|
|
| Message/Author |
|
|
| arif özer posted on Sunday, July 15, 2018 - 6:45 am
|
|
|
Dear Authorized,
I want to compare three models through ulsmw estimator with categorical data. Which one is 4-related factors (h), the second one is 4-unrelated factors (a1), and the third one is one factor model (a2). why is chisquare 1390 for 1. diff. test (h-a1), while second diff test 326 (h-a2)?
I used difftest command as stated in mplus manual.
the scale has multifactor, in that case, Should the first diff test be greater than the second one?
Thank you very much |
|
|
|
|
Size comparisons of Difftest values cannot be made.
Also, you can use the NESTED option to check if models are nested. See
http://www.statmodel.com/download/Version%208.1%20Language%20Addendum.pdf |
|
|
| arif özer posted on Monday, July 16, 2018 - 2:42 pm
|
|
|
Thank you very much, but I'm confused that the output warns that "The chi-square value for ULSMV cannot be used for chi-square difference testing in the regular way. Instead ULSMV difference testing is done using the DIFFTEST option."
When I run the syntaxes using nested command that you suggest me, output gives me that (h-a1) is nested but (h-a2) is not.
what do you suggest in this case?
Thanks |
|
|
|
|
| Non-nested models can be compared using BIC which requires ML. |
|
|
| Back to top |
|
|
