Anyone who follows the 2012 presidential election polls likely noticed a trend: The numbers don’t make much sense. As it turns out, inconsistent polls during Election Season are pretty normal.
As Geoff Skelley from the University of Virginia’s Center for Politics points out, “Each pollster has slightly different methodologies and, sometimes, political purposes”—which can leave some confusing statistics.
For the most part, polls seem to be founded on sound research and methodology, but it is important to understand where the numbers are coming from. There are laws and technological barriers which often limit the scope of the research conducted. Without access to information or the power to study the effects of methodology, it can be incredibly difficult to glean useful information from the polls we see each week.
The New York Times recently explored the difference between polls that include people who only use cell phones and those who use house phones. According to the article one-third of American households rely on their mobile phones and do not have landlines. If a poll neglects to include cell phone users then they have effectively cut out one-third of the population, likely having an effect on the poll’s validity. It also seems that “potential voters who rely on cellphones belong to more Democratic-leaning demographic groups than those that do not , and there is reasonably strong empirical evidence that the failure to include them in polls can bias the results against Democrats.”
Take the numbers for three swing states in 2012. Two separate Fox polls put Obama +5 in Florida and +7 in Ohio while a CBS/New York Times/Quinnipack poll put Obama +4 in Virginia. According to these polls, Obama leads in these three swing states. The cellphone-inclusive polls, on the other hand, give Obama a 68 percent chance to win Florida, a 79 percent chance in Ohio, and an 80 percent chance to win Virginia; all considerably higher numbers than the aforementioned polls suggest.
At first the difference may seem insignificant, but even though Obama leads in both cases, a sizable lead and a small lead are two very different things, especially when we consider that these numbers reflect President Obama’s post-convention bounce.
Even after the methodology behind the numbers is understood, it can still be difficult to make sense of the polls. It turns out that timing, varying laws regarding polling from state to state, and political agendas of pollers also strongly affect the poll’s results.
For some extra help understanding the numbers and for some advice about how to read political polls, I spoke with Political Analyst Geoff Skelly from the University of Virginia Center for Politics .
The Numbers we’re seeing this week are perhaps a little more confusing than usual. Is it possible that we are just in a weird period before Obama’s conventions bounce begins to wane or does it look like his boost may be permanent?
Obama certainly got a convention bounce, but while the bounce has come down a little bit (as all bounces must by definition), the growth in the president’s lead does seem to have hardened a bit. At the start of the Democratic convention, Obama and Romney were essentially tied in the poll average. Now Obama holds a nearly four-point lead. If Obama is still holding this kind of a lead a week or two from now, it gets harder and harder to see how Romney gets the kind of surge he needs to get ahead with just a month to go. But this is still a close election, one that will mostly be decided by who shows up to vote. If there’s a similar (or higher) level of minority participation compared to 2008, the president probably has a good shot at winning. But if that falls off and white support falls below 40% for Obama, Romney has a shot of winning this thing.
What should we make of polls that have President Obama ahead by 14 points in Wisconsin, anywhere from 2 to 7 in Virginia and behind by 2 in New Hampshire?
You have to make sure to look at all the recent polls in one state. Of the polls taken in New Hampshire since early September, President Obama leads by an average of 1%. So what that tells us is that New Hampshire is still a battleground state. There were some polls during the summer that seemed to show Obama building a lead in the Granite State but now it’s very much neck-and-neck. And to a certain extent, that matches what we expected: we always thought Romney had a decent shot at winning states like Florida and New Hampshire.
As for Wisconsin, polling there has suddenly shown a significant swing toward Obama. This is likely a combination of a few different factors. For one thing, the boost Romney got out of picking Paul Ryan as his running mate was bound to fall off some and apparently has. But perhaps more importantly, the Wisconsin numbers may tell us that a number of Democrats who were previously unengaged have now become engaged in what’s going on. With most polls moving from registered voter polls to likely voter polls, this is meaningful – more Democrats are qualifying as “likely voters”, which somewhat dispels the notion that Republicans have a significant lead in enthusiasm for November. Still, some of these polls are probably a bit too Dem-friendly. Obama is probably up in the Badger State but not by double digits.
The Old Dominion is still very close. The difference between a three-point lead and a seven-point lead is huge so it’s better to be cautious and say that Obama probably has a small lead in Virginia.
The next 41 days promise to be full of new polls and numbers. We will see some contradictory numbers, but for the most part, as Election Day nears, the numbers typically stabilize making them a bit easier to understand.