[R] HLM - centering level 2 predictor

Daniel Malter daniel at umd.edu
Mon Jun 29 01:36:06 CEST 2009


without being an expert for the specific lmer error code, it seems that your
model matrix is singular, that is, not full rank. For that reason, X'X is
not invertible, which means that there is no unique solution to the problem.
Have you tried with x2 alone, without the interaction effect between x1 and
x2, and with a level 1 random effect (intercept) only? That would give a
hint where the singularity arises.

Best,
Daniel

-------------------------
cuncta stricte discussurus
-------------------------

-----Ursprüngliche Nachricht-----
Von: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org] Im
Auftrag von Sachi Ito
Gesendet: Sunday, June 28, 2009 6:05 PM
An: r-help at r-project.org
Betreff: [R] HLM - centering level 2 predictor

Dear R-helpers,

I'm analyzing a data with hierarchical linear model.  I have one level 1
predictor and one level 2 predictor, which looks like below:

fm1 <- lmer(y ~ 1 + x1 + x2 + x1:x2 + (1 + x1 | id.full))

where:
y is the outcome variable.
x1 is the level 1 predictor variable.
x2 is the level 2 predictor variable.
id.full is the conditioned variable.

It runs beautifully when only x1 is centered (I subtracted the mean from
each value).  However, when I also center x2 variable with the same
procedure, it gives me the following error message:

Warning message:
In mer_finalize(ans) : singular convergence (7)

I'd appreciate if someone could explain me what it means.

One of the differences between "non-centered values" and "centered values"
is that the "centered values" include negative values.  Could it be the
reason?  If so, what shall I do?

Thank you!

	[[alternative HTML version deleted]]

______________________________________________
R-help at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.




More information about the R-help mailing list