Message/Author |
|
|
I have 4 groups I'd like to compare on a common instrument. All of the indicators in the instrument are multiple choice items scored correct/incorrect (1/0). I want to do a very basic and straightforward test of MI across these groups. Can someone help me out with some advice or possibly something replicable? I'm a newbie so go lightly :-) |
|
|
Here is an example used to illustrate MI testing of a 2-factor model across 34 countries using version 7.11 (see the "slide 13" run on our website for data). Configural, metric, and scalar invariance is tested in a single run. data: file = issp.txt; variable: names = cn v21 v22 v26 v29 v35; usev = v21-v35; missing = v21-v35 (0 8 9); grouping = cn(34); analysis: estimator = ml; model = configural metric scalar; model: nat by v21 v22; pat by v26 v29-v35; output: tech1; |
|
SG posted on Thursday, August 15, 2013 - 10:38 am
|
|
|
Thank you for this! Would the same type of set up work in the case of binary outcome variables like I have? |
|
|
Yes. You would need to use the CATEGORICAL option to specify the binary variables. |
|
Back to top |