Physicians have long been aware of the importance of assessing the psychologic status of their patients as well as carrying out careful physical examinations and laboratory studies. However, the subjective nature of the interview process and the length of time necessary for a sensitive clinical interview make such assessment difficult. This difficulty led to the development of the questionnaire later known as the Minnesota Multiphasic Personality Inventory (MMIM). It was believed that a self-report questionnaire that briefly covered the topics included in a series of clinical interviews could be used as an adjunct to the physical and mental status examinations and thus could aid the diagnostician by characterizing salient features of the patient's personality.
In its present form the MMPI is a questionnaire composed of 5 50 di fièrent true-false statements that describe a variety of thoughts, feelings, attitudes, and prior life experiences as well as physical and emotional symptoms. The privacy afforded by this paper-and-pencil sell-report questionnaire, the economy and ease of administration and scoring, and the usefulness of the information provided all have contributed to the acceptance of the MMPI as a screening device and clinical tool in healthcare settings. The MMPI currently is (he most widely used and the most thoroughly researched of the objective personality assessment instruments available.1
The inception of (he MMPI occurred in 1937 through a collaboration between a psychologist. Starke R. Hathaway, and a neurologist and psychiatrist. John C. McKinley, in response to needs expressed by physicians. In one of their early papers. McKinley and Hathaway stated, '"competent internists have estimated variously thai 30% to 70% of lhe ambulatory patients who appear for medical attention, come primarily because of one or more complaints that turn out to be psychoneurotic in nature.'" figures with which there is likely to be little disagreement today.- In addition to ils clinical usefulness, the new assessment instrument was believed to have potential lor application in research - for example, in evaluating the effectiveness of insulin shock therapy which had gained widespread use at that time. Hathaway and McKinley also believed that "one might devise a personality test which, like intelligence tests, would somewhat stabilize the identification of the illness and provide an estimate of its severity. Toward this problem the MMPI research was initiated.'"5
DEVELOPMENT OF THE MMPI
Hathaway and McKinley were not the first people in the mental health field to use a self-report or questionnaire format for personality assessment; more than 50 years of such research preceded them. The earliest efforts in this area were given impetus during World War I when concern about the prevalence of mental disorders in recruits led to suggestions that draftees should be screened for emotional fitness. Although the personality assessment questionnaires developed at that time were utilized successfully during the war. they fared poorly as psychodiagnostic aids when evaluated for usefulness during peacetime. These early inventories were weakened by an excessive dependence on rational construction and on the face validity of the items. Thus, a question might be selected or composed for inclusion on the basis of the author's beliefs about neuroticism. psychosis, or mental health. Furthermore, it was believed that the more items endorsed by the subject, the greater was the strength of the trait within the individual. However, as research evidence accumulated, it became apparent that these assumptions were not tenable. There were significant contradictions among the concurrent behavior of the subjects, what they reported to interviewers about themselves, and their responses (o the items ol the questionnaire. Some of these discrepancies represented frank dissimulation or malingering; others reflected facets of interpersonal style or self-concept.
Hathaway and McKinley were aware of the faulty assumptions of face validity and rational test construction which characterized lhe early assessment efforts and pursued a different course. They believed that a subject's response to any item, or collection of items, should be viewed only as a small bit of behavior or perhaps as a behavioral sign. The meaning to be assigned to these bits or signs was to be determined empirically. Thus, a subject's endorsement of certain items might not represent historical fact or current self-concept but instead might represent a small hit of similarity to other people who had answered those items in lhe same way. This approach to questionnaire construction decreased contamination from the theoretic or unconscious bias that might affect the construction of the questionnaire or the interpretation of its results.
The first step in constructing the MMPI. originally known as the "Medical and Psychiatric Inventory" and somewhat later as the "Multiphasic Personality Scbedule," was the accumulation of more than 1 ,000 potential items. Items for this initial pool were obtained from existing psychiatric examination forms, structured interview outlines, textbooks, prior scales of personal and social attitudes, and opinions of experienced clinicians. By deleting duplicates and using clinical judgment, the pool was reduced to 504 items. Later, 9 items were dropped and 55 items related to patterns of masculinefeminine interest were added, bringing the total to its present number of 550 statements. The items in the MMPI cover 26 different symptom categories and are listed by content in the Table.
In its original form the MMPI was composed of a deck of small cards, each of which carried a single MMPl item. Each of the statements was written in the first person singular, common English idioms were used, reading level was simplified, and no interrogative sentences were used; grammatical form occasionally was sacrificed for simplicity. This deck of 550 cards was randomizd by shuffling and then given to the patient who was asked to sort the cards into boxes marked "True." "False," and "Cannot Say."
The patient's responses were then scored and the results were plotted on a graph to form a profile of different scales. Each of the scales was thought to characterize important diagnostic categories from the psychiatric literature of that day. The total score for each MMPI scale measured the degree of agreement between the patient's pattern of responses and the pattern obtained from members of homogeneous criterion groups on which the scales had been developed, each of which carried a specific psychiatric diagnosis and known behavioral characteristics. The items for each scale were selected by their ability to differentiate among normal people and medical or psychiatric patients of various types.
The MMPI also differed from prior personality assessment questionnaires by the inclusion of four additional validity scales (?, L, F, and K). These validity scales were constructed to provide information about the confidence the clinician might have in the profile by assessing the patient's degree of cooperation, comprehension, tendencies toward frank dissimulation or malingering, and general response style.
CONSTRUCTION OF MMPI SCALES
Homogeneous criterion groups of psychiatric patients were used in developing the nine basic clinical scales of the MMPI.3-10 The original scales, listed in the order in which they appear on the MMPI profile sheet and with the psychiatric nomenclature of 1940 are: Scale 1, Hs, hypochondriasis; Scale 2. D. depression; Scale 3, Hy. hysteria; Scale 4, Pd. psychopathic deviate; Scale 6, Pa. paranoia; Scale 7, Pt, psychasthenia; Scale 8, Sc, schizophrenia; and Scale 9, Ma, hypomania. A number of items related to interest patterns were added later and contributed to the development of scale 5, Mf, masculinityfemininity. Subsequently, scale 0, Si. social introversionextroversion was developed on a normal sample of high school and college students." This scale, the only one to be derived from a nonpsychiatric sample, was found to be so useful that it was added to the MMPI profile as the last of the 10 standard clinical scales.
MMPI CONTENT CATEGORIES
An empirical approach to item selection and scale construction has been the hallmark of MMPI research. Typically, three steps were involved in the selection of items composing each clinical scale. First, items were tentatively included if there was a statistically significant difference between the percentage of people responding "true" in a criterion group of patients and in the normal sample. Second, responses to the items from the preliminary scale were contrasted with the response frequencies in a group of miscellaneous psychiatric patients but excluding those carrying the same diagnosis as the criterion group, to further select items. The third and final step was the cross validation of each scale by its application to a new series of clinical cases. Usually, additional validation was carried out by a comparison with responses in a group of general medical patients who carried no psychiatric diagnosis.
Scoring was performed by accumulating the number of "true" or "false" responses that were made in the scoreable direction. Differential weighing of some items to reflect their clinical usefulness or relative importance diagnostically was also investigated by Hathaway and McKinley who noted that:
we tried a variety of item-weight systems. Nothing that we could find improved discrimination enough even to compensate for the increased complexity of scoring in contrast to the simplicity of unit weights . . . We lost test power whenever we did anything but add empirically pertinent items. Further, adding 'extra-good items" did not work appreciably better than adding items with only moderate reliability. We were eventually driven to the simple generalization which had provided our initial point of departure: the more differential items a person answered like some criterion group, the more like the group he appeared in other ways - without there being any requirement thai the items belong together statistically, show a difference of extra-high reliability, or have a recognizable rational validity.12
The scales that were developed corresponded to the patient groups of importance in medical and psychiatric practice at that time and were primarily viewed as an aid in reaching diagnostic decisions necessary for management.
DEVELOPMENT OF THE MMPJ VALIDITY SCALES
Because earlier questionnaires or inventories for assessing personality functioning had been developed by "armchair" or intuitive methods rather than by empirical item selection (ie, based on external criterion groups as with (he MMPf clinical scales), they were very susceptible to faking. Needless to say, this unfortunate attribute reduced their clinical usefulness because either conscious efforts at deceiving the clinician or unconscious self-deception had the same invalidating effect. It is obvious that subjects can exert considerable influence on MMPI scores if they wish to do so. This influence might range from straightforward deception or malingering to unconscious factors that could lead the subject toward responding in a socially desirable direction.
Solutions offered to increase reliability and validity in spite of defensiveness or "faking bad" included such methods as exhorting the client to respond with frankness, disguising the significance of the items, rewording questions so that neither true nor false was a consistently normal or abnormal response, using subtle types of items, not requesting the client's name until the bottom of the response form after it had been entirely completed, and using a card-sorting task rather than requiring the subject to make a written response.
In addition, some efforts to assess the degree of conscious or unconscious distortion, in either the positive or negative direction, were carried out but yielded no data to support their usefulness.
Hathaway and McKinley used a different approach, basically following the same empirical procedures used in constructing the MMPI clinical scales. In general, items for the validity scales were selected by procedures that assessed the degree of item shift when subjects completed the MMPI under ordinary conditions and then completed it a second time under specific instructions for faking or bias.
A set of more sophisticated studies were carried out later by contrasting response patterns from normally functioning individuals whose MMPI profiles were abnormal with the response patterns obtained from the records of hospitalized psychiatric patients whose MMPI profiles were within normal limits.
DESCRIPTION OF MMPI VALIDITY AND CLINICAL SCALES
Cannot Say (?)
This validity index is not a scale in the usual sense of the term but simply represents the total number of items that have not been answered by the subject, including those marked both true and false. A high score suggests that the subject may have been hesitant or uncomfortable with the ambiguity of many of the items and thus felt unable to respond, or it may simply represent an index of refusal to complete the task for other reasons.
Scale L (L)
An elevated score on the 1 5 items composing scale L suggests an effort, usually a conscious one, to create the particular impression of being a very "good" person in the sense of having high moral, social, and ethical values.
Scale F (F)
As the score on the 64 F items increases, the greater is the likelihood that some factor has operated to invalidate the whole questionnaire. This might include poor reading comprehension or general reading ability, mental confusion, a deliberate desire to fake psychiatric disturbance, random marking of responses, or perhaps an error in scoring.
Scale K (K)
This 30-item scale measures a more subtle type of psychologic defensiveness than scale L does. A moderate elevation indicates a view of the self as being welladjusted, capable, and confident. A very high elevation suggests an extremely positive degree of social and emotional well-being which is likely to be a denial of the subject's actual status.
Scale 1 (Hs)
An elevation on scale 1 (33 items) suggests an undue concern with the state of one's body and possible preoccupation with symptoms of physical illness. In addition, high scorers are often typified by a sour or pessimistic view of life.
Scale 2 (D)
An elevation on scale 2 (60 items) indicates feelings of frank depression, sadness, pessimism, guilt, and passivity and a tendency to give up hope easily.
Scale 3 (Hy)
An elevation on scale 2 (60 items) seems to represent a continuum of psychologic maturity, with an elevated score characterizing people who tend to be self-centered, demanding, and superficial in their relationships with others and being prone to utilize denial as a means of coping with life stresses.
Scale 4 (Pd)
At moderate levels, scale 4 (50 items) describes interpersonal assertiveness and a desire for nonconformity, whereas marked elevations usually characterize feelings of angry rebelliousness and a lack of identification with usual social mores.
Scale 5 (Mf)
Scale 5 (60 items) provides an index of the range of interests held by the individual and, for both sexes, the degree to which the subject identifies with traditional male or female roles.
Scale 6 (Pa)
Elevations on scale 6 (40 items) can range from interpersonal sensitivity or oversensitivity to irritability or negative speculation about the possible motives or behavior of others and finally to the suspicious style of thinking that characterize a paranoid type of personality.
Scale 7 (Pt)
An elevation on scale 7 (48 items) indicates generalized feelings of anxiety and discomfort. There is often excessive rumination about personal inadequacies, either real or imagined.
Scale 8 (Sc)
This scale is the longest of those making up the basic MMPl profile (78 items) and may be viewed as an index of comfort (or lack thereof) in interpersonal relationships. Elevations are associated with feelings of alienation or detachment from the social realm, which may extend to frank mental confusion or interpersonal aversiveness with higher scores.
Scale 9 (Ma)
Scale 9 (46 items) assesses the subject's level of psychologic energy with moderate elevations being characterized by talkativeness, distractibility, and physical restlessness and higher elevations being associated with impulsivity, impatience, irritability, or rapid mood swings.
Scale 0 (Si)
An elevation of this scale (70 items) indicates social introversion and a lack of desire to be with others; persons scoring low lend to enjoy social interactions and actively seek such contacts.
Before interpreting the profile of clinical scales, the clinician evaluates the scores on scales ?, L, and F to gain a subjective impression of the degree of distortion that might be introduced from an atypical response set. While this is also true with scale K, it was also found that information from five clinical scales - 1 (Hs), 4 (Pd), 7 (Pt), 8 (Sc), and 9 (Ma) - could be improved by a weighting procedure based on the score from scale K. Thus, for these five scales, a fraction of the score from scale K was added to each before it was plotted on the profile, to increase clinical sensitivity.
It was originally believed that the elevation would occur on a single scale for a subject and thus aid in establishing a psychiatric diagnosis. This did not turn out to be the case. It was found, however, that characteristic peaks and valleys among the various MMPI scales accompanied certain patterns of thought and behavior. Specific phrases and descriptive adjectives could then be applied with relative confidence by matching the MMPI profile configuration of the subject being screened to profile patterns that had been established by research or clinical experience. Thus, for example, subjects having their highest scores on scales 4 and 9 are very likely to display some form of sensation-seeking or acting-out behavior and to be characterized by impulsivity, superficial personal relationship, and a lack of responsibility. By comparison, patients having their highest scores on scales 2, 7, and 8 are typified by a chronic difficulty in coping with the usual stresses one is likely to encounter during life and a proneness for dysphoric mood, excessive worrying, and indecision, with much difficulty in building close personal relationships. Not surprisingly, persons facing juvenile court proceedings frequently have the former profile code (no. 49) while the latter code (no. 278) is seen with greatest frequency in psychiatric practice.
In order to interpret the results of the subject's responses to the MMPI items, the raw scores for each scale are transformed to standard units. Such a transformation is necessary because the scales vary widely in the number of items they contain and because of the lack of uniformity in pattern of response to each of the scales. Thus, MMPI raw scores are transformed so that each scale has an average value of 50 and a standard deviation of 10. These standard score units allow comparisons to be made across the various MMPl scales so that the same standard score value reflects the same degree of deviation from the mean, in standard units, regardless of the scale being considered.
A T score of 70, two standard deviations above the mean, has traditionally been viewed as a clinically significant departure on the MMPI because it would represent the 97.7th percentile in a normal distribution. However, it has been found that meaningful interpretation and useful information can be obtained from T scores at lower levels of elevation.
THE ORIGINAL MINNESOTA NORMAL SAMPLE
The criterion groups of patients composing the psychiatric samples were selected, after much study, from inpatients seen at the University of Minnesota Hospitals. The responses from these homogeneous criterion groups were then compared with responses from a criterion group of normal subjects.
The original sample of normal subjects actually was composed of three subgroups described as corresponding to the census data for Minnesota in 1930. The first of these subgroups was described by Hathaway and McKinley4 as:
a normal group from the University hospital and outpatient department (724 cases). These are individuals who themselves are not ill but are bringing relatives or friends to the clinic. They constitute the bulk of our socalled normal cases. The assumption is made, of course, that these people are in good health, which may not always be the case. To help establish them as real normals we ask them whether or not they are receiving treatment for any illness. Only those who say they are not under a physician's care are accepted in this group.
The second sample was described as:
a normal group from the University Testing Bureau (265 cases). These are mainly pre-college high school graduates who came to the Testing Bureau for pre-college guidance but there arc a number o( representatives from various college classes as well.
The third sample was "a group of normals whom we were able to contact through the courtesy of the local WPA Administration (265 cases). These are all skilled workers from local projects." Dahlstrom and Welsh12 also stated that subjects in the WPA sample were all white-collar workers who were used as controls for urban background and socioeconomic level.
To summarize, these three normal control subgroups included 1,254 married and single men and women ranging in age from 16 to 65 years. In the first normal subgroup there were 107 men and 98 women ages 16-25, 233 men and 149 women ages 26-43, 69 men and 43 women ages 44-54. and 16 men and 9 women ages 55-65. Of this group, 66% of the women and 74% of the men were married. The subjects from the second normal subgroup (113 women; 152 men) obtained from the Testing Bureau were ages 16-25; all were unmarried. No information regarding age or marital status was reported for the WPA group.
Subsequently, Hathaway and Briggs'3 presented a refined version of the original Minnesota normal sample in order to improve the representativeness of the norms which were then to be used in developing additional MMPI scales, but they did not collect any new data.
The importance of these normal control groups cannot be overemphasized because as Dahlstrom et al stated. "... the performance of these men and women on each of the component scales in the MMPI is used as a basis for the norms and the test profile."14 Now, how might one characterize the typical or average subject serving as the basis for the MMPl norms? The general nature of the standardization sample can be succinctly summarized as follows:
Each subject taking the MMPI. therefore, is being compared to the way a typical man or woman endorsed those items. In 1940. such a Minnesota normal adult was about 35 years old, was married, lived in a small town or rural area, had had 8 years of general schooling, and worked at a skilled or semiskilled trade (or was married to a man with such an occupational level).14
Because Hathaway and McKinley were keenly aware of the low regard in which prior personality assessment questionnaires were held, they based their approach on the type of criterion-group methods exemplified by the work of Binel in selecting items for assessing cognitive development and by Strong's techniques for assessing patterns of occupational/vocational interest. This course of rational empiricism became the hallmark of MMPI item and scale research:
... all the items were empirically selected by contrast of criterion groups with other clinical groups and with various normal groups. All scales were tested by one or more cross-validation samples. Frequency distributions of the cross-validation samples were constructed to show separation of the criterion abnormal type from normals, from patients who were physically ill (but not obviously with mental symptoms), and finally from miscellaneous psychiatric cases having diagnoses other than the one being studied.10
A solid base of research evidence supporting the usefulness of the MMPI in medical and psychiatric settings has gradually emerged during the 45 years since it was first constructed.
However, it is clear from the tenor of the publications throughout the developmental and subsequent years that, although the MMPI was believed to be a decidedly superior assessment tool in comparison with the inventories that had preceded it, Hathaway and McKinley viewed it as representing a first-generation product that would be superseded by improved assessment methods. This has not occurred. Although some research making use of factor analytic techniques has been undertaken, no second-generation inventory of the MMPl type has emerged in general clinical use. Proliferation of special MMPI scales and some work on the importance of moderator variables have been most prominent in the literature.
The 566 items of the MMPI ( 16 items were duplicated in the booklet form to facilitate computer scoring by the earlier mark-sense devices) can now be scored to yield more scales than there are items! However, the basic profile continues to contain 3 validity and 10 clinical scales supplemented by the number representing the items left blank. Thus, with these basic scales, augmented by a few of the special scales that are gaining interest (eg, assessment of hostility as it relates to coronary heart disease), the MMPI has remained an important clinical and research tool in the fields of medicine and mental health.
1. Lubin B. Larsen RM. Malaraz JD: Patterns of psychological tesi usage in lhe United States: 1935-1982. Am Psychol 1984: 39:451-454.
2. McKinley JC, Hathaway SR: The identification and measurement of lhe psychoneuroses in medical practice: The Minnesota Multiphasic Personality Inventory. JAMA 1943: 122:161-167.
3. Hathaway SR: MMPI: Professional use by professional people .Am Psychoi 1964: 19:204-210.
4. Hathaway SR. McKinley JC: A multiphasic personality schedule (Minnesota), I. Construction of the schedule. J Psychol 1940: 10:249-254.
5 McKinley IC. Hathaway SR: A multiphasic personality schedule (Minnesota). II. A differential study of hypochondriasis. J Psychol 1940: 10:255-268.
6. Hathaway SR. McKinley IC: A multiphasic personality schedule (Minnesota). III. The measurement of symptomatic depression. J Psychol 1942: 14:73-84.
7 McKinley JC. Hathaway SR: A multiphasic personality schedule (Minnesota). IV. Psychasthenia. I Appi Psychol 1942: 26.614-624.
8. McKinley IC Hathaway SR: The Minnesota Multiphasic Personality Inventory. V. Hysteria, hypomania and psychopathic deviale. J Appi Psychol 1944; 28:153-174.
9 McKinley IC. Hathaway SR. Meehl Pt: The Minnesota Multiphasic Personality Inventory. Vl. The K scale. J Consult Clin Psychol 1948; 12:20-51.
10. Hathaway SR: Scales 5 (masculinity- femininity). 6 (paranoia), and 8 (schizophrenia), in Welsh GS. Dahlstrom WCI (edsl: Basic Readings on lhe MMPI in Psychology and Medicine Minneapolis. University of Minnesota Press. 1956. pp 104-111. '
11. Drake LK: A social I.E. scale for the Minnesota Multiphasic Personality Inventory. J Appi Psychol 1946; 30:51-54.
12. Dahlstrom W(J. Welsh GS: An MMPI Handbook: A Guide to Use in Clinical Practice and Research. Minneapolis. University ol Minnesota Press. I960. 13. Hathaway SR. tiriggs PF: Some normative daia on new MMPI scales J Clin Psychol 1957; 13:364-568.
14. Dahlstrom WG. Welsh GS. Dahlstrom LK: An MMPI Handbook: Clinical Interpretation. Minneapolis, University of Minnesota Press. 1972. vol I.
MMPI CONTENT CATEGORIES