A meta-analytic study of social desirability distortion in computer-administered questionnaires, traditional questionnaires, and interviews

A meta-analysis of social desirability distortion compared computer questionnaires with traditional paper-and-pencil questionnaires and face-to-face interviews in 61 studies (1967-1997; 673 effect sizes). Controlling for correlated observations, a near-zero overall effect size was obtained for computer versus paper-and-pencil questionnaires. With moderators, there was less distortion on computerized measures of social desirability responding than on the paper-and-pencil measures, especially when respondents were alone and could backtrack. There was more distortion on the computer on other scales, but distortion was small when respondents were alone, anonymous, and could backtrack. There was less distortion on computerized versions of interviews than on face-to-face interviews. Research is needed on nonlinear patterns of distortion, and on the effects of context and interface on privacy perceptions and on responses to sensitive questions.