MODE EFFECTS IN A SURVEY OF MEDICARE PROSTATE SURGERY PATIENTS

A 1992 national survey of 1,072 Medicare beneficiaries who had surgery for prostate cancer was designed to evaluate the way that mode of dat a collection affected responses. The samples outside of Massachusetts were randomly assigned to one of two primary data-collection strategie s: mail or telephone, In Massachusetts, the two primary modes of data collection were mail and personal interviews in respondents' homes. Of the 51 questions compared, there were 25 statistically significant di fferences between the mail and telephone responses, Differences were c oncentrated among questions asking for current self-descriptions. The mail responses indicated more problems or worse health status 22 times . When mail responses and personal interviews were compared, there wer e only nine significant differences, of which four indicated more prob lems via mail response. The results highlight the need for better gene ralizations about which questions are affected by mode and how to mini mize mode effects to facilitate multimode surveys.