Hi everyone,
I've recently shifted to using Lavaan for structural equation modeling, and I'm finding that it's pretty simple to use once you get the hang of it, and extremely powerful. I've also heard that it's similar to Mplus in terms of features and modeling procedures. With that in mind, I'd be interested in hearing from those who have experience with both programs. Are there any advantages to Mplus that I might be missing out on with Lavaan?
Thanks!
- Tim
Hi Timothy,
according to my impression (which may be outdated), lavaan is still not capable of fitting mixture and multilevel SEM. Moreover, there may be differences with regard to the speed of model fitting. However, Bayesian SEMs have been recently become available with the blavaan package. Finally, the R interface of lavaan is to my mind much more convenient. Unless you want to fit the former or very large datasets, there is probably no reason for sticking to the proprietary M+ suite.
Best, Robert
Hi Timothy,
according to my impression (which may be outdated), lavaan is still not capable of fitting mixture and multilevel SEM. Moreover, there may be differences with regard to the speed of model fitting. However, Bayesian SEMs have been recently become available with the blavaan package. Finally, the R interface of lavaan is to my mind much more convenient. Unless you want to fit the former or very large datasets, there is probably no reason for sticking to the proprietary M+ suite.
Best, Robert
Interesting discussion. Compared to Mplus and Lavaan, I wonder how does Amos weigh on important attributes?
Best wishes!
Hi Timothy;
The lavaan package in R requires writing equations. But there's no need to write equations in Mplus. MplusAutomation package in R can be used to combine Mplus and R.
Best regards.
As Robert stated, some features are not yet implemented in lavaan, such as multi-level SEM. But aside from these very few features, lavaan is equally powerful.
In my opinion, the biggest benefit of lavaan is that you have all of R at your disposal. This implies small stuff, such as no limit of characters per line of code (that always bugs the hell out of me in Mplus^^), to extremely handy stuff, such as further extraction of and computation with fit values.
The other obvious benefit is that R is free.
Best of luck,
Georg
Thanks for the feedback, everyone. Based on the responses, it does look like lavaan covers everything I need for now. Following recent links and some of the chatter on the lavaan Google group, it also looks like Yves Rosseel is working on implementing multilevel SEM in an upcoming version of lavaan: https://github.com/yrosseel/lavaan/commit/cda251d91df328aa188de8a1f3a29036379a9829
http://lavaan.ugent.be/development.html
http://www.educational-measurement.uni-jena.de/Events/Spring+School+2017/Workshops/Workshop+5_+Yves+Rosseel.html
Plus, I totally agree with Georg that the ability to incorporate other R functions alongside lavaan pushes it over the top.
I know that this question is pretty old by now. But I saw it in my search results when I was searching for Lavaan.
As of May 2018, Lavaan is capable of fitting mixture and multilevel SEM. (However, random slopes is not incorporated yet.)
This means that Lavaan can do most of what Mplus can, and due to the R environment Lavaan can do some more stuff that Mplus can't. Also, as said it is free.
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Materials Science
- Materials
- Inorganic Materials
- Ceramics