Ah, I have been tagged by Robin.
The rules:
Pick up the nearest book of 123 pages or more. (No cheating!)
Find Page 123.
Find the first 5 sentences.
Post the next 3 sentences.
Tag 5 people.
This was a close call between Error and the Growth of Experimental Knowledge, by Deborah Mayo and Risk Analysis in Engineering and Economics by Bilal M. Ayyub. EatGoEK won out by a hair, so here it is-- page 123, first 3 sentences *after* the first 5 sentences:
Now let us substitute my error statistical construal of probable success (for some specified measure of "successful outcome"). Evaluating H's probable success (or H's reliability) means evaluating the relative frequency with which applications of H would yield results in accordance with(i.e., specifiably close to) what H asserts. As complex as this task sounds, it is just the kind of information afforded by experimental knowledge of H.
This passage comes in the near conclusion of Mayo's criticism of the Bayesian approach to scientific inference (a la Howson & Urbach), and right before her presentation of her error statistical rehabilitation of Neyman-Perason statistics. I would be remiss to mention Mayo without pointing-out Cosma's excellent review of EatGoEK .
As for the five people, I tag Dani, Otis, Josh, Richard, and Michelle.