Date of this Version
Published in Field Methods 2018, vol. 30(4), pp 371-385.
Some survey questions do not apply to all respondents. How to design these questions for both eligible and ineligible respondents is unclear. This article compares full filter (FF) and quasi filter (QF) designs for autobiographical questions in mail surveys. Using data from National Health, Wellbeing, and Perspectives Study, we examine the effect of type of filter on item nonresponse rates, response errors, and response distributions. We find that QF questions are more confusing to respondents, resulting in higher rates of item nonresponse and response errors than FF questions. Additionally, FF questions more successfully identify ineligible respondents, bringing estimates closer to national benchmarks. We recommend that survey designers use FF designs rather than QF designs for autobiographical questions in mail surveys.