R help - Oct 2010 - GLM and Weights

Dear all,

I am trying to use the 'glm' package as part of a semiparametric
technique that involves weighting a likelihood in various ways, i.e.

L(theta;data)=Sum_i=1,..,n (W_i)(log L(theta;data_i))

Where W_i can be a kernel weighting function, or W_i can be an indicator of
'non-missingness' divided by a propensity score.

In a Monte Carlo exercise, the option glm(..., weights=W_i,) works very well for
the Gaussian design, but if I were to change it to a Binomial design it
doesn't (I have read previous posts in this mailing list explaining why).

My question is: Is there a way or a package that uses the glm library and allows
me to set 'weights' the way I need?

Thanks in advance for your suggestions.

	[[alternative HTML version deleted]]

there is no package. 
I think the solution is modify the variance function for the binomial with
your weights as that should propagate properly.
-- 
View this message in context:
http://r.789695.n4.nabble.com/GLM-and-Weights-tp3015732p3015930.html
Sent from the R help mailing list archive at Nabble.com.