Power of t-test

[NN_STAT]

Hello all,

I want to write an R procedure that calculates the power of a simple two samples t-test. Does anyone here has an idea how to do it ?

Thank you !

 

[trinker]

I'd say don't remake the wheel, use the pwr package:

http://cran.r-project.org/web/packages/pwr/index.html

 

[BGM]

http://math.arizona.edu/~jwatkins/r-composite.pdf

The document give some example of computing the power function of z-test.


For your problem, first you need to state your alternative hypothesis. Next, note that we can compute the power function for a given level. Also the rejection probability is a function of the nuisance parameters (the population means in the two population) so we can just express the power as a function. Some may plug in the MLE in the function as an estimate.