Hello,
I'm working with Splus 4.5 on Windows NT. I have some data with polynomial
dependencies. As I've understood, the use of poly() has statistical
advantages in the modelling of polynomial dependencies.
In the Splus-books I have (for version 3.3, 1995), it is shown, that for a
model
lmfit1_lm(y~poly(x1,3))
the coefficients corresponding to the untransformed values of the
polynomials of x can be obtained through
poly.transform(poly(x1,3),lmfit1)
This works fine.
But I have more than the covariable x1 in my model and if I try
lmfit2_lm(y~poly(x1,3)+x2+x3)
I get no result with the following function
poly.transform(poly(x1,3),lmfit2)
just an error message
Error in poly.transform(........): wrong length for coefs
I understand the reason for the error message, but how do I have to use
poly.transform() to get what I want? Or is there an other solution to get
the estimates of the coefficients for the not transformed polynomials? At
last, I have the same problem in lme-models and would be interested if the
solution in lme is the same.
I've searched the archives (but maybe not enough...). I'll report the
result to the list.
Thanks
Daniel
***************************************************
Daniel Erdin
Swiss Federal Institute of Technology
Institute of Animal Science
Animal Breeding
Research Station Chamau
CH-6331 Huenenberg
Phone: +41 41 780 10 47
Fax: +41 41 780 43 83
e-mail: erdin@inw.agrl.ethz.ch
homepage: http://www.tz.inw.agrl.ethz.ch/~erdin/
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news
|