Validity, reliability, and measurement invariance of an adapted short version of the HIV stigma scale among perinatally HIV infected adolescents at the Kenyan coast

Table 4 Multi-group confirmatory factor analysis for age and gender sub-groups

Group	Invariance	χ² (df)	p value	CFI^b	TLI^b	RMSEA^b	Δχ² (Δdf)	p value^a	ΔCFI	ΔRMSEA
Age	Configural	80.86 (70)	0.176	0.974	0.967	0.040 [0.000–0.074]	–	–	0.007	–
	Metric/weak	92.89 (79)	0.136	0.967	0.962	0.042 [0.000–0.074]	10.98 (9)	0.2774	0.001	0.002
	Scalar/strong	101.34 (88)	0.157	0.968	0.967	0.039 [0.000–0.070]	8.77 (9)	0.4583	0.001	0.003
	Strict	111.06 (98)	0.173	0.969	0.971	0.037 [0.000–0.067]	10.60 (10)	0.3897	0.001	0.002
Sex	Configural	91.84 (88)	0.369	0.992	0.992	0.021 [0.000–0.060]	–	–	0.001	–
	Metric/weak	82.13 (79)	0.383	0.993	0.992	0.020 [0.000–0.061]	10.06 (9)	0.3452	0.001	0.001
	Scalar/strong	91.76 (88)	0.371	0.992	0.992	0.021 [0.000–0.060]	9.79 (9)	0.3681	0.001	0.001
	Strict	99.60 (98)	0.436	0.997	0.997	0.013 [0.000–0.055]	8.20 (10)	0.6094	0.005	0.008

^aThe chi-square difference value is not significant. It indicated that constraining the parameters of the nested model did not significantly worsen the fit of the model. Our result indicated measurement invariance
^bCriteria for an acceptable fit were a root mean square error of approximation of < 0.06, and a comparative fit index (CFI) and a Tucker-Lewis index (TLI) of ≥ 0.90. Configural invariance—no constraints; Full metric invariance—with all factor loadings constrained equal. Scalar invariance—with all intercepts constrained equal; Strict invariance—with all factor loadings and intercepts fixed; Measurement invariance is assumed when ΔCFI is ≤ 0.01

ISSN: 2397-0642