Re-interview, Revisited

The reinterview method is the usual technique for estimating the response variance and the response bias. It is assumed that the respondent does not remember or is not influenced by the original response. The gross discrepancy rate (GDR) is the ratio of the sum of the two types of discrepancies to the total number of respondents in a yes/no type of question. The response variance is half of the GDR. Response bias is also estimated from the ratio of the net difference between the two discrepancies and the total. However this assumption that the respondent neither remembers nor influenced by the original response is not realistic. A measure of the memory effect is derived in this paper. The GDR, the sample response variance and the response bias are rectified based on this measure of the memory effect.