Night of the Poll Junkies

The obsession is back – and it's spreading. In the last week before the election, the polling has become more intense and so has the attention polls are getting from political junkies. Traffic at sites like Pollster.com, Real Clear Politics and FiveThirtyEight.com is skyrocketing, apparently not just with political operatives and reporters but with all kinds of people. And the eternal debate about whether the polls are accurate is raging, as most surveys show Barack Obama ahead of John McCain.

Surveys are eagerly consumed and bitterly criticized, often by the same people, frequently in the same sentence. The difficulties polls face in predicting elections has been dissected every possible way. If you've been paying attention in the last few weeks, you've probably learned a lot about polling concepts like the Bradley effect, the problems of surveying cell phones, identifying "likely voters" and margins of error.

Most of the criticisms of polling, however, are rooted in one basic assumption, which can be summed up like this: the most important function of surveys is to forecast elections. If they can't do that, they're not good for anything.

This is a false premise, sold to us by political partisans, the media, and people who essentially see politics as an extension of sports. They're the major consumers of survey data in our society, and to them, it's all about who wins and who loses. Surveys are just a way of keeping score. This is why so much cable news commentary is indistinguishable from color commentary at the World Series (except much, much angrier).

Just for the record, let's acknowledge that election polling is one of the most challenging tasks in survey research. Not only are pollsters trying to measure attitudes that may be shifting rapidly, but they also have to try and sort out those who vote from those who don’t, and those who say they'll vote from those who actually show up. And when it comes to surveys, being "right" means being accurate within the margin of error, usually three or four percentage points either way. In looking at a close election, it's entirely possible for a poll to be statistically "accurate" but still show the wrong person in the lead.

But the most important question surveys can answer isn't who. It's why.

Surveys are at their best when they probe people's values and attitudes, their personal experiences and perceptions of what's happening around them. They provide an indispensible and highly accurate window into what's worrying people and what matters to them. Support for a candidate can change quickly, but values don't. These values and concerns not only impact how people vote, they form the public's basis for judging the next president's policies once he's in office.

Good survey organizations do ask those questions as part of their election work (here are some great examples on foreign policy, the economy, one-party government and overall issue trends). But these questions just don't get the media attention that they deserve.

Right now surveys are showing record levels of public dissatisfaction with the direction of the country, with a staggering eight in ten saying the country's "on the wrong track." Public discontent has never been this high since modern polling began. Multiple polls also show that the The country is facing enormous challenges over the next four years. The economy is in crisis, the federal government's finances are in terrible shape, the country is fighting two wars, and crucial decisions about problems like health care and climate change can't be put off much longer.

None of those problems can be solved unless the next administration's policies are grounded in the public's sense of what's doable and what's not, what's right and what's wrong. I'm not advocating the next president "follow the polls." But I am saying that the public's voice should be just as important to governing as it is to campaigning, and that leaders will lead more effectively if they listen to the insights surveys can provide.

As the political class focuses on the limitations of polls in predicting the horserace, the powerful data surveys can provide about values and policies are at risk of being overshadowed. And if there was ever a moment for insight, rather than mere scorekeeping, this is it.