Re: Hypothesis testing, two different rates


On 23-Aug-2005, Eric Bohlman <ebohlman@xxxxxxxxxx> wrote:

> > The b parameter is -0.0012 with a standard error of 0.006469. The t
> > score is -0.19 and the Prob(t) -- probability that b could be zero --
> > is 0.85616.
>
> Semi-nitpick: it's not the probability that "b could be zero"; it's the
> probability that a sample of that size could have a calculated b with an
> absolute value of at least 0.0012 *if* the b in the "true model" were
> actually zero. More formally, it's P(abs(bhat)>=0.0012|b=0) rather than P
> (b=0|abs(bhat)<=0.0012).

I accept that amendment.

--
Phil Sherrod
(phil.sherrod 'at' sandh.com)
http://www.dtreg.com (decision tree and SVM predictive modeling)
http://www.nlreg.com (nonlinear regression)
.