|
|
 |
| Use summary data for SEM with binary ... |
 
|
| Message/Author |
|
|
| spring posted on Wednesday, November 02, 2016 - 1:15 pm
|
|
|
I am trying to run SEM model with binary observed variables. Yet, as I am using data provided by the Stats department, it is not allowed to take original data (at individual level) out of their lab.
So, would the following work?
Use Stata to obtain the tetrachoric correlation matrix (which could be taken out the lab), and run Mplus using the correlation matrix as the input.
If this would work, do I still need to specify the data is categorical? Which estimator should I use, WLSMV or MLM?
Thanks a lot! |
|
|
|
|
| Use ULSMV. |
|
|
| Back to top |
|
|
