[R] glm.fit() and binomial family

[vito muggeo]

Hi all,
I'm interested in updating glm using glm.fit() (of course my final output is
different, but the problem is in glm.fit()). Then my function is, say:

fn<-function(obj,z){
    X<-update(obj,x=T)$x
    X<-cbind(X,z)
    y<-obj$y
    fam<-family(obj)
    o<-obj$offset
    contr<-obj$control
    w<-obj$weights
    o<-glm.fit(x=X,y=y,offset=o,weights=w,family=fam,control=contr)
    o$coef}

Such function works with gaussian family, but I have some problems with
binomial one. To test it let use the Down-syndrome data in the boot library:

library(boot)
data(downs.bc)

o<-glm(r/m~age ,family=binomial,weight=m,data=downs.bc)

> fn(o,z=I(downs.bc$age^2))
 (Intercept)          age            z
-4.940844023 -0.250360022  0.006250743
Warning message:
non-integer #successes in a binomial glm! in: eval(expr, envir, enclos)

but

> update(o,.~.+I(downs.bc$age^2))$coef
      (Intercept)               age I(downs.bc$age^2)
     -2.243578933      -0.414359607       0.008624326

Of course there is some problems, probably I'm missing something in
extracting the response, but I'm stuck and not able to solve my problem.
I would like use my function with *any* glm() (with any
family,link,offset,....).
Where have I to modify my code in order to use glm.fit()?


Also I would like to use "fn(o,z=I(age^2))" without specifying the dataframe
(i.e. "fn(o,z=I(downs.bc$age^2))"). That is fn() should search its argument
z in the environment of obj and then in the global environment. Where is the
trick?
Many thanks for your help!

best,
vito

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._