select: Confidence limits for the number of true hypotheses.
In cherry: Multiple Testing Methods for Exploratory Research

Description Usage Arguments Details Value Author(s) Examples

Calculates confidence limits for the number of true hypotheses on the basis of the closed testing procedure.

1	pick (closure, reject, alpha, silent=FALSE, plot=FALSE)

`closure`	An object of class `closure`, typically created through a call to `closed` .
`reject`	A character vector identifying the hypotheses to be rejected. Must be a subvector of `hypotheses(closure)`.
`alpha`	For closure objects with adjusted p-values, specifies the value of alpha for which confidence limits are to be calculated (optional).
`silent`	If `FALSE`, prints the result to the screen.
`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"))