McDonald et al., 1996), it has been assumed that dependence on the foster care program is bad for the children. In addition, a consistent finding in the literature is that the probability of leaving the system declines with time spent in the system (Benedict et al., 1987; Courtney, 1994; Fanshel, 1971; Goerge, 1990; Jenkins, 1967). In other words, the longer a child remains in foster care, the less likely it is that the child will be able to leave. These findings, combined with substantial program growth between the eighties and nineties, prompted the enactment of the Adoption and Safe Families Act of 1997 (ASFA). The primary objective of ASFA is to minimize the duration of participation in the program. More recently, Dwyer and Noonan (2000) redefine "dependence" on the foster care to account for re-entry onto the system. In doing so they find that expediting discharges may not be ideal for all children if the objective is to minimize dependence on the system since premature discharges result in re-entry and an increased probability of dependence. In particular, some children are actually hurt by such policies. Since the analysis finds differences in unobserved heterogeneity in the effects of this policy, we cannot identify which types of children are winners versus losers. What we do learn is that not all children benefit from being rushed off the system and there is something systematically different that causes this. This unobserved heterogeneity is most likely correlated with the reasons that place the child in foster care. We have foster care to deal with children in special circumstances. The child's reasons for entry reflect these circumstances and therefore may be useful in distinguishing between "winners" and "losers" of current policy. Measurement error in those reasons make the indicators noisy controls for heterogeneity. The purpose of this study is to minimize noise in these indicators and try and pick up what is important and common among the many reasons reported. We will develop factors that may be better measures of these unobserved