Nonresponse Adjustment Using a Response Propensity Model

• Activity Number: 432
• Type: Roundtables
• Date/Time: Wednesday, August 4, 2010 : 7:00 AM to 8:15 AM
• Sponsor: Section on Government Statistics
• Author(s): Donsig Jang*+
• Companies: Mathematica Policy Research, Inc.
• Keywords: survey nonresponse ; weighting adjustment ; weighting class ; model fitting ; propensity score

Abstract

In recent years, as survey nonresponse rates have increased, survey statisticians have grown more concerned about nonresponse bias. To reduce such bias in survey estimates, it is critical to include auxiliary variables related to response propensity and/or key survey variables for weighting adjustments. Response propensity models allow the use of numerous variables for weighting adjustments, but statisticians may encounter issues in using these models, including:  Which variables to include in the model  How to limit the number of interaction terms in the model  Whether to fit the model with or without survey weights  How to derive the adjustment factors from the predicted model (using an individual predicted value or cell-based ratio factor) In this roundtable, we will lead a discussion of the practical use of response propensity models for nonresponse adjustments.