Random Selection in a National Telephone Survey: A Comparison of the Kish, Next-Birthday, and Last-Birthday Methods
Diane Binson, Jesse A. Canchola, and Joseph A. Catania
The process of randomly selecting a respondent in household surveys is one juncture in the interview session that often results in a large proportion of refusals. Do the standard methods (Kish, "last-birthday," and "next-birthday") differentially contribute to the dropout rate? At what stage in the screening process are informants/respondents more likely to drop out? The literature that addresses these questions is mixed. Utilizing a national Random Digit Dialing (RDD) sample design, we compared the differential dropout rate in the standard respondent selection methods; in particular, the magnitude of the dropout rate in each phase of the screening process. Contrary to what would be expected, we found a significant difference between the three methods in the dropout rate during the initial stages of the screening process. The highest proportion of dropout rates occurred in the screening interview before the informant was asked questions unique to one of the three selection methods. The higher rates were in the Kish and ``last-birthday' conditions, with the highest number in the Kish condition. We suggest that interviewers rather than respondents are a primary source of the higher rate of refusals when using the Kish method.
Kish; respondent selection; birthday methods.