Moderation in latent change score models

[Rena]

Hi, 

I have a multiple indicator latent change score model with 2 time points, 1 latent factor (with 3 indicators) in each time point. I want to investigate whether latent change predicts a variable at a later time point and whether this relationship is moderated by another variable with 3 categories.  

Would this be possible to do in lavaan or would it be very complicated? I am finding it hard to figure out how to create product indicators with a latent change factor created by 2 other factors (time 1 and time 2 factors).

I would really appreciate your comments on this.

Many thanks!

[Keith Markus]

Could you treat this as a 3-group analysis with the 3 groups defined by the values of the moderator?  Could equality constraints across groups then be used to test moderation?

Keith

------------------------
Keith A. Markus
John Jay College of Criminal Justice, CUNY
http://jjcweb.jjay.cuny.edu/kmarkus
Frontiers of Test Validity Theory: Measurement, Causation and Meaning.
http://www.routledge.com/books/details/9781841692203/

On Thursday, February 2, 2023 at 5:28:31 PM UTC-5 Rena wrote:

[Rena]

Hi Keith, 

Thank you for your comment. Do you mean to do something like multigroup SEM first? I tried this and unfortunately, it led to model convergence issues. 

[Keith Markus]

Rena,

I am not sure that I understand what you mean by "first".  I wondered whether you could incorporate the moderation part of your model by simultaneously fitting the model to the three groups as a 3-group analysis.  It is not obvious why that would lead to convergence problems unless you fit the model with all the cross-group constraints and these were not consistent with the data generating process.  If you still get convergence problems without cross-group constraints, then I would try fitting each group one at a time to troubleshoot where the convergence problems originate from.  Of course, if you are fallible like me, it is also worth checking for syntax errors in the 3-group model that you fit.  You may have overlooked something.