R help - Nov 2002 - Qualitative factors

Hi,
I have some doubt about how qualitative factors are coded in R. For
instance, I consider a response y, a quantitative factor x and a qualitative
factor m at 3 levels, generated as follow :

y_c(6,4,2.3,5,3.5,4,1.,8.5,4.3,5.6,2.3,4.1,2.5,8.4,7.4)
x_c(3,1,3,1,2,1,4,5,1,3,4,2,5,4,3)
m_gl(3,5)

lm(y~x+m)

Coefficients:
(Intercept)            x           m2           m3  
    3.96364      0.09818      0.44145      0.62291  

In literature, 2 usual implicit coding process are suggested : m1=0 or
m1=-m2-m3. Does R use one of these process ? (I've already read the R
documentation on this topic, and it is still not clear in my mind).
Furthermore, how can I make prediction using this model, I mean how should I
specify the new data (especially for m).

Thanks in advance for your help

Regards,

Isabelle Zabalza-Mezghani
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
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
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

There are several ways to model categorical variables, including
corner-point, sum-zero, helmert,....

In R you can select them using the options() function, namely for instance

options(contrasts = c(unordered = "contr.treatment", ordered
"contr.poly"))

or directly in the model
lm(y~x+C(m, treatment)) or lm(y~x+C(m, sum)) and so on.

The interpretation of the estimated coefficients (and their univariate Wald
statistic) depends on the used parameterization, but the LRT does not.

best,
vito


----- Original Message -----
From: "ZABALZA-MEZGHANI Isabelle" <Isabelle.zabalza-mezghani at
IFP.fr>
To: "help R (E-mail)" <r-help at stat.math.ethz.ch>
Cc: "JOURDAN Astrid" <Astrid.JOURDAN at IFP.fr>
Sent: Thursday, November 07, 2002 9:55 AM
Subject: [R] Qualitative factors

> Hi,
> I have some doubt about how qualitative factors are coded in R. For
> instance, I consider a response y, a quantitative factor x and a
qualitative
> factor m at 3 levels, generated as follow :
>
> y_c(6,4,2.3,5,3.5,4,1.,8.5,4.3,5.6,2.3,4.1,2.5,8.4,7.4)
> x_c(3,1,3,1,2,1,4,5,1,3,4,2,5,4,3)
> m_gl(3,5)
>
> lm(y~x+m)
>
> Coefficients:
> (Intercept)            x           m2           m3
>     3.96364      0.09818      0.44145      0.62291
>
> In literature, 2 usual implicit coding process are suggested : m1=0 or
> m1=-m2-m3. Does R use one of these process ? (I've already read the R
> documentation on this topic, and it is still not clear in my mind).
> Furthermore, how can I make prediction using this model, I mean how should
I
> specify the new data (especially for m).
>
> Thanks in advance for your help
>
> Regards,
>
> Isabelle Zabalza-Mezghani
> -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.
-.-.-
> 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
>
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._.
_._
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
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
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

Dear Isabelle,

Contrast coding in R for factors and ordered factors is governed by the 
contrasts option:

     > options("contrasts")
     $contrasts
             unordered           ordered
     "contr.treatment"      "contr.poly"

The default for an unordered factor is "treatment" (0/1) coding, using
the
first level as the baseline:

     > contrasts(m)
       2 3
     1 0 0
     2 1 0
     3 0 1

     > options("contrasts")
     $contrasts
             unordered           ordered
     "contr.treatment"      "contr.poly"

See ?contr.treatment for details and alternatives.

I hope that this helps,
  John

At 09:55 AM 11/7/2002 +0100, you wrote:
>Hi,
>I have some doubt about how qualitative factors are coded in R. For
>instance, I consider a response y, a quantitative factor x and a qualitative
>factor m at 3 levels, generated as follow :
>
>y_c(6,4,2.3,5,3.5,4,1.,8.5,4.3,5.6,2.3,4.1,2.5,8.4,7.4)
>x_c(3,1,3,1,2,1,4,5,1,3,4,2,5,4,3)
>m_gl(3,5)
>
>lm(y~x+m)
>
>Coefficients:
>(Intercept)            x           m2           m3
>     3.96364      0.09818      0.44145      0.62291
>
>In literature, 2 usual implicit coding process are suggested : m1=0 or
>m1=-m2-m3. Does R use one of these process ? (I've already read the R
>documentation on this topic, and it is still not clear in my mind).
>Furthermore, how can I make prediction using this model, I mean how should I
>specify the new data (especially for m).
-----------------------------------------------------
John Fox
Department of Sociology
McMaster University
Hamilton, Ontario, Canada L8S 4M4
email: jfox at mcmaster.ca
phone: 905-525-9140x23604
web: www.socsci.mcmaster.ca/jfox
-----------------------------------------------------

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