Myth One: “No president has been reelected with an unemployment rate higher than ... .” This hoary notion never made much sense. Put the unemployment rate and the incumbent’s vote percentage on a graph and you can immediately see that the two bear almost no relationship to each other.
Two main reasons explain this. One is that the unemployment rate sometimes goes up when times are getting better and down when things get worse. As an economy improves, more people start looking for work, sometimes leading to a temporary increase in the unemployment rate; conversely, when things get worse and discouraged workers drop out of the market, the rate can drop.
The other reason is that voters respond much more to the direction of change than they do to the status quo. In other words, whether voters perceive the economy as getting better or worse matters more than the economy’s absolute level. And the number of jobs created in the economy provides a much better measure than the unemployment rate.
Myth Two: “In an equally divided country, self-identified Democrats and Republicans should be roughly equal in polls.” This misconception launched a thousand debates over whether published polls were “skewed” in favor of Democrats. It also caused some Republican pollsters to fool themselves -- and presumably their clients -- by weighting poll results to match a preconceived idea of what the balance between the two parties “should” be.
In the end, the election returns proved them wrong. As exit polls showed, voters who identified themselves as Democrats outnumbered those who identified as Republicans 38% to 32% this year, about the same margin that the major pre-election polls showed.
Asking voters which party they identify with is one of the most important parts of any poll. Pollsters use party identification to analyze other data, looking at how self-identified Democrats and Republicans view issues differently. And, of course, party identification powerfully predicts voting behavior. Republicans who weighted their polls to narrow the partisan gap ended up producing results that made Mitt Romney and other GOP candidates appear to be in better shape than they were.
Of course, those pollsters didn’t set out to fool themselves. They were trying to solve a riddle: If Democrats and Republicans are close to parity in the results -- President Obama ended up beating Romney by about 2 percentage points in the popular vote -- how can Democrats have a six-point edge in party identification?
The answer is actually pretty simple. Over the last decade, a significant number of people who vote for Republicans have stopped identifying themselves as Republicans. That includes a lot of ardent conservatives. Instead of calling themselves Republicans, those voters now identify as independents. No similar movement has taken place among Democrats.
As a result, the group of voters who identify as independent no longer sits at the center of the political spectrum. Instead, independents heavily tilt toward the Republicans. Polls that take those independents into account and then weight their results to show more Republicans are, in effect, double counting.
The other result of that shift: A Democrat no longer has to win independents to gain a majority. Indeed, Obama lost self-identified independents by five points, 45% to 50%, according to the exit polls. He beat Romney 56% to 41% among voters who identified themselves as ideologically “moderate.”