In Bayesian statistics, a credible interval (or Bayesian confidence interval) is an interval in the domain of a posterior probability distribution used for interval estimation[1]. The generalisation to multivariate problems is the credible region. Credible intervals are analogous to confidence intervals in frequentist statistics[2].
For example, in an experiment that determines the uncertainty distribution of parameter , if the probability that lies between 35 and 45 is 90%, then is a 90% credible interval.
Credible intervals are not unique on a posterior distribution. Methods for defining a suitable credible interval include:
It is possible to frame the choice of a credible interval within decision theory and, in that context, an optimal interval will always be a highest probability density set.[3]
A frequentist 90% confidence interval of 35–45 means that with a large number of repeated samples, 90% of the calculated confidence intervals would include the true value of the parameter. The probability that the parameter is inside the given interval (say, 35–45) is either 0 or 1 (the non-random unknown parameter is either there or not). In frequentist terms, the parameter is fixed (cannot be considered to have a distribution of possible values) and the confidence interval is random (as it depends on the random sample). Antelman (1997, p. 375) summarizes a confidence interval as "... one interval generated by a procedure that will give correct intervals 95 % [resp. 90 %] of the time". [4]
In general, Bayesian credible intervals do not coincide with frequentist confidence intervals for two reasons:
For the case of a single parameter and data that can be summarised in a single sufficient statistic, it can be shown that the credible interval and the confidence interval will coincide if the unknown parameter is a location parameter (i.e. the forward probability function has the form ), with a prior that is a uniform flat distribution;[5] and also if the unknown parameter is a scale parameter (i.e. the forward probability function has the form ), with a Jeffreys' prior [5] — the latter following because taking the logarithm of such a scale parameter turns it into a location parameter with a uniform distribution. But these are distinctly special (albeit important) cases; in general no such equivalence can be made.
|