Methods for Responsive Split Questionnaire Panel Surveys

• Activity Number: 252
• Type: Contributed
• Date/Time: Monday, July 30, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Section on Survey Research Methods
• Author(s): Jeffrey Gonzalez*+
• Companies: Bureau of Labor Statistics
• Keywords: Adaptive design ; Matrix sampling ; Questionnaire development ; Respondent burden ; Survey design

Abstract

Lengthy surveys may be viewed as burdensome and yield both low data quality and high nonresponse. To address these concerns, we can reduce the length by eliminating items, but this means that some information does not get collected. An alternative is to divide a survey into subsets of items and administer each subset to subsamples of the full sample. This is called a split questionnaire (SQ) and has the benefit of collecting all original survey data. We identify a deficiency in the current set of SQ methods - incomplete use of prior information about the sample unit in the design. In applications of SQs, only characteristics of the survey items are used in the design; but, if joint consideration is given to characteristics of the items and sample unit, then there may be the potential to improve the SQ's efficiency. In this paper, we explore the extent to which jointly considering both sets of information will yield more efficient SQs. We propose methods for including prior information about the sample unit into the SQ using features of responsive design. We use concepts from survey design, experimental design, and epidemiology to evaluate the proposed new elements of our SQ design.