TABLE 4.

Statistics Used to Evaluate Model Fit

Statistics	Criteria for “Good” Fit	Model 1	Model 2a	Model 3b	Model 4c
χ2		69.13	19.76	306.11	49.47
df		12	8	56	33
p		<.001	.01	<.001	.03
CFI	>0.95	0.93	0.99	0.82	0.99
RMSEA	<0.08	0.12	0.07	0.11	0.04

df = degrees of freedom; p = significance value; CFI = comparative fit index; RMSEA = root mean square error of approximation.

^aThree paths and one covariance were added to the model, resulting in the loss of 4 df.

^bEight covariates were added to the model, and modification indices were evaluated to identify potential confounders.

^cThree potential covariates (years in the United States, married, and health insurance) and six paths were added between covariates and endogenous variables.