Dynamic of Change in Pathological PersonalityTrait Dimensions: A Latent Change AnalysisAmong at-Risk Women

Baptiste Barbot & Scott R. Hunter & Elena L. Grigorenko &

Suniya S. Luthar

Published online: 21 November 2012# Springer Science+Business Media New York 2012

Abstract This study explores longitudinally a four-factorstructure of pathological personality trait dimensions(PPTDs) to examine both its structural stability and intra-individual changes among PPTDs over time. PersonalityDisorder (PD) scales of the Millon Clinical MultiaxialInventory-III were administered to 361 low-income womenwith various psychiatric conditions (drug dependence, depres-sion), who were followed in a two-wave study over 5-years.Cross-sectional and longitudinal factor analyses outlined arobust factorial structure of PPTDs, extrinsically invariantover time, representing Negative Emotionality, Introversion,Antagonism and Impulsivity. Despite moderate rank-orderstability in the PPTDs, results also indicated substantialintra-individual variability in the degree and direction ofchange, consistent with trajectories of change in participants’clinical diagnoses. Results are discussed in light of currentdebates on the structure and dynamic of pathologicalpersonality.

Keywords Pathological personality . Personality disorders .

MCMI . Longitudinal factor analysis

Introduction

Current debates on the nature of personality disorders (PD)have included calls for an empirical structure of personalitypathology to be incorporated in the new edition of theDiagnostic and Statistical Manual of Mental Disorders(DSM-5). Toward this end, cross-sectional studies of clini-cal and non-clinical samples have sought to identify a factorstructure of pathological personality trait dimensions(PPTD) that underlie PDs. While consensus is building fora four- to five-factor structure, longitudinal evidence for thisstructure is lacking, precluding the structural debate to beinformed by developmental data. The need for longitudinalevaluation of PPTD is highlighted by a growing body oflongitudinal research reporting significant rates of change inPDs over time, a finding that calls into question the notionthat PDs are stable and enduring. Thus, ongoing develop-ments in the understanding of PD will require such studiesas the present one to determine (a) whether the structure ofpersonality pathology is stable and reproducible over time,and (b) if structural stability excludes intra-individualchanges in naturalistic and/or treatment conditions. Thepresent study integrates the structural and developmentalcontroversies of PD in a single model, through a longitudi-nal structural analysis of the PD scales of the Millon ClinicalMultiaxial Inventory III (MCMI-III; Millon et al. 2006)among a sample of at-risk women followed over 5-years(Luthar and Sexton 2007).

The Dimensional Nature of Personality Disordersand the Empirical Structure of Pathological PersonalityTraits

Proposed changes to the categorical conceptualization ofPDs in the American Psychiatric Association’s (APA)Diagnostic and Statistics Manual of Mental Disorders

B. Barbot : E. L. Grigorenko : S. S. LutharYale University,New Haven, CT, USA

S. R. HunterUniversity of California,Los Angeles, CA, USA

E. L. GrigorenkoMoscow State University,Moscow, Russia

E. L. Grigorenko : S. S. Luthar (*)Department of Counseling and Clinical Psychology,Teachers College, Columbia University,P.O. Box 133, West 120th Street,New York, NY 10027, USAe-mail: suniya.luthar@columbia.edu

J Psychopathol Behav Assess (2013) 35:173–185DOI 10.1007/s10862-012-9331-4

(American Psychiatric Association 2000) reflect a broaderconceptual question of whether to define and describe person-ality pathology dimensionally, as a quantitatively extremeexpression of normal functioning, or categorically, as a quali-tatively distinct process (Wright 2011). Limitations of thecategorical model are well-recognized and dimensional mod-els of PDs are favored by many (Eaton et al. 2011; Krueger etal. 2007; Widiger and Samuel 2005; Widiger and Simonsen2005). As noted byWesten andMuderrisoglu (2006), virtuallyall these dimensional models are “trait models” derived fromthe factor analysis of self-report data. Research on the empir-ical structure of pathological personality trait dimensions(PPTDs) has identified four broad domains appearing withreasonable consistency in cross-sectional studies. They in-clude: (a) negative emotionality or emotional dysregulation,(b) introversion or detachment, (c) antagonism, and (d) disin-hibition (Krueger et al. 2011; Livesley 2007; Sheets andCraighead 2007; Widiger and Simonsen 2005). Many studies(e.g., Tackett et al. 2008), and meta-analyses (e.g., O'Connor2005), have identified these same domains that purportedly tapthe core traits underlying pathological personality functioningthrough structural analysis of numerous self-rated pathologicalpersonality inventories.

Many have also observed that these four factors are recog-nizable as maladaptive variants of “normal” personality, asoperationalized for instance with the Five Factor Model(FFM) of personality (e.g., Lenzenweger 2010; Widiger andCosta In press). Therefore, researchers and clinicians alikehave advocated for the reconceptualization of personalitypathology within the multidimensional space of a “normal”personality model such as the FFM (e.g., Lynam in press;Markon et al. 2005; O'Connor 2005; Samuel and Gore inpress). Accordingly, meta-analyses have confirmed the rela-tionship between current DSM personality disorders and theFFM, with strong relationships observed in four of the fivedomains, excepting the domain of Openness, which overlapsonly weakly (Samuel and Widiger 2008). Consistently, themaladaptive variant of Openness (often referred to asPsychoticism, and covering major features of SchizotypalPD) is the smallest (Widiger and Simonsen 2005) and leaststable, and as such is often under-identified in factor analysesof PD inventories that are not specifically designed to capturethis domain, while the other domains appears with greatconsistency in many PD inventories.

For example, attempts to validate the theoretical model ofpersonality pathology underlying the MCMI-III have foundlittle consistency for the presumed theoretical model(Mullins-Sweatt and Widiger 2007; O'Connor and Dyce1998). In addition, questions have emerged regarding theability of this instruments to discriminate between distinctPD categories and to serve as robust diagnostic tool inpractical settings given the rather high overlap betweenMCMI-III PD scales (e.g., Rogers et al. 1999). This overlap

suggests that a careful examination of MCMI structure isneeded to reveal the higher-order constructs (PPTDs) under-lying the covariation between scale scores (PhilippeRushton and Irwing 2009). In this line, a four-factor struc-ture often appeared to be the best underlying model withMCMI data gathered from various clinical and non-clinicalsamples across cultures (e.g., Cuevas et al. 2008; Dyce et al.1997; Rossi et al. 2010). These four factors have beenobserved to align well with the four PPTDs introducedabove (Krueger et al. 2011; Rossi et al. 2010), thus furthersupporting the relevance of the four-factor model of mal-adaptive personality traits underlying PDs, as this structureis identifiable across a number of assessments of personalitypathology which sample a comprehensive range of PDfacets, such as the MCMI-III. In the present study, we seekto explore the extrinsic (i.e. structural) and intrinsic (i.e.within-individual) stability of this four-factor structure ofPD over time, as represented in the MCMI-III. Althoughthe MCMI was designed to understand the development andcourse of individual psychopathology in the context of anindividuals’ personality (Millon 1983), relatively few longi-tudinal analyses have been conducted with this instruments.

Stability and Change of Pathological PersonalityStructure and Diagnosis

Until recently, the empirical literature documenting long-termstability of personality pathology has been limited(Lenzenweger 1999, 2010). Moreover, most longitudinalstudies on PDs have exhibited methodological problems thatlimited their conclusions (Skodol 2008), including ignoringinherent unreliability in the measurement of change (Ferguson2010). A new generation of rigorous longitudinal studies ofcommunity population and clinical samples has emerged(Lenzenweger 2006; Skodol et al. 2005; Skodol 2008) andhas revitalized a debate regarding the stability of PDs (Clark2007; Krueger and Tackett 2005; Skodol 2008). Longitudinalevidence from clinical and non-clinical samples has revealedthat individual change in pathology personality traits occursfor individuals both in- and out-of-treatment, and thesechanges differ considerably across individuals (Johnson etal. 2000; Lenzenweger et al. 2004; Watson et al. 2008).

As a general trend, longitudinal studies suggest that per-sonality pathology improves over time (i.e. symptoms de-crease) at significant rates (Grilo et al. 2004, 2010; Johnsonet al. 2000; Lenzenweger et al. 2004; Lenzenweger 2010;Sanislow et al. 2009; Skodol et al. 2005; Skodol 2008).However, several studies report rank-order and mean-levelstability difference according to the diagnostic approach used:dimensional or categorical (Johnson et al. 2000; Samuel et al.2011). The dimensional system yields significantly greaterstability, especially when using self-report questionnaires as

174 J Psychopathol Behav Assess (2013) 35:173–185

opposed to structured interviews (Samuel et al. 2011). Thesefindings suggest conceptual limitations inherent in categoricalmodels of PDs, in particular, the use of arbitrary thresholds(Grilo et al. 2004) leading to measurement error in diagnosis(Clark 2007). Research has also evidenced differential stabil-ity across PD categories regardless of the method for diagno-sis (e.g., Durbin and Klein 2006; Nestadt et al. 2010). Usually,Cluster B disorders (Antisocial, Borderline, HistrionicNarcissistic) tend to exhibit the highest long-term stability(Durbin and Klein 2006).

The mechanisms by which change in personality pathol-ogy occurs are still not well understood. Lenzenweger andWillett (2007) proposed that change in PD features resultfrom changes in neurobehavioral systems that manifest aspathological personality traits. Consistently, Warner andcolleagues (2004) have proposed that personality traits(such as the PPTDs) may represent causally proximal mech-anisms that influence the expression and variability of PDover time. If so, a greater understanding of the dynamic ofchange of maladaptive personality traits is of paramountimportance for both theory and treatment (see Warner etal. 2004). Despite evidence of change in personality pathol-ogy, the DSM-IV-TR still defines PDs as “stable and endur-ing, reflecting a persistent pattern of maladaptive personalitythroughout the life course” (e.g., Skodol 2008).

Although evidence for such stability is mixed, researchershavemade limited attempts to integrate the developmental andthe structural controversies of personality traits underlyingPDs, placing the “core structure” of PPTDs (i.e., the four-factor structure outlined above) in a longitudinal perspective.Among these attempts, an integrative hypothesis suggests thatPDs may be characterized by maladaptive trait constellationsthat are stable in their structure but can change or fluctuate inseverity or expression over time (Grilo et al. 2004).Nevertheless, only Sanislow and collaborators (2009) exam-ined the longitudinal invariance of such PD latent constructs(i.e. their “extrinsic” stability), but these constructs were notaligned with the more widely acceptedmaladaptive variants ofthe FFM, that is, the core PPTDs. Such a longitudinal exam-ination would provide additional support for the empiricalstructure of personality pathology. Furthermore, longitudinalanalyses of the PPTDs will likely contribute to a better under-standing of PDs development and change.

The objective of the present study was (a) to investigate thelongitudinal stability of the four PPTDs structure by testing itsfactorial invariance over a 5-year delay (i.e., extrinsic stabili-ty), while examining (b) within-individual change in PPTDlevels (i.e., intrinsic stability). We then compared the averagelevels of PPTDs as a function of various groups (such assubstance-abusing, depressed, and a non-clinical group), aswell as within-individual change in the PPTD levels as afunction of patterns of stability, remission, and other formsof change in the clinical diagnosis as derived through an

external DSM-based assessment. This focus was meant toelucidate (c) the ability of the PPTD constructs to discriminatevarious clinical groups, and (d) their sensitivity to change inparticipant’s mental health condition, informing their useful-ness to capture the pathological personality component ofvarious psychopathological conditions beyond PDs solely.

Method

The study sample chosen for the study consisted of high-riskclinical and non-clinical mothers, as such a population isexpected to have a high prevalence of pathological person-ality traits. To address many methodological issues general-ly disregarded in the study of personality stability andchange, and in particular, the unreliability in the measure-ment of change (Ferguson 2010), both PPTD structure anddevelopment were examined at a latent level, in a structuralequation modeling framework.

Participants

Data used in the present study was collected through alongitudinal research project on maternal drug abuse, psy-chopathology, and child adaptation (Luthar and Sexton2007). The initial sample consisted of 361 low-incomemother-child dyads living in an urban area of Connecticut,who have been followed since 1996. At the launch of theproject, mothers’ age ranged from 23.5 to 55.8 (Mage0

38.2 years, SD06.2) and included women of African-American (51.5 %), Caucasian (34.2 %), Hispanic (6.3 %),Native American (0.8 %), Asian (0.3 %) or Mixed (5 %)descent (2 % of the participants were unidentified ethnicallyor culturally). Women were recruited from outpatient treat-ment facilities for substance abuse and other mental healthproblems (chiefly depression and anxiety), as well as fromcommunity settings. Therefore, the sample comprises bothclinical and non-clinical groups, including a high prevalenceof substance use disorders, historically associated with PDs(e.g., Jahng et al. 2011). “Clinical group” membership inthis study was defined by the presence of a DSM diagnosisat the time of data collection (see “Data Analyses”).

Instruments

DSM Diagnoses

The Computerized Version of the Diagnostic InterviewSchedule (DIS-IV; Robins et al. 2000) was used to assessmajor psychiatric disorders as defined in the DSM-IV. TheDIS-IV is a fully structured questionnaire designed to iden-tify the presence of psychiatric disorders by quantifyingpsychiatric symptoms endorsed by a respondent. It offers a

J Psychopathol Behav Assess (2013) 35:173–185 175

lifetime history of symptoms as well as a diagnosis ofcurrent syndrome (a disorder is counted as current if anyof its symptoms had been present within the past year). Dueto the fully structured, highly standardized nature of themeasure, non-clinicians may administer the DIS-IV withhigh level of reliability. Metric properties of the DIS-IVand previous versions have been studied extensively andshowed fair to excellent level of inter-diagnostician reliabil-ity (Kappa ranging from .40 to .77 with non-clinician inter-viewer) with various samples, including substance abusers(e.g., Dascalu et al. 2001), and strong agreement betweenother DSM instruments (e.g., Hasin and Grant 1987) sup-porting its good criterion validity (Malgady et al. 1992).

Personality Pathology

The MCMI-III (Millon et al. 2006) is a widely used, 175-item,self-report inventory which consists of 24 clinical scales includ-ing 14 pathological personality scales. The Clinical PersonalityPattern scales which represent mild to moderate personalitypathologies include: Schizoid, Avoidant, Depressive,Dependent, Histrionic, Narcissistic, Antisocial, Aggressive(Sadistic), Compulsive, Negativistic, and Masochistic scales.In addition, three severe personality pathology scales are in-cluded: Schizotypal, Borderline, and Paranoid. Across contexts(such as substance abusing samples; e.g., Hesse et al. 2012),this instrument displays strong psychometric properties includ-ing proper reliability and validity (Millon et al. 2006).

Procedure

All women who expressed interest in participation werescreened to determine eligibility for the main research pro-gram (Luthar and Sexton 2007). Women were screened forpsychotic features and excluded if they were on antipsychoticmedications and/or had a history of psychosis. Before beingenrolled in the research program, voluntary informed consentwas obtained. After each measurement occasion, participantswere compensated monetarily for their participation.

The first measurement occasion of this project (T1) oc-curred in 1996; Participants were contacted by phone andmail to take part to the follow-up (T2) and a large part of thesample returned, after an average of 60 months (5 years).For both measurement occasions, DIS-IV interviews wereconducted by trained non-clinician interviewers, andMCMI-III protocol were self-administered. Following theMCMI-III scoring guideline (Millon et al. 2006), only com-plete and valid test-profiles were used. One profile at T1 and4 profiles at T2 were invalidated. As a result, 80 % of theparent sample (N0288) presented complete data (i.e., DIS-IV diagnoses and MCMI scores) at both T1 and T2.Attrition analysis demonstrated that the prevalence of thevarious clinical groups at T1 were highly similar across

complete (with complete data at both T1 and T2) andincomplete respondents (with complete data at only T1),indicating that attrition was not biased toward a particularclinical group (χ202.4, df03, p0 .49).

Data Analyses

Grouping DSM Diagnosis and Trajectories of Changein Diagnosis

Current DSM diagnoses for both T1 and T2 (determinedwith the DIS-IV), were first stratified into four main groupsat each measurement occasion in view of planned analyses:Substance only (substance dependence without affective/anxiety disorders); affective/anxiety only (depressive or anx-iety diagnoses without substance dependence), comorbid(substance dependence and affective/anxiety diagnoses),and neither (neither substance dependence nor affective/anxiety diagnosis). Specifically, the substance only groupincluded 7.6 % of the sample at T1 and 3.1 % at T2. Theaffective/anxiety only group represented 26.7 % at T1 and15.3 % at T2. The comorbid group represented 8 % of theparticipants at T1 and 3.1 % at T2. Participant with neithersubstance dependence nor affective/anxiety diagnosis (nei-ther) included 57.6 % of the participants at T1 and 74 % atT2. Despite the apparent trend of reduction of clinical diag-nosis between T1 and T2 (i.e., increase of the prevalence ofthe neither group), these group distributions mask consider-able variability in the observed patterns of change. Figure 1depicts the four-state transition matrix between DSM

Fig. 1 Simple four-state diagram of transition in the DSM diagnosisgroups over time. Values represent transition probabilities. The thick-ness of the lines is proportional to the probabilities of transition.Autoregressive arrows represent probability of stability. Transitionpath below .05 are not represented in the diagram

176 J Psychopathol Behav Assess (2013) 35:173–185

diagnosis over time, revealing patterns of remission, stabil-ity, and transition across the clinical groups.

For example, participants assigned to the comorbid groupat T1 were mostly found in the neither group at T2. Indeed,48 % of this group was characterized by “remission,” while35 % “transited” to the group with depression or anxietydisorders without substance dependence (they assigned tothe affective/anxiety only group). Finally, only 13 % of themembers of the comorbid group at T1 remained in thisgroup at T2. Based on the main patterns of change observedin our data, we proceeded to a second classification ofparticipants into three main forms of change in the clinicalgroup: “stable”, namely participants assigned to the samegroup at both T1 and T2 (57.8 % of the sample of completecases); “remission”, participants assigned to a clinical groupat T1 (substance only, affective/anxiety only or comorbid)and then assigned to the neither group at T2 (28 %), and“newly diagnosed”, herein, participants with no diagnosis ofsubstance dependence nor affective/anxiety at T1, who fellinto one of the three clinical group at T2 (8 %). Under-represented pathways (including a total of 6.2 % of thetrajectories) were not included in analyses using this group-ing system.

Modeling MCMI-III PD Structure, Longitudinal Invariance,and Latent Change

Only the 14 MCMI-III PD scales were used in this study, aswe focus on PD organization only. Following recommenda-tions in the literature (Cuevas et al. 2008; Dyce et al. 1997;Rossi et al. 2010), MCMI-III Linearly Independent Scales(LIS) scores were used as the basis of all analyses in thisstudy. LIS scores are computed by averaging the “prototyp-al” items for each PD scale, as these items are the mostcentral indicators of the personality disorder construct(Millon et al. 2006). The resulting scores have the advantageof reducing artificial colinearity between scales withoutlosing content validity (Rossi et al. 2007). Preliminary anal-yses indicated that the LIS scores were highly congruentwith the original full-scales scores (i.e. computed followingthe MCMI-III manual) with median correlation of .96 be-tween LIS and full-scale scores at both Time 1 and Time 2.

Confirmatory Factor Analyses (CFA) in a structural equa-tion modeling approach using Full Information MaximumLikelihood estimation (FIML) were used to estimate the fitof two-, three-, and four-factor models of the MCMI-III PDstructure commonly identified in the literature. The specifi-cations of these models were the same as those described byRossi and colleagues (2010). At a second stage, the initialmodels were improved using modification indices suggest-ing which parameter could be fixed or set free (correlatederror terms were allowed only if they were tenable concep-tually). The chi-square difference test (Δχ2) was used to

estimate the statistical significance of the improvement in fitof the various models.

The model fitting the best T1 and T2 data was thenextended into a single longitudinal multivariate factor model(e.g., McArdle and Nesselroade 1994) to test for structuralinvariance over time. In addition, we implemented amultiple-common-factor latent change score (LCS) model(e.g., Ferrer and McArdle 2010; McArdle 2009) allowingestimation of mean-level change in the PPTD factors at alatent level, as well as correlations between these changefactors. This approach avoids the problems inherent in usingdifference scores such as cumulated errors and regression tothe mean (McArdle and Nesselroade 1994; McArdle 2009),that are rarely addressed in the personality-stability and -change literature (Ferguson 2010). The fit of all models wasassessed using the Chi-square tests of fit and four otherwidely used indexes: The normed fit index (NFI), the com-parative fit index (CFI), the root mean square residual(RMR), and the root mean square error of approximation(RMSEA) with its 90 % confidence interval (e.g., Kline2010). A final set of multivariate analyses of variance(MANOVA) was conducted to examine the typical profileof PPTDs associated with the four clinical groups.Following recent support for the dynamic modeling of psy-chological processes (Ferrer and McArdle 2010), trajecto-ries of change in the clinical group (stable, remission, andnewly diagnosed) were also examined using a MANOVAand Tukey post-hoc tests to further explore the associationbetween change in psychiatric condition over time, withlevel of change in the PPTD.

Results

Preliminary Analyses

Internal consistency of the MCMI-III LIS scores was mod-erate (e.g., Histrionic, Narcissistic) to high (e.g., Avoidant,Depressive), with consistent Cronbach’s alphas coefficientsacross measurement occasions, ranging from .62 to .89(MT10 .70, MT20 .71). Distributions of the LIS scores sug-gested possible violation of the assumption of a multivariatenormal data distribution required for planned factor analy-sis. Consequently, normal scores using the Rankit transfor-mation method (Solomon and Sawilowsky 2009) werecomputed and yielded distributional features much closerto the Gaussian distribution. Subsequently, the properties ofthe correlation matrices obtained from the 14 MCMI-III LISnormal scores for both T1 and T2 were examined to ensuretheir suitability for use in the factor analyses. The correla-tion matrices were not identity matrices (T1 Bartlett’s test ofsphericity 0 3177.7, df091, p< .001; T2 Bartlett’s test ofsphericity 0 2757.7, df091, p< .001) and the global

J Psychopathol Behav Assess (2013) 35:173–185 177

measures of sampling adequacy (KMO) were excellent (.94at both T1 and T2), reflecting individual measures of sam-pling adequacy (MSA) that were acceptable to excellent(ranging from .61 to .97 at T1, and from .52 to .96 at T2).Thus, the correlation matrices had the proper features to beused in factor analyses.

Underlying Pathological Personality Trait Dimensions

Table 1 display the fit indices obtained from the three initialmodels (i.e., two-, three-, and four-factor models) and theiradjusted version as suggested by the modification indices,with T1 and T2 data, separately. The adjusted models con-sist of improved versions of the initial models, adding 1 to 3modest correlated errors terms which made substantivesense. In accordance with prior results (Rossi et al. 2010),the two-factor and three-factor models were associated withpoor fit, even in their “adjusted” form. As confirmed by theΔχ2 significance, the best-fitting models for each measure-ment occasion were the models with four factors (in boththeir “initial” and “adjusted” versions), with all fit statisticsindices falling within the acceptable to excellent range(Kline 2010). This solution was parsimonious with fourfactors explaining a total of 62.1 % of the variance of the14 MCMI PD scales used in the model, with explainedvariance ranging from 19.7 % (Compulsive) to 84 %(Borderline) across indicators. Factor loadings of this four-

factor model were similar at both time points (Tucker-Burt-Wrigley-Neuhaus congruence coefficient 0 .91, .75, .83 and.86 for Factors 1, 2, 3 and 4 respectively). Thus, we deriveda consensus factor structure (cf. O'Connor 2005) from bothloading matrices, used for the sake of clarity when interpret-ing the factor composition. This structure and its relatedfactor consensus loadings are depicted in Fig. 2.

Based upon the most salient loadings estimated in thismodel (which were ordered in a similar way to those obtainedby Dyce et al. 1997; Rossi et al. 2010), we named each factoras follows: Factor 1 was labeled “Negative Emotionality” as itdisplayed high loadings on many PD scale items, in particularon the Depressive, Self-Defeating, and Avoidant scales.Factor 2 was labeled “Antagonism” and was mostly loadedwith the Antisocial, Schizoid, Paranoid and Narcissistic scaleitems. Factor 3 was identified as the “Introversion” bipolardimension, with high and negative loading on the Histrionicscale on one end, and a high, positive loading in the Schizoidscale on the other end. Finally, Factor 4 was labeled“Impulsivity”, representing Compulsivity on one end, andthe Antisocial and Borderline scales on the other end. Giventhe composition of this factor and its pattern of associationwith the other PPTDs in the tested population, this factor wasthought to represent mainly the self-damaging and self-injurious impulsivity features (rather than impulse aggressionfor example) often associated with Borderline personalitydisorders. The correlations between the factors were as

Table 1 Goodness of fit indices of the cross-sectional and longitudinal models

Model χ2 df χ2/df p Δχ2 NFI CFI RMR RMSEA (CI)

Models T1

2F initial 322.1 70 4.6 <.001 _ .90 .92 .056 .100 (.089–.111)

2F adjusted 263.6 68 3.9 <.001 <.001 .92 .94 .056 .089 (.078–.101)

3F initial 223.5 64 3.5 <.001 <.001 .93 .95 .042 .083 (.071–.095)

3F adjusted 200.4 63 3.2 <.001 <.001 .94 96 .040 .078 (.066–.090)

4F initial 118.4 55 2.2 <.001 <.001 .96 .98 .024 .057 (.043–.071)

4F adjusted 94.3 54 1.7 <.01 <.001 .97 .99 .022 .046 (.030–.061)

Models T2

2F initial 290.1 70 4.1 <.001 _ .90 .92 .054 .105 (.092–.117)

2F adjusted 255.7 69 3.7 <.001 <.001 .91 .93 .043 .097 (.082–.110)

3F initial 204.4 64 3.2 <.001 <.001 .93 .95 .034 .087 (.074–.101)

3F adjusted 173.0 62 2.8 <.001 <.001 .94 .96 .033 .079 (.065–.093)

4F initial 147.2 55 2.7 <.001 <.001 .95 .97 .026 .076 (.062–.091)

4F adjusted 86.1 52 1.7 <.001 <.001 .97 .99 .021 .048 (.029–.065)

Longitudinal models

4F free change 436.8 276 1.58 <.001 _ .93 .97 .025 .045 (.037–.053)

4F invariant 464.0 302 1.54 <.001 .40 .93 .97 .030 .043 (.035–.051)

4F latent change score 531.4 314 1.69 <.001 _ .92 .96 .071 .049 (.049–.056)

χ20chi-square; df 0 degrees of freedom; p0p value of the chi-square test; Δ χ20p value of the chi-square difference test; NFI Bentler-Bonettnormed fit index; CFI Comparative fit index; RMR root mean square residual; RMSEA root mean square error of approximation; CI090 %confidence interval of RMSEA value

178 J Psychopathol Behav Assess (2013) 35:173–185

follows: Factor 1 correlated .65, .19, and −.01 with factor 2, 3and 4, respectively. Factor 2 correlated −.51, and −.67 withfactor 3 and factor 4 respectively, and factor 3 correlated .53with factor 4 (Fig. 2).

Longitudinal Structural Invariance and Individual Change

In order to confirm longitudinally this outlined four-factorstructure of PPTD as represented in the MCMI-III, we speci-fied a single longitudinal multivariate factor model (e.g.,McArdle and Nesselroade 1994), as shown in Fig. 3a.Specifically, two versions of this longitudinal factorial modelwere tested to estimate measurement invariance over time. Inthe first model, we set free the estimation of the factor loadingsat each time point (“free change” model). The second modeltested the hypothesis of the equality of the factor loading overtime (“invariance” model), to confirm that the latent constructmeasured at each time point had the same substantive meaning(McArdle and Nesselroade 1994). Model fit indices of thesetwo alternative models are presented in Table 1. As indicated,both models displayed very similar, adequate fit to the datawith fit indices in the acceptable to good range. More impor-tantly, the measurement invariance model did not display asignificant degradation in fit as compare to the free changemodel (Δχ2027.2, df026, p0 .40; ΔCFI0 .000), holding thehypothesis of the equality of the factors loading over time.

The evidence of “extrinsic” stability of the PPTD factorstructure underlying the MCMI-III PD scores allowed theexamination of intrinsic, or within-person change in two

common-factor scores over time (McArdle and Nesselroade1994): cross-lagged correlation of two common-factor scoresindicated rank-order stability of .80, .70, .73 and .57 respec-tively for Negative Emotionality, Antagonism, Introversionand Impulsivity (Fig. 3a). In contrast, rank-order stabilitycoefficients of the 14 individual MCMI-III PD scales (at thebasis of the four PPTD constructs) yielded lower stability,with test-retest intra-class correlation coefficients rangingfrom ICC0 .38 (p<.001) for the Compulsive Scale toICC0 .69 (p< .001) for the Avoidant Scale (averageICC0 .59, SD0 .07). Given the delay between the measure-ment occasions in this test-retest analysis (5 years), stabilitycoefficients obtained with both the PPTDs and PPTD indica-tors (i.e., PD scales) are interpreted in terms of stability of theconstructs measured, rather than in terms of reproducibility ofthe scores per se (i.e., test-rest reliability). In this context,results suggest rather high levels of rank order stability inthe PPTDs constructs (latent factor scores) and moderatestability of the PD constructs (manifest scale scores).

In a second step, we built upon the longitudinal invariantmodel (Fig. 3a) to introduce new latent scores that repre-sented the latent change between two common-factor scores(one latent change factor for each PPTD construct). Thismultiple-common-factor latent change score (LCS) model(e.g., McArdle 2009) represented in Fig. 3b, fit our dataadequately (Table 1). The mean estimated in the LCS wereall different from 0 suggesting significant change over time(with MNegative Emotionality 0 −.11, p<.001; MAntagonism 0

−.12, p<.001; MIntroversion 0 .08, p0 .06; MImplulsivity 0 .10,

Fig. 2 Four factor model ofPPTD outlying the MCMI-IIIPD scores. For the sake of clar-ity, loadings under |.25| are notrepresented in the model

J Psychopathol Behav Assess (2013) 35:173–185 179

p0 .011). In addition, the variances in the LCS factors were alldifferent from 0 (p<.001) indicating substantial individualdifferences in the level and direction of change, and allowingthe estimation of inter-correlations between the four LCSwhich were as follows: Negative Emotionality LCS correlated.66, .00, and −.13 with Antagonism LCS, Introversion LCSand Impulsivity LCS respectively. Antagonism LCS correlat-ed −.65, and −.66 with Introversion LCS and Impulsivity LCSrespectively, and Introversion LCS correlated .63 withImpulsivity LCS, (Fig. 3b).

DSM Diagnoses Groups and Their Pathological PersonalityTraits Correlates

Individual factor-scores were derived for each PPTD factorand their LCS extension using the regression method. TheMANOVA on the PPTD factor scores at T1, using the clinicalgroup (substance only, affective/anxiety only, comorbid andneither) as group factor, yielded a significant effect (F [12,743.7] 0 12.78; p<.001; η20 .17). This effect is considered tobe large in size (η2< .14) as suggested by accepted

Fig. 3 a Longitudinal factorialmodel of PPTD with loadingsinvariance; bMultivariate latentdifference score model ofPPTD. Negative Emot. 0Negative Emotionality; LCS 0Latent change scores

180 J Psychopathol Behav Assess (2013) 35:173–185

benchmarks in the literature. This analysis was replicatedwith T2 data, showing a lower, yet substantial effect size(F [12, 709.4] 0 4.88; p<.001; η20 .08). This reduction ineffect size could be explained by the reduction of samplesize in some clinical groups at T2 (as patterns of remissionbetween T1 and T2 were frequent). The Tukey’s post hoctests used in complement of these MANOVAs, permitted toidentify the variables for which the groups differedsignificantly.

As indicated by the descriptive statistics in Table 2, eachclinical group was associated with significant level differ-ences on the four PPTD factors scores. The main featuresof the participants in the neither group were lower levelson all four PPTD factor score. The substance only groupwas associated with moderate scores on NegativeEmotionality, Antagonism and Impulsivity, and rather lowscores on Introversion. Finally, the affective/anxiety andcomorbid groups (with affective/anxiety disorders and sub-stance dependence) displayed similar average profiles, ex-cept on the Impulsivity factor where the comorbid grouptended to show higher levels. The one-way ANOVA sug-gested that it was mostly the Negative Emotionality factorthat differentiated the best between the four clinical groups(η20 .26).

Change in DSM Diagnosis and Associated Changesin Personality Pathology Traits

The MANOVA on the LCS (representing change betweenT1 and T2), using the Trajectories of Change in the clinicalgroup (stable, remission and newly diagnosed patterns)indicated a moderate but significant effect (F [8, 504] 0

3.46; p<.001; η20 .08). As detailed in Table 3, patterns ofstability in DSM diagnoses over time (stable group) wereassociated with mean-level stability of all the PPTD factorscores. Pattern of remission was associated with a mean-level decrease in both Negative Affectivity and Antagonismfactor scores. Finally, patterns of “newly diagnosed” (nodiagnosis of substance use nor affective/anxiety at T1 fol-lowed by any diagnosis at T2) were characterized by amean-level increase in Negative Affectivity, Introversion,and Impulsivity.

Discussion

The results presented in this study offer new insightinto the underlying structure and development of PDs.Both contributions are discussed separately, then incombination, to outline the strengths and limitations ofthis study, as well as important directions for futureresearch.

Structure of the Higher-Order Pathological Personality TraitDimensions

The PPTDs identified in our study are comparable to thosepresented by Rossi and collaborators (2010) except for thefourth factor (Impulsivity) which appeared less independentthan expected. This discrepancy may be due to the nature ofour sample (women only, with over-representation of sub-stance dependence), as the dimension of impulsivity hasbeen implicated in many forms of substance use disorder(e.g., Hicks et al. 2004). As a result (or due to other factorsnot controlled in this study), the pattern of loadingsobtained here suggest an impulsivity factor that aligns morewith self-injurious impulsivity rather than the aggressiveimpulsivity that is represented in structural analyses ofMCMI-III PD scales obtained with other samples. As awhole, the pattern of inter-correlation between PPTDs isalso consistent with a broad “general maladjustment factor”suggested by Millon and collaborators (2006) from theoverall pattern of moderate to high intercorrelations be-tween MCMI-III scales scores, further confirmed byPhilippe Rushton and Irwing (2009).

However, excluding the fifth PPTD which is underrepre-sented in the MCMI-III, our results confirm that a four-factor solution best achieves the goal of simplification with-out excessive loss of information (Dyce et al. 1997;O'Connor and Dyce 1998), as suggested by the rather highamount of total variance in MCMI-III PD scores explainedby four factors. It results in a meaningful structure ofPPTDs, which represent four delineated constructs, inter-pretable independently, and associated soundly with variousclinical conditions. Indeed, beyond the relative alignment ofour PPTD constructs with four of the five higher-order

Table 2 Average differences in PPTD factor scores as a function of the clinical group and result of the univariate analyses

PPTD factor Substance only Affective/anxiety only Comorbid Neither Univariate results

M (SD) M (SD) M (SD) M (SD) F (3, 284) p η2

Negative emotionality .05 (.90) .69 (.96) .81 (.78) −.44 (.79) 38.82 <.001 .26

Antagonism .13 (.87) .31 (1.18) .29 (.89) −.20 (.89) 5.87 <.001 .06

Introversion −.19 (.66) .27 (1.26) .33 (1.17) −.15 (.83) 4.35 <.01 .04

Impulsivity .09 (.91) .20 (1.19) .72 (1.12) −.20 (.82) 7.91 <.001 .08

J Psychopathol Behav Assess (2013) 35:173–185 181

constructs of PPTD proposed for the DSM-5 (Krueger et al.2011), our results revealed coherent average profiles ofPPTD factor scores as a function of four broad clinicalgroups identified with an external DSM-based measure(substance only, affective/anxiety only, comorbid, and nei-ther), supporting their interpretability and practical rele-vance. Low scores on the four PPTD were observed in thegroup with no present DSM diagnosis, while the groupswith substance dependence (substance only and comorbid)showed similar profiles (higher scores for the comorbidgroup) with a relative “apex” on the Impulsivity domain.The affective/anxiety only and comorbid groups showed acoherent average profile with high Negative Emotionalityassociated with high Impulsivity for the comorbid group.Together, these group differences suggest that the fourcore PPTD discriminate, albeit broadly, various forms ofpsychopathology (i.e. broad quasi-syndromal diagnosticgroups).

Rank-Order and Mean-Level Stability and Change in PPTD

The question of structure stability of PPTD was examinedlongitudinally, demonstrating the robustness of the four PPTDstructure over time, and allowing the study of these constructin a developmental perspective. As pointed out above, therather long delay between the two measurement occasions inthis study (5 years) allowed an analysis of the PPTD con-structs stability and changes. However, further study of thePPTD constructs using shorter test-retest delay would allowfor estimating the PPTD scores reproducibility, and furtherstrengthening the interpretability of their stability and changeacross longer periods. Consistent with prior studies (e.g., Griloet al. 2004, 2010), the relative stability of individual differ-ences (rank-order stability) was moderate (Impulsivity) tohigh (Negative Emotionality) given the study period. Therelatively lower rank-order stability of the Impulsivity domainmay indicate a higher sensitivity to change in this domain,possibly influenced by other factors such as differential effectsof treatment or environmental influences. We also observedthat the change on the PPTDs differ considerably acrossindividuals (as evidenced by the significant variance in theLCS factor), supporting earlier results (Johnson et al. 2000;Lenzenweger et al. 2004). As a whole, changes in the

Antagonism, Introversion, and Impulsivity domains werehighly related (with change in Antagonism being on theopposite direction than the two other domains, i.e. increasesin Introversion and Impulsivity were associated withdecreases in Antagonism, and vice versa). Interestingly,change in Negative Emotionality was only related to changein Antagonism, and was independent to change inIntroversion and Impulsivity. This new set of results appearshighly informative to better understand PPTD dynamics ofchange, which could translate into a new perspective forunderstanding the dynamic underlying change in PDs in thesetting of other psychiatric conditions.

When stability is considered at the mean level, our resultssuggest an overall pattern of decrease on the PPTD scores,and a prevalent trajectory of remission with the DSM groupchange. These results are consistent with well-establishedfindings that suggest a general trend of improvement in PDsand other psychopathology over time (Johnson et al. 2000;Lenzenweger et al. 2004; Sanislow et al. 2009; Skodol et al.2005; Skodol 2008). It is worth noting here that a majorityof participants in the present study’s clinical sub-groupswere in treatment at the time of recruitment, which wasnot controlled for in this study, due to its naturalistic design(cf., Sanislow et al. 2009). Whether this general trend ofremission was due to treatment condition or not (some of thereduction may have been the result of the treatments thatmany of the participants received), what is important for thescope of the present study is that the various patterns ofchange in the DSM diagnosis were associated with mean-ingful changes in the PPTD factors scores as well, support-ing the relative intrinsic plasticity of the four PPTDstructure.

However, some of the changes observed in the PPTDfactor scores as a function of the change in the DSM diag-nosis were not significant. For example, Introversion andImpulsivity appeared less related to change in the DSMdiagnosis. It is possible that the small representation ofparticipants in some of the groups–such as the group ofparticipants acquiring a diagnosis over time (newly diag-nosed group)–may have weakened the power of the analy-ses. On the other hand, the groups (i.e. stability, remission,and newly diagnosed) may have been too broad to capturethe variety of patterns of change in the PPTD factors

Table 3 Average T1-T2 level ofchange in the PPTD factorscores as a function of the pat-tern of change in the clinicalgroup

PPTD factor Stability Remission Newly diagnosed Univariate results

M (SD) M (SD) M (SD) F (2, 255) p η2

Negative emotionality −.03 (.39) −.27 (.36) .07 (.46) 11.78 <.001 .08

Antagonism −.04 (.44) −.22 (.37) −.08 (.41) 4.86 <.01 .04

Introversion .02 (.30) .03 (.32) .15 (.33) 1.71 ns –

Impulsivity .03 (.29) .05 (.26) .12 (.20) 1.14 ns –

182 J Psychopathol Behav Assess (2013) 35:173–185

(especially for the domains apparently more sensitive tochange, such as Impulsivity).

Consistent with this interpretation, we observed highintra-group variability among the PPTD change factors (re-vealing that within each group, individuals did not change atthe same speed). It is possible that, by disentangling varioustrajectories within our “change in DSM diagnosis groups”,the results may have been clearer. For example, the patternof changes in the PPTD factor scores may certainly bedifferent for participant who acquired a diagnosis of sub-stance abuse between T1 and T2, than for those who ac-quired a diagnosis of any affective/anxiety disorders only. Inthe study presented here, both trajectories were grouped intoa single form of change labeled “newly diagnosed.” Futurestudies with larger samples should attempt to model trajec-tories using the specific diagnosis (e.g., major depression vs.anxiety disorder) rather than our broad quasi-syndromaldiagnostic groups. However, such analyses are highly de-manding in sample size and were thus not feasible in thisstudy. Nevertheless, the proposed approach to capture thedynamic of change in PPTD as a function of the change incurrent DSM diagnosis, may offer a useful agenda for futureresearch using existing longitudinal datasets with largersample sizes.

Study Limitations and Future Directions

In addition to those already described, a number of otherlimitations warrant mention. Because “peculiarity” featuresare underrepresented in the MCMI-III and other PD inven-tories, we were not able to examine the fifth PPTD de-scribed in the literature (i.e., Psychoticism). As noted byRossi and collaborators (2010) in line with the PersonalityDisorders Work Group, future research should include thisfifth PPTD in order to represent oddity or peculiarity fea-tures (e.g., Tackett et al. 2008), that is, the negative variantof the Openness domain in the FFM. Future longitudinalresearch incorporating broader assessment of PD (such asthe Personality Psychopathology-Five, PSY-5; Harkness etal. 1995) are of importance in order to elucidate all fivedomains together, their structural stability, and dynamic ofchange.

Although the sample was diverse, including clinical andnon-clinical participants with the same high-risk back-ground (high prevalence of ethnic minorities associated withlow-income), the exclusion of participants with a history ofpsychosis from this study may have reduced the generaliz-ability of the findings, with some clinical groups that areprevalent in the community being underrepresented in ouranalyses. Similarly, only women were recruited though gen-der differences in personality construct in both cross-sectional (e.g., Schmitt et al. 2008) and longitudinal studies(e.g., Klimstra et al. 2009) are often observed. For example,

Jane and colleagues (2007) identified a differential itemfunctioning in the diagnostic features of 10 personalitydisorder categories, suggesting that men and women withequivalent levels of pathology tend to endorse items ofcommon PD inventories at different rates. This points outanother limit of this study: its reliance on a self-reportmeasure which usually maps only partially onto PD-relatedbehavior (e.g., Rogers et al. 1999), although this format iscommonly used to capture dimensional “trait models” ofmaladaptive personality (Westen and Muderrisoglu 2006).

Finally, we must note several limitations of the variousgrouping strategies used in this study related to DSM diag-nostic groups and change in diagnostic groups, and theirderivation from categorical diagnostic systems. First, al-though the DIS-IV used in this study to derived DSMdiagnosis is a measure that proved robust levels of inter-diagnostician reliability in several studies, this reliabilityindicator was not evaluated here, precluding the estimationof “interviewer effects” that may account for error in diag-nosis classification. Second, clinical diagnoses stabilitydepends at least partially on the method of assessment, andstructured interviews have yielded lower stability (Samuel etal. 2011). Third, the use of thresholds often arbitrary for acategorical diagnosis (Grilo et al. 2004) are likely to lead tomeasurement error in diagnostic and cumulated errors inmeasurement of change. Together, these limitations shouldbe considered for future studies involving larger and morediverse samples, using multiple PDs and other clinical di-agnoses techniques to assess psychopathology.

Conclusion

This study extends prior efforts to derive an empirical struc-ture of PPTD for inclusion in official nosologies, by provid-ing a unique set of longitudinal evidence of four higher-order PPTD underlying PDs and other psychiatric condi-tions. This model does not represent the “true” structure ofPDs (O'Connor 2005), but at least represents an increasinglyconsistent framework to understand pathological personalitytrait structure and dynamic of change. As a result, we wereable to achieve the first developmental analyses using thesehiger-order constructs of personality pathology traits, whichprovide an innovative understanding of PD developmentand dynamic of change. Ultimately, a better knowledge ofthis dynamic can prove useful to trigger change in themaladaptive personality component of psychopathology.To that end, we recommend extending the present researchusing longitudinal examination of PPTDs, particularly in aneffort to explore the fifth domain of Psychoticism, andexamine further relevance of a higher-order, general factorof pathological personality (cf. Philippe Rushton and Irwing2009). Finally, future research should focus on the

J Psychopathol Behav Assess (2013) 35:173–185 183

developmental relation between PPTD and other domains ofthe psychological functioning, in an effort to better understandthe mechanisms by which change in PPTD affects—and isaffected by—other features of psychological functioning.

Acknowledgment Wewould like to thank Theresa Babuscio, GabrielleSanta-Donato, Julia Chafkin, Jeremy Rothstein and Pamela Brown fortheir help with data collection, data management, and editorial assistancein the preparation of this manuscript. We also thank Gina Rossi for herdata-analytic advices. Preparation of this manuscript was supported byDA010726, DA11498 and DA14385 (PI: Luthar).

References

American Psychiatric Association. (2000). Diagnostic and statisticalmanual of mental disorders: DSM-IV-TR. American PsychiatricPublishing, Inc.

Clark, L. A. (2007). Assessment and diagnosis of personalitydisorder: perennial issues and an emerging reconceptualiza-tion. Annual Review of Psychology, 58(1), 227–257.doi:10.1146/annurev.psych.57.102904.190200.

Cuevas, L., García, L. F., Aluja, A., & García, Ó. (2008). Exploratoryand confirmatory factorial structure of the MCMI-III personalitydisorders: overlapping versus non-overlapping scales. The Euro-pean Journal of Psychiatry, 22(2), 59–68.

Dascalu, M., Compton, W., Horton, J., & Cottler, L. (2001). Validity ofDIS-IV in diagnosing depression and other psychiatric disordersamong substance users. Drug and Alchohol Dependence, 63, S37.

Durbin, C. E., & Klein, D. N. (2006). Ten-year stability of personalitydisorders among outpatients with mood disorders. Journal ofAbnormal Psychology, 115(1), 75–84.

Dyce, J. A., O'Connor, B. P., Parkins, S. Y., & Janzen, H. L. (1997).Correlational structure of the MCMI-III personality disorderscales and comparisons with other data sets. Journal of Person-ality Assessment, 69(3), 568–582.

Eaton, N. R., Krueger, R. F., South, S. C., Simms, L. J., & Clark, L. A.(2011). Contrasting prototypes and dimensions in the classifica-tion of personality pathology: evidence that dimensions, but notprototypes, are robust. Psychological Medicine, 41, 1151–1163.doi:10.1017/S0033291710001650.

Ferguson, C. J. (2010). A meta-analysis of normal and disorderedpersonality across the life span. Journal of Personality and SocialPsychology, 98(4), 659–667.

Ferrer, E., & McArdle, J. J. (2010). Longitudinal modeling ofdevelopmental changes in psychological research. CurrentDirections in Psychological Science, 19(3), 149–154.doi:10.1177/0963721410370300.

Grilo, C. M., Shea, M. T., Sanislow, C. A., Skodol, A. E., Gunderson,J. G., Stout, R. L., et al. (2004). Two-year stability and change ofschizotypal, borderline, avoidant, and obsessive-compulsive per-sonality disorders. Journal of Consulting and Clinical Psycholo-gy, 72(5), 767–775. doi:10.1037/0022-006X.72.5.767.

Grilo, C., Stout, R., Markowitz, J., Sanislow, C., Ansell, E., Skodol,A., et al. (2010). Personality disorders predict relapse after remis-sion from an episode of major depressive disorder: a 6-yearprospective study. The Journal of Clinical Psychiatry, 71(12),1629–1635. doi:10.4088/JCP.08m04200gre.

Harkness, A. R., McNulty, J. L., & Ben-Porath, Y. S. (1995). Thepersonality psychopathology five (PSY-5): constructs and MMPI-2 scales. Psychological Assessment, 7(1), 104–114. doi:10.1037/1040-3590.7.1.104.

Hasin, D. S., & Grant, B. F. (1987). Psychiatric diagnosis of patientswith substance abuse problems: a comparison of two procedures,the DIS and the SADS-L: alcoholism, drug abuse/dependence,anxiety disorders and antisocial personality disorder. Journal ofPsychiatric Research, 21(1), 7–22. doi:10.1016/0022-3956(87)90005-7.

Hesse, M., Guldager, S., & Holm Linneberg, I. (2012). Convergentvalidity of MCMI-III clinical syndrome scales. British Journal ofClinical Psychology, 51(2), 172–184. doi:10.1111/j.2044-8260.2011.02019.x.

Hicks, B., Markon, K., Patrick, C., Krueger, R., & Newman, J. (2004).Identifying psychopathy subtypes on the basis of personalitystructure. Psychological Assessment, 16(3), 276–288.doi:10.1037/1040-3590.16.3.276.

Jahng, S., Trull, T. J., Wood, P. K., Tragesser, S. L., Tomko, R., Grant,J. D., et al. (2011). Distinguishing general and specific personalitydisorder features and implications for substance dependencecomorbidity. Journal of Abnormal Psychology, 120(3), 656–669.doi:10.1037/a0023539.

Jane, J. S., Oltmanns, T. F., South, S. C., & Turkheimer, E. (2007).Gender bias in diagnostic criteria for personality disorders: anitem response theory analysis. Journal of Abnormal Psychology,116(1), 166–175. doi:10.1037/0021-843X.116.1.166.

Johnson, J. G., Cohen, P., Kasen, S., Skodol, A. E., Hamagami, F., &Brook, J. S. (2000). Age-related change in personality disorder traitlevels between early adolescence and adulthood: a community-based longitudinal investigation. Acta Psychiatrica Scandinavica,102(4), 265–275. doi:10.1034/j.1600-0447.2000.102004265.x.

Klimstra, T. A., Hale, W. W., Raaijmakers, Q. A., Branje, S. J., &Meeus, W. H. (2009). Maturation of personality in adolescence.Journal of Personality and Social Psychology, 96(4), 898–912.doi:10.1037/a0014746.

Kline, R. B. (2010). Principles and practice of structural equationmodeling (3rd ed.). New York: The Guilford Press.

Krueger, R. F., Skodol, A. E., Livesley, W. J., Shrout, P. E., & Huang,Y. (2007). Synthesizing dimensional and categorical approachesto personality disorders: refining the research agenda for DSM-Vaxis II. International Journal of Methods in Psychiatric Research,16(1), 65–73.

Krueger, R. F., Eaton, N. R., Clark, L. A., Watson, D., Markon, K. E.,Derringer, J., et al. (2011). Deriving an empirical structure ofpersonality pathology for DSM-5. Journal of Personality Disor-ders, 25(2), 170–191.

Krueger, R. F., & Tackett, J. L. (2005). Progress and innovation:personality disorders and the vanguard of psychopathology re-search. Journal of Personality Disorders, 19(5), 540–546.

Lenzenweger, M. F. (1999). Stability and change in personality disor-der features: the longitudinal study of personality disorders.Archives of General Psychiatry, 56(11), 1009–1015.

Lenzenweger, M. F. (2006). The longitudinal study of personalitydisorders: history, design considerations, and initial findings.Journal of Personality Disorders, 20(6), 645–670.

Lenzenweger, M. F. (2010). Current status of the scientific study of thepersonality disorders: an overview of epidemiological, longitudi-nal, experimental psychopathology, and neurobehavioral perspec-tives. Journal of the American Psychoanalytic Association, 58(4),741–778.

Lenzenweger, M. F., Johnson, M. D., & Willett, J. B. (2004). Individ-ual growth curve analysis illuminates stability and change inpersonality disorder features: the longitudinal study of personalitydisorders. Archives of General Psychiatry, 61(10), 1015–1024.doi:10.1001/archpsyc.61.10.1015.

Lenzenweger, M. F., & Willett, J. B. (2007). Predicting individualchange in personality disorder features by simultaneous individ-ual change in personality dimensions linked to neurobehavioralsystems: the longitudinal study of personality disorders. Journal

184 J Psychopathol Behav Assess (2013) 35:173–185

of Abnormal Psychology, 116(4), 684–700. doi:10.1037/0021-843X.116.4.684.

Livesley, W. J. (2007). A framework for integrating dimensional andcategorical classifications of personality disorder. Journal of Per-sonality Disorders, 21(2), 199–224.

Luthar, S. S., & Sexton, C. C. (2007). Maternal drug abuse versus maternaldepression: vulnerability and resilience among school-age and adoles-cent offspring. Development and Psychopathology, 19(1), 205–225.

Lynam, D. R. (in press). Assessment of maladaptive variants of fivefactor model traits. Journal of Personality. doi:10.1111/j.1467-6494-2012.00775.x.

Malgady, R. G., Rogler, L. H., & Tryon, W. W. (1992). Issues of validityin the diagnostic interview schedule. Journal of Psychiatric Re-search, 26(1), 59–67. doi:10.1016/0022-3956(92)90016-H.

Markon, K. E., Krueger, R. F., & Watson, D. (2005). Delineating thestructure of normal and abnormal personality: an integrative hier-archical approach. Journal of Personality and Social Psychology,88(1), 139–157.

McArdle, J., & Nesselroade, J. R. (1994). Using multivariate data tostructure developmental change. In S. Cohen & H. Reese (Eds.),Life-span developmental psychology: Methodological contributions(pp. 223–267). Hillsdale: Lawrence Erlbaum Associates, Inc.

McArdle, J. (2009). Latent variable modeling of differences andchanges with longitudinal data. Annual Review of Psychology,60(1), 577–605.

Millon, T. (1983). Millon clinical multiaxial inventory. Minneapolis:National Computer Systems.

Millon, T., Millon, C., Davis, R., & Grossman, S. (2006). Millonclinical multiaxial inventory-III manual (3rd ed.). Minneapolis:Pearson Assessments.

Mullins-Sweatt, S. N., & Widiger, T. A. (2007). Millon’s dimensionalmodel of personality disorders: a comparative study. Journal ofPersonality Disorders, 21(1), 42–57.

Nestadt, G., Di, C., Samuels, J. F., Bienvenu, O. J., Reti, I. M., Costa,P., et al. (2010). The stability of DSM personality disorders over12 to 18 years. Journal of Psychiatric Research, 44(1), 1–7.doi:10.1016/j.jpsychires.2009.06.009.

O'Connor, B. P., & Dyce, J. A. (1998). A test of models of personalitydisorder configuration. Journal of Abnormal Psychology, 107(1),3–16. doi:10.1037/0021-843X.107.1.3.

O'Connor, B. P. (2005). A search for consensus on the dimensionalstructure of personality disorders. Journal of Clinical Psychology,61(3), 323–345. doi:10.1002/jclp. 20017.

Philippe Rushton, J., & Irwing, P. (2009). A general factor of personalityin the millon clinical multiaxial inventory-III, the dimensional as-sessment of personality pathology, and the personality assessmentinventory. Journal of Research in Personality, 43(6), 1091–1095.

Robins, L., Cottler, L., Bucholz, K., Compton, W., North, C., &Rourke, K. (2000). Diagnostic interview schedule for DSM-IV(DIS-IV). St. Louis: Washington University.

Rogers, R., Salekin, R. T., & Sewell, K. W. (1999). Validation of themillon clinical multiaxial inventory for axis II disorders: does itmeet the daubert standard? Law and Human Behavior, 23(4),425–443. doi:10.1023/A:1022360031491.

Rossi, G., Elklit, A., & Simonsen, E. (2010). Empirical evidence for afour factor framework of personality disorder organization: multi-group confirmatory factor analysis of the millon clinical multiaxialinventory-III personality disorder scales across Belgian and Danishdata samples. Journal of Personality Disorders, 24(1), 128–150.

Rossi, G., van der Ark, L. A., & Sloore, H. (2007). Factor analysis ofthe Dutch-language version of the MCMI-III. Journal of Person-ality Assessment, 88(2), 144–157.

Samuel, D. B., & Gore, W. L. (in press). Maladaptive variants ofconscientiousness and agreeableness. Journal of Personality.doi:10.1111/j.1467-6494-2012.00770.x.

Samuel, D. B., & Widiger, T. A. (2008). A meta-analytic review of therelationships between the five-factor model and DSM-IV-TR per-sonality disorders: a facet level analysis. Clinical PsychologyReview, 28(8), 1326–1342.

Samuel, D. B., Hopwood, C. J., Ansell, E. B., Morey, L. C., Sanislow,C. A., Markowitz, J. C., et al. (2011). Comparing the temporalstability of self-report and interview assessed personality disorder.Journal of Abnormal Psychology, 120(3), 670–680. doi:10.1037/a0022647.

Sanislow, C. A., Little, T. D., Ansell, E. B., Grilo, C. M., Daversa, M.,Markowitz, J. C., et al. (2009). Ten-year stability and latentstructure of the DSM-IV schizotypal, borderline, avoidant, andobsessive-compulsive personality disorders. Journal of AbnormalPsychology, 118(3), 507–519.

Schmitt, D. P., Realo, A., Voracek, M., & Allik, J. (2008). Why can’t aman be more like a woman? sex differences in big five personalitytraits across 55 cultures. Journal of Personality and Social Psy-chology, 94(1), 168–182. doi:10.1037/0022-3514.94.1.168.

Sheets, E., & Craighead, W. E. (2007). Toward an empirically basedclassification of personality pathology.Clinical Psychology: Scienceand Practice, 14(2), 77–93. doi:10.1111/j.1468-2850.2007.00065.x.

Skodol, A. E. (2008). Longitudinal course and outcome of personalitydisorders. Psychiatric Clinics of North America, 31(3), 495–503.doi:10.1016/j.psc.2008.03.010.

Skodol, A. E., Gunderson, J. G., Shea, M. T., McGlashan, T. H.,Morey, L. C., Sanislow, C. A., et al. (2005). The collaborativelongitudinal personality disorders study (CLPS): overview andimplications. Journal of Personality Disorders, 19(5), 487–504.

Solomon, S. R., & Sawilowsky, S. S. (2009). Impact of rank-basednormalizing transformations on the accuracy of test scores. Jour-nal of Modern Applied Statistical Methods, 9(2), 448–462.

Tackett, J. L., Silberschmidt, A. L., Krueger, R. F., & Sponheim, S. R.(2008). A dimensional model of personality disorder: incorporat-ing DSM cluster A characteristics. Journal of Abnormal Psychol-ogy, 117(2), 454–459.

Warner, M. B., Morey, L. C., Finch, J. F., Gunderson, J. G., Skodol, A.E., Sanislow, C. A., et al. (2004). The longitudinal relationship ofpersonality traits and disorders. Journal of Abnormal Psychology,113(2), 217–227. doi:10.1037/0021-843X.113.2.217.

Watson, D., Clark, L. A., & Chmielewski, M. (2008). Structures ofpersonality and their relevance to psychopathology: II. Furtherarticulation of a comprehensive unified trait structure. Journal ofPersonality, 76(6), 1545–1586.

Westen, D., & Muderrisoglu, S. (2006). Clinical assessment of patho-logical personality traits. The American Journal of Psychiatry,163(7), 1285–1287.

Widiger, T. A., & Costa Jr, P. T. (In press). Integrating normal andabnormal personality structure: the five factor model. Journal ofPersonality. doi:10.1111/j.1467-6494-2012.00776.x.

Widiger, T. A., & Samuel, D. B. (2005). Diagnostic categories ordimensions? a question for the diagnostic and statistical manualof mental disorders-V. Journal of Abnormal Psychology, 114(4),494–504.

Widiger, T. A., & Simonsen, E. (2005). Alternative dimensional mod-els of personality disorder: finding a common ground. Journal ofPersonality Disorders, 19(2), 110–130.

Wright, A. G. C. (2011). Qualitative and quantitative distinctions inpersonality disorder. Journal of Personality Assessment, 93(4),370–379.

J Psychopathol Behav Assess (2013) 35:173–185 185