Cohen's small/medium/large categories for Epsilon-squared? #681

^{Originally posted by jarbet August 15, 2025}
I often use epsilon-squared ($\epsilon^2$) effect size when doing the Kruskal-Wallis Test. However, I have a question about interpreting the relative size of $\epsilon^2$. The help page of interpret_epsilon_squared says that Cohen's categories are:

However, taking a look at the Cohen 1992 paper, I'm not sure how the above was derived?

Should the cutoffs instead be from line 8 of the above Table, which are 0.02, 0.15, 0.35?