KRISTEN SOLTIS ANDERSON
When the clock struck nine on the East Coast Tuesday night, the polls closed in Wisconsin. Moments later, the news spread across Twitter that exit polls showed the race neck and neck, 50–50. Gradually, results came in that debunked this result. So what happened? Are exit polls junk?
The answer is no, exit polling isn’t junk, but it is a lousy way to estimate a final vote margin before any votes are in. Exit pollsters collect data from a carefully selected set of precincts, and then as actual returns come in, the data are weighted to match those results.
Here’s a rough example: Let’s say I’m an interviewer doing exit polling at a precinct, and during the day I interview six Walker voters and four Barrett voters. Alone, that’s not enough interviews to say with any certainty that my precinct is going 60–40 for Walker, but the “raw” data will look like that’s the result. As the night goes on, it turns out my precinct was instead split 53–47. The data from my precinct will be weighted to that result.
What the networks are doing when they go to make a call is they’re watching how the data adjust when precincts get weighted back to real returns. When enough precincts are in for which the exit-poll estimates are sound, they can call. The next morning when we wake up to read the news, the numbers being tossed around are based on the final weighting and are a great way to understand subgroups in the electorate.
The raw data are lousy for telling us instantly who the winner is in an election. But as the data evolve, they grow in accuracy and usefulness for evaluating “what happened.”
— Kristen Soltis Anderson is pollster at the Winston Group.
Though exit polls indicated a dead heat, the networks picked Governor Scott Walker to win by 10 p.m. Eastern time. The result is a body blow to government unions at a time when they have become more consequential than the remnants of the once-massive private-sector-union bloc. While the election represents a major setback for public unions (whose perks and benefits are low-hanging fruit in a time of budget crises), it demonstrates a healthy civic impulse and clear-sightedness on the part of voters. The decision to retain Governor Walker demonstrates an ability on the part of voters to discern the difference between private unions (which can be a perfectly legitimate part of a free-market negotiating process) and public unions (which are a different sort of creature). Public unions create a class of voters (government-union members) who are able to promise money and support to the people making decisions about their pay packages and work conditions. It is anything but an arm’s length transaction made in the interest of all citizens. The process highlights the way narrow interests can exploit apathy on the part of the public to gain concentrated benefits. The situation is made worse by the fact that many of the benefits (such as pension promises) don’t come due until well in the future, when feckless decision-makers have long since left office and need not face the music. The public union is a Tocquevillian nightmare.
There is little incentive (other than fiscal responsibility, an exceedingly rare virtue) for executives like Scott Walker to rein them in. But he did it anyway. Walker bet his term on drawing attention to the way public unions act against the public interest. He paid a price in having to fight a determined recall. But last night, he was vindicated. The people of Wisconsin have decided to reject the kind of government that rewards organized public employees with perks and promises well beyond those available to the vast majority of Americans working in the private sector. In so doing, they have taken an important step in equalizing the playing field between those holding government jobs and those working in the private economy that supports the government jobs.
— Hunter Baker, J.D., Ph.D., is an associate professor of political science at Union University. He is the author of The End of Secularism and Political Thought: A Student’s Guide, which comes out next month.