Statistics Department Seminar Series: Howard Wainer, Distinguished Research Scientist, National Board of Medical Examiners

Abstract

Too often in science, when we discuss studies, “bigger is better” is taken as gospel. But for exploration, big studies are often too time consuming and cumbersome, whereas small studies are easy and flexible. And, if well designed, can provide as much insight into the phenomena of interest as their bloated cousins. Moreover, in a small study we can quickly pivot to explore any interesting, unexpected phenomena that appear. In this talk I demonstrate the advantages of small studies through four examples:
-Anscombe’s Quartet
-Morris’ Three Polls
-Gelman’s Nine Voters
-Clauser & Wainer’s Two Examinees

The fourth example is discussed in greatest detail and provides insight into:
-Establishing policies for standard setting,
-Using multiple measures,
-Deciding on retaking policy.
It also introduces the Grabovsky Curve, a new tool for setting standards.