[R] using identity link for binomial familly with glm

[Prof Brian Ripley]

On Tue, 31 Jul 2001, Stephane Dupas wrote:

> -- Error in binomial(link = "identity") : identity link not available for binomial family, available links are "logit",  "probbbit", "cloglog" and "log"
>
> Hi, I have a question, dealing with this error response. I'm trying to
> make anova on percentages. The variablethat has a biological
> significance is actually the percentage itself. Is it really not
> possible, to test a simple linear probability model with glm, using
> binomial model and identity link function ? I know it's hard to deal
> with probability responses out of [0,1] but well... help me !
> There is an example of such a linear probability model in Agresti (1997)
> some software aparently do it

You can do it easily: use a quasi model.

quasi(link = "identity", variance = "mu(1-mu)")

I believe.  You will probably need to specify a good starting value.

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272860 (secr)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
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
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._