The Agreement between Clients’ and their Therapists’ Ratings of Personality Disorder Traits

Abstract

Objective: Treating clinicians provide the majority of mental health diagnoses, yet little is known about the validity of their routine diagnoses, including the agreement with clients’ self-reports. This is particularly notable for personality disorders (PDs) as the literature suggests weak agreement between therapists and clients. Existing research has been limited by a focus on PD categories and brief therapist-report measures. Furthermore, although self-reports of PD have been criticized for under-reporting, very few data have compared them to therapist-report in terms of mean-level. Method: We addressed these limitations by collecting dimensional trait ratings from 54 therapist-client dyads within outpatient clinics. The clients (52% women, 94% Caucasian, 39.8 years) provided ratings of dimensional PD traits via the Personality Inventory for DSM-5 (PID-5) while therapists (72% female, 89% Caucasian) completed the Informant version of the same measure. Results: Employing systematic measures of traits yielded higher rank-order agreement than observed in prior studies, with a median correlation of .41 across the PID-5 domains. Most interestingly, mean-level comparisons indicated that clients reported significantly higher levels of PD pathology than did their therapists. This effect was most notable for the domain of Psychoticism, which had the lowest rank-order agreement (r = .16) and the largest mean-level discrepancies. Conclusions: When clinicians utilized systematic measures of dimensional traits their agreement with client was higher than reported in past studies. Furthermore, clients reported significantly more PD pathology than was noted by their therapists suggesting concerns about invalid self-reports due to under-reporting have been overstated.

The Personality Inventory for ICD-11: Investigating Reliability, Structural and Concurrent Validity, and Method Variance

“…Next, every measure used in the present study was based on self-reports, and therefore the relations between scale scores were likely inflated due to shared method variance (Campbell & Fiske, 1959), in which the use of different methods may result in slightly different convergent and discriminant validity results. The recently developed informant version of the PiCD (Bach et al, 2019) could be used in order to leverage the advantages of multisource assessment (e.g., separating trait and method variance; more information regarding a target's personality pathology across different contexts; Campbell & Fiske, 1959) and investigate questions related to self-other agreement (e.g., Samuel, Suzuki, Bucher, & Griffin, 2018;Sleep, Lamkin, Lynam, Campbell, & Miller, 2018).…”

Focusing Narrowly on Model Fit in Factor Analysis Can Mask Construct Heterogeneity and Model Misspecification: Applied Demonstrations across Sample and Assessment Types

“…The issues of items being misassigned to factors or failing to be clear indicators of a single factor may be particularly salient when examining structural models of symptoms, given that different symptom experiences often are closely interrelated (Clark & Watson, 2019;Kotov et al, 2017). For example, items assessing "having thoughts that don't make sense to others" often are used to score thought disorder scales, but individuals with high levels of internalizing psychopathology may have worries that seem irrational to others and may strongly endorse these items as a result (Samuel et al, 2018). In such cases then, a two-factor model consisting of thought disorder and internalizing dimensions potentially could fit well if an item assessing "thoughts that don't make sense to others" is allowed to load only on thought disorder, even when such an item could have a loading of equal or stronger magnitude on internalizing.…”