!! Post-hoc tests for ANOVA and Welch ANOVA

Games-Howell, Tukey HSD and other post-hoc tests for ANOVA and Welch ANOVA. Either Games-Howell test or Tukey honestly significant difference (HSD) post-hoc tests for one-way analysis of variance (ANOVA).

posthoc_anova(
  y,
  group = NULL,
  method = c("Games-Howell", "Tukey"),
  conf_level = 0.95,
  digits = 3,
  digits_p = 3,
  p_adjust = "none",
  format_pvalue = TRUE,
  data = NULL,
  ...
)

# S3 method for formula
posthoc_anova(
  y,
  group = NULL,
  method = c("Games-Howell", "Tukey"),
  conf_level = 0.95,
  digits = 3,
  digits_p = 3,
  p_adjust = "none",
  format_pvalue = TRUE,
  data = NULL,
  ...,
  sep = " | "
)

# S3 method for default
posthoc_anova(
  y,
  group = NULL,
  method = c("Games-Howell", "Tukey"),
  conf_level = 0.95,
  digits = 3,
  digits_p = 3,
  p_adjust = "none",
  format_pvalue = TRUE,
  data = NULL,
  ...
)

# S3 method for posthoc_anova
print(
  x,
  digits = x$input$digits,
  digits_p = x$input$digits_p,
  digits_param = 2,
  ...
)

# S3 method for posthoc_anova
plot(
  x,
  ...,
  zero_line_color = "grey",
  add_p = TRUE,
  p_size = 1,
  p_color = "blue",
  p_pos_adj = 0.22,
  flip_xy = TRUE
)

Arguments

y	(numeric|formula) Either a numeric vector or formula.
group	(factor) a vector that either is a factor or can be converted into one. If y is a formula, group is ignored.
method	("games-howell"|"tukey") Name of post-hoc tests to conduct. Valid values are "tukey" and "games-howell".
conf_level	(number) Confidence level (equals to 1 - alpha, where alpha is significanve level). Number from 0 to 1. Default is 0.95.
digits	(integer) The number of digits to round data related numbers to.
digits_p	(integer) The number of digits to round p values to. Must be 2, 3 or more.
p_adjust	Any valid stats::p.adjust() method.
format_pvalue	(does not work yet) Whether to format the p values according to APA standards (i.e. replace all values lower than .001 with '<.001'). This only applies to the printing of the object, not to the way the p values are stored in the object.
data	(data frame) a data frame with data to be used in combination with formula.
...	Further arguments to methods.
sep	(character) A string with a symbol to separate group names if several grouping variables are used.
x	object to print.
zero_line_color	(character) Color for line indicating zero differences.
add_p	(logical) Flag if p values should be added.
p_size	(numeric) Font size to p-values-related text.
p_color	(character) Color for p values.
p_pos_adj	(numeric) Factor for p value position correction
flip_xy	(logical) Flag if x and y axes should be swapped.

Value

A list of three elements:

input

List with input arguments

output

List with post-hoc test results

.

Details

The main part of the function code and descriptions were imported from posthocTGH() in package userfriendlyscience.

Note

Options that carry out Games-Howell and Tukey HSD analyses are based on code of function posthocTGH() in package userfriendlyscience (version 0.7.0).

See also

• http://www.gcf.dkf.unibe.ch/BCB/files/BCB_10Jan12_Alexander.pdf

• https://rpubs.com/aaronsc32/games-howell-test

• http://aoki2.si.gunma-u.ac.jp/R/src/tukey.R

• https://stats.stackexchange.com/questions/83941/games-howell-post-hoc-test-in-r

Examples

library(biostat)

# Compute post-hoc statistics using the Games-Howell method
posthoc_anova(weight ~ Diet, data = ChickWeight, method = "Games-Howell")
#> Games-Howell test results (ANOVA post-hoc) 
#> 
#>   groups difference ci_lower ci_upper    t     df          p
#> 1    2-1     19.971    0.358   39.585 2.64 201.38  0.044 *  
#> 2    3-1     40.305   17.544   63.065 4.59 175.92 <0.001 ***
#> 3    4-1     32.617   13.452   51.783 4.41 203.16 <0.001 ***
#> 4    3-2     20.333   -6.203   46.870 1.98 229.94  0.197    
#> 5    4-2     12.646  -10.909   36.201 1.39 235.88  0.507    
#> 6    4-3     -7.687  -33.899   18.524 0.76 226.16  0.873    
#> 
#>   group   n    mean variance     sd cld spaced_cld
#> 1     1 220 102.645 3209.965 56.657   a         a_
#> 2     2 120 122.617 5127.633 71.607   b         _b
#> 3     3 120 142.950 7489.476 86.542   b         _b
#> 4     4 118 135.263 4737.392 68.829   b         _b

# Compute post-hoc statistics using the Tukey method
posthoc_anova(weight ~ Diet, data = ChickWeight, method = "Tukey")
#> Tukey HSD test results (ANOVA post-hoc) 
#> 
#>   groups difference ci_lower ci_upper    t     df          p
#> 1    2-1     19.971   -0.300   40.242 2.54 574.00  0.055 .  
#> 2    3-1     40.305   20.034   60.576 5.12 574.00 <0.001 ***
#> 3    4-1     32.617   12.235   52.999 4.12 574.00 <0.001 ***
#> 4    3-2     20.333   -2.727   43.394 2.27 574.00  0.106    
#> 5    4-2     12.646  -10.512   35.804 1.41 574.00  0.495    
#> 6    4-3     -7.687  -30.845   15.470 0.86 574.00  0.828    
#> 
#>   group   n    mean variance     sd cld spaced_cld
#> 1     1 220 102.645 3209.965 56.657   a         a_
#> 2     2 120 122.617 5127.633 71.607  ab         ab
#> 3     3 120 142.950 7489.476 86.542   b         _b
#> 4     4 118 135.263 4737.392 68.829   b         _b