1
$\begingroup$

I want to examine measurement invariance across two nationalities for a construct measured with 3 items. First one is supposed to examine the configural fit before examining more restrictive models. However, when I have a three items model, the configural model has 0 df and the statistics are:

Chi2(df) = 0.00 (df= 0)

change in Chi 2(df) = NA

P = NA

CFI = 1

RMSEA = 0.000

Can I say something about the fit from these results? How should I interpret these results and write about this in a report?

I have used lavaan and semTools (packages in R), using the following code:

Code for a CFA model with 3 items and measurement invariance

model_3_items <-'
factor =~  item1 + item2 + item3
'
fit <- cfa(model_3_items, data=data, estimator="MLM")
summary(fit, fit.measures = T, standardized=T,  modindices=TRUE)

measurementInvariance(model = model_3_items, data = data, group = "Gender", estimator="MLM" , strict=TRUE)```


Thanks, 
Gorp  


$\endgroup$

1 Answer 1

0
$\begingroup$

I found help fort his question in the very helpful lavaan google group, so I post the answer here:

"Can I say something about the fit from these results? How should I interpret these results and write about this in a report?"

"No, the model is just-identified. You have no extra information (df) with which to assess the fit of the configural model, but you can still test more constrained models in which parameters are equated."

$\endgroup$
1
  • $\begingroup$ but is it okay to test models with constraints (e.g., metric, scalar models) when support for the configural model has not been obtained?? $\endgroup$
    – llbia
    Dec 22, 2020 at 2:16

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.