160 – Using Adaptive Design and Collection Strategies to Improve Data Quality in Business Surveys
Strategies for Selecting a Follow-Up Sample of Nonrespondents in Business Surveys
Elisabeth Neusy
Statistics Canada
Wesley Yung
Statistics Canada
Jean-François Beaumont
Statistics Canada
Mike Hidiroglou
Statistics Canada
David Haziza
Université de Montréal
Follow-up of nonrespondents in business surveys is a time and resource intensive activity. Unlike household surveys where respondents are more-or-less equivalent, business surveys have influential units for whom a response is absolutely necessary in order to not adversely affect level estimates. An extensive simulation study was undertaken to compare different methods of performing follow-up for business surveys. For a fixed budget, is it best to follow-up all nonrespondents or only a sample of nonrespondents? If it is better to follow-up a sample of nonrespondents, what is the best way to select this sample? Should the sample be selected using simple random sampling (SRS), stratified SRS or probability proportional to size (PPS) sampling? The different methods were evaluated via a simulation study by comparing their Monte Carlo mean square errors.