Different results for confidence intervals using Wilcoxon paired signed-rank test

I get different confidence intervals when using the package statsExpressionsand the package rstatix in a Wilcoxon paired signed-rank test. I think statsExpressions uses rstatix, so I don't understand the difference. The first gives 0.24 - 0.884, and the second 1 - 6.

library(tidyverse)
library(BSDA)
#> Loading required package: lattice
#> 
#> Attaching package: 'BSDA'
#> The following object is masked from 'package:datasets':
#> 
#>     Orange
library(ggstatsplot)
#> You can cite this package as:
#>      Patil, I. (2021). Visualizations with statistical details: The 'ggstatsplot' approach.
#>      Journal of Open Source Software, 6(61), 3167, doi:10.21105/joss.03167
library(statsExpressions)
library(rstatix)
#> 
#> Attaching package: 'rstatix'
#> The following object is masked from 'package:stats':
#> 
#>     filter

# make long format - tidy data
d <- Speed %>% 
  gather(key = "speed", value = "score", before, after)

# Compute Paired Samples Wilcoxon Signed Rank Test

# statsExpressions
  two_sample_test(
    data = d,
    x = speed,
    y = score,
    subject.id = NULL,
    type = "nonparametric",
    paired = TRUE,
    alternative = "two.sided",
    k = 2L,
    conf.level = 0.95,
    effsize.type = "d",
    var.equal = FALSE,
    bf.prior = 0.707,
    tr = 0.2,
    nboot = 100L,
    conf.int = TRUE
  )
#> # A tibble: 1 × 14
#>   parameter1 parameter2 statistic p.value method                    alternative
#>   <chr>      <chr>          <dbl>   <dbl> <chr>                     <chr>      
#> 1 score      speed           100.  0.0229 Wilcoxon signed rank test two.sided  
#>   effectsize        estimate conf.level conf.low conf.high conf.method n.obs
#>   <chr>                <dbl>      <dbl>    <dbl>     <dbl> <chr>       <int>
#> 1 r (rank biserial)    0.675       0.95    0.240     0.884 normal         15
#>   expression
#>   <list>    
#> 1 <language>

wilcox_test(
  data = d,
  formula = score ~ speed,
  comparisons = NULL,
  ref.group = NULL,
  p.adjust.method = "holm",
  paired = TRUE,
  exact = NULL,
  alternative = "two.sided",
  mu = 0,
  conf.level = 0.95,
  detailed = TRUE
)
#> # A tibble: 1 × 12
#>   estimate .y.   group1 group2    n1    n2 statistic      p conf.low conf.high
#> *    <dbl> <chr> <chr>  <chr>  <int> <int>     <dbl>  <dbl>    <dbl>     <dbl>
#> 1     3.50 score after  before    15    15      100. 0.0229     1.00      6.00
#> # ℹ 2 more variables: method <chr>, alternative <chr>

Created on 2023-09-26 with reprex v2.0.2

I'm just pointing out that they agree in the statistic, but the disagreements are about the effect size

1 Like

They are two different test? maybe it is possible that they applied a different method to calculate the p adjust.

1 Like

The first uses a Baysiasn prior parameter, and the second doesn't.

This topic was automatically closed 42 days after the last reply. New replies are no longer allowed.

If you have a query related to it or one of the replies, start a new topic and refer back with a link.