Using Regression Trees to Model Characteristics of Nonresponse and Measurement Error in a Longitudinal Survey

• Activity Number: 349
• Type: Topic Contributed
• Date/Time: Tuesday, August 2, 2016 : 10:30 AM to 12:20 PM
• Sponsor: Survey Research Methods Section
• Author(s): Morgan Earp* and Daniell Toth and Polly Phipps and Charlotte Oslund
• Companies: Bureau of Labor Statistics and Bureau of Labor Statistics and Bureau of Labor Statistics and Bureau of Labor Statistics
• Keywords: Nonresponse ; Measurement Error ; Regression Trees ; Longitudinal ; Establishment

Abstract

Nonresponse and measurement error are common problems in survey data collection, especially in longitudinal surveys where each sample unit is asked to give multiple responses over time. With multiple response periods, there are more opportunities for nonresponse as well as measurement error. Using regression tree models and the longitudinal Job Openings & Labor Turnover Survey conducted by the US Bureau of Labor Statistics, we analyze characteristics associated with establishment nonresponse and measurement error during data collection. The characteristics include length of time in survey, establishment characteristics (size, industry, ownership, multi-establishment, etc.), and population size of the establishment's geographic location.