[R-sig-ME] meaning of lmer formula
jfox at mcmaster.ca
Thu Sep 7 14:27:22 CEST 2017
> -----Original Message-----
> From: R-sig-mixed-models [mailto:r-sig-mixed-models-bounces at r-project.org]
> On Behalf Of Don Cohen
> Sent: Thursday, September 7, 2017 12:43 AM
> To: Ben Bolker <bbolker at gmail.com>
> Cc: r-sig-mixed-models at r-project.org
> Subject: Re: [R-sig-ME] meaning of lmer formula
> Is there some documentation I should be reading about this rather than asking
> all these questions? Other than the code, that is?
Have you read the vignette on lmer() in the package?
It also appeared as a paper in the Journal of Statistical Software.
It strikes me that your questions can also be answered by just trying the various formulas that you propose and examining the output.
I hope this helps,
John Fox, Professor Emeritus
Hamilton, Ontario, Canada
> Ben Bolker writes:
> > Separate terms in lme4 formulas are always independent.
> In a formula like "out ~ in + ((a | group) + (b | group))"
> do (a | group) and (b | group) qualify as separate terms?
> > (input1|group) + (input2|group) is problematic because both terms >
> include an intercept.
> I was imagining that this duplication was removed after the formula was
> expanded into some internal form (that I'd like to see). I guess you're saying
> that's not true.
> > (0+input2|group) can be helpful, but doesn't do > what you think when the
> variable on the LHS (e.g. input2) is a factor.
> Where can I read about what that means?
> > If input1 and input2 are both numeric (or 2-level factors) then they're > two
> independent 3x3s.
> I have trouble seeing how factors make sense on the LHS.
> Another question:
> I tried an example with about a dozen inputs inside one group
> (in1 + in2 + ... | group1) and another with the same inputs for a second group,
> and both took about a minute, and then when I used both groups the run time
> went up to about 10 min.
> Is this expected and easily explained?
> R-sig-mixed-models at r-project.org mailing list
More information about the R-sig-mixed-models