[R-sig-ME] who shot the predict method (and did you shoot the deputy as well?)

Ben Bolker bbolker at gmail.com
Tue Aug 28 02:21:23 CEST 2012

Paul Johnson <pauljohn32 at ...> writes:

> It seems like I can't leave for a month or two without pieces of mount
> Rushmore falling off of lme4.
> Remember last June, I posted about some troubles with the predict
> method and you fixed it.

  Quick question: did predict() *ever* exist for the stable version
of lme4?  (I don't think so, although I may of course be mistaken ...)
  Were you working with the development (r-forge) version
of lme4 before?

  I will take a look at your code ...

  Ben Bolker

> And I had some other trouble with fishy looking variance estimates
> from models specified like (1|id) + (0 +x|id).
> I stopped fighting with that because I just couldn't understand it at
> all, but today I took it up again.
> Now the fishy looking variance estimates are completely solved
> (yeah!), but somebody stole the predict function entirely:
> > m3newdat$mm4pred <- predict(mm4, newdata = m3newdat)
> Error in UseMethod("predict") :
>   no applicable method for 'predict' applied to an object of class "mer"
> Have I gotten shunted to the wrong lme4 packaging?

More information about the R-sig-mixed-models mailing list