Iterative Weighting Procedures for School Surveys: Trimming and Raking as One Algorithm

• Activity Number: 358
• Type: Contributed
• Date/Time: Tuesday, August 3, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Section on Survey Research Methods
• Author(s): Ronaldo Iachan*+ and James G. Ross and Mirna Moloney
• Companies: ICF Macro and ICF Macro and ICF Macro
• Keywords: Raking ; Trimming ; School surveys ; Design effects

Abstract

National school surveys typically adopt post-stratification at the final stage of weighting. As post-stratum cells are often defined in terms of school type (public versus non-public), grade, gender and race/ethnicity, some cells end up very small. Trimming is usually performed earlier in the weighting process prior to post-stratification; with trimming cells defined in terms of stratification rather than post-stratification variables, trimming is less effective because weights are not highly variable within these cells. In the investigation reported here, we developed and tested an iterative procedure for the trimming and raking based on the same cells defined in terms of the factors above. We applied the procedure to data from two national surveys. A comparison of the results shows gains in DEFFs obtained from the new alternate weighting over the established weighting procedures.