Abstract
Using data from a national longitudinal survey of children referred to child protective services (NSCAW II), this article compares behavioral, child/caregiver relationship, and school performance outcomes for children residing in kinship and nonkinship settings. Up to three waves of data were gathered for each child. The analysis sample comprises 4,202 children and 10,881 observations. Two sets of regressions were completed. One represents family living settings using child-mean centered predictors while the other does so with dummy-coded variables. An advantage of the child-mean centered predictors is that omitted variables that vary between children but that are constant within each child do not introduce bias. The regressions using dummy-coded variables evidenced, on balance, somewhat better outcomes for kinship settings than for nonkinship ones. Good outcomes in these regressions were found for kinship adoption. Results with the child-mean centered regressions were more equivocal, though perhaps still favored kinship settings. Limitations and policy recommendations conclude the article.