Examining cross-cultural and cross-linguistic validity of the Personality Inventory for DSM-5 (PID-5) in a South Asian diaspora sample

The Alternative Model of Personality Disorders (AMPD) is an emerging approach to the conceptualization and diagnosis of personality disorders, developed with the aspiration of becoming a universal approach to personality disorder diagnosis and supplanting the current approach to diagnosis: Section II of the American Psychological Association’s (APA) Diagnostic and Statistical Manual of Mental Disorders (DSM). The AMPD was first introduced in Section III of the DSM-5 (APA, 2013). Generally, criterion A of the AMPD is meant to assess the severity of impairment in one’s self-functioning and interpersonal functioning; criterion B is meant to assess for the presence of pathological personality traits. As per the AMPD, personality pathology is characterized by five higher-order dimensional trait domains (Negative Affectivity [NA], Detachment [DET], Antagonism [ANT], Disinhibition [DIS], and Psychoticism [PSY]) which are each comprised of lower-order personality trait facets, of which 25 exist. To assess these personality traits, Krueger and colleagues (2012) developed the Personality Inventory for DSM-5 (PID-5) which is a 220-item self-report instrument of personality psychopathology. The items within this instrument directly correspond to the five higher-order domains and the 25 lower-order trait facets of the AMPD. A five-factor structure consistent with the AMPD conceptualization of personality psychopathology has been found through exploratory factor analyses with an American sample (Krueger et al., 2012). Watters and Bagby (2018) argue that while the higher-order structure of the PID-5 has a strong construct validity, the lower-order structure of the scale varies significantly across studies and is affected by interstitiality (i.e., the cross-loading of facets onto more than one). On its own, interstitiality within an instrument of personality traits is expected because of the heterogeneity and complexity of the construct. However, with respect to the PID-5, the factor loadings of these interstitial facets vary across samples. Watters and Bagby (2018) found evidence of a significant decrease in interstitiality when potential variability in sampling error was offset by combining multiple samples across different studies. This suggests that sample characteristics (e.g., sampling variability and sampling error) likely influence the degree of interstitiality observed. In a meta-analysis conducted by Somma and colleagues (2019), the replicability of the five-factor structure of the PID-5 was examined among many of these translations using factor analytic-based analyses. The authors found support for the replicability of the five-factor solution across both U.S. and non-U.S. samples, showing that the PID-5 factor structure is generalizable and replicable across many diverse populations. The findings presented by Somma and colleagues (2019) showcase the meaningful advancements in the field of personality psychopathology and assessment, as the findings of these studies build our understanding of the construct validity of the PID-5 as well as the AMPD across different samples, countries, and languages at the global scale. In a similar vein, Sorrel and colleagues (2021) conducted a study in which they examined the cross-cultural measurement invariance of the PID-5 in five samples from four European-Western countries (Belgium, France, Spain, and Switzerland) with a combination of undergraduate and community samples (N = 4,380). These five samples additionally differed in language as well, with three samples consisting of French-speaking participants, one containing Castilian-speaking individuals, and the final one consisting of Catalonian-speaking persons. The data from all five samples supported both configural and metric invariances of the PID-5, which provided support for consistency in factor structure across the cultures and languages, with support for the traditional five-factor solution (Sorrel et al., 2021). The authors also found support for partial scalar invariance, suggesting that several item intercepts are equal across groups, with variety in intercepts being explained by cultural and other demographic differences. Cross-cultural differences in latent means were smallest for Psychoticism, followed by detachment, with disinhibition being the domain with the largest cross-cultural differences (Sorrel et al., 2021). Overall, the authors conclude that the PID-5 structure remains complex with multiple cross loadings across domains. It is crucial to note that the samples included in both Somma et al.’s (2019) and Sorrel et al.’s (2021) studies were primarily from European or North American countries; this limits the overall generalizability exploration of the PID-5 factor structure. In fact, Somma and colleagues (2019) note that the cross-cultural replicability of the PID-5 factor structure remains limited due to the minimal, yet growing, research on this topic in non-Western cultures. Emerging research shows that other factor structures are deemed to be a better fit relative to the traditional five-factor solution for persons living in countries beyond North America (Lau et al., 2024; Montes & Sanchez, 2023; Zhang et al., 2021) as well as for persons of colour living in a Western country (Bagby et al., 2022), although the growing evidence base on this topic remains mixed even in non-Western countries (Al-Attiyah et al., 2017; Amini et al., 2021). To date, several translations of the PID-5 exist including Arabic (Al-Attiyah et al., 2017), Farsi (Lotfi et al., 2018), French (Roskam et al., 2015), Italian (Fossati et al., 2013), Dutch/Flemish (De Clercq et al., 2011), German (Zimmerman et al., 2014), Danish (Bo et al., 2016), Norwegian (Thimm et al., 2017), Hungarian (Labancz et al., 2020), Spanish (Gutiérrez et al., 2017), Portuguese (Pires et al., 2017), Czech (Riegel et al., 2018), and Polish (Rowiński et al., 2018). Al-Dajani and colleagues (2016) comment on the existing translations of the PID-5 in relation to the Standards of Educational and Psychological Testing (American Educational Research Association et al., 2016); they argue that merely translating the PID-5 from English into another language does not necessarily preserve the construct validity of the instrument. In doing so, we impose the pathological personality dimensions determined by Western tradition onto other cultures and assume that they hold the same relevance and deeper meaning across different cultures and language structures respectively. Hence, it is critical to empirically establish the psychometric properties of translated versions of the PID-5 by comparing it against the original English version. This ensures that the construct being measured reliably transfers between versions of PID-5. Although several different studies have reported that the five factors correspondent to the AMPD can be recovered in the PID-5 factor structure in different languages, only one to date has attempted to directly disentangle language from culture. Leclerc and colleagues (2025) conducted a study in which data from 768 bilingual participants who were administered the PID-5 in French and in English 1-2 weeks apart were explored; the final dataset consisted of 605 participants. The authors used this dataset to explore whether linguistic measurement invariance and stability-equivalence are achieved in this within-subjects study for two language versions of the PID-5. The authors found support for their aims as the PID-5 reached full measurement invariance and produced high stability equivalence. Leclerc and colleagues (2025) results indicate that the PID-5 is linguistically equivalent across the English and French languages with reliable stability when completed at different points in time. For the AMPD to demonstrate cultural equivalency across both Western and non-Western countries, cultures, and languages, further translations are needed especially for unique languages that are diverse in nature relative to existing translations of the scale. The within-subjects methodological design applied by Leclerc and colleagues (2025) explores the question of whether cultural or linguistic factors (or both) impact how respondent’s approach PID-5 items; this can provide support for the interpretations of PID-5 scores in an interchangeable manner. In a similar vein, our goal in the present study was to extend the examination of the cross-cultural and cross-linguistic replicability of the PID-5 factor structure in an English-Urdu bilingual population in Canada. Urdu is the national language of Pakistan – a collectivistic nation with a current population of approximately 234 million citizens (Worldometer, 2022); it is also one of the official state languages in India. In addition to Pakistan and India, the Urdu language is spoken by many individuals in counties all around the globe including the United Arab Emirates, Canada, Germany, the U.S., South Africa, Australia, Ireland, and the United Kingdom (New World Encyclopedia contributors, 2023). It is the tenth most widely spoken language globally with over 231 million speakers of the language (Eberhard et al., 2023). Several Pakistani research teams have translated shorter forms of the PID-5 from English into Urdu and administered them forensic samples within Pakistan. For instance, Ishfaq and colleagues (2022) set out to develop an Urdu translation of the PID-5 short form to explore its construct validity among 361 convicts in Pakistan, and another 509 convicts in a later study (Ishfaq & Kamal, 2024). The authors note several limitations of their data collection strategies employed in these studies, including the fact that some participants were illiterate; therefore, literacy teachers and other convicts were involved in administering and collecting test responses. This deviation from standard test administration brings into the question the psychometrics of the translated test including its vali

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.