Multiple group longitudinal analysis PreviousNext
Mplus Discussion > Growth Modeling of Longitudinal Data >
 Anthony Mancini posted on Friday, May 08, 2015 - 8:38 am
I am interested in comparing groups of people (n = 4) in terms of their patterns of growth over four time points. The focus of this analysis is a hypothesized interaction between time and group.

Ordinarily this could be done in a simple repeated measures ANOVA (with a within and between subject factors). But because of missing data, I would like to take advantage of Mplus's capacity to handle missing data.

Is there a method in Mplus using either a multiple group or a structural equation format that would accommodate this analysis? Also, would there be a way to conduct post hoc analyses, such as a simple effect of time for each group? (I imagine just examining the significance of the slope parameter by group would permit this.)

Thanks very much.
 Bengt O. Muthen posted on Friday, May 08, 2015 - 8:50 am
I would use multiple-group growth modeling and as you say test for slope mean differences across the groups, e.g. using Wald testing in Model Test.
 Anthony Mancini posted on Friday, May 08, 2015 - 10:03 am
Thank you for the quick and helpful response.

Are there syntax examples that could guide my analysis?

I have searched the website and not found any that are specifically multiple group growth model examples.
 Bengt O. Muthen posted on Friday, May 08, 2015 - 10:12 am
You can start from ex 6.18.
 Evan Fishman posted on Wednesday, September 28, 2016 - 12:14 pm
I'm looking to run a model that compares two cohorts of teachers (Cohorts 1 & 2) over 4 time points. At each time point, there are two continuous dependent variables (rating 1 & rating 2).

There is missing data, and there is a covariate at times 3 and 4.

Mainly, I want to examine the cohort x time interaction, and the time and cohort main effects. Simple slopes as well.

Teachers are clustered within school, and schools are confounded within cohort (i.e., if school has more than 1 teacher, all teachers are in same cohort).

Any help is much appreciated!
 Bengt O. Muthen posted on Wednesday, September 28, 2016 - 5:49 pm
You get a lot of modeling flexibility by representing the cohorts by multiple groups.

Use school as level 2 and teacher as level 1 where the 4*2 outcomes are spread out in wide format for the growth modeling.
Back to top
Add Your Message Here
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Options: Enable HTML code in message
Automatically activate URLs in message