The Variance of the Product of Two Independent Variables and Its Application to an Investigation Based on Sample Data

(e) In the experiments based on the 1% sample of the 1951 census data, which were described in §X of The Components of Mortality,† it was found that the relative sizes of k (which was of course 100), N and the various values of u were such that the third and fourth terms inside the bracket in expression (e) were negligible. It was therefore possible to use the approximation Variance of expected deaths = expected deaths x (I + 99mx), (f> which may be generally used in a 1% sample investigation solong as u and N are sufficiently large to justify ignoring the two terms in question.