Confused by all the polls? Pollsters explain the variation
As recently as two weeks ago, political prognosticators were looking at polls and declaring, "If Romney would have to pull off a miracle to close the gap in national polling, he has no shot at matching the president in the electoral college."
This week, with Gallup polling showing Mitt Romney with a strong national lead over President Obama, some conservatives are pointing to the Republican candidate's likely victory.
The latest polls out of Wisconsin and Iowa, meanwhile, show Mr. Obama with a modest but clear lead in the two battleground states.
For what it's worth, a CBS News/YouGov simulation this week showed that, had the election been held at that point in time, Mr. Obama would have held a slight Electoral College edge. This simulation, however -- as with every survey -- represents no more than a snapshot in time.
"Poll results reflect the mindset of voters during the time the poll was conducted," CBS News director of surveys Sarah Dutton explained. "Those voters can change their minds later, or decide not to vote, or not be able to get to the polls on Election Day to vote."
Even among polls conducted over the same span of time, different methodologies -- whether it's differences in the way survey samples are chosen or differences in the way the surveys are conducted -- can lead to different results.
"It's certainly important to understand whether a poll includes calls to cell phones, whether it's done online, whether live interviewers ask the questions, when the poll was conducted - those are all part of a poll's methodology," Dutton said. "Consumers of polls should be aware that not all polls are conducted the same way, and some utilize a more rigorous methodology than others."
Furthermore, some polls track registered voters while others track likely voters -- and pollsters may differ in how they determine who is a "likely voter." Some polls look at people who say they will definitely vote, Dutton said, while others identify likely voters based on their responses to a number of questions, such as past vote history, whether they know where their voting place is, and whether they will definitely vote in this election.
Gallup's tracking poll provides a daily, seven-day rolling average of both likely voters and registered voters. This week's figures show a noticeable difference between the two groups: Today, Romney led Mr. Obama by six points among likely voters, 51 to 45 percent. His lead among registered voters, however, was just one point, 48 percent to 47 percent. Gallup editor-in-chief Frank Newport told the Washington Post that Gallup's likely voter model "takes into account changes in the response to questions about how closely they're following and how enthusiastic they are." The relatively large difference between the likely voter numbers and registered voter numbers may be because the likely voter model is, as Newport said, "very sensitive to changes in enthusiasm."
Polling experts discourage using any single poll to predict the outcome of an election. That said, looking at poll averages, and multiple polls over time, can provide insights into broader overall trends in the election. Quinnipiac University Polling Institute assistant director Peter Brown emphasized that it's only worth comparing trends from the same polling outfit -- there would be no use, for instance, in comparing a Gallup poll with a CBS News poll because of the different methodologies.
This month, there have been some clear trends.
"Clearly, two weeks and three days ago before the first debate, there were an awful lot of people who, according to the polls, thought Mr. Romney's chance of victory were small," Brown said. "Obviously, the polls have changed. Overall, Mr. Romney's certainly doing better than before the first debate."
