Illustration of Power of a Test
A reader read my discusssion of Beta Error. He ran the analysis, to determine the power of a test under certain null hypotheses.
If the “true” trend in climate from 2001-2007 was 0 C/century, and we wished to run a hypothesis test on an incorrect “null” hypothesis that the trend was 1.5C/century, we would only falsify 10% of the time.
That is, 90% of the time, we would fail to falsify 1.5 C/century, even though that is wrong.
