Reasoning about Luck: Probability and its Uses in Physics, by Vinay Ambegaokar. New York: (c)1996 Cambridge University Press, ISBN 0-521-44217-6 (hc) - ISBN 0-521-44737-2 (pbk):
Vinay Ambegaokar is Goldwyn Smith Prof. of Physics, Cornell University, Ithaca, NY
From p. 7:
“The technical use of the word probability applies in general to a special class of situations. It presupposes a repeatable experiment or observation with more than one possible outcome controlled by chance, which means that before the fact, precisely which outcome will occur is neither known nor deducible.”
From pp. 8-9
"It is worth making the point that the concept of probability makes sense only when there is some understanding or working hypothesis about what is going on.
"This is well illustrated by a story told by my colleague Michael E. Fisher. When he was a student at King’s College, London, a man seeking advice from someone skilled in statistics was directed to him. The inquirer’s research task required that he enter in a ledger the results of urine analyses from a large number of London hospitals. He had begun to suspect, and was excited by the suspicion, that there were slightly more entries showing abnormally high sugar at the upper as opposed to the lower end of each page.
"Fisher asked the man if he had any hypotheses to explain the supposed effect. He did not. Did he start a new page every morning? Did the results come to him from different hospitals in some special order, or did he, himself, order the entries according to some plan or intuition? Were emergency tests treated differently from routine ones? Did the results come in batches of a given size?
"After many such questions, all of which were answered in the negative, Fisher concluded that it would be pointless to pursue the matter further, since there was not the remotest reason being offered for why reading the entries from top to bottom was in any way different from reading them in any other order. Even if the supposed connection between the top of the page and high sugar withstood statistical scrutiny, it would have to be treated as meaningless.”