Talk:Sample Size, Effect Size, and Power

>Different people offer different advice regarding how to interpret the resultant >effect size, but the most accepted opinion is that of Cohen (1992) where 0.2 is >indicative of a small effect, 0.5 a medium and 0.8 a large effect size.

The effect size advice is (problematic). The page focusses on standardized effect size and then only on one type (the d family) and ignores the r family (which crops up in SPSS much more than the d family). In general unstandardized effect sizes using meangful units are preferable to standardized ones.

However, the most innacurate bit is the quote about "the most accepted opinion" being the use Cohen's guidelines on size of effect. First I know of no serious statistician who thinks Cohen's definitions of small, medium and large effect size are useful outside of a very narrow context (if at all). Even Cohen suggested them as a last resort rule-of-thumb if you have no other information to go on.

Thom

Comments welcome
Hi Thom,

Yes - this page needs some work. The current content is less than ideal. You are welcome to edit the page in any way you see fit, but keep in mind SPSSwiki is not a stats site, so much as a SPSS workbook. While accuarcy of information is very important, the purpose is to show how you can do it with SPSS.

Thanks again for the input!

Cheers,

--Zoole 01:06, 27 January 2006 (UTC)