1
$\begingroup$

I am a PhD student in the field of economics. As part of my research work I am setting a SEM model. I have designed two models that present acceptable goodness-of-fit statistics respectively. Nevertheless, I’ve got doubts about which one would be my best choice. Model 1 would be my preferred option as it incorporates more observable variables than Model 2. However the fit statistics of Model 1 seem “poorer” than those of Model 2. For instance, the p-value of the Chi-Square of Model 1 is 0.001 < 0.05.

Do you think that the fit statistics of Model 1 are acceptable, or should I clearly incline my decision for Model 2?

The SEM graphs and fit results are as follows:

enter image description here

enter image description here

$\endgroup$

1 Answer 1

1
$\begingroup$

Model 2 is a big improvement over model 1. Model one is acceptable, by pretty much all measures, model two is barely acceptable by several measures.

Do the modification indices tell you anything useful about why dropping obs2 improves fit so much.

(Also, that's a slightly unusual way of drawing a second order CFA.)

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.