After showing a statistical tie for months, the presidential polls have taken a dramatic turn — at least for a moment. As I write this, CNN has Obama ahead by 9 and Rasmussen has Romney ahead by 4, with both polls taken during the exact same time period. What is going on?
I have written about the methodology of polling before, pointing out how raw data is massaged before numbers are released to the public. As an overly simple example, suppose the raw data showed that 56 percent of poll respondents were women in a state where only 51 percent of the voters were women. The pollster would randomly throw out responses from women until the ratios matched, thus "weighting" the final report. Obviously, CNN and Rasmussen weighted their samples in different ways.
Also, Rasmussen applied an additional element to its sample. While CNN polled RVs (registered voters) Rasmussen went for LVs (likely voters) by asking questions designed to screen out those who probably will not vote even though they are registered to do so. Historically, RV polls tend to tilt more towards the Democrats than LV polls, but a discrepancy this large — 13 points — is very unusual.
Which brings another set of polls into the picture — those that attempt to measure enthusiasm level.
These are taken less often, but throughout the campaign, they have reported a higher level of interest and enthusiasm among Republicans than Democrats this year. Many Democratic observers did not expect this, early on. They smugly told me, as Romney was being pummeled during the primaries, that he could never attract hard core conservatives to his banner; turned off by him, they would simply stay home. I countered by saying that Obama's name on the ballot would be all the encouragement conservatives needed to turn out. So far, polls seeking to test enthusiasm say that that is so.
If it is true that Obama has a bigger enthusiasm problem on the left than Romney has on the right, that could explain part of the current gap in the polls. Registered voters who preferred Obama in the CNN sample may have been filtered out by the "likely voter" screen in the Rasmussen sample, making it the more accurate measure of turnout. Team Obama's actions suggest that they fear that this is true: their massive personal attacks on Romney seem to be less about swaying wavering independents than they are about firing up lukewarm Democrats.
Polls are not the only method of measuring voter sentiment and predicting voter behavior. Information gathered from focus groups can also serve as an important component of political analysis, and such information often surprises political experts. In a recent news report, journalists who were steeped in all of the daily twists, turns and details of the campaign battle found that many focus group participants had not paid attention to any of it. All they knew about Romney were three things: he was a Mormon, he was rich and he was not Obama.1 comment on this story
So, if you believe the majority of the current polls, you can be optimistic about Obama. His lead in the projected electoral vote has widened and he now stands at 247, just 23 shy of victory, in spite of bad economic news. However, if you believe that the enthusiasm gap can be decisive — in 2008, it tilted strongly towards Obama, even as polls were showing that race as even — then you can be optimistic about Romney. Elections are determined by those who vote in November, not those who respond to polls in August.
Robert Bennett, former U.S. Senator from Utah, is a part-time teacher, researcher and lecturer at the University of Utah's Hinckley Institute of Politics.