Self-disclosure on Army surveys: Survey procedures and respondent beliefs related to candidness.

Effects of Amount, Format, and Location of Demographic Information on Questionnaire Return Rate and Response Bias of Sensitive and Nonsensitive Items

“…Apparently, the possibility of format-induced response bias increases as question content becomes more sensitive. In this regard, Wilson and Rosen (1975) found that job satisfaction was the least sensitive of five topics they studied (the other four were use of marijuana, use of heroin, abuse of alcohol, and dislike of members of other races). This would seem to imply that if the format manipulation can produce response bias with a topic of relatively low sensitivity such as job satisfaction, even greater response bias might be expected in surveys dealing with more sensitive material.…”

“…This particular operational definition was utilized since it is analogous to the condition of non-anonymity used in much of the response bias research cited earlier. That is, when subjects in an organization are required to identify themselves on questionnaires (Butler, 1973;Fuller, 1974;Wilson and Rosen, 1975), they thereby assume the risk that their responses may be disclosed to their administrative superiors. Data gathered by Wilson and Rosen (1975) show that possible disclosure to superiors is a very real concern of persons who answer surveys on which they can be identified.…”

“…However, if the gathering of this additional information causes return rate and/or response bias distortions in the data, it may be advisable to omit the additional demographic questions. Wilson and Rosen (1975) feel that the demographic data gathered on anonymous surveys should be greatly reduced or eliminated in order to ensure that respondents will feel that they are anonymous.…”

“…However, these advantages may be offset by the possibility that asking for specific numbers may reduce respondents' perceived anonymity such that return rate and response bias distortions result. For this reason, there have been suggestions that demographic questions should not ask for specific figures, but only for the checking of categories (Wilson and Rosen, 1975).…”