The original MMPI was developed in the late 1930s using an empirical keying approach, which means that the clinical scales were derived by selecting items that were endorsed by patients known to have been diagnosed with certain pathologies. The difference between this approach and other test development strategies used around that time was that it was atheoretical (not based on any particular theory) and thus the initial test was not aligned with the prevailing psychodynamic theories of that time. The atheoretical approach to MMPI development enabled the test to capture aspects of human psychopathology that were recognizable and meaningful despite changes in clinical theories.
The first major revision of the MMPI was the MMPI-2, which was standardized on a new national sample of adults in the United States and released in 1989. It is appropriate for use with adults 18 and over. Subsequent revisions of certain test elements have been published, and a wide variety of subscales was also introduced over many years to help clinicians interpret the results of the original clinical scales, which had been found to contain a general factor that made interpretation of scores on the clinical scales difficult. The current MMPI-2 has 567 items, all true-or-false format, and usually takes between 1 and 2 hours to complete. There is an infrequently used abbreviated form of the test that consists of the MMPI-2's first 370 items. The shorter version has been mainly used in circumstances that have not allowed the full version to be completed (e.g., illness or time pressure), but the scores available on the shorter version are not as extensive as those available in the 567-item version.
A version of the test designed for adolescents, the MMPI-A, was released in 1992. The MMPI-A has 478 items, with a short form of 350 items.
A new and psychometrically improved version of the MMPI-2 has recently been developed employing rigorous statistical methods that were used to develop the RC Scales in 2003. The new MMPI-2 Restructured Form (MMPI-2-RF) has now been released by Pearson Assessments. The MMPI-2-RF produces scores on a theoretically grounded, hierarchically structured set of scales, including the RC Scales. The modern methods used to develop the MMPI-2–RF were not available at the time the MMPI was originally developed. The MMPI-2-RF builds on the foundation of the RC Scales, which have been extensively researched since their publication in 2003. Publications on the MMPI-2-RC Scales include book chapters, multiple published articles in peer-reviewed journals, and address the use of the scales in a wide range of settings.
The original clinical scales were designed to measure common diagnoses of the era. While the descriptions of each type were originally used in assessment, the current practice is to use the numbers only.
|Number||Abbreviation||Description||What is Measured|
|1||Hs||Hypochondriasis||Concern with bodily symptoms|
|3||Hy||Hysteria||Awareness of problems and vulnerabilities|
|4||Pd||Psychopathic Deviate||Conflict, struggle, anger, respect for society's rules|
|5||MF||Masculinity/Femininity||Stereotypical masculine or feminine interests/behaviors|
|6||Pa||Paranoia||Level of trust, suspiciousness, sensitivity|
|7||Pt||Psychasthenia||Worry, Anxiety, tension, doubts, obsessiveness|
|8||Sc||Schizophrenia||Odd thinking and social alienation|
|9||Ma||Hypomania||Level of excitability|
|0||Si||Social Introversion||People orientation|
Codetypes are a combination of the one or two highest-scoring clinical scales (ex. - 8, 48). Codetypes and interaction of clinical scales can be quite complex and require specialized training to properly interpret.
The validity scales in the MMPI-2 RF are minor revisions of those contained in the MMPI-2, which includes three basic types of validity measures: those that were designed to detect non-responding or inconsistent responding (CNS, VRIN, TRIN), those designed to detect when clients are over reporting or exaggerating the prevalence or severity of psychological symptoms (F, Fb, Fp, FBS), and those designed to detect when test-takers are underreporting or downplaying psychological symptoms (L, K)). A new addition to the validity scales for the MMPI-2 RF includes an over reporting scale of somatic symptoms scale (Fs).
|Abbreviation||New in version||Description||Assesses|
|?||1||"Cannot Say"||Questions not answered|
|L||1||Lie||Client "faking good"|
|F||1||Infrequency||Client "faking bad" (in first half of test)|
|Fb||2||Back F||Client "faking bad" (in last half of test)|
|VRIN||2||Variable Response Inconsistency||answering similar/opposite question pairs inconsistently|
|TRIN||2||True Response Inconsistency||answering questions all true/all false|
|F-K||2||F minus K||honesty of test responses/not faking good or bad|
|S||2||Superlative Self-Presentation||improving upon K scale, "appearing excessively good"|
|Fp||2||Psychiatric Infrequency||Frequency of presentation in clinical setting|
|Fs||2 RF||Infrequent Somatic Response||Overreporting of somatic symptoms|
To supplement these multidimensional scales and to assist in interpreting the frequently seen diffuse elevations due to the general factor (removed in the RC scales) were also developed, with the more frequently used being the substance abuse scales (MAC-R, APS, AAS), designed to assess the extent to which a client admits to or is prone to abusing substances, and the A (anxiety) and R (repression) scales, developed by Welsh after conducting a factor analysis of the original MMPI item pool.
Dozens of content scales currently exist, the following are some samples:
|Es||Ego Strength Scale|
|OH||Over-Controlled Hostility Scale|
|MAC||MacAndrews Alcoholism Scale|
|MAC-R||MacAndrews Alcoholism Scale Revised|
|APS||Addictions Potential Scale|
|AAS||Addictions Acknowledgement Scale|
|SOD||Social Discomfort Scale|
|TPA||Type A Scale|
|MDS||Marital Distress Scale|
Unlike the Content and Supplementary scales, the PSY-5 scales were not developed as a reaction to some actual or perceived shortcoming in the MMPI-2 itself, but rather as an attempt to connect the instrument with more general trend in personality psychology. The five factor model of human personality has gained great acceptance in non-pathological populations, and the PSY-5 scales differ from the 5 factors identified in non-pathological populations in that they were meant to determine the extent to which personality disorders might manifest and be recognizable in clinical populations. The five components were labeled Negative Emotionality (NEGE), Psychoticism (PSYC), Introversion (INTR), Disconstraint (DISC) and Aggressiveness (AGGR).
Like many standardized tests, scores on the various scales of the MMPI-2 and the MMPI-2-RF are not representative of either percentile rank or how "well" or "poorly" someone has done on the test. Rather, analysis looks at relative elevation of factors compared to the various norm groups studied. Raw scores on the scales are transformed into a standardized metric known as T-scores (Mean or Average equals 50, Standard Deviation equals 10), making interpretation easier for clinicians.
The MMPI-2 and MMPI-2-RF should only be scored and interpreted by individuals with graduate level training in either clinical, experimental or I/O psychology, and who have received specialized MMPI-2 training. Individuals who are not trained in psychological assessment and scoring should not attempt to score or interpret the MMPI-2, as accurate scoring and interpretation requires knowledge of the test itself, standardized testing theory, the various subscales in combination (test profile) and correspondence of results to diagnosis. Test manufacturers and publishers ask test purchasers to prove they are qualified to purchase the MMPI/MMPI-2/MMPI-2-RF and other tests.
Some questions have been raised about the RC Scales and the forthcoming release of the MMPI-2-RF, which eliminates the older clinical scales entirely in favor of the more psychometrically appealing RC scales. The replacement of the original Clinical Scales with the RC scales has not been met with universal approval, and has warranted enough discussion to prompt a special issue of the academic Journal of Personality Assessment (Vol 87, Issue 2, October 2006) to provide each side with a forum to voice their opinions regarding the old and new measures.
Individuals in favor of retaining the older Clinical scales have argued that the new RC scales are measuring pathology which is markedly different than that measured by the original clinical scales. This claim is not supported by results of research, which has found the RC scales to be cleaner, more pure versions of the original clinical scales because 1) the interscale correlations are greatly reduced and no items are contained in more than one RC scale and, 2) common variance spread across the older clinical scales due to a general factor common to psychopathology is parsed out and contained in a separate scale measuring demoralization (RCdem). Critics of the new scales argue that the removal of this common variance makes the RC scales less ecologically valid (less like real life) because real patients tend to present complex patterns of symptoms. However, this issue is addressed by being able to view elevations on other RC scales that are less saturated with the general factor and, therefore, are also more transparent and much easier to interpret.
Critics of the RC scales assert they have deviated too far from the original clinical scales, the implication being that previous research done on the clinical scales will no longer be relevant to the interpretation of the RC scales and the burden of proof should be on the RC scales to demonstrate they are clearly superior to the original clinical scales. Proponents of the RC scales assert that research has adequately addressed those issues with results indicating that the RC scales predict pathology in their designated areas better than their concordant original clinical scales while using significantly fewer items and maintaining equal to higher internal consistency reliability and validity, and are as good or better at identifying the core elements of the original clinical scales; further, unlike the original clinical scales, the RC scales are not saturated with the primary factor (demoralization, now captured in RCdem) which frequently produced diffuse elevations and made interpretation of results difficult; finally, the RC scales have lower interscale correlations and, in contrast to the original clinical scales, contain no interscale item overlap.
In March 2008 a front page article in the Wall Street Journal exposed what it claimed to the lack of scientific validity of the "fake bad" scale, which is used in courts as argument for malingering in injury litigation. According to the article two Florida judges barred use of the scale after special hearings on its scientific validity.
The article reports that the scale was developed by psychologist Paul Lees-Haley, who works mainly for defendants (insurance companies etc.) in personal injury cases. The article reports that in 1991 Lees-Haley paid to have an article supportive of his scale published in Psychological Reports, which the Wall Street Journal described as "a small Montana-based medical journal." The scale was introduced in MMPI after a review of the literature. This review was considered flawed by its critics because at least 10 of 19 studies reviewed were done by Lees-Haley or other insurance defense psychologists, while 21 other studies critical of the test were excluded from the review.
One of the critics of the "fake bad" scale is retired psychologist James Butcher, who found that more than 45 percent of psychiatric patients he studied had Fake Bad Scale scores of 20 or more, which according to the "fake bad" scale meant they were malingering. Butcher contends that it is unlikely that so many psychiatric patients misled doctors. The article qoutes Butcher concluding:
This is great for insurance companies, but not great for people.
The article claimed that disagreeing with statements like "My sex life is satisfactory" or "I seldom or never have dizzy spells" earned malingering points for the test takers.