Nonresponse: A Crippling Bias in Mail Survey Research?

Abstract

Social scientists have long understood that response rates in survey research are fundamental to the representativeness of a sample. It is believed that late respondents or those who do not respond to various survey methodologies represent a distinct group of individuals that bias samples in unknown ways. Although little research has been done to examine such differences, this article analyzes a research project that was able to secure a 99% response rate after a tenacious followup that lasted over a year. All but one respondent was found eight years after the initial group was studied which allowed for comparison between early responders, who would have been identified by routine followup procedures, and those who, under typical methods, would have constituted the nonresponders. It was concluded that assumptions widely held by social scientists may not be valid, although certain clear differences were found to distinguish the groups.