Table 3.

Measurement invariance of bifactor model across gender, educational level, and setting using entire sample.

	χ2 or (Δχ2)	df or (Δdf)	p-value	CFI or (ΔCFI)a	RMSEA or (ΔRMSEA)a
Gender (male vs. female)
1. Configural	41.25	102	1.00	1.000	0.000
2. Loadings constrained	65.02	125	1.00	1.000	0.000
3. Loadings and intercepts constrained	72.51	135	0.98	1.000	0.000
2. vs. 1.	(23.77)	(23)	0.42	(0.000)	(0.000)
3. vs. 2.	(7.49)	(10)	0.68	(0.000)	(0.000)
Educational level (≤elementary vs. ≥junior high)
1. Configural	49.80	102	1.00	1.000	0.000
2. Loadings constrained	92.32	125	0.99	1.000	0.000
3. Loadings and intercepts constrained	104.67	135	0.98	1.000	0.000
2. vs. 1.	(42.52)	(23)	0.008	(0.000)	(0.000)
3. vs. 2.	(12.35)	(10)	0.26	(0.000)	(0.000)
Setting (community vs. institution)
1. Configural	38.39	102	1.00	1.000	0.000
2. Loadings constrained	71.76	125	1.00	1.000	0.000
3. Loadings and intercepts constrained	77.91	135	1.00	1.000	0.000
2. vs. 1.	(33.37)	(23)	0.07	(0.000)	(0.000)
3. vs. 2.	(6.16)	(10)	0.80	(0.000)	(0.000)
Age (< 70 years vs. ≧ 70 years)
1. Configural	52.54	102	1.00	1.000	0.000
2. Loadings constrained	80.23	125	0.999	1.000	0.000
3. Loadings and intercepts constrained	82.48	135	1.00	1.000	0.000
2. vs. 1.	(27.69)	(23)	0.23	(0.000)	(0.000)
3. vs. 2.	(2.25)	(10)	0.99	(0.000)	(0.000)

^aCFI, comparative fit index; RMSEA, root mean square error of approximation.