ABSTRACT
The supple rigor of statistical analysis is often held up these days as a scientific alternative to speculation, interpretation, and debate. Although the social and behavioral sciences do indeed provide appropriate material for quantitative investigation, their effectiveness even as tools of expert analysis generally requires an engagement with ordinary human concerns, where mathematical rigor is typically out of reach. Indeed, this wider perspective is likely needed not just for political reasons, as gestures of respect for pluralism and democratic governance, but also because the results of a technical analysis will often be tested and may well be undermined by public reactions. The cooptation of quantified rules, exploiting their vulnerability to unintended consequences, is often more deadly than outright opposition. This chapter is concerned especially with neoliberal systems of incentives, which may be weakened or even collapse in the face of what I call “exploitable ambiguity.” The vulnerabilities of statistics appear in certain characteristic forms that we can examine and sometimes even anticipated. The hazards of ignoring them go beyond their (perhaps) unbecoming hubris. Too narrow a focus on what is “data-driven” and the rejection of alternative forms of expertise is often self-defeating. At the same time, there are important reasons to bring lay citizens into statistical discussions. Effective democracy, in modern times, often requires intelligent attention to statistics, communicating numerical information that matters in an imaginatively accessible form.
