Two groups with different numbers of indicators

Chacolex · Post by **Chacolex** » Tue Apr 30, 2013 9:30 pm

Hi,

I am working on my doctoral dissertation and I am having trouble with figuring out how to load my data, as I have a group of supervisors (N=49) who have completed 90 questions and their subordinates (N=280) who only completed 35 questions. I want to compare answers across certain sub-scales, but I can't upload the CSV with both sets of data in one file due to the empty cells across crewmembers.

Is there a way to structure a CSV file so that I can include all, or does anyone know how I can get around this?

Thanks!

Hengkov · Post by **Hengkov** » Fri May 03, 2013 8:46 am

Hi,
Run multigroup analysis using permutation or non-parametric approach.

Regards,
Hengky

Chacolex · Post by **Chacolex** » Tue May 07, 2013 8:40 pm

Thanks!

christian.nitzl · Post by **christian.nitzl** » Wed May 08, 2013 10:07 am

Nevertheless, it is important to pay attention to measurement invariance. That means, that at least you have to measure your latent constructs with the same indicators for the different groups. Please check this forum for the topic of measurement invariance and you will find some important hints.
Greetings,
Christian

Hengkov · Post by **Hengkov** » Wed May 08, 2013 10:16 am

Hello Nitzl,
Yes, right.
But I think almost case in multigroup analysis have invariance. So, Smith-Satterwaite test just solution. We need develop another procedure for multigroup invariance.

Best Regards,
Hengky

christian.nitzl · Post by **christian.nitzl** » Wed May 08, 2013 10:24 am

You right, we really need better procedures. But “at least” you need the “same” indicators for the measurement when using a multigroup analyses. In the next step we have to check if the “loadings” are significantly different with such a test you mentioned e.g. “parameter free” test.
Greetings,
Christian

Hengkov · Post by **Hengkov** » Wed May 08, 2013 11:04 am

Yes, I know Nitzl. ;-)