A Typology of kinship care: Latent class and confirmatory analyses of kinship foster family characteristics

Background and Purpose: Over the past 20 years, empirical examinations of the characteristics of kinship care (e.g., Barth et al., 2007; Lewis & Fraser, 1987), and the outcomes of children placed with kin (e.g., Goerge, 1990; Wulczyn, 2003), have treated kinship foster families as an homogeneous population. Few, if any, attempts have been made to differentiate kinship foster families from one another, or to explore how child outcomes may vary across different types of kinship foster families (e.g., Zinn, 2008). The present study attempts to address both these issues. First, using a sample of kinship foster families from Illinois, attempts were made to identify distinct classes of kinship families based on several other distinguishing foster family characteristics, including the nature of foster families' relationships with the children in their care and several foster family demographic characteristics. Second, the significance and meaning of these identified classes were explored via several sets of confirmatory analyses, including an examination of differences in the characteristics and outcomes of children placed with each identified class of kinship family. Methods: The study sample included 1,312 randomly selected children placed in kinship and non-kinship foster homes in Illinois as of March 1, 2005. Data were drawn from a web-based survey of child welfare caseworkers (90.9% response rate) and administrative data from the Illinois Department of Child and Family Services. Subsequent placement histories for children and foster families were available through December 31, 2007. Distinct subgroups of kinship providers were indentified through latent class analysis (McCutcheon, 1987), which uses a set of observed manifest variables to differentiate subjects with respect to one or more unobserved latent classes. The specific manifest variables used for this study included foster parents' age and relationship with the child in their care, foster family structure (i.e., partner status, number and age of resident non-foster children), and foster parents' employment status. Results: Latent class analyses with the manifest variables described above identified several distinct classes of kinship foster families, including (a) lone non-grandparents with one or more resident non-foster children, (b) unemployed grandparents with few or no resident non-foster children, and (c) a small, but homogeneous, group of working grandparents with pre- or elementary school resident non-foster children. Confirmatory analyses suggest that the distribution of other kinship family characteristics (e.g., foster parent race, sufficiency of family income) differs significantly across these identified latent classes. Further, significant differences were found with respect to the characteristics of children (e.g., race, mental health status) placed with each class of kinship family. Finally, the likelihood of certain placement outcomes (e.g., placement stability) was found to differ significantly across latent kinship classes. Implications: The results described above support the hypothesis that there is substantial heterogeneity within the population of kinship foster families in Illinois, and that this heterogeneity serves to differentiate children's placement outcomes. These findings also suggest that treating kinship foster families as a single, homogenous population may obscure important differences within kinship care and, as a result, confound the observed relationships between kinship status and child outcomes.