|
Message/Author |
|
JackeJR posted on Tuesday, May 10, 2016 - 12:59 am
|
|
|
I am developing a MIMIC model and would like to clarify on the steps to develop the model. I have the following model and I want to test for measurement invariance for AGE: Step 1 MODEL f1 BY y1-y4; f2 BY y5-y8; f3 BY y9-y12; f1-f3 ON AGE@0; y1-y12 ON AGE@0; OUTPUT: STANDARDIZED MODINDICES(3.84); Step 2 Examine the modification indices and add 1 path at a time. Question: 1. Is what I outlined correct? OR should we be adding all the direct and indirect effects and see which is significant? 2. For Step 2 in what order should I be adding the paths? Indirect then direct or vice versa? Do I a) build up the model incrementally i.e. add path 1 -> check chi-sq diff -> if better, add paths 1 and 2 and repeat until all paths have been added OR b) should I add each of the path in the modification index separately, do a chi-sq difftest to determine which makes significant changes and then have a final model with all the significant paths? |
|
|
Start with the model that you think is closest to the best model. Then check the modindices. |
|
Back to top |
|
|