Message/Author 

pyen posted on Saturday, January 02, 2010  6:04 pm



Hi, I couldn't find principal axis factoring in Mplus. I saw someone mentioned that ULS is similar to principal axis factoring, is that right? I did principal axis factoring+oblimin in SPSS, and ULS+oblimin in mplus, but the factor loadings are different. Did I miss anything? Thanks!! 


ULS is the same as the MINRES method described in Harman's factor analysis book. The equivalence is pointed out e.g. by Joreskog (2007) in the book Factor Analysis at 100. Principalfactoring is an older and different method. I would use ULS or ML with Geomin rotation in Mplus. 


To add on to this....so if indicators have marked nonnormality...use ULS in Mplus (i.e., as an alternate to ML)? In short, if my indicators are nonnormally distributed (i.e., some have marked skewness and kurtosis) what would be the most optimal estimator in Mplus (for EFA)? 


No, I would not hesitate to use ML also for nonnormal continuous outcomes. The sample correlations are the same as for ULS and the ML fitting function is more thorough in getting Sigmahat close to R so estimates will be ok. MLR (or MLM) provides nonnormality corrected chisquare and SEs. Only if the nonnormality amounts to floor or ceiling effects would I switch away from the usual approach (ML or ULS) and consider the outcomes as say censored. 

nanda mooij posted on Friday, December 04, 2015  5:11 am



Hi Dr. Muthen, I also have a question about the principal axis factoring method. I have to compare the results from an analysis in SPSS with principal axis factoring and a analysis in MPlus with summary data (correlation matrix) with all categorical variables. When I do this analysis, it says that due to the categorical variables and the summary data, ULS is used as estimator. Can't I use another estimator? How big is the difference between ULS and PAF? Can I compare these analysis? Thanks very much! 


ULS should be good here. There shouldn't be a big difference between the two. 

Back to top 