PDA

View Full Version : Help needed - estimating effect size

that_guy_
08-01-2008, 12:13 PM
Hi everyone,

I'd really appreciate any help/guidance in solving a little problem of mine.

I need to know how to estimate an effect size (cohen's d) when the standard deviation is not reported?

You may assume that my stats knowledge is fairly limited in your explanation lol.

TG.

that_guy_
08-02-2008, 07:03 AM
noone have any suggestions?

Andy McKenzie
08-04-2008, 11:34 PM
To use Cohen's D, you're going to need some sort of standard deviation info. However, it is possible that you could estimate the standard deviation, if you assume that your distribution is normal. This means that your data isn't skewed towards one direction or the other.

If this is the case, then you can use the formula

standard deviation estimate = (largest possible value - smallest possible value) / 6

This is not as reliable, but it could be used as an estimate. Note that you could divide by 4 too, and it would yield a more conservative estimate.

that_guy_
08-05-2008, 02:00 PM
That's great, sounds simple enough, do you have a reference for this type of equation so i can reference it in my text ?