The Technical Appendix to the Urban Charter School Report contains material that explains the design of the Urban Charter School Study, the sampling approach used to select the cities for inclusion in the study and detail about the statistical techniques employed to conduct the study.
The technical appendix contains six sections:
"Model Selection", discusses how CREDO chose to use the Virtual Control Record (VCR) technique employed in this paper and the relative benefits and drawbacks to this and other commonly used analytic methods.
"Developing the CREDO Model," explains the development of the CREDO regression model and describes how comparisons are made across different states and testing regimes. This section also explores the feedback CREDO has received on the VCR method since the release of our original National Charter School Study in 2009 and how it has been incorporated into our analytic process.
"Data," discusses how test scores across all states were standardized, as well as the reasons for indicator variable omission where necessary.
"Defining Urbanity," discusses how CREDO chose which regions and schools to include in this report.
"Additional Estimations" presents tests of the robustness of CREDO’s modeling specification.
"Model Results" contains the full regression output from the primary (aggregated urban region) regressions.