Psychometric properties of the Brazilian version of the Big Five Inventory

Abstract Introduction There is growing interest in the fields of psychiatry and psychology in investigating the relationship between personality and psychopathology. The Big-5 is a model developed to investigate five personality dimensions: Extroversion, Agreeableness, Conscientiousness, Neuroticism, and Openness. In the present study, we describe the process of translation into Brazilian Portuguese and adaptation of a free tool to evaluate the Big-5 model: The Big-5 Inventory (BFI). The instrument has 44 items with a Likert response scale ranging from 1 to 5. Objectives To translate and adapt the BFI into Brazilian Portuguese. Methods The adaptation was conducted in the following steps: 1) Translation, 2) Evaluation Committee, 3) Back-translation, 4) Pilot study, 5) Evaluation Committee, and 6) Application. The sample comprised 490 participants from various regions of Brazil. The participants’ ages ranged from 18 to 71 years, most of them had completed high school (62.9%), and the majority were women (75%). Results A model with the following fit indexes was found: χ2/df: 1.954; goodness fit index (GFI): 0.924; comparative fit index (CFI): 0.920; and root mean square error of approximation (RMSEA): 0.044. Conclusion The results are suggestive that the Brazilian version of this instrument has good psychometric properties and represent a cost-free option for investigating associations with the Big-5 in psychiatry.


Introduction
One of the best-known constructs in the study of personality is the Big Five Factors theory (Big-5), which involves five dimensions of personality: Neuroticism, Conscientiousness, Agreeableness, Openness, and Extraversion. 1 Each trait encompasses different characteristics of an individual: Openness is related to creativity and imagination; Conscientiousness is related to organization and reliability; Extroversion is a tendency towards sociability and assertiveness; Agreeableness is a tendency towards prosocial attitudes and altruism; and Neuroticism is a tendency to sadness and negative emotions. 2 Researchers have been investigating the association of the Big Five model with psychiatric disorders, psychological well-being and general well-being. [3][4][5][6] A recent systematic review of stroke patients showed that personality may indicate prognosis for treatment. The study demonstrated that post-stroke patients who had high Neuroticism scores were more likely to be affected by depression. 7 Another study identified a strong association between one's ability to delay gratification and the Conscientiousness trait, which is especially relevant to health problems such as overweight, drug abuse and risky sexual behavior. 8 In Brazil, the most used inventory for evaluating the Big-5 model is the NEO Personality Inventory (NEO-PI-R), prepared by Costa and McRae,9 adapted to Brazilian Portuguese by Carmen Flores Medonza, and published by Vetor. 10 Its use in research is limited, however, because it is restricted to psychologists, and use is prohibited for other professionals whose research could greatly benefit from assessment of the Big-5 model, such as psychiatrists, neurologists, and neuroscientists. Furthermore, the NEO-PI-R can only be acquired by purchasing it for a fee, which poses an additional difficulty, especially at a time when the country's research funding is suffering from drastic budget cuts.
The Big-5 Inventory (BFI) is an instrument for use in research, consisting of 44 Likert-type questions, designed to assess the five major personality factors.
The BFI is an instrument that has been used in research in several countries, 2 for example: in France with higher education students with an average age of 21 years 11 ; in the Bolivian Amazon area with farmers aged from 20 to 88 years 12 15 ; and in New Zealand with construction workers aged from 20 to 50 years. 16 Based on the evidence presented above, it is necessary to make a free instrument available to assess the five major personality traits, which would be very useful for researchers in Brazil as well. 17

Methods
The present study consisted of the translation and cross-cultural adaptation of the original instrument from English (North American) into Brazilian Portuguese. The instrument is free to use for research purposes.
The translation process was based on the general guidelines described by Hungerbuhler  were presented on a slide, and participants were asked which items they agreed with and which items they would suggest changing.
6. Final administration: the definitive version of the instrument was administered and validated with confirmatory factor analysis (CFA).

Sample
The sample size was based on 10:1 ratio, 17   Composite reliability (CR) and average variance extracted (AVE) were also analyzed, both of which enable us to assess the quality of the instrument. 22 Acceptable reference values for CR and AVE are greater than 0.7 22 and greater than or equal to 0.5, 23 respectively. Cronbach's alpha was also calculated and we set the reference value at > 0.7.

Results
The original, translated, and back-translated versions of the Big Five Inventory are shown in Table 1 Table 2). Model M6 was estimated without the aforementioned items, but with inclusion of two second-order factors.
In M1, a single factor was used containing all items.
In M3, all items that had factor loadings less than 0.3 were removed (2,12,22,35,37). In M7, items that presented residual covariances were removed until the model was adjusted. * There were no changes between the original version and the back translation. † Changes were made between the original version and the back translation or from the adjusted synthesis version to the definitive version.    The results in Table 4    Regarding reliability, it is known that CR is a more accurate indicator of precision than Cronbach's alpha, because CR factor loads are free to vary among themselves, whereas in Cronbach's alpha, factor loads are fixed as equal. The CR is therefore able to produce better adjusted accuracy indices. All factors had values above 0.7, which indicates homogeneity between items.
Regarding the AVE, all results were above the reference value (0.5), except for the Openness factor, which had a borderline index. This means that most factors (latent variables) explain more than half of the variance of all of the items they contain, according to Valentini and Damásio. 32 The validated model (M8) retained 25 items and maintained the five factors that support the Big Five theory. 2 Church and Burke 33 point out difficulties with use of CFA in personality instruments, since there are restrictions to assessment of the personality structure.
The original model (M3), for example, had low fit indices. Other models observed in the literature also failed to achieve adequate fit indices, Benet-Martinez and John, 24 Danu, 34 Little et al., 35 Marsh et al. 36 The present study has some limitations, including the following: the sample was selected by convenience and was non-probabilistic; and the sample contained majorities of females and northeast Brazilian participants. Invariance was not a study goal because groups are not balanced by gender or region. We suggest that future studies perform invariance analysis.

Conclusion
Research associated with personality has increased considerably, which highlights the need for a measurement instrument that can provide more accurate measurements. The Big Five Inventory is used worldwide and proves to be this instrument.
There is also a need for free instruments to support research producing evidence, since this will facilitate replicability and increase researchers' access to studies involving personality. In Brazil there is a need for more studies involving personality and this instrument could substantially contribute to increase such research.