All the polls showed that Barack Obama would win. He was comfortably ahead of his chief rival. The RealClearPolitics average had him up by 8.3 percentage points. The result was foreordained. Only his margin of victory was in dispute.
Except that Obama didn’t win. He lost January’s New Hampshire primary by 2.6 percentage points.
Nobody should expect a repeat of that pollster’s fiasco tonight. Estimating turnout is more difficult in a primary than in a general election. Yet it is also true that even presidential polls are not oracles. Some polls get it wrong; and sometimes, polls get it really wrong.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
ADVERTISEMENT
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Just scan the various surveys on
RealClearPolitics or Pollster.com. Their results won’t all be correct. The ABC News/
Washington Post daily tracking poll has Obama up by 11 percentage points. Rasmussen Reports has him up by six. Diageo/Hotline has him up by five.
By contrast, on the eve of Election Day four years ago, the pre-election polls’ findings were largely similar. CBS News had Bush ahead by 2. 1 percentage points; Harris had Bush up by 1; and Pew had Bush up by 3. Their results were also largely accurate. Bush ended up winning by 2.5 percentage points. (The exit polls, which consistently overstated Kerry’s support, were a different story.)
The polls will likely predict the winner this year, too. Obama is by all accounts the favorite. But, and this is essential to keep in mind, the polls might well be wrong. Steven Schier, a political scientist at Carleton College, told the
Washington Post that he estimates that McCain has a 25-percent chance of winning. (That sounds about right to me).
If Obama loses, how could the polls have botched the job again? The answer will have nothing to do, as it did in January, with the small window of time (five days) that pollsters had to gauge voter sentiment. Rather, it will have to do with pollsters’ failure to survey two groups accurately.
The first group is working-class whites. Defined as those with less than a four-year college degree, this bloc represents the plurality of America’s voters.
Obama’s troubles with this group during the primary season were well publicized; he was blown out in the Kentucky and West Virginia primaries, two states with heavily blue-collar populations, and lost by significant margins in the Ohio and Pennsylvania primaries. To be sure, Obama’s relationship with the group has improved; an economy on the verge of recession will do that for a Democratic nominee. Yet his level of support remains fairly weak. And it is almost certain to be weaker than polls indicate.
The reason: Joe Sixpack (or Joe the Plumber) is the voter least likely to appear in a pollster’s survey. He can’t be reached or doesn’t want to be reached. As Jon Cohen — the
Washington Post’s polling director — has noted, poorer, less educated whites who tend to hold somewhat less positive views toward African Americans are also harder to get on the phone than those who have higher incomes and more formal education.
This phenomenon surely overstates Obama’s lead in the polls. Al Gore lost the non-college-educated white by 17 percentage points; John Kerry lost it by 23 percentage points. And arguably both Gore, he of the People against the Powerful slogan, and Kerry, the decorated Vietnam veteran, were better tribunes for the white-working class than the former law school professor.
The second group whose votes are likely to be misstated is young voters. For Obama, those who are 18 to 29 years old are the opposite of blue-collar whites: he has attracted a lot of their support, most famously in the Democratic primaries. According to Gallup, he enjoys a 60-to-37 lead with this group of voters.
