Identifying Socially Desirable Responses in Personality Inventory

The application of self-reported personality test in a competitive environment is found to be susceptible to biased responding (Hirsh Peterson, 2008). This biased response causes difficulty in assessing an applicant’s true scores in a standard selection process (Ellingson, Sackett Hough, 1999). Therefore, this study aimed to detect the socially desirable responses from artificially differential response condition using a personality inventory. An experimental repeated measure design involved 521 students responding twice to the shortened International Personality Item Pool (IPIP) under honest and socially desirable instructions. The analysis of mean, score distribution and Rasch model’s outfit indices were examined in differentiating the socially desirable responses from the honest responses. The socially desirable condition was found to have a higher mean compared to the honest condition. The percentage of respondents in the socially desirable condition obtaining scores with two standard deviations (2SD) above the mean was two times the honest condition. Additionally, two-third of the respondents with outfit values greater than 2.0 logits were from the socially desirable condition. Based on the findings in this study, it can be concluded that the score deviation greater than 2SD above the mean and the outfit values exceeding 2.0 logit are indications of high possibility of socially desirable responses. Therefore, test users for high-stake decisions can examine extreme high scores and an individual’s inconsistent response as an initial detection of the socially desirable responses. This would help to minimize the issue of social desirability in high-stake testing.