Statistics, these days, play some role in nearly everything–including democratic politics (the contest through voting over what policies to adopt or at least who gets to decide what policies to adopt) and thus democracy itself (the correspondence between the distribution of preferences or underlying interests, on the one hand, and the policies adopted, on the other). Vote counts, whether in legislatures or elections, are statistics. Polling data are also statistical, as are the associated apparatuses of sampling and inference. Policy-relevant statistics like unemployment rates, health care costs per capita, and rates of illegal immigration may affect opinions and votes. Statistical models can help explain why voters and politicians behave as they do.

For understanding, statistics are a blessing. For democracy, however, the blessing is dilute. Many people have trouble understanding statistics and giving them their due. We tend to overweigh the most recent or vivid observations. A record snow storm belies global warming. Further, the value of statistics depends on their truth value. In politics, statistics may be invented, misstated, or misleadingly constructed or cited. Under a sufficiently generous definition of “small business,” a tax on the very wealthy will indeed fall on a high proportion of “small businesses.”

The blessing is also mixed–hence my subtitle, “friend and foe.” Even accurate, accurately understood statistics can distract or seduce as well as enlighten. Media polls, telling the public what it thinks and how it is likely to vote, fuel “bandwagons.” As the primary means of “calling the race” or keeping score,” they also provide the basis of “horse race” or “sporting event” coverage.

This should give the flavor of my remarks. I may add something about possible innovations, including Deliberative Polling, to provide new ways in which statistics may benefit democracy.

Keywords: Statistics; Democracy; Elections; Polling

Biography: Professor Luskin is Associate Professor of Government and Director of the Center for Deliberative Opinion Research at the University of Texas at Austin and Research Advisor at the Center for Deliberative Democracy at Stanford University. He has taught in the ICPSR Summer Program at the University of Michigan and the ECPR Summer School at the University of Essex and taught or worked in a research capacity at Princeton University, Stanford University, Sciences Po-Paris, the Université de Paris I, and the Center for Advanced Study in the Behavioral Sciences at Stanford University, among other places. He has served on the editorial boards of Political Analysis and the American Political Science Review. His research interests center on mass politics, notably including deliberation and political knowledge, and on statistical models and methods. He helped design Deliberative Polling, with which he has been involved from its inception.