Ann Selzer, as you may know, is among the best pollsters in the business. In fact, we have her firm, Selzer & Co., rated as the single best pollster out of the 32 companies that we evaluated. Polling for the Des Moines Register, she nailed both the Democratic and Republican outcomes in the Iowa caucus, races that other pollsters had a great deal of trouble with.
Selzer’s polls, as you also may know, have tended to have significantly better numbers for Barack Obama than most other agencies. In Indiana, for instance, Selzer has the race at Obama +3, whereas most other polling firms show John McCain with a (small) lead. In Michigan, while several polls have shown Obama with a fairly large lead, Selzer’s poll for the Detroit Free Press pegs the race at Obama +13, the largest Obama lead in any current poll of the state.
Clearly this is not random variance; the Obama “house effect” is highly statistically significant. So I asked Selzer what she might be doing differently from other pollsters.
Selzer thinks that a lot of pollsters may be undercounting the youth vote, and potentially also the black vote. Young voters are becoming harder and harder to reach. They are in the habit of screening their phone calls. More problematically still, a great number of them (roughly 50 percent of voters under 30) rely principally or exclusively on cellphones, which most pollsters (including Selzer) will not call.
Pollsters can attempt to work around this problem by weighting the young voters they are able to reach more heavily; indeed, it is imperative that they make at least some attempt at weighting if they want to produce accurate results. But Selzer says she knows of at least one prominent polling firm — she would not mention them by name — which is not weighting by age groups at all.
Moreover, many of the pollsters that do weight by age group may be doing so — to her mind — in the wrong way. Specifically, they tend to use the 2004 election as a benchmark, when 17 percent were aged 18-29. Selzer uses census bureau data as her benchmark instead; among American adults aged 18 and up, about 22 percent age 18-29. This might not seem like a large difference, but given Obama’s strong performance among young voters, it makes a difference of about 1.5 points in the net Obama-McCain margin.
Mind you, Selzer is not necessarily assuming that 22 percent of the electorate will be under-30 voters. She is using that as her starting point, and then using her likely voter screen to refine her turnout estimate. For instance, suppose that 60% of voters aged 18-29 pass her likely voter screen, as do 70% of voters aged 30 and up. In this case, the algebra would dictate that 19.5% of her likely voter electorate would be age 18-29:
... Percentage of % Passing Likely
Group Population Voter Screen
=================================================
Age 18-29 22% (A) 60% (B)
Age 30+ 78% (C) 70% (D)
=================================================
... Percentage of
Group Electorate Equation
=================================================
Age 18-29 19.5% (A*B)/((A*B)+(C*D))
Age 30+ 80.5% (C*D)/((A*B)+(C*D))
There is nothing particularly difficult about this algebra. But that may not be preventing some pollsters from getting it wrong. They may fix the youth voter figure at 17%, regardless of what their turnout model says (and ignoring the fact that youth voter turnout increased by 52% as a share of the Democratic primary electorate). Worse yet, they may start with the 17% and then apply their likely voter model, which has the effect of double-counting young voters’ lower propensity to turn out. Or they may simply not stratify their sample by age at all, which creates even worse problems.
A parallel problem could very well be in effect among groups like African-Americans and Latinos. And Selzer told me that pollsters are having another, very peculiar problem with the black vote. Specifically, many respondents, but especially (she believes) black voters, are refusing to disclose their race to interviewers. This wreaks havoc with turnout models, and particularly poorly-designed turnout models.
Selzer is taking a big gamble on this election, as her results have tended to stand out from those of other pollsters. But she is meticulous in how she does her polling, and Selzer’s polls have not had any particular partisan lean in previous election cycles. Don’t be surprised if her gamble pays off again.