bsem with binary outcome variable

[Rosie]

Hello,

I saw that there was a post from 2018 that suggested blavaan (bsem) was not yet ready for using categorical data. I'm wondering if, for the function bsem, this has since changed to be able to use a binary outcome variable in an SEM?

Many thanks for any guidance on this!

Rosie

[Terrence Jorgensen]

I saw that there was a post from 2018 that suggested blavaan (bsem) was not yet ready for using categorical data. I'm wondering if, for the function bsem, this has since changed to be able to use a binary outcome variable in an SEM?

Not yet. 

Terrence D. Jorgensen

Assistant Professor, Methods and Statistics

Research Institute for Child Development and Education, the University of Amsterdam

http://www.uva.nl/profile/t.d.jorgensen

[CLee]

Hello,

Can we use binary outcome variable now?

Many thanks.

Lee

[CLee]

Just to make sure, by "outcome variable", I mean the observed variables that feed into the latent variables.

[Mauricio Garnier-Villarreal]

Lee

Not yet. The good news is that this is one of the next big additions to blavaan. But, its hard to say when it will be available

As a temporary alternative, in this github I have a series of Stan code (and user friendly functions) for some IRT models, they might be useful

https://github.com/maugavilla/IRTStan

[CLee]

Thank you. I just looked into IRT. It seems that it requires a large sample size, which we do not have (less than 90). Is your code also applicable for such small sample size?

[Mauricio Garnier-Villarreal]

Lee

Couriouly, we are just working on submitting a paper about estimation methods for multidimensional IRT. My Bayesian code worked well even with small sample sizes, the average bias is around 0, and the model priors constrainted the parameter space to prevent improper solutions. The major issue is that the bias variability increases at lower sample sizes, mainly due to sampling variability as an small sample might vary further from the overall population

I cover several of these model and methods in my BSEM summer course in the Stats Camp, for BSEM I use blavaan, and for IRT I use the syntax models from the github

https://www.statscamp.org/summer-camp/advanced-sem

Hope this helps

[Yuanyuan Hu]

Hi,

Following the previous posts, does the blavaan now support categorical data? For example by specifying "categorical = TRUE"?

Best,

Yuan

[Mauricio Garnier-Villarreal]

Yes, with the same argument as lavaan, it is ordered=T