We examine the conceptual and empirical merits of concerns Flett et al. recently raised about the Almost Perfect Scale-Revised (APS-R), specifically that items on the APS-R Discrepancy should be separated into a "Pure" Discrepancy factor and a Dissatisfaction factor. Limitations in the logic and findings of that critique are summarized. We replicate and extend Flett et al.'s study with results from two samples: (a) college freshmen STEM students ( N = 279) and (b) doctoral students in a national sample ( N = 529). Confirmatory factor analyses indicated that the alternative measurement models could be fit to the data, but were not practical improvements over the original APS-R factor model: Alternative discrepancy factors failed to demonstrate discriminant validity, nor did they have meaningfully different patterns of associations with numerous criterion variables (i.e., stress, emotion regulation, rumination, adult attachment, and life satisfaction). Thus, a data-based answer to the question of how perfectionistic discrepancy should be assessed is to stay the course with confidence using the original APS-R.
Keywords: confirmatory factor analysis; discriminant validity; perfectionism.