Table 1. Goodness-of-fit tests for models retaining 1 to 20 factors from an analysis of 93 traits (N = 13,138).
Number of Factors | Residual df | Chi-square related tests | Residual correlations | Variance explained | ||||
χ2 | RMSEA | CFI | SRMR | >0.1 | >0.05 | |||
Null | 4278 | 602645 | 0.103 | 0.000 | 0.262 | 73.7 | 93.9 | 0.0 |
1 | 4185 | 347777 | 0.079 | 0.426 | 0.084 | 12.7 | 37.0 | 44.2 |
2 | 4093 | 278850 | 0.071 | 0.541 | 0.067 | 7.0 | 26.7 | 55.8 |
3 | 4002 | 230499 | 0.066 | 0.621 | 0.055 | 4.5 | 18.4 | 64.5 |
4 | 3912 | 198199 | 0.061 | 0.675 | 0.048 | 2.8 | 14.7 | 70.7 |
5 | 3823 | 172549 | 0.058 | 0.718 | 0.041 | 2.2 | 10.1 | 76.5 |
6 | 3735 | 155107 | 0.056 | 0.747 | 0.037 | 1.9 | 8.6 | 80.7 |
7 | 3648 | 139399 | 0.053 | 0.773 | 0.033 | 1.8 | 6.5 | 84.7 |
8 | 3562 | 124820 | 0.051 | 0.797 | 0.030 | 1.5 | 6.0 | 87.6 |
9 | 3477 | 111664 | 0.049 | 0.819 | 0.028 | 1.2 | 4.9 | 90.3 |
10 | 3393 | 101096 | 0.047 | 0.837 | 0.026 | 1.1 | 4.1 | 92.7 |
11 | 3310 | 89025 | 0.044 | 0.857 | 0.024 | 0.9 | 3.6 | 95.1 |
12 | 3228 | 82315 | 0.043 | 0.868 | 0.022 | 0.7 | 3.0 | 97.0 |
13 | 3147 | 76152 | 0.042 | 0.878 | 0.020 | 0.6 | 2.3 | 98.9 |
14 | 3067 | 71111 | 0.041 | 0.886 | 0.018 | 0.4 | 2.0 | 100.7 |
15 | 2988 | 67511 | 0.041 | 0.892 | 0.017 | 0.2 | 1.7 | 102.3 |
16 | 2910 | 64403 | 0.040 | 0.897 | 0.016 | 0.2 | 1.4 | 103.6 |
17 | 2833 | 62037 | 0.040 | 0.901 | 0.014 | 0.2 | 1.3 | 104.9 |
18 | 2757 | 58874 | 0.039 | 0.906 | 0.013 | 0.1 | 1.1 | 106.0 |
19 | 2682 | 56837 | 0.039 | 0.909 | 0.012 | 0.1 | 0.9 | 107.0 |
20 | 2608 | 55531 | 0.039 | 0.912 | 0.012 | 0.1 | 0.7 | 107.8 |
CFI = Comparative Fit index.
RMSEA = Root mean square error of approximation.
SRMR = Standardised root mean square residual.
All the model χ2 values were highly significant (p<0.0001). For the 7 factors in this study, the overall fit statistics suggested a reasonable to good fit. Examination of the 66 residual correlations >0.1 showed that they clustered within the factor structure: 6 related to Factor 2 variables 6–24 m (6.6% of all inter-correlations within this group), 7 (19.4%) related to Factors 1/2 variables 30–42 m, 8 (7.6%) related to Factor 1 variables 57 m–9y, 4 (8.9%) to Factor 3 variables, 10 (12.8%) to Factor 4, 7 (19.4%) to Factor 5, 12 (21.8%) to Factor 6 and 12 (18.2%) to Factor 7.
The variance explained is expressed as a percentage of the sum of the communalities. This latter measure can exceed 100% due to the presence of negative eigenvalues. Residual correlations >0.1 and >0.05 are reported as a percentage of the 4278 pairwise correlations.