Thanks for the gracious reply. But I reiterate
my warning against casting significance test reports
in confidence terminology.
Nick
[email protected]
White, Justin
> I am sorry for the misinformation. After I read Nick's response, it
> jogged my memory from Intro to Econometrics. It's been a
> while. Sorry.
>
> What I should have said is this:
> Let's say you have a p-value of 0.0890 from an F-test. This tells us
> that we can expect coefficients as extreme as observed in 8.9
> out of 100
> random samples given the null hypothesis is true. We can use this
> information to either reject or fail to reject the null based on
> personal confidence criterion. In this case we can reject
> the null at a
> 91.1% confidence interval.
Nick Cox
> Note that this is wrong. The P-value is emphatically not
> the probability that the null hypothesis is true.
> The P-value is the probability of getting results as or
> more extreme than those observed _if_ the null hypothesis
> is true and if the associated assumptions are correct.
>
> In any case, the picture here that there is a spike of probability
> corresponding to a zero test statistic does not match basic
> facts about the sampling distribution, which in this kind of
> problem is continuous.
White, Justin
> > Here is how to interpret a p-value. Let's say you have a p-value of
> > 0.0890 from an F-test. This tells us that given the data
> > sample, we can
> > expect the estimated coefficients to be jointly equal to zero in 8.9
> > times out of 100. This is known as Type-1 error.
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/