Pick {cherry} | R Documentation |
Calculates confidence limits for the number of true hypotheses on the basis of the closed testing procedure.
pick (closure, reject, alpha, silent=FALSE, plot=FALSE)
closure |
An object of class |
reject |
A character vector identifying the hypotheses to be rejected. Must be a subvector of |
alpha |
For closure objects with adjusted p-values, specifies the value of alpha for which confidence limits are to be calculated (optional). |
silent |
If |
plot |
Whether a a confidence distribution should be plotted. Only available for closure objects with adjusted p-values. |
The function pick
calculates a confidence interval for the number of true hypotheses among a selected set of hypotheses.
The function returns the upper confidence limit for the number of true hypotheses among the set reject
. The lower confidence limit is always equal to 0. If closed
was called with alpha=NA
, a confidence distribution is plotted and returned.
Jelle Goeman: j.j.goeman@lumc.nl
# Example: the birthwt data set from the MASS library
# We want to find variables associated with low birth weight
library(MASS)
fullfit <- glm(low~age+lwt+race+smoke+ptl+ht+ui+ftv, family=binomial, data=birthwt)
hypotheses <- c("age", "lwt", "race", "smoke", "ptl", "ht", "ui", "ftv")
# Define the local test to be used in the closed testing procedure
mytest <- function(hyps) {
others <- setdiff(hypotheses, hyps)
form <- formula(paste(c("low~", paste(c("1", others), collapse="+"))))
anov <- anova(glm(form, data=birthwt, family=binomial), fullfit, test="Chisq")
res <- anov$"Pr("[2] # for R >= 2.14.0
if (is.null(res)) res <- anov$"P("[2] # earlier versions
res
}
# perform the closed testing
cl <- closed(mytest, hypotheses)
summary(cl)
# how many variables among a chosen set are associated with the response?
pick(cl, c("ht", "lwt", "smoke", "ui"))