ULS in Amos vs. WLS in MPLUS PreviousNext
Mplus Discussion > Confirmatory Factor Analysis >
Message/Author
 Ke Anne Zhang posted on Thursday, February 19, 2009 - 7:26 am
I am conducting a CFA with ordinal data (0-3 Likert scale; n = 425), and am wondering which method would produce the most reliable and valid results. First, I used ULS in Amos, which only output the chi-square, GFI, AGFI, RMR, NFI, RFI, and PNFI. Every model showed a good fit according to this procedure. Then, I reran the CFA using WLSMV in Mplus, which was able to output the chi-square, CFI, TLI, RMSEA, and SRMR. However, none of the models showed a good fit according to this procedure.

Would you have any recommendations as to which results are more trustworthy?
 Linda K. Muthen posted on Thursday, February 19, 2009 - 11:18 am
In Amos the variables are being treated as continuous variables. In Mplus, they are being treated as categorical variables. Treating them as categorical should yield the most valid results.
 Ke Anne Zhang posted on Thursday, February 19, 2009 - 11:23 am
Dr. Muthen,

Thank you for such quick advice!

Anne
 Ke Anne Zhang posted on Friday, February 20, 2009 - 9:02 am
In a sample as large as mine, with Likert scales... Is there ANY reason at all to prefer ULS over WLSMV?
 Ke Anne Zhang posted on Friday, February 20, 2009 - 9:06 am
I should have also mentioned that I ran all of the models in MPLUS using ULS as well as WLSMV. The results from both methods were quite similar to each other in terms of how well the models were fitting in comparison to each other.

Thanks,

Anne
 Linda K. Muthen posted on Saturday, February 21, 2009 - 8:38 am
I know of no reason to prefer ULS over WLSMV.
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: