Analysis Plan and Project Code

Estimate baseline racial disproportionality in child welfare referrals and acceptances from July 2014 to June 2017.

1. Using data from the most recent American Community Survey 5-year estimates (2012-2016), generate population estimates for children in Charlottesville by race and calculate proportions by race, incorporating the uncertainty represented by the 90% margins of error. See prep_basedisp.R for details. The detailed tables in the 5-year estimates were estimates for sex by age for
   • The combined population (B01001)
   • White alone population (B01001A)
   • Black or African-American alone population (B01001B)
   • American Indian and Alaska Native alone population (B01001C)
   • Asian alone population (B01001D)
   • Native Hawaiian and Other Pacific Islander alone population (B01001E)
   • Some other race alone population (B01001F)
   • Two or more races population (B01001G)

2. Using the CWS referral data from June 2014-July 2016 from DSS, generate counts of children referred to CWS by race for each year (with years defined as June-July). Calculate proportions by race. See prep_basedisp.R for details.

3. Generate visualization of proportions of children by race in Charlottesville and proportions of children referred to CWS (see analysis_basedisp.R).

4. Calculate racial disproportionatilty by race, the ratio of % of [racial cateogry] children referred to CWS to % of [racial category] children in population (e.g., % referred to CWS who are white/% children in Cville who are white). Generate point estimates and estimates for lower and upper bounds of estimated population proportions (see analysis_basedisp.R).

5. Pull Charlottesville Census tract data for child population and poverty rates; generate maps for child population, poverty rates, referral rates. See prep_tract.R.

Estimate racial disparity in post-referral decision points, modeling decision outcomes as a function of race and other characteristics.

1. Reporter relation – racial proportions by reporter relation
2. Screened in – referred children screened in (accepted as a case) as a function of race and covariates
3. Investigation – accepted/screened in children assigned to investigation or assessment (alternative response) as a function of race and covaraites
4. Substantiation – outcome of investigations for investigated cases as a function of race and covariates
5. Level of substantiaion – level of finding among substantiated cases

See prep_postreferral.R, figures_postreferral.R, and analysis_postreferral.R.

Estimate racial disparity in foster care outcomes, modeling outcomes as a function of race and other characteristics.

1. Initial placement type – among children placed in out-of-home care, type of placement by race and other characteristics
2. Number of placements – among children placed in out-of-home care, number of placements by race and other characteristics
3. Time in care – among children placed in out-of-home care, duration of time in care by race and other characteristics
4. Exit – among children placed in out-of-home care more than a year prior to June 30, 2018, exit from the system within year by race and other characteristics
5. Nature of exit – among children in out-of-home care who have exited, how/why they exit by race and other characteristics