:2. Solution. That the observed significaance level is 4% does
not mean or imply that the chance of the null hypothesis being
correct is 4%, nor that the chance of it being incorrect is 4%.
It does imply that if the null hypothesis (not the alternative
hypothesis!) is true, then the chances of the test statistic
turning out to be as extreme or more extreme than it did is only
4%, and so something did happen which had a chance of only one
out of twenty of happpening. This is not beyond the bounds of
possibility, but is rather difficult to explain as a chance
variation.