Standard error of proportion data

[tickbite]

Hi!

I am trying to compute the standard error for binomial proportion data, using GLM. To get the mean, I used:
> tapply(predict(model,type="response"),species,mean)
But I couldn't figure out how to get a proper standard error for this data. Can anyone help?

(My model contains diseased and not diseased individuals from samples of different sizes.)

Thanks!

[Masteras]

predict(model,type="response") just that. i do not recall if it is better to write type="response" or the other thing.