General Description of the MMPI/MMPI-2
The Minnesota Multiphasic Personality Inventory (MMPI) and its revised forms, the MMPI-2 (Butcher, Graham, Ben-Porath, Tellegen, Dahlstrom, & Kaemmer, 2001), the MMPI-A (Butcher, Williams, Graham, Archer, Tellegen, Ben-Porath, & Kaemmer, 1992), and the MMPI-2 Restructured Form (MMPI-2-RF; Ben-Porath & Tellegen, 2008) are the most widely used objective personality adjustment inventories in the world (Archer, 2005; Butcher, 2010; Friedman, Lewak, Nichols, & Webb, 2001; Greene, 2011; Lubin, Larsen, & Matarazzo, 1984; Lubin, Larsen, Matarazzo, & Seever, 1985). Even from its beginnings, the test has enjoyed popularity and has been ranked as a leading personality instrument (Archer, 1997, 2005; Butcher, 2010; Friedman, Webb, & Lewak, 1989). Taught in the majority of clinical psychology training programs (Dahlstrom & Moreland, 1983;
Friedman, Webb, Smeltzer, & Lewak, 1989; Watkins, 1991), the MMPI is used by most psychologists who conduct assessments as part of their clinical and consulting practice (Archer et al., 2006; Dahlstrom, 1992b; Moreland & Dahlstrom, 1983; Watkins, Campbell, Nieberding, & Hallmark, 1995). The MMPI and its successors have also been the subject of extensive research. Butcher (2010) estimated the number of references for the MMPI, MMPI-2, and MMPI-A at 19,000. Greene (2011) conducted an electronic search of the psychology databases in January, 2010, using the search term “MMPI,” yielding 24,171 citations, and the search term “MMPI-2” produced 4,216 citations.
The widespread use and longevity of the MMPI is attributable to several factors, including its simplicity of scoring and administration, an objective response format important for research designs, a large item pool (from which at least 800 additional scales have been derived), many useful applications, the inclusion of validity scales for determining the examinee’s level of cooperation, numerous translations into other languages, and thousands of empirically established correlates. In fact, it is difficult to imagine many settings in which psychologists perform assessment and treatment functions where the MMPI in its various forms has not been used. Inpatient and outpatient mental health facilities often employ psychologists to make diagnostic and treatment decisions, and the MMPI-2/MMPI-A/MMPI-2-RF is typically included in psychological test batteries or administered independently. Many clinicians routinely use the instrument to assess new patients for psychopathology in their office practices and to help formulate treatment plans, or to conduct child custody evaluations (Alan Jaffe, personal communication, May 31, 2013). Psychologists also use the test in medical settings to evaluate the presence of psychological components in physical complaints and to assist in predicting response to various treatments (Osborne, 1979).
2 Development of the MMPI and MMPI-2
Industrial-organizational and clinical psychologists, whose responsibilities involve matching individuals to particular employment positions or screening individuals for psychopathology, also use the MMPI-2. For example, because the MMPI-2 is sensitive to emotional maladjustment in individuals other than identified patient groups, it is often used in personnel selection situations in which high-risk occupations require careful screening of applicants (Butcher, Ones, & Cullen, 2006; Davis & Rostow, 2004; Davis, Rostow, Pinkston, Combs, & Dixon, 2004). People in positions of public trust, such as airline flight crews, law enforcement officers, firefighters, emergency dispatchers, nursing staff, ministerial candidates, and nuclear power plant operators, are typically administered the MMPI-2, not only for selection purposes but also for fitness-for-duty evaluations.
Research applications with the MMPI-2, either as the sole subject of study or as one of the major dependent measures in investigation, range from cross-cultural studies of response patterns to evaluating treatment effects to making forensic decisions (Friedman, Lewak, Nichols, & Webb, 2001; Rothke & Friedman, 1994; Rothke, Friedman, Dahlstrom, Greene, Arrendondo, & Mann, 1994; Rothke, Friedman, Jaffe, Greene, Wetter, Cole, & Baker, 2000). The test is widely used in both criminal and civil forensic settings, as the MMPI-2 is often admitted as evidence in court (Pope, Butcher, & Seelen, 2006).
The original MMPI consisted of 566 numbered statements, each of which could be answered True or False on an answer sheet, or, in the original deck form, the test takers sorted the cards into True, False, or Cannot Say categories. An item was scored Cannot Say if it was marked or sorted both True and False, left blank, or placed into the Cannot Say pile in the original card form. There are several methods for scoring a completed test.
The answer sheet is either hand scored or read by an optical scanner. Some MMPI users enter the responses from the answer sheet into a computer programmed to score and/or interpret the test. Some test takers enter their responses directly into a computer using a keyboard, which instantly scores the test.
For the original MMPI, 13 standard scales are scored, regardless of the scoring method. The following are the original three validity scales and 10 standard clinical scales. Note that the convention adopted here is that the eight scales numbered 1–4 and 6–9 will be referred to as the basic clinical scales or, more simply, as the basic scales.
When speaking of the entire set of clinical scales displayed on the main profile form, including Masculinity-Femininity (Mf, or Scale 5) and Social Introversion (Si, or Scale 0), we refer to the standard clinical scales or more simply, the standard scales.
Original MMPI Validity Scales t Lie (L)
t Infrequency (F) t Correction (K) Clinical Scales
t Scale 1 Hypochondriasis (Hs) t Scale 2 Depression (D) t Scale 3 Hysteria (Hy)
t Scale 4 Psychopathic Deviate (Pd) t Scale 5 Masculinity-Femininity (Mf)
t Scale 6 Paranoia (Pa) t Scale 7 Psychasthenia (Pt) t Scale 8 Schizophrenia (Sc) t Scale 9 Hypomania (Μa) t Scale 0 Social Introversion (Si)
The validity scales were developed to assist in recognizing test records produced by uncooperative or deceptive participants with various test-taking attitudes (e.g. faking good or faking bad) or participants who had difficulty comprehending or reading the test items. The clinical scales were developed primarily to assist in identifying the type and severity of abnormal psychiatric conditions. A secondary goal was to provide an objective means of estimating therapeutic effects and other changes in the status of patients’ conditions over time (Dahlstrom, Welsh, & Dahlstrom, 1972). The raw scores on each of the validity and clinical scales are converted to standard Τ-scores by plotting them onto a profile form, thereby rendering the raw scores comparable. It is the pattern of these T-scores that is usually interpreted. The T-scores provide the clinician with an opportunity to examine how the test taker compares to different populations, including, most importantly, a standardized group of normal individuals. A more detailed description of T-scores is provided later in this chapter. Initially, individual scales were interpreted to assist in psychodiagnosis, but experience showed that combinations of scales were better predictors of personality characteristics, so test users began focusing on patterns of scale scores versus individual scale elevations. “Profile” became the term used to focus on the eight basic clinical scales, and eventually, “Codetype” or “Code Pattern” analysis became the way in which interpretation of the test was described.
Historical Development of the MMPI
In an important historical overview of the MMPI, W. Grant Dahlstrom (1992b) summarized a series of studies by Carney Landis and his colleagues at the New York State Psychiatric Institute in the 1930s (Landis & Katz, 1934; Landis, Zubin, & Katz, 1935;
Page, Landis, & Katz, 1934; see also Greene, 2011); these studies strongly reinforced the skepticism at that time of professionals relying on personality testing as an aid in assessment and diagnosis because existing personality tests lacked validity. Specifically, the objective tests available at the time relied on individuals’ willingness and capacity to accurately report their feelings and experiences and also depended on their reading and intellectual ability to comprehend the inventory. These concerns sensitized clinicians to the need for improving the validity of objective personality tests.
Despite the tenor of the times, or perhaps because of it, Starke Rosencrans Hathaway and J. Charnley McKinley teamed up to develop a new inventory capable of overcoming the limitations inherent in the existing personality tests. Hathaway, a clinically experienced physiological psychologist, and McKinley, a neuropsychiatrist and head of the Department of Psychiatry and Neurology, both at the Medical School at the University of Minnesota, were joined in their later efforts by Paul E. Meehl. Hathaway and McKinley were originally motivated to design a test that could serve as an aid: “in diagnosing persons classified as constitutional psychopathic inferiors” (Hathaway, 1939, p. 117); assist “in assessing the psychological factors associated with physical problems or disease seen in a
4 Development of the MMPI and MMPI-2
medical practice” (McKinley & Hathaway, 1943, p. 161); and, as a corollary, “measure the effectiveness of insulin therapy” (Hathaway, 1964, p. 204) in schizophrenia, which was in widespread use by the late 1930s. The test also came to be seen as an aid in determining levels of psychiatric impairment and changes in the patients’ condition over time, as well as in measuring the effects of psychotherapy (Dahlstrom et al., 1972).
Although the MMPI was originally published in 1942 (Nichols, 2011), the authors actually began their work on the test in the late 1930s. By the end of April, 1943, the MMPI was generating enough revenue to motivate the Psychological Corporation to become its licensed distributor. In 1951 the MMPI arrived in its final form with the addition of the Si scale (Si), published in 1946 (Drake, 1946; also in Butcher, 2000). There were several subsequent revisions of the original MMPI Manual through 1983.
The first article describing the inception of the MMPI was published by Hathaway and McKinley in 1940: “A Multiphasic Personality Schedule (Minnesota): I. Construction of the Schedule.” Hathaway and McKinley strove to correct many of the problems hampering the effectiveness of most previously existing personality inventories. These earlier personality inventories were typically constructed on a rational basis with a focus on content validity, but lacked scales designed to measure the participant’s test-taking attitude (e.g. defensiveness or over-reporting of symptoms). One such inventory was the Woodworth (1920) Personal Data Sheet, also called the Psychoneurotic Inventory. World War I created a strong need to screen for maladjustment among draftees, and Woodworth developed a 116-item self-rating scale to detect neurotic maladjustment. The items consisted of statements that Woodworth believed reflected neurotic symptoms. If a participant answered a certain number of items in the neurotic direction, a psychiatric interview was conducted. A fundamental assumption inherent in the test was that the items measured what Woodworth assumed they measured.
Items were chosen on rational grounds; that is, if the items appeared content-relevant to neuroticism, they were included on the scale. Over time, it became clear that items selected on a rational basis did not always indicate deviant behavior in the way Woodworth expected. Another unwarranted assumption in the Personal Data Sheet, and similar tests, was that the participant would, and could, honestly and accurately describe him- or herself. This is not always the case; self-deception and social desirability factors operate to influence the person’s responses to test items. According to Hathaway (1965) and Nichols (2011), the test was limited in its success because prospective soldiers who feared combat or otherwise considered themselves in need of evaluation were inclined to declare their vulnerabilities on the test. The items were apparently too obvious in their intent to detect neuroticism. Although the Woodworth Personal Data Sheet was not completed early enough to allow its use before World War I ended, it did set the stage for other similarly constructed inventories that achieved widespread use after the war (Anastasi, 1982).
The Bell (1934) Adjustment Inventory and the Bernreuter (1933) Personality Inventory were derived from the Woodworth Personal Data Sheet and were also criticized for their excessive reliance on a rational approach to test construction and for the face-valid nature of the test items (Colligan et al., 1989). Landis and Katz (1934) and others contributed to the demise of the Bernreuter Personality Inventory by demonstrating a lack of discriminant validity between diagnostic groups. Psychotic patients, not just neurotics, showed elevations, thereby misclassifying them as neurotic (Greene, 2011).
The Humm-Wadsworth Temperament Survey (Humm & Wadsworth, 1935) was a methodological precursor of the MMPI and was cited in Hathaway’s (1939) article,
“The Personality Inventory as an Aid in the Diagnosis of Psychopathic Inferiors.” It was the first personality questionnaire to use the actual responses of psychiatric patients to determine the direction in which items should be scored and their suitability for scale development (Nichols, 2011). The test consisted of 318 items and provided scores for seven scales. Examination of the Humm-Wadsworth items show a remarkable resemblance to at least 27 percent of the 550 MMPI items (D. S. Nichols, personal communication, April 1, 1999). Meehl (1989) also noted that many items on the MMPI were adopted from the 318-item Humm-Wadsworth Inventory. The seven components of temperament identified on the inventory were: Normal, Antisocial, Manic, Depressed, Schizoid Autistic, Paranoid, and Epileptoid. Greene (2011) provides an excellent critique of the test and states that by the early 1950s there were a total of 14 critiques on the instrument, focusing on themes of how “problem” and “satisfactory” employees had similar profiles, and how statistical analyses of new data sets were incongruent with data reported by Humm and Wadsworth. As Greene (2011) points out, even though the published research on the test disappeared by the mid-1950s, Humm and Wadsworth were innovative and contributed to aspects of the MMPI’s development.
Because of Hathaway’s reservations about the rational approach to inventory construction, he used an empirical method (the criterion keying method) to construct the MMPI. Nichols (2011) illuminates Hathaway’s motivation for adopting this approach to constructing the MMPI. He described Hathaway as a “thoroughgoing pragmatist”
with a deep distrust of theory and an abiding belief in practical experience:
The method of contrasted groups provided Hathaway with a practical means of avoiding theory and sidestepping rational or intuitive guidance in the selection of the items for the MMPI scales. Hathaway did not pretend to know how different kinds of patients would respond to his items. The method of contrasted groups allowed him a satisfactory way of finding out: it allowed him to ask them.
(Nichols, 2011, p. 2) As Anastasi (1982) noted, the MMPI represents the outstanding example of the empirical criterion construction methodology. In this method of contrasted groups, test items are administered to two or more groups of participants—a criterion group selected for homogeneity with respect to a certain diagnosis, cluster of features, traits, or other characteristics (e.g. schizophrenia), and a normal comparison group that does not share the same characteristics or shares them only in base-rate amounts. Items to which the criterion and comparison groups respond statistically differently are included on the scale being developed, and items to which the responses of the two groups are similar are not included. Scales constructed in this fashion are typically named after the criterion group. As Butcher (2010) noted, because items are selected based on prediction of criterion variables, the scales will be heterogeneous in their content. Scoring the scales is accomplished by assigning one point to each item answered in the direction that is marked significantly more frequently by the criterion participant; that is, if a higher proportion of individuals with hysteria than normal answered an item True, a True response to that item would earn one point on the Hysteria scale, and a False response
6 Development of the MMPI and MMPI-2
would be given zero points. The higher the raw score a person receives on a scale, the more items he or she has answered in the direction of the criterion group.
Using this contrasted group or criterion keying method, Hathaway and McKinley began their construction of the MMPI by compiling more than 1,000 self-reference statements from a wide variety of sources, including psychiatric examination forms, psychiatric textbooks, previously published attitude and personality scales, clinical reports and case summaries, and their own clinical experience. From these resources, they initially adopted an item pool of 504 separate statements that could be answered True or False. McKinley and Hathaway (1943) later added 55 items primarily related to masculinity-femininity, but eventually eliminated nine more items, resulting in a final pool of 550 separate items. Because 16 items were repeated to facilitate early mechanical scanning, the final published booklet version of the MMPI contained 566 items, with the 16 repeated items scored no more than once on the scales on which they appeared (Scales 6, 7, 8, and 0). The 16 repeated items were scored the first time they appeared in the test. The 550 statements were divided into 26 content areas, including phobias, religious attitudes, general health—including medical and neurological symptoms—
political and social attitudes, family, educational and occupational experiences, and items associated with an overly virtuous self-presentation (Dahlstrom et al., 1972). The items were judged to be easily readable, written in the first person declarative form with simplified wording based on contemporary word-frequency tables. “Brevity, clarity, and simplicity were occasionally given precedence over grammatical precision. Common English slang and idioms were used, but esoteric or specialized language was avoided”
(Nichols, 2011, p. 3). Using the pool of 550 items, Hathaway and McKinley proceeded to construct scales by contrasting the responses of normal and clinical criterion groups.
The normal reference group consisted of 724 friends and relatives of patients being seen at the University of Minnesota Hospital outpatient department who were willing to complete the test. These normals were all White, as few ethnic minority groups other than native Americans lived in Minnesota at that time and belonged to what was termed
“the underprivileged” classes, and came from all parts of the state (Dahlstrom, Welsh,
& Dahlstrom, 1972; McKinley & Hathaway, 1940). According to Dahlstrom and Welsh (1960), “the subjects were approached in the halls and in waiting rooms of the hospital, and invited to participate in the research project” (p. 44). Additional normal and patient groups consisted of high school graduates attending pre-college conferences at the University of Minnesota (n = 265), medical patients from the University of Minnesota Hospital (n = 254), skilled Work Project Administration (WPA) personnel (n = 265), and inpatients with varied diagnoses in the psychiatric unit (then called the Psychopathic Unit) at the University of Minnesota Hospital (n = 221). Normal participants (other than the medical patients) who were then under the care of a physician were excluded from the normative samples; all other participants were included. Dahlstrom et al. (1972, pp.
7–8) pointed out the importance of the normal reference groups by stating that:
the performance of these men and women on each of the component scales in the MMPI is used as the basis for the norms in the test profile. 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 eight years of general schooling,
and worked at a skilled or semiskilled trade (or was married to a man with such an occupational level).
Hathaway and McKinley found their original sample of normals to correspond well in age, gender, and marital status to the Minnesota population in the 1930s census (Dahlstrom et al., 1972). However, it is now generally accepted that “the original MMPI norm group appears to have over-represented lower educational and occupational groups” (Dahlstrom, 1993, p. 9). It is important to note that while the original fixed reference group consisted of 724 normals, Hathaway and Briggs (1957), in the course of providing normative data on a number of additional MMPI scales, replaced many incomplete or defective protocols of the original normals. Their modified sample of 225 men and 315 women became the basis for the T-scores printed on the MMPI standard profile form. T-scores derived from norms in An MMPI Handbook (Dahlstrom et al., 1972, 1975) and other MMPI reference works do not necessarily match precisely the T-scores of the “purified” sample. Subsequently, corrected tables appeared in the MMPI-2 Manual revision (Butcher et al., MMPI-2001).
The clinical criterion (abnormal) groups consisted of carefully selected psychiatric patients, and participants represented the following major diagnostic categories:
The clinical criterion (abnormal) groups consisted of carefully selected psychiatric patients, and participants represented the following major diagnostic categories: