Show Summary Details
Page of

(p. 122) Personality Dysfunction and Trait Extremity: Conceptually, but Not Empirically Distinct? 

(p. 122) Personality Dysfunction and Trait Extremity: Conceptually, but Not Empirically Distinct?
Chapter:
(p. 122) Personality Dysfunction and Trait Extremity: Conceptually, but Not Empirically Distinct?
Author(s):

Lee Anna Clark

, Elizabeth J. Daly

, Stephanie Larew

, Hallie Nuzum

, Thomas Kingsbury

, Jaime L. Shapiro

, Xia Allen

, and Eunyoe Ro

DOI:
10.1093/med-psych/9780190227074.003.0006
Page of

date: 19 March 2019

Despite their many well-documented problems (e.g., see Widiger & Simonsen, 2005), the personality disorder (PD) criteria in the main text (Section II) of the Diagnostic and Statistical Manual of Mental Disorders, fifth edition (DSM-5-II; APA, 2013; hereafter “PD-II”) are identical to those in DSM-IV (American Psychiatric Association [APA], 1994). The PD-II diagnostic types are considered “qualitatively distinct clinical syndromes” and are defined conceptually as personality traits that “deviate markedly from the expectations of the individual’s culture” and “are manifested in at least two of the following areas: cognition, affectivity, interpersonal functioning, or impulse control.” Personality traits, in turn, are defined as “enduring patterns of perceiving, relating to, and thinking about the environment and oneself that are exhibited in a wide range of social and personal contexts” or, for short, “enduring patterns of inner experience and behavior” (APA, 2013, p. 645).

Criterion A of each of the 10 PD-II diagnostic types first describes a characteristic pattern and then lists seven-to-nine specific indicators of the pattern, some minimum number of which must be exhibited for a diagnosis. For some PD-II types, however, two distinct patterns comprise Criterion A, and there is no clear mandate to ensure that indicators of both are present. For example, the Criterion A description of schizotypal PD is “social and interpersonal deficits marked by acute discomfort with, and reduced capacity for, close relationships as well as by cognitive or perceptual distortions and eccentricities of behavior” (APA, 2013, p. 655). Of the nine specific indicators, only two (arguably three) (p. 123) specifically represent social or interpersonal deficits, but standard procedure is to make the diagnosis as long as the minimum of five indicators is present, regardless of whether any indicator of social or interpersonal deficits is manifested. Further, there is not always a clear correspondence between the Criterion A pattern description and its purported specific indicators. For example, the Criterion A description for borderline personality disorder (BPD) is “instability of interpersonal relationships, self-image, and affects, and marked impulsivity” (APA, 2013, p. 663), but it is unclear how “chronic feelings of emptiness” reflects this pattern. Thus, although the PD-II types are conceptualized in terms of traits, they typically are diagnosed via specific indicators that relate inconsistently to the defining trait patterns.

Alternative Model of Personality Disorder

In contrast, the alternative model of PD in DSM-5, Section III (“Emerging Models and Measures”), hereafter PD-III, defines PD as “moderate or greater impairment in personality (self/interpersonal) functioning”1 and “one or more pathological personality traits” (APA, 2013, p. 761). Moreover, PD-III diagnosis is directly aligned with this definition. Specifically, PD-III provides descriptors for rating impairment level in four areas of personality functioning (identity and self-direction for self functioning, and intimacy and empathy for interpersonal functioning) on a 5-point scale. Unlike the implied normal-pathological dichotomy of PD-II, these five impairment levels are understood to represent a continuous distribution of functioning from 0 (no impairment) to 4 (extreme impairment).

For assessing personality traits, PD-III provides definitions of 25 trait facets, organized in five trait domains—Negative Affectivity, Detachment, Antagonism, Disinhibition, and Psychoticism—all of which are also understood to represent continuous distributions ranging from low to high levels. Although PD-III offers no specific guidance for assessing trait elevation, it does provide a set of self-report, other-report, and clinician-rating scales for this purpose (e.g., Krueger, Derringer, Markon, Watson, & Skodol, 2012; Markon, Quilty, Bagby, & Krueger, 2013) that use a 4-point (0–3) scale with ratings of two or higher considered pathological.

Much has been made of the fact that PD-III distinguishes personality dysfunction and traits conceptually (e.g., Livesley, 2012; Wakefield, 2013), based on the argument that trait extremity per se is not necessarily pathological (Livesley & Jang, 2005). However, there has been limited discussion of the fact that PD-III Criterion B requires not one or more extreme traits, but “one or more pathological traits” (APA, 2013, p. 761). Nonetheless, given that every trait researched to date has been shown to be continuously distributed in the population, “pathological traits” presumably doesn’t mean that the traits themselves are pathological (because, if that were the case, then it would be possible for an individual to exhibit a normal level of a pathological trait, which is nonsensical). Rather, what is meant by “pathological traits” is that the person’s trait level or expression is pathological. However, the term expression is conceptually problematic, because the phrase (p. 124) pathological trait expression is confounded with personality-functioning impairment. Thus, more neutral terms such as level, elevation, or extremity are preferable, although their use then requires empirical determination of one or more cut points for designating a given trait level as pathological (see Finn, 1982, for a discussion of the pitfalls of using single, fixed cut points for this determination and for clinical decision making).

A further difficulty with the current PD-III model is that there is clear descriptive overlap between PD-III personality impairment and “pathological traits.” To give but two examples: (1) Criterion A identity impairment is reflected in Criterion B trait-domain Negative Affectivity, specifically its emotional-lability facet, and (2) Criterion A impaired intimacy is reflected in Criterion B trait-domain Detachment, particularly its intimacy-avoidance facet (see Table 6.1 for details). This overlap also raises measurement issues: Shared content increases scale intercorrelations and confounds efforts to investigate the distinction between trait extremity and personality impairment. Thus, although distinguishing trait extremity and personality impairment is attractive conceptually, it is difficult empirically, and would be impossible if they are linked inherently. If distinguishing trait extremity and personality impairment is simply empirically difficult, then there is much measure-development work to do, whereas if their linkage is inherent, that is, if trait extremity is intrinsically pathological, then separate assessment of personality dysfunction and traits would be neither necessary nor desirable. It is possible that the only way to adjudicate these issues is to attempt their separate assessment and either succeed or fail in the effort.

Table 6.1 Content Overlap of Personality Dysfunction and Trait Definitions with Examples in Level of Personality Functioning Scale Descriptors

Personality Functioning Domain and Definition

Trait Domain/Facet and Definition

Level of Personality Functioning Scale

Identity

  • Capacity for, and ability to regulate, a range of emotional experience

Emotional lability (NA)

  • Instability of emotional experiences and mood

  • Emotions that are easily aroused, intense, and/or out-of-proportion to events and circumstances

1-Strong emotions may be distressing, associated with a restriction in range of emotional experience.

3-Emotions may be rapidly shifting or a chronic, unwavering feeling of despair.

Self-direction

  • Pursuit of coherent and meaningful short-term and life goals

  • Utilization of constructive and pro-social internal standards of behavior

Impulsivity (DIS)

  • Acting on a momentary basis without a plan or consideration of outcomes

  • Difficulty establishing and following plans.

3-Has difficulty establishing and/or achieving personal goals.

4-Has poor differentiation of thoughts from actions, so goal-setting ability is severely compromised, with unrealistic or incoherent goals.

Empathy

  • Comprehension and appreciation of others’ experiences and motivations

  • Understanding the effects of one’s own behavior on others

Callousness (ANT)

  • Lack of concern for the feelings or problems of others

  • Lack of guilt or remorse about the negative or harmful effects of one’s actions on others

2-Is generally unaware of or unconcerned about the effect of own behavior on others; or has an unrealistic appraisal of own effect.

Intimacy

  • Depth and duration of connection with others;

  • Desire and capacity for closeness.

Intimacy avoidance (DET)

  • Avoidance of close or romantic relationships, interpersonal attachments, and intimate sexual relationships.

4-Desire for affiliation is limited because of profound disinterest or expectation of harm. Engagement with others is detached, disorganized, or consistently negative.

note: NA = negative affectivity; DIS = disinhibition; ANT = antagonism; DET = detachment.

Measuring Personality Functioning and Traits

Considering self-report, trait measurement has a long history, and there are many well-established measures with strong psychometric properties. In contrast, assessment of personality dysfunction is a relatively new enterprise. To address this issue, the field’s inclination likely will be to develop personality-functioning measures that minimize content overlap with established trait measures. However, it also is worth attempting to modify existing—or to develop new—trait measures to reflect trait extremity more purely, without the “contamination” of personality dysfunction.

Turning to clinician ratings, there are few established measures of either personality dysfunction or traits, which represents an opportunity to try to distinguish severity of personality functioning impairment from trait levels through new scale-development projects. Before embarking on any measure-modification or development projects, however, we need to understand better how personality functioning and traits are interrelated. To this end, Clark and Ro (2014) analyzed interrelations among multiple self-report measures of personality dysfunction and traits, along with other measures typically included in the concept of functioning.

They first examined interrelations among the functioning measures and replicated the three dimensions reported by Ro and Clark (2013) of poor versus good social/interpersonal functioning, self-pathology versus well-being/life (p. 125) (p. 126) satisfaction, and basic functioning (e.g., self-care, mobility, poor health/environment). They then factored the trait measures and found a hierarchical structure which, at the five-factor level, reflected four of the familiar “Big Five” factors—neuroticism/negative affectivity (N/NA), (low) sociability, disinhibition, and (dis)agreeableness—plus a fifth factor they termed “rigid goal engagement.” This factor reflected obsessive-compulsive (OC) traits and can be identified as the dimension that in the ICD-11 (World Health Organization, 2018) section on personality disorders is termed “anankastia” (i.e., the noun form of anankastic, the British adjective for describing OCPD traits). Finally, when all functioning and trait measures were co-factored, a hierarchical structure again emerged which, at the five-factor level, included two factors that blended personality and functioning—internalizing (N/NA converged with the self-pathology vs. well-being/life satisfaction factor) and externalizing (social/interpersonal dysfunction was joined by traits low sociability and disagreeableness), whereas the last three factors reflected either just personality traits (disinhibition and rigid goal engagement/anankastia, respectively) or just functioning (basic functioning).

Thus, the analyses indicated that personality dysfunction overlapped with some, but not all, trait dimensions. Specifically, more behavioral traits—disinhibition and rigid goal engagement/anankastia—were largely independent of personality functioning, whereas affective and interpersonal personality traits—N/NA, (low) sociability, and antagonism–agreeableness—were intertwined with personality dysfunction, with the traditional functioning domain of quality-of-life/satisfaction forming the opposite end of the broad N/NA–self-pathology factor.

Overview

In this chapter, we report on a second study with, again, an extensive self-report battery of personality functioning, trait, and other functioning measures, partially overlapping with that of the first study. In addition, we used a single semistructured interview to rate both the PD-II and PD-III criteria. We examine how, and how well, personality functioning and trait measures separately (via raw correlations) and jointly (via multiple regressions) predict dimensional ratings of the PD-II types—specifically, aggregated criterion ratings—which blend personality functioning and traits. As predictors, we used a set of traits that were hypothesized a priori (specifically, in 2011 by the first author [LAC]2, before trait-facet assignments were made for the PD-III types) to relate to each of the PD-II types’ criteria (shown in Table 6.4). For example, the avoidant PD criterion “views self as socially inept, personally unappealing, or inferior to others” (APA, 1994, p. 665) was mapped to depressivity (“feelings of inferior self-worth,” APA, 2013, p. 779), whereas the narcissistic PD criterion “has a grandiose sense of self-importance” (APA, 1994, p. 661) was mapped to grandiosity (“believing that one is superior to others and deserves special treatment . . . feelings of entitlement”; APA, 2013, p. 780). All trait facets that mapped to any criterion were (p. 127) included as predictors for a given PD type. We detail our analytic strategy at the end of the Method section.

Table 6.4 Correlations and Prediction of Interview-Rated Personality Disorder From Hypothesized Trait Facets and Personality Dysfunction

CRF (Interview)

PID-5 (Self-report)

Facet

r

R

Sig. Beta

r

R

Sig. Beta

sidp-iv antisocial pd

Irresponsibility

.78

1.41

.44

1.70

Deceitfulness

.70

.53

.40

Impulsivity

.70

.72

.44

Manipulativeness

.65

.60

.34

Callousness

.60

.39

Risk taking

.54

.69

.43

1.68

Hostility

.43

.39

.26

Self-dysfunction

.45

.21

Interpersonal-dysfunction

.42

.29

.51

Adj. R: Final Model

.87

.53

sidp-iv avoidant pd

Withdrawal

.60

1.69

.49

1.73

Anxiousness

.60

1.53

.37

Depressivity

.53

.85

.44

Intimacy avoidance

.35

.19

Self-dysfunction

.40

.51

1.66

Interpersonal-dysfunction

.28

.47

Adj. R: Final Model

.72

.56

sidp-iv borderline pd

Emotional lability

.70

1.61

.57

1.60

Impulsivity

.59

1.43

.47

1.29

Hostility

.52

.52

.45

Depressivity

.50

.47

.53

Separation insecurity

.47

.43

.46

Suspiciousness

.39

.45

Cognitive and perceptual dysregulation

.37

.39

.47

Self-dysfunction

.69

1.00

.57

1.39

Interpersonal-dysfunction

.49

.39

Adj. R: Final Model

.86

.65

sidp-iv narcissistic pd

Grandiosity

.78

2.60

.40

1.34

Manipulativeness

.58

1.07

.35

1.53

Attention seeking

.55

.78

.42

Callousness

.54

.22

Self-dysfunction

.39

.07

Interpersonal-dysfunction

.52

.73

.15

.49

Adj. R: Final Model

.86

.47

sidp-iv obsessive-compulsive pd

Perseveration

.66

2.05

.22

Rigid perfectionism

.58

1.34

.41

2.09

Intimacy avoidance

.09

.12

Restricted affectivity

–.03

.15

Self-dysfunction

.21

.07

Interpersonal-dysfunction

.24

.11

Adj. R: Final Model

.72

.41

sidp-iv schizotypal pd

Unusual beliefs and experiences

.66

1.22

.42

1.38

Cognitive and perceptual dysregulation

.60

.95

.30

Eccentricity

.57

.85

.23

Suspiciousness

.53

.43

.29

Withdrawal

.46

.48

.36

Intimacy avoidance

.42

.33

.35

.67

Restricted affectivity

.24

.19

Self-dysfunction

.50

.26

Interpersonal-dysfunction

.54

.38

.71

Adj. R: Final Model

.81

.52

sidp-iv dependent pd

Separation Insecurity

.76

2.20

.44

1.12a

Submissiveness

.68

1.83

.36

.90a

Anxiousness

.38

.42

Depressivity

.37

.37

.—

Self-dysfunction

.45

.41

.69a

Interpersonal-dysfunction

.25

.09

a

Adj. R: Final Model

.82

.52

sidp-iv histrionic pd

Attention seeking

.79

2.84

.51

1.85

Emotional lability

.40

.68

.29

.74

Self-dysfunction

.28

.14

Interpersonal-dysfunction

.28

-.03

Adj. R: Final Model

.82

.54

sidp-iv paranoid pd

Suspiciousness

.78

2.19

.53

2.14

Hostility

.54

.52

.36

Withdrawal

.39

.33

.31

Unusual beliefs and experiences

.42

.23

Self-dysfunction

.53

.46

.27

Interpersonal-dysfunction

.58

.46

1.07

Adj. R: Final Model

.81

.54

sidp-iv schizoid pd

Withdrawal

.70

1.11

.47

.80

Intimacy avoidance

.64

.61

.46

.96

Restricted affectivity

.55

1.33

.28

Anhedonia

.48

.26

Self-dysfunction

.32

.21

Interpersonal-dysfunction

.50

.38

.44

.45

Adj. R: Final Model

.81

.54

note: Beta weights are from the final equation.

a Including interpersonal dysfunction, beta weights are 1.07, .74, 1.08, –.61, respectively.

CRF = Clinicians’ Rating Form; PID-5 = Personality Inventory for DSM-5 (Krueger et al., 2012); Sig. = significant; r = correlation; R = multiple correlation; — = nonsignificant; Adj. = adjusted; Dysfx = dysfunction.

Method

The study method and sample from which this subset was drawn were described in Clark et al. (2015). We refer readers there for more detailed descriptions.

Participants and Procedure

Participants were 164 “high-risk” community adults and 135 psychiatric outpatients. The community adults were screened for two or more positive responses (the recommended cut-point) on the Iowa Personality Disorder Screen (IPDS), which has good sensitivity and specificity in identifying PD (Langbehn et al., 1999). Unless otherwise noted, all reported analyses are based on these 299 participants.

Patients were referred primarily from a community mental health center and by local practitioners; a small minority were recruited using listservs, newsletters, and mass emails sent to University of Notre Dame staff, faculty, and graduate students, as well as by word of mouth, as long as we could verify current mental health patient status. Hereafter, we call these the “high-risk” and “patient” subsamples.

Demographics

Mean age was 48.0 years (SD = 12.4; range = 18–84); most (60%) participants were female; 27% were racial/ethnic minorities, of which most (70%) were Black/African American. A minority (35%) was employed and 26% were on disability; 51% had an annual family income below $20,000, and only 22% over $50,000. Approximately one third of participants each were married or living with a partner (38.5%), divorced/separated or widowed (34.5%), or single/never married (27%). The subsamples differed significantly in sex, age, racial/ethnic composition, employment status, income level, and relationship status, but not average educational level. Briefly, the patient subsample had proportionately fewer women, was older and more racially diverse, was more likely to be on disability (vs. employed), had lower average income, and was more likely to be single/never married, whereas the high-risk subsample was more likely to be married or living with a partner.

Procedure

Participants came to our research facility and gave written informed consent before beginning the study. Most completed the questionnaires alone on a computer; the rest were assisted by a team member. Participation required ~6 hours; interviews were interspersed with computer sections to maintain interest and (p. 128) reduce fatigue. All procedures were approved by the University of Notre Dame Institutional Review Board.

Self-Report Measures

Personality Functioning

We used three measures that had developed relatively recently at the time of data collection (2012-early 2014) to assess personality functioning. We standardized all scales to put them on the same metric and then averaged the self- and interpersonal-pathology scales from each measure, respectively, to derive overall measures of self and interpersonal pathology.

General Assessment of Personality Disorder

For the General Assessment of Personality Disorder (GAPD; Livesley, 2010), we used a 15-item self-pathology (α‎ = .92) and an 11-item interpersonal-pathology scale (α‎ = .90) derived from the GAPD via replicated factor analysis. Respondents rate items considering how they “usually are, think, feel, believe, or act” using a 5-point Likert-type scale (very unlike me—very like me).

Measure of Disordered Personality Functioning

For the Measure of Disordered Personality Functioning (MDPF; Parker et al., 2004), Parker’s non-coping (self) and non-cooperativeness (interpersonal) factors were assessed with 11-item (α‎ = .88) and 12-item (α‎ = .88) scales, respectively. The MDPF uses a 4-point Likert-type format (definitely false—definitely true) and a general time frame.

Severity Indices of Personality Problems–Short Form

For the Severity Indices of Personality Problems–Short Form (SIPP; Verheul et al., 2008), we used two scales—Identity (9 items, α‎ = .89) and Relationships (11 items, α‎ = .85)—developed via replicated factor analysis to assess self and interpersonal functioning, respectively. The SIPP uses a 4-point Likert scale (fully agree—fully disagree) with a past-3-months time frame.

Personality Traits

The Personality Inventory for DSM-5 (PID-5; Krueger et al., 2012) has 220 items rated using a 4-point scale from 0 (very false or often false) to 3 (very true or often true). It assesses 25 trait facets organized in five domains and was developed in conjunction with the PD-III model. Mean α‎ = .85 (range = .71 [callousness] to .95 [cognitive-and-perceptual dysregulation]).

Interview-Based Measures

We administered the Structured Interview for DSM-IV Personality (SIDP-IV; Pfohl, Blum, & Zimmerman, 1997) and from it both scored the 10 PD-II diagnoses (p. 129) and PD-III Criteria A (personality impairment) and B (pathological personality traits). The PD-II criteria were scored on a 4-point Likert-type scale (0 = no to minimal evidence of the criterion to 3 = prominent personality feature). Criterion scores were aggregated to yield a dimensional score for each PD-II diagnosis.

Following standard administration and earlier described scoring of the SIDP, the interviewer also provided ratings of the PD-III Criterion A and Criterion B. To rate PD-III Criterion A, we used the PD-III Levels of Personality Functioning Scale (LPFS), which details five impairment levels (0 = little or no impairment to 4 = extreme impairment) for the PD-III components of self-pathology (identity and self-direction) and interpersonal pathology (empathy and intimacy). These then were aggregated to yield self-pathology, interpersonal pathology, and overall personality-dysfunction scores. To rate PD-III Criterion B, we used the Clinicians’ Rating Form (CRF) which uses a 4-point scale (0 = not at all like the person to 3 = extremely like the person) to rate all 25 PD-III trait facets.

Data-Analytic Strategy

We first examined the zero-order correlations between dimensional ratings of each PD-II type with their hypothesized PD-III traits and with personality dysfunction—both overall and for self- and interpersonal pathology separately—using both interview and self-report measures, respectively. We then conducted a series of multiple regression analyses to predict each PD-II type from its hypothesized PD-III traits and personality dysfunction. For each type, we first entered all hypothesized traits and then used backward regression until only traits with significant (p < .05) beta weights remained. Next, we did the same with self- and interpersonal dysfunction. Finally, we performed two sets of hierarchical multiple regressions, first entering self- and/or interpersonal dysfunction (depending on what was significant in the backward regressions), followed by the significant trait set for each PD-II type and then vice versa, to determine the relative predictive power of traits versus personality dysfunction in predicting the PD-II types.

Results

Reliability and Convergent/Discriminant Validity

PD-III Ratings

We first examined the interrater reliabilities (N = 28) for the LPFS and CRF (PD-III Criterion A and B, respectively), shown in the first numerical column of Table 6.2. The mean intraclass coefficient (ICC) for the four PD-III Criterion A ratings was .72 (range = .67 [empathy] to .76 [self-direction]), whereas that for the facets was .67 (range = .35 [eccentricity] to .91 [separation insecurity].3 The quality of these ratings is highly encouraging, especially given that they were made based on information gathered with an interview designed for the PD-II-type diagnoses. (p. 130) (p. 131) As measures are developed specifically to assess the PD-III constructs, interrater reliability may be expected to improve.

Table 6.2 Interrater Reliabilities and Convergent Validity of Dimensional Ratings

Measure

ICC

Convergent rs

personality functioning

Overall (Four Components’ Mean)

.83

.48

Self-pathology (Components’ Mean)

.77

.50

    Identity

.71

    Self-direction

.76

Interpersonal dysfunction (Components’ Mean)

.77

.43

    Empathy

.67

    Intimacy

.73

trait facets

Negative Affectivity

Emotional lability

.85

.55

Anxiousness

.64

.35

Separation insecurity

.91

.44

Submissiveness

.48

.36

Hostility

.78

.52

Perseveration

.55

.25

Detachment

Withdrawal

.76

.59

Intimacy avoidance

.63

.42

Anhedonia

.57

.47

Depressivity

.72

.54

Restricted affectivity

.49

.27

Suspiciousness

.66

.42

Antagonism

Manipulativeness

.80

.44

Deceitfulness

.61

.42

Grandiosity

.73

.32

Attention seeking

.84

.51

Callousness

.59

.34

Disinhibition

Irresponsibility

.85

.39

Impulsivity

.82

.49

Distractibility

.44

.29

Risk taking

.68

.22

Psychosis

Unusual beliefs and experiences

.57

.37

Eccentricity

.35

.13

Cognitive and perceptual dysregulation

.65

.24

Trait Mean

.69

.40

note: Rigid perfectionism omitted due to errors in the ratings.

ICC = intraclass correlation coefficient (N = 28 pairs of interviews); convergent r = interviewer–self-report correlations.

We then examined the convergent/discriminant correlations between the interview-based and self-report ratings of personality functioning and traits; the convergent validities are shown in the last column of Table 6.2. The personality functioning measures, both individually and when aggregated, showed a good convergent/discriminant pattern (i.e., correlated most strongly with their counterpart and significantly less with the other domain); the aggregated convergent/discriminant correlations averaged .57 and .33, respectively.

For the 25 facets, average convergence (mean r = .40; range = .13 [eccentricity] to .59 [withdrawal]) was considerably higher than the average discriminant correlation (mean r = .13); 68% of the convergent correlations were the highest for both sets of ratings. In another 20%, the convergent correlation was the highest for one of the two paired measures (e.g., CRF anhedonia correlated most strongly with PID-5 anhedonia [r = .47], but PID-5 anhedonia correlated slightly stronger [r = .50] with CRF depressivity). Thus, only three facets—anxiousness, eccentricity, and cognitive-and-perceptual aberration—did not converge either way, suggesting that these three PID-5 scales may benefit from modification to reflect the facet definition more closely and/or that the facet definitions themselves need refinement.

PD-II Type Diagnoses

We next examined the interrater reliabilities and internal consistencies of the 10 PD-II-type diagnoses. The ICCs (shown in the first numerical column of Table 6.3) were quite strong: M = .87, range = .73 (OCPD) to .96 (avoidant PD). We then calculated Cronbach’s alpha for the set of traits hypothesized to correspond to each of the 10 PD-II types, both without Criterion A (i.e., traits only), and then adding the Criterion A ratings (LPFS) or aggregated self-report scale score (shown in the middle portion of Table 6.3). Internal consistencies (p. 132) for the self-report scales were higher than for the interview-based ratings for both the traits alone (i.e., Criterion B; means = .74 and .68, respectively) and with Criterion A included (means = .83 and .76, respectively), which might be expected given that the self-report scores already were reliable aggregates of multiple items. Including Criterion A increased internal consistency substantially (average alpha increased from .68 to .76 for the interview-based ratings and from .74 to .83 for the self-report scales), indicating that the traits shared considerable common variance with personality dysfunction.

Table 6.3 Reliabilities of Dimensional Ratings of Interview- and Self-Report Scale-Based Personality Disorder Types

Construct (No. of Traits)

PD-II ICC

PD-III Alpha

PD-III Mean Interitem Correlation

CRF

PID-5

CRF

PID-5

B

A+B

B

A+B

B

A+B

B

A+B

Antisocial PDa (7)

.83

.85

.86

.86

.86

.45

.47

.47

.47

Avoidant PD (4)

.96

.77

.81

.74

.84

.46

.52

.42

.57

Borderline PD (7)

.86

.75

.80

.74

.89

.30

.36

.29

.54

Narcissistic PD (4)

.94

.76

.78

.71

.78

.44

.47

.38

.47

Obsessive-Compulsive PD (4)

.73

.41

.57

.58

.71

.15

.25

.26

.38

Schizotypal PD (7)

.85

.75

.80

.80

.85

.30

.36

.36

.45

Dependent PD (4)

.90

.75

.79

.78

.83

.43

.48

.47

.55

Histrionic PD (2)

.92

.37

.58

.79

.88

.23

.41

.65

.79

Paranoid PD (4)

.85

.68

.78

.73

.81

.35

.47

.40

.52

Schizoid PD (4)

.89

.75

.78

.71

.81

.43

.47

.38

.52

Mean

.87

.68

.76

.74

.83

.35

.43

.41

.53

note:

a Adult criteria. Conduct disorder criteria α‎ = .96. PD types common to PD-II-III listed first, then PD-III-only PDs; in alphabetical order, respectively.

PD-II-III = DSM-5, Section II–III personality disorder types, respectively; CRF = Clinician’s Rating Form; PID-5 = Personality Inventory for DSM-5 (Krueger et al., 2013); ICC = intraclass coefficient (N = 28 pairs of interviews); PD = personality disorder.

Because some of the alpha coefficients were rather low (e.g., .37 and .41 for interview-based HPD and OCPD, respectively) and because the number of traits hypothesized to map to the PD-II types ranged from 2 to 7, we also examined the average interitem correlations (AICs). None of the values was below Clark and Watson’s (1995) recommended range of .15 to .50, indicating that the low alpha values derived, at least in part, from the small number of traits mapped to certain PD-II types. Interestingly, one of the interview-based sets of ratings, and (p. 133) seven (35%) of the self-report scale sets exceeded the recommended maximum. Indeed, even the average AIC—averaged across both the self-reported traits and the personality pathology measures was .53 (see the last column of Table 6.3). These results again underscore the overlap between PD-III Criterion A and B, especially when assessed via self-report measures.

Relations of PD-III Traits and Personality Dysfunction to PD-II-Type Dimensions

Zero-Order Correlations

The first and third numeric columns in Table 6.4 present the zero-order correlations of each PD-II type’s dimensional ratings with its hypothesized traits and with self- and interpersonal dysfunction, for interview-rated and self-reported measures, respectively. The first and third numeric columns in the left and right halves of Table 6.5 present the averages of the trait and self-and-interpersonal dysfunction correlations, respectively, for each PD-II type for interview-rated and self-reported measures, respectively. In both tables, the six PD types common to Sections II and III are listed first, followed by the Section-II-only types, alphabetized within each subgroup. The average trait correlations were generally somewhat higher than those for personality dysfunction, and the interview-based correlations were higher than those for self-report: For traits and personality dysfunction, respectively, interview-based mean rs = .58 and .43, and self-report mean rs = .38 and .29.

Table 6.5 Summary of Zero-Order Correlations and Multiple Regressions Predicting Dimensional Ratings of DSM-5, Section II Personality Disorders PD-III Traits and Self- and Interpersonal Dysfunction

Personality Disorder

Clinicians’ Rating Form

Personality Inventory for DSM-5

Traits

PDys

Δ‎R2

Traits

PDys

Δ‎R2

Mean r

Ra

Mean r

Ra

+PDys

+Traits

Final R

Mean r

Ra

Mean r

Ra

+PDys

+Traits

Final R

Antisocial

.64

.87

.44

.47

.00

.54

.87

.39

.52

.25

.29

.01

.20

.53

Avoidant

.53

.72

.34

.40

.00

.36

.72

.38

.53

.49

.55

.04

.02

.56

Borderline

.51

.85

.60

.69

.02

.27

.86

.49

.65

.49

.57

.01

.10

.65

Narcissistic

.62

.85

.46

.53

.02

.46

.86

.35

.46

.11

.14

.01

.20

.47

Obsessive-compulsive

.36

.72

.22

.23

.00

.47

.72

.23

.41

.09

.10

.00

.15

.41

Schizotypal

.51

.81

.52

.56

.00

.34

.81

.31

.51

.32

.38

.01

.13

.52

Dependent

.57

.82

.36

.45b

.00

.47

.82

.40

.51

.26

.41c

.01

.07

.51d

Histrionic

.63

.82

.28

.28

.00

.60

.82

.41

.54

.09e

.18

.01

.27

.54

Paranoid

.56

.80

.56

.60

.00

.38

.81

.37

.54

.37

.46

.04

.12

.57

Schizoid

.60

.81

.41

.50

.01

.41

.81

.37

.54

.33

.44

.01

.10

.54

Mean

.58

.81

.43

.48

.004

.43

.82

.38

.52

.29

.36

.015

.14

.53

note:

a For hypothesized traits or self/interpersonal dysfunction, respectively, remaining at p < .05 using backward regression. See text for details.

b R = .46 including interpersonal dysfunction with a negative beta weight.

c R = .45 including interpersonal dysfunction with a negative beta weight.

d R = .52 including interpersonal dysfunction with a negative beta weight.

e Of absolute values; interpersonal dysfunction correlation is negative.

PD = personality disorder; PDys = personality dysfunction; Δ‎R2= R2 change after adding significant self/interpersonal dysfunction or hypothesized traits to the other, respectively; mean r = average of zero-order correlations, for all hypothesized traits or self-and-interpersonal dysfunction, respectively; R = multiple R; based on adjusted R2s.

Correlations between the trait ratings and the PD-II-type diagnoses can be interpreted similarly to each other across methods, because both the interview-based and self-report measures were derived directly from the PD-III trait-facet definitions. Specifically, the interviewers rated the trait facets based on their definitions, and the self-raters completed the PID-5, which was designed to reflect these same definitions. Of course, the targets for both sets of raters were the interviewers’ PD-II diagnostic ratings, so interviewers had the advantage of using the same material to rate the PD-II diagnoses and PD-III traits, whereas the self-raters completed the PID-5 scales and we examined their relations with the interviewers’ PD-II-type ratings. Thus, it is not surprising that the interview-based correlations are higher. Nonetheless, the moderate to moderately strong correlations of the self-report scales with the corresponding PD-II-type ratings provide evidence of the PID-5 scales’ convergent validity with interview-based ratings.

In contrast, the meaning of the personality dysfunction–PD-II correlations is somewhat different for the interview-based ratings versus self-report scales, because the measures used by the two sets of raters were rather different. Interviewers rated personality dysfunction using the PD-III definitions and their elaboration in the LPFS, whereas self-raters responded to personality-functioning scales that had been developed prior to DSM-5’s publication. Moreover, again, interviewers used the same material to make both PD-II and PD-III ratings. (p. 137) (p. 136) (p. 135) (p. 134) (p. 138) Thus, the interview-based correlations more or less directly indicate the degree to which PD-III personality functioning is inherent in the PD-II types’ criteria. For example, the interview-based results suggest that personality dysfunction is an important aspect of PD-II-BPD (mean r = .60), but it is much less important in PD-II-OCPD (mean r = .22). Interestingly, these findings are consistent with those of Sharp et al. (2015), who found, using bifactor analysis, that BPD criteria loaded strongly on a general PD factor (mean = .64), whereas OCPD criteria did not (mean = .27).

For the self-report scales, however, the correlations reflect the extent to which the personality-functioning scales’ content overlaps with that of the PD-II-types’ criteria, rather than directly tapping the PD-III personality-functioning definitions. As mentioned, this is because these scales were developed prior to PD-III’s development. However, the PD-III personality-functioning definitions may have been influenced by the personality-functioning scales’ content, because the authors of two of the three measures (GAPD and SIPP) were original members of the DSM-5 Personality and Personality Disorders Work Group (P&PDWG). Moreover, the results are reasonably parallel across methods. For example, as with the interview-based ratings, the personality-functioning scales correlated moderately strongly with ratings of PD-II-BPD criteria (mean r = .49), but minimally with ratings of PD-II-OCPD (mean r = .09). Moreover, the interview-self-report convergent correlations (see Table 6.2) were similar for self- and interpersonal pathology (mean rs =.50, .43, respectively) versus traits (mean rs =.40). Finally, the mean correlational pattern of the PD-II types across methods (i.e., correlating respective Table 6.5 columns) was actually higher for personality pathology than for traits (rs = .63 vs. 42, respectively). Thus, although the meaning of the correlations may be different conceptually, the values were comparable empirically.

Interestingly, the average predictive power was not significantly higher for the six PD types common to Sections II and III (i.e., antisocial, avoidant, borderline, narcissistic, OC, and schizotypal PD) compared to those that are only in Section II (i.e., paranoid, schizoid, histrionic, and dependent PDs). For traits, the mean values were .60 and .58, respectively, whereas for overall personality dysfunction, self-dysfunction, and interpersonal dysfunction, they were, respectively, .47, .43, and .42 versus .44, .40, and .41. Thus, from the perspective of personality functioning and traits, there is no reason for some PD-II types to have been included and others excluded in PD-III.

Multiple Regressions Predicting PD-II-Type Dimensions From PD-III Personality Dysfunction and Traits

The second and fourth columns of Table 6.4 present the beta weights for the final regression equations predicting dimensional ratings of each PD-II type from its significant hypothesized traits plus personality dysfunction, for the interview ratings and self-report scales, respectively. The final adjusted multiple Rs also are provided. Several aspects of these results are noteworthy. First, more of the hypothesized traits were significant predictors for the interview-based ratings (77%) than the self-reports (38%), perhaps because the trait intercorrelations were somewhat stronger among the self-report scales than interview-based (p. 139) ratings (mean rs= .34 vs. .22). Second, personality dysfunction was more likely to contribute to predicting the PD-II types for self-report than interviews (eight vs. four PD-II types). This may be because interviewers based their ratings on a single interview and/or because members of the P&PDWG developed the definitions for both PD-III Criteria A and B, whereas the trait and personality-functioning scales were distinct measures developed by different researchers. Third, when personality dysfunction did contribute to the equation, only self- or interpersonal dysfunction was a significant predictor, never both.4

Summaries of the steps that led to these final equations are shown in Table 6.5. Specifically, the second and fourth columns (for each half: the interviews and self-reports, respectively) give the multiple Rs for the hypothesized traits and personality pathology, respectively. The fifth and sixth columns in each half of the table give the change in R2 when personality dysfunction was added to traits and vice versa, and the last columns give the final adjusted multiple Rs (as mentioned, also shown in Table 6.4). Again, several aspects of the results are noteworthy.

First, for the clinician ratings, traits strongly predicted the PD-II diagnoses; average multiple R = .81 (range= .72 [OCPD] to .87 antisocial PD [ASPD]). The mean ICC for the PD-II types was .87, so these results suggest that the PD-III trait ratings predicted almost all of the reliable variance of the PD-II types. Using self-reports, traits’ predictive power was moderately strong, averaging .53 (range= .41 [OCPD] to .65 [ASPD]), again indicating good convergent validity of self-reports with interview-based ratings. The second striking finding is that when both traits and personality dysfunction were entered into the equations, the traits always significantly incremented the predictive power over personality dysfunction (the sixth columns in each of the left and right halves of Table 6.5), again more strongly for the interview-based than the self-report measures (mean Δ‎R2= .43 and .14, respectively). In contrast, personality dysfunction provided little to no additional predictive power over traits for either interview ratings or self-report scales (mean Δ‎R2= .004 and .015; maximums= .02 and .04, respectively), although even these small increments were significant in three cases for the interviews and all but one case in self-reports.

Third, overall predictive power was notably higher for the interview-based ratings versus self-report measures (mean rs = .82 vs. .53), clearly a method-based difference but, again, the results generally indicate good convergent validity of self-report with interview-based ratings of the PD-II types. Finally, there was again no difference in the overall predictive power for the six PD types common to Sections II and III versus those that are only in Section II (mean rs both = .82 for the interview-based ratings and .53 and .54 for the self-reports).

Comparing A Priori Hypothesized Traits and Traits Comprising the PD-III Types in Predicting PD-II Type Dimensions

The traits used to predict the PD-II types in the preceding sections were based on a priori hypotheses and are similar, but not identical, to the traits comprising the six PD-III types (except for ASPD and OCPD, for which they are identical). To determine whether the a priori hypotheses or the “official” PD-III traits better predicted ratings of the PD-II types, we reran the analyses starting with the traits comprising (p. 140) the PD-III types for the four PD types with a nonidentical trait set (i.e., avoidant, borderline, narcissistic, and schizotypal PDs). In all cases, the hypothesized trait set yielded either the same model after nonsignificant traits were eliminated or a higher multiple R, although in most cases the difference was small, never exceeding .02. Of course, these trait sets are hardly independent, as the hypotheses were finalized by a member of the DSM-5 P&PDWG with its input. Also, in one case (NPD), manipulativeness was hypothesized and contributed to predicting PD-II NPD, but it was deliberately omitted by the DSM-5 P&PDWG because its inclusion increased comorbidity with ASPD, whereas here we examined only predictive validity.

Structural Placement of Personality Dysfunction

Another way to examine relations between the PD-II types and PD-III personality dysfunction is through broad structural analyses. To this end, we ran two principal-axis factor analyses with varimax rotation, one using interview-based ratings and the other self-reports.

Interview-Based Ratings of PD-II Types

For the interview-based analyses, we factored the 10 PD-II-types’ dimensional scores with the ratings of PD-III self- and interpersonal dysfunction. We extracted three factors, as indicated by parallel analysis. The results (see Table 6.6), reflect (p. 141) the traditional DSM PD-category cluster structure with two exceptions: BPD split across all three clusters, loading most strongly on the third factor, and OCPD did not load on any dimension (although when a fourth factor was extracted, it was the sole marker). These findings again are consistent with those of Sharp et al. (2015), whose results suggest that the BPD criteria lie at the core of personality pathology, whereas those of OCPD are more peripheral.

Table 6.6 Principal-Axis Factor Analysis of the PD-II Types Scored Dimensionally and PD-III Self- and Interpersonal Dysfunction

Measure

Psychoticism

Externalizing

Internalizing

Schizoid PD

.73

–.12

.04

Schizotypal PD

.70

.23

.13

Interpersonal Dysfunction

.68

.41

.26

Paranoid PD

.61

.25

.23

Obsessive-compulsive PD

.21

.17

.13

Narcissistic PD

.30

.71

–.06

Histrionic PD

–.06

.70

.12

Antisocial PD

.20

.56

.17

Dependent PD

.02

.15

.64

Self Dysfunction

.27

.52

.59

Borderline PD

.50

.39

.59

Avoidant PD

.34

–.25

.54

notes: Self and Interpersonal Dysfunction (shown in italics) comprise Criterion A of the PD-III General Diagnostic Criteria. All loadings > .35 are in bold.

PD = personality disorder; II, III = Sections II and III, respectively, of the Diagnostic and Statistical Manual of Mental Disorders (5th ed.).

Despite the parallelism with the traditional DSM PD cluster structure, we labeled these factors psychoticism, externalizing, and internalizing, terms used very commonly in the psychopathology-structure literature (e.g., Bagby et al., 2014; Caspi et al., 2014), to facilitate linking the PD and broader psychopathology literatures. Interestingly, interpersonal dysfunction split across the psychoticism and externalizing factors, loading most strongly on the former, whereas self-dysfunction split across the externalizing and internalizing factors, loading slightly more strongly on the latter. This pattern indicates that these two aspects of personality functioning may have somewhat different relations to personality pathology and, by extension, may relate differently to psychopathology in general. One interpretation of the fact that interpersonal dysfunction does not load on the internalizing factor is that internalizing psychopathology may have a less deleterious effect on interpersonal relations. By the same token, however, it is puzzling that psychoticism was not strongly related to self pathology, because it seems reasonable that being high on psychoticism would have a disruptive influence on one’s identity and self-direction. This finding needs further investigation.

Structure of PD-III Personality Dysfunction and Traits

To examine structural relations between PD-III personality dysfunction and traits, we ran two principal-axis factor analyses with varimax rotation, one using the interview-based ratings and the other the self-report scales, each including the 25 PID-5 facet scales and measures of self- and interpersonal dysfunction. In both cases, we extracted four factors, based on the results of parallel analyses. The results are shown in Table 6.7. Although there were a number of cross-loadings, clear N/NA, antagonism + disinhibition, detachment, and psychoticism factors emerged in both analyses, and the facets’ primary loadings showed the same pattern for all but four facets.

Table 6.7 Principal-Axis Factor Analysis of the PD-III Facet Scales and Measures of Self- and Interpersonal Dysfunction Using Interview Ratings and Self-Report Scales

Measure

N/NA

Antagonism

Detachment

Psychoticism

SR

INT

SR

INT

SR

INT

SR

INT

Depressivity

.82

.67

.01

.04

.35

.37

.09

.00

Self dysfunction

.83

.63

.05

.37

.41

.38

.14

.31

Anxiousness

.77

.63

–.03

–.01

.10

.34

.29

.23

Distractibility*

.72*

.30

.08

.33

.13

.10

.20

.21

Emotional lability

.66

.53

.10

.26

.12

.07

.39

.26

Separation insecurity

.64

.67

.19

.10

–.06

.03

.24

.12

Submissiveness

.46

.54

.17

–.08

–.17

–.06

–.01

.04

Suspiciousness*

.38

.14

.30

.21

.30

.44

.31

.48

Deceitfulness

.13

–.01

.78

.84

.19

.05

.14

.05

Manipulativeness

.00

.02

.73

.78

.04

.00

.17

.11

Risk taking

.01

–.02

.69

.55

–.06

.07

.00

.10

Callousness

.08

–.05

.68

.70

.43

.19

.11

.23

Attention seeking

.11

–.03

.64

.48

–.31

–.22

.26

.33

Impulsivity

.42

.27

.60

.69

.04

–.04

.17

.01

Grandiosity

–.06

–.16

.56

.46

.06

.02

.46

.52

Irresponsibility

.46

.22

.53

.71

.23

.06

.03

.01

Hostility

.37

.17

.39

.42

.33

.27

.40

.34

Anhedonia*

.67*

.32

–.11

.12

.48

.59

–.02

.08

Interpersonal dysfunction

.22

.24

.22

.40

.82

.57

.08

.42

Withdrawal

.26

.25

–.11

–.15

.77

.73

.17

.07

Restricted affectivity

–.09

–.12

.19

.04

.54

.44

.16

.04

Intimacy avoidance

.12

.09

–.02

–.01

.52

.75

.13

.06

Perseveration*

.61*

.32

.16

.06

.08

–.06

.49

.46

Rigid perfectionism

.40

.12

.30

.22

.26

.25

.63

.51

Cognitive and perceptive dysregulation

.17

.22

.06

–.09

.14

–.03

.62

.43

Unusual beliefs and experiences

.24

.23

.25

.19

.15

.28

.60

.46

Eccentricity

.38

.02

.29

.23

.28

.31

.49

.44

note: Loadings ≥ .35 are in bold. Results for personality (self and interpersonal) dysfunction are in italics.

* Facets and loading that showed a different pattern between self- and interview raters.

N/NA = Neuroticism/Negative Affectivity; SR = self-report; INT = interview-based ratings.

In each of these four cases, the facet loaded most strongly on the N/NA factor in the self-report analysis (range = .61–.72, except suspiciousness, for which its N/NA loading, although still its highest, was only .38), but it did not load on this factor in the interview-based analysis. Instead, in the interview-based analysis, (1) distractibility had no strong loading on any factor; and (2) suspiciousness loaded most strongly on the detachment and psychoticism factors (.44 and .48, respectively); whereas (3) perseveration loaded moderately strongly on psychoticism (.49 and .46 in the self-report and interview-based analyses, respectively) and (4) anhedonia loaded moderately strongly on detachment (.48 and .59, respectively), the difference being that these were primary loadings in the interview-based analysis and secondary loadings in the self-report analysis. This pattern emerged, in part, because the N/NA factor was considerably larger in the (p. 142) (p. 143) self-report than the interview-based analysis, perhaps because internal distress is more salient to individuals experiencing it than to observers whose information derives largely from an interview.

Turning to the personality-dysfunction measures, in both analyses, self-dysfunction loaded strongly (.63–.83) on the N/NA factor and had a moderately strong loading (.38–.41) on the detachment factor. However, in the interview-based analysis, it also loaded moderately strongly (.37) on the antagonism/disinhibition factor. Interpersonal pathology, on the other hand, showed a somewhat different loading pattern in the two analyses. It marked the detachment factor in both analyses, but quite strongly (.82) and only that factor in the self-report analysis, whereas in the interview-based analysis, the loading was more moderate (.57) and the scale also loaded moderately strongly (.40–.42) on the antagonism and psychoticism factors. These results suggest that, in relation to traits, self-dysfunction is rated similarly by both self- and interview raters, except that self-raters link self-dysfunction more strongly with N/NA-factor traits, and only interviewers link self-dysfunction with antagonism. In contrast, interpersonal dysfunction is linked with traits differently by the two types of raters: Self-raters link interpersonal dysfunction only with detachment-factor traits, whereas interview raters link it broadly with all traits except N/NA-factor traits.

The Importance of Severity

Over the past two decades, multiple researchers (e.g., Caspi et al., 2014; Clark, Watson, & Reynolds, 1995; Tyrer, 2005; Tyrer et al., 2011) have noted that severity is a very important—but not well-defined or understood—dimension of psychopathology. An important question relevant to the issue of severity is the degree to which trait extremity and personality dysfunction are related in general, that is, not in relation to specific traits or trait factors, but overall. Our data allow us to examine this question by computing aggregate measures of trait extremity across all 25 PD-III facets. We did so, separately for self- and interview raters, and first tested whether such an aggregated measure could be interpreted as a metric of overall trait extremity/severity by computing internal consistency indices for the 25 facets in both the self-reports (α‎ = .92; AIC = .32) and interview-based (α‎ = .86; AIC = .20) ratings. These high values confirmed the interpretability of a composite index.

We then correlated these aggregates with self- and interview-based measures of self- and interpersonal dysfunction. It is clear from the results (see Table 6.8) that, using current measures, trait extremity and personality dysfunction are strongly overlapping constructs (rs = .76 and .80 for self-report and interview-ratings, respectively), with their overlap indexing “severity,” broadly speaking. Indeed, the correlations are sufficiently strong that they largely explain why there was little to no increment in predicting PD-II types by personality dysfunction, above and beyond traits. Moreover, it also is clear that self-dysfunction is the stronger component in the correlation of personality dysfunction with trait extremity, at least for the self-ratings. Importantly, this is the case even though the within-method part-whole correlations of self- and interpersonal dysfunction with overall personality dysfunction are virtually the same (i.e., .90 vs. .88 and .92 vs. .93).

Table 6.8 Correlations Among Overall Trait Extremity and Personality Dysfunction—Overall, Self, and Interpersonal, for Self-Report and Interview-Based Measures

Personality Dysfunction and Trait ExtremityConceptually, but Not Empirically Distinct?

note: Boxed correlations are between traits and personality dysfunction (overall and separately for self and interpersonal domains). Within-method correlations are in bold. Cross-method convergent correlations are in underlined italics.

SR = self-report; INT = interview based.

(p. 144) The cross-method difference in the nature of the personality dysfunction ratings noted earlier (i.e., based directly on PD-III for the interview-based ratings only) is supported in these analyses, in that the cross-method trait-extremity correlation (.57), for which both sets of ratings were based on the PD-III facets, was stronger than any of the other cross-method correlations (rs ranged from .43 to .50; although the only statistically significant difference among the four correlations was .57 vs. .43). In any case, the strong correlations between PD-III Criteria A and B for both types of measures represent a very strong measurement challenge to any theory-based distinction between them.

Discussion

Meta-analyses (Samuel & Widiger, 2008; Saulsman & Page, 2004) have shown that normal-range personality-trait measures relate systematically to PD-II types. (p. 145) Moreover, dimensional interview-based measures of PD-II-types and trait profiles show a clear convergent/discriminant pattern (e.g., Miller, Reynolds, & Pilkonis, 2004; Miller, Bagby, & Pilkonis, 2005; Samuel, Connolly, & Ball, 2012), with convergence averaging ~.40-.50. Our results were similar (mean r = .53, range = .41–.65) using the PID-5, which was designed to assess the 25 PD-III facets’ more extreme high ends. Although critics have countered that this convergence level is too low to consider traits as a replacement for traditional diagnoses, it is comparable to that obtained between structured and unstructured clinical interviews (mean r = .42, range= .32–.51; Miller, Few, & Widiger, 2012).

Further, we obtained considerably stronger convergence (mean R = .81, range = .72–.87) using an interview-based trait measure. This value is slightly higher than the convergence between two interview-based measures (.77; Miller et al., 2012). The “too low” values reported previously thus seemingly reflect method variance, rather than any inadequacy of traits per se to assess PD. Interestingly, all 10 PD-II types were equally well predicted, thus mitigating concerns that four PD-II types are not represented in PD-III. Although they are not specifically depicted, they are easily diagnosed using two or three PD-III traits in the presence of self- and/or interpersonal dysfunction.

Given (1) the infamous problems with the PD-II types, (2) clear evidence that traits not only can capture the personality pathology they represent but can do so in a way that is clinically useful (e.g., Bach, Markon, Simonsen, & Krueger, 2015; Clark et al., 2015; Glover, Crego, & Widiger, 2012), and (3) over a 6- to 10-year follow-up, “a system that integrates normal and pathological traits generally showed the largest validity coefficients in predicting a host of important clinical outcomes” (p. 1705), we contend that it is time to focus PD research on advancing our understanding entirely from a dimensional perspective, particularly that of its core components, personality dysfunction and extreme traits.

“Borderline” Personality Disorder

There is no question that BPD captures the vast majority of current clinical and research attention on PD. Our results, along with those of Sharp et al. (2015), offer some reasons for why that may be. First, when we factor analyzed interview-based ratings of the PD-II types plus personality dysfunction, BPD ratings split across all three factors. Indeed, the PD-II BPD criteria reflect not only all three major PD factors—“transient stress-related paranoid ideation or severe dissociative symptoms” taps psychoticism, “impulsivity in at least two areas that are potentially self-damaging” reflects externalizing, and “affective instability due to a marked reactivity of mood” taps internalizing—but also both self-dysfunction (“identity disturbance: markedly and persistently unstable self-image or sense of self”) and interpersonal dysfunction (“a pattern of unstable and intense interpersonal relationships”) (APA, 2013, p. 663). We investigated this further by correlating interviewers’ ratings of overall personality dysfunction and mean trait extremity with the PD-II type ratings and found that BPD had the strongest correlation in (p. 146) each case, correlating .64 with overall personality dysfunction and .71 with mean trait extremity (vs. the next highest values of .61 and .53, respectively.)

It also is noteworthy that the PD-II BPD criteria do not reflect a clear set of traits, but rather a mix of traits (e.g., impulsivity), symptoms (e.g., chronic feelings of emptiness), specific behaviors (e.g., recurrent suicidal behavior), and personality dysfunction (e.g., identity disturbance). This is in contrast to other PD types, such as paranoid or avoidant PD, in which one or a few traits predominate (respectively, suspiciousness and anxiousness + detachment). Thus, BPD most likely garners the most clinical and research attention not because it is a particular syndrome, but because it is essentially a synonym for complex, severe PD. Recognition of this fact and acknowledgment that complex, severe PD is heterogeneous, reflecting diverse trait combinations, will facilitate research into both etiology and treatment.

Limitations

Our study has several limitations: (1) We used a single PD interview, and given the generally low agreement among PD interviews (median kappa = .35; Clark, Livesley, & Morey, 1997), our results need replication with another interview. (2) Our self-report personality-dysfunction measures were developed prior to PD-III and do not specifically provide for its subdomains. However, post-hoc analyses indicated that four of the six scales we used to create the self- and interpersonal-dysfunction indices correlated most strongly with the four interview-based subdomain scores compared to other functioning measures in our battery, ranging from .31 (MDPF Non-cooperativeness with Empathy) to .47 (GAPD Self-pathology with Identity), indicating the possibility of developing at least somewhat distinct self-report scales for the four subdomains. (3) Three PD-III facets—anxiousness, eccentricity, and cognitive-and-perceptual aberration—did not converge between the interview-based ratings and self-report scales, and five others showed less than optimal convergence. Thus, these facets may benefit from further clarification and improved measurement in multiple modalities.

Summary, Conclusions, and Future Directions

Taken together, our findings indicate clearly that personality dysfunction and trait extremity are strongly interrelated in existing measures of these constructs. Moreover, even though personality dysfunction is not an explicit construct in PD-II-types’ conceptualization, both interview-based and self-ratings of personality dysfunction correlated moderately strongly with PD-II-type diagnoses (mean rs = .46 and .32, respectively, for overall personality dysfunction). Further, using interview-based trait ratings, our analyses indicate even stronger relations between PD-II types and personality traits, to the point that they account for (p. 147) virtually all their reliable variance (mean R = .81), whereas measures of personality dysfunction had little to no additional predictive power (mean R = .004).

Personality dysfunction is a relatively new research domain, however, so the extent to which a strong interrelation with traits is inherent in these constructs versus a measurement-based artifact remains unknown. That is, there is content overlap in the definitions and descriptions of existing personality-dysfunction and trait measures, and we do not know how strongly these constructs would relate if measures were developed based on nonoverlapping definitions. Nor do we know whether it even is possible to define these constructs without overlapping content. If these constructs are inherently intertwined, there may be no, or only a few, ways in which personality dysfunction can be manifested independently of trait expression. Thus, we have much to learn regarding the conceptualization and measurement of personality dysfunction. Determining whether trait measures can be developed that are—at least in terms of their content—distinct from personality dysfunction and vice versa is a primary future research task.

References

American Psychiatric Association. (1994). Diagnostic and statistical manual (4th ed.). Arlington, VA: Author.Find this resource:

American Psychiatric Association. (2013). Diagnostic and statistical manual (5th ed.). Arlington, VA: Author.Find this resource:

Bach, B., Markon, K., Simonsen, E., & Krueger, R. F. (2015). Clinical utility of the DSM-5 alternative model of personality disorders: Six cases from practice. Journal of Psychiatric Practice, 21(1), 3–25. doi:10.1097/01.pra.0000460618.02805.efFind this resource:

Bagby, R. M., Sellbom, M., Ayearst, L. E., Chmielewski, M. S., Anderson, J. L., & Quilty, L. C. (2014). Exploring the hierarchical structure of the MMPI-2-RF personality psychopathology five in psychiatric patient and university student samples. Journal of Personality Assessment, 96(2), 166–172. doi:10.1080/00223891.2013.825623Find this resource:

Caspi, A., Houts, R. M., Belsky, D. W., Goldman-Mellor, S., Harrington, H., Israel, S., . . . Moffitt, T. E. (2014). The p factor: One general psychopathology factor in the structure of psychiatric disorders? Clinical Psychological Science, 2(2), 119–137. doi:10.1177/2167702613497473Find this resource:

(p. 148) Clark, L. A., Livesley, W. J., & Morey, L. (1997). Personality disorder assessment: The challenge of construct validity. Journal of Personality Disorders, 11, 205–231.Find this resource:

Clark, L. A., & Ro, E. (2014). Three-pronged assessment and diagnosis of personality disorder and its consequences: Personality functioning, pathological traits, and psychosocial disability. Personality Disorder: Theory, Research, & Treatment, 5(1), 55–69. doi:10.1037/per0000063Find this resource:

Clark, L. A., Vanderbleek, E., Shapiro, J., Nuzum, H., Allen, X., Daly, E., . . . Ro, E. (2015). The brave new world of personality disorder-trait specified: Effects of additional definitions on prevalence and comorbidity. Psychopathology Review, 2(1), 52–82. doi:10.5127/pr.00Find this resource:

Clark, L. A., & Watson, D. B. (1995). Constructing validity: Basic issues in objective scale development. Psychological Assessment, 7, 309–319.Find this resource:

Clark, L. A., Watson, D., & Reynolds, S. K. (1995). Diagnosis and classification in psychopathology: Challenges to the current system and future directions. Annual Review of Psychology, 46, 121–153. doi:10.1146/annurev.psych.46.1.121Find this resource:

Finn, S. E. (1982). Base rates, utilities, and DSM-III: Shortcomings of fixed-rule systems of psychodiagnosis. Journal of Abnormal Psychology, 91(4), 294–302. doi:10.1037/0021–843X.91.4.294Find this resource:

Glover, N. G., Crego, C., & Widiger, T. A. (2012). The clinical utility of the five-factor model of personality disorder. Personality Disorders: Theory, Research, and Treatment, 3(2), 176–184. doi:10.1037/a0024030Find this resource:

Krueger, R. F., Derringer, J., Markon K. E., Watson D., & Skodol, A. E. (2012). Initial construction of a maladaptive personality trait model and inventory for DSM-5. Psychological Medicine, 42(9), 1879–1890. doi:10.1017/S0033291711002674Find this resource:

Langbehn, D. R., Pfohl, B. M., Reynolds, S., Clark, L. A., Battaglia, M., Cadoret, R., . . . Links, P. (1999). The Iowa Personality Disorder Screen: Development and preliminary validation of a brief screening interview for non-antisocial DSM personality diagnoses. Journal of Personality Disorders, 13, 75–89.Find this resource:

Livesley, J. (2010). General assessment of personality disorder. Port Huron, MI: Sigma Assessment Systems.Find this resource:

Livesley, J. (2012). Tradition versus empiricism in the current DSM-5 proposal for revising the classification of personality disorders. Criminal Behaviour and Mental Health, 22(2), 81–90. doi:10.1002/cbm.1826Find this resource:

Livesley, W. J., & Jang, K. L. (2005). Differentiating normal, abnormal, and disordered personality. European Journal of Personality, 19(4), 257–268. doi:10.1002/per.559Find this resource:

Markon, K. E., Quilty, L. C., Bagby, R. M., & Krueger, R. F. (2013). The development and psychometric properties of an informant-report form of the personality inventory for DSM-5 (PID-5). Assessment, 20(3), 370–383. doi:10.1177/1073191113486513Find this resource:

Miller, J. D., Bagby, R. M., & Pilkonis, P. A. (2005). A comparison of the validity of the five-factor model (FFM) personality disorder prototypes using FFM self-report and interview measures. Psychological Assessment, 17(4), 497–500. doi:10.1037/1040–3590.17.4.497Find this resource:

Miller, J. D., Few, L. R., & Widiger, T. A. (2012). Assessment of personality disorders and related traits: Bridging DSM-IV-TR and DSM-5. The Oxford handbook of personality disorders (pp. 108–140). New York, NY: Oxford University Press. doi:10.1093/oxfordhb/9780199735013.013.0006Find this resource:

(p. 149) Miller, J. D., Reynolds, S. K., & Pilkonis, P. A. (2004). The validity of the five-factor model prototypes for personality disorder in two clinical samples. Psychological Assessment, 16(3), 310–322. doi:10.1037/1040-3590.16.3.310Find this resource:

Parker, G., Hadzi-Pavlovic, D., Both, L., Kumar, S., Wilhelm, K., & Olley, A. (2004). Measuring disordered personality functioning: To love and to work reprised. Acta Psychiatrica Scandinavica, 110, 230–239.Find this resource:

Pfohl, B. M., Blum, N., & Zimmerman, M. (1997). Structured Interview for DSM-IV Personality (SIDP-IV). Washington, DC: American Psychiatric Association.Find this resource:

Ro, E., & Clark, L. A. (2013). Interrelation between psychosocial functioning and adaptive- and maladaptive-range personality traits. Journal of Abnormal Psychology, 122(3), 822–835. doi:10.1037/a0033620Find this resource:

Samuel, D. B., Connolly, A. J., & Ball, S. A. (2012). The convergent and concurrent validity of trait-based prototype assessment of personality disorder categories in homeless persons. Assessment, 19(3), 287–298. doi:10.1177/1073191112444461Find this resource:

Samuel, D. B., & Widiger, T. A. (2008). A meta-analytic review of the relationships between the five-factor model and DSM-IV-TR personality disorders: A facet level analysis. Clinical Psychology Review, 28(8), 1326–1342. doi:10.1016/j.cpr.2008.07.002Find this resource:

Saulsman, L. M., & Page, A. C. (2004). The five-factor model and personality disorder empirical literature: A meta-analytic review. Clinical Psychology Review, 23, 1055–1085. doi:10.1016/j.cpr.2002.09.001Find this resource:

Sharp, C., Wright, A. G. C., Fowler, C., Frueh, C., Oldham, J., & Clark, L. A. (2015). The structure of personality pathology: Both general (“g”) and specific (s’s) factors? Journal of Abnormal Psychology, 124(2), 387–398. doi:10.1037/abn0000033Find this resource:

Verheul, R., Andrea, H., Berghout, C. C., Dolan, C., Busschbach, J. J. V., van der Kroft, P. J. A., . . . Fonagy, P. (2008). Severity Indices of Personality Problems (SIPP-118): Development, factor structure, reliability, and validity. Psychological Assessment, 20, 23–34. doi:10.1037/1040-3590.20.1.23Find this resource:

Tyrer, P. (2005). The problem of severity in the classification of personality disorder. Journal of Personality Disorders, 19(3), 309–314. doi:10.1521/pedi.2005.19.3.309Find this resource:

Tyrer, P., Crawford, M., Mulder, R., Blashfield, R., Farnam, A., Fossati, A., . . . Reed, G. M. (2011). The rationale for the reclassification of personality disorder in the 11th revision of the international classification of diseases (ICD-11). Personality and Mental Health, 5(4), 246–259. doi:10.1002/pmh.190Find this resource:

Wakefield, J. C. (2013). DSM-5 and the general definition of personality disorder. Clinical Social Work Journal, 41(2), 168–183. doi:10.1007/s10615-012-0402-5Find this resource:

Widiger, T. A., & Simonsen, E. (2005). Alternative dimensional models of personality disorder: Finding a common ground. Journal of Personality Disorders, 19(2), 110–130. doi:10.1521/pedi.19.2.110.62628Find this resource:

World Health Organization (2018). International Classification of Diseases. Available athttps://icd.who.int. (p. 150)

Notes:

1. Thus, we use the terms “personality impairment” and “personality dysfunction” interchangeably.

2. Members of the P&PDWG contributed to these predictions, but LAC takes full responsibility for selection and (mis)predictions of the final set.

3. The ICC for rigid perfectionism is not provided due to errors in the rating data.

4. With one exception: When interpersonal dysfunction was included with a negative beta weight predicting Dependent PD in self-report.