Message/Author |
|
|
Hello, is there a way to analyze measurement invariance and the mean structure with trend data? I have multiple datasets of the same cohort at different timepoints and want to compare CFA models. Each dataset at each timepoint consists of about 3000 cases which are nearly the same persons but not linked together as in panel studies. So for each case there is only one valid measurement (at one timepoint) but on the aggregate level I have repeated measurement of the same variables. I suspect that it is not possible to do multigroup CFA because the data is pooled and the assumption of independence is violated. What can I do? Thank you for your answers Andreas |
|
|
You test measurement invariance across time. See the multiple indicator growth example in the Topic 3 or 4 course handouts on the website. The beginning of that example shows how to do this. |
|
|
Hello, thank you for your answer. I looked at the examples (NLSY, GBG) and I suspect that they all base on true panel data? Isn't it true that you always need panel data to do this kind of analyzes? My problem is that I don't have panel data. We asked the same kohort (in fact nearly the same pupils) in several years. But the cases are not linked together. So in my thinking I can neither test invariance over time because I have no true panel data nor can I make a multigoup analysis because the samples (my groups) are not independent but strongly pooled? |
|
|
The same people should be repeatedly measured for this method. You say you have nearly the same pupils in which case the ones who were not measured at each time point could be represented by missing data. If you do not have the same students, you can do a multiple group analysis. |
|
Back to top |