Using Paradata from Interviewer Voices to Examine Nonresponse Error and Measurement Error

• Activity Number: 491
• Type: Topic Contributed
• Date/Time: Wednesday, August 6, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Survey Research Methods Section
• Author(s): Frauke Kreuter*+ and Fred Conrad and Jose Benki and Alexandra Birg
• Companies: University of Maryland and University of Michigan and University of Michigan and University of Munich
• Keywords: interviewer ; nonresponse ; measurement ; speech rate ; multilevel models ; total survey error

Abstract

When potential survey respondents decide whether or not to participate in a telephone interview, they may consider what it would be like to converse with the interviewer who is currently inviting them to respond. This includes for example how the interviewer sounds, speaks and interacts. Using a corpus of 1380 audio-recorded survey invitations this presentation will discuss how vocal properties of interviewers and householders, such speech rate, pause duration, gap duration and vocal pitch affect householders' decision to participate in a survey. Of particular interest is the increasing convergence between interviewers and respondents speech rate and its positive association with the participation decision. Such convergence and other vocal properties are factors that later might lead to correlated response variance, often interpreted as measurement error due to interviewers.