Recent Research in Reinterview Procedures

The Bureau of the Census makes frequent use of reinterview procedures to measure the quality of census and survey results. The kind of reinterview procedure used can itself affect the measurement of quality. For example, does a time lag between the original interview and the reinterview affect the measurements of quality? Also, if the reinterviewer is given access to the original responses, are the measurements of quality affected? By comparing estimates of the simple response variance and estimates of the bias for several characteristics from three samples of the 1960 Census enumerated population, some light is thrown on these questions. It was found that for the purpose of evaluating the quality of a census or survey, the best procedure is one in which the reinterview is relatively close in time to the original interview and one in which the reinterviewer does not have access to the original responses.