|
|
Rmsea with unbalanced group sizes |
|
Message/Author |
|
|
In his artcle about multilevel sem modeling, Heck (2001) states that "with unbalanced group sizes, the rmsea is incorrect, so only the chi-square can be used to estimate the overall model fit." is this still the case--that the rmsea in the mplus printout (with multilevel modeling) should not be trusted? thanks. |
|
|
He is referring to the MUML estimator which is ML only in the balanced case. For the Mplus maximum likelihood estimators, which we probably did not have when he wrote that, RMSEA is just fine. |
|
Back to top |
|
|