|
Activity Number:
|
449
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 7, 2003 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Social Statistics Section
|
| Abstract - #301523 |
|
Title:
|
Competing Risks Survival Model in Foster Care Research
|
|
Author(s):
|
Tiandong Li*+ and David R. Judkins and George Gabel and Madeleine Kimmich
|
|
Companies:
|
Westat and Westat and Westat and Human Services Research Institute
|
|
Address:
|
1650 Research Blvd., Rockville, MD, 20850,
|
|
Keywords:
|
observational study ; recidivism ; counterfactual imputation
|
|
Abstract:
|
To evaluate the effects of Ohio's foster care waiver program, Ohio Department of Job and Family Services provided data on foster care placements for 14 demonstration and 14 comparison counties. A competing risks survival model for the duration and outcomes of first placements was constructed on this dataset. This was done by fitting separate proportional hazard models for each exit type: reunification, adoption, runaway, custody to relative and other. The competing risks model was then used to both impute the outcomes of censored first placements under actual conditions and the outcomes of all first placements in the demonstration counties under the counterfactual condition of no waiver. The completed actual file was then compared with the simulated counterfactual file to estimate the effects of the waiver. To validate the model and imputation procedure, outcomes of all first placements were also imputed under actual conditions. A related model and imputation procedure for the duration and outcomes of first reunions was also validated. This paper presents the results of these validations.
|
