Contributions to the Dimensional Assessment of Personality Disorders Using Millon's Model and the Millon Clinical Multiaxial Inventory (MCMI–III)

Abstract

For over 35 years, Mllion's (1996) model of personality and the Millon Clinical Multiaxial Inventory (Millon, 1977, 1987, 2006) have been useful resources for clinicians to understand and assess personality disorders (PDs) and clinical syndromes in psychiatric patients. In this article, we highlight significant features of the model and test that have proved valuable to personologists in their quest for a more satisfactory taxonomy of PDs based on continuously distributed traits. We also describe Millon's (1996) prototypal domain approach to personality that combines dimensional and categorical elements for the description of PDs and their normal counterparts.

ACKNOWLEDGMENTS

Participation of Stephen Strack was supported by the U.S. Department of Veterans Affairs. We thank Susan Stanton and Brandon Yakush for helpful comments on an early draft of the manuscript.

Notes

¹The word prototype is commonly used in the English language to mean the model on which something is based or formed. Applied to personality science, the term refers to the most common features or properties of members of a category, in this case a personality type, style, or disorder. A personality prototype is a theoretical ideal or standard against which real people can be evaluated. All of the prototype's properties are assumed to characterize at least some members of the category, but no one property is necessary or sufficient for membership in the category. It is possible that no actual person would match the theoretical prototype perfectly. Different people approximate it to different degrees, and the closer a person comes to matching all the definitional criteria, the more closely that person typifies the concept (Rosch, 1978). Millon's (1996, 2006) taxonomy describes personality in terms of prototypes, but for stylistic purposes, we use the terms type, style, disorder, and prototype interchangeably in this report.

²Should the American Psychiatric Association adopt a diagnostic system that included normal personality, the MCMI would have to be revised to measure a broader range of normal traits. Current MCMI–III PD scales assess mostly abnormal features.

³The concept of personality subtypes does not negate the validity or importance of personality prototypes any more than color blends negate the reality of the primary colors from which they are derived.

⁴Although the MCMI–III and other assessment instruments can be summarized as being accurate or inaccurate for making particular diagnoses within particular parameters, the research literature is still far from complete in terms of assaying all of the variables that affect clinical diagnosis such as state–trait effects, comorbidity, and neurological factors. As such, readers are cautioned not to overgeneralize the empirical findings presented here and elsewhere regarding the MCMI–III's diagnostic accuracy.

⁵ Specificity is also frequently used as an estimate of a measure's diagnostic accuracy. This is defined as the number of patients who are determined by clinicians as not having a particular disorder who are also shown not to have the disorder by the measure. Research on the MCMI–I demonstrated that the PD scales have high specificity (range = .85–.95; Gibertini et al., 1986), and this was confirmed for the MCMI–III (M for all scales = .97; Hsu, 2002). Specificity is similar to negative predictive power, which we present in Table 2.

⁶ Whereas Millon (1996) hypothesized that many phenotypically expressed personality characteristics will show continuous distributions in both normal and abnormal samples (e.g., introversion–extraversion), other features important to personality classification (such as active–passive orientation) will not necessarily emerge in dimensional form. Because of this, Millon (1996) believed that factor analysis and dimensionalized traits alone cannot serve as an adequate foundation for a taxonomy of PDs—too many elements would be lost.

⁷In this regard, it is unfortunate that the vast majority of published studies on Millon's (for example, 1996) body of work have focused on his assessment measures rather than his model of personality and psychopathology. We strongly encourage research on his model that is independent of his measures because assessment instruments such as the MCMI are imperfect and limited in how they operationalize many theoretical variables.