Abstract
SUMMARY: The use of Fieller’s theorem for determining confidence limits for a ratio α of bivariate normal means is a standard procedure in biological assay problems (Fieller, 1940, 1944; Finney, 1952, Chapter 2). Occasionally situations arise wherein several successive samples of correlated measurements are involved, and then particular interest attaches to the corresponding problem of confidence limits for the common ratio of multivariate means based on successive samples. For example, in clinical trials on experimental subjects concerning the effectiveness of a certain therapeutic substance as compared with a control or placebo, correlated clinical measurements may be obtained at a number of participating research laboratories. An approximate method is also discussed by Clarke (1956) for an example in microbiological assay.

This publication has 3 references indexed in Scilit: