wald.test: Makes wald test

In timereg: Flexible Regression Models for Survival Data

Makes wald test

Description

Makes wald test, either by contrast matrix or testing components to 0. Can also specify the regression coefficients and the variance matrix. Also makes confidence intervals of the defined contrasts. Reads coefficientes and variances from timereg and coxph objects.

Usage

wald.test(
  object = NULL,
  coef = NULL,
  Sigma = NULL,
  vcov = NULL,
  contrast,
  coef.null = NULL,
  null = NULL,
  print.coef = TRUE,
  alpha = 0.05
)

Arguments

object	timereg object
coef	estimates from some model
Sigma	variance of estimates
vcov	same as Sigma but more standard in other functions
contrast	contrast matrix for testing
coef.null	which indeces to test to 0
null	mean of null, 0 by default
print.coef	print the coefficients of the linear combinations.
alpha	significance level for CI for linear combinations of coefficients.

Author(s)

Thomas Scheike

Examples

data(sTRACE)
# Fits Cox model 
out<-cox.aalen(Surv(time,status==9)~prop(age)+prop(sex)+
prop(vf)+prop(chf)+prop(diabetes),data=sTRACE,n.sim=0)

wald.test(out,coef.null=c(1,2,3))
### test age=sex   vf=chf
wald.test(out,contrast=rbind(c(1,-1,0,0,0),c(0,0,1,-1,0)))

### now same with direct specifation of estimates and variance
wald.test(coef=out$gamma,Sigma=out$var.gamma,coef.null=c(1,2,3))
wald.test(coef=out$gamma,Sigma=out$robvar.gamma,coef.null=c(1,2,3))
### test age=sex   vf=chf
wald.test(coef=out$gamma,Sigma=out$var.gamma,
	  contrast=rbind(c(1,-1,0,0,0),c(0,0,1,-1,0)))

timereg documentation built on Sept. 11, 2024, 8:35 p.m.