Confidence Intervals
Jerzy Neyman (1894-1981) developed a method for
hypothesis
tests with
Egon Pearson (1885-1980) and
William Gosset (1876-1937). Their method used a sample
statistic to determine whether a specified value was plausible for
a parameter. The appendix of Neyman's work On the Two Different
Aspects of the Representative Method
, presented in 1934,
discussed sample survey analysis. In it, he explained a way to
determine the precision of a statistic as an estimate of a
parameter, using a collection of plausible values. Many of his
contemporaries understood his straightforward method, which uses
probability distributions. However, the meaning of probability in
the context of the interval was ambiguous since a parameter is
fixed, meaning the probability that a parameter is in an interval
is either $100$ percent or $0$ percent. Neyman tried to avoid
confusion by avoiding the word probability
in its name
altogether, calling the construct a confidence interval
.