Depending on which presidential campaign poll you look at, President Obama either has a 9 point lead nationally over Mitt Romney, a 7 point lead, a 4 point lead, a 1 point lead, or its tied, or Romney has a 1 point lead over the President, or a 2 point lead. Huh?
Welcome to campaign polls. It’s not quite an exact science. In fact, some would argue it is as much an art as it is a science. Consider the national polls represented in the Real Clear Politics average, which is updated daily. In their latest polls, Fox News has Obama up by 9, CNN has him up by 7, while Rasmussen and Gallup have Romney leading by 2 points. They can’t all be right. And they’re not.
Part of the reason for the discrepancy is the voter samples they are using to perform their surveys. For example, most media polls derive their samples from lists of all registered voters. While others, notably Rasmussen, derive their samples from lists of “likely” voters. Likely voters are those voters with a demonstrated track record of voting in previous elections, compared to registered voters, which is simply everyone who is registered to vote in a given state.
An identical set of survey questions given to these two different samples will yield different results. On average, that difference ranges from 2 to 4 points, but in some cases can be as much as 5 to 8 points. Likely voters tend to have higher incomes, therefore polls built on likely voter samples will lean Republican (since higher income voters tend to vote Republican). By contrast, polls built on registered voter samples will normally lean Democrat.
So you have two alternate realities based on different data sets and a different set of assumptions. So who is right? It depends completely on turnout – who actually shows up at the polls this Nov. 6. That is a factor that is unknown, because it hasn’t happened yet. (Duh) So pollsters have to guess. That’s not quite fair, it is an educated guess, based on past elections and a mountain of data. But it is still a guess.
It’s very hard to accurately predict who will actually show up on election day, and in what percentages they will show up. Most polling organizations have a mixed record at best. And turnout can be effected by events, sometime late in the campaign – economic news, debate performances, attack ads, etc. Turnout was likely effected in 2000 when the Bush DUI story broke just before election day. Karl Rove and GOP activists very likely altered the 2004 election day turnout in key swing states by pushing for amendments on ballots in those states supporting the defense of marriage act.
I suspect that one of the reasons Mitt Romney selected Paul Ryan as his VP is a hope that it would spur conservative Republican turnout this November in crucial states, such as VA, NC, OH, and FL.
If the election were held today, President Obama would likely win a very close election. But the election is not being held today. If presidential elections were held in August, then Jimmy Carter and George H.W. Bush would both have been two term presidents, and John Kerry would be completing his second term this year.
There is a lot of campaigning yet to be done and the polls will bounce around until October. Only then will we be able to get a good read on who is likely to win in November.
What’s the old saying? Voters really don’t start paying attention in significant numbers until after the World Series anyway.