exact2x2 (version 1.6.4.1)

exact2x2-internal: Internal functions for exact2x2. Not to be called by user.

Description

The function exact2x2Pvals can calculate p-values for a vector of odds ratios. The function exact2x2CI is the code that calculates the confidence intervals for the two-sided Fisher's exact test and Blaker's exact test. The functions binomMeldCalcInt and binomMeldCalcMC are called by binomMeld.test.

Usage

fisherCalcMidp(x,or,alternative,conf.int,conf.level)
exact2x2Pvals(x, or, relErr=1+10^(-7),tsmethod = "minlike", alternative="two.sided")
exact2x2CI(x, tsmethod="minlike", conf.level=0.95, tol=0.00001, orRange=c(10^-10,10^10))
mcnemar.exact.calc(bb,cc,or,alternative,tsmethod="central",conf.level=.95, midp=FALSE)
binomMeldCalcInt(x1,n1,x2,n2,nullparm=NULL, 
           parmtype=c("difference","oddsratio","ratio"),
           conf.level=0.95, conf.int=TRUE,
           alternative=c("two.sided","less","greater"), midp=FALSE, eps=10^-8)
binomMeldCalcMC(x1,n1,x2,n2,nullparm=NULL, 
                           parmtype=c("difference","oddsratio","ratio"),
                           conf.level=0.95, conf.int=TRUE,
                           alternative=c("two.sided","less","greater"),
                           midp=FALSE,nmc=10^6)

Arguments

x

matrix representing 2 by 2 table

or

odds ratio, may be a vector

relErr

relative error. This is used to handle true ties on the computer. (see details).

tsmethod

either "minlike","blaker", "central"

conf.int

logical, calculate CI?

conf.level

confidence level

tol

tolerance

orRange

range for search for odds ratio confidence interval

alternative

indicates the alternative hypothesis and must be one of "two.sided", "greater" or "less".

midp

logical. Do midp adjustment?

Value

Output from exact2x2Pvals is a LIST, with

or

vector of odds ratios

pvals

vector of two-sided p-values

Output from exact2x2CI is a confidence interval with attributes: conf.level and conf.int.prec (a list of the bounds on the precision of the limits).

Details

P-values for both the two-sided Fisher's exact and Blaker's exact test add probabilities from the opposite tail if either the cumulative probabilities (for Blaker's test) or the probabilities (Fisher's test) are less than or equal to those of the observed tail. Since sometimes the p-value at odds ratio=1 is important, we may have problems if the opposite tail is some very small different probability due to computer rounding, when mathematically the probabilities are exactly the same and should be included. To get around this problem fisher.test uses relErr so that it chooses all d<= d[i]*relErr and if mathematically d[i] is equal to another value in d but there is a slightly computer rounding error, that value will be included. We use the same tactic here.

The function mcnemar.exact.calc is just a simple call to binom.test with p=.5.

See Also

exact2x2