[From Bill Powers (2008.12.10.0658 MST)]
I was looking up the meaning of “effect size” and found the
following table, which shows the relationships between CHohen’s d (effect
size expressed as difference between means divided by standard deviation)
and r (Pearson’s r, correlation).
Note that a correlation of 0.371 goes with an effect size that is called
“large”, and one of 0.243 goes with “medium.” This is
rather sad. If I were working in a field where those definitions applied,
I would be terribly discouraged.