Convergent validity of methods for assessing socially desirable responding in personality items

Abstract Methods for detecting socially desirable responding (SDR) differ in terms of ease of administration and may also differ in their capacity to capture SDR. Several methods have been used to capture desirable self-evaluations, among which are the better-than-average evaluation, fake-good instructions, social desirability scale value (SDSV) and the Marlowe-Crowne Social Desirability Short Form (MCSD-SF). The purpose of this research was to evaluate agreement between these methods in two studies, using personality items. Data for Study I was collected with internet surveys sent out to university students. The surveys contained a test of the better-than-average effect on the Big Five Inventory (BFI) items, the original BFI, the MCSD-SF, the BFI with fake-good instructions and instructions to evaluate the desirability (SDSV) of each BFI item. The findings suggest that three of the four methods capture socially desirable responding in personality items (fake-good, SDSV and MCSD-SF), with correlations between methods ranging between r = .80 and .98 (indicating high convergence between methods). However, the better-than-average manipulation did not produce more SDR indicating that it cannot be used to evaluate the desirability of item content. Study II was conducted to confirm the convergent validity of the remaining three methods (fake-good, SDSV and MCSD-SF). Results supported the convergent validity of the three methods with correlations between methods ranging between r = .84 and .98.