Abstract:
|
With consistent decline of survey response rates over the past decade, the need increases for survey researchers to gather useful auxiliary variables to reduce bias. One straightforward source is the set of interviewer observations of characteristics of sampled units. Compared with area-level characteristics, which researchers have available for reducing bias due to screener non-response, characteristics at the dwelling unit-level may provide more information and result in a weight adjustment that could potentially have an impact on reducing bias further. These observations, however, usually vary greatly among interviewers, and tend to lack the quality needed for the survey data producers. To investigate the level of quality and usefulness of interviewer observations, this study assesses completeness, consistency, and potential for bias reduction in the Food Acquisition and Purchasing Survey Pilot Study. This paper will shed light on the dwelling unit characteristics that are more likely to be problematic, the potential value added beyond area-level characteristics for non-response adjustments, and ways to improve upon consistency of the observations.
|