[R-sig-ME] Mixed model for compositional data

[Stéphanie Périquet]

Thanks for your reply Thierry.

We did try this as a first approach. The model was over dispersed and 
zero-inflated so is hard to fit but we are getting there using R2OpenBUGS.

Best,
Stéphanie

On 08/11/16 10:31, Thierry Onkelinx wrote:
> Dear Stéphanie,
>
> A workaround would be to model the numbers per item and use the (log 
> of) the total number per night as an offset. Then the model parameters 
> are essentially ratio's.
>
> no effects of season and moonlite: number ~ offset(log(total)) + item 
> + (1|animal)
> only season: number ~ offset(log(total)) + item*season + (1|animal)
> season:moonlight: number ~ offset(log(total)) + item*season*moonlight 
> + (1|animal)
>
> The model fit will not be guaranteed to sum to 1 per night and animal.
>
> Best regards,
>
> Thierry
>
> ir. Thierry Onkelinx
> Instituut voor natuur- en bosonderzoek / Research Institute for Nature 
> and Forest
> team Biometrie & Kwaliteitszorg / team Biometrics & Quality Assurance
> Kliniekstraat 25
> 1070 Anderlecht
> Belgium
>
> 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
>
> 2016-11-08 8:40 GMT+01:00 Stéphanie Périquet 
> <stephanie.periquet at gmail.com <mailto:stephanie.periquet at gmail.com>>:
>
>     Hi all,
>
>     I have repeated measures of individual diet composition (I
>     actually have
>     the number of each item eaten and the duration of each
>     observation) and
>     I would like to model the effect of variables such as season and
>     moonlight on diet composition (not the actual number of item eaten).
>     I gathered that I should use compositional but the packed
>     'compostions'
>     doesn't allow the inclusion of random factors.
>
>     Does anyone has experience whit such data and analyses and point me in
>     the right direction?
>
>     Thanks a lot in advance for your replies!
>
>     Regards,
>     Stéphanie
>     --
>     *Stéphanie PERIQUET (PhD) * - Bat-Eared Fox Research Project
>     /Dept of Zoology & Entomology/
>     /University of the Free State, Qwaqwa Campus/
>     /Private Bag X13/
>     /9866 Phuthaditjhaba, Free State/
>     /South Africa/
>     *Cell: +27 79 570 2683 <tel:%2B27%2079%20570%202683>*
>     ResearchGate
>     <https://www.researchgate.net/profile/Stephanie_Periquet
>     <https://www.researchgate.net/profile/Stephanie_Periquet>> &
>     Twitter <https://twitter.com/ShumbAfrica
>     <https://twitter.com/ShumbAfrica>>
>
>     Kalahari bat-eared foxes on Twitter
>     <https://twitter.com/kal_batearedfox
>     <https://twitter.com/kal_batearedfox>>
>
>     //////
>
>             [[alternative HTML version deleted]]
>
>     _______________________________________________
>     R-sig-mixed-models at r-project.org
>     <mailto:R-sig-mixed-models at r-project.org> mailing list
>     https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>     <https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models>
>
>


-- 
*Stéphanie PERIQUET (PhD) * - Bat-Eared Fox Research Project
/Dept of Zoology & Entomology/
/University of the Free State, Qwaqwa Campus/
/Private Bag X13/
/9866 Phuthaditjhaba, Free State/
/South Africa/
*Cell: +27 79 570 2683*
ResearchGate <https://www.researchgate.net/profile/Stephanie_Periquet> & 
Twitter <https://twitter.com/ShumbAfrica>








Kalahari bat-eared foxes on Twitter <https://twitter.com/kal_batearedfox>

//////