Modeling the Difference in Interview Characteristics for Different Respondents

John S. Dixon

Abstract

The survey process as measured by paradata may be different for different respondents. Those initially reluctant to participate may be convinced by the interviewer to cooperate, or the reluctance may produce attrition. Difficulty to contact respondents may be related to interviewer effort and busy respondent schedules, or it may be a form of reluctance. This paper will use mixed models to attempt to identify subsets of respondents whose paradata relates differently to survey outcomes and measures of survey quality.