An official website of the United States government
An increased nonresponse rate is a major problem not only in household surveys in the U.S. and abroad but also in establishment surveys (Christianson and Tortora, 1995) and economic censuses (Ambler and Mesenbourg, 1992). An increasing nonresponse rate in establishment surveys has been of primary concern to government bureaus collecting data from businesses or firms, as a high response rate is considered an important component of data quality, and the effort to reduce nonresponse mounts the survey cost. It is important for survey researchers to reduce as much nonresponse as possible, because nonresponse affects the reliability of statistical estimates by introducing bias. This paper reports the findings from a recent field experiment conducted by the Bureau of Labor Statistics (BLS) that was designed to evaluate a set of nonresponse-reducing techniques for an establishment mail survey. The nonresponse reducers investigated are advance letters and reminder/thank you letters. Among various design techniques (e.g., personalization, stamped return envelope, first outgoing postage, sponsorship, financial incentives, questionnaire color, and etc.), the combined use of advance and reminder/thank you letters have been found most effective in reducing nonresponse rates for voluntary household surveys and censuses (Dillman et al., 1993), but have not been tested yet for establishment surveys.