Cohen's Effect Size Interpretation

[UK_Professor]

Cohen’s guidelines for interpreting small, medium, and large effect sizes are given as points (e.g., d of .2 = small, .5 = medium, .8 = large) rather than ranges. I have seen one book indicate that anything less than.2 is small, anything above .2 but less than .5 is medium, and anything at or above .5 is large, but I have not seen anyone else interpret Cohen’s guidelines in this way (everyone seems to just duplicate his points rather than provide ranges) and I don’t see any indication in Cohen’s book about how to interpret effect sizes falling between the given points. What is the "correct" (read: accepted) was to interpret Cohen’s guidelines: as upper-limits to each effect size, as that one book indicated, or as mid-points for each effect size? For example, is d = .25 a medium effect because it is over .2, or is it a small effect because it is closer to .2 than .5?

Thanks,
Jason

[longrob]

I always thought those "guidelines" were a bit arbitrary, and certainly shouldn't be considered as upper/lower limits. If you were to use that interpretation it would cause difficulty where values were very close to the limits: 0.195 is small, while 0.205 is medium ? It seems much more sensible to treat them like midpoints of a band. In the example you gave, I would say it's a "fairly small" effect but in essence it's not really unlike classifying p values as providing "strong evidence", "very strong evidence" etc.