Warwick Edinburgh Mental Well-Being Scale (WEMWBS): measurement invariance across genders and item response theory examination

Table 2 WEMBWS across men and women (N = 386)

	χ²	df	p	CFI	ΔCFI	TLI	RMSEA	Δ RMSEA	SRMR	BIC	AIC
Configural: loadings + intercepts free	399.266	154	0.233	0.924		0.910	0.091		0.049	13,076	13,076
Metric: loadings fixed + intercepts free	415.117	167	0.001*	0.923	0.001	0.916	0.088	0.003	0.061	13,014	12,734
Scalar: loadings + intercept fixed	476.726	180	0.064	0.908	0.015	0.907	0.092	0.004	0.067	12,999	12,769
Partial invariance	438.214	179	0.001*	0.915	0.008	0.907	0.092	0.000	0.065	13,043	12,758

* = Statistically significant p < 0.05. Partial invariance achieved by freeing intercept 5. The model is regarded as acceptable if the chi-square is not significant. However, this is disregarded when the sample size exceeds 200. The Comparative Fit Index (CFI) compares the examined model of interest with the null model. The Tucker Lewis Index (TLI) is computed by the division of the chi square for the target model and the null model by their corresponding df vales (relative chi squares), which are then subtracted from each other, and their difference is finally divided by the relative chi square for the null model minus 1. The Root Mean Square Error of Approximation (RMSEA) represents the square root of the average or mean of the covariance residuals. The Bayesian Information Criterion (BIC) expresses the log of a Bayes factor of the target model compared to the saturated model. Finally, the Akaike information criterion (AIC) is regarded as an information theory goodness of fit measure applicable when maximum likelihood estimation is used [5]. After freeing the intercept for one item (Item 5; “I’ve had energy to spare”), partial scalar invariance was supported

ISSN: 2050-7283