It is widely recognized that the categorical approach to PDs in DSM-5 [3] has serious flaws. However, with the introduction of DSM-5, an alternative and dimensional model of PDs based on pathological personality traits and personality dysfunction is provided, which people are free to choose. The PID-5 [4] is currently the primary instrument to assess the five trait domains and 25 maladaptive personality trait facets of the DSM-5 AMPD. This 220-item inventory has shown adequate psychometric properties in clinical and nonclinical samples, in different age groups and in different countries [1]. Recently, an abbreviated form of the PID-5 with 100 items has been developed [33]. The goal of the present study was to investigate the reliability, structure, and criterion validity of the PID-5-SF in two Norwegian samples. In the first sample, the PID-5-SF was derived from the original PID-5, whereas in the second sample – the replication sample -, the PID-5-SF was used as a standalone instrument to obtain validity estimates that are not affected by biases caused by scoring the two forms from the same administration (cf. [45]).
The score reliability of the Norwegian PID-5-SF was overall good in terms of internal consistency, mean inter-item correlations, and mean item-total correlations. In the derivation sample, the mean alpha coefficients were .87 (domains) and .80 (facets), respectively. In the replication sample, the mean Cronbach’s alphas were .87 for the domains and .79 for the facets, respectively. This is remarkable given the small number of items per scale and aligns with previous findings [9, 10, 33]. However, in the present investigation, comparatively low internal consistencies were found for Perceptual dysregulation and Irresponsibility (.60 and .61 in the derivation sample and .66 and .59 in the replication sample, respectively). A similar alpha for the Irresponsibility scale of the PID-5-SF (.63) was reported by Bach et al. [9, 10].
The factor structure of the Norwegian PID-5-SF used as a standalone instrument showed similarity with the original PID-5 form. The factor congruence coefficients were .88 (Negative affectivity), .86 (Detachment), .87 (Psychoticism), .80 (Antagonism), and .88 (Disinhibition) with an average of .86. According to Lorenzo-Seva and Ten Berge [32], congruence coefficients in the range .85-.94 indicate fair similarity, and factors can be assumed equal when the values are above .95. Thus, the results suggest that the factors obtained in the analyses of the short and original Norwegian PID-5 displayed adequate similarity with the exception of Antagonism. Overall, fairly high factor congruency coefficients of the Norwegian PID-5-SF with the original PID-5 and the PID-5-SF in the US [30, 33] were found. Some scales of the PID-5-SF had their highest loadings on other factors than expected from the proposed structure of the inventory [30]. In both samples, Rigid perfectionism loaded on Negative affectivity (instead of Disinhibition) and Perseveration on Disinhibition (instead of Negative affectivity). Further, in the derivation sample, Suspiciousness loaded on Psychoticism (instead of Detachment or Negative affectivity) and Attention seeking on Detachment (instead of Antagonism). In the replication sample, Intimacy avoidance and Withdrawal loaded on Psychoticism (instead of Detachment) and Attention seeking on Disinhibition. However, these deviations have previously been observed in studies on the PID-5. Rigid perfectionism has repeatedly shown to load on Negative affectivity [12, 13, 15, 34, 43, 55]. In the Wright and Simms [55] study on the PID-5 and related measures, Perseveration loaded on Disinhibition almost as high as on Negative affectivity (.35 and .37, respectively). With regard to Suspiciousness, Bastiaens et al. [12, 13] found that this facet loaded nearly equally high on Psychoticism, Negative affectivity and Detachment. As in the present study, Attention seeking loaded about equally high on Detachment (low) and Antagonism in the investigation by Wright and Simms [55]. Substantial cross loadings of Intimacy avoidance and Withdrawal on Psychoticism have been previously reported by Maples et al. [33] and Wight and Simms [55]. Maples et al. [33] also found that Attention seeking loaded on Disinhibition.
The criterion validity of the PID-5-SF was investigated by examining the relationships with the dimensions of the FFM and dysfunctional beliefs associated with the DSM-IV/DSM-5 PD categories. Further, the similarity of these associations between the original form of the Norwegian PID-5 and the short form was examined to test if the nomological network of the original PID-5 is maintained by the short form (cf. [33]). In line with previous studies on the PID-5 and FFM (e.g., [18, 23, 55]), the PID-5 domains of the original and short form were strongly associated with the FFM dimensions in both samples: Negative affectivity with Neuroticism, (low) Detachment with Extraversion, (low) Antagonism with Agreeableness, and (low) Disinhibition with Conscientiousness. In the present study, Psychoticism was significantly related to Openness, but showed also significant associations with the remaining FFM dimensions. Findings on the relationships between Psychoticism and Openness have been mixed so far. In accordance with the results of the current study, Thomas et al. [48] and De Fruyt et al. [19] reported significant Psychoticism-Openness associations in student samples. On the other hand, several other studies have found only weak or near zero correlations between Psychoticism and Openness (e.g., [41, 51, 59]). Importantly for the purpose of the present study, when used as a standalone instrument, the profile agreement of the PID-5-SF with the original form across the FFM-dimensions was high with a mean of .94.
Further, strong conceptually meaningful associations between the PID-5 scales of the original and short form and pathological personality beliefs were found in both samples. For example, paranoid beliefs were strongly related to Suspiciousness and Schizoid beliefs to Intimacy avoidance. Antisocial beliefs predicted highly Callousness and Deceitfulness. Borderline beliefs had significant relationships with PID-5 facets from all domains, but were especially associated with Depressivity, Anxiousness, Anhedonia, Emotional lability, and Suspiciousness. Histronic beliefs were associated with Attention seeking. Narcissistic beliefs predicted primarily Grandiosity. Avoidant beliefs were most strongly related to Depressivity and Anxiousness. Dependent beliefs were primarily associated with Separation insecurity. Obsessive-compulsive beliefs were a strong predictor of Rigid perfectionism. These results are in line with the findings of Hopwood et al. [24, 25] and suggest that the cognitive perspective on PDs can be integrated with the DSM-5 section III trait model. In the replication sample, the profile agreement of the original and short form of the PID-5 was high, averaging .70 for the PID-5 domains and .61 for the PID-5 facets. It should be noted that the profile agreement was very low or even negative for several scales, including Hostility, Restricted affectivity, Perceptual dysregulation, Deceitfulness, Manipulativeness, and Risk taking.
Taken together, the findings of the present study regarding reliability, structure, and criterion validity suggest that the Norwegian PID-5 short form is a parsimonious, overall internally consistent, and structurally valid measure of the trait criterion of the DSM-5 AMPD. Fairly similar factor structures of the original PID-5 and the PID-5-SF, and, for the majority of scales, similar associations with external criteria suggest that the knowledge base that has been built around the original PID-5 can be largely applied to the shortened version. These results are in accordance with and supplement the findings of previous investigations on the PID-5-SF [9, 10, 33] and support its use in research and clinical practice. The brevity of the PID-5-SF, while retaining the comprehensiveness of the original version, makes it easier to include the pathological personality traits of the DSM-5 AMPD in clinical assessment. Widiger and Samuel [52] recommended for the assessment of the DSM-IV-TR PDs to use first a self-report inventory for screening purposes, followed by a structured interview. In a similar way, the PID-5-SF can serve as a short screening instrument used prior to an interview-based assessment, e.g., the structured interview that is currently being developed for the assessment of the traits system (criterion B) along with rating of functioning (criterion A; [22]). Although concerns regarding the clinical utility of the DSM-5 AMPD have been raised when the model was developed [58], findings support its clinical usefulness and acceptability in routine clinical practice. In a field trial of the DSM-5, the clinical utility ratings of the proposed diagnostic criteria for PDs were among the highest [39]. The pathological traits of the DSM-5 AMPD have been found to be superior to the DSM-IV-TR/DSM-5 PD categories with respect to clinicians’ ratings of ease of use, communication with patients, usefulness for describing an individual’s personality problems and global personality, and treatment planning [38]. Furthermore, the DSM-5 AMPD predicts treatment decisions (e.g., level of treatment, type of psychotherapeutic or pharmacological treatment) better than the DSM-IV-TR/DSM-5 PD categories [36]. Examples of how the DSM-5 AMPD can be used in clinical practice are provided by Skodol, Morey, Bender, and Oldham [44] and Bach, Markon, Simonsen, and Krueger [11].
A limitation of the present study is the use of a convenient nonclinical sample consisting of university students. This group is obviously rather homogeneous with respect to age, educational level, and socioeconomic status. Although the DSM-5 AMPD personality traits are assumed to be continuously distributed [3], the variance of the distribution of these traits is likely restricted in university student samples, which may affect the generalizability of the findings. Ideally, the present study is extended and replicated in more heterogeneous samples, including patients within mental health care. Another limitation of the current investigation is the relatively low sample size of the replication sample. Further, this study used only self-reported data, which may have involved a risk for artificially high correlations between measures due to shared method variance. Importantly, as few items of the original PID-5 and none of the PID-5-SF items are reversed scored and the items describe undesirable traits, these instruments are particularly prone to the effects of acquiescence responding and social desirability responding [7]. As a consequence, the alpha reliabilities and the associations with other self-report measures can be inflated [7]. It is therefore possible that the results of the present study would have been different if reports from multiple informants (e.g., spouse, parents, or siblings) had been available. More definitive findings would likely have been obtained if it had been possible to also administer structured interviews, informant-reports or clinician ratings of DSM-5 traits. Thus, we recommend that ongoing research on the Norwegian PID-5 use informant or clinician reports of DSM-5 traits, which are currently available and free to use [5, 38].