0
$\begingroup$

I am trying to fit a mixed model using the library lme4, and is emerging the following error.

fixed-effect model matrix is rank deficient so dropping 10 columns / coefficients.

In this case, I'm using the Effects window and time as factors. If they are considered numeric, the error becomes

fixed-effect model matrix is rank deficient so dropping 2 columns / coefficients.

Follow the syntax used.

Modelo1 <-lmer (EREF GROUP ~ + Time + window +
                GROUP:Time +
                GROUP:Window + 
                Time:Window + Time:Window:GROUP + (1 | Subject),
                data = datalong2)

From what I see, it is not estimating doi elements of interaction between GROUP: WINDOW.

Does anyone have suggestions to solve this problem?

$\endgroup$
  • 1
    $\begingroup$ Have you tried running lmer(EREF ~ Time + Window + GROUP:Time + GROUP:Window + Time:Window + Time:Window:GROUP + (1 | Subject), data = datalong2)? $\endgroup$ – Pavel Apr 22 '15 at 9:57
  • $\begingroup$ Not yet, but I'll try. $\endgroup$ – Matheus Henrique Dal Molin Rib Apr 22 '15 at 12:17
  • $\begingroup$ while framed as an R question, I believe the essential issue here is statistical. $\endgroup$ – Glen_b -Reinstate Monica Apr 23 '15 at 0:07
1
$\begingroup$

This looks like it's just a typo, i.e. specifying GROUP: + Window instead of GROUP:Window. It would be easier and less error-prone instead to specify ~Time*Window*GROUP-GROUP ... (are you sure you meant to leave out the main effect of GROUP?)

$\endgroup$
  • $\begingroup$ Removing some interactions, the model would look like, lmer(Eref ~GRUPO+Janela+Time+GRUPO:Janela+(1|Subject), data=datalong1, REML=FALSE), But the following message appears: fixed-effect model matrix is rank deficient so dropping 2 columns / coefficients What to do? $\endgroup$ – Matheus Henrique Dal Molin Rib Apr 23 '15 at 12:47

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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