Comparative Validation of a Partial (Versus Full) Randomized Response Technique: Attempting to Control for Social Desirability Response Bias to Sensitive Questions
Author(s): Bill Thornton, Sat Gupta
Abstract: Social desirability response bias is prevalent in interview or questionnaire surveys involving personally sensitive questions. The present research provided for a comparative evaluation of a partial and full randomized response technique (RRT) as a statistical technique to circumvent this tendency to present oneself in a socially desirable manner. Also included for validation purposes were face-to-face, assured anonymity, and bogus pipeline conditions (the latter previously demonstrated as a means to elicit more accurate, truthful responding to sensitive items). Consistent with previous computer simulation research (Gupta et al., 2002), the present data based on actual survey respondents similarly suggests the effectiveness of a partial RRT in providing estimates closer to "accurate" responses (obtained in the anonymous and bogus pipeline conditions) and with less error of estimation than that possible using a full RRT. This advantage was evident on personally sensitive items; on non-sensitive items, there was no apparent advantage in utilizing any strategy intended to circumvent response bias.
Pages: 214-224
Download Full Article: Click here