Gabriel posted on Tuesday, October 04, 2005 - 7:49 pm
Dear all, I am trying to fit al longitudinal factor model in several groups (school types). The problem is that that the number of measurement occasions differs across groups. I have three occasions for the first two groups but only two occasions for the second two groups. Is there a possibility for fitting such a model in Mplus (three factors with five indicators = 15 manifest variables in the first two groups, and two factors with five indicators = 10 manifest variables in the last two groups)? As far as I know Mplus requires that the same variables are measured in all groups. I hope that there is a “trick” that can be applied to this problem.
bmuthen posted on Saturday, October 08, 2005 - 7:54 pm
Different number of measurement occasions across groups is not a problem, but different measures across groups is a problem. It would be ok if at least some measures were the same across all groups (or at least pairs of groups), but if not you don't have the possibility of having your growth outcome on the same scale which is necessary if you want to compare growth exactly across groups. I say "exactly" because you can always do separate growth analyses in the different groups - with different outcome measures - and qualitatively compare the results.