\name{combine.p}
\alias{combine.p}
\title{Combine probabilities from tests of significance}
\usage{combine.p(x, y = NULL, method = c("fisher", "winer", "stouffer"))}
\arguments{
\item{x}{a vector of p.values}
\item{y}{a vector of degrees of freedom for Winer method}
\item{method}{a character string indicating which method to use.
One of "fisher", "winer", or "stouffer". No abbreviations
accepted, defaults to "fisher".}
}
\description{
\code{combine.p} combines probabilities from multiple significance
tests using one of Fisher, Winer or Stouffer methods.
}
\details{
Uses one of three methods to combine probabilities from multiple
significance tests.
if \code{method} is \code{"fisher"}, the probabilities given in
\code{x} are tested against the null hypothesis that they are chi
square distributed with 2k degrees of freedom, where k is the number
of tests.
if \code{method} is \code{"winer"}, the probabilities given in
\code{x} are converted to t-scores, summed and divided by the square
root of the degrees of freedom for each t after each df has been
divided by df-2. See Wolf (1986, pp 20) for rationale.
if \code{method} is \code{"stouffer"}, the probabilities given in
\code{x} are converted to z-scores, summed and divided by the square
root of the number of tests. See Wolf (1986, pp 20) for rationale.
}
\value{
A list with class \code{"htest"} containing the following
components:
\item{statistic}{the value the chi square test statistic.}
\item{parameter}{the degrees of freedom.}
\item{p.value}{the p-value for the test.}
\item{method}{a character string indicating the method.}
}
\references{
Ronald A. Fisher (1958),
\emph{Statistical Methods for Research Workers}, 13th ed.
New York: Hafner Publishing.
Section 21.1
Pages 99--101.
Robert R. Sokal & F. James Rohlf (1995),
\emph{Biometry}, 3rd ed.
New York: W. H. Freeman & Company.
Section 18.1.
Pages 794--797.
Fredric M. Wolf (1986),
\emph{Meta-analysis: quantitative methods for research synthesis}
Quantitative Applications in the Social Sciences, vol 59
Newbury Park: Sage Publications
Pages 18--23.
}
\examples{
## Example from Fisher (1958) p 100.
> combine.p(c(.145,.263,.087))
Fisher's combined probability
data:
Chi Square = 11.4169, X-squared df = 6, p-value = 0.07631
## Example from Sokal & Rohlf p 795)
> combine.p(c(.073,.086,.10,.08,.06))
Fisher's combined probability
data:
Chi Square = 25.4249, X-squared df = 10, p-value = 0.004596
## Example from Wolf (1986) p 22
> combine.p(c(.004,.97,.024,.06),method="stouffer")
Stouffer's combined probability
data:
Z score = 2.1517, No df = NA, p-value = 0.01571
## Example from Wolf (1986) p 22
> combine.p(c(.004,.97,.024,.06),c(80,60,200,20),method="winer")
Winer's combined probability
data:
Z score = 2.1601, No df = NA, p-value = 0.01538
>
}
\keyword{htest}