Hi there,
I want to test for measurement invariance in Stata 16.
I want to follow the approach of testing different steps of nested models, starting with a model were all parameters are freely estimated in both groups - ginvariant(none) - and then in each step restricting one more parameter (group) to be equal across groups. (e.g., https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5145197/)
To identify the model, I use the Reference-Group Method, i.e. set latent means to 0 and latent variances to 1 in first group, estimate freely in second group (http://www.agencylab.ku.edu/~agencyl...d,%202006).pdf).
I did that before in earlier versions of stata and got, when using "estat gof, stats(all)" afterwards, the Chi2 values and other fit indices, at least as far as I remember. I need those values so that I can compare the model with the free estimates to the model with one more parameter constrained via the Chi2 difference test and Delta CFI.
However, when I use "estat gof, stats(all)" in Stata 16, I only get some indices (AIC, BIC, SRMR, CD), but not Chi2 and CFI.
So what can I do to either get the proper fit indices or to test measurement invariance in another (scientifically correct) way in Stata 16?
Here is the command for the first model I want to test:
sem (Latent1 -> L1_1, ) (Latent1 -> L1_2, ) (Latent1 -> L1_3, ) (Latent2 -> L2_1, ) (Latent2 -> L2_2, ) (Latent2 -> L2_3, ), covstruct(_lexogenous, diagonal) group(orgsize) ginvariant(none) byparm vce(oim) var(1: Latent1@1) var(1: Latent2@1) mean(1: Latent1@0) mean(1: Latent2@0) cov( Latent1*Latent2) nocapslatent
The commend itself works well and I get proper results.
But as said, when using "estat gof, stats(all)" afterwards, I don't get Chi2 values. I think in earlier versions of Stata, I got them...
It would be great if anyone could help me with that. Thanks a ton in advance!!!
Kind regards
Franzi
I want to test for measurement invariance in Stata 16.
I want to follow the approach of testing different steps of nested models, starting with a model were all parameters are freely estimated in both groups - ginvariant(none) - and then in each step restricting one more parameter (group) to be equal across groups. (e.g., https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5145197/)
To identify the model, I use the Reference-Group Method, i.e. set latent means to 0 and latent variances to 1 in first group, estimate freely in second group (http://www.agencylab.ku.edu/~agencyl...d,%202006).pdf).
I did that before in earlier versions of stata and got, when using "estat gof, stats(all)" afterwards, the Chi2 values and other fit indices, at least as far as I remember. I need those values so that I can compare the model with the free estimates to the model with one more parameter constrained via the Chi2 difference test and Delta CFI.
However, when I use "estat gof, stats(all)" in Stata 16, I only get some indices (AIC, BIC, SRMR, CD), but not Chi2 and CFI.
So what can I do to either get the proper fit indices or to test measurement invariance in another (scientifically correct) way in Stata 16?
Here is the command for the first model I want to test:
sem (Latent1 -> L1_1, ) (Latent1 -> L1_2, ) (Latent1 -> L1_3, ) (Latent2 -> L2_1, ) (Latent2 -> L2_2, ) (Latent2 -> L2_3, ), covstruct(_lexogenous, diagonal) group(orgsize) ginvariant(none) byparm vce(oim) var(1: Latent1@1) var(1: Latent2@1) mean(1: Latent1@0) mean(1: Latent2@0) cov( Latent1*Latent2) nocapslatent
The commend itself works well and I get proper results.
But as said, when using "estat gof, stats(all)" afterwards, I don't get Chi2 values. I think in earlier versions of Stata, I got them...
It would be great if anyone could help me with that. Thanks a ton in advance!!!
Kind regards
Franzi
