The aim of this study was to explore the psychometric properties of the Norwegian version of the TIPI (N-TIPI), in terms of internal consistency and structural validity. Psychometric properties of the N-TIPI were compared with the original version (Gosling et al., 2003; Ehrhart et al., 2009) as well as with four non-English translations: German (Muck et al., 2007), French (Storme et al., 2016), Portuguese (Nunes et al., 2018), and Spanish (Renau et al., 2013). These non-English versions were selected based on them being properly validated. It was a priori assumed that it was more suitable to compare the Norwegian translation with other European languages than for instance with Bangla, Indonesian, or Persian translations.

This study was designed as a cross-sectional psychometric assessment of the N-TIPI, conducted in a sample of 5,009 Norwegian master graduates.

Data were obtained from the Norwegian graduate survey, a digital survey conducted among Norwegian graduates with a master degree or equivalent in 2019, 3 years upon graduation, by the Nordic Institute for Studies in Innovation, Research and Education (NIFU; Skjelbred, 2019). The eligible sample included all individuals who had graduated from Norwegian higher education institutions in 2016 (n = 12,578) as well as Norwegian citizens who had graduated from higher education institutions abroad in 2015 and 2016 (n = 5,018). Hence, a total of 17,596 individuals were invited, and 6,188 (35.2%) provided informed consent and agreed to participate. According to Skjelbred et al. (2019), individuals who agreed to participate were substantially representative of the population. However, 1,179 graduates failed to respond on all relevant study items and were excluded, leaving a final study sample of 5,009 individuals.

Females constituted the majority of the sample (62.4%), individuals aged 28–30 years represented the most common age group (49.3%), and social sciences/law (22.1%) and natural/technical sciences (21.7%) were the most prevalent fields of study among the participants. Characteristics of the study sample are presented in Table 1.

The N-TIPI was administered to participants by means of a digital survey. The 10 items are designed to measure the FFM personality domains (E, A, C, ES, and O), each domain with two items (Gosling et al., 2003). Participants were asked to indicate, on a seven-point Likert scale (1 = disagree strongly; 2 = disagree moderately; 3 = disagree a little; 4 = neither agree nor disagree; 5 = agree a little; 6 = agree moderately; and 7 = agree strongly), the extent to which they agreed that a set of 10 descriptive statements applied to them. For instance, “extraverted, enthusiastic” (Norwegian: “utadvendt, entusiastisk”) and “reserved, quiet” (Norwegian: “reservert, stille”) represented the indicators of the E domain. An overview of items, item wordings, and response categories for the N-TIPI and the original TIPI is presented in Table 2.

The five subscales were constructed with the following procedure (Gosling et al., 2003): First, reverse-scored items (N-TIPI-2, N-TIPI-4, N-TIPI-6, N-TIPI-8, and N-TIPI-10) were recoded, ensuring that higher scores indicated higher domain values. Second, a mean score was calculated for each subscale.

The 10 N-TIPI items were analyzed descriptively by calculating means (M) and standard deviations (SD), separately for the 10 items as well as for the five subscales. Descriptive statistics for the N-TIPI subscales were compared with results reported for the original TIPI (Gosling et al., 2003), the German TIPI-G (Muck et al., 2007), the French TIPI-F (Storme et al., 2016), the Spanish TIPI-SPA (Renau et al., 2013), and the Portuguese TIPI-P (Nunes et al., 2018). The N-TIPI pattern of subscale correlations was calculated and compared to those reported for the original TIPI (Ehrhart et al., 2009), and for the TIPI-G and TIPI-F.

Internal consistency of the N-TIPI was assessed by calculating inter-item correlations (Pearson r), Cronbach’s α, and Spearman-Brown coefficients (S-B) for each of the five two-item subscales. N-TIPI estimates of internal consistency were compared with those reported for the original TIPI, TIPI-G, TIPI-F, TIPI-SPA, and TIPI-P. Although S-B coefficients are generally considered more appropriate than Pearson r and Cronbach’s α for estimating internal consistency of two-item scales (Eisinga et al., 2013), the two latter statistics were also applied in this study to enable comparisons with the other TIPI versions.

The structural validity of the N-TIPI was assessed with exploratory principal component analysis (PCA) using orthogonal (varimax) rotation. Parallel analysis (Horn, 1965) and visual scree plot inspection (Cattell, 1966) were performed to aid in determining the adequate number of components to extract. It was a priori defined that a fully satisfactory component structure had to meet the following six criteria:

An exploratory analysis was chosen in order to investigate whether it was possible to generate a FFM with a bottom-up (data driven) approach that satisfied the abovementioned six criteria, rather than to simply examine properties of a pre-defined five-factor solution (confirmatory analysis). Orthogonal rotation was chosen based on an a priori assumption that components would not be correlated at ≥0.32 (Tabachnick and Fidell, 2013).

All analyses were performed with IBM SPSS version 27, with the exception of parallel analysis that was conducted with an engine developed by Patil et al. (2017). The analytical procedures were based on the COSMIN guidelines for evaluating internal consistency and structural validity of measurement instruments (Terwee et al., 2012).

Participants were assured confidentiality and informed that participation was voluntary. Written informed consent was obtained from all participants (see Skjelbred et al., 2019 for more details).

Means and standard deviations for the N-TIPI items are presented in Table 3.

Of the N-TIPI subscales, the highest mean scores were found for C (M = 5.78; SD = 1.07), followed by O (M = 5.33; SD = 0.99), ES (M = 5.15; SD = 1.24), E (M = 4.88; SD = 1.45), and A (M = 4.77; SD = 1.09). Descriptive statistics for the N-TIPI subscales – compared with the original TIPI, TIPI-G, TIPI-F, TIPI-SPA, and TIPI-P – are presented in Table 4.

As shown in Table 4, the N-TIPI subscales demonstrated a comparable rank order of means and standard deviations as the original TIPI, TIPI-G, and TIPI-SPA, i.e., with higher mean scores for C and O than for ES, E, and A. TIPI-F and TIPI-P deviated from this pattern by having higher mean scores for A and C than for O, ES, and E.

All five N-TIPI subscales were significantly correlated, yet none of the subscales were correlated ≥0.32. In line with the original TIPI, all subscale associations were positive and the correlation between E and O was the strongest (r = 0.28, p < 0.001 for the N-TIPI). Correlations between N-TIPI subscales – compared with the original TIPI, TIPI-G, and TIPI-F – are presented in Table 5.

As shown in Table 5, the pattern of correlations between the N-TIPI subscales was quite comparable to those reported for the original TIPI, TIPI-G, and TIPI-F.

Internal consistency was highest for the E subscale (r = 0.61, p < 0.001; α = 0.75; S-B = 0.76), followed by ES (r = 0.47, p < 0.001; α = 0.62; S-B = 0.64), C (r = 0.47, p < 0.001; α = 0.61; S-B = 0.62), O (r = 0.27, p < 0.001; α = 0.41; S-B = 0.43), and A (r = 0.22, p < 0.001; α = 0.35; S-B = 0.36). Internal consistency for the N-TIPI subscales – compared with the original TIPI, TIPI-G, TIPI-F, TIPI-SPA, and TIPI-P – are presented in Table 6.

As evident in Table 6, N-TIPI subscale inter-item correlations (range: r = 0.22 to 0.61) and estimates of Cronbach’s α (range: α = 0.35 to 0.75) were quite comparable to those reported for the original TIPI (r = 0.28 to 0.61; α = 0.40 to 0.74) and TIPI-G (α = 0.42 to 0.67). The internal consistency of N-TIPI was somewhat higher than results reported for the TIPI-F (r = 0.13 to 0.52; α = 0.22 to 0.69), TIPI-SPA (r = 0.16 to 0.46; α = 0.08 to 0.71), and TIPI-P (α = 0.39 to 0.72).

The N-TIPI demonstrated satisfactory structural validity. The 10 items were deemed suitable for PCA (Table 7), as indicated by a statistically significant Bartlett’s test of sphericity (p < 0.001) and a KMO of 0.56. The PCA identified five components with λ exceeding 1.0 (C₁ = 2.24; C₂ = 1.63; C₃ = 1.29; C₄ = 1.06; C₅ = 1.03).

A parallel analysis (Table 8) indicated that only components with an λ of ≥1.01 should be retained. Hence, parallel analysis supported the extraction of five components (for the sixth component, the randomly generated λ exceeded the corresponding λ in the data: λ6_random = 0.99; λ6_data = 0.86).

Although somewhat unclear, the scree plot (Figure 1) indicated an inflection between the fifth and sixth components, supporting the extraction of five components.

As shown in Table 7, the five-component structure explained a total of 72.4% of the variance in the data, and each of the five components explained more than 10% of the variance (C₁ = 22.36%; C₂ = 16.25%; C₃ = 12.89%; C₄ = 10.64%; C₅ = 10.25%). The 10 N-TIPI items loaded pairwise on the five components, theoretically in accordance with the FFM personality domains. Each item loaded substantially (≥0.40) on its corresponding FFM component without any cross-loadings. Each item communality (h²) reached ≥0.20 (range: h² = 0.48 to 0.84).

In sum, the results showed that the N-TIPI met the six a priori defined criteria for a fully satisfactory component structure. First, preliminary analyses yielded suitability for PCA. Second, a five-component structure was supported by PCA, parallel analysis, and visual scree plot inspection. Third, the five-component structure explained more than 50% of the variance in the data and each of the components explained at least 10% of the variance. Fourth, the items loaded pairwise on the five components in accordance with the FFM personality domains and each item loaded substantially on its corresponding FFM component without any cross-loadings. Finally, each item communality reached ≥0.20.

This is the first study to systematically explore psychometric properties of the Norwegian version of the TIPI (N-TIPI). We were able to demonstrate acceptable internal consistency and satisfactory structural validity of the N-TIPI. The study sample was large (n = 5,009) and substantially representative of the eligible sample (Skjelbred et al., 2019). However, certain limitations should be kept in mind when interpreting results from this study. First, the sample consisted solely of individuals who had completed a master’s degree or equivalent, and participants were thus far higher educated than the general Norwegian population. Second, due to the study’s cross-sectional design and certain data limitations (N-TIPI was the only FFM instrument in the survey), we were not able to assess test–retest reliability of the N-TIPI or convergent validity with other validated FFM instruments.

This study represents an important step on the path to a fully validated Norwegian version of the TIPI. Future research on the N-TIPI could benefit from utilizing general population samples, exploring test–retest reliability and convergent validity with other validated Norwegian FFM instruments, i.e., the NEO-PI-R, NEO-FFI, and BFI (Martinsen et al., 2003, 2005; Engvik and Føllesdal, 2005). A large and growing number of TIPI versions have been and are being developed, and this brief FFM instrument is widely utilized in research. Future research on the TIPI could benefit from secondary research efforts (e.g., systematic or scoping reviews) focusing on providing an overview of translated and validated versions and their psychometric properties.

This was the first study to systematically explore psychometric properties of the N-TIPI. The N-TIPI demonstrated acceptable internal consistency and satisfactory structural validity. Although further research on the N-TIPI is warranted, the instrument stands out as feasible when it is essential to minimize participants’ response burden in studies that aim to explore personality as one among several concepts or utilize personality traits as covariates.