Exit polling has been one of the top ways to analyze election results. They are used to determine how key segments of voters cast their ballot in a given election. They can help give a rough idea of the final margin. Election observers in foreign countries with questionable reliability of election results use exit polls as a method to counteract fraud. Exit polls are not perfect tools, particularly on Election Day. In fact, the first exit poll releases can be misleading and even downright harmful to gauging what is going on.
Traditionally, the main problem with exit polls on Election Day has been that they are unweighted. When the results are compiled from all of the surveys around the country, human error and certain response biases make the raw results less than perfect. The surveyors who are hired by the firms running the exit polls can make mistakes, they can fail to show up for work, and they may not follow all of the procedures. Also, certain groups of people are more likely to respond to surveyors and this can distort the results.
The most notorious instance of misleading exit polls was in 2004. The unweighted exit polls were fantastic for John Kerry. These early returns had Kerry defeating George W. Bush by double digit margins in key states like Ohio, Pennsylvania, and Michigan. Democratic figures that had access to the results were ecstatic while Republicans who knew what was occurring, like Karl Rove, were pointing out how inaccurate these numbers were. Democratic confidence was unfounded, as the original set of exit polls were way off both the final polling before the election, and the final result, which gave Bush a three point popular vote victory.
Even this year, unweighted exit polls proved to be very inaccurate during the Wisconsin recall election. The polls going into the June recall generally showed Scott Walker with a mid-single digits lead over Tom Barrett. The first exit poll only showed Walker up one point, giving great hope to Democratic partisans who were previously resigned to defeat and great angst to Republicans who thought a Walker victory was a done deal. But as the results started coming in, they were indicative of a solid Walker victory. It soon became apparent that the actual results were similar to the polls, and Walker ended up winning by seven points.
In addition to these older problems, changes to the exit polling process this year could increase the errors to the early exit polls. This year, the consortium of media sources that handle the exit polls are switching from reporting the exit poll results as a percentage of precincts tabulated to a percentage of the expected vote. This change makes sense when considering the increase in early voting, which makes reporting by precinct somewhat inaccurate. Even if a change is necessary, the guess at the “expected vote” is going to be tough to get right and that could also throw off election night reporting.
Another thing that could make election night full of misinformation is that this year nineteen states will not be reporting exit poll data beyond a simple mention of who someone voted for. Every contested state will still have full exit polling, but the absence of nineteen states will require a rebalancing to ensure that subsamples of electoral groups are properly represented. This could pose problems for interpreting results on the night of the election and will be even worse for breaking down the results later.
Exit polls once properly weighted are quite valuable tools for looking at an election. However, the inevitable early election eve leaks of exit poll results that you will see on Twitter are nearly useless and could actually be false information. The best way to understand what is going on Election Day is to watch hard returns come in, precinct by precinct.
Chris Palko works as an assistant media analyst at Smart Media Group, a Republican political media buying agency in Alexandria, Va. He is a graduate of American University and George Washington University’s Graduate School of Political Management.
