Calculate Cohen's d for each variable across groups, comparing each group to the overall sample. Supports continuous and categorical (one-hot encoded) variables.
Arguments
- data
A data frame.
- group
Character, name of the grouping variable (factor).
- vars
Character vector of variable names to compare.
- onehot
Logical. If
TRUE(default), categorical variables are one-hot encoded before computing Cohen's d.- col_labels
Optional named character vector for renaming variables in the output (e.g.
c(age = "Age (years)")).
Value
A data frame with columns: Variable and one column per
group level, containing Cohen's d values.
Examples
# Basic: continuous variables
stat_cohen(iris, group = "Species",
vars = c("Sepal.Length", "Sepal.Width"))
#> Variable setosa versicolor virginica
#> 1 Sepal.Length -1.1324106 0.1214746 0.9487421
#> 2 Sepal.Width 0.8772742 -0.7024173 -0.2028886
# With column label mapping
stat_cohen(iris, group = "Species",
vars = c("Sepal.Length", "Sepal.Width"),
col_labels = c(Sepal.Length = "Sepal L.", Sepal.Width = "Sepal W."))
#> Variable setosa versicolor virginica
#> 1 Sepal L. -1.1324106 0.1214746 0.9487421
#> 2 Sepal W. 0.8772742 -0.7024173 -0.2028886
# Mixed variables with one-hot encoding
df <- data.frame(
group = factor(sample(c("A","B","C"), 100, TRUE)),
age = rnorm(100, 50, 10),
sex = factor(sample(c("M","F"), 100, TRUE)),
stage = factor(sample(c("I","II","III"), 100, TRUE))
)
stat_cohen(df, group = "group", vars = c("age", "sex", "stage"))
#> Variable A B C
#> 1 age 0.152936711 -0.04949980 -0.03634952
#> 2 sex_M -0.009060075 -0.03737443 0.03041184
#> 3 stage_I 0.017522616 0.07194424 -0.05930769
#> 4 stage_II 0.007510451 0.03094770 -0.02524949
#> 5 stage_III -0.024969200 -0.10397398 0.08286605
# Without one-hot (numeric variables only)
stat_cohen(df, group = "group", vars = "age", onehot = FALSE)
#> Variable A B C
#> 1 age 0.1529367 -0.0494998 -0.03634952