الگوهای خلق و خوی و تغییرات مرتبط با آسیب شناسی اختلال شخصیت

Abstract This study examined mood and mood variation in relation to varying forms and degrees of personality disorder (PD) pathology. Mood experiences of 98 psychotropic medication-free individuals were repeatedly assessed over a 4-day period. Persons with PDs (n = 57) generally displayed neutral to moderately positive moods; however, overall mood valence was less positive when compared to those without PDs (n = 41). Mood ratings demonstrated moderate covariations with anxious-fearful (A-F) PD traits but little or no association with erratic-emotional-dramatic (E-D) and odd-eccentric (O-E) PD traits once common variance among PD dimensions was removed. For PD diagnostic categories, the presence of avoidant and/or depressive PDs was most strongly associated with negative mood. When dimensional scores based on specific PD trait features were considered, avoidant, depressive, borderline, passive-aggressive, obsessive-compulsive, dependent, paranoid, and schizoid PD traits demonstrated the most reliable associations with negative mood. Apart from borderline PD features, traits associated with other E-D cluster PDs displayed little or no associations with mood quality. Consistent with previous research, mood variability emerged as an internally consistent and stable individual difference variable. Mood variability, however, was not generally associated with PD diagnostic categories or traits. Implications of this study’s findings are considered in relation to the conceptual modeling of PDs. ALTHOUGH THE personality disorders (PDs) of the Diagnostic and Statistical Manual of Mental Disorders (DSM) have been the object of numerous studies in recent years, 1 some areas of inquiry within the field have not been well-researched, such as the association of PD pathology with mood. This is particularly striking given that PDs have commonly been associated with depressive 2 and 3 and anxiety 4 disorders, and that anger, loneliness, and emptiness are commonly reported experiences among some individuals with PDs. 5 Relatedly, the facilitation of emotion tolerance and regulation skills is emphasized in many therapies for PDs. 6, 7 and 8 Consequently, emotions and moods are of central importance in the conceptualization and treatment of PD pathology, even though the nature of their relationships has not been well-researched. To address this relative gap in knowledge, the present research investigated moods and mood variability associated with PD pathology. The experience of neuroticism or negative mood is common among those with various forms of psychopathology,9 and 10 including many of the PDs.11, 12 and 13 Positive moods, however, are more frequently experienced than negative moods in nonpathological samples,14, 15 and 16 particularly among those who are extraverted.17 and 18 Whereas neuroticism or negative mood appears to be a central feature of many of the PDs, the broad personality dimension of introversion-extraversion discriminates some PDs from others. Anxious-fearful (A-F; avoidant, dependent, obsessive-compulsive) and odd-eccentric PDs (O-E; paranoid, schizoid, schizotypal) tend to be associated with introversive tendencies, while erratic-emotional-dramatic PDs (E-D; histrionic, narcissistic, borderline, antisocial) have been associated with extraversive tendencies.12, 13 and 19 Introversion and extraversion have also been associated with low and high levels of positive affect, respectively.17 The personality dimensions of trait anxiety and impulsivity also discriminate among some classes of PD pathology, most notably A-F from E-D PDs.12 and 20 Although trait anxiety has consistently been associated with negative mood,17 impulsivity has been reported to be unrelated to positive mood18, 21 and 22 and to have unreliable or modest associations with negative mood.21 and 22 Research has further suggested that extraversion’s association with positive mood is largely a product of the sociability, not impulsivity, component of extraversion.2 and 22 Research on mood variability has suggested that within- or intraperson mood variability is a temporally stable and cross-situationally consistent individual difference variable23, 24 and 25 that is unreliably associated with major personality characteristics such as neuroticism and extraversion.23, 26, 27, 28 and 29 Observations such as these have led some to conclude that mood variability is a unique personality trait.23 and 29 As such, mood variability may reflect the effects of dispositional rather than situational variability on moods. With regard to PDs, some have suggested that persons with E-D PDs, in particular, have difficulty regulating emotions or tend to be more emotionally reactive.5, 7, 30 and 31 However, limited research on the association of mood variability and PDs has yielded inconsistent findings. Two studies have found borderline and/or schizotypal traits to display modest to moderate associations with mood variability as assessed by self-report or rating measures completed at 12- to 24-hour intervals on multiple occasions.29 and 32 Similarly, persons with borderline PD have been observed to demonstrate similar degrees of self-reported mood lability but higher levels of self-reported affective intensity when compared to bipolar II patients.33 Self-reports of mood instability have also been observed to be greater among persons with borderline PD when compared to those with other PDs; however, self-reports of affective intensity reportedly did not differ between borderline and nonborderline PD patients.34 In these latter studies, retrospective self-reports of mood variability and intensity were based on self-report questionnaires administered on a single occasion. Retrospective self-reports of experiences such as mood may be adversely influenced by memory biases and response sets such as social undesirability, perhaps more so than indicators based on samples of immediate experience taken in an ongoing fashion in naturalistic environments.35 and 36 The primary purpose of this research was to evaluate the degree and nature of associations among mood, mood variability, and PD pathology. The experience sampling method (ESM) was employed to assess mood and mood variability because this approach is particularly appropriate for the naturalistic assessment of experiences with minimal intrusion.37, 38 and 39 Because individuals’ experiences are sampled in environmental contexts where they naturally occur, this assessment approach maximizes ecological validity and is suited to the description of patterns of daily experience.37 and 38 This method is also less influenced by response sets or biases commonly associated with retrospective questionnaire assessments,36 and has been demonstrated to be a reliable and valid approach for the assessment of mood, cognition, personality constructs, and behavior.37, 39 and 40 Although the ESM method has been used to assess the experiences of a number of clinical groups,41, 42 and 43 it has not been systematically applied to the study of experiences associated with PD pathology. In the present investigation, both categorical (i.e., PD diagnosis present v absent) and dimensional representations of PD pathology were considered. In the case of PD dimensions, the degree of symptom presence was determined for each participant for each of 12 PD diagnostic categories and each of three DSM-IV 44 symptomatological PD clusters (i.e., the O-E, E-D, and A-F clusters). Multivariate research has provided moderate empirical support for this clustering scheme. 19, 45, 46, 47 and 48 Two primary hypotheses were examined in this research. First, based on previous observations,13 persons with PDs were expected to demonstrate less favorable moods than those without PDs. Given that negative moods are more commonly experienced among persons who are anxious17 and less common among those who are extraverted17 and 18 or impulsive,21 and 22 the degree of negative moods observed was hypothesized to vary as a function of PD dimensions. Specifically, negative mood was hypothesized to covary with A-F (anxious) PD characteristics and show little association with other PD features once common variance associated with A-F PD features was removed. Second, consistent with other research studies,23, 24 and 25 mood variability was expected to emerge as a stable individual difference variable among study participants. Despite the observation that mood variability has unreliable associations with common personality dimensions,23, 26, 27, 28 and 29 there is a suggestion that some PD traits, particularly those associated with E-D and O-E pathology, may be associated with mood variability.29, 32, 33 and 34 Given the inconsistent findings on the association between mood variability and personality, this study explored the tentative hypothesis that some E-D and O-E PD traits, most notably borderline and schizotypal PD traits, may demonstrate associations with mood variability.

Results Inter-rater reliability of PD diagnoses and dimensional scores One third of the 149 SCID-II interviews (n = 50) were reassessed from audiotapes by a second rater, a doctoral candidate in clinical psychology, for purposes of establishing inter-rater reliability. Interviews reviewed for reliability purposes were randomly selected with the constraint that three-quarters (n = 38) corresponded to individuals with at least one PD. Overall, inter-rater reliability was very good. For categorical judgments across the 12 PDs that occurred with sufficient frequency (i.e., rated as present by the interviewer and reliability assessor on 10 or more occasions; n = 8 of 12 PD concepts), the median kappa value was .91 (range, .85 to .95). At the cluster level of analysis, corresponding kappa values were .92 for the A-F cluster, .87 for the E-D cluster, and .86 for the O-E cluster. Finally, kappa for the diagnosis of any PD (including provisional PDs) across participants was .90. For dimensional scores, the median intraclass correlation coefficient (ICC Formula 1,1)61 for the 12 PDs was .97 (range, .87 to .99). For the dimensional scores associated with each of the 3 PD clusters, ICCs were .99 for A-F, .98 for E-D, and .88 for O-E. Finally, the ICC that corresponded to the degree of agreement in the dimensional score based on all PD features (including those for provisional PDs) was .93. Farmer and Chapman49 provide additional reliability and diagnostic accuracy data associated with this sample. Representativeness of retained sample relative to entire sample As noted, 98 of 149 participants (66%) provided usable ESM data. When participants who provided usable data were contrasted with those who did not, few differences were observed. Both groups evidenced similar proportions of individuals who met criteria for at least one PD, χ2(1) = 0.02, not significant (NS) (56.9% v 58.2% for noncompleters and completers, respectively). A set of 11 chi-square analyses (one for each PD concept with the exception of dependent PD, a diagnosis that was not observed in the present research) revealed an absence of any significant effect for the presence versus absence of specific PDs and the provision of usable ESM data. When analyzed at the dimensional level, there were no significant differences in the three cluster dimensional scores (O-E, E-D, and A-F) as a function of the provision of usable ESM data. When dimensional scores for individual PD concepts were considered, only one of 12 contrasts revealed a significant difference. This was observed for obsessive-compulsive PD, whereby those who produced usable ESM data had a higher dimensional score for this PD category (mean = .36) than those who did not (mean = .30), t(147) = 1.98, P < .05. No significant sex difference between groups was noted, χ2(1) = 0.53, although there was a significant age difference, F(1, 147) = 5.61, P < .05. Those who provided usable ESM data were significantly older (mean = 29.49) than those who did not (mean = 25.75). PD diagnostic and dimensional score characteristics of sample Table 1 reports the diagnostic characteristics and mean dimensional scores for the 98 persons who provided usable ESM data. Those individuals who met diagnostic criteria for at least one PD (n = 57, or 58%) had, on average, 2.28 (SD = 1.67) diagnosed PDs. Those with PDs did not significantly differ from those without PDs in terms of age, F(1, 96) = 0.50, or in sex distribution, χ2(1) = 2.42. Table 1. Personality Disorder Diagnoses and Mean Dimensional Scores (N = 98) Frequency (%) of Diagnosis Dimensional Score Mean for Sample Anxious-fearful cluster 36 (36.7) 0.25 Avoidant 17 (17.3) 0.26 Dependent 0 (0.0) 0.12 Obsessive-compulsive 26 (26.5) 0.36 Odd-eccentric cluster 19 (19.4) 0.24 Paranoid 13 (13.3) 0.28 Schizotypal 7 (7.1) 0.29 Schizoid 2 (2.0) 0.15 Erratic-dramatic cluster 28 (28.6) 0.23 Histrionic 5 (5.1) 0.19 Narcissistic 11 (11.2) 0.27 Borderline 17 (17.3) 0.34 Antisocial 9 (9.2) 0.14 Provisional personality disorders Passive-aggressive 8 (8.2) 0.25 Depressive 15 (15.3) 0.34 Any personality disorder 57 (58.1) — All personality disorders — 0.24 Table options Factor analysis of PD dimensional scores To investigate the degree to which PD concepts displayed the dimensionality implied in the DSM-IV PD clustering scheme, a factor analysis of PD dimensional scores was performed with varimax (orthogonal) rotation, with factors extracted according to the principle components method and the eigenvalue >1.0 rule. Provisional PD concepts (i.e., depressive and passive-aggressive) that have not been incorporated into the cluster scheme were excluded from this analysis. As displayed in Table 2, three factors were extracted that accounted for 64.5% of the cumulative variance in dimensional PD ratings (34.5% for factor 1; 19.2% for factor 2; and 10.9% for factor 3). This analysis essentially supported the DSM-IV PD clustering scheme. Factor 1 was defined by each of the E-D PD concepts (histrionic, narcissistic, borderline, antisocial). PD concepts that largely defined factor 2 were from the O-E cluster (schizoid, schizotypal, paranoid) as well as avoidant PD from the A-F cluster. Factor 3 was defined by two of the A-F PDs (dependent and compulsive). Avoidant PD also had a moderate (.51) loading on this factor. Table 2. Factor Analysis of Personality Disorder Dimensional Scores (N = 98): Orthogonal Rotation Diagnostic Concept Factor 1 Factor 2 Factor 3 Histrionic (HST) .84 Narcissistic (NAR) .79 Borderline (BRD) .60 .52 Antisocial (ANT) .62 −.40 Paranoid (PAR) .58 Schizotypal (SZT) .46 .57 Schizoid (SZD) .81 Avoidant (AVD) .68 .51 Dependent (DEP) .74 Obsessive-compulsive (COM) .69 NOTE. All loadings ≤ −.40 or ≥ .40 are presented. The highest loading for a PD concept is bolded. Table options Intercorrelations among personality constructs Intercorrelations of PD cluster dimensional scores Given the general support for the DSM-IV PD clustering scheme from the factor analytic findings described above, intercorrelations among the PD cluster dimensional scores were computed in order to determine the degree of covariation among these cluster concepts. The correlations of A-F and O-E (r = .51, P < .001) and E-D and O-E (r = .53, P < .001) were moderate and significant, whereas the correlation for A-F and E-D (r = .15) was not significant. Pearson and partial correlations corresponding to PD cluster dimensional scores and personality measures Table 3 displays the Pearson and partial correlations among PD cluster dimensional scores and personality measures. Because of the moderate associations that some cluster dimensional scores share, partial correlations are also presented for each PD cluster once common variance associated with the remaining two clusters was removed. A-F dimensional scores were most strongly associated with anxiety, neuroticism, and introversion, but were not significantly associated with impulsivity. In contrast, E-D dimensional scores were most strongly associated with impulsivity, extraversion, and neuroticism, but were not significantly related to anxiety. Patterns of correlations for O-E dimensional scores suggested a tendency toward introversion, neuroticism, and anxiety, with a modest significant negative partial correlation obtained for impulsivity. The overall pattern of correlations suggests that neuroticism is not a particularly good discriminator among the PD clusters. However, anxiety and impulsivity discriminated A-F cluster scores from E-D cluster scores. Finally, O-E correlations showed a similar pattern, albeit attenuated, to those obtained for A-F. Table 3. Correlations (Pearson and partial) Among Personality Disorder Cluster Dimensional Scores and Personality Measures (N = 98) Personality Measures PD Cluster Anxious-Fearful (A-F) Erratic-Dramatic (E-D) Odd-Eccentric (O-E) Pearson Partial Pearson Partial Pearson Partial EPI extraversion −.47‡ −.41‡ .40‡ .63‡ −.24∗ −.39‡ IMP composite −.08 −.09 .61‡ .64‡ .12 −.22∗ EPI neuroticism .56‡ .44‡ .49‡ .36‡ .58‡ .21∗ ANX composite .77‡ .67‡ .12 −.18 .55‡ .33‡ NOTE. Partial correlations are computed once the shared variance associated with dimensional scores from the remaining two personality disorder clusters is removed. Abbreviations: EPI, Eysenck Personality Inventory; IMP, impulsivity composite; ANX, anxiety composite. †P < .01 ∗ P < .05 ‡ P < .001. Table options Overview of experience sampling data Average number of usable responses On average, participants provided 25.54 (SD = 2.85) usable ESFs in response to page signals. Consequently, across the 98 participants, a total of 2,503 usable ESFs were obtained. Persons with one or more PDs were as likely to respond to pages as those without PDs, F(1, 96) = 0.02, NS. Average response time The mean response time to pages was 1.67 minutes (SD = 1.25). Persons with PDs responded to pages as promptly as those without PDs, F(1, 96) = 0.00, NS. Specific mood experiences among those with and without PDs Averaged mood ratings For each participant, ratings for the 20 mood pairs were separately aggregated and then averaged across experience samples. Such an approach yields more stable or reliable indices of trait mood.62 These averaged scores were then used in the computation of an overall sample mean, as well as means for specific subsamples. Table 4 presents the means and standard deviations of averaged mood ratings for the sample as a whole, as well as for two contrast groups, those with and without PDs. In the interpretation of these data, mean values greater than 0.00 indicate a tendency toward the endorsement of the positively valenced mood of the pair, whereas means less than 0.00 indicate a tendency toward the endorsement of the negatively valenced mood. Table 4. Means and Standard Deviations of ESM Mood Ratings for the Entire Sample and as a Function of Personality Disorder Status Mood or State Entire Sample (N = 98) Contrasts t Without PD (n = 41) With PD (n = 57) Mean (SD) Mean (SD) Mean (SD) Sleepy/awake 1.15 (0.88) 1.18 (0.81) 1.13 (0.93) NS Confused/clear 0.83 (0.82) 1.04 (0.76) 0.67 (0.83) 2.23∗ Bored/interested 0.69 (0.74) 0.91 (0.73) 0.54 (0.72) 2.50∗ Restless/calm 0.44 (0.82) 0.77 (0.85) 0.21 (0.72) 3.53‡ Sad/happy 0.40 (0.76) 0.75 (0.77) 0.15 (0.65) 4.19‡ Anxious/carefree 0.00 (0.68) 0.23 (0.61) −0.23 (0.67) 3.57‡ Angry/cheerful 0.42 (0.67) 0.75 (0.71) 0.19 (0.53) 4.48‡ Lonely/sociable 0.44 (0.61) 0.73 (0.53) 0.23 (0.59) 4.34‡ Ashamed/proud 0.20 (0.51) 0.37 (0.64) 0.08 (0.36) 2.57∗ Dull/alert 0.89 (0.80) 1.08 (0.78) 0.75 (0.79) 2.10∗ Tense/relaxed 0.57 (0.94) 1.00 (0.77) 0.25 (0.93) 4.22‡ Disagreeable/agreeable 0.46 (0.63) 0.71 (0.65) 0.28 (0.57) 3.45‡ Rejected/supported 0.30 (0.57) 0.55 (0.68) 0.11 (0.39) 3.68‡ Threatened/safe 0.72 (0.87) 1.02 (0.95) 0.51 (0.75) 2.85† Dissatisfied/satisfied 0.20 (0.75) 0.49 (0.81) 0.00 (0.64) 3.17† Weak/strong 0.33 (0.70) 0.58 (0.74) 0.16 (0.61) 3.07† Empty/fulfilled 0.12 (0.61) 0.47 (0.59) −0.13 (0.51) 5.46‡ Stubborn/cooperative 0.31 (0.57) 0.48 (0.66) 0.19 (0.46) 2.41∗ Passive/active 0.34 (0.58) 0.48 (0.48) 0.24 (0.63) 2.00∗ Closed/open 0.43 (0.67) 0.62 (0.62) 0.29 (0.68) 2.53∗ NOTE. When variances were unequal between groups based on Levene’s test, pooled variance estimates were used in the evaluation of group differences. Negative means indicate stronger association with left term of the pair; positive means indicate stronger association with right term of the pair. When a Bonferroni correction is applied to the critical alpha level, only those t values that are significant at P < .001 remain significant at the P < .05 level. ∗ P < .05 † P < .01 ‡ P < .001. Table options For all 20 pairs of terms, average ratings were positive (or, in the case of anxious/carefree, neutral), indicating a predominance of slightly to moderately positive mood experiences across the entire sample. When individuals who had at least one diagnosable PD were compared with those who did not have a PD, significant differences in moods were apparent. Although those with a PD demonstrated a tendency to experience more neutral or positive moods than negative ones, they had lower mood ratings in comparison to the no PD group. Overall mood variability Mood variability was indexed by the within-subject variances derived from the mood ratings of individual participants over multiple occasions. As such, the basic datum employed for each participant referred to his or her average extremity of mood change.24 Previous research has suggested that mood variability is a stable individual difference characteristic.23, 24 and 25 To evaluate whether this was the case in this study, a 20 × 20 correlation matrix of averaged within-subject mood variances was computed. A total of 190 nonredundant correlations were derived from this matrix, and of these, all were positive and 95% were significant (correlation range, .09 to .74; mean correlation = .48). Consequently, mood variability was a stable individual difference variable in this research. Based on this analysis, the degree of mood variability experienced by those with and without PDs was contrasted. When averaged within-subject variances were contrasted for those with and without PDs, no significant differences were noted for any of the 20 pairs of mood terms. Within-subject mood variances for individual mood terms were also considered for aggregation. Prior to aggregation, an internal consistency analysis was performed on the averaged within-subject mood variances for the 20 mood terms. The result alpha coefficient was .94; therefore, averaged within-subject variances associated with mood terms were highly internally consistent. Mood variances for each of the 20 mood terms were subsequently summed, which resulted in a single index of average mood variability. This index was then evaluated in relation to the presence versus absence of any PD, whereby no significant difference in mood variability was noted, t(96) = 0.12. Factor analysis of mood ratings As a precursor to an investigation of mood in relation to specific PDs and associated dimensional scores, a factor analysis was conducted to identify and reduce any underlying dimensionality associated with mood ratings. In this analysis, promax (oblique) rotation was specified and factors were extracted according to the principle components method and eigenvalue >1.0 rule. Table 5 presents the results from this analysis. Table 5. Factor Analysis of Averaged Mood Ratings (N = 98): Oblique Rotation Mood or State Dimensions of Mood Factor 1: Unpleasant v Pleasant Factor 2: Activated Unpleasant v Unactivated Pleasant Factor 3: Unactivated Unpleasant v Activated Pleasant Sleepy/awake .93 Confused/clear .73 Bored/interested .69 Restless/calm .91 Sad/happy .70 Anxious/carefree .79 Angry/cheerful .69 Lonely/sociable .82 Ashamed/proud .97 Dull/alert .85 Tense/relaxed .88 Disagreeable/agreeable .64 Rejected/supported .97 Threatened/safe .62 Dissatisfied/satisfied .72 Weak/strong .73 Empty/fulfilled .79 Stubborn/cooperative .82 Passive/active .62 Closed/open .59 NOTE. All loadings ≤−.40 or ≥.40 are presented. Table options Three factors were extracted that accounted for 75.7% of cumulative variance in averaged mood ratings (59.4% for factor 1; 10.6% for factor 2; and 5.8% for factor 3). Factor 1 was largely defined by items that corresponded to Larsen and Diener’s63 and Russell’s58 unpleasant-pleasant dimension. Items with the highest loadings on this factor emphasized mood states associated with interpersonal disunion versus connection (e.g., rejected/supported, ashamed/proud, lonely/sociable). Factor 2 resembled Larsen and Diener’s63 activated unpleasant-unactivated pleasant dimension, and was defined primarily by mood pairs that contrasted anxious tension versus contentment (e.g., tense/relaxed, restless/calm, anxious/carefree). Factor 3, which demonstrated a correspondence to Russell’s58 arousal-sleep dimension and Larsen and Diener’s63 unactivated unpleasant-activated pleasant dimension, was largely defined by mood pairs that contrasted deactivation versus activation (e.g., dull/alert, sleepy/awake, bored/interested). Each of the 20 mood pairs substantially loaded (i.e., .59 or higher) on one of these factor dimensions. Subscales based on these factor-analytically derived dimensions were generated, whereby mood pairs that loaded highest on a given factor were included in the subscale that corresponded to that factor. The unpleasant-pleasant subscale based on the first rotated factor contained 11 items and demonstrated excellent internal consistency (α = .97) as did the four-item activated unpleasant-unactivated pleasant subscale based on the second rotated factor (α = .90). Finally, the five-item subscale generated from the third rotated factor, unactivated unpleasant-activated pleasant, displayed very good internal consistency (α = .87). Prior to performing the analyses that follow, subscale scores that corresponded to the mood dimensions identified in the factor analysis were standardized within the sample. As a consequence, means reported below are based on standard scores. Mood experiences in relation to PD diagnostic categories Averaged mood ratings Contrasts in mood experiences were performed at three different levels of categorical analysis: (1) the presence versus absence of individual PD diagnostic categories, (2) the presence versus absence of a cluster-level diagnosis, and (3) the presence versus absence of any PD. At the level of individual PD diagnostic categories, six of 12 PDs (avoidant, obsessive-compulsive, paranoid, narcissistic, borderline, and depressive) were diagnosed 10 or more times within the study sample. Scores that corresponded to each of the three mood dimensions were contrasted for those with and without these PDs, separately for each PD. Table 6 notes the discriminating and nondiscriminating mood experiences associated with each of these PD concepts. Table 6. Contrasts in Mood Experience as a Function of Personality Disorder Categories and Specific Personality Disorder Diagnostic Concepts Personality Disorder Category or Concept No. of Participants With/Without Diagnosis Dimensions of Mood: Means (SD) Factor 1: Unpleasant v Pleasant Factor 2: Activated Unpleasant v Unactivated Pleasant Factor 3: Unactivated Unpleasant v Activated Pleasant Specific PD Concepts a Avoidant Present 17 −.80 (.67) −.83 (.89) −.62 (.84) Absent 81 .17 (.98)‡ .17 (.94)‡ .13 (.96)† Obsessive-compulsive Present 26 −.38 (.88) −.44 (.86) −.19 (.96) Absent 72 .14 (1.01)∗ .16 (1.00)† .07 (1.01) Narcissistic Present 11 −.20 (.66) .04 (.70) .26 (1.11) Absent 87 .03 (1.04) .00 (1.03) −.03 (.99) Borderline Present 17 −.30 (.99) −.35 (1.11) −.03 (1.15) Absent 81 .06 (1.00) .07 (.97) .01 (.97) Paranoid Present 13 −.51 (.84) −.59 (.85) −.54 (1.05) Absent 85 .08 (1.00)∗ .09 (.99)∗ .08 (.97)∗ Depressive Present 15 −.88 (.72) −1.05 (.88) −.46 (1.12) Absent 83 .16 (.96)‡ .19 (.90)‡ .08 (.96) Cluster Diagnosis Any A-F PD Present 36 −.48 (.84) −.51 (.90) −.34 (.93) Absent 62 .28 (.99)‡ .30 (.94)‡ .20 (.99)∗ Any E-D PD Present 28 −.22 (.82) −.21 (.94) −.01 (1.08) Absent 70 .09 (1.06) .08 (1.02) .00 (.97) Any O-E PD Present 19 −.50 (.77) −.56 (.92) −.42 (1.00) Absent 79 .12 (1.02)∗ .13 (.98)† .10 (.98)∗ Any Diagnosis of a PD Present 57 −.34 (.81) −.34 (.89) −.22 (.99) Absent 41 .47 (1.06)‡ .47 (.96)‡ .31 (.95)† NOTE. All means are expressed as standard scores. For all significant contrasts as determined by the t-test statistic, the presence of a PD diagnostic category was associated with less favorable moods. When a Bonferroni correction was applied to the critical alpha level to adjust for all contrasts performed, only those t values significant at the P < .0017 level remained significant at the P < .05 level. ∗ P < .05 † P < .01 ‡ P < .001. a Individual PD diagnostic categories in this column only include those diagnosed as positive on at least 10 occasions within the sample. Table options When a Bonferroni correction was applied to the critical alpha level to adjust for the number of contrasts performed, only those t values that were significant at the P < .0017 level remained significant at the P < .05 level. Results indicated that individuals with avoidant and/or depressive PDs were more likely relative to those without either of these conditions to experience less pleasant moods associated with the factor 1 (unpleasant v pleasant) and factor 2 (activated unpleasant v unactivated pleasant) but not factor 3 (unactivated unpleasant v activated pleasant) mood dimensions. At the cluster level of analysis (from which depressive and passive-aggressive PDs are excluded), those with any A-F PD compared to those without a PD from this cluster had less pleasant moods associated with factors 1 and 2. Finally, those with any PD compared to those without a PD displayed less pleasant factor 1 and 2 related moods. Overall mood variability When mood variability was examined in relation to the presence versus absence of specific PD diagnostic concepts observed with sufficient frequency (i.e., 10 or more occasions within the sample), presence versus absence of cluster-level PD diagnoses, or the presence versus absence of any PD, no significant differences were observed in any analysis. Mood experiences in relation to PD dimensional scores Averaged mood ratings ESM mood rating data were also analyzed with reference to PD dimensional scores. Dimensional representations of PD concepts have been shown to possess greater psychometric advantages relative to the diagnostic categorization of these conditions.64 Because passive-aggressive and depressive PDs have not been assigned to a symptomatological cluster due to their provisional status, they were excluded from cluster-based analyses. Also, because of the moderate association that the O-E dimensional score had with dimensional scores derived from the remaining two clusters, both Pearson and partial correlations between PD cluster scores and averaged mood ratings are presented in Table 7. Table 7. Correlations (Pearson and partial) Among Personality Disorder Cluster Dimensional Scores and Averaged Mood Ratings (N = 98) Mood Dimension PD Cluster Anxious-Fearful (A-F) Erratic-Dramatic (E-D) Odd-Eccentric (O-E) Pearson Partial Pearson Partial Pearson Partial Factor 1: unpleasant v pleasant −.50‡ −.40‡ −.27† −.16 −.39‡ −.07 Factor 2: activated unpleasant v unactivated pleasant −.49‡ −.42‡ −.23∗ −.16 −.31† .02 Factor 3: unactivated unpleasant v activated pleasant −.39‡ −.35‡ −.18 −.14 −.21∗ .06 NOTE. Negative correlations indicate stronger association with less pleasant moods. Partial correlations are computed once the shared variance associated with dimensional scores from the remaining two PD clusters is removed. ∗ P < .05 † P < .01 ‡ P < .001. Table options When Pearson correlations for PD cluster scores and mood dimension scores were considered, a fairly reliable pattern was noted. All correlations for A-F cluster dimensional scores and mood ratings were significant and moderately negative; that is, as the number of A-F features increased, values for each of the three mood dimensions decreased. When E-D cluster dimensional scores were correlated with mood dimension scores, two of the three correlations were significant although all were relatively modest (range of r values, −.18 to −.30). Finally, correlations of O-E dimensional scores with mood dimensions were all negative and significant. A somewhat different pattern emerged, however, when partial correlations were considered. These correlations were based on cluster dimensional scores once the shared variance associated with dimensional scores from the remaining two clusters was removed. When partial correlations were computed for A-F dimensional scores and mood ratings, correlations with each of the mood dimensions continued to be moderately negative and significant, although attenuated relative to the corresponding Pearson correlations. Greater change was apparent when partial correlations for E-D and O-E with mood dimensions were considered, as none of the resultant correlations were significantly different from zero. Table 8 displays the correlations that the 12 PD dimensional scores shared with the three mood dimensions. All of the individual A-F PDs had significant and negative correlations with each of the three mood dimensions, with avoidant PD consistently displaying the largest association with mood quality. While none of the O-E PDs demonstrated an association with factor 3 moods, they generally correlated negatively and significantly with the other mood dimensions. With the exception of borderline PD, which significantly and negatively correlated with each of the three mood dimensions, all other PDs within the E-D cluster either had no association or only a very modest association with mood quality. Noteworthy is the observation that two of the three correlations that borderline PD traits shared with mood remained significant once common variance that borderline PD dimensional scores had with A-F dimensional scores was statistically removed (partial r values: factor 1 = −.29, P < .01; factor 2 = −.27, P < .01; factor 3 = −.13, NS). Consequently, negative moods associated with borderline PD traits cannot be fully accounted for by the presence of comorbid A-F PD features. Finally, both the depressive and passive-aggressive provisional PD concepts correlated significantly and negatively with each of the mood dimensions. Table 8. Specific Personality Disorder Dimensional Scores Correlated With Dimensions of Mood Experience Personality Disorder Dimension Dimensions of Mood Factor 1: Unpleasant v Pleasant Factor 2: Activated Unpleasant v Unactivated Pleasant Factor 3: Unactivated Unpleasant v Activated Pleasant Avoidant −.46‡ −.46‡ −.36‡ Dependent −.28† −.27† −.20∗ Obsessive-compulsive −.32† −.31† −.26† Histrionic −.02 −.04 −.12 Narcissistic −.22∗ −.11 −.12 Borderline −.42‡ −.40‡ −.25∗ Antisocial −.06 −.05 .03 Paranoid −.32† −.29† −.18 Schizotypal −.26∗ −.14 −.19 Schizoid −.32† −.28† −.11 Depressive −.48‡ −.49‡ −.33† Passive-aggressive −.35‡ −.32† −.28† NOTE. Negative correlations indicate stronger association with unpleasant moods; positive correlations indicate stronger association positive moods. ∗ P < .05 † P < .01 ‡ P < .001. Table options Overall mood variability At the cluster dimensional level of analysis, nonsignificant correlations were obtained for A-F (r = −.17), E-D (r = .04), and O-E (r = −.11) dimensional scores. Partial correlations were also computed that examined the degree of association between mood variability and individual PD cluster dimensional scores once shared variance with the remaining two PD cluster dimensional scores was statistically removed. Similarly, all partial correlations associated with mood dimensions and PD cluster score dimensions once common variance among the PD clusters was removed were nonsignificant. When dimensional scores were considered for each of the 12 PD concepts, only one significant association with mood variability was observed, and this was a negative association between the avoidant PD dimensional score and overall mood variability (r = −.23, P < .05). This finding indicated that as avoidant PD traits became more pronounced, the degree of mood variability became attenuated. Of particular note was a near-zero correlation between borderline (r = .02, NS) and schizotypal (r = −.14, NS) PD dimensional scores and the mood variability aggregate index.