Sequential Bernoulli Sampling Plans Re-Examined

Abstract

Connections between the so called efficient and inefficient Bernoulli sampling plans are explored, via the concept of sufficiency of statistical experiments. From this point of view, many inefficient plans turn out to be quite useful. We also take up the problem of comparison of two arbitrary plans and provide a variety of results. The paper ends with a few conjectures on some naturally raised questions, yet to be satisfactorily settled.