Psychometric properties of full and reduced length forms of the balanced inventory of desirable responding under honest and faking conditions

Socially desirable responding (SDR) occurs when respondents answer questionnaires untruthfully, either to give overly positive or overly negative impressions. The Balanced Inventory of Desirable Responding (BIDR; Paulhus, 1991) is the most commonly administered inventory of SDR that includes separate scales to measure Impression Management and Self-Deceptive Enhancement.

In this study, polytomous and dichotomous scoring methods for full and shortened forms of the BIDR were rigorously compared in distributional characteristics, score consistency, convergent/discriminant validity, factor structure, and effectiveness in separating honest and faked responses. Shortened forms included all those identified in the research literature plus two new forms designed specifically to detect faking good and faking bad.

Results showed that polytomous scoring was more effective than dichotomous scoring in distinguishing individual differences for norm-referencing purposes, but frequently not as accurate in distinguishing honest from faked responses as was properly targeted dichotomous scoring. Variability and reliability of scores was greater for the full-length form, but scores for shortened forms often provided stronger factor model fits and greater accuracy in classifying honest and faked responses. Shortened forms developed by Asgeirsdottir et al. (2016) and by the author provided best classification accuracy rate for detecting faking good, and an alternative shortened form developed by the author provided best classification accuracy rate for detecting faking bad. More detailed recommendations are provided for which forms and scoring methods best serve the intended uses of BIDR scores, with broader implications discussed for self-report measures in general.