Skip to main content
Open Access Publications from the University of California

Multiply Imputed Sampling Weights for Consistent Inference with Panel Attrition


This chapter demonstrates a new methodology for correcting panel data models for attrition bias. The method combines Rubin's Multiple Imputations technique with Manski and Lerman's Weighted Exogenous Sample Maximum Likelihood Estimator (WESMLE). Simple Hausman tests for the presence of attrition bias are also derived. We demonstrate the technique using a dynamic commute mode choice model estimated from the University of California Transportation Center's Southern California Transportation Panel. The methodology is simpler to use than standard maximum likelihood-based procedures. It can be easily modified to use with many panel data estimation and forecasting procedures.

Main Content
For improved accessibility of PDF content, download the file to your device.
Current View