Meta-Analysis: Transforming F scores to Cohen's d


New Member
I need to convert F statistics from 1-Way & from 2-Way ANOVAs into Cohen's d. The formula in my handbook:

d = (2 * the square root of F) /over/ (the square root of df(error))

My question is: Is df(error) equal to (N - k) - (n - 1)? Or does "error" denote something else?

If anyone has a link to an example of a conversion like this worked through, that would also be extremely helpful & appreciated.

I don't know a lot about meta-analysis but (N-k)-(n-1) is dferror for a repeated-measures anova, whereas (N-k) is df for the error term (often called dfwithin, may also sometimes be called dferror, I can't remember for sure) for an independent-measures anova.
Last edited: