[R] GLM Contingency table regressions

Martin Maechler maechler at stat.math.ethz.ch
Mon Aug 12 10:33:23 CEST 2002

>>>>> "PaulMJ" == Paul M Jacobson <pmj at jciconsult.com>
>>>>>     on Sat, 10 Aug 2002 15:02:55 -0400 writes:

    PaulMJ> I have a contingency table with a 0/1 variable (fwr)
    PaulMJ> that flags a particular condition, a weight variable
    PaulMJ> (cnt) and other variables (e.g. zz) characterizing
    PaulMJ> that observation of the table.  I am trying to use
    PaulMJ> GLM. By converting the variables to factors and
    PaulMJ> running the following regression:

    PaulMJ> fwr1<-factor(fwr)
    PaulMJ> zz1<-factor(zz1)
    PaulMJ> res1<-glm(fwr1 ~ zz1,weights=cnt)

(spaces around "<-" would improve readibility}

    PaulMJ> Is that appropriate?

not quite.
Assuming that you really want to predict (a function of) the
probability   P[ fwr = 1 | zz1 ]
you need at least to add

	 family = binomial

to your glm() arguments.
The thing you did above was least-squares regression with a 0/1 variable
which is probably not what you wanted.

Martin Maechler <maechler at stat.math.ethz.ch>	http://stat.ethz.ch/~maechler/
Seminar fuer Statistik, ETH-Zentrum  LEO C16	Leonhardstr. 27
ETH (Federal Inst. Technology)	8092 Zurich	SWITZERLAND
phone: x-41-1-632-3408		fax: ...-1228			<><
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

More information about the R-help mailing list