پیش بینی اختلالات شخصیت MCMI-III با مدل شخصیت زاکرمن
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|38472||2007||11 صفحه PDF||سفارش دهید||محاسبه نشده|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Personality and Individual Differences, Volume 42, Issue 7, May 2007, Pages 1311–1321
Abstract The current study explores the relationships between the basic dimensions of the Zuckerman personality model (measured through the ZKPQ) and the 14 personality disorder scales of the millon clinical multiaxial inventory third edition (MCMI-III). The total sample comprised 673 subjects, of whom 50% were university students and 50% were subjects from the general population. Statistical analysis followed a similar methodology used by Dyce and O’Connor (1998) and O’Connor (2005). The principal component analysis showed that the five ZKPQ dimensions were associated with the personality disorder scales in different patterns. Linear regression analyses indicated that ZKPQ scales were very good predictors of the three DSM personality disorder clusters, and in particular of cluster B. Also, the global prediction power of Zuckerman’s dimensions was highly similar to that demonstrated for the NEO-PI-R in previous studies. LOESS graphical analysis showed additional information about nonlinear relationships between normal personality and personality disorders. The discussion focuses on the strengths and weaknesses of Zuckerman’s approach to account for personality disorders
Introduction DSM Axis-II personality disorders are defined as “rigid and maladaptive traits” (DSM-IV-VR; American Psychiatric Association, 2000). Following this definition, these disorders may be understood as extreme and maladaptive variants of normal personality traits (Widiger & Costa, 1994). In fact, personality dimensional models have always emphasized relationships between normal traits and personality disorders (PDs). Within this tradition, the five-factor model (FFM) began to look prominent as an alternative to the categorical model in the classification of PDs (Costa & Widiger, 2002). Two meta-analyses developed by Ostendorf, 2000 and Saulsman and Page, 2004 have supported relationships between personality disorders and the FFM, with special attention to the studies using the NEO-PI-R (Costa & McCrae, 1992). Personality disorders were assessed through different procedures: structured interview based on DSM criteria, scales derived from the Minnesota multiphasic personality inventory (MMPI), or different inventories like the millon clinical multiaxial inventory (MCMI). The results indicated that: (1) disorders defined by emotional distress such as Paranoid, Schizotypal, Borderline, Avoidant and Dependent are positively associated with Neuroticism, (2) disorders linked with high gregariousness like Histrionic and Narcissistic correlate positively with Extraversion, (3) disorders characterized by shyness and reclusive qualities such as Schizoid, Schizotypal, and Avoidant are negatively related to Extraversion, (4) disorders implying interpersonal difficulties (Paranoid, Schizotypal, Antisocial, Borderline and Narcissistic) correlate negatively with Agreeableness, (5) conscientiousness shows positive associations with disorders characterized by orderliness such as Obsessive–Compulsive, and negative associations with recklessness disorders like Antisocial and Borderline, and (6) the Openness dimension is associated with no personality disorders (Saulsman & Page, 2004). In general, it is demonstrated that the broader NEO-PI-R scales (i.e. dimensions) account for around 34% of the variance in personality disorders, this percentage being larger when the narrower NEO-PI-R subscales (i.e. facets) are used as independent variables (Dyce & O’Connor, 1998). In contrast with the research effort in the area of the relations between personality disorders and the FFM or other normal personality models such as Eysenck‘s PEN (Jang, Livesley, & Vernon, 1999) and Cloninger’s Temperament and Character (Svrakic, Whitehead, Przybeck, & Cloninger, 1993), Zuckerman’s Alternative Five-Factor Model has attracted little attention with regard to this topic. Zuckerman developed his model as an alternative to the FFM (Zuckerman et al., 1993 and Zuckerman et al., 1991), including five basic personality dimensions: impulsive-unsocialized sensation seeking (ImpSS), neuroticism–anxiety (N-Anx), aggressivity–hostility (Agg-Hos), activity (Act) and sociability (Sy). Impulsive-unsocialized sensation seeking has been the most studied dimension in relation to the question of personality disorders. In fact, the Antisocial Personality Disorder is supposed to be the extreme pathological end of this trait dimension (Zuckerman, 1999). Also, some research has related this dimension to other cluster B personality disorders due to the shared broad characteristic of impulsiveness of these disorders. Thus, for instance, Ball, Carroll, and Rounsaville (1994) found that sensation seeking correlated with antisocial personality and lifetime drug abuse and Thornquist and Zuckerman (1995) reported significant correlations between ImpSS and Agg-Host with the Total score of the Psychopathy Check List (Hare, 1991). Also, patients diagnosed with borderline personality score higher on this dimension than patients with “non-cluster B personality disorders” and normal controls (Reist, Haier, DeMet, & Chicz-DeMet, 1990). Also, However, this dimension has not been widely studied in personality disorders other than antisocial personality, and the other four dimensions defined in the model have hardly been investigated at all (Zuckerman, 1996). One noteworthy exception is the study by Wang, Du, Wang, Livesley, and Jang (2004), in which Chinese versions of the ZKPQ and the Dimensional Assessment of Personality Pathology—Basic Questionnaire (DAPP-BQ) were applied to a sample of 149 healthy university subjects. The results support the associations between N-Anx and 11 of 18 DAPP-BQ scales. ImpSS was related to both Dissocial and Impulsive Misconduct factors, with Agg-Hos being highly related to the former factor. Sy loaded negatively on the “inhibition” factor, and Act was mainly related to one scale only (Compulsivity). However, this study presents two limitations, which makes further research necessary: (1) the size of the sample, and (2) the use as a dependent variable of a questionnaire which relates to, but does not directly address, the measuring of all personality disorders described on the DSM-IV Axis II. Also, Pearson correlations and factor analyses are the only methods to have been used in most studies about associations between personality traits and personality disorders, irrespective of the theoretical normal personality model. However, correlations may be misleading if these associations are not linear. Note that measures of PDs were developed with an almost exclusive focus on the high end of the PD continua/um. High-scoring individuals are thus likely to have personality disorders, but the characteristics of low-scoring individuals remain unknown. These individuals are characterized only by the absence, or by diminished amounts, of the personality disorder. If low scores on measures of personality disorders represent normality and psychological health, then these associations, if they exist, may not be linear. For instance, a possible pattern would be to begin departing from average scores on traits only when scores on personality disorders are relatively high. Recently, O’Connor (2005) outlined a procedure to detect graphically whether the relations between the FFM and PDs are linear or not. They reported that, most of the time, low scores on particular PDs were accompanied by normal-range scores on FFM variables, so low PD scores do not generally represent some kind of personality deviance in the other direction. In spite of this promising approach, to our knowledge, it has never been applied to another personality model such as Eysenck’s, Cloninger’s, or Zuckerman’s. In general, the relationships between Zuckerman’s model and personality disorders are still largely unknown. This personality model is of particular interest, however, providing as it does an alternative five-factor structure that includes, among others, a dimension of Impulsive sensation seeking that is of particular clinical significance. The current study seeks therefore to fill this gap by exploring the relationships between Zuckerman’s personality model and MCMI-III personality disorder scales. A similar procedure to Dyce and O’Connor (1998) was developed. Also, a graphic analysis was conducted to detect possible nonlinear relationships between Zuckerman’s personality traits model and personality disorders. To do this, the LOESS, non-parametric, local area, polynomial regression procedure was used following O’Connor’s (2005) specifications.
نتیجه گیری انگلیسی
. Results 3.1. Descriptive, distribution statistics and reliability Table 1 shows descriptives and reliabilities for ZKPQ and MCMI-III scales and the three DSM-IV personality disorders clusters formed after the corresponding scales of the MCMI-III. Note that the kurtosis and skewness report a normal distribution for seven scales (values between −1 and +1) but not for the rest of the scales. Alpha reliabilities oscillated between 0.64 and 0.79 for Millon’s scales, between .72 and .82 for Clusters, and between 0.71 and 0.85 for the ZKPQ scales. Table 1. Descriptives, skewness (S), kurtosis (K) and alpha (α) of MCMI-III scales and ZKPQ dimensions a MCMI-III scales Min Max M SD S K α Schizoid 0 21 4.63 3.74 1.26 1.71 .68 Avoidant 0 22 4.05 4.11 1.44 2.06 .77 Dependent 0 23 6.11 4.55 1.04 .71 .76 Histrionic 0 24 16.29 4.90 −.43 −.39 .74 Narcissistic 3 28 13.76 3.98 .35 .35 .64 Antisocial 0 21 5.54 3.89 .88 .69 .68 Sadistic 0 22 6.04 4.66 .91 .39 .74 Obsessive–Compulsive 2 25 14.83 4.81 −.25 −.40 .65 Passive–Aggressive 0 25 7.51 4.83 .74 .16 .76 Masochistic 0 22 2.68 3.30 1.76 3.42 .77 Schizotypal 0 21 2.98 3.70 1.82 3.89 .76 Borderline 0 24 4.96 4.44 1.16 1.21 .76 Paranoid 0 24 4.45 4.57 1.33 1.67 .79 Depressive 0 23 4.08 4.72 1.49 2.00 .76 Cluster A 0 66 12.06 10.25 1.53 2.84 .82 Cluster B 11 87 40.53 11.49 .34 .13 .72 Cluster C 6 62 25.04 9.67 .81 .51 .72 ZKPQ–ImpSS 0 19 8.55 4.38 .14 −.84 .82 ZKPQ–N/Anx 0 19 7.52 4.54 .47 −.69 .85 ZKPQ–Agg/Hos 0 17 7.31 3.45 .14 −.46 .73 ZKPQ–Act 0 17 7.91 3.34 .09 −.59 .71 ZKPQ–Sy 0 17 9.09 3.57 −.25 −.42 .78 a ImpSS: impulsive sensations seeking; N/Anx: neuroticism–anxiety; Agg-Hos: aggressivity–hostility; Act: activity; Sy: sociability. Cluster A: Schizotypal + Schizoid + Paranoid. Cluster B: Histrionic + Narcissistic + Antisocial + Borderline. Cluster C: Avoidant + Dependent + Compulsive. Table options 3.2. Principal component analyses Five factors were extracted following a Varimax Principal Components analysis including the 5 ZKPQ scales and the 14 Millon scales following the same procedure as in Dyce and O’Connor (1998). The Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy was 0.90, and Bartlett’s test of sphericity (BTS) yielded approx. χ2 = 8689, 676; df: 171 (p < 0.001). Note that both KMO and Bartlett’s test of sphericity indicate that factor analysis is appropriate. As can be observed in Table 2, every ZKPQ dimension loads on a different factor. The first one is defined by N-Anx and 10 of the 14 MCMI-III scales (excluding Histrionic, Narcissistic, Antisocial and Obsessive–Compulsive disorders). The second one is defined by Imp-SS and Antisocial, Narcissistic, Sadistic, Borderline and Obsessive–Compulsive (the latter in negative) disorders. Sy loaded on the third one with Histrionic and Avoidant (the latter in negative) scales. The fourth one was formed by Agg-Hos and the Sadistic disorder (and Passive–Aggressive and Narcissistic with lower loadings). Finally, Act, Narcissistic, Bordeline, Obsessive–Compulsive and Histrionic scales loaded on the fifth factor. Table 2. Principal component analysis with Varimax rotation including MCMI-III scales and ZKPQ dimensions MCMI-III scales I II III IV V Schizoid .46 .03 −.75 .00 .11 Avoidant .77 −.09 −.40 −.07 −.12 Dependent .86 −.02 .07 −.18 −.04 Histrionic −.23 .23 .78 .10 .30 Narcissistic −.27 .44 .17 .32 .61 Antisocial .27 .83 −.01 .23 .12 Sadistic .49 .43 −.10 .50 .29 Obsessive–Compulsive .03 −.80 −.14 −.09 .37 Passive–Aggressive .77 .25 −.08 .35 .05 Masochistic .83 .13 −.20 .01 −.01 Schizotypal .79 .20 −.24 .06 .17 Borderline .78 .41 .01 .18 .01 Paranoid .67 .11 −.29 .25 .39 Depressive .87 −.02 −.13 .10 −.06 ZKPQ–ImpSS .11 .73 .30 .07 .14 ZKPQ–N/Anx .70 −.16 .10 .33 −.34 ZKPQ–Agg/Hos .12 .24 .07 .87 .06 ZKPQ–Act .05 −.13 .23 .01 .67 ZKPQ–Sy .01 .20 .82 −.01 .15 % Accounted variance 32.9 14.5 12.9 8.3 8.1 (a) ImpSS: impulsive-unsocialized sensations seeking; N/Anx: neuroticism–anxiety; Agg-Hos: aggressivity–hostility; Act: activity Sy: sociability. (b) Loadings >.40 are in boldface. Table options 3.3. Linear and logistic regression analyses Following Dyce and O’Connor (1998) procedure, several regression analyses (stepwise method at a 0.0001 necessary significance level to be included in the equation) were conducted taking ZKPQ dimensions as independent variables and every MCMI-III scale and the three clusters successively as dependent ones. Moreover, two extreme groups were formed. The criterion for selection was to obtain a z score equal to or higher than +1 SD for the high group, or equal to or lower than −1 SD for the low group. Further, a logistic regression was conducted to explore which ZKPQ dimensions contribute significantly to the classification between extreme groups. Regression results are shown in Table 3. Percentage of variance ranged between 19% and 52%. N-Anx was presented in most of the equations, except for Antisocial and Obsessive–Compulsive disorders. Cluster A disorders are mainly characterized by social shyness and an extreme distrust and suspicion of others (Schizoid, Schizotypal, and Paranoia), although other dimensions add specificity, such as ImpSS and Agg-Hos to the Schizotypal disorder. Regarding Cluster B, ImpSS and Agg-Hos play the most important role, although other scales are specifically related to some disorders of this cluster; for instance, N-Anx to Bordeline or Sy to Histrionic. Cluster C is almost entirely predicted by N-Anx, although low levels of ImpSS were relevant as well. With minor exceptions, logistic regression equations are formed by the same variables and identical sign. Percentages of right classification were between 76% (Obsessive–Compulsive) and 92% (Antisocial). Percentages were 85%, 97%, and 85% for clusters A, B, and C, respectively. Note ImpSS, Agg-Hos, and Sy dimensions predicted almost perfectly the group classification for cluster B. Table 3. Linear multiple and logistic regression results analyzing the contribution of ZKPQ dimensions to the prediction of MCMI-III scores and extreme groups (1 ± SD)a MCM-III scale Linear multiple regression Logistic regression R R2 Significant dimensions at last step %b Schizoid .58 .35 Sy-, N-Anx+ Sy- 87 Avoidant .57 .32 N-Anx+, Sy- Sy-, N-Anx+ 87 Dependent .54 .30 N-Anx+ N-Anx+, 83 Histrionic .69 .48 Sy+, ImpSS+, N-Anx- Sy+, ImpSS+ 89 Narcissistic .61 .37 ImpSS+, N-Anx-, Agg-Hos+, Act+ Agg-Hos+, N-Anx-, ImpSS+ 88 Antisocial .63 .40 ImpSS+, Agg-Hos+ ImpSS+, Agg-Hos+ 92 Sadistic .61 .37 Agg-Hos+, N-Anx+, ImpSS+ ImpSS+, Agg-Hos+ 85 Obsessive–Compulsive .54 .29 ImpSS-, Act+ ImpSS- 76 Passive–Aggressive .64 .41 N-Anx+, Agg-Hos+ N-Anx+, Agg-Hos+ 89 Masochistic .55 .30 N-Anx+ N-Anx+ 84 Schizotypal .54 .29 N-Anx+, ImpSS+, Sy- N-Anx+, ImpSS+, 85 Borderline .63 .40 N-Anx+, ImpSS+, Agg-Hos+ N-Anx+, Agg-Hos+, ImpSS+ 88 Paranoid .44 .19 N-Anx+, Agg-Hos+ N-Anx+, Agg-Hos+ 77 Depressive .65 .52 N-Anx+ N-Anx+ 91 Cluster A .53 .27 N-Anx+, Sy-, Agg-Hos+ N-Anx+, Agg-Hos, Sy- 85 Cluster B .72 .52 ImpSS+, Agg-Hos+, Sy+ ImpSS+, Agg-Hos+, Sy+ 97 Cluster C .52 .27 N-Anx+, ImpSS- ImpSS-, N-Anx+ 85 a ImpSS: impulsive-unsocialized sensation seeking; N/Anx: neuroticism–anxiety; Agg-Hos: aggression–hostility; Act: activity; Sy: sociability. Cluster A: Schizotypal + Schizoid + Paranoid. Cluster B: Histrionic + Narcissistic + Antisocial + Borderline. Cluster C: Avoidant + Dependent + Compulsive. b Percentage of right classification. Table options 3.4. Graphical analysis The graphical analysis procedure described above (O’Connor, 2005) was conducted to analyse possible nonlinear relationships between ZKPQ scales and the three DSM personality disorders clusters. Previously, ZKPQ raw scores were transformed to z scores, and scores on the three clusters were converted to T scores using the average and standard deviation of the sample itself. With regard to the graphical results ( Fig. 1), subjects at the high end of cluster A presented a profile of high scores on N-Anx (and also on Agg-Hos) and low ones on Sy. For Cluster B ( Fig. 1), Imp-SS, Agg-Hos, and Sy were quite high. Also, Act and N-Anx presented a tendency to be above the average. Finally, for Cluster C ( Fig. 1), graphical analysis highlighted the deviance of N-Anx for the high end of the continuum, with the remaining dimensions showing negative scores, especially Sy. LOESS plots for DSM personality disorders clusters (T scores) after ZKPQ ... Fig. 1. LOESS plots for DSM personality disorders clusters (T scores) after ZKPQ dimensions (z scores).