In this research, the sample was non-probabilistic, and the type of sampling was intentional. Clinical psychologists, psychiatrists, psychological and psychiatric care centers, registered in the phone book of the Yellow Pages in the city of Medellin (Colombia) were contacted and were asked to inform their patients about a research project that was underway with psychological and psychiatric patients in the city, which sought to validate a personality test. The criteria for selecting patients who would like to collaborate with the research were the following: (1) to be, at the time of selection under psychological, psychiatric treatment, or both; (2) to have attended minimum primary school; (3) to be between 18 and 60 years of age; and (4) not to be hospitalized at the time of selection.

The patients who agreed to cooperate with the research were informed about the project and its objectives and, after that, if they agreed to be part of the research, signed an informed consent and filled out a short socio-demographic survey. The patients also filled the MINI (International Neuropsychiatric Interview; Sheehan et al., 2000, 1998) to establish their clinical diagnoses. A total of 341 patients filled out the PID-5 (Krueger et al., 2012), of which 284 also filled the PBQ-SF (Butler et al., 2007) (table 2). One of the researchers (AF) returned the results of the evaluation of each patient to the clinical psychologists and psychiatrists who had sent them, so that they, in turn, could explain the results to their patients.

Table 2 Table 2 Demographic characteristics of the sample

Personality Inventory for DSM-5 (PID-5), Adult (Krueger et al., 2012). This is a 220-item test that evaluates 25 facets contained in 5 domains, which was translated into Spanish using the following procedure suggested for Ruiz, Gómez and Londoño (2001): two expert and certified translators were asked to do the translation of each of the 220 test items independently. Once the two translations were ready, a third translator was asked to assess which of the two translations was the best for each item. This translator could make remarks to the translations. In addition, if it was considered that none of the two translations reflected the original item, this translator could propose a new translation. The test was developed in Spanish under the judgment of this third translator, and then a fourth translator was asked to re-translate the test into English. Subsequently, two of the authors of this article (AF and NHL) compared the re-translation with the original version. This way, the final version of the PID-5 into the Spanish spoken in Colombia was obtained.

Each item of the PID-5 (Krueger et al., 2012) is graded on a scale of 0 (if the statement is "very false or frequently false") to 3 (if the statement is "very certain or with certain frequency"). Each facet contains from 4 to 14 items. The allocation of items in facets was made according to the proposal in Appendix B of Krueger et al. (2012) the supplementary material, as well as the grading of every facet (which is the arithmetic mean of it, in other words, the sum of the scores of all the constituent items of the facet, divided by the number of them). The grading of the domains was obtained from the arithmetic mean of the constituent facets of each domain. As there are 3 organizations of the facets in domains (Krueger et al., 2012; American Psychiatric Association, 2013; 2015), each domain has three grades, depending on the model of facet organization into domains (Table 1). It is important to clarify that for both the organization of facets in domains in Model 1 (Krueger et al., 2012), as in Model 2 (American Psychiatric Association, 2013a), the scores for restricted affectivity and rigid perfectionism were reversed, so that they could be included in the domains of negative affect and disinhibition, respectively.

Personality Belief Questionnaire, Short Form (PBQ-SF) (Butler et al., 2007). This test is the short version of the Personality Belief Questionnaire (Beck & Beck, 1991), a test developed to operationalize the beliefs identified by Beck (Beck et al., 1995, 2005). The Personality Belief Questionnaire, Short Form (PBQ-SF) (Butler et al., 2007), is a 65-item test that evaluates 10 beliefs associated with personality disorders. Each belief is evaluated with 7 questions, and each question is scored on a Likert scale of 0 to 4. The only questions that are repeated, and therefore are shared with other beliefs, are the beliefs of borderline personality disorder. The test evaluates the following beliefs: schizoid, paranoid, antisocial, narcissistic, histrionic, limit, avoidant, dependent, obsessive-compulsive and passive-aggressive. This test was validated in Colombian non-clinical population by Londoño, Calvete and Palacio (2012). The solution was very satisfactory with excellent adjustment rates, χ² (1854; N= 665)= 2720; p< 0.001; RMSEA= 0.027 (95% IC= 0.025; 0.029); CFI= 1. The internal consistency coefficients were appropriate, between 0.58 and 0.96.

After having the test translated, as described above, clinical psychologists, psychiatrists, psychological and psychiatric care centers listed in the Yellow Page directory in the city of Medellin (Colombia) were contacted and asked to refer their patients for the research. To do it, an event was held in a hotel in the city of Medellin, to which many of the psychologists and psychiatrists listed in the directory were invited. One of the researchers (AF) gave a lecture on the research project. After that, the professionals were invited to send those patients that met the following inclusion criteria, to participate in the project: to be undergoing psychological and/or psychiatric treatment, to have attended a minimum of primary school, to be between 18 and 60 years of age, and not be hospitalized at the time of selection. In the end, 341 patients filled out the PID-5 (Krueger et al., 2012), the MINI (International Neuropsychiatric Interview; Sheehan et al., 2000, 1998) and 284 of which also filled the PBQ-SF (Butler et al., 2007) (table 2). The patients’ tests were evaluated, and the results were delivered; after evaluating the tests, the results were delivered in a preset letter to the therapists, so that they could explain the results to their patients. Then, a patient database was developed, and the data were processed using the IBM SPSS Statistics 20.0 and LISREL 9.1

The structure of the PID-5 (Krueger et al., 2012) was assessed. Confirmatory factor analyses were conducted using LISREL 9.2 (Jöreskog & Sörbom, 2013) via Weighted Least Square (WLS) method using the polychoric matrix and the asymptotic covariance of the indicators. Due to the high number of items in the PID-5, three item-parcels were used as indicators of the 25 first-order latent variables (Little, 2013). Items were assigned to parcels after conducting an exploratory factor analysis with all the items corresponding to a latent variable, so that factor loadings were balanced within parcels. We used the procedure described by Little, Cunningham, Shahar, and Widaman (2002). Using the loadings as a guide, we started by using the three items with the highest loadings to anchor the three parcels. The three items with the next highest item-to-construct loadings were added to the anchors in reverse order. The highest loaded item from among the anchor items was matched with the lowest loaded item from among the second selections. If more items were available, the basic procedure was continued by placing lower loaded items with higher loaded parcels. This procedure was repeated for each first-order latent variable. Thus, in total 75 parcels were used as indicators of the 25 first-order latent variables.

Based on the above structure, several alternative hierarchical models were tested. Model 1 consisted of a hierarchical model in which five broader factors (negative affectivity, detachment, antagonism, disinhibition, and psychoticism) explained the associations among the 25 first-order factors. Model 2 was similar to model 1, but some first-order factors were allowed to load into two different second-order factors (American Psychiatric Association, 2013a). Finally, Model 3 consisted of a hierarchical model in which the above five second-order factors explained 15 first-order factors (emotional lability, anxiousness, separation insecurity, withdrawal, anhedonia, intimacy avoidance, manipulativeness, deceitfulness, grandiosity, irresponsibility, impulsivity, distractibility, unusual beliefs and experiences, eccentricity, perceptual dysregulation). Following the recommendations of several statisticians (Hu & Bentler, 1999; Little, 2013), the goodness of model fit was evaluated using the Comparative Fit Index (CFI), the NonNormative Fit Index (NNFI), and the Root Mean Square Error of Approximation (RMSEA). Generally, a good fit is indicated by CFI and NNFI values of 0.9 or higher, and RMSEA values lower than 0.06.

Table 3 Table 3 Factor loadings of Exploratory Factor Analysis (EFA) of items and Confirmatory Factor Analysis (CFA) of parcels

Table 3 displays the factor loadings obtained in a series of exploratory factor analyses with items of each facet. All factor loadings were higher than 0.40 but a few ones. The exceptions were items 11, 198 and 90 in Insensibility, items 3 and 195 in Risk-Taking, item 96(reverse) in Anxiety, item 142 (reverse) in Deceitfulness, item 201 in Irresponsibility, and item 20 in Withdrawal.

Next, based on the above item-parcels, we conducted a Confirmatory Factor Analysis, which indicated that the 25 first-order factor structure showed good adjustment to the data, χ²(2400, n = 341) = 2473, RMSEA = 0.001(90% CI: 0; 00.015), CFI = 1, NNFI = 1. Factor loadings ranged between 0.76 and 0.99. This model served as a baseline to test the hierarchical models of the PID-5. Table 3 displays the factor-loadings for item-parcels.

Next, three alternative hierarchical models were tested. The procedure proposed by Byrne (2012) was used. Model 1 consisted of five broader factors (negative affectivity, detachment, antagonism, disinhibition and psychoticism) that explained the associations among the 25 first-order factors. Adjustment indexes were excellent for this model, χ²(2665, n = 341) = 3384, RMSEA = 0.028 (90 % CI: 0.025; 0.031), CFI = 0.99, NNFI = 0.99. This model increased χ² significantly, ∆χ²(265,n = 341)=911, p < 0.001. However, the change in CFI was within the cut-off of 0.01 proposed by Cheung and Rensvold (2002), which indicates that the adjustment of both models is similar. Table 4 displays factor loadings for the second-order structures. All coefficients were adequate. Model 2 was similar to Model 1, but some first-order factors were allowed to load into two different second-order factors. This model showed excellent adjustment indexes, χ²(2661, n = 341) = 3350, RMSEA = 0.028 (90 % CI:0.025; 0.030), CFI = 0.99, NNFI = 0.99. When compared to the baseline model, this model significantly increased χ², ∆χ²(261, n = 341) = 877, p < 0.001.Once again, the change in CFI was within the cut-off. However, the functioning of some items in Model 2 was poor. The restricted affectivity (lack of) facet loaded negatively on negative affectivity and the hostility facet loaded negatively on antagonism. In addition, two factor loadings in the detachment domain were low: Depressivity (0.39) and Suspiciousness (0.33). Comparison between Model 1 and Model 2 indicated a significant ∆χ²(4, n = 341) = 34, p < 0.05. Finally, Model 3, consisting of 15 first-order factors explained by the five second-order factors, showed excellent adjustment indexes, χ²(920, n = 341) = 1550, RMSEA = 0.045 (90 % CI: 0.041; 0.049), CFI = 0.99, NNFI= 0.98. All second-order factor loadings ranged between 0.60 and 1. This model was not compared by changes in χ² because it was not nested into the other models.

Table 4 Table 4 Second-order factor loadings for hierarchical models

All Cronbach Alpha Coefficients (Table 5) of the PID-5 (Krueger et al., 2012) facets have a value greater than 0.7, ranging from 0.71 (irresponsibility) to 0.94 (eccentricity and depressivity). Domain coefficients also have excellent internal consistency, ranging from 0.87 (disinhibition Models 1 and 2) to 0.96 (detachment Model 1).

Table 5 Table 5 Cronbach Alpha of the facets and domains of the PID-5

The t test was performed to examine gender differences in the scores of the PID-5 (Krueger et al., 2012). There were no statitiscally significant differences in the scores for men and women, except in the following facets and domains, in which men had a score significantly higher than women: grandiosity (mean for men = 1.01, SD for men = 0.62, mean for women = 0.75, SD for women = 0.59, t = 3.95, p < 0.001, Effect Size = 0.43), irresponsibility (mean for men = 0.88, SD for men = 0.57, mean for women = 0.73, SD for women = 0.53 t = 2.53, p = 0.012, Effect Size = 0.27), manipulativeness (mean for men = 0.92, SD for men = 0.62, mean for women = 0.70, SD for women = 0.64, t = 3.27, p < 0.001, Effect Size = 0.35), risk taking (mean for men = 1.32, SD for men = 0.58, mean for women = 1.16, SD for women = 0.57, t = 2.50, p = 0.013, Effect Size = 0.28). , antagonism Model 1 (mean for men = 0.85 SD for men = 0.41 mean for women = 0.71, SD for women = 0.45, t = 2.80, p = 0.005, Effect Size = 0.32), disinhibition Models 1 and 2(mean for men = 1.30, SD for men = 0.37, mean for women = 1.21, SD for women = 0.39, t = 2.30, p = 0.022, Effect Size = 0.23), antagonism Model 2 (mean for men = 0.9, SD for men = 0.4, mean for women = 0.80, SD for women = 0.46, t = 2.16, p = 0.037, Effect Size = 0.23) and antagonism Model 3 (mean for men = 0.90, SD for men = 0.49, mean for women = 0.71, SD for women = 0.51, t = 3.40, p < 0.001, Size Effect = 0.38).

In two facets the scores for women were significantly higher than for men: emotional lability (means for men: 1.36, SD for men: 0.73, means for women: 1.56, SD for women: 0.76, t = -2.34, p = 0.020, Size Effect: -0.27), intimacy avoidance (means for men = 0.52, SD for men = 0.54, means for women = 0.76, SD for women = 0.73, t = -3.36, p < 0.001, Size Effect = -0.36).

To evaluate the concurrent validity of PID-5 (Krueger et al., 2012), the results were correlated with a recognized test that measures personality disorders as it is PBQ-SF (Butler et al., 2007). Cronbach Alpha Coefficients in this test were excellent, ranging from 0.71 (avoidant) to 0.9 (paranoid), which indicates good internal consistency. The correlations between the facets and domains of the PID-5 (Krueger et al., 2012) and beliefs in the personality disorders of the PBQ-SF (Butler et al., 2007) were very consistent with provisions of the alternative model of the personality disorders of the DSM-5 (American Psychiatric Association, 2013a). The traits with higher correlations within each personality disorder generally correspond to the predictions in the traits model of the DSM-5 (American Psychiatric Association, 2013a) (Table 6).

Table 6 Table 6 Correlations between Beliefs of Personality Disorders and facets and domains of PID-5 Note.

**Correlation is significant at the 0.01 level (bilateral)

* Correlation is significant at the 0.05 level (bilateral)

Once the models were confirmed using the confirmatory factor analysis, which showed the excellent internal consistency and concurrent validity of the PID-5 (Krueger et al., 2012), the percentiles required by the clinician or researcher to locate any test score related to the studied sample were developed (Colombian clinic) (Table 7).

Table 7 Table 7 Percentiles of the facets and domains of PID-5

Cont. Table 7 Cont. Table 7 Percentiles of the facets and domains of PID-5

Cont. Table 7 Cont. Table 7 Percentiles of the facets and domains of PID-5