Who Gets a CASA?: Assessing Patterns of Selection Bias in the Appointment of Court Appointed Special Advocates

Background:

Prior research examining the effectiveness of Court Appointed Special Advocates (CASA) as an intervention for improving the outcomes of children in foster care has been hindered by the prevalence of selection bias, because judges may select children to receive CASA representation based on perceived need or other non-random characteristics. Because selection bias poses a strong threat to internal validity, researchers have struggled to isolate the effects of CASA services on child and case outcomes. To advance a research agenda with a stronger capacity for causal inference, the present study examines patterns of selection bias in the CASA appointment process among a multi-year population sample of foster children in Texas.

Methods:

Administrative child welfare data from the Texas Department of Family and Protective Services was linked with CASA case management data from 68 of 71 local CASA program affiliates in Texas, creating a two-year entry cohort of children who were placed in substitute care in Texas during fiscal years 2013 and 2014 (N=31,371). The dependent variable is a binary indicator of whether each child was selected to receive CASA services. An array of child-, family-, and case-level predictor variables were used to examine their association with CASA appointment.

Unadjusted bivariate chi-square tests were conducted for every predictor against the outcome to examine crude observed differences between groups. Mixed-effects logistic regression was used to identify independent predictors of CASA appointment while controlling for all factors and accounting for data clustering at the court/judge level.

Results:

Just over half of children in the sample (N=17,603; 56.1%) were selected to receive CASA services. Multivariate regression findings suggest that the factors which increase the odds of CASA appointment signal more complex case circumstances. Controlling for all covariates, CASA is significantly more likely to be appointed to:

• Children in sibling groups (3+ siblings: OR: 2.08; p<.000)
• Older children (age 13-17: OR: 1.52; p<0.000)
• Children whose caregivers have more risk factors (3+ risk factors: OR: 1.21; p=.002)
• Children involved in three or more prior child welfare cases (OR: 1.23; p<.000)
• Children who experienced three or more maltreatment types leading to removal (OR: 1.37; p<.000)
• Children whose initial placement is a non-relative setting (OR: 1.15; p<.000)

Implications:

Consistent with prior research, findings indicate that CASA volunteers are appointed to cases representing more serious or complicated circumstances. At the policy level, detailed understanding of the aggregate characteristics of children selected for CASA can inform training and program improvement for CASA programs. At the research level, these findings lay the groundwork for a rigorous quasi-experimental outcome study that will substantially improve upon prior research on CASA effectiveness. Specifying multi-level factors that predict selection to CASA will facilitate a robust propensity score analysis that can approximate random assignment and minimize the selection bias that has hampered prior efforts to identify the causal effects of CASA services on child and case outcomes.