Interviewer Effects on Survey Questionnaire Response Times: A Hierarchical Bayesian Analysis of Multivariate Survival Paradata

• Activity Number: 616
• Type: Contributed
• Date/Time: Thursday, August 7, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Survey Research Methods Section
• Author(s): Hiroaki Minato*+
• Companies: U.S. Energy Information Administration
• Keywords: hierarchical Bayesian data analysis ; multivariate survival analysis ; paradata ; survey questionnaire response time ; Residential Energy Consumption Survey (RECS)

Abstract

Survey research often covers a range of topics in one large survey and related questions are usually grouped to form sections in the questionnaire. With a computer-assisted survey interview instrument, the time spent to complete each section of the questionnaire may be measured. Our interest is to understand the completion-rate variability due to interviewers.

With some language from biostatistics, we conceptualize multiple response times, "treatments" (e.g., interviewers, interview times), and hierarchical covariates. The multivariate survival data framework is used to model the relationship between multiple "failure" rates and survey treatments and covariates. And, we adopt the hierarchical Bayesian approach in analyzing the multivariate, multilevel data and making inference on interviewer effects on the multiple completion rates.

As an illustration, the 2009 Residential Energy Consumption Survey (RECS) data along with the paradata on survey questioning are examined.