Abstract
Both personality impairment and maladaptive-range traits are necessary for diagnosis in the alternative model of personality disorder. We clarified personality impairment-trait connections using measures of the interpersonal problems circumplex and personality traits among adult outpatients (N = 351) with major depressive disorder receiving cognitive therapy (CT). The trait scales’ circumplex projections were summarized by elevation (correlations with general interpersonal problems), amplitude (specific relations to the circumplex dimensions of dominance and affiliation), and angle (predominant orientation in the two-dimensional circumplex). Most trait scales showed hypothesized circumplex relations, including substantive elevation (e.g., negative temperament, mistrust), amplitude (e.g., aggression, detachment), and expected angles (e.g., positive temperament and manipulativeness oriented toward overly nurturant/intrusive or domineering/vindictive problems, respectively), that were stable across time during CT. These results revealed meaningful and consistent impairment-trait connections, even during CT when mean depressive affect decreased substantially.
Relations between personality pathology and depression are strong, complex, and incompletely understood. Clinical questions include the validity of personality pathology assessed during major depressive episodes, when mood disturbance may bias or obscure personality assessment (Skodol, 2014). Related theoretical questions include differentiation of personality pathology from depression (Klein et al., 2014), including the stability of trait-dysfunction systems across changes in depression severity. To advance the literature on these issues, we assessed trait dimensions with the Schedule for Nonadaptive and Adaptive Personality, second edition (SNAP-2; Clark et al., 2014) and dysfunction with the Inventory of Interpersonal Problems—Circumplex version (IIP-C; Horowitz et al., 2000) before and after acute-phase cognitive therapy (CT) for major depressive disorder (MDD). We examined the configuration and stability of trait-dysfunction relations during CT, when patients experienced large decreases in depressive mood.
Personality, particularly personality traits, are conceptualized as relatively stable patterns of affect, behavior, and cognition (ABC). In the Diagnostic and Statistical Manual of Mental Disorders–Fifth Edition (DSM-5; American Psychiatric Association, 2013), Section III—the alternative model of personality disorder (AMPD)—personality impairment (dysfunction) and maladaptive-range personality traits are the main criteria necessary for diagnosis. Personality dysfunction in the AMPD involves self- and interpersonal impairment. Self-impairment includes identity disturbance and poor self-directedness, whereas interpersonal impairment reflects low empathy and difficulty in close personal relationships (i.e., intimacy). Personality traits in the AMPD include five domains: negative affectivity, detachment, antagonism, disinhibition, and psychoticism. The 11th edition of the International Classification of Diseases (ICD-11) presents a similar model of personality disorder (Reed, 2018; World Health Organization, 2019).
The AMPD and ICD-11 PD models have inspired research on the empirical connections between personality dysfunction and maladaptive-range traits, revealing considerable overlap between the two domains (e.g., Bastiaansen et al., 2016; Berghuis et al., 2014; Calabrese & Simms, 2014; Clark & Ro, 2014; McCabe & Widiger, 2020; Morey et al., 2020). In cross-sectional factor analyses, trait and dysfunction measures often (but not always) load on the same factors in community and clinical samples (e.g., Berghuis et al., 2012; Calabrese & Simms, 2014; Clark & Ro, 2014). Similarly, when testing prediction of criterion measures (e.g., traditional personality disorder diagnoses, daily functioning), maladaptive-range trait and personality dysfunction scales often have limited incremental predictive power over one another (Bastiaansen et al., 2016; Berghuis et al., 2014; Calabrese & Simms, 2014). Recently, Morey et al. (2020) demonstrated that one reason for these findings may be that measures of maladaptive-range traits essentially reflect a combination of normal-range trait variance plus variance related to dysfunction. That is, dysfunction is inherent in items of maladaptive-range trait measures. Another likely reason for substantive correlations between trait and dysfunction measures is shared ABC content (Nuzum et al., 2019). Importantly, because traits vary in their ABC content, different traits are more and less relevant across various functioning domains.
Focusing on the domain of interpersonal relationships, interpersonal functioning is often conceptualized and measured using a circumplex model (e.g., Gurtman, 2009; Widiger, 2010; Wiggins & Pincus, 2002). A depiction of the interpersonal circumplex appears in Figure 1. The two-dimensional interpersonal circumplex is anchored by dimensions of dominance (the vertical axis) and affiliation (the horizontal axis), and often divided into eight octants. Each octant marks a progressive blend of interpersonal style associated with low/high dominance and low/high affiliation. Interpersonal problems can range from mild to severe for any orientation in the circumplex. Previously, the interpersonal circumplex has been shown to have systematic relations with measures of the five-factor model of personality (e.g., Du et al., 2020; Traupman et al., 2009). In the current study, we extended this literature by examining relations of trait dimensions assessed by the SNAP-2 with interpersonal dysfunction assessed with the IIP-C. For example, the SNAP-2 aggression scale falls in the domineering-vindictive octant, suggesting that people who score high in aggression often display problems in interpersonal functioning associated with domineering and vindictive behavior (see Figure 1).

Personality scales with a defined orientation in the interpersonal problems circumplex (i.e., amplitude ≥ .15) early in cognitive therapy.
Placement of a trait scale in the interpersonal circumplex can be estimated from that scale’s correlations with measures of problems in each octant (see Figure 2). In particular, these eight correlations are transformed into four indices in the structural summary method (e.g., Gurtman & Balakrishnan, 1998; Wright et al., 2012). To the extent that the trait relates to problems with dominance and/or affiliation, the magnitude of the correlations of the trait scale with octant measures follow a sine curve reflecting the positions of the octants in the circumplex space, and the curve’s R2 fit value is higher. The curve’s elevation is the target scale’s average correlation with the interpersonal octant scales and reflects the extent to which the target scale correlates with nonspecific interpersonal problems or distress. The amplitude is the distance between this elevation and the peak (or nadir, if the elevation is negative) of the curve and marks the extent to which the target scale shows well-differentiated relations to specific types of interpersonal problems. Finally, the location of the curve’s peak (or nadir, whichever has the larger absolute value) indicates the target scale’s orientation or angle in the circumplex space. Considering a trait scale’s relations to the interpersonal circumplex, fit R2 ≥ .80, absolute elevation ≥ .15, and amplitude ≥.15 can be interpreted as substantive (Wright et al., 2012; Zimmermann & Wright, 2017).

Correlations of the trait aggression scale with the interpersonal problems scales in circumplex order.
The structural summary method of quantifying a trait scale’s relations to interpersonal style or problems has notable strengths. The amplitude and angle indices efficiently summarize a trait’s extent and type of interpersonalness, respectively. Moreover, because a trait measure’s amplitude and angle are separate from its elevation, the trait measure’s connection to a particular interpersonal style is theoretically separated from the level of nonspecific distress. This feature of the structural summary method applied to the IIP-C helps address important issues in personality assessment—namely, that scores on trait measures reflect both unstable state and stable trait variance (e.g., Griens et al., 2002; Morey et al., 2020; Noordhof et al., 2018; Zimmermann et al., 2017), such that controlling state effects may increase the validity of trait measures (e.g., Clark et al., 2003; Naragon-Gainey et al., 2013; Vittengl et al., 2014). The structural summary method applied to the IIP-C helps differentiate state and trait variance because the IIP-C’s general distress factor (reflected in elevation) is notably more state-like (i.e., changeable) than dominance and affiliation factors (reflected in amplitude and angle) with trait-like stability over time (Renner et al., 2012; Vittengl et al., 2003). In this context, we considered the consistency of trait measures’ relations to interpersonal style both before and after CT for MDD, during which decreases in depressive states are often large (e.g., Craighead et al., 2015).
Past research addressing associations of SNAP-2 and related trait measures with the interpersonal problems circumplex informed the current study. The SNAP-2 trait scales display a three-factor structure, anchored by negative temperament, positive temperament, and disinhibition (Clark et al., 2014). Research with other instruments suggests that negative temperament/neuroticism scales and facets tend to show high elevation (i.e., more interpersonal problems, in general), variable amplitude (i.e., differing amounts of interpersonalness), and angles in the exploitable to submissive (e.g., for scales tapping depression and anxiety) or domineering to vindictive (e.g., for scales tapping aggression and hostility) ranges of the circumplex (e.g., Du et al., 2020; Hopwood et al., 2013; Wright & Simms, 2016; Wright et al., 2012). Positive temperament/extraversion scales and facets tend to show low elevation (i.e., fewer interpersonal problems, in general), high amplitude (i.e., substantive interpersonalness), and angles in the overly nurturant to intrusive range of the circumplex. Finally, disinhibition/low conscientiousness scales and facets show variable elevation, amplitude, and angles ranging considerably from intrusive to vindictive portions of the circumplex.
Research testing relations of the SNAP-2 scales with the interpersonal problems circumplex has been limited. In an earlier sample of outpatients with recurrent MDD who received CT, the IIP-C scales were submitted to factor analysis and a three-factor solution representing nonspecific interpersonal distress, dominance, and affiliation was retained (Vittengl et al., 2003). At the end of CT, most SNAP-2 scales correlated > .30 (range = |.27| to |.70|) with the nonspecific interpersonal distress factor; three scales—(low) aggression, dependency, and (low) detachment—correlated > .30 (range = |.33| to |.53|) with affiliation; and three scales—manipulativeness, exhibitionism, and impulsivity—correlated > .30 (range = |.31 to |.33|) with dominance. The latter two sets of correlations suggest that these six SNAP-2 scales may have a defined orientation in the circumplex space among MDD patients. Consistent with this possibility, in research relating the SNAP to a short form of the IIP-C in a normative undergraduate sample (Hopwood et al., 2009, 2013), five of these six SNAP scales (impulsivity was the exception) showed amplitude ≥ .15 and angles broadly consistent with research reviewed above for other personality measures. Important gaps in past research that are addressed in the current study include (a) formal interpersonal structural analysis of the SNAP-2 scales in a clinical sample and (b) evaluation of the stability of personality trait scales’ interpersonal structural summaries across treatment of MDD with CT.
Past research with the SNAP-2 (Vittengl et al., 2014) and IIP-C (Renner et al., 2012) in the current sample also informed our new analyses. From early to late in CT, the SNAP-2 temperament and trait scales showed moderate to high retest stability (median r = .77, range .66-.87). Even so, all SNAP-2 scales showed statistically significant change in mean levels (median |d| = 0.42, range 0.14-0.94), with the majority of change attributable to concurrent decreases in depressive symptoms (d = 2.27). The IIP-C factors nonspecific interpersonal distress (.54), dominance (.68), and affiliation (.74) also showed moderate retest correlations during acute-phase CT. However, nonspecific interpersonal distress decreased notably during acute-phase CT (d = 0.90), whereas dominance increased only slightly (d = 0.14) and affiliation did not change significantly (d = 0.03). Thus, interpersonal style was relatively stable during CT.
In sum, past research in a smaller sample provided initial estimates of the SNAP-2 scales’ projections in the interpersonal problems circumplex, and we examined replication of these findings in a larger CT sample (Jarrett & Thase, 2010). In addition, earlier analyses of the current sample showed aspects of both stability and change in SNAP-2 scales and, separately, also in the IIP-C scales. Here, we tested projections of the SNAP-2 scales in the interpersonal problems circumplex, and in novel analyses, the stability of the trait-dysfunction relations during CT for MDD. Testing stability of trait-dysfunction relations across a mood-changing intervention was relevant to theories that view trait-dysfunction systems as central to personality disorder.
We hypothesized that most SNAP-2 scales (specifically negative temperament, mistrust, manipulativeness, aggression, self-harm, dependency, (low) positive temperament, detachment, and propriety) would correlate with nonspecific interpersonal distress, represented by the elevation of the SNAP-2 scales in the circumplex space. Second, we hypothesized that six SNAP-2 scales (manipulativeness, aggression, dependency, exhibitionism, detachment, and disinhibition) would show a differentiated orientation in the circumplex space, represented by substantive amplitude of these SNAP-2 scales. Third, we hypothesized that the orientation of interpersonally differentiated SNAP-2 scales would be stable from early to late in CT, represented by minimal changes in angle.
Method
Participants
Data for the current analyses were drawn from the acute phase of a multiphase clinical trial (Jarrett & Thase, 2010). Participants were outpatients who were self- or practitioner-referred. Inclusion criteria were (a) written informed consent for evaluation and treatment; (b) met DSM-IV criteria for recurrent MDD (American Psychiatric Association, 2000); (c) previously remitted between depressive episodes, ≥1 depressive episode with complete interepisode recovery, or antecedent dysthymic disorder; and (d) 17-item Hamilton Rating Scale for Depression (HRSD; Hamilton, 1960) score ≥14. 1 Exclusionary criteria were (a) severe or poorly controlled concurrent medical disorders that could cause depression; (b) psychotic or organic mental disorders, bipolar disorder, active substance dependence, or primary obsessive–compulsive or eating disorders; (c) unable to complete questionnaires in English; (d) active suicide risk; (e) <18 or >70 years old; (f) previous nonresponse to ≥8 weeks of CT or 6 weeks of fluoxetine; or (g) pregnant or planned to become pregnant during the first 11 months after intake. The Structured Clinical Interview for DSM-IV (First et al., 1996) was used for psychiatric diagnosis. Patients (N = 351) who entered the study after the SNAP-2 was added to the acute-phase protocol are included in this report. Their age was M = 43.3 (SD = 12.4) years; 67.8% were women; 1.4% were Asian, 12.3% Black, 4.0% Hispanic, 80.3% White, and 2.0% other races/ethnicities; and they had completed M = 15.0 (SD = 3.0) years of education.
Acute-Phase CT
During the acute-phase, all participants were treated with CT only (Beck et al., 1979). The 12-week CT protocol lasted 12 weeks, with 2 additional weeks allowed for rescheduling, and included 16 or 20 CT sessions. For the first 4 weeks, patients received two sessions/week. Thereafter, patients with ≥40% reduction in HRSD scores received one session/week for 8 weeks (16 total sessions), whereas patients with less early improvement received two sessions/week for 4 weeks and then one session/week for 4 weeks (20 total sessions). 2 More sessions were provided to patients with less early improvement to increase their chances of response and eligibility for later phases of the clinical trial (see Jarrett & Thase, 2010). The cognitive therapists (N = 16) achieved and maintained Cognitive Therapy Scale (Young & Beck, 1980) scores ≥40 to demonstrate competence, submitted session videotapes for review, participated in group supervision weekly, and received feedback on weaknesses and strengths.
Measures
Schedule for Nonadaptive and Adaptive Personality–Second Edition (SNAP-2)
The SNAP-2 (Clark et al., 2014) is a 390-item, factor analytically derived self-report inventory that uses a true–false format to assess 15 dimensions relevant to normal and disordered personality (see Table 1). 3 In past research, the SNAP-2 scales have demonstrated good internal consistency (median alphas of .79 to .92 in student, adult, and patient samples), test–retest reliability (e.g., mean r = .87 over a mean interval of 49 days in normal adults), and discriminant validity (mean interscale r = ~|.20|; Clark et al., 2014). The SNAP’s three temperament-trait scales (negative temperament, positive temperament, disinhibition) reflect the instrument’s factor structure, and multiple studies have supported the scales’ validity for assessing personality traits across the adaptive–maladaptive spectrum (e.g., Morey et al., 2012; Pryor et al., 2009; Ready & Clark, 2002; Stepp et al., 2012; Vittengl et al., 2003). Participants completed the SNAP early (Week 1) and again late (Week 12) during acute-phase CT.
Descriptions of SNAP-2 Scales and Their Empirical Relations to Interpersonal Problems.
Note. Scale descriptions copyright © 2014, Lee Anna Clark. For each SNAP-2 scale, low and high scorers are at least 1 SD below and above the normative mean, respectively. Respondents with more extreme scores will match the SNAP-2 scale descriptions to increasing degrees and/or in more situations.
Inventory of Interpersonal Problems–Circumplex Version
The Inventory of Interpersonal Problems (Horowitz et al., 1988) contains 127 statements describing behaviors, thoughts, and feelings that may be problematic in significant relationships. Respondents describe their interpersonal problems by rating each item on a scale from 1 (not at all) to 5 (extremely). From these items, we scored the 64-item IIP-C (Alden et al., 1990) that includes eight, eight-item scales (overly nurturant, intrusive, domineering, vindictive, cold, socially avoidant, nonassertive, exploitable). As depicted in Figure 1, the IIP-C scales reflect progressive blends of dominance and affiliation in a two-dimensional structure. The IIP-C scales have demonstrated acceptable internal consistency (median alpha = .81, range .76 to .88) and test–retest reliability (median r = .73, range .56 to .83, over 1 week) in normative samples (Horowitz et al., 2000). The IIP-C scales’ validity is supported by moderate convergence with measures of depression (median r = .38, range = .33 to .43), anxiety (median r = .36, range = .31 to .39), and social adjustment (median r = .30, range .16 to .49; Horowitz et al., 2000). Principal components analyses have supported the circumplex structure of the IIP-C (Alden et al., 1990; Horowitz et al., 2000; Vittengl et al., 2003). Patients completed the IIP-C before (diagnostic intake) and approximately 1 week after acute-phase CT.
Depression
Patients completed the Beck Depression Inventory (Beck et al., 1961) and Inventory of Depressive Symptomatology–Self-Report (Rush et al., 1996), and clinicians administered the HRSD and Current MDD section of the Structured Clinical Interview for DSM-IV (First et al., 1996) at the diagnostic evaluation, weekly during CT, within 7 days of the last CT session, and any time a patient exited the protocol. We aggregated the Beck Depression Inventory, Inventory of Depressive Symptomatology–Self-Report, and HRSD to form a robust depressive symptom severity index, because these three measures mark the same symptom severity construct in acute-phase CT (Vittengl et al., 2005). Specifically, we standardized the measures based on their distributions at the intake assessment using a T-score metric and averaged them. Categorical treatment response was defined a priori as the absence of a major depressive episode and an HRSD score ≤12 at exit from acute-phase CT (Jarrett & Thase, 2010).
Statistical Analyses
We analyzed the intent-to-treat sample of patients who entered the clinical trial after the SNAP-2 was added to the assessment battery (N = 351). Early and late in CT, respectively, 1.5% and 31.8% of IIP-C data, and 7.2% and 34.5% of SNAP-2 data, were missing among these patients. Data were missing due to attrition from the study and patient assessment fatigue. To analyze data from all patients in the sample, including cases with missing data, we conducted analyses using full information maximum likelihood estimation and included pre-CT depressive symptom severity (no missing data), post-CT depressive symptom severity (19.7% missing), and CT response (no missing data) variables because these variables were available for greater proportions of the sample and correlated broadly with the SNAP-2 and IIP-C in the current sample (Vittengl et al., 2014, 2018).
In preliminary analyses, the IIP-C demonstrated circumplex structure in the current sample, replicating past research (Horowitz et al., 2000). Using a latent variable modeling approach, we fit strict circumplex models (equal spacing of the IIP-C octant scales around the circumplex and equal communalities; Wendt et al., 2019). The models included general distress factors, on which all scales loaded, plus dominance and affiliation factors, on which scales loaded according to their circumplex position (Figure 1). Circumplex fit was good in a model constraining pre- and post-CT factor loadings to be identical (goodness of fit index [GFI] = .962, comparative fit index [CFI] = .949, root mean square error of approximation [RMSEA] = .066). In a model allowing pre- versus post-CT loadings to vary, circumplex fit was similarly good (GFI = .964, CFI = .951, RMSEA = .064), and pre/post-CT factor congruence was high (Tucker coefficients > .999). Thus, circumplex structure appeared largely invariant over CT.
We computed correlations of each SNAP-2 scale with the eight IIP-C scales, both early and late in CT and used these correlations to compute the IIP-C structural summary of each SNAP-2 scale (Gurtman & Balakrishnan, 1998; Wright et al., 2012). In particular, we computed the R2 fit of the correlations to a sine function, the elevation (average correlation), amplitude (differentiation), and angle (orientation). We computed these structural summary values early and late in CT, as well as the change in the structural summary values. Changes in structural summary values were computed as the late-CT estimate minus the early-CT estimate, analogous to a paired-samples t test. We computed bootstrapped confidence intervals for elevation, amplitude, and angle, early and late in CT, and the parameters’ changes. Full information maximum likelihood correlations, 1,000 bootstrapped samples, and bias-corrected and accelerated confidence intervals were computed using the psych (Revelle, 2018) and boot (Canty & Ripley, 2019) packages in R 3.6.1 (R Core Team, 2019). We did not report bootstrapped confidence intervals for R2, based on their demonstrated inaccuracy (Zimmermann & Wright, 2017).
We focused on each SNAP-2 scale’s elevation, amplitude, and angle to quantify its relations to the IIP-C. For elevation estimates, as well as changes in elevation, amplitude, and angle, we considered the result statistically significant when the 95% confidence interval excluded zero.
Point estimates of amplitude are always positive because this quantity can be defined as a distance from the origin of the circumplex space. To estimate the expected value for amplitude when a trait scale has no real relation to the IIP-C, we conducted a Monte Carlo study. In particular, we drew 10,000 samples of N = 351 from a simulated multivariate normal population with zero correlations among three random variables, two representing independent dominance and affiliation dimensions of the interpersonal circumplex, and one representing a trait unrelated to this circumplex. In the simulation, we computed amplitude as the square root of (rdom2 + raff2), where rdom and raff are correlations of the trait scale with dominance and affiliation circumplex dimensions (Zimmermann & Wright, 2017), and examined the resulting distribution of amplitude values. The average or expected value of amplitude under the null hypothesis was .07; similarly, the median was .06. Consequently, when the 95% confidence interval for amplitude exceeded .07, we considered the result to be statistically significant.
In addition to statistical significance, we considered the observed magnitude of structural summary values. We considered absolute elevation and amplitude values ≥.15 to be substantive in size (Wright et al., 2012). Similarly, we viewed changes ≥.15 in elevation and amplitude from early to late in CT as substantive. We reported a SNAP-2 scale’s angle only when the scale showed a clear circumplex orientation marked by amplitude ≥.15, early and/or late in CT. In all instances where amplitude ≥.15, we observed fit R2 ≥ .80 in the current sample. Also consistent with past research (Hopwood et al., 2009), we interpreted changes in angle ≥ 23° (i.e., more than half of a 45° octant) as substantive because they would often suggest a qualitative shift in the predominant type of problem.
Results
The SNAP-2 scales’ interpersonal circumplex structural summary values, early and late in CT, appear in Table 2. See the appendix for changes in these values from early to late in CT.
Interpersonal Problems Structural Summary of SNAP-2 Scales.
Note. N = 351. 95% confidence intervals are in square brackets. Angles provided for a SNAP-2 scale only when amplitude ≥.15 early and/or late in cognitive therapy. Elevation and amplitude values that are significant at p < .05, two-tailed, appear in
The first hypothesis was that specific SNAP-2 scales (negative temperament, mistrust, manipulativeness, aggression, self-harm, dependency, (low) positive temperament, detachment, and propriety) would correlate significantly with nonspecific interpersonal distress, represented by the scales’ elevation values. As shown in Table 2, this hypothesis was supported for all nine scales both early and late in CT. Of these 18 statistically significant elevation values, 14 were substantive (i.e., ≥.15). In addition to the hypothesized relations, (low) exhibitionism and (low) entitlement also had statistically significant elevation early (but not late) in CT, but both values were relatively small (<.15). Only disinhibition and impulsivity did not show significant elevation, early or late in CT.
The second hypothesis was that particular SNAP-2 scales (manipulativeness, aggression, dependency, exhibitionism, detachment, disinhibition) would show significantly differentiated orientations in the circumplex space, represented by the scales’ amplitude. Considering these six SNAP-2 scales early and late in CT, 10 of 12 hypothesized amplitude values were statistically significant, and these same 10 amplitude values were all substantive in size (i.e., ≥.15), as shown in Table 2. Specifically, aggression, dependency, exhibitionism, and detachment had substantive amplitude both early and late in CT; manipulativeness and disinhibition had substantive amplitude early but not late in CT. In addition to the SNAP-2 scales hypothesized, positive temperament and entitlement also showed significant and substantive amplitude both early and late in CT.
For SNAP-2 scales with amplitude ≥.15 early and/or late in CT, we computed the angle in the circumplex space (see Table 2). Both early and late in CT, positive temperament was between overly nurturant (0°) and intrusive (45°); exhibitionism and entitlement were between intrusive (45°) and domineering (90°); manipulativeness, aggression, and disinhibition were between domineering (90°) and vindictive (135°); and detachment was between cold-hearted (180°) and socially avoidant (225°). These scales’ orientations early in CT are shown in Figure 1. Table 1 contains descriptions of the SNAP-2 scales’ observed relations to interpersonal problems.
Our third hypothesis was that the interpersonally differentiated SNAP-2 scales would show stable angles from early to late in CT. Among the eight SNAP-2 scales with amplitude ≥.15 early and/or late in CT (manipulativeness, aggression, dependency, positive temperament, exhibitionism, entitlement, detachment, and disinhibition), the median absolute change in angle was 8° (range 2° to 13°), well below the threshold of a substantive change of 23°. Moreover, none of the changes in angle was statistically significant, with the exception of the change in aggression’s angle from 118° to 130°. Nonetheless, aggression’s orientation remained in the domineering-vindictive octant of the circumplex both early and late in CT, as noted above.
The SNAP-2 scales’ elevation and amplitude values were also largely stable from early to late in CT. Two of 15 SNAP-2 scales (negative temperament and aggression) showed statistically significant increases in elevation, but neither change was substantive (≥.15). Moreover, no SNAP-2 scales showed statistically significant or substantive (≥.15) changes in amplitude (see Table 2 and the appendix).
Discussion
Clarifying relations between personality traits and dysfunction may help advance definitions and diagnosis of personality disorder. The current analyses tested correlations of the SNAP-2’s 15 scales with the IIP-C, both early and late in CT for recurrent MDD. Using the structural summary method, we found that most SNAP-2 scales correlated with the general level of interpersonal problems, and many SNAP-2 scales also corresponded with a predominant interpersonal style or type of problem. Furthermore, observed trait-dysfunction relations were largely stable across CT, when large decreases in depressive symptoms were common. These results expanded the literature showing robust trait-functioning connections relevant to psychopathology versus well-being.
The current results suggest that a wide range of personality traits are relevant to the general level interpersonal distress or dysfunction among adult outpatients with recurrent MDD. This broad finding replicates past research with other personality measures in different patient and normative populations (Du et al., 2020; Hopwood et al., 2013; Wright & Simms, 2016; Wright et al., 2012). On the SNAP-2, we found that all trait scales marking the first two factors were correlated with greater interpersonal problems, in general. That is, higher negative temperament and related traits (mistrust, manipulativeness, aggression, self-harm, eccentric perceptions, dependency) and lower positive temperament and related traits (exhibitionism, entitlement, [high] detachment) correlated with greater elevation in the interpersonal problems circumplex. Also consistent with past research, elevation findings for SNAP-2 scales loading primarily on the disinhibition factor were mixed. The SNAP-2 disinhibition and impulsivity scales did not correlate significantly with general interpersonal problems, whereas the SNAP-2 propriety and workaholism scales both correlated with greater problems.
Our analyses also revealed more specific relations of traits measured with the SNAP-2 with particular types of interpersonal problems, separate from the general level of interpersonal problems. Consistent with the SNAP-2 scales’ content (Clark et al., 2014) and past research (Vittengl et al., 2003; Hopwood et al., 2009, 2013), we found that high scorers on the entitlement and exhibition scales tended to report more interpersonal problems involving their domineering and intrusive behavior. Similarly, high scorers on aggression, manipulativeness, and disinhibition tended to report more problems involving their domineering and vindictive behavior. Positive temperament and detachment were oriented in roughly opposite directions in the circumplex, with high scorers on positive temperament reporting more problems involving their overly nurturant and intrusive behavior, whereas high scorers on detachment reported more problems with cold and socially avoidant behavior. Finally, high scorers on the SNAP-2 scale of dependency reported more problems due to exploitable and overly nurturant behavior.
The SNAP-2 scales were oriented in five of eight octants (and all quadrants) of the interpersonal circumplex. However, interpersonal problems coverage by the SNAP-2 scales was somewhat stronger in the domineering (top) half rather than in the submissive (lower) half of the circumplex, consistent with past research using different trait measures (e.g., Wright & Simms, 2016). The addition of trait scales relating to the five-factor model agreeableness dimension might improve coverage of the submissive half of the circumplex, primarily from 270° (nonassertive), through 315° (exploitable), to 360/0° (overly nurturant; Du et al., 2020), when this coverage is needed to meet assessment goals in particular clinical or research settings. Full assessment of the submissive portion of the interpersonal problems circumplex might be particularly valuable in studies of social anxiety or schizotypy, for example.
Few changes in the relations between SNAP-2 scales and interpersonal problems were statistically significant, and no changes were substantive in size, from early to late in CT. Thus, the trait-functioning connections were largely stable over a period of 4 months when many CT patients experienced large decreases in depressive mood. This pattern of results suggests that trait-functioning relations reflect an enduring personality system, as opposed to connections inflated or distorted due to loading with transient distress. Our findings in this regard are broadly similar to past research showing that, under some conditions, traditional personality disorder diagnoses can be made reliably during major depressive episodes (e.g., Morey et al., 2010), and stable trait variance can be partitioned from changeable state variance to improve the validity of personality assessment (e.g., Naragon-Gainey et al., 2013).
The nature of the sample, design, measures, and analyses limit conclusions from this study. The sample of adult outpatients with recurrent MDD was treated by trained and supervised cognitive therapists. Thus, generalization of findings to other patient, treatment, or normative populations is uncertain. Similarly, both personality traits and interpersonal problems were assessed with well-established self-report instruments. Different self-report measures of personality and interpersonal functioning or assessment methods (e.g., interviews, informant reports) might have produced different results. Although the sample was relatively large, especially for psychotherapy studies, the study was not designed to detect trait-functioning relations of specific magnitudes. Finally, several parameters were estimated for each SNAP-2 scale, both pre- and post-CT, increasing the probability of false-positive findings and the need for replication of the current findings in future research.
Because personality disorder involves both maladaptive-range traits and dysfunction (American Psychiatric Association, 2013; World Health Organization, 2019), clarifying connections between traits and functioning is important from both theoretical and clinical perspectives. The current results show that a well-established trait measure, the SNAP-2, has both broad and specific relations to interpersonal problems. The stability of these relations over time and mood shifts during treatment for depression supports the idea that such connections are attributable to operation of an enduring personality system rather than to transitory distress. Consequently, assessment with the SNAP-2 and IIP-C may advance research and clinical practice drawing on modern definitions of personality disorder (Krueger et al., 2014).
Footnotes
Appendix
Changes in Interpersonal Problems Structural Summary of SNAP-2 Scales.
| Scale | R2 fit | Elevation | Amplitude | Angle |
|---|---|---|---|---|
| Negative temperament | .52 | .14 [.06, .22] | .05 [.00, .17] | |
| Mistrust | .18 | .09 [.00, .18] | .03 [−.03, .11] | |
| Manipulativeness | −.02 | .07 [−.03, .17] | −.04 [−.14, .03] | −13 [−47, 16] |
| Aggression | −.02 | .12 [.03, .21] | −.03 [−.12, .04] | 12 [1, 25] |
| Self-harm | .24 | .08 [−.02, .18] | .04 [−.04, .13] | |
| Eccentric perceptions | .26 | .06 [−.03, .15] | .06 [−.02, .15] | |
| Dependency | .03 | .03 [−.05, .11] | −.02 [−.10, .05] | 3 [−19, 28] |
| Positive temperament | .00 | −.11 [−.19, −.03] | −.01 [−.07, .05] | 10 [−11, 30] |
| Exhibitionism | .07 | .07 [−.04, .17] | −.04 [−.13, .04] | 11 [−12, 32] |
| Entitlement | .01 | .02 [−.07, .12] | .06 [−.02, .14] | −6 [−33, 24] |
| Detachment | .03 | .04 [−.04, .12] | .05 [.00, .11] | −6 [−19, 7] |
| Disinhibition | .00 | .01 [−.09, .11] | −.06 [−.14, .01] | 2 [−46, 40] |
| Impulsivity | .00 | −.01 [−.12, .10] | .00 [−.08, .06] | |
| Propriety | −.02 | .05 [−.03, .13] | −.02 [−.10, .05] | |
| Workaholism | −.42 | .04 [−.07, .13] | −.06 [−.15, .03] |
Note. N = 351. Changes computed as late–early cognitive therapy. 95% confidence intervals are in square brackets. Angles provided for a SNAP-2 scales only when amplitude ≥ .15 early and/or late in cognitive therapy.
Declaration of Conflicting Interest
The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Dr. Vittengl is a paid reviewer for UpToDate. Dr. Clark is author and copyright holder of the Schedule for Nonadaptive and Adaptive Personality, second edition (SNAP-2), and its family of measures. Fees for commercial and funded, noncommercial usage licenses support student research; unfunded, noncommercial research and clinical usage licenses are required, but free of charge. Dr. Thase has consulted with and/or served on advisory boards for Alkermes, Allergan (includes Forest Laboratories), AstraZeneca, Cerecor, Johnson & Johnson (includes Janssen), Lundbeck, MedAvante, Merck, Moksha8, Otsuka, Pfizer Pharmaceuticals, Shire, Sunovion, and Takeda; he has received grant support from Alkermes, Allergan (includes Forest Laboratories), Assurerx, Johnson & Johnson, Takeda, the Agency for Healthcare Research and Quality, Patient Centered Outcomes Research Institute and the NIMH. He has equity holdings for MedAvante, Inc., and has received royalties from American Psychiatric Publishing, Inc. (APPI), Guilford Publications, Herald House, and W.W. Norton & Company, Inc. Dr. Thase’s spouse is an employee of Peloton Advantage, which does business with several pharmaceutical companies. Dr. Jarrett is a paid consultant to the NIH, NIMH, and UpToDate. She has equity holdings in Amgen, Johnson and Johnson, and Procter and Gamble. Her medical center charges fees for the cognitive therapy she provides to patients.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This report was supported by grants K24 MH001571, R01 MH58397, R01 MH69619 (to Robin B. Jarrett, PhD) and R01 MH58356 and R01 MH69618 (to Michael E. Thase, M.D.) from the National Institute of Mental Health (NIMH). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIMH or the National Institutes of Health.
