Sunday, 16 January 2005 - 10:30 AM

This presentation is part of: Findings from the National Survey of Child and Adolescent Well-Being (NSCAW): Applying Innovative Methods to Understanding Services and Outcomes for Maltreated Children

Assessing the Impact of Kinship Foster Care vis-à-vis Foster Care on Children’s Well-being

The increasing use of kinship foster care has drawn attention among child welfare researchers and policy makers, particularly with regard to its merits and demerits. Scattered evidences suggest that kinship foster parents differ from other foster parents in tending to be older, less well off financially, to have more health problems and to be single parents; they are much more likely to be African American and to receive fewer services; and they are more likely to maintain the children’s ties with their biological parents.

To evaluate the well-being of children receiving kinship foster care vis-à-vis those receiving foster care, one must control for selection bias. The parties involved in the selection process (e.g., families, police, child welfare workers, courts), factors affecting the process of decision making should be carefully controlled for in evaluations concerning the impacts of services on children’s well-being. Two selection biases are likely to exist in this process. One is the bias pertaining to bureaucratic selection - placing (or not placing) a child into kinship foster care is because the child has (or does not have) capable kin parent. The other is the bias pertaining to kin parent’s own selection – it’s likely that a kin denies serving as a foster parent because the child has excessive behavioral problems.

In this presentation we will report findings of a propensity score analysis using NSCAW data. Approximately 180 kinship foster children and 260 foster children comprise the study sample. The study will examine two outcome measures: the change of behavioral problems or CBCL and the timing of transitions to permanency (i.e., reunification, guardianship, or adoption).

The analysis involves the following two steps to control for selection biases. First, we will conduct a logistic regression analysis predicting propensity of using kinship foster care. Predictors of the logistic regression will include child and family characteristics. Based on the estimated propensity scores, we will re-sample the study children by matching kinship foster children with foster children. Three matching algorithms will be used in the resampling: nearest neighbor matching (caliper matching), Mahalanobis-metric matching with propensity score, and Heckman’s local linear matching.

Based on the new sample, we will conduct a second-stage analysis that: (1) examines the change of CBCL from baseline to 18 months by using Heckman’s difference-in-differences approach; and (2) investigates the timing of transition to permanency by using event history analysis. Preliminary analysis using PS2 Match procedures has confirmed that the above plan is feasible. The promise and limitations of propensity score matching, of various types, with these data will be discussed.

See more of Findings from the National Survey of Child and Adolescent Well-Being (NSCAW): Applying Innovative Methods to Understanding Services and Outcomes for Maltreated Children
See more of Symposium

See more of Celebrating a Decade of SSWR (January 13 - 16, 2005)