Search

No research results are any better than the quality of the data. As scientists, survey researchers must share the problems of measuring, understanding, and controlling bias in their data. The potential sources of bias in survey research are many and varied. They may appear at any one or all of the several stages of the data gathering process. For example, they may occur in the selection of the sample, in the development of measuring instruments, in securing the responses of the persons in the sample, in the respondents replies, in the recording of the responses, and in processing the recorded responses. Even beyond the data collection process, bias may occur, as in the misapplication of statistics and in the interpretation of data. Progress has been made toward identifying and controlling sources of bias in some of these stages of the research process. Perhaps most progress has been made in the sampling stage and in the statistical application stage. Much advance has been made also in the reduction of error at the instrumentation stage. However, considerable work remains to be done on the problems of error control, especially biasing error, at all stages of the research between instrumentation and statistical analysis. This includes getting full response from the sample, accurate answers to the questions asked, and accurate recording and processing of responses. The accumulation of bias from these and other possible sources may be referred to as total or aggregate bias. The total bias may then be divided for analytical purposes into the following types: (1) sampling bias, (2) instrumentation bias, (3) non-response bias, (4) respondent bias, (5) interviewer bias, and (6) processing bias. This paper is an effort to deal with the problem of respondent bias in survey research. That respondent bias is a potent force in diminishing the validity and reliability of results from survey research is well recognized. The recognition of this condition has led to an enormous amount of activity aimed at isolating factors which determine this weakness of research results. These studies, by and large, have dealt with the problem of determining the direction of respondent bias, its magnitude, the speearle variables which produce respondent bias, and the effect of such bias on research results. [ABSTRACT FROM AUTHOR]