DTK.test {DTK}R Documentation

Dunnett's Modified Tukey-Kramer Pairwise Multiple Comparison Test

Description

Conducts a pairwise multiple comparison test (using the C procedure) for mean differences with unequal sample sizes and no assumption of equal population variances.

Usage

DTK.test(x = "data vector", f = "factor vector", a = "alpha level")

Arguments

x

Numeric data vector.

f

Factored level vector.

a

Alpha, significance level. DEFAULT=0.05

Details

Input data as vectors.

Value

[[1]]

"a" or the alpha significance level

[[2]]

Matrix containing the pair-waise comparisons as row names and the pairwise mean differences and lower and upper confidence interval values in columns, respectively

Note

In the case of equal sample sizes and equal population variances, Dunnett's test (the T3 Procedure) produces slightly wider (i.e. more conservative) confidence intervals than the Tukey-Kramer procedure. This is because of differences in the degrees of freedom used for determining the Studentized Range values. In cases where variances are unequal, however, the Tukey-Kramer test, which uses the pooled variance, will spread variance across levels and produce misleading results.

Author(s)

Matthew K. Lau, Department of Biological Sciences, Northern Arizona University, AZ

References

Dunnett,C.W. (1980) Pairwise Multiple Comparisons in the Unequal Variance Case. Journal of the American Statistical Association. 75 (372): 796-800.

See Also

DTK.plot, gl.unequal, TK.test, TukeyHSD, qtukey

Examples

x=c(rnorm(25,5,2),rnorm(30,5,5),rnorm(35,15,5))
f<-gl.unequal(n=3,k=c(25,30,35))
DTK.result<-DTK.test(x=x,f=f,a=0.05)
DTK.result
DTK.plot(DTK.result)

[Package DTK version 3.5 Index]