1
$\begingroup$

In the cLDA (constrained longitudinal analysis) analysis the baseline value is already taken as a response, so it doesn't have to be included on the right side of the model.

But in MMRM ANCOVA (mixed-model for repeated measures; not a mixed model - it's just generalized least square) it's common to adjust the values (either raw or change from baseline) for the baseline on the right side. But should be then the baseline included in the response variable? If not, then how otherwise calculate contrasts vs. baseline (without doing it via modelling change)?

Improve this question
$\endgroup$
3
  • $\begingroup$ Can you please give the full names for the acronyms? $\endgroup$
    – Erik Ruzek
    Commented Nov 17, 2022 at 22:26
  • $\begingroup$ By the way, technically a MMRM can be expressed as a mixed model (with a multivariate random effect vector with one component per timepoint), it's just that it's more typically explained like a linear model with correlated residuals. $\endgroup$
    – Björn
    Commented Oct 11 at 11:24
  • $\begingroup$ Also, see this answer, that already seems to address your question. $\endgroup$
    – Björn
    Commented Oct 11 at 11:26

1 Answer 1

Reset to default
0
$\begingroup$

Typically you have change from baseline in the response variable, and you can still have baseline as a covariate in the MMRM. Does that help?

Improve this answer
$\endgroup$

Not the answer you're looking for? Browse other questions tagged or ask your own question.