mantelHaenszel: Cochran-Mantel-Haenszel Chi-Squared Test for Count Data

In felix-hof/biostatUZH: Misc Tools of the Department of Biostatistics, EBPI, University of Zurich

Cochran-Mantel-Haenszel Chi-Squared Test for Count Data

Description

Compute Cochran-Mantel-Haenszel chi-squared test of the null that two nominal variables are conditionally independent in each stratum, assuming that there is no three-way interaction.

Usage

mantelHaenszel(exposure, outcome, stratum)

Arguments

exposure	Binary variable coding whether patient was exposed (1) or not (0).
outcome	Binary variable coding outcome of patient.
stratum	Factor object with at least 2 levels identifying to which stratum the corresponding elements in exposure and outcome belong.

Value

tab	Table that counts numbers of strata for each case-control combination.
test.stat	Test statistic for chi^2 test.
p.val	p-value of chi^2 test.

Author(s)

Kaspar Rufibach
[e-mail]

References

Agresti, A. (2002). Categorical data analysis. Wiley, New York,

See Also

Similar functionality is provided in mantelhaen.test and clogit. See the examples below for a comparison.

Examples

# generate data
set.seed(1977)
data <- data.frame(exposure = rep(c(1, 0, 0, 0, 0), 41),
                   outcome = sample(x = c(rep(1, 62), rep(0, 5 * 41 - 62))),
                   stratum = rep(1:41, each = 5))

# via conditional logistic regression
logreg <- clogit(outcome ~ exposure + strata(stratum), method = "approximate",
                 data = data)
summary(logreg)

# R function in library 'stats'
mh <- with(data, mantelhaen.test(x = outcome, y = exposure, z = stratum))

# this function
mH <- with(data, mantelHaenszel(exposure = exposure,
                                outcome = outcome,
                                stratum = stratum))

# compare p-values
summary(logreg)$coef[5]
mh$p.value
mH$p.val

felix-hof/biostatUZH documentation built on Sept. 27, 2024, 1:48 p.m.