[R-sig-Epi] random-effects modeling

[BXC (Bendix Carstensen)]

You would not use a Poisson distribution, you would use a binomial, since the number of positives is not in (0,\inf) but in (0,broilers).

So you would de a random-effects logistic regression.
I guess an attractive approach would be to use the MCMCglmm package that will do this sort of model using an MCMC approach.

Best regards,
Bendix Carstensen

> -----Original Message-----
> From: r-sig-epi-bounces at stat.math.ethz.ch 
> [mailto:r-sig-epi-bounces at stat.math.ethz.ch] On Behalf Of 
> dieter.anseeuw
> Sent: 17. august 2010 07:29
> To: r-sig-epi at stat.math.ethz.ch
> Subject: [R-sig-Epi] random-effects modeling
> 
> Hi all,
> 
> 
> 
> Being new to analysing epidemiologic data, I have a question 
> about the following data set:
> 
> A friend has inspected three randomly chosen farms (random 
> factor 'farm'). At each farm three randomly chosen series of 
> chickens (random factor 'flock') were each inspected for the 
> presence of a certain bacteria. The contaminated chickens 
> were counted (response variable 'positives'). The sample 
> sizes per flock are given by 'broilers'. We want to have a 
> look at within-broilers, within-farm and between-farm variability.
> 
> 
> 
> It seems to me that we have a random-effects model, in which 
> the factor 'flock' is nested within the factor 'farm'. Am I 
> correct so far?
> 
> Now, since the response variable yields count data, fitting 
> the model should be done using Poisson regression. Correct?
> 
> 
> 
> Could somebody help me out with (an example of) such an analysis?
> 
> 
> 
> Many thanks in advance,
> 
> Dieter
> 
> 
> 
> Here is the dataset used:
> 
> 
> 
> broilers<-data.frame(farm=c("FA","FA","FA","FB","FB","FB","FC"
,"FC","FC"), flock=c("a","b","c","d","e","f","g","h","i"), > broilers=c(50, rep(25,8)), positives=c(7,2,0,7,2,0,0,5,2))
> 
> 
> 
> 
> 
> --
> 
> Dieter Anseeuw
> 
> Katho Campus Roeselare
> 
> Wilgenstraat 32
> 
> 8800 Roeselare Belgium
> 
> 
> 
> Direct phone: +32 51 23 29 68
> 
> http://www.katho.be/hivb
> 
> http://www.linkedin.com/in/dieteranseeuw
> 
> 
> 
> 
> 	[[alternative HTML version deleted]]
> 
> _______________________________________________
> R-sig-Epi at stat.math.ethz.ch mailing list 
> https://stat.ethz.ch/mailman/listinfo/r-sig-epi
>