Package Status Usage GitHub References
CRAN_Release_Badge Travis Build Status Daily downloads badge GitHub version Website
CRAN Checks AppVeyor Build Status Weekly downloads badge Forks Rdoc
minimal R version lifecycle Monthly downloads badge Github Issues vignettes
GitHub code size in bytes Coverage Status Total downloads badge Github Stars DOI
Licence Codecov test coverage HitCount Last-changedate GitHub last commit
status Covrpage Summary Gitter chat Project Status contributions welcome

Introduction

pairwiseComparisons provides a tidy data friendly way to carry out pairwise comparison tests.

It currently supports post hoc multiple pairwise comparisons tests for both between-subjects and within-subjects one-way analysis of variance designs. For both of these designs, parametric, non-parametric, and robust statistical tests are available.

Installation

To get the latest, stable CRAN release (0.1.4):

install.packages(pkgs = "pairwiseComparisons")

You can get the development version of the package from GitHub (0.1.4.9000). To see what new changes (and bug fixes) have been made to the package since the last release on CRAN, you can check the detailed log of changes here: https://indrajeetpatil.github.io/pairwiseComparisons/news/index.html

If you are in hurry and want to reduce the time of installation, prefer-

# needed package to download from GitHub repo
install.packages(pkgs = "remotes")

# downloading the package from GitHub
remotes::install_github(
  repo = "IndrajeetPatil/pairwiseComparisons", # package path on GitHub
  dependencies = FALSE, # assumes you have already installed needed packages
  quick = TRUE # skips docs, demos, and vignettes
)

If time is not a constraint-

remotes::install_github(
  repo = "IndrajeetPatil/pairwiseComparisons", # package path on GitHub
  dependencies = TRUE, # installs packages which pairwiseComparisons depends on
  upgrade_dependencies = TRUE # updates any out of date dependencies
)

Summary of types of statistical analyses

Following table contains a brief summary of the currently supported pairwise comparison tests-

Between-subjects design

Type Equal variance? Test p-value adjustment?
Parametric No Games-Howell test Yes
Parametric Yes Student’s t-test Yes
Non-parametric No Dwass-Steel-Crichtlow-Fligner test Yes
Robust No Yuen’s trimmed means test Yes
Bayes Factor No No No
Bayes Factor Yes No No

Within-subjects design

Type Test p-value adjustment?
Parametric Student’s t-test Yes
Non-parametric Durbin-Conover test Yes
Robust Yuen’s trimmed means test Yes
Bayes Factor No No

Examples

Here we will see specific examples of how to use this function for different types of

  • designs (between or within subjects)
  • statistics (parametric, non-parametric, robust)
  • p-value adjustment methods

Between-subjects design

# for reproducibility
set.seed(123)
library(pairwiseComparisons)

# parametric
# if `var.equal = TRUE`, then Student's *t*-test will be run
pairwise_comparisons(
  data = ggplot2::msleep,
  x = vore,
  y = brainwt,
  type = "parametric",
  var.equal = TRUE,
  paired = FALSE,
  p.adjust.method = "bonferroni"
)
#> # A tibble: 6 x 8
#>   group1  group2  mean.difference p.value significance label                                 test.details    
#>   <chr>   <chr>             <dbl>   <dbl> <chr>        <chr>                                 <chr>           
#> 1 carni   herbi            0.542    1     ns           list(~italic(p)[ adjusted ]== 1.000 ) Student's t-test
#> 2 carni   insecti         -0.0577   1     ns           list(~italic(p)[ adjusted ]== 1.000 ) Student's t-test
#> 3 carni   omni             0.0665   1     ns           list(~italic(p)[ adjusted ]== 1.000 ) Student's t-test
#> 4 herbi   insecti         -0.600    1     ns           list(~italic(p)[ adjusted ]== 1.000 ) Student's t-test
#> 5 herbi   omni            -0.476    0.979 ns           list(~italic(p)[ adjusted ]== 0.979 ) Student's t-test
#> 6 insecti omni             0.124    1     ns           list(~italic(p)[ adjusted ]== 1.000 ) Student's t-test
#>   p.value.adjustment
#>   <chr>             
#> 1 Bonferroni        
#> 2 Bonferroni        
#> 3 Bonferroni        
#> 4 Bonferroni        
#> 5 Bonferroni        
#> 6 Bonferroni

# if `var.equal = FALSE`, then Games-Howell test will be run
pairwise_comparisons(
  data = ggplot2::msleep,
  x = vore,
  y = brainwt,
  type = "parametric",
  var.equal = FALSE,
  paired = FALSE,
  p.adjust.method = "bonferroni"
)
#> # A tibble: 6 x 11
#>   group1 group2  mean.difference    se t.value    df p.value significance label                                
#>   <chr>  <chr>             <dbl> <dbl>   <dbl> <dbl>   <dbl> <chr>        <chr>                                
#> 1 omni   herbi             0.476 0.255   1.32   20.9       1 ns           list(~italic(p)[ adjusted ]== 1.000 )
#> 2 omni   carni            -0.066 0.061   0.774  21.1       1 ns           list(~italic(p)[ adjusted ]== 1.000 )
#> 3 omni   insecti          -0.124 0.057   1.55   17.2       1 ns           list(~italic(p)[ adjusted ]== 1.000 )
#> 4 herbi  carni            -0.542 0.25    1.54   19.4       1 ns           list(~italic(p)[ adjusted ]== 1.000 )
#> 5 herbi  insecti          -0.6   0.249   1.70   19.1       1 ns           list(~italic(p)[ adjusted ]== 1.000 )
#> 6 carni  insecti          -0.058 0.027   1.53   10.7       1 ns           list(~italic(p)[ adjusted ]== 1.000 )
#>   test.details      p.value.adjustment
#>   <chr>             <chr>             
#> 1 Games-Howell test Bonferroni        
#> 2 Games-Howell test Bonferroni        
#> 3 Games-Howell test Bonferroni        
#> 4 Games-Howell test Bonferroni        
#> 5 Games-Howell test Bonferroni        
#> 6 Games-Howell test Bonferroni

# non-parametric
pairwise_comparisons(
  data = ggplot2::msleep,
  x = vore,
  y = brainwt,
  type = "nonparametric",
  paired = FALSE,
  p.adjust.method = "none"
)
#> # A tibble: 6 x 8
#>   group1  group2       W p.value significance label                                   test.details                      
#>   <chr>   <chr>    <dbl>   <dbl> <chr>        <chr>                                   <chr>                             
#> 1 carni   herbi   -0.8     0.942 ns           list(~italic(p)[ unadjusted ]== 0.942 ) Dwass-Steel-Crichtlow-Fligner test
#> 2 carni   insecti -2.36    0.342 ns           list(~italic(p)[ unadjusted ]== 0.342 ) Dwass-Steel-Crichtlow-Fligner test
#> 3 carni   omni    -1.72    0.619 ns           list(~italic(p)[ unadjusted ]== 0.619 ) Dwass-Steel-Crichtlow-Fligner test
#> 4 herbi   insecti -2.40    0.325 ns           list(~italic(p)[ unadjusted ]== 0.325 ) Dwass-Steel-Crichtlow-Fligner test
#> 5 herbi   omni    -0.948   0.908 ns           list(~italic(p)[ unadjusted ]== 0.908 ) Dwass-Steel-Crichtlow-Fligner test
#> 6 insecti omni     1.61    0.667 ns           list(~italic(p)[ unadjusted ]== 0.667 ) Dwass-Steel-Crichtlow-Fligner test
#>   p.value.adjustment
#>   <chr>             
#> 1 None              
#> 2 None              
#> 3 None              
#> 4 None              
#> 5 None              
#> 6 None

# robust
pairwise_comparisons(
  data = ggplot2::msleep,
  x = vore,
  y = brainwt,
  type = "robust",
  paired = FALSE,
  p.adjust.method = "fdr"
)
#> # A tibble: 6 x 10
#>   group1  group2    psihat conf.low conf.high p.value significance label                                
#>   <chr>   <chr>      <dbl>    <dbl>     <dbl>   <dbl> <chr>        <chr>                                
#> 1 insecti omni    -0.0556   -0.184     0.0728   0.969 ns           list(~italic(p)[ adjusted ]== 0.969 )
#> 2 carni   herbi   -0.0530   -0.274     0.168    0.969 ns           list(~italic(p)[ adjusted ]== 0.969 )
#> 3 carni   omni     0.00210  -0.151     0.155    0.969 ns           list(~italic(p)[ adjusted ]== 0.969 )
#> 4 herbi   omni     0.0551   -0.173     0.283    0.969 ns           list(~italic(p)[ adjusted ]== 0.969 )
#> 5 carni   insecti  0.0577   -0.0609    0.176    0.969 ns           list(~italic(p)[ adjusted ]== 0.969 )
#> 6 herbi   insecti  0.111    -0.0983    0.320    0.969 ns           list(~italic(p)[ adjusted ]== 0.969 )
#>   test.details              p.value.adjustment  
#>   <chr>                     <chr>               
#> 1 Yuen's trimmed means test Benjamini & Hochberg
#> 2 Yuen's trimmed means test Benjamini & Hochberg
#> 3 Yuen's trimmed means test Benjamini & Hochberg
#> 4 Yuen's trimmed means test Benjamini & Hochberg
#> 5 Yuen's trimmed means test Benjamini & Hochberg
#> 6 Yuen's trimmed means test Benjamini & Hochberg

Within-subjects design

# for reproducibility
set.seed(123)

# parametric
pairwise_comparisons(
  data = bugs_long,
  x = condition,
  y = desire,
  type = "parametric",
  paired = TRUE,
  p.adjust.method = "BH"
)
#> # A tibble: 6 x 8
#>   group1 group2 mean.difference  p.value significance label                                 test.details    
#>   <chr>  <chr>            <dbl>    <dbl> <chr>        <chr>                                 <chr>           
#> 1 HDHF   HDLF            -1.11  1.00e- 3 ***          list(~italic(p)[ adjusted ]== 0.001 ) Student's t-test
#> 2 HDHF   LDHF            -0.474 7.07e- 2 ns           list(~italic(p)[ adjusted ]== 0.071 ) Student's t-test
#> 3 HDHF   LDLF            -2.14  7.64e-12 ***          list(~italic(p)[ adjusted ]<= 0.001 ) Student's t-test
#> 4 HDLF   LDHF             0.637 5.47e- 2 ns           list(~italic(p)[ adjusted ]== 0.055 ) Student's t-test
#> 5 HDLF   LDLF            -1.03  1.39e- 3 **           list(~italic(p)[ adjusted ]== 0.001 ) Student's t-test
#> 6 LDHF   LDLF            -1.66  6.67e- 9 ***          list(~italic(p)[ adjusted ]<= 0.001 ) Student's t-test
#>   p.value.adjustment  
#>   <chr>               
#> 1 Benjamini & Hochberg
#> 2 Benjamini & Hochberg
#> 3 Benjamini & Hochberg
#> 4 Benjamini & Hochberg
#> 5 Benjamini & Hochberg
#> 6 Benjamini & Hochberg

# non-parametric
pairwise_comparisons(
  data = bugs_long,
  x = condition,
  y = desire,
  type = "nonparametric",
  paired = TRUE,
  p.adjust.method = "BY"
)
#> # A tibble: 6 x 8
#>   group1 group2 statistic  p.value significance label                                 test.details       
#>   <chr>  <chr>      <dbl>    <dbl> <chr>        <chr>                                 <chr>              
#> 1 HDHF   HDLF        4.78 1.44e- 5 ***          list(~italic(p)[ adjusted ]<= 0.001 ) Durbin-Conover test
#> 2 HDHF   LDHF        2.44 4.47e- 2 *            list(~italic(p)[ adjusted ]== 0.045 ) Durbin-Conover test
#> 3 HDHF   LDLF        8.01 5.45e-13 ***          list(~italic(p)[ adjusted ]<= 0.001 ) Durbin-Conover test
#> 4 HDLF   LDHF        2.34 4.96e- 2 *            list(~italic(p)[ adjusted ]== 0.050 ) Durbin-Conover test
#> 5 HDLF   LDLF        3.23 5.05e- 3 **           list(~italic(p)[ adjusted ]== 0.005 ) Durbin-Conover test
#> 6 LDHF   LDLF        5.57 4.64e- 7 ***          list(~italic(p)[ adjusted ]<= 0.001 ) Durbin-Conover test
#>   p.value.adjustment   
#>   <chr>                
#> 1 Benjamini & Yekutieli
#> 2 Benjamini & Yekutieli
#> 3 Benjamini & Yekutieli
#> 4 Benjamini & Yekutieli
#> 5 Benjamini & Yekutieli
#> 6 Benjamini & Yekutieli

# robust
pairwise_comparisons(
  data = bugs_long,
  x = condition,
  y = desire,
  type = "robust",
  paired = TRUE,
  p.adjust.method = "hommel"
)
#> # A tibble: 6 x 10
#>   group1 group2 psihat conf.low conf.high  p.value significance label                                
#>   <chr>  <chr>   <dbl>    <dbl>     <dbl>    <dbl> <chr>        <chr>                                
#> 1 HDLF   LDHF   -0.701  -1.71       0.303 6.20e- 2 ns           list(~italic(p)[ adjusted ]== 0.062 )
#> 2 HDHF   LDHF    0.5    -0.188      1.19  6.20e- 2 ns           list(~italic(p)[ adjusted ]== 0.062 )
#> 3 HDLF   LDLF    0.938   0.0694     1.81  1.36e- 2 *            list(~italic(p)[ adjusted ]== 0.014 )
#> 4 HDHF   HDLF    1.16    0.318      2.00  1.49e- 3 **           list(~italic(p)[ adjusted ]== 0.001 )
#> 5 LDHF   LDLF    1.54    0.810      2.27  1.16e- 6 ***          list(~italic(p)[ adjusted ]<= 0.001 )
#> 6 HDHF   LDLF    2.10    1.37       2.82  1.79e-10 ***          list(~italic(p)[ adjusted ]<= 0.001 )
#>   test.details              p.value.adjustment
#>   <chr>                     <chr>             
#> 1 Yuen's trimmed means test Hommel            
#> 2 Yuen's trimmed means test Hommel            
#> 3 Yuen's trimmed means test Hommel            
#> 4 Yuen's trimmed means test Hommel            
#> 5 Yuen's trimmed means test Hommel            
#> 6 Yuen's trimmed means test Hommel

Using pairwiseComparisons with ggsignif to display results


# needed libraries
library(ggplot2)
library(pairwiseComparisons)
library(ggsignif)

# converting to factor
mtcars$cyl <- as.factor(mtcars$cyl)

# creating a basic plot
p <- ggplot(mtcars, aes(cyl, wt)) + geom_boxplot()

# using `pairwiseComparisons` package to create a dataframe with results
(df <-
  pairwise_comparisons(mtcars, cyl, wt, messages = FALSE) %>%
  dplyr::mutate(.data = ., groups = purrr::pmap(.l = list(group1, group2), .f = c)) %>%
  dplyr::arrange(.data = ., group1))
#> # A tibble: 3 x 12
#>   group1 group2 mean.difference    se t.value    df p.value significance label                                
#>   <chr>  <chr>            <dbl> <dbl>   <dbl> <dbl>   <dbl> <chr>        <chr>                                
#> 1 4      8                1.71  0.188    6.44  23.0   0     ***          list(~italic(p)[ adjusted ]<= 0.001 )
#> 2 6      4               -0.831 0.154    3.81  16.0   0.008 **           list(~italic(p)[ adjusted ]== 0.008 )
#> 3 6      8                0.882 0.172    3.62  19.0   0.008 **           list(~italic(p)[ adjusted ]== 0.008 )
#>   test.details      p.value.adjustment groups   
#>   <chr>             <chr>              <list>   
#> 1 Games-Howell test Holm               <chr [2]>
#> 2 Games-Howell test Holm               <chr [2]>
#> 3 Games-Howell test Holm               <chr [2]>

# using `geom_signif` to display results
p +
  ggsignif::geom_signif(
    comparisons = df$groups,
    map_signif_level = TRUE,
    tip_length = 0.01,
    y_position = c(5.5, 5.75, 6),
    annotations = df$label,
    test = NULL,
    na.rm = TRUE,
    parse = TRUE
  )

Code coverage

As the code stands right now, here is the code coverage for all primary functions involved: https://codecov.io/gh/IndrajeetPatil/pairwiseComparisons/tree/master/R

Contributing

I’m happy to receive bug reports, suggestions, questions, and (most of all) contributions to fix problems and add features. I personally prefer using the GitHub issues system over trying to reach out to me in other ways (personal e-mail, Twitter, etc.). Pull Requests for contributions are encouraged.

Here are some simple ways in which you can contribute (in the increasing order of commitment):

  • Read and correct any inconsistencies in the documentation

  • Raise issues about bugs or wanted features

  • Review code

  • Add new functionality (in the form of new plotting functions or helpers for preparing subtitles)

Please note that this project is released with a Contributor Code of Conduct. By participating in this project you agree to abide by its terms.