Proxy Response in Social Surveys

Abstract

A common method of reducing costs in both personal and mail surveys is to require one member of a household to provide information about himself personally and about others in the household by proxy. In personal interviewing, where the cost of interviewer time in locating an individual forms a substantial proportion of the final cost of the survey, the use of proxy respondents has a greater relative value than in postal work, where once an address has been selected the cost of approaching an individual there remains constant. As we preferred to conduct personal interviews in our study (cf. Chapter 5), for reasons of response rate and quality of data collected, we decided to test the effect of using proxy respondents in a personal interview. We were concerned to discover not only the extent of the information loss which might result from using proxy respondents to provide information about all members of a household, but of equal importance was the need to investigate what kinds of biased reporting, if any, might result, and whether such bias or random loss could be reduced by selecting a particular household member as respondent.