1
A Content Cluster Approach to MMPI-2 Interpretation
For some contemporary users, MMPI-2 interpretation involves reviewing the patient score profile in conjunction with an expertly crafted, computergenerated report. The algorithms underlying interpretive reports variously draw on actuarially derived information. Other users begin with an extended profile of scores obtained either by computer scoring or hand scoring, after which they apply an interpretive strategy based on their training or on a defining text or both. Still others approach the task more or less intuitively, perhaps even idiosyncratically.
The content cluster approach was derived almost exclusively by study of an inpatient population that is typical in many ways of the severely and persistently mentally ill and/or chemically dependent patients encountered in state-supported care settings. Its application to patients from other settings may require retiming and adjustment.
A cluster refers to a group of scales that correlate more strongly with one another than they do with scales assigned to other clusters. Scales within a cluster also share content characteristics that make them recognizable as differing ways of measuring aspects of a common domain. Because the manifest content of many of their items tends toward being transparent, they may be subject to the influence of response styles. Nevertheless, Content scales contain other items that are less transparent and, consequently, less subject to response style influence.
Clusters in use draw on scales of diverse origin: standard clinical scales and new Content scales (Butcher et al., 1989); Indiana Rational Scales (Levitt, 1989); Tryon, Stein, and Chu Cluster Scales (Chu, 1966; Stein, 1968; Tryon, 1966); Wiggins Content Scales (Wiggins, 1966); Harris-Lingoes Subscales (Harris & Lingoes, 1968); Indiana Personality Disorder Scales (Levitt & Gotts, 1995); and a number of individual or stand-alone scales. In addition, clusters contain the following groupings of scales that make their initial appearance here: Impulsivity Components Scales and Critical Inquiry Scales. The Psychosocial Development Scales are of a different order, penetrating and permeating the content clusters.
Clusters are arranged and labeled in the following manner:
I. |
Validity |
II. |
Depressive Spectrum Conditions |
III. |
Anxiety Spectrum Conditions |
IV. |
Thought Process Disorders |
V. |
Cognitive Disorders |
VI. |
Somatic Concerns |
VII. |
Self-Esteem |
VIII. |
Social Maladjustment & Social Anxiety |
IX. |
Anger, Hostility, Suspicion, Cynicism, Resentment |
X. |
Behavior Controls & Norms Violations |
XI. |
Virtue & Morality |
XII. |
Stimulus Seeking & Drive |
XIII. |
Family Relationship Issues |
XIV. |
Social Vulnerability & Alienation |
XV. |
Gender & Sexuality |
XVI. |
Dependency-Dominance |
XVII. |
Personality Disorders |
XVIII. |
Prognosis & Risk Management |
The residues of past psychosocial crises appear in six new scales that reliably examine Eriksonās (1963) psychosocial crises using MMPI-2:
Feels Cared for/Loved vs. Neglected/Disliked (Cared for vs. Neglected)
Autonomous/Self-Possessed vs. Self-Doubting/Shamed (Autonomous vs. Doubting)
Initiating/Pursuing vs. Regretful/Guilt-Prone (Initiating vs. Guilt-Prone) Industrious/Capable vs. Inadequate/Inferior (Industrious vs. Inadequate)
Life-Goal Oriented/Ego Identity Secure vs. Directionless/Confused (Ego Identified vs. Confused)
Socially Committed/Involved vs. Disengaged/Lonely (Involved vs. Disengaged)
The Psychosocial scales are shown in later analyses to offer a latticelike framework within which to view the varieties of normality and psychopathology. In some instances, they may offer insights into the origins of particular disordered patterns of thinking, feeling, and behaving. Next, consider the research activities and results that support the foregoing assertions.
PREVIEWING UPCOMING CHAPTERS AND APPENDICES
Chapters 2ā4 review the literature, including the interpretation of scales in isolation from one another. Chapters 5ā7 present the methods used in the present study, scale updates and revisions, and new scales created for MMPI-2. Essential results of that work appear as Appendices A-C. Chapters 8ā10 explore what is known clinically in terms of the psychosocial framework at its points of intersection with the content clusters. These latter chapters further expose limitations of the DSM approach vis-Ć -vis the MMPI-2. As a result, a selected subset of DSM findings for the database sample is set apart in Appendix D. Chapter 11 juxtaposes Rorschach findings with those available from MMPI-2 and other data sources and identifies occasions when their conjoint use can clarify patient condition. Chapter 12 pulls into unified focus the pragmatics of applying the content cluster approach.
2
Assessing the Quality of the Test Record: The Validity and Response Style Scales
The Validity scales are some of the best known and most widely used scales for the MMPI-2. Before a test profile is analyzed, the clinician uses the Validity scales to lay the foundation for building the psychological profile from the remaining clinical and special scales. The term validity is somewhat misleading. Validity in psychological testing refers to whether the tool measures what it claims to measure (Greene, 2000). The Validity scales on the MMPI-2 capture the respondentās response style. These scales help the examiner to determine if the respondent exaggerated or underreported psychopathology, attempted to fake bad or fake good in the answers, or answered in an inconsistent or random manner. Many reasons exist for the myriad of response styles. A respondent could be trying to appear more pathological to receive disability compensation. In another scenario, respondents could attempt to minimize their pathology in a child custody evaluation to appear more stable and psychologically healthy. The Validity scales have become more than a measure of test accuracy. They provide useful clinical data on the individual respondent.
Hathaway and McKinley (1940) were aware that dissimulation might occur on a verbal inventory. They attempted to develop several validity indicators via internal measures that would point to an individual who was not responding honestly. Sixty years later, this approach is still being touted as one of the cornerstones of the MMPI-2. In fact, the Validity scales have become a powerful tool in diagnostic measurements. In a study by the American Psychological Associationās Psychological Assessment Work Group (Meyer et al., 2001 ), effect sizes were compared between psychological assessments and medical assessments. The popular assumption would seem to be that medical tests would far outweigh the validity of psychological measures. However, this was shown to be untrue: The MMPI-2 Validity Scales demonstrated some of the largest discriminative abilities of the measures compared. The validity measures on the MMPI-2 for identifying malingering of psychopathology produced an effect size of .74, whereas the effect size of an exercise electrocardiogram to diagnose coronary artery disease was only .58. Users of the MMPI-2 can be confident of impressions formed based on the Validity scales in both the clinical and forensic settings.
This chapter overviews the traditional Validity scales: L, F, K, and Cannot Say (?). The Mean Elevation (ME) of the Clinical scales has had past use, but is not commonly used in modern validity profiles. It is updated here. The MMPI-2 brought further improvements, including Variable Response Inconsistency (VRIN; Butcher et al., 1989), True Response Inconsistency (TRIN; Butcher et al, 1989), and Infrequency Back (FB) for the second half of the questions (Butcher et al., 2001) and Infrequency-Psychopathology (Fp; Arbisi & Ben-Porath, 1995). Some older measures of response style are reexamined for their value along with more recent ones. These include the Wiggins (1959) Social Desirability Scale (Sd) and the Edwards (1957) Social Desirability Scale (So). New approaches to consistency measurement are treated at the end of this chapter.
TRADITIONAL VALIDITY MEASURES
Lie Scale (L)
The Lie (L) scale was constructed by the developer of the MMPI on a completely nonempirical, rational basis. It consists of 15 statements whose content is such that a truthful respondent has a high probability of responding āTrueā to any one of the statements. The intent of the L scale was that it should provide an index of the extent to which patients attempt to answer the MMPI items by selecting those responses that represent them most favorably to others (Hathaway & McKinley, 1951). However, the gambit is rather obvious to anyone with at least average intelligence and a somewhat developed amount of psychological mindedness. Normative studies have indicated that the average raw score is between 3 and 5 (Greene, 2000). The L scale is, however, not an effective measure for detecting underreporting of psychopathology in psychologically sophisticated persons.
An investigation by Lebovits and Ostfeld (1967) showed that L scale scores and educational level are negatively correlated. The same can be seen with respondents of lower socioeconomic (SES) status (W.G.Dahlstrom & Tellegen, 1993). The respondent from within a low SES and/or low education group may not be purposefully trying to falsify their scores when the L scale is elevated. Thus, interpretation must be formulated within its proper cultural context.
Interpretation of the L scale requires sensitivity to the assessment context in which it was given....