[R] estimating p values bootstrap regression

Roberto Edmundo Munguia Steyer munguiar at ecologia.edu.mx
Thu May 26 06:36:06 CEST 2005


Dear R-users:

I have a basic question.

I am trying to estimate the gradients of sexual selection on the males of a 
waterbug species.I performed a bootstrap analysis on a multiple regression 
using library boot and obtained the estimated coefficents and their respective 
standard errors. 

eri<-read.table("D:/efi/er2.txt",header=T)
grad1<-eri
grad1$fit<-fitted(gradi)
grad1$res<-resid(gradi)
grad.fit<-function(data){
mod<-lm(data$huevrel~data$slongitud*data$sabdomen*data$saparcap, 
weights=data$presencia, data=eri)
c(coef(mod),
summary(mod)$coef[,2]^2)}

gmodfun<-function(d,i){
d$huevrel<-d$fit+d$res[i]
grad.fit(d)}
grandi<-boot(grad1,gmodfun, R=999)
grandi


I was wondering if there is a way to estimate p-values in order to know wich 
independent variables are significant. Would be correct to add the bias to the 
regression coefficients and divide with de SE to obtain a t value and use pt 
function eg (1 - pt(abs(1.51),211))*2 in order to get the p values? 

Thanks a lot for your help.

Roberto
I use R 2.1, Windows XP.
 

Departamento de Biologia Evolutiva
Instituto de Ecología, A.C.
Km 2.5 Carretera antigua a Coatepec
Ap. Postal 63 (excepto mensajería)
Xalapa, Veracruz 91000
MEXICO
Tel. (52)(228)8421800 ext.3009


 


---------------------------------------------
Instituto de Ecología, A. C.
http://www.ecologia.edu.mx




More information about the R-help mailing list