Personality Functioning Scale: A Scale to Assess DSM-5’s Criterion A Personality Disorders

The Section III on Emerging Measures and Models included in the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders, introduces a hybrid alternative approach, dimensional-categorical, to diagnose personality disorders. The Criterion A establishes the assessment of the impairment in personality functioning in terms of two dimensions: self and interpersonal. The present study was aimed at developing a short scale to measure both dimensions. The sample was composed of 342 adults from Buenos Aires city and its outskirts, with ages ranging from 19 to 82 years old (M = 39.90, SD = 13.75). Data were gathered using the Personality Functioning Scale, developed in this study, as well as the Personality Inventory for DSM‐5 Brief Form, the Mental Health Continuum Short Form, and the Symptom Check List-27. A principal components analysis conducted on 28 items found 2 factors, interpersonal and self. Internal consistency, estimated by ordinal Alphas, achieved values between .92 and .86 whilst Cronbach’s Alphas were .88 and .87. Significant and positive correlations between the Personality Functioning Scale scores on the one hand, and the Personality Inventory for DSM‐5 Brief Form scores and the Symptom Check List-27 score on the other, were found. Negative correlations between PFS scores and the Mental Health Continuum Short Form were calculated. As a result, a short scale with adequate psychometric features, suitable to assess Criterion A in adult Argentinian population has been developed.

on Canadian community and clinical samples. A 4-factor-24-item model was confirmed by the factor analysis procedure. A second-order analysis obtained one general personality pathology factor. Regarding the scores' reliability, Cronbach's Alphas indicating internal consistency ranged from .60 to .92, and stability reliability found Pearson's r coefficients between .63 and .91. Significant differences in items' scores by group (community sample versus clinical sample) were verified, except for two items. However, authors maintained such items in the final version of the scale. Besides, correlations between SIFS' scores and the Criterion B domains, measured by PID-5, were positive and high (r = .49 to r = .81).
Finally, the unique study conducted on adolescents (12 to 18-year-old children) deserves a special mention (Goth, Birkhölzer, & Schmeck, 2018). It developed the Levels of Personality Functioning (LoPF-Q 12-18), composed of 97 items, whose features were analysed on clinical and community samples from Switzerland, Austria, and Germany. Exploratory factor analysis reported a 4-factor structure explaining 39.9% of the variance. Most of the items (72.2%) showed loadings higher than .30 in only one factor. Statistically significant differences were found in average scores obtained by a clinical sample versus a school sample. Internal consistency ranged from .76 to .92. Table 1 summarises features of the scales above mentioned. at developing a scale to measure personality functioning in terms of the self and interpersonal dimensions introduced in DSM-5. It is also aimed at analysing its psychometric features in order to make the scale suitable to be used in local population.

Method Design
A non-experimental and cross-sectional design was conducted.

Material Sociodemographic Survey
It gathers information on gender, age, educational status, and occupation.

Personality Functioning Scale (PFS)
28 items to assess impairment in personality regarding self and interpersonal functioning were developed. A 4-point Likert response scale from Totally false to Very true was employed.

Personality Inventory for DSM-5 Brief Form (PID-5-BF)
The short local and adapted version of PID-5-BF (Góngora & Castro Solano, 2017b) is composed of 25 items using a 4-point Likert response from Very false or Often false to Very true or Often true. Five major domains are measured: Negative Affectivity, Disinhibition, Detachment, Antagonism, Psychoticism. An exploratory factor analysis found five factors. Their internal consistency ranged between α = .86 to α = .90.

Mental Health Continuum Short Form (MHC-SF)
MHC-SF (Keyes, 2005 (Casullo & Pérez, 2008;Derogatis, 1983). It is composed of 27 items responded by a 5-point Likert scale (Not at all to Extremely), assessing psychopathological unspecified symptoms suffered in the last week. Its 1-dimension structure was verified by a confirmatory factor analysis, and its internal consistency, estimated by Cronbach's Alpha was .92.

Procedure
Considering previous studies and the DSM-5 guidelines, items to assess impairments in personality functioning in terms of self and interpersonal domains were developed. Five experts examined those items following recommendations for short scales' development (Stanton, Sinar, Balzer, & Smith, 2002). 28 items were selected according to how well they met DSM-5's criteria (American Psychiatric Association, 2013a). Later, a pilot study, aimed at examining the items clarity, was run with 20 adults. Although all of the items remained, a few minor modifications in the statements were made. The Appendix contains the version developed in Spanish.
After conducting the procedures above described, data were gathered in individual sessions. Psychology undergraduates collaborated under the supervision of trained psychologists. Participants volunteered and no retribution was given, neither economical nor academic. An informed consent signed by the participants guaranteed confidentiality of results and anonymity. Participants were told about the feasibility of ceasing participation at any moment of the process. Eight answer sheets, which showed many non-responded items were left out.

Data Analysis
Construct validity evidences of the scale were examined by means of a minimum rank factor analysis. A polychoric covariance matrix was employed since the response scale was ordinal (Muthén & Kaplan, 1985).
Such analyses were conducted using the FACTOR software (Lorenzo-Seva & Ferrando, 2013). Additionally, Cronbach's Alphas were also determined. The Attenuation Index was used to estimate which percentage of the theoretical internal consistency decreased when Cronbach's Alpha was calculated (Dominguez-Lara, 2018).
Pearson's r coefficients between the scale´s scores, the short version of the PID-5 scores, the SCL-27 score and de MHC score were calculated, in order to analyse criterion validity evidences. The effect sizes of the correlations were examined regarding the following cut-off points: small = .10, medium = .30, large = .50 (Cohen, 1988

Construct Validity
A minimum rank factor analysis with a direct oblimin rotation was performed, in order to examine construct validity evidences. Prior to that, a polychoric correlations matrix was generated. The parallel analysis method (Horn, 1965) suggested the retention of two factors. Items 14 and 27 were excluded since they showed Stover, Liporace, & Castro Solano cross-loadings that are less than .10, different from the loading on the factor of interest. The sample size was adequate for the calculations which were performed (KMO = .923; Bartlett's Sphericity Test: χ 2 = 3389.1, df = 325; p < .01). The explained total variance was 51.33%, mostly by the first factor (43.73%), whereas the second factor explicative percentage was, comparatively, much lower (7.59%). Regarding the content of the items, factors were identified as Interpersonal (F1) and Self (F2). Table 2 shows factorial loadings higher than .40 in bold. The correlation between factors was positive and large (r = .663; p < .01).

Internal Consistency
Internal consistency was adequate either when it was calculated by ordinal Alphas or when Cronbach's Alphas were used, resulting higher with the former (Table 3). The Attenuation Index was below 30% (Dominguez-Lara, 2018). That indicates that both Alpha coefficients are dependable.

Criterion Validity
Positive and large associations among the five PID-5 BF domains and the PFS dimensions were found, except for the association between the Interpersonal dimension and the Negative Affectivity dimension, which was medium. As for SCL-27 score, positive associations were reported: medium with the PFS Interpersonal dimension and large with the PFS Self dimension. The correlations between PFS scores and the MHC scores were significant and negative. They were small with Social Well-being and moderate with the remaining ones (Table 4).

Discussion
The study was aimed at developing a short scale to assess impairments in personality according to the Criterion A introduced in the DSM-5's Section III (American Psychiatric Association, 2013a). Based on Morey's studies (Morey, 2017) and following Stanton's recommendations for short scales' development (Stanton et al., 2002), 28 items were examined by a minimum rank factor analysis, obtaining a 26-item-2-factor structure. The exam of the content of the items composing the factors suggested that there is a correspondence between the impairment dimensions related to the Self and to the Interpersonal Relationships. The number of factors has been a controversial issue since 1-factor, 2-factor, 4-factor and 8-factor structures were reported (e.g., Zimmermann et al., 2019). DSM-5 introduces two major dimensions (Self and Interpersonal). The Self-Functioning one is composed of two lower rank dimensions (Identity and Self-Direction), whilst the Interpersonal Functioning major dimension involves the Empathy and Intimacy minor dimensions. However, such classification was only verified by Goth et al. (2018) and Gamache et al. (2019).

Stover, Liporace, & Castro Solano
The present study, like those conducted by Hutsebaut et al. (2016) and Weekers et al. (2019), found two factors, following the estimations of the parallel analysis (Horn, 1965). Besides, the fact that double factorial loadings were absent must be highlighted since such issue had been negatively pointed out by Goth et al. (2018) and Siefert et al. (2019) about their results. Consequently, construct validity evidences for the PFS scale seem adequate since they are theoretically coherent with DSM-5, and because of the psychometric quality of the items as well. Although two factors were found, the correlations between them were large. Hence, further confirmatory studies should examine the likelihood of a unique underlying factor as reported by Morey (2017) and Hopwood et al. (2018).
Regarding scores' reliability, internal consistency was estimated by means of two statistics: Cronbach's Alpha, following previous studies, and ordinal Alpha coefficients which is more appropriate since data were ordinal.
Both estimators reached adequate values (Cortina, 1993). Such result represents an improvement regarding previous studies run on short scales, where those coefficients were barely up to .65 (e.g., Gamache et al., 2019;Hutsebaut et al., 2016).
As for criterion validity evidences, correlations between EFP and PID-5-BF were large. Significant and high values verify the link between Criterion A and Criterion B. Such results are reasonable since they correspond to the main criteria to diagnose PDs according to the DSM-5's Section III. These findings add strength to the evidence previously reported by Hopwood et al. (2018) and by Gamache et al. (2019) about correlations higher than .50 between scales measuring Criterion A and Criterion B. Associations between SCL-27 score and the EFP major dimensions scores were analysed. Pearson´s r with the EFP Self dimension score was large, whilst it was medium regarding EFP Interpersonal dimension score. Two issues should be underlined.
On the one hand, the SCL-27's items content does not assess symptoms involving interpersonal relationships.
On the other, values follow the same correlational pattern between LPFS-BF and BSI scores (BSI is a scale similar to SLC-27; Hutsebaut et al., 2016). Beyond the differences among the values achieved by Pearson's r coefficients between PFS dimensions scores and the SCL-27 score, evidence supporting the link between psychopathological distress experienced during the last week and personality dysfunctions was verified. The PFS obtained negative correlations with the MHC scores. As reported by Huprich et al. (2018) and Siefert et al. (2019), psychological well-being is usually the opposite pole of impairments in personality functioning.
About the weaknesses of the study, the use of a convenience community sample restricts the conclusions' generalisation. Even though DSM-5 (American Psychiatric Association, 2013a) states that the dimensional model is useful to assess either individuals from the community population or suffering from personality dysfunctions, adding a clinical sample would have been optimal in order to test such assertion. Besides, even when the answer sheets were carefully reviewed, the lack of validity scales arises as an issue since it might generate biased responses.
As regards of the limitations above described and taking into consideration that DSM-5 was originally aimed at diagnosing psychopathological disorders, further studies should be focused on analysing psychometric features of the PFS in clinical samples. Replicating the studies on the structure of the scale in other countries in Latin America would also be a matter of interest, as well as the development and adaptation of such scale in other languages.
In sum, a scale to assess Criterion A, suitable to be used in local population given their linguistic adaptation to the Spanish spoken in Argentina, with adequate validity evidences and appropriate reliability results has been Personality Functioning Scale developed. This is the first scale in Spanish developed to assess Criterion A, as a first step in the analysis of the new model to assess PDs presented in the DSM-5's Section III (American Psychiatric Association, 2013a).
From the methodological standpoint, the ordinal nature of items was taken into consideration, calculating polychoric correlational matrices and ordinal Alphas, thus ameliorating previous studies' weaknesses. As for clinicians, 28 items enable shorter testing sessions as well as easier and faster scoring processes compared to longer versions, involving from 80 to 132 items (e.g., Hopwood et al., 2018;Huprich et al., 2018;Morey, 2017).
The theoretical contribution here presented, referred to the emerging models introduced in DSM-5 deserves to be pointed out since the notion of impairments in personality functioning according to Criterion A has been analysed in a Latin American sample for the first time.

Funding
The authors have no funding to report.