GEE.trend.test: GEE-based trend test

In CorrBin: Nonparametrics with Clustered Binary and Multinomial Data

GEE-based trend test

Description

GEE.trend.test implements a GEE based test for linear increasing trend for correlated binary data.

Usage

GEE.trend.test(cbdata, scale.method = c("fixed", "trend", "all"))

Arguments

cbdata	a CBData object
scale.method	character string specifying the assumption about the change in the overdispersion (scale) parameter across the treatment groups: "fixed" - constant scale parameter (default); "trend" - linear trend for the log of the scale parameter; "all" - separate scale parameter for each group.

Details

The actual work is performed by the geese function of the geepack library, which is required for this feature to work. This function only provides a convenient wrapper to obtain the results in the same format as RS.trend.test and SO.trend.test.

The implementation aims for testing for increasing trend, and a one-sided p-value is reported. The test statistic is asymptotically normally distributed, and a two-sided p-value can be easily computed if needed.

Value

A list with components

statistic	numeric, the value of the test statistic
p.val	numeric, asymptotic one-sided p-value of the test

Author(s)

Aniko Szabo, [e-mail]

See Also

RS.trend.test, SO.trend.test for alternative tests; CBData for constructing a CBData object.

Examples

data(shelltox)
if (require(geepack)){
 GEE.trend.test(shelltox, "trend")
}

CorrBin documentation built on Sept. 11, 2024, 8:25 p.m.