R help - Sep 2014 - Error in quantile.default(resids) : missing values and NaN's not allowed if 'na.rm' is FALSE

About a year ago I ran some analyses using lmer. The general syntax was:

mlm <- lmer(var1 ~ (1|dyad) + var2 + var3*var4*var5, na.action=na.exclude);
summary(mlm)

The model ran fine and I saved the output. I've recently turned back to
those analyses, however, the model no longer runs. I get the following
error:

Error in quantile.default(resids) :
  missing values and NaN's not allowed if 'na.rm' is FALSE

I've searched the online forums and found that this topic has not been
touched upon since 2006, and at that time someone had indicated that it was
a bug that had been resolved. I am using the most current version of lme4,
so if the bug is fixed I am unsure why I am experiencing this problem.

I have since run the model using lmer's default for handling missing data
(na.omit), but found the results do not match what I had saved previously
when using na.exclude.

If anyone has an insight here I'd be very appreciative.

Kind regards,

Chad

-- 
Chad M. Danyluck
PhD Candidate, Psychology
University of Toronto
Lab: http://embodiedsocialcognition.com


?There is nothing either good or bad but thinking makes it so.? - William
Shakespeare

	[[alternative HTML version deleted]]

Chad Danyluck <c.danyluck <at> gmail.com> writes:
> 
> About a year ago I ran some analyses using lmer. The general syntax was:
> 
> mlm <- lmer(var1 ~ (1|dyad) + 
> var2 + var3*var4*var5, na.action=na.exclude);
> summary(mlm)
> 
> The model ran fine and I saved the output. I've recently turned back to
> those analyses, however, the model no longer runs. I get the following
> error:
> 
> Error in quantile.default(resids) :
>   missing values and NaN's not allowed if 'na.rm' is FALSE
> 
> I've searched the online forums and found that this topic has not been
> touched upon since 2006, and at that time someone
>  had indicated that it was
> a bug that had been resolved. I am using the most current version of lme4,
> so if the bug is fixed I am unsure why I am experiencing this problem.
   You missed

https://mailman.stat.ethz.ch/pipermail/r-sig-mixed-models/2014q3/022616.html

and the subsequent thread, which describes a recent issue.

This was fixed two weeks ago, at

https://github.com/lme4/lme4/commit/e68b7475d1d254bfbdae6cd3efb2d9c8dbe0b899

but hasn't made it into a released version of lme4, although at least
a Windows binary should be available @ http://lme4.r-forge.r-project.org/repos
(and you can install from Github via devtools::install_github() if you
have compilation tools available on your machine).
> I have since run the model using lmer's default for handling missing
data
> (na.omit), but found the results do not match what I had saved previously
> when using na.exclude.
> 
> If anyone has an insight here I'd be very appreciative.
> 
> Kind regards,
> 
> Chad

  Perhaps worth noting that this bug only affects the summary print
method, not anything in the actual model fitting process.