In the context of estimation of an unknown parameter, a statistic whose value provides information incidental to the estimation process. The most usual case is where the sample size is not fixed. For example, we might wish to know the proportion of a sweet pea mixture that has red flowers. We plant N seeds, but our estimate will be based on the ancillary statistic n, the number of seeds that germinate and produce flowers. The term was introduced by Sir Ronald Fisher in 1925.
Subjects: Probability and Statistics.