Re: Hypothesis testing with proportions


Am 27.07.2005 um 11:21 wurde geantwortet auf:
"m00es" <m00es@xxxxxxxxx> schrieb:
> A variance stabilizing transformation (in this particular case, the
> arcsin transformation) is often applied before analyzing proportions in
> such a manner. Therefore, take each proportion and apply
>
> arcsin( sqrt(p) )
>
> which removes, approximately, the dependency of the variance on the
> mean.
>
> Best,
>
> m00es

Sorry I don't get it. What does this function do? Am I supposed to apply this
function to each proportion and then do the t-test with the new set of numbers
with H0: arcsin (sqrt(0.5))?
.