Using Interviewer Observations Related to the Interview Process for Measurement Error and Nonresponse Error Adjustment

• Activity Number: 577
• Type: Contributed
• Date/Time: Wednesday, August 3, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Section on Survey Research Methods
• Author(s): Rachel Anne Orlowski*+ and Ashley Bowers and Wayne E. Baker
• Companies: Wayne State University and University of Michigan and University of Michigan
• Keywords: measurement error ; nonresponse error ; interviewer observations ; paradata

Abstract

Interviewer estimates of household characteristics have been used for nonresponse error adjustment (Kreuter et al., 2010) but rarely have interviewer observations of respondent behavior and attitudes related to the interview process been exploited for survey error adjustment (Tarnai & Paxson, 2005). In addition to adjustment purposes, these paradata also can guide interviewer training efforts and case assignment based on interviewer skills. In panel designs, these observations from prior waves can be used for adjustment and management in future waves. We analyze interviewer assessments of respondent understanding of the questions and respondent attitude toward the interview from a rotating panel telephone survey with a battery of attitudinal items. Indicators of measurement error, including selection of the same response for most of a question series, socially desirable reporting, and reliability of multi-item scales, are modeled as a function of interviewer assessments of respondent understanding and attitude. We also examine whether interviewer assessments can explain indicators of nonresponse error, such as missing data rates and participation in a future survey wave.