Everything you ever wanted to know about the Minnesota Multiphasic Personality Inventory...
The MMPI is important not so much because it is a good test (it isn't very), but because it is the most widely used personality test in the United States. There is a recent revision of the MMPI (the MMPI-2); my reading of the reviews of the revision is that not much has changed.
The MMPI was first published in 1943. The authors were Hathaway and McKinley. At the time, psychologists were not involved in psychotherapy. Their primary role was to administer tests and to attach diagnostic labels to patients. It was noted that people tended to disagree about diagnostic judgments. The most prominent tests of the time were called "projective." The projective test offers an ambiguous stimulus to the person, and the subject responds to it. The scoring of projective tests rests heavily on the scorer and what he or she makes of the subject's responses.
The MMPI was an attempt to make a more objective test. The questions in the MMPI are easy to comprehend, and all the respondent (patient) must do is say that a particular statement is true of them, is not true of them, or does not apply to them.
The MMPI was one of the first tests to use criterion keying of items. Prior tests were based on theoretical notions of symptoms, underlying causes of disturbance, etc. It was noted at the time that such items often did a poor job of discriminating among diagnostic categories, that is, items which theoretically should diagnose people as belonging to one category or another failed to show difference among groups. Criterion keying rests on using known groups to choose items and develop proper keys.
The authors collected lots of items form sources such as psychological and psychiatric case histories and reports, textbooks, earlier published scales (the vacuum cleaner technique). The authors reduced the item pool of over 1000 statements to 504 by eliminating redundant items. They then selected criterion groups. The normal group was mostly made up of relatives and visitors to patients in the University of Minnesota Hospitals. [Why might this be a problem in interpreting subsequent scores?] The clinical groups were, of course, the people whom the normals were coming to visit. The clinical subjects were divided into groups based on agreed upon diagnostic labels. Whenever there was doubt about a particular patient's label or the patient had more than one label, that patient was not used. [Why did they do this? Was it a good idea?] The group's labels were hypocondriasis, depression, hysteria, psychopathic deviate, paranoia, psychasthenia, schizophrenia, and hypomania. (These will be defined later.)
The keys for the items were developed in several steps.
1. The 504 items were given to each of the clinical groups and to the normals.
2. Items that discriminated among both the clinical groups and the normals were kept for the scale in question. This was done separately for each label.
3. The clinical scales were cross validated by comparing a new group of normals and clinical labels (e.g., depressives) on the relevant scale. The cross validation did not attempt to show differences among the clinical labels.
At a later time, two new scales were added to the MMPI. The first was the Masculinity-Femininity (Mf) scale, which was intended to discriminate between male homosexuals and heterosexuals. Not enough items could be identified which discriminated between the two groups, however, so items were added which discriminated between heterosexual men and heterosexual women. The second scale is an attempt to measure social introversion (Si). The items in this scale successfully discriminated between female college students who participated in many social extracurricular activities and female college students who were not very socially participative. [What problems do you see with these scales?]
Validity Scales (lie scales)
The authors also developed four scales to detect "deviant test-taking attitudes."
1. The "Cannot Say" scale. This is the simple frequency of the number of items omitted or marked both true and false. Large numbers of missing items call the scores on all other scales into question.
2. The L scale, originally called the "Lie" scale. This was an attempt to assess naive or unsophisticated attempts by people to present themselves in an overly favorable light. These items were rationally derived rather than criterion keyed. An example item is "I do not read every editorial in the newspaper every day." Only somebody playing Batman will say "false" to this item.
3. The F scale. This is a deviant, or rare response scale. The approach was to look at items which are rarely endorsed by normal people. If less than 10 percent of the normals endorse the item, but you do, your F count goes up. "All laws should be eliminated."
4. The K scale. This scale was an attempt to assess more subtle distortion of response, particularly clinically defensive response. The K scale was constructed by comparing the responses of a groups of people who were known to be clinically deviant but who produced normal MMPI profiles with a group of normal people who produced normal MMPI profiles (no evidence of psychopathology in both). [Do you see a problem with this?] The K scale was subsequently used to alter scores on other MMPI scales. It was reasoned that high K people give scores on other scales which are too low. K is used to boost the scores on other scales.
After a decade or so of experience with the MMPI, people realized that it didn't do what it was designed to do very well. That is, scores on the scales didn't tell the clinician what label to assign a patient. Many patients had high scores on several scales, and some normals had a high score on at least one of the scales. Some of the scales turned out to be highly correlated, indicating that the test did not have the factor structure intended by the authors. As a result, the original scale names have been replaced by numbers, so that the individuals with high scores on various scales will be less likely to be inappropriately labeled. The current convention is:
1 Hypochondriasis -- unrealistic health fear
2 Depression, got the blues
3 Hysteria -- paralysis, false pregnancy
4 Psychopathic Deviate -- no conscience
5 Masculinity-Femininity -- crummy scale of feminine interests
6 Paranoia -- persecution & delusion of grandeur
7 Psychasthenia -- Anxiety, OC, rumination
8 Schizophrenia -- serious thought disorder; hallucinations
9 Hypomania -- manic reaction
0 Social Introversion -- crummy scale of shyness
When you see authors mention an MMPI 49' or some number, the numbers refer to those above (that is, high scores in this case on scales 4 and 9 -- this would be a dangerous person).
Well, the MMPI doesn't do what it was intended to do. What is it good for? Clinical psychologists use it in their practices. Virtually every clinical psychologist is taught to use the MMPI in graduate school. Current thought is that the scales must mean something, and that research and experience with people who show certain scores give meaning to the scores.
It takes the average college student (normal) about 1.5 hours to complete the MMPI. It is either scored with a cutout key or by computer. Raw scores are converted to T scores (M=50; SD=10) with the K conversion (you will see an example shortly). The mean is the average keyed score for normals.
Interpretations of high scores for the scales
1 Hs. Preoccupation with the body and concomitant fears of illness. Descriptors: has excessive bodily concern; could have somatic delusions; complains of chronic fatigue, pain & weakness.
2 D. Poor morale, lack of hope in the future, general dissatisfaction with one's own life. Descriptors: feels blue, unhappy; pessimistic about the future; self-deprecatory; harbors guilt feelings; refuses to speak.
3 Hy. Involuntary psychogenic loss or disorder of function. Descriptors: reacts to stress and avoids responsibility through development of physical symptoms; has headaches, chest pains, weakness, tachycardia, anxiety attacks; has symptoms which appear and disappear suddenly; lacks insight concerning causes of symptoms.
4 Pd. Social deviance or amorality. Descriptors: has difficulty in incorporating values and standards of society; engages in asocial or antisocial behavior such as lying, cheating, stealing, sexual acting out, and excessive use of alcohol or other drugs; rebellious toward authority figures.
5 Mf. Feminine interests and attitudes. Descriptors: (male) conflicted about sexual identity; insecure in masculine role; effeminate. (female) rejects traditional female role; has stereotypic masculine interests in work, sports, hobbies; active, vigorous, assertive.
6 Pa. Paranoid symptoms such as feelings of persecution, grandiose self concepts, etc. Descriptors: manifests frankly psychotic behavior; has disturbed thinking; has delusions of persecution and/or grandeur; feels mistreated or picked on.
7 Pt. Obsessive-Compulsive (Psychasthenia). Descriptors: experiences turmoil and discomfort; anxious, tense, agitated; worried, apprehensive; high-strung, jumpy; introspective, ruminative; obsessive in his/her thinking; has compulsive behaviors.
8 Sc. Schizophrenia. Descriptors: may manifest blatantly psychotic behavior; confused, disorganized, disoriented; has unusual thoughts or attitudes; delusions; has hallucinations; does not feel a part of social environment.
9 Ma. Hypomania -- elevated mood, accelerated speech & motor activity, flights of ideas. Descriptors: manifests excessive, purposeless activity; has accelerated speech; may have hallucinations, delusions of grandeur; emotionally labile; may be confused.
10 Si. Withdrawl from social contacts. Descriptors: socially introverted; more comfortable alone or with a few close friends; reserved, timid, shy, retiring.
There are endless other scales which have been developed subsequently from the same pool of items. Also, the scores are not interpreted in isolation; rather the profile as a whole will be interpreted so that someone will be a 49, and someone else will be a 123.