In statistics, the reference class problem is the problem of deciding what class to use when calculating the probability applicable to a particular case. For example, to estimate the probability of an aircraft crashing, one might use the frequency of crashes of all aircraft, of this make of aircraft, of aircraft flown by this company in last ten years, etc. Any case is a member of very many classes, in which the frequency of the attribute of interest (such as crashing) differs, and reference class problem discusses which is the most appropriate to use.
More formally, many arguments in statistics take the form of a statistical syllogism:
- X proportion of F are G
- I is an F
- I is a G
F is called the "reference class" and G is the "attribute class" and I is the individual object. How is one to choose an appropriate class F?
In Bayesian statistics, the problem arises at that of deciding on a prior probability for the outcome in question (or when considering multiple outcomes, a prior probability distribution).
Read more about Reference Class Problem: History, Legal Applications
Famous quotes containing the words reference, class and/or problem:
“I think, for the rest of my life, I shall refrain from looking up things. It is the most ravenous time-snatcher I know. You pull one book from the shelf, which carries a hint or a reference that sends you posthaste to another book, and that to successive others. It is incredible, the number of books you hopefully open and disappointedly close, only to take down another with the same result.”
—Carolyn Wells (1862–1942)
“All this class of pleasures inspires me with the same nausea as I feel at the sight of rich plum-cake or sweetmeats; I prefer the driest bread of common life.”
—Sydney Smith (1771–1845)
“Any solution to a problem changes the problem.”
—R.W. (Richard William)