Modeling the Balanced Inventory of Desirable Responding: Evidence in Favor of a Revised Model of Socially Desirable Responding

Abstract

Paulhus (1984) proposed a 2-factor model of socially desirable responding (SDR) and created the Balanced Inventory of Desirable Responding (BIDR) to capture the 2 dimensions: self-deceptive enhancement and impression management. However, the 2-factor model has yet to be supported via confirmatory factor analysis (CFA) of the BIDR. Paulhus and Reid (1991) proposed a revised model of SDR, which bifurcated the self-deceptive enhancement dimension into 2 factors: self-deceptive enhancement (SDE; positively keyed items) and self-deceptive denial (SDD; negatively keyed items). Thus, it was considered useful to test the revised SDR model on the BIDR, as this might have implications for the manner in which the BIDR should be scored. Additionally, as the BIDR subscales might be positively correlated, it was considered useful from a both a theoretical and practical perspective to test the possibility that the BIDR might measure a general SDR process. Based on a sample of 466 adults, Paulhus and Reid's revised model of SDR was largely supported when tested via CFA on a bifactor model, which included a first-order general SDR factor and 2 nested factors. Thus, applied researchers might consider using total BIDR composite scores, a self-deceptive enhancement composite score (positively keyed items), and impression management composite scores. However, ideally, researchers would use a bifactor model to test substantive hypotheses, as the bifactor model partitions true score variance into unique sources, which facilitates less ambiguous interpretations of effects.

Notes

There is a commercially available version of the BIDR known as the Paulhus Deception Scales (PDS; Paulhus, 1998). It is also known as the BIDR version 7. The only difference between version 6 and version 7 of the BIDR is that two items within the SDE scale were changed. As the items associated with the BIDR version 6 were published in a book chapter (i.e., Paulhus, 1991), it appears to be much more frequently used in research than version 7.

The manual associated with the PDS (Paulhus, 1998) states that the questionnaire can also be used based on a 5-point rating scale ranging from 1 (not true to 5 (very true). In this case, a response of 5 is recoded into 1 and all other responses are coded 0.

Technically, a tetrachoric correlation is a special case of a polychoric correlation (Flora & Curran, 2004).

There is also an unpublished CFA investigation of the BIDR based on a sample of 219 undergraduate students (Callaway, 2009). The items were scored dichotomously, and similarly to Leite and Beretvas (2005), a CFA on polychoric correlations with WLSMV estimation was performed. The oblique two-factor model again failed to be found to be associated with acceptable levels of model fit (CFI = .75, TLI = .76). Unfortunately, Callaway did not report the correlation between the SDE and IM latent variables.

To help avoid confusion, the first 20 items within the BIDR that have until now been referred to as SDE will now be referred to as SDS. Although SDS stands for self-deceptive scales, within the context of a single dimension, it could be considered to stand for self-deceptive style. The acronym SDE will be reserved for the subdomain within the SDS portion of the BIDR proposed by Paulhus and Reid (1991).

Zumbo et al. (2007) conducted their simulation study based on data with points ranging from 2 to 7. They used the term ordinal Cronbach's α across the whole range (including 2 points) and this practice is continued in this article.

As described in the Method section, the ω estimates were derived from a model with 12 item parcels as indicators rather than the 40 individual items. As Model 7r was found to be the most acceptable for the dichotomously scored data, the coefficient ω estimates were estimated from an item parcel version of that model, which was found to be associated with acceptable levels of model close-fit: χ²(45) = 69.18, p = .012, RMSEA = .034, CFI = .978, TLI = .968. Furthermore, all of the factor loadings were statistically significant (p < .05). Full results are available on request. A 12-item-parcel model version of the continuously scored data was not estimated, as no continuously scored 40-item model was found to be associated with acceptable levels of model close-fit.