[R-sig-ME] Mixed model on few individuals

Thierry Onkelinx thierry.onkelinx at inbo.be
Tue Mar 6 10:20:08 CET 2018


Dear Simon,

Interesting problem. 3 individuals is too small for a random effect of
individual see
http://bbolker.github.io/mixedmodels-misc/glmmFAQ.html#should-i-treat-factor-xxx-as-fixed-or-random

However you still have 51 trials. So that would be relevant to include as a
random effect. The azimuth and elevation are your response variables? In
that case you rather have temporal autocorrelation. The nlme package has
several models for temporal autocorrelation among the residuals. The INLA
package provides models where the random effects have temporal
autocorrelation.

Best regards,


ir. Thierry Onkelinx
Statisticus / Statistician

Vlaamse Overheid / Government of Flanders
INSTITUUT VOOR NATUUR- EN BOSONDERZOEK / RESEARCH INSTITUTE FOR NATURE AND
FOREST
Team Biometrie & Kwaliteitszorg / Team Biometrics & Quality Assurance
thierry.onkelinx at inbo.be
Havenlaan 88 bus 73, 1000 Brussel
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
///////////////////////////////////////////////////////////////////////////////////////////

<https://www.inbo.be>

2018-02-26 9:28 GMT+01:00 Simon POTIER <sim.potier at gmail.com>:

> Dear all,
>
> First, I will try to be the most comprehensible, if not, do not hesitate to
> ask for more information.
>
> I am working on a project with collaborators, and I am trying to analyse
> the data. For one question, however, I do not feel confident with my
> method, but I do not know how I can analyse the data in a different way.
>
> We are working on visual fixation of a prey while hunting, with a new
> method of high accuracy. This method is very interesting but implies that
> we cannot use many individuals (because of different things that are not of
> interest here I think).
>
> My major problem here is that I have "only" 3 individuals and 51 trials
> (almost equilibrate per individual).
> I also have 3 different conditions (i.e. 3 types of prey). All these
> conditions have been repeated for the 3 individuals.
> In total, the dataset is relatively large as I have more than 6000 points
> (large for behavioural study), but, as I wrote, with only 3 individuals.
>
> Regarding this type of dataset, I am wondering whether I can use mixed
> model to compare the visual fixation according to the different conditions.
> So here are my two questions :
>
> 1) Can I use mixed models with only 3 individuals? If not, do you know if I
> can use another model? Or should I compare each individual independently?
>
> 2) My data are spatio-temporal (visual fixation in times: One Azimuth and
> one elevation every X sec). How can I implement this autocorrelation in my
> model?
>
> I am sincerely grateful if someone can be of any help. Sorry If similar
> question has been posted before, I did not find it.
>
> Best regards,
>
> Simon Potier
>
>         [[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-mixed-models at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>

	[[alternative HTML version deleted]]



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