Accessing a slot in the results of wilcox_test - r

Question

I'm doing a test on some data:
wilcox_test(y ~ x, distribution="exact", conf.int=TRUE)
# Exact Wilcoxon Mann-Whitney Rank Sum Test
#
# data: y by x (1, 2)
# Z = 1.8732, p-value = 0.06106
# alternative hypothesis: true mu is not equal to 0
# 95 percent confidence interval:
# 0 2
# sample estimates:
# difference in location
# 1
The number displayed under difference in location is the Hodges-Lehmann estimator. I want to assign it to a variable for further calculation. But I do not know:
How can I access the value of the "difference in location"?
I looked at str(wilcox_test(y ~ x, distribution="exact", conf.int=TRUE)), which gives me:
Formal class 'ScalarIndependenceTestConfint' [package "coin"] with 7 slots
..# confint :function (level)
..# conf.level : num 0.95
..# nullvalue : num 0
..# distribution:Formal class 'ExactNullDistribution' [package "coin"] with 7 slots
.. .. ..# q :function (p)
.. .. ..# d :function (x)
.. .. ..# support :function ()
.. .. ..# parameters: list()
.. .. ..# pvalue :function (q)
.. .. ..# p :function (q)
.. .. ..# name : chr "exact distribution (via Streitberg-Roehmel algorithm)"
..# statistic :Formal class 'ScalarIndependenceTestStatistic' [package "coin"] with 14 slots
.. .. ..# alternative : chr "two.sided"
.. .. ..# teststatistic : Named num 1.87
.. .. .. ..- attr(*, "names")= chr "1"
.. .. ..# standardizedlinearstatistic: Named num 1.87
.. .. .. ..- attr(*, "names")= chr "1"
.. .. ..# linearstatistic : num 4246
.. .. ..# expectation : Named num 3875
.. .. .. ..- attr(*, "names")= chr "1"
.. .. ..# covariance :Formal class 'Variance' [package "coin"] with 1 slot
.. .. .. .. ..# variance: Named num 39334
.. .. .. .. .. ..- attr(*, "names")= chr "1"
.. .. ..# xtrans : num [1:124, 1] 1 1 1 1 1 1 1 1 1 1 ...
.. .. .. ..- attr(*, "dimnames")=List of 2
.. .. .. .. ..$ : chr [1:124] "1" "2" "3" "4" ...
.. .. .. .. ..$ : chr "1"
.. .. .. ..- attr(*, "assign")= int 1
.. .. ..# ytrans : num [1:124, 1] 114.5 93.5 9 50 114.5 ...
.. .. .. ..- attr(*, "assign")= int 1
.. .. .. ..- attr(*, "dimnames")=List of 2
.. .. .. .. ..$ : NULL
.. .. .. .. ..$ : chr ""
.. .. ..# xtrafo :function (data, numeric_trafo = id_trafo, factor_trafo = f_trafo, ordered_trafo = of_trafo,
surv_trafo = logrank_trafo, var_trafo = NULL, block = NULL)
.. .. ..# ytrafo :function (data)
.. .. ..# x :'data.frame': 124 obs. of 1 variable:
.. .. .. ..$ x: Factor w/ 2 levels "1","2": 1 1 1 1 1 1 1 1 1 1 ...
.. .. ..# y :'data.frame': 124 obs. of 1 variable:
.. .. .. ..$ y: num [1:124] 7 5 0 2 7 1 0 9 1 7 ...
.. .. ..# block : Factor w/ 1 level "0": 1 1 1 1 1 1 1 1 1 1 ...
.. .. ..# weights : num [1:124] 1 1 1 1 1 1 1 1 1 1 ...
..# estimates : list()
..# method : chr "Wilcoxon Mann-Whitney Rank Sum Test"
But wilcox_test(...)#estimates only returns:
list()
which appears empty.
So where is the 1 stored that is listed under difference in location in the results???
Additional explanation:
This works:
hle <- wilcox.test(x, y, paired = TRUE, correct = FALSE, conf.int = TRUE)$estimate
But that is a different test and not appropriate to my data.

Answer 1

I had to do a little bit of digging to figure this out since the difference in location value is nested down a couple of levels, but I'll walk you through it using an example object provided in the help file for wilcox_test - the object is named wt, and I'll include it below. For things like this, I usually start by inspecting the show method of the class of interest -
R> class(wt)
[1] "ScalarIndependenceTestConfint"
attr(,"package")
[1] "coin"
R> getMethod("show","ScalarIndependenceTestConfint")
Method Definition:
function (object)
{
x <- object
stat <- x#statistic#teststatistic
names(stat) <- "Z"
dist <- x#distribution
cld <- class(dist)
attributes(cld) <- NULL
distname <- switch(cld, AsymptNullDistribution = "Asymptotic",
ApproxNullDistribution = "Approximative", ExactNullDistribution = "Exact")
dataname <- varnames(x#statistic)
ci <- confint(object, level = object#conf.level)
RET <- list(statistic = stat, p.value = x#distribution#pvalue(stat),
alternative = x#statistic#alternative, method = paste(distname,
x#method), data.name = dataname, conf.int = ci$conf.int,
estimate = ci$estimate)
if (length(x#nullvalue))
RET$null.value = c(mu = x#nullvalue)
if (length(x#estimates))
RET <- c(RET, x#estimates)
class(RET) <- "htest"
print(RET)
invisible(RET)
}
<environment: namespace:coin>
The line of interest is the last item in the RET list - estimate = ci$estimate, which was calculated above as ci <- confint(object, level = object#conf.level). So with the example object, wt.ci <- confint(wt, level=.95), you can do
R> wt.ci$estimate
difference in location
-0.305
which matches
R> wt
Exact Wilcoxon Mann-Whitney Rank Sum Test
data: pd by age (12-26 Weeks, At term)
Z = -1.2247, p-value = 0.2544
alternative hypothesis: true mu is not equal to 0
95 percent confidence interval:
-0.76 0.15
sample estimates:
difference in location
-0.305
From the help file ?wilcox_test():
water_transfer <- data.frame(
pd = c(0.80, 0.83, 1.89, 1.04,
1.45, 1.38, 1.91, 1.64, 0.73, 1.46,
1.15, 0.88, 0.90, 0.74, 1.21),
age = factor(c(rep("At term", 10), rep("12-26 Weeks", 5))))
##
wt <- wilcox_test(pd ~ age, data = water_transfer,
distribution = "exact", conf.int = TRUE)