Short answer: It's the difference divided by the standard deviation of
the difference.
What a t-test does is subtracts one score from the other, to give a
difference score. Then it tests that difference against zero.
So the effect size (in Cohen's d terms) is the difference from zero
divided by the standard deviation. However, there are lots of other
ways to calculate effect sizes - Andy Field always turns then into r
(in his Discovering Statistics book(s)), and that's OK too.
J
2009/12/2 Sally Keegetta <[log in to unmask]>:
> Could anyone advise me about the calculation of effect size for paired sample
> t-tests?There seems to be some debate about the best way to do it and I
> would be interested to know what people think.
> Thanks,
> Sally
>
--
Jeremy Miles
Psychology Research Methods Wiki: www.researchmethodsinpsychology.com
|