The Capriciousness of Small Numbers
This paper presents a graph for use in assessing confidence
intervals of quality control findings based on counting. A
number of practical examples ar used to demonstrate the
low informative value of small counts and to show the extent
to which tills can be improved through increasing the size
of the sample examined.
Th graph is applicable not only to counts obtained in
connection with quality control in the traditional sense but
generally to all types of statements based on counting, for
example to the resuits of opinion polls.