Skip to main content
. Author manuscript; available in PMC: 2012 May 1.
Published in final edited form as: Med Care. 2011 May;49(5):480–488. doi: 10.1097/MLR.0b013e31820fb944

Table 2.

Goodness of Fit for SF36 Multigroup Factorial Invariance Testing

Model Description RMSEA (90% CI) SRMR CFI χ2 df p* Ref ΔRMSEA ΔCFI Δχ2 Δdf p* for Δχ2
1 Unconstrained Model 0.028 (.027–.030) 0.059 0.937 4358 2152 0.001
2 Weak Invariance (1st Order Factor Weights) 0.029 (.028–.030) 0.057 0.931 4664 2248 0.001 1 0.001 −0.006 306 96 0.000
3 Weak Invariance (1st & 2nd Order Factor Weights)** 0.029 (.028–.030) 0.058 0.929 4731 2265 0.001 2 0.000 −0.002 67 17 0.000
4 Strong Invariance (Intercepts) 0.032 (.031–.033) 0.057 0.909 5526 2370 0.001 3 0.003 −0.020 795 105 0.000
5 3 Group Strong Invariance (B=W=HS not HE) 0.032 (.031–.033) 0.058 0.911 5422 2335 0.001 3 0.003 −0.018 691 70 0.000
6 3 Group Strong Invariance (B=W=HE not HS) 0.030 (.029–.031) 0.058 0.923 5029 2335 0.001 3 0.001 −0.006 298 70 0.000
7 Error Variances/Covariances(B=W=HE not HS) 0.031 (.030–.032) 0.060 0.915 5362 2415 0.001 6 0.001 −0.008 333 80 0.000

Notes:

*

The bootstrapped Bollen – Stine p-value is reported because of significant (p<.01) multivariate non-normality.

**

Factor weights/covariances are constrained equal for Blacks, Whites and Hispanic English (Hispanic Spanish are unconstrained).

Also termed strict invariance.