Hi,

I've been using G*Power to perform power analyses. For within-subject t-tests. G*Power requires you to specify the effect size in the form dz for these within-subject (dependent) t-tests rather than the standard Cohen's d used for between subject tests. It provides the formula for dz as:

dz = |μz |/σz = |μx − μy|/√(σx2 + σy2 − 2ρxy · σx · σy)

Unfortunately, I'm taking the effect size from an old paper that didn't include means and standard deviations. It provides the effect size in the form of partial eta-squared. I know it is possible to convert an eta-squared value to d but I need to convert it to dz.

Does anyone know the formula for computing dz from eta-squared? I'd be hugely grateful for any assistance.

Ryan.