Multiple group growth curve-complete ...
Message/Author
 Michelle Tollit posted on Sunday, May 09, 2010 - 11:51 pm
Hi,
We are trying to run a multiple group growth curve model. In the first group we have data on young people aged 10-17. In the second group, we only have data on students from the ages of 13-17, and so therefore the second group has complete missing data on the variable of interest from ages 10-12.

When we try to run the multiple group growth curve, Mplus does not seem to be able to run the model when a group has complete missing data on a variable at one or more time points (ie ages). Could you please advise how we might be able run this model taking into account the problem just described.
Many thanks.
 Linda K. Muthen posted on Monday, May 10, 2010 - 8:03 am
For multiple group analysis you need the same variables in each group. You would need to compare on the shared set of variables.
 Michelle Tollit posted on Monday, May 10, 2010 - 6:53 pm
Hi Linda,

Following on from my previous post, I would just like to clarify a couple of things.

The variable of interest is “LSC” and is structured by age (as indicated by _10=age 10, _11=age 11 _12=age 12 etc. as seen below):

LSC_10, LSC_11, LSC_12, LSC_13, LSC_14, LSC_15, LSC_16, LSC_17

Both of the groups that we want to include in the multiple group analysis were measured on the same variables, however because the second group (who are an older cohort) were only surveyed over 2 waves, compared with the first group who were surveyed over 6 waves, all of the second group have 3 variables where there is no data on them (ie. age LSC_10-LSC_12), as the first wave of data collection for this group occurred when most students were aged 13/14.

So, are you saying that, as the first group has data at all 8 age groups as presented above, then the second group should also have data at all 8 age points in order to run the multiple group analysis?

If this is the case, could you offer any suggestions regarding how we may be able to run these growth curves, taking into account these 2 groups.

Many thanks
Michelle
 Linda K. Muthen posted on Tuesday, May 11, 2010 - 9:47 am
If these are cohorts from the same population, you could analyze the two cohorts together. If you want to compare them, you can do so only the the time points they have in common.