INTRODUCTION

This chapter offers a philosophical and historical perspective on the development of the concept of “variability” in the last few decades. The goal is not to provide a treatment of modern mathematical techniques for the analysis of data indicating variability. The recent literature contains excellent works that document analytical tools and are eminently usable by risk analysis practitioners (Cullen and Frey, 1999; Thompson, 1999). Rather, the object is to reflect on the significance and prospects for quantitative assessment of “variability” as an intellectual innovation that has emerged in part from the interdisciplinary fusion of ideas, techniques, and social needs for information for decision making that is the discipline of risk analysis.

Briefly, the nub of the innovation is distinguishing real variation among things from measurement errors, other sources of uncertainties, and stochastic fluctuations (such as the numbers of cosmic rays arriving at a detector in a specified interval). Where uncertainties reflect the imperfections in available information about the world (and are often seen as an annoying fog that obscures investigators' ability to demonstrate differences among groups, but with no real consequence or interest in itself; Hattis, 1996), real quantitative variation among things/people has real implications for differential behavior among the “things” being studied. Such differences can, for example, take the forms of (a) the relative risks to those things/people and (b) the relative desirability of devoting resources to some things rather than to others (if, for example, the things are categories such as industries that could be the subjects of safety inspections or other resource-consuming activities) (Hattis and Goble, 1994).