RankRegV {rankreg} | R Documentation |
Compute the variance-covariance matrix of the rank estimating function at beta for the censored AFT model. It also return the chi square value for testing H0: betanull.
RankRegV(y, d, x, beta, betanull = beta, type="Gehan")
y |
a vector containing the censored responses in the AFT model. |
d |
a vector of 1's and 0's. censoring indicator. 1(uncensor), 0(censored). Both y and d should be of length n. |
x |
the design matrix, of size n by q. Should not have a column of 1s. |
beta |
a vector of dimension q. Usually it should be the solution
of the estimation equation, from rankreg() . |
betanull |
the null value (under H0) of beta to be tested. |
type |
either equal to "Gehan" (default) or "Logrank". |
It returns the chi square value and P-value which are based on
EF(betanull)*Vtest(beta)^(-1)* EF(betanull)
.
The input beta
usually should be the solution
of the estimating function.
This solution can be obtained by using rankaft
.
But other values of beta may also needed sometime.
When betanull
equals beta
then the chi square
statistic should be zero (since EF(betanull)=0
)
and the Pvalue should be 1.
A list with VEF
, which is the variance estimator of the
estimating function at beta
;
chisqvalue
which
is the quadratic form EF(betanull)*VEF^(-1)*EF(betanull)
;
and Pval
that is the P-value from the chi square quantile.
Mai Zhou.
Kalbfleisch, J. and Prentice, R. (2002) {em The Statistical Analysis of Failure Time Data.} 2nd Ed. Wiley, New York. (Chapter 7)