rank.test {npsm}

General scores rank test for two sample problem

Description

A generalization of the Wilcoxon rank-sum test where a score function is applied to the ranks. Any scores from Rfit can be used as well as user defined. Default is to perform a Wilcoxon analysis.

Usage

rank.test(x, y, alternative = "two.sided", scores = Rfit::wscores, 
  conf.int = FALSE, conf.level = 0.95)

Arguments

x

m x 1 vector

y

n x 1 vector

alternative

one of 'two.sided', 'less', or 'greater'

scores

an object of class scores

conf.int

logical indicating if a confidence interval should be estimated

conf.level

desired level of confidence for interval

Details

Test is based on T = sum_i a(R(y_i)) where R is the rank based on the combined sample and a(t) = varphi(t/(N+1)). Confidence interval, if requested, is based on call to Rfit.

Value

statistic

Standardized value of test statistics

Sphi

Test statistic

p.value

p-value

conf.int

confidence interval for shift in location

estimate

point estimate for shift in location

Author(s)

John Kloke, Joseph McKean

References

Hettmansperger, T.P. and McKean J.W. (2011), Robust Nonparametric Statistical Methods, 2nd ed., New York: Chapman-Hall.

See Also

wilcox.test

Examples

rank.test(rt(20,1),rt(22,1)+0.2)