extracting p values for main effects of binomial glmm
I'd like to suggest that the phrase "we can't discuss main effects in the presence of a statistically significant interaction" isn't so cut-and-dry. It depends. If the size of the main effects is far greater than additional interaction effect, then one can certainly talk about main effects. The catch is knowing about "practical" or "subject matter" significance as it is not solely a statistical issue. It is the "interpretation" of results that can be problematic and not necessarily the fault of SAS or R or any other software package that provides the results. Jim -----Original Message----- From: R-sig-mixed-models [mailto:r-sig-mixed-models-bounces at r-project.org] On Behalf Of Ken Beath Sent: Wednesday, March 04, 2015 2:57 PM To: Megan Kutzer Cc: r-sig-mixed-models at r-project.org Subject: Re: [R-sig-ME] extracting p values for main effects of binomial glmm That is what I though you meant. In that case you can't discuss main effects at all, as the effect of diet, for example, is different for each combination of infection status and day. SAS and some other software will attempt to give results but they aren't usually sensible.
On 5 March 2015 at 09:44, Megan Kutzer <makutzer at gmail.com> wrote:
No, sorry, the model is Diet + infection status + day and all the two way interactions and the 3 way interaction. On 4 Mar 2015 23:34, "Ken Beath" <ken.beath at mq.edu.au> wrote:
Did yo mean to have interactions between all 3 as "Diet * Infection status * Day". With interactions it isn't possible to test for the effect of main effects. On 5 March 2015 at 07:11, Megan Kutzer <makutzer at gmail.com> wrote:
Hi,
I'm fairly new to mixed models and have done a lot of reading
without much success. Unfortunately there is no one at my
institution who is really familiar with them so I thought I would
try this list.
I'm running a binomial generalized linear mixed effects model and I
need p-values for the main effects. I know this isn't entirely
correct with this type of model but my supervisor wants the
p-values!
The model is:
glmer (Proportion hatched ~ Diet * Infection status * Day +
(1|SubjectID) +
(1|Day), family=binomial)
where,
Proportion hatched = cbind(Offspring, Eggs-Offspring) Diet is a
factor with 2 levels Infection status is a factor with 4 levels Day
is a factor with 3 levels
Using Subject ID number and Day as random effects is supposed to
control for pseudoreplication in the model, although I am not
entirely sure that this is specified in the correct way. I wanted to
include experimental replicate here too but the model failed to converge.
My question is: is there a way to get p-values for the main fixed
effects of Diet, Infection and Day?
If you need more specific model information or the model output I
would be happy to provide it.
Thanks,
Megan
[[alternative HTML version deleted]]
_______________________________________________ R-sig-mixed-models at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
-- *Ken Beath* Lecturer Statistics Department MACQUARIE UNIVERSITY NSW 2109, Australia Phone: +61 (0)2 9850 8516 Building E4A, room 526 http://stat.mq.edu.au/our_staff/staff_-_alphabetical/staff/beath,_ken / CRICOS Provider No 00002J This message is intended for the addressee named and may contain confidential information. If you are not the intended recipient, please delete it and notify the sender. Views expressed in this message are those of the individual sender, and are not necessarily the views of the Faculty of Science, Department of Statistics or Macquarie University.
-- *Ken Beath* Lecturer Statistics Department MACQUARIE UNIVERSITY NSW 2109, Australia Phone: +61 (0)2 9850 8516 Building E4A, room 526 http://stat.mq.edu.au/our_staff/staff_-_alphabetical/staff/beath,_ken/ CRICOS Provider No 00002J This message is intended for the addressee named and may...{{dropped:16}}