JOURNAL OF PERSONALITY ASSESSMENT, 72(3), 365–389 Copyright © 1999, Lawrence Erlbaum Associates, Inc.

Millon’s Dimensional Polarities Thomas A. Widiger Department of Psychology University of Kentucky

Millon has presented a grand and rich theory of personality disorder. Fundamental to this theory are three polarity dimensions (active–passive, pleasure–pain, and self–other) from which the personality disorders are said to be derived. This article reviews these polarity dimensions with respect to their conceptual relation with the personality disorder nomenclature, their assessment, and their empirical support.

One of the more common criticisms of the American Psychiatric Association’s (APA) classification of personality disorders in the Diagnostic and Statistical Manual of Mental Disorders (4th ed. [DSM–IV]; APA, 1994) is the absence of a unifying theoretical model. A variety of models have been proposed, including (but not limited to) the lexical five-factor model (Costa & Widiger, 1994), the interpersonal circumplex model of Kiesler (1996), the interpersonal and object-relational three-circumplex model of Benjamin (1993), the temperament and character seven-factor model of Cloninger (Cloninger & Svrakic, 1994), the three- and seven-factor models of Tellegen (Tellegen & Waller, in press), and the neurochemical dysregulation four-factor model of Siever (Siever & Davis, 1991). Millon’s theoretical perspective warrants particular attention as he has been an influential participant in the development of various editions of the personality disorder nomenclature. The inclusion of the avoidant personality disorder in the third edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM–III; APA, 1980; Gunderson, 1983) and the deletion of the passive–aggressive personality disorder in DSM–IV (Gunderson, 1998) are due in large part to proposals and suggestions of Millon. The Millon Clinical Multiaxial Inventory–III (MCMI–III; Millon, Millon, & Davis, 1994) might be the most favored self-report inventory among practicing clinicians for the assessment of the DSM–IV personality disorders (Watkins, Campbell, Nieberding, & Hallmark, 1995). Millon’s (1981) text on DSM–III may also be the best known and most widely cited text on personality disorders. The book is indeed a classic within the personality disorder literature. The more

366

WIDIGER

recently updated edition by Millon and Davis (1996) remains the major scholarly source for historical and current information concerning the DSM–IV personality disorders. The purpose of this article is to review Millon’s theoretical model, particularly with respect to its logic, assessment, and empirical support. Millon’s model for the personality disorders is rich in its depth and creativity, and it has not perhaps received the attention from researchers that it deserves. Strack is to be commended for developing this symposium. My presentation raises a number of concerns regarding the model, but these comments are presented in the spirit of generating fruitful discussion and additional informative research.

THE THREE POLARITIES Central to Millon’s (1969, 1981) theoretical model are three polarity dimensions from which the personality disorders are said to be derived (Millon & Davis, 1996). Freud (1915/1953) proposed that “our mental life as a whole is governed by three polarities, namely, the following antitheses: subject (ego)–object (external world), pleasure–pain, active–passive” (p. 113). “Some 50 years later Millon (1969) utilized these same polarities [self–other, pleasure–pain, and active–passive] in constructing a series of eight basic personality patterns” (Millon, 1981, p. 46). One can readily imagine eight personality patterns being derived from the crossing of the three polarities of self versus other, pleasure versus pain, and active versus passive. “Eight basic coping patterns … were derived by combining the nature (positive or pleasure vs. negative or pain), the source (self vs. others), and the instrumental behaviors (active vs. passive)” (Millon & Davis, 1996, p. 67). One could, for example, have a self-oriented person who actively seeks pleasure, a self-oriented person who passively seeks pain, and an other-oriented person who passively seeks pain. However, one potential misunderstanding of Millon’s model is that the three polarities are not in fact used in this manner to derive the eight personality patterns. The eight patterns that Millon derived are instead an imbalanced or uneven mixture of the three polarities. Millon’s (1969) original biosocial-learning model of personality described eight personality disorders derived from two (or three) constructs: “A classification scheme based on (1) the nature and source of reinforcements and (2) styles of instrumental behavior provides a theoretical framework for what we believe to be the eight principal pathological personality patterns” (p. 195). Four personality patterns are derived from the nature and source of reinforcements (detached, dependent, independent, and ambivalent) and these four are crossed with two variants of instrumental behavior (active, passive) to yield eight personality patterns. The two variants of instrumental behavior (active, passive) do represent cleanly the active–passive polarity, but the detached, dependent, independent, and ambiv-

MILLON’S DIMENSIONAL POLARITIES

367

alent patterns do not represent a clean crossing of the two self–other and pleasure–pain polarities. Millon (1969) asked rhetorically on what basis a relevant, fruitful distinction among reinforcements could be made, and he answered: Two possibilities present themselves. First, whether or not the patient does find certain events to be reinforcing. And second, what the nature of these reinforcements are, that is, whether he seeks primarily to achieve positive reinforcements, or to avoid negative reinforcements. (p. 193)

The positive versus negative nature of these reinforcements does appear to be the pleasure–pain polarity. However, only one of the four personality patterns, the detached, is associated explicitly with the nature of the reinforcement, and even this pattern is not described as either purely pleasure or pain oriented. “Patients either characterized by deficits in the capacity to experience reinforcement or oriented primarily to negative reinforcers, display a detached pattern” (Millon, 1969, p. 194). The detached pattern is the presence of either the pain polarity or the absence of an interest in either pleasure or pain. “Both groups share a deficit capacity to sense pleasurable reinforcers, although one is hyperreactive to pain. Millon describes both of these as detached patterns” (Millon, 1981, p. 59). Two of the three remaining personality patterns, the dependent and independent, appear to be clean representations of the self–other polarity. “The distinction between others and self as the primary reinforcement source underlies our categories of dependent and independent personalities” (Millon, 1969, p. 194). “Dependent personalities have learned that feeling good, secure, confident, and so on—that is, those feelings associated with pleasure or the avoidance of pain—are best provided by others” (Millon, 1981, p. 59). “Independent personality patterns, in contrast, are characterized by a reliance on the self” (Millon, 1981, p. 59). The fourth pattern, ambivalent, represents persons who can neither be classified as either self- or other-oriented. “Such clear-cut commitments are not made by all patients” (Millon, 1969, p. 195). “Some people, those whom Millon speaks of as ambivalent, remain unsure as to which way to turn; that is, they are in conflict as whether to depend on themselves for reinforcement or on others” (Millon, 1981, p. 59). In sum, the active–passive, self–other, and pleasure–pain polarities could be used to identify eight personality patterns, but it is unclear if the eight personality patterns Millon derived from these polarities are in fact logical derivations from these three polarities (O’Connor & Dyce, 1998). Table 1 indicates how the eight personality patterns relate to the three polarities. The active–ambivalent and passive–ambivalent personality patterns represent both poles of the self–other polarity rather than one of the two poles; there are only two pain-oriented patterns rather than four (both of which can be as detached from pain as they are oriented to pain);

368

Active–detached? (avoidant)

Pain

Pleasure

Pleasure

Other

Active–independent Active–dependent (antisocial) (histrionic) Active–ambivalent (passive–aggressive)

Self

Active

Pain Passive–detached (schizoid)

Pain

Pleasure

Pleasure

Other

Passive–independent Passive–dependent (narcissistic) (dependent) Passive–ambivalent (compulsive)

Self

Passive

TABLE 1 The Eight Basic Personality Patterns With Respect to the Three Polarities

Pain

MILLON’S DIMENSIONAL POLARITIES

369

and there are (perhaps) no personality patterns to represent the two other–pain orientations. One minor note is that the active–detached and the passive–detached patterns are classified in Table 1 as self-oriented, but this might not be correct. The self–other orientation of the detached patterns was not explicitly described by Millon (1969, 1981), but the clinical descriptions of these avoidant (active–detached) and schizoid (passive–detached) personality disorders would suggest a placement opposite to the other–orientation of the histrionic (active–dependent) and dependent (passive–dependent) personality disorders. Alternatively, the avoidant (active–detached) personality disorder might be considered to be other-oriented, as the avoidant personality is said to be motivated in large part by the avoidance of negative reinforcements from others. However, if the active–other–pain orientation cell is filled by the active–detached pattern, the active–self–pain then becomes empty (Millon & Davis, 1996, subsequently classified the avoidant personality disorder as being average on the self–other polarity). Revisions for DSM–III–R The polarities and the personality patterns derived from them were subsequently revised by Millon (1987) to include a discordant personality pattern that covered the addition of the self-defeating and sadistic personality disorders in the Diagnostic and Statistical Manual of Mental Disorders (3rd ed., rev. [DSM–III–R]; APA, 1987): The theory is based on derivations from a simple combination of a few variables or constructs. Essentially, it posits 10 basic styles of personality functioning that can be formed logically from a 5 × 2 matrix consisting of two basic dimensions. Previously, 8 basic styles were posited from a 4 × 2 matrix and served as the basis for the MCMI–I basic personality disorder scales. (Millon, 1987, p. 18)

Five personality styles crossed with the active–passive polarity does logically result in 10 personality patterns. “Combining the five sources of primary reinforcement together with the two instrumental or coping patterns results in 10 basic personality styles” (Millon, 1987, p. 18). As Millon et al. (1994) indicated, the theoretical model “posits basic styles of personality functioning that can be formed logically from a 5 × 2 matrix consisting of two basic dimensions” (p. 10). What remains confusing, however, is how the one dimension that concerns the source of reinforcement and defines five personality patterns represents logically the two active–passive and pleasure–pain polarities. The new discordant pattern can be conceptualized as a pain-oriented disposition. “Circumstances which lead patients to reverse positive or negative reinforcements, or to substitute pain for pleasure are termed discordant personalities” (Millon, 1987, p. 18). But the new

370

WIDIGER

discordant pattern did not actually fill the empty cells of the original model. It may have added instead more ambiguity and complexity. Table 2 provides Millon’s (1987) descriptions of 10 personality patterns with respect to the two dimensions (Table 2 also includes the three more severely dysfunctional variants of the 10 basic personality patterns). The descriptions provided in Table 2 suggest that the pleasure–pain distinction is irrelevant to six of the personality patterns (i.e., the active and passive variants of the dependent, independent, and ambivalent personality patterns). This did not appear to be the intention of the original version of the model, although their placement with respect to pleasure–pain was not explicit. The revision is equally unclear regarding the placement of the detached (schizoid and avoidant) and discordant (self-defeating and sadistic) patterns with respect to the self–other distinction. The 1987 revision of the model, as presented in Table 2, suggests that the self–other distinction is largely irrelevant to the understanding of the avoidant and schizoid personality disorders. Millon (1987) also presented the model in the form of a circumplex, as indicated in Figure 1. A circumplex will be cleanly defined by two dimensions, but there appear to be a number of important inconsistencies between the Figure 1 circumplex presentation and the Table 2 tabular presentation. First, the constructs are substantively revised. The circumplex is defined by the dimensions of affiliation (autonomy and enmeshment) and emotionality (impassive and expressive), not by the dimensions of source of reinforcement (self–other and pleasure–pain) and coping pattern (active–passive).

TABLE 2 Representation of DSM–III–R Personality Patterns With Respect to Polarities by Millon (1987) Pathology Domain Self–Other

Reinforcement Source

Pain–Pleasure

Other + Pleasure –

Self + Other –

Self ↔ Other

Pain ↔ Pleasure

Pleasure – Pain ±

Discordant

Detached

Instrumental coping style–interpersonal pattern Passive variant

Dependent

Independent

Ambivalent

Dependent

Narcissistic

Compulsive

Active variant

Histrionic

Antisocial

Dysfunctional variant

Borderline

Paranoid

Self-defeating (masochistic) Passive– Aggressive aggressive (sadistic) Borderline or paranoid

Schizoid Avoidant Schizotypal

Note. DSM–III–R = Diagnostic and Statistical Manual of Mental Disorders (3rd ed., rev). From Manual for the MCMI–II (2nd ed., p. 19), by Theodore Millon, 1987, Minneapolis, MN: National Computer Systems. Copyright 1987 by Theodore Millon. Reprinted with permission.

MILLON’S DIMENSIONAL POLARITIES

371

FIGURE 1 Circumplex presentation of two-dimensional model by Millon (1987). From Manual for the MCMI–II (2nd ed., p. 20), by Theodore Millon, 1987, Minneapolis, MN: National Computer Systems. Copyright 1987 by Theodore Millon. Reprinted with permission.

There is some correspondence between the constructs. The emotionality dimension, with the two poles of impassivity and expressivity, would correlate with a passivity–activity dimension; and the circumplex affiliative dimension, with the two poles of autonomy and enmeshment, does appear to correspond with the self–other polarity. However, it is then unclear how the pleasure–pain polarity is being represented. More important, the placement of individual personality disorders within the figure is inconsistent with the tabular presentation, due in part perhaps to the difficulty representing the two self–other and pleasure–pain polarities within one dimension and the substantive changes to the constructs. For example, the dependent (passive–dependent) personality disorder is described as passive within Table 2, but neutral with respect to the impassive versus expressive dimension within Figure 1. The narcissistic (passive–independent) and self-defeating (passive–discordant) personality disorders are described as passive within Table 2 but are described as active

372

WIDIGER

within Figure 1. Personality disorders placed next to each other within the figure should be similar with respect to the dimensions of the model. The histrionic and narcissistic personality disorders are placed next to one another within the figure, yet are opposite one another with respect to the fundamental active–passive polarity.

Evolutionary Model “In 1990, Millon reconceptualized his model of personality and its disorders” (Millon & Davis, 1996, p. 69). The shift was from a largely ontogenetic theory of neuropsychological stages to a more evolutionary, phylogenetic theory of human and even biological species development (Millon, 1990). “In his 1990 book, Millon attempted to illustrate the universals he judged were fundamental to all spheres of evolution’s progression, including those of human existence” (Millon & Davis, 1996, p. 70). The evolutionary model posits four fundamental polarities of evolutionary development: (a) aims of existence (life enhancement and preservation), (b) modes of adaptation (ecological accommodation and modification), (c) strategies of replication (reproductive individuation and nurturance), and (d) processes of abstraction: Each evolutionary phase is related to a different stage of ontogenetic development (Millon, 1969). For example, life enhancement–life preservation corresponds to the sensory-attachment stage of development in that the latter represents a period when the young child learns to discriminate between those experiences that are enhancing and those that are threatening. (Millon & Davis, 1996, p. 100)

The broadening of the model to include phases of evolution across all species did not require significant revision to the original three polarities. “What was most gratifying in this [1990] exploratory search was the close correspondence between the 1969 biosocial-learning theory and the key elements of the evolutionary model” (Millon & Davis, 1996, p. 70). The three polarities of pleasure–pain, active–passive, and self–other correspond to three evolutionary phases of existence, adaptation, and replication, respectively. For example, the aim of existence is life enhancement and life preservation. “Events that extend life and preserve it correspond largely to metaphorical terms such as ‘pleasure’ and ‘pain,’ that is, eliciting positive sensations and emotions, on the one hand, and eschewing negative sensations and emotions, on the other” (Millon, 1990, p. 51). The evolutionary accommodation and modification phase corresponds to the active–passive polarity: “Plants best characterize the mode of ecologic accommodation, an essentially passive style. … Animals, in contrast, typify what has been termed adaptation via ecologic modification, an essentially active style which intervenes and transforms the surrounds” (Millon, 1990, p. 65). The third phase of evolutionary develop-

MILLON’S DIMENSIONAL POLARITIES

373

ment, strategies of replication (propagation and nurturance), corresponds to the self–other polarity. For example Males tend to be self-oriented owing to the fact that competitive advantages that inhere within themselves maximize the replication of their genes. Conversely, females tend to be other-oriented owing to the fact that their competence in nurturing and protecting their limited progeny maximize the replication of their genes. (Millon, 1990, p. 78)

Table 3 provides Millon’s (1990, p. 2) classification of the DSM–III–R personality disorders from the perspective of the evolutionary model. This tabular presentation does indeed indicate the consistency of the phylogenetic and ontogenetic polarities. Tables 2 and 3 are consistent with one another. One minor point is the absence of a representation of the fourth phase of phylogenetic and ontogenetic development (abstraction) that results in a new polarity of thinking versus feeling. More important, however, is that the conceptual ambiguity of including the pleasure–pain and self–other polarities within one dimension has remained unclarified. Millon (1990) continued to refer to the model as a 5 × 2 two-dimensional personality structure, from which 10 personality patterns are derived: It was both feasible and productive to employ the key dimensions of the bipolar model to make the clinical features of the ten basic “styles” of personality functioning more explicit, from the actively pain-sensitive avoidant to the passively self-centered narcissist. (p. 129)

However, it is now even more difficult to conceptualize the existential aims (pleasure–pain) and replication strategies (self–other) as defining a single dimension of evolutionary development or affiliative personality functioning. These two phases of development are even said to be separated ontogenetically by the occurrence of the adaptation (active-phase) phase. Millon and Davis (1996) Millon and Davis (1996) more recently provided a clarification of the model that does indicate the placement of each of the 10 DSM–IV personality disorders on each of the three active–passive, pleasure–pain, and self–other dimensions. Table 4 presents a summary of these descriptions. For example, it was unclear in Millon (1987, 1990) whether the pleasure–pain dimension was relevant to the dependent, histrionic, antisocial, and narcissistic personality disorders (see Table 2). Millon and Davis (1996) indicated that the histrionic, narcissistic, and dependent personality disorders are average with respect to the pleasure–pain dimension.

374

Schizoid Avoidant Schizotypal

Pleasure – Pain – +

Self – Other +

Dependent Histrionic Borderline

Self–Other

Narcissistic Antisocial Paranoid

Self + Other –

Reproductive Propagation

DSM–III–R Personality Disorders Self-defeating Sadistic Borderline–paranoid

Pleasure Pain

Pleasure–Pain

Life Preservation

Compulsive Passive–aggressive Bordeline–paranoid

Self Other

Reproductive Nurturance

Replication Strategy

Note. DSM–III–R = Diagnostic and Statistical Manual of Mental Disorders (3rd ed., rev.). From Toward a New Personology: An Evolutionary Model (p. 128), by Theodore Millon, 1990, New York: Wiley. Copyright © 1994 John Wiley & Sons, Inc. Reprinted by permission of John Wiley & Sons, Inc.

Passive: accomodation Active: modification Dysfunctional

Adaptation Mode

Pathology deficiency, imbalance, or conflict

Polarity

Life Enhancement

Existential Aim

TABLE 3 Representation of DSM–III–R Personality Disorders With Respect to the Evolutionary Model

MILLON’S DIMENSIONAL POLARITIES

375

Lawrence Erlbaum Associates does not hold electronic rights to Table 4. Please see the print version.

However, the specification by Millon and Davis (1996) also highlights the ambiguities, as the personality disorders are now classified with respect to both poles of the polarity dimensions, as if each of the poles refers to an independent dimension. For example, the schizoid personality disorder is said to be low (or weak) in the pursuit of pleasure and in the avoidance of pain. On the one hand, this does appear to be a valid description of this disorder, consistent with the clinical description of anhedonia (APA, 1994): “They are unmotivated to seek out joy and gratification, are unable to view life enthusiastically, and also, they experience none of the distressing affects of life, such as sadness, anxiety, and anger” (Millon & Davis, 1996, p. 228). However, classifying a disorder as being low on both pleasure and pain might be problematic to the validity of pleasure–pain as a single polarity dimension. Persons can not be both tall and short with respect to the dimension of height, both brilliant and retarded with respect to the dimension of intelligence, or both introverted and extroverted with respect to the polarity dimension of introversion versus extroversion. Persons who are highly elevated on one pole of a single dimension (e.g., highly extroverted) are logically considered to be low with respect to the other pole (i.e., not at all introverted). One might interpret being low on both pleasure and pain as being simply average with respect to the dimension. Persons who are neither highly introverted nor

376

WIDIGER

highly extroverted would be classified within the middle of a distribution of introversion versus extroversion. However, this does not appear to be the intention of describing the schizoid personality disorder as being low on both pleasure and pain. It is the histrionic, narcissistic, and dependent personality disorders that are described in Table 4 as being average with respect to pleasure and pain. The schizoid personality disorder is not said to be average on both pleasure versus pain; the schizoid personality disorder is low on pleasure and pain. In sum, Table 4 does not describe the DSM–IV personality disorders along three polarity dimensions of pleasure–pain, active–passive, and self–other. Table 4 describes a placement along six independent dimensions of pleasure, pain, active, passive, self, and other. Millon and Davis (1996) described the model as consisting of three active–passive, self–other, and pleasure–pain polarity dimensions (or as two circumplical dimensions of affiliation and emotionality). “Every individual acquires personologic dispositions representing a balance or predilection toward one of the two polarity inclinations” (Millon & Davis, 1995, p. 638). However, Table 4 is describing a placement along six independent dimensions. This might be a mistaken interpretation of Table 4, as it is difficult to imagine that the most dysfunctional personality disorder, borderline, is neither high nor low on any one of the six (or three) dimensions (i.e., borderline is described in Table 4 as being average on all six dimensions), but this does appear to be the logical understanding of the description of the disorders with respect to the polarity dimensions as presented in Table 4. ASSESSMENT OF THE THEORETICAL MODEL Millon has authored a number of personality and clinical inventories (Millon & Davis, 1994). These inventories are said to provide operational measures of the theoretical model, and studies have been published using these inventories to provide empirical studies concerning the validity of the model. Space limitations prohibit a discussion of all of the instruments. Only two, the MCMI–III (Millon et al., 1994) and the Millon Index of Personality Styles (MIPS; Millon, 1994), are discussed within this article. Two additional inventories of particular importance to the assessment of the theoretical model are the Personality Adjective Check List (PACL; Strack, 1987, 1991b) and the Millon Behavioral Health Inventory (MBHI; Millon, Green, & Meagher, 1982), as the PACL and the MBHI provide scales to assess the eight personality styles or coping strategies (respectively) that are derived logically from the theoretical model. MCMI The MCMI–III is perhaps the most appropriate instrument with which to assess empirically Millon’s theoretical model of personality functioning, as it is the instru-

MILLON’S DIMENSIONAL POLARITIES

377

ment that is most directly concerned with the assessment of the central focus of the theory, personality disorders. However, it might be difficult to use the MCMI–III (or any prior edition) to conduct research on the theoretical model, as none of the scales within the MCMI–III corresponds to the fundamental constructs of the model. There are no self–other, pleasure–pain, or active–passive scales within the MCMI–III, nor any scales to assess the circumplical emotionality and affiliation dimensions. It might be possible to derive such scales from MCMI–III items. The MCMI–I items were said to be written to assess (or represent) the theoretical model: “The clinical characteristics of each scale were derived from theory-based formulations of personality types” (Millon, 1977, p. 3). “Eight basic styles were posited from a 4 × 2 matrix and served as the basis for the MCMI–I basic personality disorder scales” (Millon, 1987, p. 18): Each of [the personality disorder] scales was constructed as an operational measure of a syndrome derived from a theory of personality and psychopathology (Millon, 1969, 1981). As such, the scales and profiles of the MCMI measure theory-derived variables directly and quantifiably. (Millon, 1983b, p. 1)

This purported consistency with the theoretical model has continued up to and including the MCMI–III: “In generating new items, the goal was to produce statements that were congruent with the instrument’s guiding theoretical system” (Millon et al., 1994, p. 17). Few diagnostic measures have either been based on or have evolved from clinical theory. The MCMI is different. Each of its personality disorder … scales was constructed as an operational measure of a syndrome derived from a theory of personality and psychopathology (Millon, 1969, 1981, 1990). (Millon & Davis, 1994, p. 103)

To the extent that the MCMI–III personality disorder scales are indeed operational measures of Millon’s theoretical model for these personality disorders, it should not be difficult to provide self–other, pleasure–pain, and active–passive scales from the personality disorder scales in a manner consistent with the theoretical model for these personality disorders. One can indeed discern some representation of the theoretical model within Millon’s descriptions of the personality disorders and perhaps within the scales. For example, the passive accommodation and other nurturance orientation of the dependent personality disorder is evident in the description of the disorder as involving a lack of confidence, lack of initiative, and abdication of responsibilities, along with isolation anxiety and excessive dependency. Widiger, Williams, Spitzer, and Frances (1985) indicated how items from the MCMI Antisocial and Histrionic scales appeared to be closer to Millon’s (1981) descriptions of the aggressive (active–self) and gregarious (active–other)

378

WIDIGER

personality patterns than to the DSM–III antisocial and histrionic personality disorders, respectively. The MCMI and MCMI–II Obsessive–Compulsive scales have often obtained zero to negative correlations with other measures of this personality disorder (e.g., Streiner & Miller, 1988) due in part perhaps to its closer adherence to Millon’s (1981) formulation of this disorder than to the description provided within the DSM–III or the DSM–III–R (Widiger & Corbitt, 1993). However, many (if not most) of the MCMI–III items were also written or rewritten to represent the DSM–IV diagnostic criteria rather than the self–other, pleasure–pain, and active–passive theoretical constructs. In the section of the most recent edition of the manual (MCMI–III) that is concerned with item development, examples of items for each of the scales are provided. All of the examples refer to DSM–IV diagnostic criteria. No mention is made in the presentation of examples of the theoretical constructs from which the scales were purportedly derived. For example, the following is the description of the MCMI–III items for the schizoid personality disorder scale: Scale 1 (Schizoid). The second DSM–IV schizoid criterion is “almost always chooses solitary activities.” The comparable test item (Item 27) is “When I have a choice, I prefer to do things alone.” The third DSM–IV criterion is “has little, if any, interest in having sexual experiences with another person.” In the test this is phrased “I’ve always had less interest in sex than most people do” (Item 46). The fourth DSM–IV criterion is “takes pleasure in few, if any activities.” In the test this became “Few things in life give me pleasure” (Item 148). (Millon et al., 1994, p. 17)

It is possible that many of the other schizoid items were written to assess or represent the self–other, pleasure–pain, and active–passive polarities, but no examples are provided in the test manual. It is then difficult to determine whether the MCMI–III Schizoid scale is an assessment of Millon’s theoretical model for this personality disorder or simply a scale developed to assess the DSM–IV diagnostic criteria for this disorder. Millon, however, is perhaps somewhere in between a rock and a hard place. A major purpose of the MCMI–III is to provide a valid assessment of the DSM–IV personality disorders (Millon et al., 1994). Its success as a clinical instrument is dependent primarily on its validity as a measure of the DSM–IV personality disorders, and most of the diagnostic criteria for the DSM–IV personality disorders were not based on Millon’s (1969, 1981; Millon & Davis, 1996) theoretical model. Millon (1983a) did write initial drafts of the diagnostic criteria for the DSM–III personality disorders, but the final versions of these criteria sets, with the exception of the avoidant personality disorder, appear to have been derived largely from other sources (Frances, 1980; Widiger et al., 1985). Substantial revisions were also made to these criteria sets for DSM–III–R, and these revisions were not based on Millon’s theoretical model (Widiger, Frances, Spitzer, & Williams, 1988). Further revisions were made

MILLON’S DIMENSIONAL POLARITIES

379

for DSM–IV, and these revisions were again more heavily influenced by other research and other theoretical models, with the exception of the passive–aggressive personality disorder (Gunderson, 1998; Millon, 1993; Widiger, Mangine, Corbitt, Ellis, & Thomas, 1995). It would be difficult for any instrument to be both fully congruent with the DSM–IV personality disorders and with a particular theoretical model for these personality disorders, given the diverse representation of alternative theoretical models within DSM–IV (Livesley, 1998; Millon & Davis, 1996). As a result, perhaps the MCMI–III is being unduly criticized from both sides of the fence. Widiger et al. (1985) were critical of the extent to which the MCMI was more congruent with Millon’s (1981) theoretical model than with the DSM–III. This article, in contrast, is critical of the extent to which the MCMI–III fails to represent the theoretical model. It is perhaps inconsistent or excessive to criticize the MCMI for being too close to the theory, and the MCMI–III for being too distant from the theory. On the other hand, this article is not so much a criticism of the failure of the MCMI–III to represent the theoretical model but of a failure to make the extent of this representation explicit, clear, or testable. MIPS Millon (1994) developed a self-report personality instrument that does provide scales with which to assess the active–passive, self–other, and pleasure–pain polarity dimensions. The instrument, the MIPS, provides scales to assess 24 personality dimensions. “The first set of scales of the MIPS measure the three polarity pairs of the theory directly (e.g., pleasure–pain, active–passive, and self–other)” (Millon & Davis, 1994, p. 107). The three polarity dimensions are assessed by six scales: Enhancing, Preserving, Modifying, Accommodating, Individuating, and Nurturing. The titles of the six scales refer to the evolutionary aims, but they are coordinated explicitly with the three polarities. The Enhancing and Preserving scales correspond to pleasure versus pain, respectively; Modifying and Accommodating correspond to activity versus passivity, respectively; and Individuating and Nurturing correspond to self versus other, respectively. For example, “whether individual human males and females … give evidence of a balance between dispositions toward self and other is gauged by the third pair of Motivating Aims of the MIPS, Individuating versus Nurturing” (Millon, 1994, p. 20). The presence of six scales to assess the three polarity dimensions, however, is somewhat confusing. “The model comprised three bipolar dimensions: positive versus negative reinforcement (pleasure/pain); self versus other as the source of reinforcement, and the instrumental styles of active versus passive” (Millon, 1994, p. 13). One might then have expected the presence of three scales, each of which assessed a single bipolar dimension, comparable to an introversion versus extroversion scale. For example, the Individuating and Nurturing scales concern “two

380

WIDIGER

broad orientations [that] reflect a fundamental bipolarity that exists in nature and expresses itself in two contrasting aims of human motivation/emotion” (Millon, 1994, p. 20). The two scales assess A frequently studied dimension that contrasts striving, self-indulgent, enterprising, courageous, competitive, ambitious, dominating, self-assured, and independent traits, at the one extreme, with love-oriented, altruistic, nurturant, intimate, harmony-seeking, warm, trusting, and cooperative behaviors, at the other. (Millon, 1994, p. 20)

It is unclear why two scales are used to assess one dimension with contrasting poles, as persons might then be classified as being high (or low) on both poles of the one dimension. The presence of two scales is consistent, however, with the six-dimensional model presented by Millon and Davis (1996). It is hypothetically possible, using the MIPS to assess the six dimensions, to have persons with a borderline personality disorder obtain an average score on all six dimensions or to have schizoid persons obtain a low score on both the Enhancing (pleasure) and Preserving (pain) dimensions. On the other hand, it should also be noted that the two scales that assess each of the respective polarities overlap substantially. For example, many (but not all) of the items keyed false on the Preserving (pain) scale are keyed true on the Enhancing (pleasure) scale (and vice versa). Thirty-one percent of the Enhancing (pleasure) and Preserving (pain) items are shared; 38% of the Modifying (active) and Accommodating (passive) items are shared; and 24% of the Individuating (self) and Nurturing (other) items are shared. In a sample of 1,000 persons, the Enhancing (pleasure) and Preserving (pain) scales correlated –.87 and the Modifying (active) and Accommodating (passive) scales correlated –.80 (Millon, 1994), due in large part to the scale overlap; the correlations were –.46 and –.32, respectively, without the overlapping items. However, the Individuating (self) and Nurturing (other) scales correlated only –.47 (–.19 without the overlapping items; Millon, 1994), suggesting perhaps a significant degree of independence of the selforiented and other-oriented poles of the self–other polarity dimension.

EMPIRICAL SUPPORT FOR THE THEORETICAL MODEL Studies with the MIPS might provide the most direct assessment of the validity of Millon’s theoretical model, given the presence within the MIPS of scales to assess explicitly the active–passive, self–other, and pleasure–pain polarities. Table 5 provides the correlations of these MIPS scales with the five domain scales from the NEO–Personality Inventory (NEO–PI; Costa & McCrae, 1985) obtained from a sample of 60 community adults (48% female; Millon, 1994).

MILLON’S DIMENSIONAL POLARITIES

381

TABLE 5 Correlations of MIPS Polarity Dimensions With NEO–PI Domain Scales NEO–PI Scale MIPS Scales Enhancing (pleasure) Preserving (pain) Modifying (active) Accommodating (passive) Individuating (self) Nurturing (other)

N

E

O

A

C

–.70* .80* –.46* .62* .00 .09

.48* –.41* .49* –.51* –.02 .38*

.17 –.10 .22 –.19 .04 .02

–.02 –.04 –.16 .06 –.47* .29

.27 –.19 .53* –.45* .20 .22

Note. MIPS = Millon Index of Personality Styles (Millon, 1994); NEO–PI = NEO–Personality Inventory (Costa & McCrae, 1985); N = neuroticism; E = extraversion; O = openness; A = agreeableness; C = conscientiousness. N = 60. From the Manual of the Millon Index of Personality Styles. Copyright © 1994 by Dicandrien, Inc. Reproduced by permission of the publisher, The Psychologial corporation. All rights reserved. No part of the items may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without express written permission from the publisher. *p < .01.

It is evident from Table 5 that MIPS Preserving (pain) relates heavily to NEO– PI Neuroticism. The MIPS test manual provides correlations with many other personality inventories, including the Sixteen Personality Factor Questionnaire (Cattell, Eber, & Tatsuoka, 1970), the Myers–Briggs Type Indicator (Myers & McCaulley, 1985), the California Psychological Inventory (Gough, 1987), and the Personality Research Form (Jackson, 1985). The highest correlations obtained by the MIPS Enhancing and Preserving scales were with NEO–PI Neuroticism. Nevertheless, MIPS Enhancing and Preserving also correlated substantially with NEO–PI Extraversion. One would have expected and Millon (1994) did obtain a close correspondence of the Modifying (active) versus Accommodating (passive) polarity dimension with Extraversion versus Introversion, given that activity versus passivity is one of the facets within this domain of personality functioning (Costa & McCrae, 1992). However, the Modifying and Accommodating scales correlated as highly with NEO–PI Neuroticism and Conscientiousness. The results provided in Table 5 also indicate the relative independence of the MIPS Individuating (self) and Nurturing (other) scales, with the MIPS Individuating (self) scale correlating negatively with NEO–PI Agreeableness and MIPS Nurturing (other) correlating positively with NEO–PI Extraversion. Of most importance to the validity of the theoretical model would be the correlations of the six (or three) MIPS polarity dimension scales with a measure of personality disorders. For example, the correlations of the MIPS Enhancing (pleasure), Preserving (pain), Modifying (active), Accommodating (passive), Individuating

382

WIDIGER

(self), and Nurturing (other) scales with the MCMI–III personality disorder scales would provide a direct and explicit test of the extent to which the schizoid personality disorder is in fact low on Enhancement and Preservation, low on Modification, high on Accommodation, average on Individuation, and low on Nurturance; whether, in contrast, the dependent personality disorder is average with respect to Enhancement and Preservation, low (or lower) on Individuation and high on Nurturance, but also low on Modification and high on Accommodation; and whether the borderline personality disorder is average along all six (or three) dimensions. The MIPS manual provides the correlations of the MIPS scales with a large number of personality instruments. However, no correlations are provided with any measure of personality disorders, including the MCMI–III. It is possible that other participants within this symposium will be aware of, or might provide, data concerning the correlation of the MIPS scales with the MCMI–III or with another measure of the personality disorders. There have been studies with various editions of the MCMI personality disorder scales that have purportedly concerned Millon’s (1969, 1981) theoretical model. However, to the extent that the MCMI personality disorder scales were written to assess the DSM perspective on the personality disorders rather than Millon’s perspective, these studies may not be particularly informative with respect to the validity of the model. Given the ambiguity of the representation of the theoretical model within the MCMI–III personality disorder scales, the relevance of a finding with an MCMI–III scale with respect to Millon’s (1981) theoretical model for that disorder is equally ambiguous. For example, Streiner and Miller (1988) obtained a significant correlation of .61 between the MCMI and the Morey, Waugh, and Blashfield (1985) Minnesota Multiphasic Personality Inventory (MMPI; Hathaway et al., 1989) histrionic personality disorder scales. This finding might be supportive of Millon’s (1969, 1981) theoretical model for histrionic personality disorder as an expression of an active–other–pleasure stage of ontogenetic or phylogenetic development. However, to the extent that the active, other, and pleasure components of the disorder are not represented within the MCMI Histrionic scale, the correlation of this scale with the MMPI Histrionic scale, or with any other measure, is ambiguous with respect to its relevance to the validity of the theoretical model. Factor analytic studies of the MCMI (or any other personality disorder instrument) could be used to assess the validity of the theoretical model, and many MCMI factor analytic studies have indeed been conducted (e.g., Choca, Retzlaff, Strack, Mouton, & Van Denburg, 1996; Craig & Bivens, 1998; Dyce, O’Connor, Parkins, & Janzen, 1997; Strack, 1991a; Strack, Lorr, & Campbell, 1990; Strack, Lorr, Campbell, & Lamnin, 1992). There are a number of methodological issues that are potentially problematic in this research, including the overlap among the scales, the range and representativeness of the population studied, whether to include the scales that assess the more severe variants of the 8 (or 10) basic personality patterns, whether to include the scales that assess Axis I mental disorders, and

MILLON’S DIMENSIONAL POLARITIES

383

whether to factor analyze the scales or the items (Craig & Weinberg, 1993; Dyce et al., 1997). However, a more fundamental point is that most of this research has not in fact been concerned with Millon’s theoretical model. “Relationships among the scales of a test should follow a pattern that corresponds to the structural features proposed by the instrument’s theory” (Millon, 1987, p. 128). The structural theory from which the personality disorder scales are derived consists of the two (or three) dimensions (or polarities) of affiliation (self–other, pleasure–pain, or both) and emotionality (active–passive). However, these two (three or six) dimensions are not considered or addressed in most MCMI factor analytic research. For example, Millon (1987) summarized the factor analytic research that had been conducted with the MCMI–I and indicated that this research “lends support to the stability of the theory’s structural model” (p. 137). However, no reference was actually made to the theory’s structural model in the discussion of the findings. For example, a four-factor solution was said to include (a) general maladjustment, (b) paranoid behavior and thinking, (c) schizoid behavioral detachment and thought versus sociability and self-confidence, and (d) social restraint and conformity versus aggression and rebellion. No reference was made to the self–other, pleasure–pain, or active–passive distinctions, or how these polarity dimensions related to the factors that were obtained. Choca et al. (1996) factor analyzed the items from each of the MCMI–II (Millon, 1983b) personality disorder scales to determine whether there were factors within each scale that corresponded to Millon’s theoretical model for each respective disorder and concluded that “generally the findings fit in well with Millon’s conceptualization of the personality” (p. 379). However, no mention was made in this study of the three polarity dimensions from which each of the personality disorders was theoretically derived. Choca et al. assessed instead the fit with Millon’s (1987; Millon & Davis, 1996) clinical descriptions of the personality disorders with respect to their behavioral presentation, interpersonal conduct, cognitive style, expressive mood, self-image, unconscious mechanisms, internalized content, and intrapsychic organization. For example, one factor (Introversion) defined the 35 MCMI–II Schizoid items, and three factors (Submissiveness, Agreeableness, and Respectful Sociability) defined the 37 MCMI–II Dependent items. Choca et al. did not discuss how the three Dependent factors related to the activity–passivity, pleasure–pain, or self–other polarity dimensions, or whether the one Introversion factor for the 37 Schizoid items was a confirmation of a passive–pain–detached polarity configuration for this personality disorder. Retzlaff and Gibertini (1987) may have been the first to assess empirically the structural validity of the theoretical model through a factor analysis of MCMI data. They indicated that no prior study had attempted to do so. “Although the MCMI was designed to represent this theoretical 4 × 2 matrix, no study has explored the degree of fit between the theory and the inventory” (p. 589). Retzlaff and Gibertini explored the factor structure of the eight basic personality scales with five different

384

WIDIGER

participant samples and with a correlation matrix derived solely from the extent of the overlap among the scales. Support was obtained for a three-factor solution, but this solution was inconsistent with the theoretical model. The three factors were identified by Retzlaff and Gibertini as Aloof–Social, Aggressive–Submissive, and Lability–Restraint. “No particular factor was found to represent locus of reinforcement” (p. 593). The failure to obtain a solution consistent with the theory is particularly remarkable for the analysis of the overlap among the scales, given that this overlap was purportedly governed by the theoretical model. The authors concluded that “the MCMI personality scales, though laudable in many respects, may not be appropriate measures of the constructs posited by Millon’s theory” (p. 593). The goal of a subsequent study by Strack et al. (1990) “was to evaluate the proposed circumplical model” (p. 354), consisting of the two dimensions of affiliation and emotionality, within samples of patients and students. A circumplical model was obtained when variance due to a possible response bias was removed, and one of the two dimensions did appear to be consistent with Millon’s (1987) emotionality dimension. “The horizontal axis of Aggressive (6B) and Narcissistic (5) versus Avoidant (2) and Schizotypal (S) [was] quite similar to Millon’s Expressive–Impassive dimension, especially given the position of the Histrionic (4) and Schizoid (1) scales” (Strack et al., 1990, p. 357). However, support was not obtained for affiliation (which includes both the self–other and pleasure–pain polarities). “In his model, Antisocial (6A) is contrasted with Dependent (3)” (pp. 357–358), whereas the findings contrasted the Borderline and Compulsive scales “and probably represents a dimension of Impulsivity versus Compulsivity” (p. 357). No reference was made in this study to the three polarity dimensions. Dyce et al. (1997) conducted a factor analysis of the MCMI–III personality disorder scales, and compared the results with previous factor analyses. They indicated that the correlation matrices obtained with the MCMI (Millon, 1977), the MCMI–II (Millon, 1983b), and the MCMI–III (Millon et al., 1994) were reasonably consistent over time and settings. “The structure of these personality scales has remained the same across the … recent versions of the test” (Dyce et al., 1997, p. 578). Dyce et al. (1997) concluded that “the overlapping scales are thus consistent with Millon’s evolving theory” (p. 578). However, this conclusion is somewhat surprising, as Dyce et al. also indicated that a four-factor solution was the most consistent and compelling structural representation of the MCMI–III personality disorder scales and that “there are few obvious parallels between Millon’s theory and the solutions” they obtained (Dyce et al., 1997, p. 580). They suggested that the four-factor solution was more consistent with the lexical five-factor model of personality (Costa & Widiger, 1994). The most thorough effort to empirically test Millon’s theoretical model was conducted by O’Connor and Dyce (1998). O’Connor and Dyce coded each of the personality disorders with respect to the model’s theoretical constructs to construct a target factor-loading matrix to depict the relative position of the personal-

MILLON’S DIMENSIONAL POLARITIES

385

ity disorders as predicted by the model. They then conducted 12 independent principal-axes common factor analyses on the correlation matrices among the personality disorders using a variety of samples and assessment instruments provided by nine previously published studies (two of which were from MCMI–II data, using both the overlapping and nonoverlapping scales). They extracted as many factors as were specified by the target model. The personality disorder matrices were then rotated to a least squares fit to the target matrices, using both orthogonal and oblique rotations. Congruence coefficients for the degree to which the rotated matrices conformed to the model-based target matrices were then computed. An initial difficulty experienced by O’Connor and Dyce (1998) was how to code the personality disorders from the perspective of Millon’s model, given the ambiguities already described. For example, “we used [Table 2] to construct a target matrix that represents the theory, and a number of peculiarities were immediately encountered” (O’Connor & Dyce, 1998, p. 6), including the ambiguities noted earlier regarding the number of dimensions that are involved and how to code the ambivalent patterns with respect to the self–other polarity. Their initial set of analyses addressed the description of the personality disorders using the three polarity dimensions. However, the target matrix that represented the personality disorders as described in Table 2 produced quite low congruence coefficients, none of which were statistically significant for any of the 12 correlation matrices. Even the correlations among the personality disorders provided by the MCMI–II as reported in the test’s manual were inconsistent with the theoretical model as depicted in Table 2. O’Connor and Dyce (1998) also tested the circumplex model, as depicted in Figure 1. Again, none of the congruence coefficients were significant. In contrast, Kiesler’s (1996) interpersonal circumplex model for the personality disorders obtained significant congruence coefficients for 4 of the 12 correlation matrices, 2 of which were the correlation matrices obtained by the MCMI–II. “The results of these tests indicate very little support for Millon’s circumplex structure” (O’Connor & Dyce, 1998, p. 8). O’Connor and Dyce also developed a six-factor target matrix that represented the most recent version of the model by Millon and Davis (1996), as depicted in Table 4. “The fit coefficients were [again] low and none were significant” (O’Connor & Dyce, 1998, p. 7). O’Connor and Dyce (1998) tried other possible solutions. “The target pattern that produced the best fit coefficients that was also at least roughly consistent with the model is a seven-factor matrix” (p. 6). This version of the model did obtain significant coefficients for 6 of the 12 correlation matrices provided in the studies by Hyler and Lyons (1988), Kass, Skodol, Charles, Spitzer, and Williams (1985), Morey et al. (1985), Zimmerman and Coryell (1989), and, perhaps most important, the 2 correlation matrices provided in the MCMI–II test manual (Millon, 1987). However, it still failed to obtain significant congruency coefficients for 6 correlation matrices provided by the studies of Klein et al. (1993), Livesley and Jackson

386

WIDIGER

(1986), Moldin, Rice, Erlenmeyer-Kimling, and Squires-Wheeler (1994), and Widiger, Trull, Hurt, Clarkin, and Frances (1987). In contrast, the theoretical models of Torgersen and Alnaes (1989), Widiger, Trull, Clarkin, Sanderson, and Costa (1994), and Cloninger and Svrakic (1994) obtained significant and substantial congruence coefficients for all 12 correlation matrices. In addition, O’Connor and Dyce (1998) indicated that the seven-factor model they derived “stretches the logic underlying Millon’s theory” (p. 6).

CONCLUSIONS Millon’s theoretical model for the DSM–IV personality disorders is tremendously rich and grand in its coverage. The theory is in many respects quite compelling, thorough, and detailed. Millon’s development of this theory is a tremendous service to the field that warrants considerable attention from other theorists and researchers. However, this review also raises some concerns regarding how the personality patterns are logically derived from the three active–passive, self–other, and pleasure–pain polarity dimensions, or from the two circumplical affiliation and emotionality dimensions. These constructs are probably quite important to our understanding of the DSM–IV personality disorders, with respect to both their etiology and their description (Millon & Davis, 1996). The affiliation, emotionality, self–other, active–passive, and pleasure–pain constructs are evident within a number of alternative models for the personality disorders (e.g., Benjamin, 1993; Costa & Widiger, 1994; Kiesler, 1996; Tellegen & Waller, in press). However, the particular organization of these constructs within Millon’s theoretical model might be somewhat problematic, at least with respect to their relation to the personality disorders. ACKNOWLEDGMENTS I thank David Berry for his comments concerning an earlier version of this article.

REFERENCES American Psychiatric Association. (1980). Diagnostic and statistical manual of mental disorders (3rd ed.). Washington, DC: Author. American Psychiatric Association. (1987). Diagnostic and statistical manual of mental disorders (3rd ed., rev.). Washington, DC: Author. American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author.

MILLON’S DIMENSIONAL POLARITIES

387

Benjamin, L. S. (1993). Interpersonal diagnosis and treatment of personality disorders. New York: Guilford. Cattell, R. B., Eber, H. W., & Tatsuoka, M. M. (1970). Handbook for the Sixteen Personality Factor Questionnaire (16 PF). Champaign, IL: Institute for Personality and Ability Testing. Choca, J., Retzlaff, P., Strack, S., Mouton, A., & Van Denburg, E. (1996). Factorial elements in Millon’s personality theory. Journal of Personality Disorders, 10, 377–383. Cloninger, C. R., & Svrakic, D. M. (1994). Differentiating normal and deviant personality by the seven-factor personality model. In S. Strack & M. Lorr (Eds.), Differentiating normal and abnormal personality (pp. 40–64). New York: Springer. Costa, P. T., & McCrae, R. R. (1985). The NEO Personality Inventory manual. Odessa, FL: Psychological Assessment Resources. Costa, P. T., & McCrae, R. R. (1992). Revised NEO Personality Inventory (NEO–PI–R) and NEO Five-Factor Inventory (NEO–FFI) professional manual. Odessa, FL: Psychological Assessment Resources. Costa, P. T., & Widiger, T. A. (Eds.). (1994). Personality disorders and the five-factor model of personality. Washington, DC: American Psychological Association. Craig, R. J., & Bivens, A. (1998). Factor structure of the MCMI–III. Journal of Personality Assessment, 70, 190–196. Craig, R. J., & Weinberg, D. (1993). MCMI: Review of the literature. In R. J. Craig (Ed.), The Millon Clinical Multiaxial Inventory: A clinical research information synthesis (pp. 23–70). Hillsdale, NJ: Lawrence Erlbaum Associates. Dyce, J. A., O’Connor, B. P., Parkins, S. Y., & Janzen, H. L. (1997). Correlational structure of the MCMI–III personality disorder scales and comparisons with other data sets. Journal of Personality Assessment, 69, 568–582. Frances, A. J. (1980). The DSM–III personality disorders section: A commentary. American Journal of Psychiatry, 137, 1050–1054. Freud, S. (1953). The instincts and their vicissitudes. In J. Strachey (Ed. and Trans.), Standard edition (Vol. 14, pp. 109–114). London: Hogarth. (Original work published 1915) Gough, H. G. (1987). California Psychological Inventory: Administrator’s guide. Palo Alto, CA: Consulting Psychologists Press. Gunderson, J. G. (1983). DSM–III diagnoses of personality disorders. In J. Frosch (Ed.), Current perspectives on personality disorders (pp. 20–39). Washington, DC: American Psychiatric Press. Gunderson, J. G. (1998). DSM–IV personality disorders: Final overview. In T. A. Widiger, A. J. Frances, H. A. Pincus, R. Ross, M. B. First, W. Davis, & M. Kline (Eds.), DSM–IV sourcebook (Vol. 4, pp. 1123–1140). Washington, DC: American Psychiatric Association. Hathaway, S. R., McKinley, J. C., Butcher, J. N., Dahlstrom, W. G., Graham, J. R., & Tellegen, A. (1989). Minnesota Multiphasis Personality Inventory test booklet. Minneapolis, MN: Regents of the University of Minnesota. Hyler, S. E., & Lyons, M. (1988). Factor analysis of the DSM–III personality disorder clusters: A replication. Comprehensive Psychiatry, 29, 304–308. Jackson, D. N. (1985). Personality Research Form manual (3rd ed.). Port Huron, MI: Research Psychologists Press. Kass, F., Skodol, A. E., Charles, E., Spitzer, R. L., & Williams, J. B. W. (1985). Scaled ratings of DSM–III personality disorders. American Journal of Psychiatry, 142, 627–630. Kiesler, D. J. (1996). Contemporary interpersonal theory and research: Personality, psychopathology, and psychotherapy. New York: Wiley. Klein, M. H., Benjamin, L. S., Rosenfeld, R., Treece, C., Husted, J., & Greist, J. H. (1993). The Wisconsin Personality Disorders Inventory: Development, reliability, and validity. Journal of Personality Disorders, 7, 285–303.

388

WIDIGER

Livesley, W. J. (1998). Suggestions for a framework for an empirically based classification of personality disorder. Canadian Journal of Psychiatry, 43, 137–147. Livesley, W. J., & Jackson, D. N. (1986). The internal consistency and factorial structure of behaviors judged to be associated with DSM–III personality disorders. American Journal of Psychiatry, 143, 1473–1474. Millon, T. (1969). Modern psychopathology: A biosocial approach to maladaptive learning and functioning. Prospect Heights, IL: Waveland. Millon, T. (1977). Millon Multiaxial Clinical Inventory manual. Minneapolis, MN: National Computer Systems. Millon, T. (1981). Disorders of personality. DSM–III: Axis II. New York: Wiley. Millon, T. (1983a). The DSM–III: An insider’s perspective. American Psychologist, 38, 804–814. Millon, T. (1983b). Millon Clinical Multiaxial Inventory manual (3rd ed.). Minneapolis, MN: National Computer Systems. Millon, T. (1987). Manual for the MCMI–II (2nd ed.). Minneapolis, MN: National Computer Systems. Millon, T. (1990). Toward a new personology: An evolutionary model. New York: Wiley. Millon, T. (1993). Negativistic (passive–aggressive) personality disorder. Journal of Personality Disorders, 7, 78–85. Millon, T. (1994). MIPS: Millon Index of Personality Styles manual. San Antonio, TX: Psychological Corporation. Millon, T., & Davis, R. D. (1994). Millon’s evolutionary model of normal and abnormal personality: Theory and measures. In S. Strack & M. Lorr (Eds.), Differentiating normal and abnormal personality (pp. 79–113). New York: Springer. Millon, T., & Davis, R. D. (1995). The development of personality disorders. In D. Cicchetti & D. J. Cohen (Eds.), Developmental psychopathology (Vol. 2, pp. 633–676). New York: Wiley. Millon, T., & Davis, R. D. (1996). Disorders of personality. DSM–IV and beyond. New York: Wiley. Millon, T., Green, C. J., & Meagher, R. B. (1982). Millon Behavior Health Inventory manual (3rd ed.). Minneapolis, MN: National Computer Systems. Millon, T., Millon, C., & Davis, R. D. (1994). MCMI–III manual. Minneapolis, MN: National Computer Systems. Moldin, S. O., Rice, J. P., Erlenmeyer-Kimling, L., & Squires-Wheeler, E. (1994). Latent structure of DSM–III–R Axis II psychopathology in a normal sample. Journal of Abnormal Psychology, 103, 259–266. Morey, L. C., Waugh, M. H., & Blashfield, R. K. (1985). MMPI scales for DSM–III personality disorders: Their derivation and correlates. Journal of Personality Assessment, 49, 245–251. Myers, I. B., & McCaulley, M. H. (1985). Manual: A guide to the development and use of the Myers– Briggs Type Indicator. Palo Alto, CA: Consulting Psychologists Press. O’Connor, B. P., & Dyce, J. A. (1998). A test of models of personality disorder configuration. Journal of Abnormal Psychology, 107, 3–16. Retzlaff, P. D., & Gibertini, M. (1987). Factor structure of the MCMI basic personality scales and common-item artifact. Journal of Personality Assessment, 51, 588–594. Siever, L. J., & Davis, K. L. (1991). A psychobiological perspective on the personality disorders. American Journal of Psychiatry, 148, 1647–1658. Strack, S. (1987). Development and validation of an adjective check list to assess the Millon personality types in a normal population. Journal of Personality Assessment, 51, 572–587. Strack, S. (1991a). Factor analysis of MCMI–II and PACL basic personality scales in a college sample. Journal of Personality Assessment, 57, 345–355. Strack, S. (1991b). Manual for the Personality Adjective Check List (PACL) (rev. ed.). South Pasadena, CA: 21st Century Assessment. Strack, S., Lorr, M., & Campbell, L. (1990). An evaluation of Millon’s circular model of personality disorders. Journal of Personality Disorders, 4, 353–361.

MILLON’S DIMENSIONAL POLARITIES

389

Strack, S., Lorr, M., Campbell, L., & Lamnin, A. (1992). Personality disorder and clinical syndrome factors of MCMI–II scales. Journal of Personality Disorders, 6, 40–52. Streiner, D. L., & Miller, H. R. (1988). Validity of the MMPI scales for DSM–III personality disorders: What are they measuring? Journal of Personality Disorders, 2, 238–242. Tellegen, A., & Waller, N. G. (in press). Exploring personality through test construction: Development of the Multidimensional Personality Questionnaire. In S. R. Briggs & J. M. Cheek (Eds.), Personality measures: Development and evaluation (Vol. 1). Greenwich, CT: JAI. Torgersen, S., & Alnaes, R. (1989). Localizing DSM–III personality disorders in three-dimensional structural space. Journal of Personality Disorders, 3, 274–281. Watkins, C. E., Campbell, V. L., Nieberding, R., & Hallmark, R. (1995). Contemporary practice of psychological assessment by clinical psychologists. Professional Psychology: Research and Practice, 26, 54–60. Widiger, T. A., & Corbitt, E. M. (1993). The MCMI–II personality disorder scales and their relationship to DSM–III–R diagnosis. In R. J. Craig (Ed.), The Millon Clinical Multiaxial Inventory: A clinical research information synthesis (pp. 181–202). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Widiger, T. A., Frances, A. J., Spitzer, R. L., & Williams, J. B. W. (1988). The DSM–III–R personality disorders: An overview. American Journal of Psychiatry, 145, 786–795. Widiger, T. A., Mangine, S., Corbitt, E. M., Ellis, C. G., & Thomas, G. V. (1995). Personality Disorder Interview–IV: A semistructured interview for the assessment of personality disorders. Odessa, FL: Psychological Assessment Resources. Widiger, T. A., Trull, T. J., Clarkin, J. F., Sanderson, C., & Costa, P. T. (1994). A description of the DSM–III–R and DSM–IV personality disorders with the five-factor model of personality. In P. T. Costa & T. A. Widiger (Eds.), Personality disorders and the five-factor model of personality (pp. 41–56). Washington, DC: American Psychological Association. Widiger, T. A,. Trull, T. J., Hurt, S. W., Clarkin, J. F., & Frances, A. J. (1987). A multidimensional scaling of the DSM–III personality disorders. Archives of General Psychiatry, 44, 557–563. Widiger, T. A., Williams, J. B. W., Spitzer, R. L., & Frances, A. J. (1985). The MCMI as a measure of DSM–III. Journal of Personality Assessment, 49, 366–378. Zimmerman, M., & Coryell, W. (1989). DSM–III personality disorder diagnoses in a nonpatient sample. Archives of General Psychiatry, 46, 682–689.

Thomas A. Widiger Department of Psychology University of Kentucky 115 Kastle Hall Lexington, KY 40506–0044 Received January 18, 1999