
X^{2}(df)

CAIC*

CFI^{£}

GFI^{ψ}

NNFI***

ECVI& (90% CI)

RMSEA**(90% CI)


Model 1

3,775.47 (1524)

4,672.78

0.66

0.72

0.64

10.47 (10.01–10.96)

0.06 (0.059–0.066)

Model 2

1,962.45 (990)

3,256.24

0.83

0.75

0.82

6.06 (5.74–6.40)

0.05 (0.041–0.056)

Model 3

1,145.28 (705)

2,230.40

0.91

0.90

0.90

3.78 (3.55–4.04)

0.04 (0.036–0.044)

 Note. Model 1 had seven domains and 59 questions and was based on Bekiari et al. [11]. Model 2 had six domains (excluding work environment) and 50 questions and was based on Eltayeb et al. [2]. Model 3 had six domains (excluding work environment) and 41 questions (nine questions excluded: one from work station, five from body posture, two from break time, and one from social support).
 *CAIC: consistent with the Akaike information criterion.
 **RMSEA: root mean square error of approximation; “90% CI” = 90% confidence interval for RMSEA.
 ***NNFI: nonnormed fit index.
 ^{£}CFI: comparative fit index.
 ^{ψ}GFI: goodnessoffit index.
 &ECVI: expected crossvalidation index and 90% CI.