Assessment of Measurement Error and Nonresponse Error Using Respondent-Provided Paradata from Paper Self-Administered Questionnaires

• 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): Heather M. Schroeder*+ and Russell W. Stark and Ashley Bowers and Roger D. Goddard and Robin Tepper Jacob
• Companies: University of Michigan and University of Michigan and University of Michigan and Texas A & M University and University of Michigan
• Keywords: paradata ; measurement error ; nonresponse error ; paper self-administered questionnaires

Abstract

Paradata analysis was introduced as a powerful tool to reduce error in web surveys and interviewer-administered surveys (Kreuter, 2010), and researchers are now harnessing the power of paradata analysis in paper self-administered surveys as well. Features of the environment in which self-administered surveys are completed can be measured using respondent-provided paradata. Interview setting features such as the presence of others have been shown to affect reporting in interviewer-administered surveys (see Tourangeau & Yan, 2007, for a review). We also expect that the survey setting may introduce measurement variability in paper self-administered surveys. Paradata including the presence of others, consultation with others in answering the questions, and location(s) of completion were collected from teachers and principals in selected schools as part of a self-administered survey. We assess whether features of the survey setting contribute to measurement error and nonresponse error as indicated by variation in the reporting of principals' and teachers' effectiveness, straightlining, reliability of multi-item scales, and missing data rates.