Beyond Mars and Venus: An Examination of Gender Differences in Academic Performance of High School Students - Innovative Management in Information and Production - page 33

Tabl e 4 Summary statistics of fit for four models

Std

x 2

Model ( p value) RMR GFI AGFI NFI

Model 1 4697.85 0.045 0.903 0.773 0.923 0.190

( df = 9 ) (0.000)

Model 2 3697.4 0.12 0.979 0.926 1.000 0.093

( df = 6 ) (0.000)

Model 3 2162.1 0.041 0.996 0.992 0.977 0.0407

( df = 8 ) (0.000)

Model 4 1824.80 0.012 0.997 0.992 0.971 0.0327

( df = 8 ) (0.000)

Tabl e 5 Test of invariance for males and females

x 2 x 2

Hypothesis Δ P value RMSEA GFI NFI

A: H ∑ 1502 0.000 0.05 0.90 0.95

B: H n = 2 1896 394 0.000 0.23 0.75 0.84

3.1 Model Fit

Tab le 4 shows the statistics for testing the fit of the four models. For each of the
models, chi-square values were large, with the one-factor model being the largest.
Because of the large sample size used in this analysis, all of the chi-square tests
were highly significant. The descriptive indices show that in general, all models
fit reasonably well. But it appears that models 3 and 4 provide better fit than
the other two models. In order to improve the model fit, the modification indices
were examined for each model. Unfortunately, most of the modification indices
suggested do not make much sense from a substantive point of view, and some
of the estimated values of the parameters cannot be easily interpreted. Even if
some of the modification indices seem to make sense, there exists the identification
problem given there are only six variables included in this study. Thus, no model
refinement was made, and model 4 was chosen as the baseline model to do the
group comparison.

3.2 Testing Invariance

Tab le 5 shows the results for the test of group invariance across genders. Although
the chi-square statistic for testing the invariance of covariance matrices was
significant, two of the descriptive model-fit indices, GFI and NFI, indicate that two