by mcfanda@gmail.com » Tue Dec 29, 2020 4:10 pm
HI
you can use GAMLj Mixed model, which is a module of jamovi.
Assume you have time=0 vs 1, and group (0 vs 1). Your data should be in the long format, so you also have a column with ID for patients. Every patient will have two rows in the dataset, one for time=0 and the other for time=1.
Launch a GAMLj mixed model, set the dependent variable and put time and group as covariates. Go to covariates scaling and set both to "None". Crucially, the fixed effects should be: time and time*group, meaning no main effect of group. You'll notice that in the Fixed Effects panel the software puts all main effect . Remove the main effect of group, this leaves in only the effect of time. On the left panel of Fixed Effects, select both group and time, click on the second arrow dropdown icon, and select "All 2 way". Now you have on the right only the main effect of time and the interaction. That is the cLDA model you're looking for.
Set the random components as "intercept|ID".
Look at the results, the interaction effect `group:time` is the constrained effect of group that you get in a cLDA. All standard errors and F tests are all equal to a cLDA
Here detailed explanations https://datascienceplus.com/taking-the-baseline-measurement-into-account-constrained-lda-in-r/