Just as interesting as determining what features did matter in predicting outcomes was finding features that did not. For instance, the child's gender, the primary family's structure, and the foster family structure had little to no bearing on successful outcomes.
We've also been able to identify additional family segments in need. Relative foster homes are not as successful as thought, even with greater socioeconomic status; non-relative homes and institutions have the most "failed" cases, regardless of funding.