[R-sig-ME] interpreting Std. error from glmer output
ONKELINX, Thierry
Thierry.ONKELINX at inbo.be
Wed Sep 16 10:38:29 CEST 2009
Dear Christine,
Be carefull about that. The SE of a sum is NOT the sums of the SE! But
the variance of a sum is the sum of the variances minus the covariance.
The easiest option would be to refit the model without intercept if you
want the 'total' SE for each treatment.
HTH,
Thierry
------------------------------------------------------------------------
----
ir. Thierry Onkelinx
Instituut voor natuur- en bosonderzoek / Research Institute for Nature
and Forest
Cel biometrie, methodologie en kwaliteitszorg / Section biometrics,
methodology and quality assurance
Gaverstraat 4
9500 Geraardsbergen
Belgium
tel. + 32 54/436 185
Thierry.Onkelinx at inbo.be
www.inbo.be
To call in the statistician after the experiment is done may be no more
than asking him to perform a post-mortem examination: he may be able to
say what the experiment died of.
~ Sir Ronald Aylmer Fisher
The plural of anecdote is not data.
~ Roger Brinner
The combination of some data and an aching desire for an answer does not
ensure that a reasonable answer can be extracted from a given body of
data.
~ John Tukey
-----Oorspronkelijk bericht-----
Van: r-sig-mixed-models-bounces at r-project.org
[mailto:r-sig-mixed-models-bounces at r-project.org] Namens Christine
Griffiths
Verzonden: dinsdag 15 september 2009 18:29
Aan: William Morris
CC: r-sig-mixed-models at r-project.org
Onderwerp: Re: [R-sig-ME] interpreting Std. error from glmer output
Thank you for your help. Because I am using glmer with Binomial family I
cannot calculate 95% CIs using the mcmcsamp function and have not been
able to find a way to do this. Hence my reason to look at the SE.
I probably didn't explain myself clearly enough. What I meant is, is the
SE for a fixed effect the difference from the intercept, just as the
mean for a fixed effect needs to be calculated as a difference from the
intercept?
Your definition earlier seemed to support that it is a difference and so
I need to calculate the SE by summing the value given with the
intercept.
--On 15 September 2009 23:32 +1000 William Morris <wkmor1 at gmail.com>
wrote:
> Well, it depends.
>
> It depends on what you mean by deviance, you should clarify this (here
> is a start http://en.wikipedia.org/wiki/Deviance_%28statistics%29). In
> general, deviance is used as a measure of model fit and usually
> encountered as a component of Information criteria.
>
> Do you need to take the uncertainty (SE) in model estimates into
> account? It is probably a good idea if you are going to make
> predictions based on model estimates to also calculate predictions at
> the 95CI limits.
>
> On 15/09/2009, at 10:26 PM, Christine Griffiths wrote:
>
>> Thank you. So to clarify, I do not need to calculate the deviance of
>> the standard error from the intercept standard error, in the way that
>> I would do for the estimate?
>>
>> Cheers
>> Christine
>>
>> --On 15 September 2009 21:38 +1000 Will Morris <wkmor1 at gmail.com>
>> wrote:
>>
>>> The SE is a measure of the models uncertainty about the parameter
>>> estimates, it takes into account your sample size as well as sample
>>> variance. +_2*SE is usually a good estimate of the 95% confidence
>>> interval. In other words your treatment effect for treatment2 is
>>> probably somewhere between -.6 and -.86.
>>>
>>>
>>> On Tue, Sep 15, 2009 at 8:05 PM, Christine Griffiths
>>> <Christine.Griffiths at bristol.ac.uk> wrote:
>>>
>>> I want to plot my predictions from a model and use the standard
>>> error output as a measure of dispersion as I am unable to calculate
>>> confidence intervals with mcmcsamp as I have a binomial
>>> distribution.
>>>
>>> I know that the estimates are deviations from the intercept.
>>> Fixed effects below:
>>> Estimate Std. Error z value Pr(>|z|)
>>> (Intercept) 2.90836 0.34041 8.544 < 2e-16 ***
>>> treatment2 -0.73507 0.12986 -5.660
>>> 1.51e-08
>>> ***
>>> treatment3 -1.20052 0.12371 -9.705 < 2e-16 ***
>>>
>>> So the estimate for treatment 2 is 2.9 + -0.73. Are standard errors
>>> also deviations from the intercept? i.e. 0.34 + 0.13 for treatment
>>> 2?
>>>
>>> Many thanks
>>> Christine
>>>
>>> _______________________________________________
>>> R-sig-mixed-models at r-project.org mailing list
>>> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>>>
>>>
>>>
>>>
>>> --
>>> Will Morris
>>> Masters of Philosophy candidate
>>> Vesk Plant Ecology Lab
>>> The School of Botany
>>> The University of Melbourne
>>> Australia
>>> Phone: +61 3 8344 0120
>>> http://www.botany.unimelb.edu.au/vesk/
>>
>>
>>
>> ----------------------
>> Christine Griffiths
>> PhD student
>> School of Biological Sciences
>> University of Bristol
>> Woodland Road
>> Bristol BS8 1UG
>> Tel: 0117 9287593
>> Fax 0117 3317985
>> Christine.Griffiths at bristol.ac.uk
>> http://www.bio.bris.ac.uk/research/mammal/tortoises.html
>
>
>
> Will Morris
> Masters of Philosophy candidate
> Vesk Plant Ecology Lab
> The School of Botany
> The University of Melbourne
> Australia
> Phone: +61 3 8344 0120
> http://www.botany.unimelb.edu.au/vesk/
----------------------
Christine Griffiths
PhD student
School of Biological Sciences
University of Bristol
Woodland Road
Bristol BS8 1UG
Tel: 0117 9287593
Fax 0117 3317985
Christine.Griffiths at bristol.ac.uk
http://www.bio.bris.ac.uk/research/mammal/tortoises.html
_______________________________________________
R-sig-mixed-models at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
Druk dit bericht a.u.b. niet onnodig af.
Please do not print this message unnecessarily.
Dit bericht en eventuele bijlagen geven enkel de visie van de schrijver weer
en binden het INBO onder geen enkel beding, zolang dit bericht niet bevestigd is
door een geldig ondertekend document. The views expressed in this message
and any annex are purely those of the writer and may not be regarded as stating
an official position of INBO, as long as the message is not confirmed by a duly
signed document.
More information about the R-sig-mixed-models
mailing list