Maartje Boer

CHAPTER 3 88 Summary EFA results (Tables A3.3 and A3.4) We evaluated the EFAs based on empirical eigenvalues and parallel analysis using the calibration samples for each country. The number of factors with empirical eigenvalues higher than one denotes the number of potential factors to retain. We compared the empirical eigenvalues values with 1,000 randomly generated eigenvalues, based on the same number of items and sample size of the respective country. The number of factors to retain was determined by the number of factors where the 95 th percentile random data eigenvalues did not exceed the empirical eigenvalues (Ledesma & Valero- Mora, 2007). In 42 out of 44 countries, results from the EFA identified one factor with an eigenvalue higher than one, suggesting a one-factor solution in these countries (Table A3.3). For Sweden and Ukraine, two factors showed an eigenvalue higher than one, suggesting a two-factor solution. However, the parallel analysis did not replicate this finding, because the empirical eigenvalue of only the first factor exceeded its 95 th random eigenvalue. Thus, also in these two countries, a one-factor solution was supported. Nevertheless, for each country, we estimated the model estimates from the one-, two-, and three-factor solutions. Model fit was evaluated based on the Comparative Fit Index (CFI), Tucker Lewis Index (TLI), Root Mean Square Error of Approximation (RMSEA), and Standardized Root Mean Square Residual (SRMR) (CFI/TLI: ≥ 0.90 acceptable, ≥ 0.95 good; RMSEA: ≤ 0.08 acceptable, ≤ 0.06 good; SRMR: ≤ 0.10 acceptable, ≤ 0.08 good) (Hu & Bentler, 1999). We did not rely on the Chi-square statistic given its sensitivity to large sample sizes (Fabrigar et al., 1999). In all 44 countries, the model fit of the one-factor solution was good according to all fit indices, because the lowest observedCFI and TLI were 0.964 and 0.952 and the highest RMSEA and SRMR 0.055 and 0.062, respectively (Table A3.4). We also evaluated the quality of the one-factor solution, whereby the quality of the factor was considered good when there were at least five items with significant ( p < 0.05) factor loadings higher than 0.50 (Costello & Osborne, 2005). In all countries, this requirement was fulfilled (Table A3.4). More specifically, in 30 out of 44 countries, all nine factor loadings exceeded 0.50. In 12 countries, there was one factor loading lower than 0.50 (although not lower than 0.42), and in two countries, there were two factor loadings lower than 0.50 (although not lower than 0.44). Thus, in all countries, the model fit and quality of the one-factor model was good.