Identifying Establishment Characteristics of Potential Nonresponse Bias

Morgan Earp, Daniell Toth, Polly A. Phipps, and Charlotte Oslund

Abstract

This paper attempts to identify establishment characteristic associated with lower response rates and nonresponse bias in the Job Openings and Labor Turnover Survey of the U.S. Bureau of Labor Statistics. Using regression trees, we identify subgroups of establishments, based on establishment characteristics, that are least likely to respond at each phase of the data collection process, as well as those that contribute to overall nonresponse bias. The results of our regression tree models can be used to develop strategies for increasing participation using responsive design and/or improving post adjustment methods such as weighting.