The Comparability of MMPI and MMPI-2 Scales and Profiles

The revised form of the Minnesota Multiphasic Personality Inventory (MMPI-2) incorporates a number of changes that necessitate an investigation into the comparability of its scale scores and clinical profile to those of the original MMPI. In the current study, differences between scores obtained by 189 college students who completed both the original and revised forms of the MMPI were compared with differences between scores obtained by 188 other students who were administered the original form twice. Results indicated substantial congruence between the cross-administration stability of scores and profiles obtained by the two groups.

Following an extensive restandardization project, a revised and updated form of the Minnesota Multiphasic Personality Inventory (MMPI-2; Butcher, Dahlstrom, Graham, Tellegen, & Kaemmer, 1989 ) has been completed. ¹ As outlined and detailed in its manual, a number of changes have been incorporated in the revised instrument. One such change involved the deletion of a small number of objectional items from some of the validity, clinical, and supplemental scales. These items included content that over the years has been found to be objectionable by users of the MMPI (e.g., religious items, items having to do with bowel and bladder movements). Sixty-eight of the remaining items were rewritten to update obsolete language, correct sexist or male-oriented language, or otherwise make the items more readable and understandable.

A major change in the MMPI-2 involved the development of new norms for all of the scales and a new type of T score, uniform T scores, for the eight clinical scales ( Butcher et al., 1989 ). Briefly, uniform T scores were developed so that all eight clinical scales would be similarly distributed. This was achieved by identifying a composite or average distribution among the eight scales and adjusting each individual scale's distribution to fit this composite distribution. The new norms, based on a national census-matched sample of 2,600 men and women, were found to have an attenuating effect on MMPI-2 T scores so that the restandardization committee recommended lowering the criterion for establishing interpretive elevation in a scale from a T score of 70 to one of 65.

Ben-Porath and Butcher (1989) demonstrated the psychometric stability of the 68 rewritten MMPI-2 items by comparing their contribution to the scales on which they are scored to that of the originally worded items. We conducted the current study in order to evaluate the effect of other changes introduced in the MMPI-2 on the stability of the 13 basic scales and 8 supplemental scales that were retained in the MMPI-2.

A total of 403 undergraduate college students volunteered to participate in this study in return for extra credit in an introductory psychology class. Six subjects failed to return for the second session in this study. Of the 397 remaining subjects, those who produced more than 15 "cannot says," a raw score greater than 20 on F (Infrequency), or a raw score greater than 20 on the MMPI-2 F _B (Back- F ) validity scale, which is designed to detect unusual responding in the latter part of the booklet ( Butcher et al., 1989 ), were not included in this study. Twenty subjects met one or more of these exclusionary criteria, leaving a total of 199 women and 178 men. Subjects' ages ranged from 17 to 46 years, with a mean of 19.8 years.

Two forms of the MMPI were administered in this study. The first was the regularly used group form, and the second was the experimental form MMPI-AX , which is a 704-item form developed for the purpose of the MMPI Restandardization Project ( Butcher et al., 1989 ). The AX form included all 550 of the original MMPI items plus an additional 154 experimental items. The revised MMPI-2 contains 567 items, all of which appear on the MMPI-AX . T scores for all of the subjects who completed Form AX were derived by scoring only those items that appear on the MMPI-2 and using the new norms.

Subjects were randomly assigned to one of two groups. Those in one group completed the original form of the MMPI twice to serve as the baseline for comparing differences between the original and revised forms. Roughly half of the subjects in the second group completed Form AX first and then the original MMPI, and the second half were administered the two forms in the opposite order. The average interval between administrations was 9 days, with a range of 1 to 2 weeks.

The first set of analyses conducted in this study was designed to examine the stability of scores on the 13 basic validity and clinical scales and the 8 supplemental scales that were retained in the MMPI-2 ( Butcher et al., 1989 ). Test—retest correlations were computed and compared for raw scores on all 21 scales in the two groups separately for men and women. All of these correlations appear in Table 1 .

Correlations between scores obtained on administration of the original form twice (O—O) and scores obtained on the original and revised forms (O—R) were compared by means of a Z test for the significance of the difference between two Pearson product-moment correlations. The level of significance was set at a two-tailed level of .01, which, given that there were 42 such tests conducted, would tend to overidentify discrepancies between the stabilities of scores in the two groups. Of the 42 comparisons that were conducted, only 2 correlations, on scales F and Es (Ego Strength) for women, differed significantly.

With respect to the difference found for the F scale, examination and comparison with the correlations found for the men in our sample suggests that a somewhat higher than expected level of stability for the O—O group of women might account for this finding. The correlation obtained by the O—R women is more in line with the ones found for both groups of men. The level of stability found on F for the O—R women in our sample is also well within the range of stability typically found for F with the original MMPI in normal samples with a 1-week test—retest interval ( Dahlstrom, Welsh, & Dahlstrom, 1975 ; Graham, 1987 ), in the MMPI-2 normative sample ( Butcher et al., 1989 ), and in a national sample of college students ( Butcher, Graham, Dahlstrom, & Bowman, in press ).

The attenuation in stability found for Es appears to reflect a bona fide change in the stability of this scale because a similar (though nonsignificant) pattern was found for the men in our sample. Such a change is not unexpected because the Es scale lost more items on the revised form than any other scale (a total of 14).

A second set of analyses conducted in this study was designed to provide a more detailed examination of the stability of the clinical profile in the revised form. For this purpose, several characteristics of the clinical profile (made up of the eight traditional clinical scales) were compared in the two samples. Table 2 presents the results of these comparisons.

The first characteristic of the clinical profile compared in the two samples was the percentage of subjects with the same highpoint in both administrations. For both men and women these percentages did not differ significantly (on a Z test of the significance of the difference between two proportions), and they were comparable with those reported in previous studies of the stability of MMPI high-points ( Graham, 1987 ).

The second characteristic of the clinical profile compared in the two samples was the stability of two-point codes (i.e., the two highest scores on the profile regardless of elevation). Here, the difference in two-point stability between the O—O and O—R women approached statistical significance in a two-tailed test ( Z = 1.92, p = .054). A comparison of the women's results with the percentages of two-point stability found for the men in our sample suggests that the O—O women in our current sample achieved a higher than usual level of stability, whereas the O—R women's level of stability was quite similar to that found in previous studies of the 1-week stability of two-point codes in the original MMPI ( Graham, 1987 ).

A third characteristic of the clinical profile compared in the two samples was the percentage of subjects classified as having a normal-limits profile, defined as one in which T scores on all eight clinical scales fell below 70 in the original MMPI and below 65 using the MMPI-2 uniform T scores in the revised form. This reflects the recommendation by the restandardization committee to begin interpreting elevations at a level of 65 with the MMPI-2 norms ( Butcher et al., 1989 ). These percentages (shown in Table 2 ) did not differ significantly on a Z test of the differences between two proportions.

A fourth characteristic of the profile compared between the two samples was the percentage of subjects producing the same number of elevated scales in both administrations of the MMPI. These percentages (also shown in Table 2 ) did not differ significantly based on the same statistical test.

A final analysis conducted in order to compare the clinical profiles produced by the two samples involved the computation of the D ² index, which, as described by Wiggins (1973) , provides a measure of the similarity in elevation, shape, and scatter between two profiles. D ² was computed by summing the squares of the differences between scores on each of the eight scales across the two administrations for each subject. To reflect the change in the criterion for elevation from 70 in the MMPI to 65 in the MMPI-2, 5 points were added to the MMPI-2 uniform T scores. The means and standard deviations found for the men and women in our sample appear in Table 2 . A t test comparing the D ² index found for the O—O and O—R groups showed the difference between the groups to be nonsignificant in both men and women: t (176) = 1.42, and t (197) = 1.48, respectively.

With respect to the clinical profile, here too our findings demonstrate the comparability of the revised form when applying the MMPI restandardization committee's recommendation to lower the criterion for an elevated T score from 70 to 65 when using the MMPI-2 uniform T cores ( Butcher et al., 1989 ). Despite slight modifications in the clinical scales and the use of newly normed and somewhat differently derived T scores, differences between profiles obtained by subjects who were administered the original and revised forms did not exceed those found in subjects who completed the original MMPI twice.

A frequently raised question when college students are used as subjects in psychological investigations is what extent the findings obtained in such studies generalize to other samples. A study recently conducted by Butcher et al. (in press) demonstrates substantial similarity on the MMPI-2 between a national sample of college students and the MMPI-2 normative sample. On the basis of these findings it appears safe to assume that the results reported in the current study will likely generalize to other normal populations. Further studies will be needed to ascertain whether these findings also generalize to clinical samples.

1

The revised MMPI-2 and manual were published in August 1989. The data presented in this article were collected subsequent to the completion of data collection for the restandardization project and are not part of that study.

Table 1. Test-Retest Correlations for Basic and Supplemental Scales

Table 2. Comparison of Profile Characteristics