p_wilcox.test: p-value from Wilcox test simulation

In Spower: Power Analyses using Monte Carlo Simulations

p-value from Wilcox test simulation

Description

Simulates data given one or two parent distributions and returns a p-value. Can also be used for power analyses related to sign tests.

Usage

p_wilcox.test(
  n,
  d,
  n2_n1 = 1,
  mu = 0,
  type = c("two.sample", "one.sample", "paired"),
  exact = NULL,
  correct = TRUE,
  two.tailed = TRUE,
  parent1 = function(n, d) rnorm(n, d, 1),
  parent2 = function(n, d) rnorm(n, 0, 1)
)

Arguments

n	sample size per group
d	effect size passed to parent functions
n2_n1	sample size ratio
mu	parameter used to form the null hypothesis
type	type of analysis to use (two-sample, one-sample, or paired)
exact	a logical indicating whether an exact p-value should be computed
correct	a logical indicating whether to apply continuity correction in the normal approximation for the p-value
two.tailed	logical; use two-tailed test?
parent1	data generation function for first group. Ideally should have SDs = 1 so that d reflects a standardized difference
parent2	same as parent1, but for the second group

Value

a single p-value

Author(s)

Phil Chalmers [e-mail]

Examples

# with normal distributions defaults d is standardized
p_wilcox.test(100, .5)
p_wilcox.test(100, .5, type = 'paired')
p_wilcox.test(100, .5, type = 'one.sample')

# using chi-squared distributions (standardizing to 0-1)
p_wilcox.test(100, .5, type = 'one.sample',
   parent1 = function(n, d) rchisq(n, df=10) - 10 + d)
p_wilcox.test(100, .5,
   parent1 = function(n, d) (rchisq(n, df=10) - 10)/sqrt(20) + d,
   parent2 = function(n, d) (rchisq(n, df=10) - 10)/sqrt(20))

Spower documentation built on April 4, 2025, 5:11 a.m.