Interval estimation

Abstract In Chapters 2 and 3 we discussed point estimation at some length. However, as was noted in Chapter 1, a point estimate on its own is of little use—some measure of its precision is also necessary. This leads naturally to the idea of interval estimation, which is of great practical importance. Nevertheless, we devote less space in this book to interval estimation than to either point estimation or hypothesis testing. This is because the theory underlying interval estimation is closely related to that already covered for point estimation and hypothesis testing, and the theory is most conveniently developed in these latter contexts. In fact, there is a close relationship between interval estimation and hypothesis testing, so that many of the ideas of hypothesis testing carry over directly to interval estimation; we show this later for some of the more useful ideas. The approach used in most of this chapter is known as the frequentist approach.