[R] binomial vs quasibinomial

ilai keren at math.montana.edu
Tue Feb 7 22:28:01 CET 2012


Not really an R question, now is it ? more like pure stats. I'm
guessing you didn't get an answer because this list can't tell you how
to analyze your data (or in your case, approve an incorrect analysis).

Regarding the part of your question that is R related, I think you may
be confused on what the dispersion parameter is. In summary.glm it is
reported in the line above the null deviance: "(Dispersion parameter
for quasibinomial family taken to be ...)". You said "the dispersion
factor is the same for the quasibinomial...". I doubt that for any
real data it gets estimated at 1 (which was assumed for the binomial
family model). Also, since you say your p-values are higher on the
second model, that leads me to believe it is indeed taken to be
greater than 1. Check the two models again.

Regards



On Tue, Feb 7, 2012 at 4:11 AM, Jhope <jeanwaijang at gmail.com> wrote:
> After looking at 48 glm binomial models I decided to try the quasibinomial
> with the top model 25 (lowest AIC). To try to account for overdispersion
> (residual deviance 2679.7/68 d.f.)  After doing so the dispersion factor is
> the same for the quasibinomial and less sectors of the beach were
> significant by p-value. While the p-values in the binomial were more
> significant for each section of the beach. -- telling me more about the
> beach.
>
> Is this ok? Can I just look at the binomial glm model 25 and look at its
> p-values for beach sections and forget about the quasibinomial model 25?
>
> J
>
>
> Call:
> glm(formula = cbind(Shells, TotalEggs - Shells) ~ Sector:Veg:Aeventexhumed,
>    family = quasibinomial, data = data.to.analyze)
>
>
> --
> View this message in context: http://r.789695.n4.nabble.com/binomial-vs-quasibinomial-tp4364371p4364371.html
> Sent from the R help mailing list archive at Nabble.com.
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.



More information about the R-help mailing list