|
|
Question on multiple group analysis |
|
Message/Author |
|
|
I'm conducting multiple group analysis (using sex, family structure, and age separately) and am unsure of how to use the analysis to find if a difference exists between the two groups. I've ran the initial analysis using Type = mgroup and placing a number in parentheses after each line of the model which I assumed holds all the pathways equal. Is this correct? Also, do I delete these numbers (one at a time) and run the analyses again to find out if any pathways are different? I'm not sure how to find out exactly if there is a difference in the model for each group based on running the analyses with the Type=mgroup and holding each pathway equal to one another. Can you offer any direction in how to do this? Thank you. |
|
|
If you are interested in testing for measurement invariance, see the section on this topic after the discussion of multiple group analysis in Chapter 13. See also the MODEL TEST command which carries out Wald tests. |
|
|
Hi Bengt and Linda, I thought about reporting the results of the Wald test as an F-test [i.e. (F(XX, X) = Y.YY, p = .YY)]. While regarding the Ys values it is clear what to report, the problem arises with the Xs (df). How shall the result look like with: 1) sample N =800 2) two-groups model 3) 7 estimates tested across models (7 df in the Wals output)? or is there another way to report the Wald test? thank you |
|
|
Here's how we report the Wald test: Wald Test of Parameter Constraints Value 398.752 Degrees of Freedom 1 P-Value 0.0000 |
|
Back to top |
|
|