Skip to main content

Demography rules — not so fast!

The science of public opinion polling (and not just the predictions of statistician Nate Silver) was the big winner in the 2012 run for the White House. Scientists should not merely rejoice, but tactfully remind family, friends, and foes alike that this was science in action.

Yet the pundits and political operatives have jumped on the demography bandwagon with much exuberance. Through statistical analysis, historical and contemporary, voting trends were disaggregated into subpopulations and translated into a set of probabilities that were remarkably accurate. 

This says nothing, however, about what underlay those probabilities. Demography is not destiny. To demarcate attributes like gender, race, and ethnicity — independent variables that may contribute to outcomes — is not the same as declaring them alone as decisive. No election is so simply determined. That is the fallacy of demography. 

Citizens don't vote because they have a particular attribute. Rather, they are targeted for some message that, along with a plethora of other variables, may prompt a particular behavior, like voting for one candidate (or party) and not another. 

In other words, certain demographic characteristics correlate with certain voting preferences. But to assert that "categories" of people (a statistical concept not to be confused with "groups," a sociological concept) automatically favor a candidate is to ignore free will, individual differences, the human capacity to compartmentalize, and common sense. 

Science warns us that within a margin of error people will act in predictable ways. Science is not very good, however, at explaining the myriad reasons an outcome occurred. We don't measure enough, or with sufficient precision, to pinpoint why something happened.

Instead, we jump on the most obvious sources of any particular observation and inflate its importance qualitatively if not quantitatively, thereby diminishing the other factors that may be as (or even more) responsible. 

To trumpet demography — or its opposite, what an anthropologist has termed "demographobia" — as the key to understanding the 2012 election results is to discount geography, money, ideology, and a raft of other variables that may correlate with the known characteristics of registered voters. The clever pollsters have figured this out. So they estimate and predict better. But the narrative for the observed outcomes still leaves many mysteries and misleads us to simplistic conclusions of why things turned out as they did.

Science must correct this, too. Celebrate the precision, quash the talk that we now "know" why we were so "right." Other elections are likely to restore our fallibility, if not humility.