Analysis of Nonresponse in Student Surveys

• Activity Number: 528
• Type: Contributed
• Date/Time: Thursday, August 11, 2005 : 10:30 AM to 12:20 PM
• Sponsor: Section on Survey Research Methods
• Author(s): Ronaldo Iachan*+ and William H. Robb and James G. Ross and Katherine H. Flint and Alan Bloch
• Companies: ORC Macro International, Inc. and ORC Macro International, Inc. and ORC Macro International, Inc. and ORC Macro International, Inc. and Centers for Disease Control and Prevention
• Keywords: Non-response ; Item non-response ; Student non-response ; School surveys ; Weight adjustments

Abstract

This paper investigates nonresponse in school surveys at the levels of student and item nonresponse with a focus on whether hard-to-reach students differ significantly from other students along key dimensions. The probability of student nonresponse will be modeled as a function of school characteristics (merged from the school sampling frame), student demographics, number of absences, and whether the student was administered a makeup survey (absent initially). Propensity (logistic) models will help identify the best predictors of nonresponse. As a byproduct of multivariate models, we investigate weighting approaches that incorporate these predictors in weighting adjustments. We compare alternate weighting procedures in their effectiveness in reducing the potential bias of nonresponse and variance impacts. At the item level, additional variables include position in the survey and type of item. Finally, we examine multilevel models for item and student nonresponse using these same predictors at the student and school levels, models that take into account between-item and between-student correlations.