Personality inventories were commonly used during World War I to screen military recruits for emotional disturbances. The inventories were successful at this task, and during the 1920s and 1950s several new personality inventories were developed. Attempts were made to increase the validity of these new inventories but for the most part they were not found to be useful in a psychiatric setting. They measured very general personality variables and were not able to distinguish between different types of psychiatric patients. The Humm-Wadsworth Temperament Scale1 published in 19)5 was developed specifically for use with psychiatric patients. The items were tested for their ability to distinguish different types of psychiatric patients. Consideration was also given to the number of symptoms claimed by a given patient. Too large or too small a number suggested a deviant response set. which would invalidate the lest. These seemed to be worthwhile innovations but. when used with psychiatric patients, the Humm-Wadsworth Scale did not work very well.
Frustrated by the lack of a useful personality inventory, the authors of what became the MMPI decided to develop a test that would focus on the diagnostic needs in a psychiatric setting. Most earlier tests had been developed on college students, and the personality characters measured had little relevance for psychiatric diagnosis. The MMPI was planned with the goal of measuring characteristics (hat were important to the then current psychiatric diagnostic nomenclature.
Earlier inventories had used items that were selected rationally to measure certain personality characteristics. The authors of the MMPI followed a different approach. They picked a large and varied group of questions and initially did not divide them into separate scales. They then developed the clinical scales by contrasting the responses of a normal group with the responses of patients with a specific psychiatric diagnosis. Once the items that differentiated between these two groups were identified, they were gathered into a scale that was then tested on a new sample of patients to see if it could distinguish between patients and normals. One result of this technique was the discovery of'"subtle items." Tlie.se are items that do not appear to be related to the personality dimension being measured. F:or example, a patient confronted with the item "I think Lincoln was greater than Washington" would be hard put to know what is measured by the item. Yet this is an item that appears on an MMPI scale. Another result of the empiric method of item selection was that each scale varied greatly in item content. Because the patient groups were different from the normals in many ways, these differences provided a variety of items.
Although the MMPI was not based on a specific theory of personality, its basic assumptions are consistent with trail theory, which assumes that there are enduring personality characteristics (traits) that, if known, can allow us to describe and predict a person's behavior.2 Naturally, the test's authors were aware that these enduring personality characteristics could be influenced by situational variables. This is a limiting factor in the validity of personality tests. It is assumed that people who answer the items in a similar way would be likely to show other similarities. Like all psychologic tests, the MMIM is based on the assumption that from study of a small sample of behavior (such as responses to test items) we can generalize to other aspects of the subject's life.
The test's developers originally had hoped that, when the test was given to a psychiatric patient, the patient would obtain a high score on the scale that corresponded to the appropriate diagnosis. Unfortunately, most patients had elevations on more than one scale. Therefore, it was necessary to develop a method of interpreting these various combinations of high scores. Following the original assumptions, test profiles that had similar patterns were collected. The subjects who had produced those profiles were then examined for similarities in personality characteristics. Several of the early MMPI books described the personality characteristics associated with certain profile types.3
Because patients generally do not produce an elevation on only a single scale, it was necessary for clinicians to develop a format for interpreting the test results. First, the validity scales are examined to determine whether or not the patient understood the items and took the test with an appropriate attitude. The validity scales also provide some information about personality characteristics. Next, the profile is examined for the presence of combinations of scale elevations that have been reported as indicating the presence of certain traits. Then, individual scale scores are used to modify the interpretation.
In his article. '"Wanted - a good cookbook." Mechl suggested that rules be constructed to define certain MMPI profile types and then patients with that type of profile should be studied carefully.4 If the shared characteristics were listed, when another patient with a profile of that type appeared, those characteristics could be attributed to the new patient. When they evaluated this approach. Halbower5 and Meehl4 concluded that the cookbook predicted more accurately than did a group of clinicians who interpreted the MMPI profiles. A fewcookbooks were developed and were found useful in MMPI interpretation.0,7 However, clinical experience soon demonstrated that onlv a small percentage of the profiles in a given selling would match the rules for one of the classic code types. As a result, the cookbook could be used to interpret only a few of the profiles seen. The logical extension of the cookbook approach was automated profile interpretation, a development discussed elsewhere in this issue by Swenson.
USES OF THE M MPI
The MMPI is frequently used to aid in the description of personality. It is assumed that there are enduring personality characteristics (traits) that influence the way a person responds to various situations - ie. if we assign the trait "dependency" to a person, we assume that he or she will act in a more or less dependent way in various situations. In assigning a trait, we are describing the person in terms of the likelihood that the person will behave in a particular way. A person having a strong tendency to behave in that way is considered high on that trait: one with a lesser tendency to act in that way would be described as scoring low on that trait.
Early research on MMPI code types was done by taking a group of people with a certain code type and examining their behavior, either through the use of observer ratings or by personality tests. The result usually was a list of personality characteristics or traits possessed by people who had a given code type. It was assumed that identification of personality traits would lead to prediction of behavior. Many psychologists interested in personality have argued thai behavior is much more variable from situation to situation than trait theorists suggest.8 As a result, they suggest that trait theory does not predict behavior very well. Mischel stressed the importance of situational variables in controlling behavior.8 This criticism overlooks the faci that trail theorists never assumed that traits were the only factors influencing behavior. Instead, they see traits as rellecting a probability that the person will respond in a certain way in a wide range of situations but with the reservation that the trait may not be expressed at a given time because of extrinsic factors. Because personality tests are not able to measure the situational variables, this attenuates the fit of the predictions they make. However, it does not eliminate their ability to predict. As Eyscnck stated. "No physicist would put such a silly question as: which is more important in melting a substance - the situation (heat of the flame) or the nature of the substance!"4
Another aspect of personality description is the identification of the defense mechanisms favored by the person. This is accomplished in MMPI interpretation by knowledge of the defense mechanisms used by patients in the original group on which the scales were developed. It is said, for example, that patients with elevations on scales I (Hs) and 3 (Hy) rely heavily on repression, denial, and somatization as defenses. Patients with elevation on scale b (Pa) frequently use projection, and patients with elevations on scales 4 (Pd) and 9 (Ma) lend to act out and become more active when stressed. Use of the MMPI to identify common defense mechanisms can help in evaluating how a person will respond to psychiatric intervention and psychotherapy.
Some personality researchers have attempted to develop profile classification rules or statistical procedures for classifying profiles into general types of abnormality - eg. psychotic, neurotic, or personality disorder. An early attempt was made by Meehl and Dahlstrom.10 Their sequential rules, developed in 1950J. were for the purpose of discriminating psychotic from neurotic profiles. They were to be used for research purposes and not for routine clinical evaluation. The various rules are applied sequentially to the profile until il is categorized as neurotic or psychotic. If none of the rules applies, the profile is labeled "indeterminate." This method was developed so that researchers could use the MMPI to identify homogeneous groups of psychotic and neurotic patients objectively.
Goldberg developed a linear equation based on the scores of certain MMPl scales to identify psychotic and neurotic profiles." His equation worked as well as the more complex Meehl-Dahlstrom rules. Naturally, use of these procedures is of limited interest to the clinician, who needs more information than a general classification into psychotic or neurotic categories.
Fortunately, the personality descriptions based on the traits that are found to be associated with the various MMPI profile types and the defenses presumed to he used by the patient make it possible to provide useful and detailed personality information which, when combined with information from the clinical evaluation, can aid in diagnosis.
An important issue in the assessment of patients is evaluation of suicide potential. Even among psychiatric patients, suicide is a rare event. Any index used in a personality test is likely to have a large number of false-positives and therefore will be incorrect much of the time. There is also evidence that the level of suicidal risk fluctuates from day to day. This makes it difficult to evaluate research in which the MMPI was obtained several weeks or months before the suicide actually occurred.
examination of the profile code types of patients in the MMPI atlas12 who committed suicide indicates that 55"o of the suicidal patients had scale 2 (D) as either the highest or next to the highest scale in their piotile. Unfortunately, because of the base rale problem, most of the patients with scale 2 as one of the highest scales did not commit or attempt suicide. It is difficult, therefore, to use this knowledge for the benefit of an individual patient, it is generally assumed by clinicians, however, that elevations on scale 2(D) suggest an increased risk of suicide.
Attempts to compare the profiles of patients who later commit suicide with those of patients who do not have yielded inconsistent findings. Some of the reasons for these inconsistencies are that varying times had elapsed between completion of the MMPI and the suicide and that the studies were done in settings with different suicide rates.1' Attempts to develop a new MMPI scale to predict suicide also have failed. There are some suggestions that it is possible to predict which patients will threaten suicide but not possible to predict which ones will successfully commit suicide.
At times the MMPl is used as part of an assessment to determine whether or not a patient will become violent. As for suicide potential, the potential for violence is only partly a function of intrapersonal variables that can be measured by personality tests. Violence occurs in the context of a complex interaction of situational and interpersonal variables. MMPI results can predict such things as hostility and poor impulse control but cannot assess the various situational factors that lead to acting out of hostility. From a large number of studies of violence in criminal populations. Megargee and associates identified two types of violent offenders.14 The first type was characterized by undcrcontroUcd hostility - ie. poor impulse control and tendency to act out hostility with little provocation. The other group was characterized by overcontrolled hostility. This group oí offenders would tolerate provocation in a relatively passive way for a time and then would explode into violence. The investigators developed the overcontrolled hostility scale (OH) for the MMPI to identify this group of offenders.
Gilbersladt and Duker" found that one of the common profile types in a VA hospital population was marked by poorly controlled anger and temper outbursts. These patients had primary elevations on scales 4 (Pd) and 3 (Hy). These patients with the "4-5 profile" were described as having aggressive outbursts and making homicidal threats toward family members. Interestingly. VA hospital patients with this type oï profile also had a high rate of suicide attempts. Gilberstadt and Duker speculated that these patients turned their poorly controlled aggression inward. This group of patients also drank heavily. Therefore, it is possible that the lowered impulse control was in part a reaction to alcohol.
Because the MMPI cannot measure factors in the environment, a statement about potential for violence from the lest needs to restrict itself to statements regarding level of hostility and degree of impulse control.
Burgeoning interest in chemical dependency and the proliferation of alcoholism treatment centers has stimulated research into personality factors leading to alcoholism. Many early studies compared the profiles of alcoholics with those of other groups. However, these studies did not consider the effect that alcohol consumption had on personality. Lanyon published a mean profile lor a group of alcoholics that had an elevation on scale 4 (Pd). indicating nonconformity and lack of impulse control.15 In addition to this elevation, other authors reported secondary elevations on the neurotic scales of the MMPI. Gilberstadt and Duker identified the "2-7-4 prolile" type with elevations on scales 2 (D). 4 (Pd). and 7 (Pt) in their VA hospital sample.«5 Heavy drinking was present in 96% of patients with this profile. They stated that in these patients chronic severe alcoholism was associated with anxiety, tension, inferiority feelings, and guilt. These patients were also described as hostile and dependent.
Naturally, these mean profiles conceal wide variations in the profiles of individual alcoholics. Some individual profiles indicate marked neurotic features, and others suggest personality disorder as a major contributor to the addictive behavior.
Loper et al"' used MMPI profiles obtained during a college orientation session to compare 52 freshmen who. 15 years later, entered alcoholism treatment to 148 of their classmates who did not. The pre-alcoholism profiles of the men who later entered treatment were marked by elevations on scales 4 (Pd) and 9 (Ma). These profiles were generally within normal limits but suggested reduced impulse control, nonconformity, sociability, and thrill seeking.
Several investigators attempted to develop scales that would identify alcoholics in a clinical setting. They compared patients hospitalized for alcoholism with normals and with other types of psychiatric patients. Although several of the early scales could distinguish between alcoholics and normals. MacAndrcw and Geertsma found that they could not distinguish between alcoholics and other types of psychiatric patients.'7 They concluded that these early scales were indices of general maladjustment. MacAndrcw developed his own scale by comparing the MMPI responses of 500 male alcoholics and 500 male psychiatric outpatients. After eliminating items that related directly to alcohol consumption, he found 49 items that discriminated between the alcoholics and the psychiatric patients.18 His scale has found wide clinical acceptance even though it has some difficulty distinguishing between alcoholics and people engaged in non-alcoholism-related antisocial behavior. Lachar and associates indicated that the MacAndrcw Scale was not useful in discriminating between alcoholics and drug addicts.19
The use of the MMPI to studv treatment outcome has been reviewed by Hollon and Mandell.21' They concluded that on the whole these studies are mildly supportive of the MMPI as a measure of change. In most studies, successful treatment was followed by changes on the scales that rellect neurotic symptoms. This includes decreases on scales I (Hs). 2 (D). 3 (Hy). and 7 (Pt). There also are indications that scores on scales K and Es (ego strength) increase with successful treatment. These findings are not surprising because psychotherapy is generally considered to work best on patients with neurotic symptoms.
Hollon and Mandell also reported thai the MMPI is sensitive to treatment-related changes in schizophrenics. In this group, scales F. b (Pa). and 8 (Sc) are most likely to change. These findings occurred in studies using psychotherapeutic approaches and studies using somatic treatment.
In a study in which schizophrenic patients were divided into groups of those who improved and those who did not improve with treatment. May found greater reductions on scales F. 6 (Pa). 7 (Pt). and 8 (Sc) in those schizophrenics who responded to various treatment approaches compared to schizophrenics who did not benefit from treatment.21
In many clinical settings in which children are evaluated, it is standard practice to give the MMPI to parents, with the aim ol studying the influence of parental characteristics and the effects of childhood psychopathology on parents. In an extensive study of this practice. Lachar and Sharp reached three conclusions.2- They thought that the relationship between parental psychopathology and children's problem behavior was stronger for daughters than for sons. They also concluded that mother-child interactions had a greater influence on children's behavior than did father-child interactions. They reached the more general conclusion that parents of behaviorally disturbed children appear to be more disturbed than parents in general on the MMPI but not as disturbed as adult psychiatric patients.
The MMPI is useful in several aspects of psychiatric practice. It aids in early personality description and is useful, along with other information, to help evaluate potential for suicide or violence. Some special scales have been developed to help identify patients who are chemically dependent. The test can also help reflect treatment outcome.
The MMPI is best used in the context of other information. It should be viewed as providing the clinician with hypotheses that can be verified with other methods.
A major problem with the MMPl is that the norms were originally developed during the 1950s on a group of people from Minnesota. The median individual in the original reference group, which corresponded to the 1930 census, had an 8th grade education, was married.
lived in a small town or on a farm, and was a clerk or skilled tradesman. The article in this issue by Colligan and Uf ford presents new norms which will help the clinician who wishes to apply the MMPI to a present-day population.
In addition to the establishment of a new reference sample as reported by Colligan and associates.2' clinical norms would also be useful. The psychiatric diagnostic nomenclature has changed twice since the MMPI was first developed. L)SM-IlI has more specific criteria for defining pathologic groups than was true in the past. New clinical scales for diagnostic groups based on DSM-III or the research diagnostic criteria should be developed.
1 . Humm DG. Wads wort h CiW The Humm-Wadsworth Temperament Scalo Aiti I Psychiatry 1935; 12:163-200.
2. Pervin LA: Personality: Theory, Assessment, and Research, cd 5 New York. John Wiley & Sons. I 980.
5 Dahlstrom WG. Welsh OS: An MMPI Handbook: A Guide Io Use in Clinical Practice and Research Minneapolis. University ol Minnesota Press. 1960.
4. Meehl PE: Wanted- A good cookbook Am Psychol 1956: 11:263-272.
3. Halbower WG. cited by Meehl PK: Wanted - A good cookbook. Am Psychol 1956: 11:263-275
6 Oilberstadl H. Duker I : Handbook for Clinical and Actuarial MMPI Interpretation Philadelphia, WB Saunders Company, 1965.
7 Marks PA. Seeman W. Actuarial Description of Abnormal Personality: An Atlas for Use With the MMPI Baltimore. Williams & Wilkins Company. 1963
8. Mischvl W Personality and Assessment New York, John Wiley & Sons. 1968.
9. Kysenck II. cited by Pervin LA: Personality: Theory, Assessment, and Research. ed 3. New York, lohn Wiley & Sons. 1980
10. Meehl PK. Dahlstrom WG Objective conligural rules for discriminating psychotic from neurotic MMPI profiles. G Consult Clin Psychol 1960: 24373-587.
11 Goldberg LR: The diagnosis ol psychosis versus neurosis from the MMI'I Read al the Third Animal MMPI Simposium. Minneapolis. Minnesota. April 4 to 5, 1968.
12. Hathaway SR. Meehl PK: An Atlas for the Clinical Use of the MMPI Minneapolis, University ol Minnesota Press. 1931
13. Clopton IR: The MMPI and suicide, in Newmark CS tedi: MMPI: Clinical and Research trends. New York. Praeger Publishers. 1979.
14. Megargee Kl. Cook PE, Mendelsohn CiA: Development and validation of an MMPI scale ol assaultiveness in overconl rolled individuals. J Abnorm Psychol 1967: 72:519-528.
15. Lanyon Rl: A Handbook of MMPI Croup Profiles Minneapolis. University of Minnesota Press. 1968.
16. Koper RCi. Kammeier ML. Hoffman H: MMPI characteristics ol college freshman males who later became alcoholic. J Abnorm Psychol 1975: 82:159-162.
17. MacAndrew C. Geertsma RH: A critique of alcoholism scales derived from the MMPI. Ouart J Stud Alcohol 1964; 25:68-76.
18. MacAndrew C: The differentiation of male alcoholic outpatients from nonalcoholic psychiatric outpatients by means ol the MMPI. J Stud Alcohol 1965: 26:258-246.
19. Lachar D, Berman W. Oriseli IL. et al: The MacAndrew Alcoholism Scale as a general measure ol substance misuse. J Stud Alcohol 1976; 57: 1609-1615
20 Hollon S. Mandell M: Use ol the MMPI in the evaluation of treatment effects, in Butcher IN (ed): New Developments in the Use of ihe MMPI Minneapolis. University ol Minnesota Press. 1979.
21. May PRA : The results ol treatment ol schizophrenia - An overview , in May PRA ted): Treatment of Schizophrenia: A Comparative Study of live Treatment Methods. New York. Science House. 1968.
22. Lachar K. Sharp |R: Use of parents' MMPIs in the research and evaluation of children: A review of the literature and some new data, in Butcher JN (ed): New Developments in the Use of the MMPI. Minneapolis. University of Minnesota Press. 1979
23. Colligan RC. Osborne D. Swenson WM. et al: The MMPI A Contemporary Normative Study, New York. Praeger Publishers. 1983.