Further Evidence on the Validity of the MMPI–2 Restructured Clinical (RC) Scales: Addressing Questions Raised by Rogers, Sewell, Harrison, and Jordan and Nichols

Abstract

The reviews by Rogers, Sewell, Harrison, and Jordan (2006/this issue), and by Nichols (2006/this issue) offer markedly contrasting appraisals of the MMPI-2 Restructured Clinical (RC) Scales introduced by Tellegen et al. (2003). The one common feature is that both reviews draw on the same atypical MMPI-2 (Butcher, Dahlstrom, Graham, Tellegen, & Kaemmer, 1989) data set for their empirical analyses, with results warranting critical scrutiny. Rogers et al.'s critique provides an evaluation of the RC Scales from the perspective of Jackson's (1970) method of test development. One significant issue in Rogers et al.'s review concerns social desirability, prompting us to clarify our own views on this topic. We also highlight and discuss problems associated with Rogers et al.'s use of the unrepresentative data set. Nichols's polemical critique neglects empirical and theoretical support for demoralization as a central construct and misconstrues as "construct drift" the purposeful process of developing the RC scales. Nichols's criticisms and proposals overlook requirements for assessing syndromes and for construct validation and even rudiments of scale development. Our reply incorporates evidence, including new findings, refuting his criticisms and confirming that demoralization is a pervasive MMPI dimension, that the RC Scales capture the major distinctive features of the original Clinical Scales, and that they generate correspondingly meaningful validity patterns.

What Do the MMPI–2 Restructured Clinical Scales Reliably Measure? Answers From Multiple Research Settings

“…These results raise concerns about whether this enterprise was successful in the case of RC7 because little variance seems to exist other than the variance shared with a first factor marker. When viewed in light of the sizable correlations reported by Tellegen et al (2006) between RCd and RC7, it becomes apparent that RC7 might be more saturated with Demoralization variance than was previously realized. The advantage of fewer items in RC7 and RCd (both consisting of 24 items in contrast to the 37 items that comprise A) is offset by the fact that both scales are redundant measures of A, and their combined 48 items are included in the overall set of RC scales.…”

“…In response to Nichols' (2006) earlier demonstration of high correlations between RC scales and extant MMPI-2 scales, Tellegen et al (2006) noted a high level of item overlap between the RC scales and the standard MMPI-2 scales. The same concern might be raised here given the substantial item overlap between the extant scales and the RC scales used in our analyses.…”

“…However, item overlap between established scales and newly developed scales does not minimize the concerns raised by these analyses; the substantial item overlap is further support for the view that five of the RC scales replicate well-established MMPI-2 scales. Moreover, to remove overlapping items and to calculate correlation coefficients without these items (as suggested by Tellegen et al, 2006) artificially deflates the correlation between these scales as they are put into real-world usage. For example, scores on RC1 and HEA are based on the 27 and 36 items that (respectively) comprise those scales, so the data presented in Table 2 reflect the relationship between the scores as used by clinicians.…”

“…Both Caldwell (2006) and Nichols have concluded that the RC scales were simply redundant measures of existing MMPI-2 scales. Tellegen et al (2006) responded that the high correlations reported by Nichols (2006) (m)ay be attributed to his reliance on a problematic data set and his failure to correct for the spurious effects of correlated measurement error owing to item overlap that resulted in inflated correlations (some of which reach or exceed the limit set by the reliabilities of the scales in question). (p. 167) In marked contrast to the high correlations between RC scales and existing MMPI-2 Content, Supplementary, and PSY-5 scales are the more modest relationships between several RC scales and the parent scales from which a "distinctive core" was distilled.…”

Profiles of body image disturbance and their external correlates

“…Previous work in psychological research indicates that demoralization can have a pervasive influence on other psychological constructs and influence the measurement of those constructs (Tellegen et al, 2006). In this sample general distress in participants may have influenced the measurement of other psychological constructs including measures of BID.…”