A comparison of World-Wide Web and paper-and-pencil personality questionnaires.

Does the manifestation of response set effects in World-Wide Web (WWW) questionnaire responses differ from that in paper-and-pencil (PP) questionnaire responses? Two thousand six hundred and forty-nine volunteers responded to a WWW questionnaire, and 458 volunteers responded to a PP questionnaire consisting of five personality scales. Five response sets were examined. For four of them--random response, item nonresponse, extreme response, and acquiescent response--no statistically significant differences between PP and WWW data were found. The PP administration elicited a statistically higher number of errors (uncodable responses), although eta 2 was less than .02. An analysis of interitem and interscale correlations did not differentiate WWW and PP data. Data from the Marlowe-Crowne Social Desirability Scale, the Perfectionist Self-Presentation Scale, and the Computer Anxiety Scale did not differentiate WWW and PP data in terms of either means or scale reliabilities. It was concluded that WWW data may be comparable to PP data and that the WWW is a potentially useful and valid data collection tool.