Poster #3 - Adjusting for Under-Reporting of CPS Involvement in the FFCWS: Implications for Empirical Analyses

• In Event: Saturday Poster Luncheon

Abstract

Child protective services (CPS) involvement is common among American families; over a third of all children and more than half of Black children experience at least one investigation by age 18 (Kim et al., 2017). However, studying the causes and consequences of CPS involvement is challenging due to data limitations. National data on children involved with CPS lack a counterfactual group, while state-level administrative data often miss key family and child factors. The Future of Families and Child Wellbeing Study provides detailed data on both CPS involvement and family and child factors but relies on self-reports which may underreport CPS involvement. In this paper, we develop a Bayesian method to correct for this underreporting and estimate the probability of CPS involvement by a given age. We then apply the adjusted measure in an empirical analysis and compare the estimates to those produced when using the unadjusted measure. We find modest differences in the estimates produced by each measure, suggesting that empirical estimates based on unadjusted self-reports of CPS involvement may lead to somewhat erroneous substantive conclusions regarding relations of CPS involvement with potential precursors thereto and consequences thereof. 

Authors

• Lawrence M. Berger, University of Wisconsin-Madison
  Non-Presenting Co-Author

• Tia M Dickerson, Columbia University
  Non-Presenting Co-Author

• Hye-Min Jung, Columbia University
  Presenting Author

• Margaret M. C. Thomas, The University of Chicago
  Non-Presenting Co-Author

• Jane Waldfogel, Columbia University
  Non-Presenting Co-Author