Print

Print
Hi VM,

I'm unfamiliar with this formula (do you really mean multiplying N1 by N2?). Regardless, the answer to this can be long, but in short, in the original book Cohen didn't go into all possible cases of the GLM. In PALM, Cohen's d is computed as COPE/sqrt(variance of the residuals).

The relationship of this with the t statistic depends on a certain term that is computed as (C'*(M'*M)^(-1)*C)^-(1/2), where M is the full design matrix (effects of interest and nuisance) and C is the contrast.

All the best,

Anderson


On 15 February 2017 at 15:13, neuroimage analyst <[log in to unmask]> wrote:
Thank you, Anderson.

Since PALM returns both t stat and degrees of freedom, I calculated Cohen's d using

(N1+N2)×t-stat/(sqrt (dof×N1×N2)) using N for my 2 groups  but this isn't the same as returned by Cohen's d.csv. I will greatly appreciate if you could comment on why that would be the case.

Thanks 

Regards 

-VM

On Feb 15, 2017 2:29 AM, "Anderson M. Winkler" <[log in to unmask]> wrote:
Hi VM,

It's the coefficient of variation: https://en.wikipedia.org/wiki/Coefficient_of_variation

Actually it's a bit more general than the Wikipedia article suggests. It's here the reciprocal of Cohen's d.

All the best,

Anderson


On 15 February 2017 at 00:24, neuroimage analyst <[log in to unmask]> wrote:
Hi, Anderson,

What does *_cfvar* signify? It is returned when --saveglm is on.

Thanks

Regards

--VM