cui bunny posted on Tuesday, December 05, 2006 - 5:38 am
For testing groups invariance, If the groups are dependent, but the cases cannot be matched, i.e. the same survey instrument is filled out at two time periods. Many of the subjects from time 1 also fill out the survey at time 2; however, anonymity requirements make it impossible to match the respondents over time. If one were to carry out a test for invariance between the two groups (time periods) ignoring the fact that the groups are dependent, what would be the effect of the dependence on the invariance tests?
I don't think you can look at the two data sets together. You can't use them a independent groups because they contain some of the same observations. And you can't treat them as repeated measures because you can't match them. Also, if you can't match them, you can use only covariance matrices as input not individual observations.
I am deleting your other post of this question. Please post only once.