4/13 Simms Notes
Main purpose of the article
The study evaluates the psychometric properties of the Restructured Clinical (RC) scales for the MMPI–2.
The authors wanted to see whether the RC scales improve on long-standing problems in the original MMPI–2 clinical scales, especially:
conceptual overlap
item overlap
heterogeneity within scales
poor discriminant validity
Why the study was needed
The original MMPI/MMPI–2 clinical scales were developed mainly through empirical criterion-keying, which created scales that often:
shared many overlapping items
correlated too highly with each other
contained mixed or multidimensional content
Because of that, the meaning of a moderate elevation on a clinical scale could be unclear.
The RC scales were designed by Tellegen and colleagues to solve these problems, but at the time of this paper there had been no published independent evaluation of them.
Core criticism of the original clinical scales
The article emphasizes two main weaknesses of the standard clinical scales:
item overlap across scales, which inflates intercorrelations
heterogeneous content within scales, which makes interpretation difficult
Example given in the article:
Scale 3 (Hysteria) mixes somatic complaints with content suggesting good adjustment and social conformity
so the same T score could reflect very different underlying item patterns
How the RC scales were developed
Tellegen et al. created the RC scales in several stages:
first, they extracted a broad nonspecific distress factor called Demoralization (RCd)
then they identified a more coherent core component of each clinical scale
then they built “seed scales”
finally they selected MMPI–2 items that maximized convergent and discriminant validity
The goal was to create more distinct, more homogeneous, and more interpretable scales.
The RC scales included
The final RC scales were:
RCd: Demoralization
RC1: Somatic Complaints
RC2: Low Positive Emotions
RC3: Cynicism
RC4: Antisocial Behavior
RC6: Ideas of Persecution
RC7: Dysfunctional Negative Emotions
RC8: Aberrant Experiences
RC9: Hypomanic Activation
A major point is that RC3 differs sharply from original Scale 3/Hysteria, because the restructuring removed much of the somatic content and left a cynicism-focused scale.
Main samples used
The study examined two independent samples:
Psychology clinic clients: N = 285
Military veterans: N = 567
This cross-sample design allowed the authors to test whether results replicated across different populations.
Measures used for validation
Participants completed the MMPI–2.
The study also used:
the SNAP (Schedule for Nonadaptive and Adaptive Personality) for trait and personality disorder dimensions
the SCID for DSM–IV Axis I diagnoses in the military veteran sample
These measures allowed the authors to compare the RC scales and standard clinical scales against both:
more unidimensional trait measures
more multidimensional diagnostic criteria
Overall descriptive pattern
Table 1 shows that mean T scores for both scale sets were mostly in the 50s and low 60s, with the clinic sample generally showing more elevation than the veteran sample.
This fits expectations given that the clinic sample was more actively treatment-seeking.
Internal consistency findings
The RC scales were as internally consistent as or slightly more consistent than the standard clinical scales.
This is especially notable because the RC scales used, on average, about 60% fewer items.
Median alphas for the eight substantive RC scales were:
.83 in the clinic sample
.77 in the veteran sample
Comparable medians for the standard clinical scales were:
.78 in the clinic sample
.76 in the veteran sample
Measurement efficiency
The RC scales generally improved measurement efficiency.
The article highlights especially large reliability gains for RC6 and RC9:
these improved from alphas of about .60 in the original versions to around .80 in the RC versions
This supports the argument that the RC scales are more homogeneous and efficient.
Convergence with original clinical scales
Most RC scales correlated strongly with their original parent scales.
This suggests they retained a meaningful link to the original scales.
The major exception was RC3, which correlated only about .12 to .14 with original Scale 3 across samples.
This means RC3 and Hysteria are not interchangeable.
Discriminant validity / intercorrelations
The RC scales were generally less intercorrelated than the standard clinical scales.
Mean intercorrelations:
RC scales: about .42
standard scales: about .60 in the clinic sample and .55 in the veteran sample
This indicates that the RC scales are more distinct from one another and therefore easier to interpret separately.
Specific examples of improved distinctiveness
Several highly overlapping standard clinical scale pairs became less redundant in the RC set.
Examples:
Scales 7 and 8 correlated .91, but RC7 and RC8 were lower at about .65/.62
Scales 2 and 7 correlated .79/.76, but RC2 and RC7 were lower at about .55/.47
Scales 4 and 8 correlated .72/.72, but RC4 and RC8 dropped to about .35/.33
These are important signs of improved discriminant validity.
Important caution: RCd and RC7
Even though the RC scales were less intercorrelated overall, they still retained a noticeable nonspecific component.
The clearest example is the very high correlation between RCd (Demoralization) and RC7 (Dysfunctional Negative Emotions):
about .78 to .79 across samples
The authors note that this suggests substantial overlap and raises questions about whether both scales are needed in their current form.
General pattern of personality correlates
Both the standard and RC scales still showed some nonspecific relations with broad negative affectivity / neuroticism.
However, the RC scales showed:
fewer very large correlations
more moderate and conceptually focused correlations
This suggests that the restructuring reduced, but did not eliminate, broad distress contamination.
Validity pattern comparison
For most scale pairs, the validity patterns of the RC and original clinical versions were fairly similar.
All but one RC scale showed moderate to high similarity in validity pattern with its parent scale.
Again, the exception was RC3, which showed a markedly different pattern from Scale 3.
Meaning of RC3
RC3 appears to measure cynicism and mistrust, not the traditional Hysteria construct.
Its strongest SNAP correlate was Mistrust, not broad distress or somatic complaint.
The authors repeatedly stress that RC3 is conceptually different from Scale 3 and should not be interpreted the same way.
RC6 and RC8 findings
RC6 (Ideas of Persecution) was more specifically related to Mistrust, which fits its intended meaning.
RC8 (Aberrant Experiences) was more specifically related to Eccentric Perceptions than the original Scale 8.
In contrast, original Scale 8 correlated broadly with many SNAP traits, suggesting more diffuse psychopathology rather than a clearer target construct.
These results support improved conceptual clarity for RC6 and RC8.
RC4 and RC9 findings
RC4 (Antisocial Behavior) was more specifically related to:
disinhibition
aggression
manipulativeness
antisocial and borderline personality pathology
Compared with original Scale 4, RC4 appeared much less saturated with nonspecific distress.
RC9 (Hypomanic Activation) also showed links to externalizing and interpersonal dominance/manipulation-related traits, including:
manipulativeness
aggression
exhibitionism
some unusual perceptual experiences
The authors note that RC4 and RC9 together may capture much of the traditional 4–9/9–4 profile pattern, but in a clearer way.
RC2 findings
RC2 (Low Positive Emotions) showed a cleaner pattern than original Scale 2 (Depression).
Both related to negative temperament and low positive temperament, but RC2 was more specifically tied to:
low positive affect
anhedonia
detachment
This fits contemporary models of depression better than the broader and more diffuse original Scale 2.
SNAP personality disorder correlates
The article also related the scales to SNAP personality disorder dimensions.
The RC scales generally produced clearer and somewhat more specific relations with relevant PD dimensions.
The article specifically notes that the RC scales showed an improvement in prediction of these more complex criteria as well, not just simple trait measures.
SCID diagnostic correlates
In the military veteran sample, the SCID was used to examine relations with Axis I disorder categories.
Results were mixed:
Depressive disorders were predicted strongly by Scale 2 and RCd
RC2 predicted depression, but less strongly than both RCd and Scale 2
Anxiety disorders were predicted somewhat better by original Scale 7 than by RC7
Somatoform disorders were predicted similarly by Scale 1 and RC1
Substance use disorders were predicted better by RC4 than by original Scale 4
Interpretation of the SCID results
The SCID findings suggest that the RC scales do not uniformly outperform the original clinical scales on all diagnostic outcomes.
In some areas, especially depression and anxiety, the older scales or RCd remained as good or better predictors.
The strongest RC advantage at the diagnostic level was for substance use disorders, especially through RC4.
Incremental validity
The authors tested whether the RC scales added predictive value beyond the original clinical scales, and vice versa.
Overall, the RC scales showed somewhat greater incremental utility than the standard scales.
For SNAP trait and PD measures, the RC scales generally added more than the original scales did.
The article states that the RC scales’ incremental effect in predicting SNAP PD scales was:
about 2.3 times larger in the veteran sample
about 1.4 times larger in the clinic sample
However, for the SCID disorder classes, incremental differences were much smaller and usually nonsignificant except for substance use disorders.
Main conclusion about interchangeability
A central conclusion is that the RC scales and standard clinical scales cannot be used interchangeably.
Even when they correlate strongly, they do not always mean the same thing.
This is especially true for RC3, and to some degree also for scales where restructuring substantially narrowed the construct.
Response to critics
The article acknowledges criticism from Nichols, who argued that:
the RC scales may be redundant with other MMPI–2 scales
the construction of RCd is questionable
the validation criteria used might unfairly favor simpler, more unidimensional scales
Simms and colleagues try to address this by validating the RC scales against both:
more discrete trait measures
more complex diagnostic criteria like SCID disorder classes
Their results suggest the RC scales do show real psychometric strengths, but not universal superiority.
Authors’ overall interpretation
The RC scales appear to be:
more efficient
more distinct
often more conceptually clear
sometimes more incrementally useful
But they still retain some broad distress contamination, especially around RCd and RC7.
They also require new interpretive research, because old MMPI/MMPI–2 interpretive lore does not always transfer directly.
Most important takeaways by scale
RC1: a coherent somatic complaints scale; useful for somatoform features
RC2: cleaner measure of low positive affect / anhedonia than Scale 2
RC3: major reconceptualization; measures cynicism/mistrust, not classic Hysteria
RC4: clearer measure of antisocial/disinhibited/externalizing tendencies
RC6: more specific persecutory ideation / mistrust content
RC7: broad negative emotionality, but may overlap too much with RCd
RC8: clearer measure of aberrant perception/experience
RC9: hypomanic activation with links to manipulativeness, aggression, and exhibitionism
Bottom-line conclusion
The article concludes that the RC scales show several psychometric advantages over the original MMPI–2 clinical scales:
comparable or better reliability
less intercorrelation
clearer construct relations
somewhat greater incremental validity
At the same time, they are not simply shorter replacements for the original scales.
They should be interpreted as related but distinct measures, with special caution around RC3 and the overlap between RCd and RC7.