The five-factor model (FFM) of personality (McCrae & Costa, 1985) proposes that human personality can be described in terms of five broad traits: openness, which is the willingness to embrace ideas, values, feelings, actions, and aesthetics; conscientiousness, which is the tendency to be cautious, follow rules, be hardworking, and follow through with actions; extraversion, which is the tendency to be sociable, talkative, and have a dominant style; agreeableness, which is the tendency to prioritize social harmony over other considerations, and to be cooperative and sympathetic in dealing with others; and neuroticism, which is the tendency to worry excessively and experience negative emotions such as sadness and anxiety.

Since the FFM emerged as the dominant model of personality in the psychology literature, numerous instruments have been created to measure these personality factors, for example, the Neuroticism, Extraversion, and Openness Personality Inventory (NEO-PI; Costa & McCrae, 1985) and its various revisions and adaptations, such as Goldberg’s 50-Bipolar Self-Rating Scales and 100 Unipolar Markers of the Big Five personality dimensions (Goldberg, 1992) and The Big Five Aspects Scale (DeYoung et al., 2007). These instruments provide a detailed view of personality through the lens of FFM traits. There are also shorter scales, which include the 44-item Big Five Inventory (BFI; John et al., 2008), the 40-item Big Five Mini-Markers (Saucier, 1994), the 20-item Mini-International Personality Item Pool (Mini-IPIP; Donnellan et al., 2006), and the Ten Item Personality Inventory (TIPI; Gosling et al., 2003). Although these shorter instruments are more practical for a quick personality overview, in terms of construct validity not all have demonstrated sound psychometric properties across a broad range of population cohorts (Clark & Watson, 2019).

Students’ personality traits are potentially important constructs in education, because they predict many aspects of the learning process, such as preferences regarding the learning environment and approach (Holen et al., 2015). This study was conducted in the Singapore context, where consideration of personality traits in education aligns with the Singapore Ministry of Education’s 21st century skills vision that students should develop both desirable personal attributes and academic capabilities through their education. Although this vision has been embraced at various higher education institutions in Singapore, progress toward this goal is likely to be hampered by limited access to practical validated measures that can be used for ongoing monitoring in the affective domain. Instruments such as the NEO-PI (Costa & McCrae, 1985) are well-validated but provide a level of detail not required for broad monitoring purposes. Conversely, existing shorter instruments have exhibited inconsistent psychometric properties across studies, most of which have not been validated in the Singapore context.

Singapore is a multicultural country that has both native English speakers and a high proportion of residents for whom English is a second language. The items in most five-factor instruments were developed for native English speakers. For example, some single descriptive adjectives that are used may be difficult for participants to understand, particularly for those with English as a second language (Saucier, 1994). Other instruments, such as the BFI (John et al., 2008), are made up of short phrases incorporating words (e.g., imperturbable) that would also be difficult for participants whose native language is not English to understand. Thus, our goal in this study was to develop and provide a preliminary validation of a brief FFM instrument with simplified language content that is suitable for use in the Singapore context.

Method

Participants

To evaluate the internal structure of the instrument, students at two polytechnics (identified as A and B) in Singapore responded to the items in a single sitting (Polytechnic A, n = 421; Polytechnic B, n = 271). Data screening was then conducted to remove partially completed surveys and clearly disengaged responses (i.e., choosing the same response for all items). The final sample comprised 351 participants from Polytechnic A (157 men, 187 women, 7 no gender indicated) and 196 participants from Polytechnic B (85 men, 109 women, 2 no gender indicated). Participants’ ages ranged between 16 and 25 years (M_age = 18.00, SD = 1.57, Polytechnic A; M_age = 18.88, SD = 1.43, Polytechnic B). All participants were of Asian ethnicity, with the majority being Singaporeans (89.9% Polytechnic A, 83.7% Polytechnic B).

Procedure

The BFFI was completed online and hosted on the Qualtrics platform. Online links were sent via email. Participants were encouraged to complete the survey in one sitting to avoid possible abandonment. They were offered no incentives to participate in the survey, and were told that their participation was voluntary and that they could withdraw from the survey at any time.

Instrument Development

The Brief Five Factor Instrument (BFFI) we developed comprises 20 items and has a 7-point, bipolar statement-based response format. To construct the instrument, we examined items in the long and short Big Five instruments listed in the introduction. We then generated an item pool to present the same essential content as these measures, but with a bipolar format. Bipolar response scales have been found to reduce acquiescence bias and to produce better model fit compared to Likert-based response formats (Friborg et al., 2006). The BFFI items are listed in Appendix A. Responses are scored from 1 to 7, with high scores indicating higher levels of the attribute indicated in the subscale title.

Results

To evaluate the internal structure of the BFFI, we conducted an exploratory factor analysis (EFA) and an confirmatory factor analysis (CFA). An EFA was first conducted on data from the 196 Polytechnic B participants to provide a preliminary assessment of internal structure (a random choice was made between the samples for the EFA/CFA procedures). To cross-validate the EFA results, we conducted a CFA on data from the 351 Polytechnic A participants. All procedures associated with the EFA were conducted with SPSS 25.0, and those associated with the CFA were conducted with LISREL 8.80.

Exploratory Factor Analysis

Descriptive statistics and bivariate correlations for the Polytechnic B data are presented in Appendix B. We used maximum likelihood extraction in the EFA of Polytechnic B participants’ BFFI scores (n = 351). We rotated the factors extracted to approximate a simple structure using the direct Oblimin method, allowing factors to correlate in the rotation. The criteria used to determine the number of retained factors were the Kaiser criterion of eigenvalues over 1.0, the Cattell scree plot, and a parallel analysis of obtained eigenvalues.

Data were screened to ensure compliance with all relevant assumptions prior to the EFA being conducted. No significant departure from normality was identified in the item distribution based on a comparison of skewness and kurtosis coefficients with Kline’s (2005) criteria (skewness value < 3.0 and kurtosis value < 8.0). Relationships between all item score pairs were deemed to be linear through visual examinations of bivariate scatterplots. We identified no univariate outliers using standard (z) scores (all z scores < 3.0), and no significant multivariate outliers at the .001 level based on the obtained Mahalanobis distance, chi squared values. Results from the Kaiser–Meyer–Olkin (KMO) and Bartlett’s sphericity tests also showed that assumptions of factorability had been met. The case-to-item ratio, which was 9.8, was adequate to yield reliable estimates of correlations among the variables.

The EFA identified five distinct factors, based on Kaiser’s criterion, the scree plot, and a parallel analysis. These factors accounted for 64.11% of the item variance. Communalities and rotated factor loadings are shown in Table 1. Given that the factor rotation was oblique, we used the pattern matrix to interpret the factor item loadings from the EFA. As indicated, all items loaded most strongly on their proposed factor, with minimal cross-loadings on other factors. Cronbach’s alpha for the five factors were .74 for the BFFI C subscale, .78 for the BFFI subscale, .82 for the BFFI N subscale, .86 for the BFFI E subscale, and .64 for the BFFI O subscale. Thus, the first four were well above the minimum level of .70 recommended by Nunnally (1978). As Cronbach’s alpha obtained for the BFFI O subscale was lower than this minimum, this suggested that the openness factor was factorially more complex than the other factors. Collectively, however, our results indicated that the proposed five-factor BFFI model was tenable in the Polytechnic B sample.

Confirmatory Factor Analysis

To cross-validate the EFA results, we conducted a CFA using scores from Polytechnic A (n = 351). Descriptive statistics and bivariate correlations for the Polytechnic A sample are shown in Appendix C. Prior to this analysis, we also conducted data screening to ensure that all relevant assumptions for CFA in terms of normality, linearity, factorability, and the absence of outlying univariate scores and multivariate score sets were met. These analyses all produced satisfactory results. Given the high case-to-item ratio of 17.55, the sample size was also deemed large enough to yield reliable estimates.

The overall chi square obtained for the five-factor model was χ²(160) = 403.59, p < .05 (χ²/df = 2.52). The goodness of fit (GFI) and adjusted goodness of fit (AGFI) values were .90 and .86, respectively, indicating that the proportion of variance accounted for by the estimated population covariance was just below recommended minimum levels (GFI ≥ .95 and AGFI ≥ .90). The normed fit index and nonnormed fit index values (.94 and .96, respectively) were within the recommended range (i.e., .95, or improvement of fit by 95% relative to the null model). The root mean square error of approximation, indicating the square root of the difference between the residuals of the sample covariance matrix and the hypothesized model (obtained value = .07), and the standardized root mean square residual, indicating the standardized difference between the observed correlations and the predicted correlations (obtained value = .06), were also within the recommended ranges (≤ .08 for both). The comparative fit index (obtained value = .96) and χ²/df (obtained value = 2.52) were also within the recommended range for determining good fit (≥ .90 and ≤ 3.0, respectively). Thus, the CFA results for the BFFI supported the proposed five-factor structure.

Discussion

In this study we aimed to provide a preliminary evaluation of a brief FFM instrument designed for use in the Singapore education context. Because the BFFI comprises only four items to measure each FFM personality factor, the instrument is very quick to administer, making it a viable option both for practical situations and for research in which multiple measures must be employed. A comparison of the psychometric properties of the BFFI with validation data published on other currently available brief instruments (indices from Boyle, 2008) is shown in Table 2.

As the results show, the BFFI compared well with the 44-item BFI, according to comparable fit indices. In comparison to the 20-item Mini-IPIP, which comprised the same number of items, four of the five BFFI subscales exhibited higher internal consistency, with similar fit indices (although the BFFI had a higher CFI). According to the fit indices, the BFFI outperformed both the BFI-15 and the TIPI. In a comparison of the BFFI with the NEO-PI-R and Goldberg 50-Bipolar Self-Rating Scales, the BFFI had comparable internal consistency for four subscales, and better fit indices. These results show that this preliminary validation of the internal structure of the BFFI was favorable.

As noted, the alpha result for the openness factor (BFFI_O) from the Polytechnic B sample was lower than those obtained for the other four BFFI factors. Boyle (2008) reviewed a number of studies that showed FFM openness was multifactorial, rather than unidimensional, and that this factor, as operationalized in all the developed FFM instruments, might represent an oversimplification of this personality trait. The complex nature of the openness factor may explain its relatively low alpha result in this study. As the factor was operationalized in a similar manner to that used in other FFM instruments, however, it was deemed tenable to retain the factor in the overall instrument.

Further research is needed for the validation of the BFFI in contexts other than Singaporean polytechnics. It would also be useful to validate the BFFI by the correlation of the scores on this scale with those obtained through established instruments such as the NEO-PI-R or the Mini-IPIP. A study in which the BFFI subscales are correlated with measures of theoretically related constructs (e.g., anxiety scales for the BFFI_N subscale), and of the functioning of the BFFI in student cohorts from other age groups (e.g., those at secondary school) would also be useful for validation purposes. Although our results are preliminary, they support the promise of the BFFI for research and monitoring exercises in the Singapore education context.

References

Boyle, G. J. (2008). Critique of the five-factor model of personality. In G. J. Boyle, G. Matthews, & D. H. Saklofske (Eds.), The SAGE handbook of personality theory and assessment, Vol. 1. Personality theories and models (pp. 295–312). Sage Publications.
https://doi.org/10.4135/9781849200462.n14

Clark, L. A., & Watson, D. (2019). Constructing validity: New developments in creating objective measuring instruments. Psychological Assessment, 31(12), 1412–1427.
https://doi.org/10.1037/pas0000626

Costa, P. T., Jr., & McCrae, R. R. (1985). The NEO Personality Inventory manual. Psychological Assessment Resources.

DeYoung, C. G., Quilty, L. C., & Peterson, J. B. (2007). Between facets and domains: 10 aspects of the Big Five. Journal of Personality and Social Psychology, 93(5), 880–896.
https://doi.org/10.1037/0022-3514.93.5.880

Donnellan, M. B., Oswald, F. L., Baird, B. M., & Lucas, R. E. (2006). The Mini-IPIP scales: Tiny-yet-effective measures of the Big Five factors of personality. Psychological Assessment, 18(2), 192–203.
https://doi.org/10.1037/1040-3590.18.2.192

Friborg, O., Martinussen, M., & Rosenvinge, J. H. (2006). Likert-based vs. semantic differential-based scorings of positive psychological constructs: A psychometric comparison of two versions of a scale measuring resilience. Personality and Individual Differences, 40(5), 873–884.
https://doi.org/10.1016/j.paid.2005.08.015

Goldberg, L. R. (1992). The development of markers for the Big-Five factor structure. Psychological Assessment, 4(1), 26–42.
https://doi.org/10.1037/1040-3590.4.1.26

Gosling, S. D., Rentfrow, P. J., & Swann, W. B., Jr. (2003). A very brief measure of the Big-Five personality domains. Journal of Research in Personality, 37(6), 504–528.
https://doi.org/10.1016/S0092-6566(03)00046-1

Holen, A., Manandhar, K., Pant, D. S., Karmacharya, B. M., Olson, L. M., Koju, R., & Mansur, D. I. (2015). Medical students’ preferences for problem-based learning in relation to culture and personality: A multicultural study. International Journal of Medical Education, 6, 84–92.
https://doi.org/10.5116/ijme.558e.6451

John, O. P., Naumann, L. P., & Soto, C. J. (2008). Paradigm shift to the integrative Big Five trait taxonomy: History, measurement, and conceptual issues. In O. P. John, R. W. Robins, & L. A. Pervin (Eds.), Handbook of personality: Theory and research (pp. 114–158). Guilford Press.

Kline, R. B. (2005). Principles and practice of structural equation modeling (2nd ed.). Guilford Press.

McCrae, R. R., & Costa, P. T. (1985). Updating Norman’s “adequate taxonomy”: Intelligence and personality dimensions in natural language and in questionnaires. Journal of Personality and Social Psychology, 49(3), 710–721.
https://doi.org/10.1037/0022-3514.49.3.710

Nunnally, J. C. (1978). Psychometric theory (2nd ed.). McGraw-Hill.

Saucier, G. (1994). Mini-Markers: A brief version of Goldberg’s unipolar Big-Five markers. Journal of Personality Assessment, 63(3), 506–516.
https://doi.org/10.1207/s15327752jpa6303_8

Appendix A

Item Statements in the Brief Five-Factor Instrument

Appendix B

Boyle, G. J. (2008). Critique of the five-factor model of personality. In G. J. Boyle, G. Matthews, & D. H. Saklofske (Eds.), The SAGE handbook of personality theory and assessment, Vol. 1. Personality theories and models (pp. 295–312). Sage Publications.
https://doi.org/10.4135/9781849200462.n14

Clark, L. A., & Watson, D. (2019). Constructing validity: New developments in creating objective measuring instruments. Psychological Assessment, 31(12), 1412–1427.
https://doi.org/10.1037/pas0000626

Costa, P. T., Jr., & McCrae, R. R. (1985). The NEO Personality Inventory manual. Psychological Assessment Resources.

DeYoung, C. G., Quilty, L. C., & Peterson, J. B. (2007). Between facets and domains: 10 aspects of the Big Five. Journal of Personality and Social Psychology, 93(5), 880–896.
https://doi.org/10.1037/0022-3514.93.5.880

Donnellan, M. B., Oswald, F. L., Baird, B. M., & Lucas, R. E. (2006). The Mini-IPIP scales: Tiny-yet-effective measures of the Big Five factors of personality. Psychological Assessment, 18(2), 192–203.
https://doi.org/10.1037/1040-3590.18.2.192

Friborg, O., Martinussen, M., & Rosenvinge, J. H. (2006). Likert-based vs. semantic differential-based scorings of positive psychological constructs: A psychometric comparison of two versions of a scale measuring resilience. Personality and Individual Differences, 40(5), 873–884.
https://doi.org/10.1016/j.paid.2005.08.015

Goldberg, L. R. (1992). The development of markers for the Big-Five factor structure. Psychological Assessment, 4(1), 26–42.
https://doi.org/10.1037/1040-3590.4.1.26

Gosling, S. D., Rentfrow, P. J., & Swann, W. B., Jr. (2003). A very brief measure of the Big-Five personality domains. Journal of Research in Personality, 37(6), 504–528.
https://doi.org/10.1016/S0092-6566(03)00046-1

Holen, A., Manandhar, K., Pant, D. S., Karmacharya, B. M., Olson, L. M., Koju, R., & Mansur, D. I. (2015). Medical students’ preferences for problem-based learning in relation to culture and personality: A multicultural study. International Journal of Medical Education, 6, 84–92.
https://doi.org/10.5116/ijme.558e.6451

John, O. P., Naumann, L. P., & Soto, C. J. (2008). Paradigm shift to the integrative Big Five trait taxonomy: History, measurement, and conceptual issues. In O. P. John, R. W. Robins, & L. A. Pervin (Eds.), Handbook of personality: Theory and research (pp. 114–158). Guilford Press.

Kline, R. B. (2005). Principles and practice of structural equation modeling (2nd ed.). Guilford Press.

McCrae, R. R., & Costa, P. T. (1985). Updating Norman’s “adequate taxonomy”: Intelligence and personality dimensions in natural language and in questionnaires. Journal of Personality and Social Psychology, 49(3), 710–721.
https://doi.org/10.1037/0022-3514.49.3.710

Nunnally, J. C. (1978). Psychometric theory (2nd ed.). McGraw-Hill.

Saucier, G. (1994). Mini-Markers: A brief version of Goldberg’s unipolar Big-Five markers. Journal of Personality Assessment, 63(3), 506–516.
https://doi.org/10.1207/s15327752jpa6303_8

Elaine Chapman, Graduate School of Education, The University of Western Australia, M428, 35 Stirling Highway, Crawley, WA 6009, Australia. Email: [email protected]