With huge samples, adherence to best practices, a strong likely voter model, and an excellent track record, there’s a case that Pew Research is the best public pollster. Although Pew showed Obama with a clear lead among registered voters throughout the spring and summer, Pew found Romney taking a 4 point lead among likely voters after the first presidential debate and showed a tied race last week.
Pew’s depiction of a tied or perhaps even Romney-leaning national race was perhaps the best argument in the case to doubt the state polls. Many state pollsters do not employ the most rigorous approaches, so the dissent of Pew and the other national pollsters was sufficient to cast an extra layer of doubt on Obama’s advantage in the state polls.
This afternoon, Pew Research released their final pre-election survey, showing Obama ahead by 3 points, 48-45 or 50-47 after allocating undecided voters. Pew’s likely voter screen and efforts to allocate undecided voters have been quite successful over the last two presidential elections, correctly showing Bush and Obama with 51-48 and 52-46 victories, respectively.
The survey assumes that 58 percent of the voting eligible population will cast a ballot in 2012, down slightly from 60 and 61 percent in 2004 and 2008. According to Pew, turnout will advantage Romney, who possesses more engaged voters who are more likely to vote. Much of the decline in turnout is among young voters, who are projected to decline from 15 to 13 percent of the vote, compared to Pew’s final survey in 2008. As a result, Pew finds a four point gap between Obama's 7 point lead among registered voters and his 3 point lead among those likely to participate on Tuesday. Despite an older electorate and a decline in turnout among young voters, Pew projects the electorate to remain as diverse as it was four years ago, with the white share of likely voters actually declining one point from 75 to 74 percent of the electorate. As mentioned earlier, demographic changes enable the electorate to remain as diverse as it was four years ago, even if turnout rates decline.
Pew judges that the president holds 39 percent of white voters, enough to win reelection if minorities again represent 26 percent of the electorate and offer 80 percent of their support to the president. Although Pew shows Hispanics again representing 7 percent of the electorate, they support Obama by a 66-27 margin, a wider margin than Obama’s 67-32 win in 2008. Obama’s gains over the last few weeks is primarily due to resurgence in the blue states. Obama now leads by 21 points in the Northeast and 8 points in the West, up from last week’s 9 point lead in the Northeast and a tied race in the West.
It would be one thing if the Pew Research poll was an outlier, but the national polls have shifted modestly but clearly in Obama’s direction since the final presidential debate. Today, NBC/WSJ--another highly regarded pollster--showed Obama ahead by one point, 48-47. On average, Obama now leads by 1 point since the final debate, with just 2 older surveys out of the 18 polls showing Romney ahead by any margin at all.
Regardless of whether a 1 point advantage in the national polls qualifies as a meaningful lead for the president, it is at the very least sufficient to neutralize the argument that there is an unusual discrepancy between the state and national polls. Depending on the estimate, the state polls show Obama ahead by anywhere from 1.5 to 2 points. With the national and state polls in somewhat better alignment, one of the primary causes for doubting the state polls has become less persuasive.