Part of results from ULS and DWLS estimators are different

[Chao Xu]

Hi all,

I estimated my bifactor model with these two estimators. The results turn out to be very similar to each other for the most part. But there are several factor loading estimates differ quite a lot. Under these circumstances, is there a guideline as to which of these estimators should be chosen (i.e., more reliable)?

I noticed one paper on Structural Equation Modeling in 2009 suggests ULS over DWLS (see below). So any thoughts on this issue?

https://www.tandfonline.com/doi/full/10.1080/10705510903203573

Thank you,

Chao

[Terrence Jorgensen]

any thoughts on this issue?

I find it more useful to post general questions on SEMNET, which has a much wider set of experts who might have read something more recent that could be informative for you.

http://www2.gsu.edu/~mkteer/semnet.html

Terrence D. Jorgensen

Postdoctoral Researcher, Methods and Statistics

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

UvA web page: http://www.uva.nl/profile/t.d.jorgensen

[Chao Xu]

Hi Terrence,

Thanks for your reply. I appreciate your help as always.

I posted the issue here because I am using different least squares estimators in lavaan. So I figure it might be appealing to other lavaan users if I put up my questions here for discussion. But anyway, thanks for the suggestion as to posting my questions on a more specified forum.

Chao