jamovi

Hey everyone,
I’m running a complicated path analysis with a ton of variables. I need all my mediators to be allowed to correlate in the model. I built in all the pairs using the interface but when I run it, it only includes one of the pairs instead of including all of them. I can tell because the relationships aren’t showing up in the estimates or the path model.
Help!

Hi
could you share an example of the model not working?

Well I can tell it's 'failed' because none of the covariance are in the path model. Obviously my fit is heinous but it's because the variables are related to one another and the system is not allowing them to covary. There are no estimates for the specified covariance relationships and the variables are clearly not being allowed to covary in the path model or there would be arrows connecting them (double-sided arrows), but there are none. It's sort of hard to see in my attachment but almost all the arrows are pointing to engagem, the only covariance that actually came through is between SupervRelat and SocSup

I just took a look at the actual R syntax and it looks like Jamovi is only using the last covariance listed in the syntax?

Hi, I have checked the module with a model with 30 covariances among variables and the model is estimated correctly. Maybe there is a problem with the specific model you are estimating. I am afraid, however, that it is not clear what the model should be. Could you explain how your model should be and how it is estimated?

Yes--thank you. Here is my model. The variables within the circles should be allowed to covary with one another.

Hi
you were right in spotting the issue with covariances, that was indeed there. A new version (0.9.0) is coming up next week that fixes the issue. If you need your results sooner, you can use the SEM module to achieve the same outcome. Here is the syntax of a model equivalent to yours.