There are several R packages with functionality to calculate effect size. Recently I start to use the package pwr. Taking pwr.t.test for example (within the context of two sample t-test), here it says once sample size, significance level and power are specified, effect size will be determined.
I have two questions:
What kind of the effect size does
pwr.t.testreturn?As far as I know, Cohen's d is one type of effect size. It is calculated purely by difference of means and standard deviation of the two samples. If
pwr.t.testreturns Cohen's d or similar metrics, why do we need to specify power and significance level?
Correct me if I am wrong. Thank you.
Best Answer
pwr.t.test does not calculate effect size at all, but statistical power. The package is called "pwr" because it is meant for power analysis. From that package, cohen.ES can compute effect sizes.