Desirability

[mattansb]

library(effectsize)

data("desirability")

head(desirability)
#>   participant_id poop chocolate
#> 1          088cf    4         6
#> 2          bcae1    1         7
#> 3          d21b0    1         7
#> 4          2019a    1         7
#> 5          c1b73    1         7
#> 6          e8e9d    1         7

repeated_measures_d(Pair(chocolate, poop) ~ 1, data = desirability)
#> d (rm) |       95% CI
#> ---------------------
#> 6.16   | [5.59, 6.73]
#> 
#> - Adjusted for small sample bias.

# More options:
repeated_measures_d(Pair(chocolate, poop) ~ 1, data = desirability, mu = 3.4)
#> d (rm) |       95% CI
#> ---------------------
#> 2.14   | [1.90, 2.37]
#> 
#> - Adjusted for small sample bias.
#> - Deviation from a difference of 3.4.
repeated_measures_d(Pair(chocolate, poop) ~ 1, data = desirability, alternative = "greater")
#> d (rm) |      95% CI
#> --------------------
#> 6.16   | [5.68, Inf]
#> 
#> - Adjusted for small sample bias.
#> - One-sided CIs: upper bound fixed at [Inf].

# Other methods
repeated_measures_d(Pair(chocolate, poop) ~ 1, data = desirability, method = "av")
#> d (av) |       95% CI
#> ---------------------
#> 6.15   | [6.02, 6.28]
#> 
#> - Adjusted for small sample bias.
repeated_measures_d(Pair(chocolate, poop) ~ 1, data = desirability, method = "b")
#> Becker's d |       95% CI
#> -------------------------
#> 8.20       | [7.67, 8.73]
#> 
#> - Adjusted for small sample bias.
repeated_measures_d(Pair(chocolate, poop) ~ 1, data = desirability, method = "d")
#> Cohen's d |       95% CI
#> ------------------------
#> 6.15      | [5.85, 6.46]
#> 
#> - Adjusted for small sample bias.
repeated_measures_d(Pair(chocolate, poop) ~ 1, data = desirability, method = "z")
#> d (z) |       95% CI
#> --------------------
#> 4.27  | [3.99, 4.55]
#> 
#> - Adjusted for small sample bias.

# d_z is the same as Cohen's d for one sample (of individual difference):
cohens_d(chocolate - poop ~ 1, data = desirability)
#> Cohen's d |       95% CI
#> ------------------------
#> 4.27      | [3.99, 4.55]

^{Created on 2025-10-17 with reprex v2.1.1}

[DominiqueMakowski]

Lol funniest shit I've seen in a while

[ianhussey]

Hi - person who collected the poop/chocolate data here.

Could I suggest that you instead use the following data subset and variables in the examples?

https://github.com/ianhussey/not-so-simple-preferences/blob/main/data/processed/data_processed_subset_with_exclusions.csv

The dataset you're currently using doesn't a) apply the various attention checks and other exclusions and b) uses the desirability items. While that's the one i posted about on bsky, for various reasons I suspect that its not a good benchmark, and not the one we'll be communicating later as a good example of a maximum positive control effect size. Instead I suggest you use the three-item evaluations scale, which instead estimates "how positive or negative is a bowl of chocolates relative to a toilet filled with human poop". These columns were called 'classic' in the original data, but I've renamed them to "evaluations" here for you, selected only the relevant columns, and applied the exclusions.

Suggested citation:
Hussey, I., & Cummins, J. (2025). (Not so) simple preferences. https://github.com/ianhussey/not-so-simple-preferences

[mattansb]

Thanks @ianhussey!

I've updated the data and docs. The docs are here:

https://github.com/easystats/effectsize/pull/685/files#diff-9e9409e340d36be06d00cbeefbb4a7d7a2a3e980a75fd541c8a863f4e6c5f77fR87

Here is the updated example from ?repeated_measures_d()

library(effectsize)

data("preferences")
# Is chocolate preferred over... poop?

repeated_measures_d(Pair(chocolate, poop) ~ 1, data = preferences)
#> d (rm) |       95% CI
#> ---------------------
#> 5.58   | [5.01, 6.15]
#> 
#> - Adjusted for small sample bias.

# More options:
repeated_measures_d(Pair(chocolate, poop) ~ 1, data = preferences,
                    mu = 3.4)
#> d (rm) |       95% CI
#> ---------------------
#> 1.53   | [1.32, 1.73]
#> 
#> - Adjusted for small sample bias.
#> - Deviation from a difference of 3.4.
repeated_measures_d(Pair(chocolate, poop) ~ 1, data = preferences,
                    alternative = "greater")
#> d (rm) |      95% CI
#> --------------------
#> 5.58   | [5.11, Inf]
#> 
#> - Adjusted for small sample bias.
#> - One-sided CIs: upper bound fixed at [Inf].

# Other methods
repeated_measures_d(Pair(chocolate, poop) ~ 1, data = preferences,
                    method = "av")
#> d (av) |       95% CI
#> ---------------------
#> 5.57   | [5.43, 5.71]
#> 
#> - Adjusted for small sample bias.
repeated_measures_d(Pair(chocolate, poop) ~ 1, data = preferences,
                    method = "b")
#> Becker's d |       95% CI
#> -------------------------
#> 5.17       | [4.82, 5.53]
#> 
#> - Adjusted for small sample bias.
repeated_measures_d(Pair(chocolate, poop) ~ 1, data = preferences,
                    method = "d")
#> Cohen's d |       95% CI
#> ------------------------
#> 5.58      | [5.29, 5.86]
#> 
#> - Adjusted for small sample bias.
repeated_measures_d(Pair(chocolate, poop) ~ 1, data = preferences,
                    method = "z")
#> d (z) |       95% CI
#> --------------------
#> 3.56  | [3.32, 3.80]
#> 
#> - Adjusted for small sample bias.

^{Created on 2025-10-20 with reprex v2.1.1}

[bwiernik]

Came here really hoping the WORDLIST update was to add "poop", but was alas disappointed

[mattansb]

@ianhussey okay to merge?

[ianhussey]

yes fine by me!