1 Faculdade de Medicina, Universidade Luterana do Brasil, Canoas RS , Brazil, Faculdade de Medicina, Universidade Luterana do Brasil, Canoas, RS, Brazil.
2 Centro de Pesquisa Clínica, Hospital de Clínicas de Porto Alegre, Porto Alegre RS , Brazil, Centro de Pesquisa Clínica, Hospital de Clínicas de Porto Alegre, Porto Alegre, RS, Brazil.
3 Grupo de Pesquisa em Inovações e Intervenções em Qualidade de Vida, UFRGS, Porto Alegre RS , Brazil, Grupo de Pesquisa em Inovações e Intervenções em Qualidade de Vida, Universidade Federal do Rio Grande do Sul (UFRGS), Porto Alegre, RS, Brazil.
Find articles by Victória Machado Scheibe2 Centro de Pesquisa Clínica, Hospital de Clínicas de Porto Alegre, Porto Alegre RS , Brazil, Centro de Pesquisa Clínica, Hospital de Clínicas de Porto Alegre, Porto Alegre, RS, Brazil.
3 Grupo de Pesquisa em Inovações e Intervenções em Qualidade de Vida, UFRGS, Porto Alegre RS , Brazil, Grupo de Pesquisa em Inovações e Intervenções em Qualidade de Vida, Universidade Federal do Rio Grande do Sul (UFRGS), Porto Alegre, RS, Brazil.
4 Faculdade de Medicina, Universidade Federal de Ciências da Saúde de Porto Alegre, Porto Alegre RS , Brazil, Faculdade de Medicina, Universidade Federal de Ciências da Saúde de Porto Alegre, Porto Alegre, RS, Brazil.
Find articles by Augusto Mädke Brenner2 Centro de Pesquisa Clínica, Hospital de Clínicas de Porto Alegre, Porto Alegre RS , Brazil, Centro de Pesquisa Clínica, Hospital de Clínicas de Porto Alegre, Porto Alegre, RS, Brazil.
3 Grupo de Pesquisa em Inovações e Intervenções em Qualidade de Vida, UFRGS, Porto Alegre RS , Brazil, Grupo de Pesquisa em Inovações e Intervenções em Qualidade de Vida, Universidade Federal do Rio Grande do Sul (UFRGS), Porto Alegre, RS, Brazil.
4 Faculdade de Medicina, Universidade Federal de Ciências da Saúde de Porto Alegre, Porto Alegre RS , Brazil, Faculdade de Medicina, Universidade Federal de Ciências da Saúde de Porto Alegre, Porto Alegre, RS, Brazil.
Find articles by Gianfranco Rizzotto de Souza2 Centro de Pesquisa Clínica, Hospital de Clínicas de Porto Alegre, Porto Alegre RS , Brazil, Centro de Pesquisa Clínica, Hospital de Clínicas de Porto Alegre, Porto Alegre, RS, Brazil.
3 Grupo de Pesquisa em Inovações e Intervenções em Qualidade de Vida, UFRGS, Porto Alegre RS , Brazil, Grupo de Pesquisa em Inovações e Intervenções em Qualidade de Vida, Universidade Federal do Rio Grande do Sul (UFRGS), Porto Alegre, RS, Brazil.
5 Faculdade de Psicologia, UFRGS, Porto Alegre RS , Brazil, Faculdade de Psicologia, UFRGS, Porto Alegre, RS, Brazil.
Find articles by Reebeca Menegol6 Centro de Pesquisa em Psicologia, Universidade Autónoma de Lisboa, Lisboa , Portugal, Centro de Pesquisa em Psicologia, Universidade Autónoma de Lisboa, Lisboa, Portugal.
Find articles by Pedro Armelim Almiro2 Centro de Pesquisa Clínica, Hospital de Clínicas de Porto Alegre, Porto Alegre RS , Brazil, Centro de Pesquisa Clínica, Hospital de Clínicas de Porto Alegre, Porto Alegre, RS, Brazil.
3 Grupo de Pesquisa em Inovações e Intervenções em Qualidade de Vida, UFRGS, Porto Alegre RS , Brazil, Grupo de Pesquisa em Inovações e Intervenções em Qualidade de Vida, Universidade Federal do Rio Grande do Sul (UFRGS), Porto Alegre, RS, Brazil.
7 Departamento de Psiquiatria, UFRGS, Porto Alegre RS , Brazil, Departamento de Psiquiatria, UFRGS, Porto Alegre, RS, Brazil.
8 Programa de Pós-Graduação em Psiquiatria e Ciências do Comportamento, UFRGS, Porto Alegre RS , Brazil, Programa de Pós-Graduação em Psiquiatria e Ciências do Comportamento, UFRGS, Porto Alegre, RS, Brazil.
Find articles by Neusa Sica da Rocha1 Faculdade de Medicina, Universidade Luterana do Brasil, Canoas RS , Brazil, Faculdade de Medicina, Universidade Luterana do Brasil, Canoas, RS, Brazil.
2 Centro de Pesquisa Clínica, Hospital de Clínicas de Porto Alegre, Porto Alegre RS , Brazil, Centro de Pesquisa Clínica, Hospital de Clínicas de Porto Alegre, Porto Alegre, RS, Brazil.
3 Grupo de Pesquisa em Inovações e Intervenções em Qualidade de Vida, UFRGS, Porto Alegre RS , Brazil, Grupo de Pesquisa em Inovações e Intervenções em Qualidade de Vida, Universidade Federal do Rio Grande do Sul (UFRGS), Porto Alegre, RS, Brazil.
4 Faculdade de Medicina, Universidade Federal de Ciências da Saúde de Porto Alegre, Porto Alegre RS , Brazil, Faculdade de Medicina, Universidade Federal de Ciências da Saúde de Porto Alegre, Porto Alegre, RS, Brazil.
5 Faculdade de Psicologia, UFRGS, Porto Alegre RS , Brazil, Faculdade de Psicologia, UFRGS, Porto Alegre, RS, Brazil.
6 Centro de Pesquisa em Psicologia, Universidade Autónoma de Lisboa, Lisboa , Portugal, Centro de Pesquisa em Psicologia, Universidade Autónoma de Lisboa, Lisboa, Portugal.
7 Departamento de Psiquiatria, UFRGS, Porto Alegre RS , Brazil, Departamento de Psiquiatria, UFRGS, Porto Alegre, RS, Brazil.
8 Programa de Pós-Graduação em Psiquiatria e Ciências do Comportamento, UFRGS, Porto Alegre RS , Brazil, Programa de Pós-Graduação em Psiquiatria e Ciências do Comportamento, UFRGS, Porto Alegre, RS, Brazil.
Correspondence: Neusa Sica da Rocha Rua Ramiro Barcelos, 2350 90035-903 - Porto Alegre, RS - Brazil Tel.: +5551996665055 E-mail: rb.ude.apch@ahcorn
DisclosureNo conflicts of interest declared concerning the publication of this article.
Received 2021 Jul 23; Accepted 2021 Nov 2.This is an Open Access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
This article has been corrected. See Trends Psychiatry Psychother. 2023 May 02; 45: e20230001.The Eysenck Personality Questionnaire Revised – Abbreviated (EPQR-A) consists of 24 items for assessment of the three fundamental personality traits (psychoticism, extraversion, and neuroticism) and a validity scale (lie scale). Our objectives were to assess the psychometric properties of a version of this instrument culturally adapted for Brazil.
321 participants were recruited using a non-probabilistic method.
Internal consistencies ranged from minimally acceptable to respectable, except for the psychoticism domain. Higher neuroticism scores were associated with higher depression and anxiety scores, higher extraversion scores were associated with lower levels of depression symptoms, and higher psychoticism scores were associated with higher levels of depression symptoms.
Our findings describe sustainable psychometric properties for the Brazilian Portuguese version of EPQR-A.
Keywords: Personality assessment, psychometrics, P-E-N model, confirmatory factor analysis, validation study, Brazil
According to Eysenck, personality can be defined as a more or less stable and enduring organization of a person’s character (conative behavior system), temperament (affective behavior system), intellect (cognitive behavior system), and physique (bodily configuration and neuroendocrine endowment), which determines their unique adjustment to the environment. 1 Eysenck’s personality model considers the existence of what he called superfactors, dimensions, or traits. According to this model, the three fundamental dimensions of personality are psychoticism (P), extraversion (E), and neuroticism (N). Each of these dimensions is expressed in terms of a continuum, and people can be classified at any point of the scales, from extremes to median points. 2
In the E dimension, people are shy and retracted on one side (introversion) and sociable and uninhibited on the other (extraversion). The same happens with the N dimension: the neurotic or emotionally unstable personality is located at one extreme and the emotionally stable personality lies at the other. Individuals with high N scores are overly emotional, anxious, and depressed, frequently experience feelings of guilt, have low self-esteem, and tend to suffer from psychosomatic disorders. The opposite happens with stable individuals, who are typically calm, steadfast, easygoing, and able to control their emotions. The P dimension is characterized by impulsivity on one side and impulse control on the other. The main characteristics of high P scores are hostility, cruelty, lack of empathy, and non-conformism. Eysenck believed that high levels of P are linked to increased vulnerability to psychosis and considered that the biological bases of personality could provide an explanation for certain behaviors through the physiological functioning of the central nervous system. 2 , 3
The Eysenck Personality Questionnaire (EPQ) contains 90 items 4 and the Revised Eysenck Personality Questionnaire (EPQ-R) contains 100 items. 3 The EPQ has the same factorial structure as the EPQ-R, but the original version had some psychometric limitations related to the P domain and the revised version of the questionnaire was the result of efforts to fix this problem. Reliability indices for the new P domain were improved and achieved acceptability, but were nevertheless not as high as those for the other domains. It should however be remembered that the P scale explores characteristics such as hostility, cruelty, little evidence of socialization, and lack of empathy, which may cause the lower reliability levels. 3
One of the consequences of the continuous development and improvement of these scales was a progressive increase in questionnaire size. This increase can be explained by the introduction of additional personality items forming a lie/social desirability scale (L) 3 – to facilitate detection of faking – and by the psychometric principle that larger size increases questionnaire reliability. 5 , 6 Although longer tests measure constructs more accurately, there are certain practical disadvantages to using them. There are numerous situations in which a research project would benefit from inclusion of a personality measure, but 90 or 100 items of additional information would increase the general questionnaire to an undesirable size. In contrast, shorter tests, even those with reliable psychometric properties, inevitably have a more limited measurement capability, but can be more easily incorporated into longer assessment protocols, complementing the data obtained.
Eysenck et al. 3 developed a short form of the EPQ-R, the Eysenck Personality Questionnaire Revised – Short Form (EPQR-S), for use in adults. In this version, the four subscales (the N, E, and P dimensions and the L scale) each contain 12 items, making a total of 48 items. The authors reported reliability for men and women, respectively, of 0.84 and 0.80 for N, 0.88 and 0.84 for E, 0.62 and 0.61 for P, and 0.77 and 0.73 for social desirability (L). Although the EPQR-S was developed explicitly “for use when time is very limited,” some might still consider the 48-item questionnaire too long and cease using personality variables in their research for reasons of convenience. 3
The Eysenck Personality Questionnaire Revised – Abbreviated (EPQR-A) is a 24-item inventory comprising four 6-item subscales (E, N, P, and L) that was developed for researchers to use when time is limited. 7 Researchers administered all of the EPQ and EPQR-S items to a sample of 685 undergraduate students in England, Canada, and Australia, 3 , 4 analyzing the data using item-total correlations for each of the four subscales of the EPQR-S. The six items with the highest item-total correlations for each of the subscales were selected for inclusion in the EPQR-A. The reliability of EPQR-A subscales was demonstrated by internal consistency levels. Satisfactory levels of internal reliability were found for the E (0.74-0.84), N (0.70-0.77), and L (0.59-0.65) subscales. However, unsatisfactory levels were found for the P scale (0.33-0.52). Concurrent validation of the EPQR-A subscales was performed by examining their association with the original EPQ subscales. The correlations between the two forms of measuring E, N, and L ranged from 0.84 to 0.90. A considerably lower correlation was found between the two P scales (0.44-0.52). 7
The EPQ has been adapted to Portuguese, and validated for the Brazilian population. The validation study compared personality structures in Brazilian and English men and women. The sample consisted of 636 Brazilian men and 760 Brazilian women, who were compared to 500 English men and 500 English women. Authors found that identical factors emerged for both the English and Brazilian populations and that intercorrelations of reliability scales were similar for both groups. 8
The Portuguese version of the EPQ-R consists of 70 items with dichotomous responses distributed over four scales: N (23 items), E (20 items), P (nine items), and L (18 items). The values obtained for internal consistency were: N (0.87, “very good”), E (0.83, “very good”), P (0.55, “unacceptable”), and L (0.78, “respectable”), according to the criterion established by DeVellis. 9 Test-retest reliability showed adequate (“very good” to “respectable”) temporal stability over four to eight weeks (according to the same criterion): N (0.86), E (0.89), P (0.72), and L (0.86). 10 Psychometric studies of the EPQ-R suggest adequate psychometric properties when using both Classical Test Theory and Item Response Theory. 11 , 12
Although some of these instruments, such as the EPQ (90 items), have been studied in the Brazilian population, 7 the main psychometric properties of the EPQR-A (24 items) have not yet been evaluated. Moreover, other instruments for personality assessment that have been validated, such as the Personality Inventory for the 5th edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-V) (PID-5) (220 items) 13 and the Dimensional Clinical Personality Inventory 2 (IDCP-2) (206 items), 14 are composed of many items, limiting their use in long research protocols. Even the other versions of Eysenck’s questionnaires are still lengthy, which may restrict their usage in studies and hinder data collection. Thus, our objectives were to validate this brief instrument for measurement of personality dimensions so that it can be easily incorporated into longer assessment protocols, culturally adapting a Brazilian Portuguese version of the EPQR-A and assessing its psychometric properties in terms of convergent, discriminant, and construct validity. We chose a depression scale (Patient Health Questionnaire [PHQ-9]) and an anxiety scale (Generalized Anxiety Disorder [GAD-2]) to investigate evidence of the validity of the N scale of our instrument, considering that the relationship of neuroticism as a predictor of depression and anxiety symptoms is well established in literature. 15 - 17
Our sample comprised 321 individuals, of whom 266 were women (82.9%) and 55 were men (17.1%). Ages ranged from 18 to 74, with a mean of 44.85 (standard deviation [SD] = 13.733) and a median of 44 years. Majorities of our sample were of white ethnicity (90.3%), were married or cohabiting (61.7%), had a paid occupation (63.2%), and had graduate education (57.6%).
Depression symptoms were assessed with the PHQ-9 and anxiety symptoms were assessed using the GAD-2 questionnaire. Personality dimensions were assessed using the EPQR-A.
The PHQ-9 is a nine-item scale that assesses the intensity and degree of incapacitation of nine depression symptoms according to the depression criteria described in the DSM-V. 18 , 19 Each item has four possible answers (not at all, several days, more than half the days, and nearly every day), scored from 0 to 3 points. A depression episode was defined as presence of five or more items of the PHQ-9, at least one of which was the first or second item. The Cronbach’s alpha coefficient for the PHQ-9 was 0.88 in our sample.
The GAD-2 is a two-item questionnaire on the anxiety symptoms “feeling nervous, anxious, or on edge” and “not being able to stop or control worrying.” 20 Each item has four possible answers (not at all, several days, more than half the days, and nearly every day), scored from 0 to 3 points. Generalized anxiety was defined as when the participant’s GAD-2 score (sum of the scores for both items) was greater than or equal to 3. The Cronbach’s alpha coefficient for the GAD-2 was 0.86 in our sample.
The EPQR-A consists of 24 items divided into four scales: N, E, P, and L. Each scale has six items, each of them with a dichotomous response format (yes or no); each answer is scored specifically, according to the scale, as 0 or 1. Scoring for each question is predetermined because some questions have reverse coding. 7
Participants were recruited by sharing the research protocol via social media in a non-probabilistic method. We conducted an online self-report survey in order to avoid spread of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Data on age, gender, education, marital status, ethnicity, occupation, depression and anxiety symptoms, and personality were collected from July to August of 2020.
The online questionnaire was presented in Google Forms (Google, Mountain View, CA, USA) to facilitate participants’ access. The main social media platforms used for sharing the online form were Facebook and WhatsApp (Facebook Inc., Menlo Park, CA, USA). Participants declared they were over 18 years of age and completed an online informed consent form. All responses were anonymous and optional; each participant could stop answering or refuse to answer at any point of the questionnaire.
Our first step was to contact the instrument developer, Leslie J. Francis, in order to obtain permission to translate the EPQR-A instrument. A three-person bilingual team handled the translation process. First, one person independently produced an English to Brazilian Portuguese translation. Second, a second member of the group produced an independent back-translation from Brazilian Portuguese to English and the third member of the team produced a second independent English to Brazilian Portuguese translation. Finally, the three-person team compared the two Brazilian Portuguese translations and came up with a final Brazilian Portuguese version of the EPQR-A, shown in Supplementary Material S1, available online-only. Once the questionnaire was ready, a 10-person group conducted a debriefing process with the final Brazilian version. This group included a team of expert psychiatrists in quality of life questionnaires and their applications, a psychologist, and medical and psychology students. Issues of lexical and cultural equivalence of the questionnaire items were discussed. The questionnaire’s stages are shown in Supplementary Material S2 (online-only).
Descriptive data were presented using sums, means, SDs, ranges, and percentages. Our analyses were based on assessment of the questionnaire’s construct validity (measured by confirmatory factor analysis [CFA]), convergent validity (evaluated using Pearson’s correlation coefficients), discriminant validity (assessed with t tests for independent samples), and reliability (assessed in terms of internal consistency). Internal consistency was evaluated by calculating Cronbach’s alpha and McDonald’s omega for each of the EPQR-A domains. 21 Normality was assessed with the Kolmogorov-Smirnov test. However, due to our sample size, all variables were considered parametric. 22 The t test for independent samples was used to compare EPQR-A domains between gender and age groups. The age variable was dichotomized using its median value for the independent samples t test. We calculated Cohen’s d for sample effect size when performing the t test and interpreted the results according to the Marôco criteria. 23 , 24 We used Pearson’s bivariate correlation test to evaluate the correlation coefficients between EPQR-A domains and PHQ-9 and GAD-2 scores. The t test for independent samples was also used to compare mean EPQR-A domain scores between depressed and non-depressed subjects and anxious and non-anxious individuals. Furthermore, linear regression was used to assess whether there were linear relationships between EPQR-A domains and gender (coded 0 for female and 1 for male), age, PHQ-9 scores, and GAD-2 scores.
Confirmatory factor analysis was used to examine the factor structure of the Brazilian EPQR-A (construct validity). The four-factor model (N, E, P, L) assessed by the EPQR-A was estimated using the maximum likelihood method, which is a well-established factor structure for the PEN model of personality measures. 12 , 25 , 26 As Furnham et al. 25 concluded: “the EPQ factors are strongly replicable across all 34 countries; that is, the original UK data can be replicated using data from any countries.” The goodness-of-fit indices considered were: chi-square (χ 2 ); ratio of chi-square to degrees of freedom (χ 2 /df); comparative fit index (CFI); Tucker-Lewis index (TLI); standardized root mean square residual (SRMR); and root mean square error of approximation (RMSEA). A confirmatory model has a good fit when the ratio χ 2 /df < 3, CFI >0.95 (CFI > 0.90 is acceptable), TLI > 0.95 (TLI > 0.90 is acceptable), SRMR < 0.08, and RMSEA < 0.06. 27 - 29 Statistical significance was defined at a 95% confidence interval (95%CI). All statistical analyses were performed using IBM SPSS Statistics version 22.0 (Armonk, NY, 2020), IBM SPSS Amos version 20.0 (Arbuckle, 2011) and JASP version 0.14.1 (University of Amsterdam, 2020).
The present study was approved by the ethics committee at the Hospital de Clínicas de Porto Alegre, in Porto Alegre, southern Brazil, and by the Brazilian National Committee of Research Ethics under CAAE no. 30487620.7.0000.5327, in accordance with the Declaration of Helsinki.