Presidential polling possibly problamatic
CNN’s newest poll of polls puts Democrat Barack Obama up over Republican John McCain, 48-43, which is where they stood prior to the first presidential debate. Polls can be big news at this point in an election cycle, and while all the usual caveats apply, this year’s contest may involve some factors which makes the data less reliable than we’d expect.
The first variable could be the so-called Bradley effect, named for Los Angeles mayor Tom Bradley. An African American, Bradley lost the 1982 gubernatorial election in California by 1.2% despite leading in pre-election polling. Some suspect that many poll respondents who said they intended to vote for him only dis so because they didn’t want to appear racist. If the Bradley effect is in effect, then Barack Obama’s actual support may be lower than polls indicate; some people who are actually leaning McCain or who really are totally undecided may be saying they’ll support the Democrat just so they don’t feel like the pollster is judging them as a racist. In the anonymity of the voting booth, this pressure wouldn’t be there. Helps: McCain
People who exclusively use cell phones may also be influencing polls by not influencing them. Voters who don’t have a land line are much more likely to be young, minorities, and lower income, all demographics among which Obama has more support, but they are less likely to be questioned by a pollster. (Younger people are also less likely to be married and to own a home, factors which correlate with somewhat more conservative political views.) However, pollsters can attempt to correct for this bias; they usually ask demographics questions of their respondents and can tell if a particular demographic is significantly underrepresented in their samples. If a demographic is underrepresented they can then weigh the responses of those people more heavily to try to approximate their numbers in the general population. But research indicates that even correcting for age, people who exclusively use cell phones tend to support Obama more heavily than McCain. Some polling experts, however, believe the difference is negligible or non-existent, given their ability to correct for sampling bias. Helps: Obama
Then, even if the polls are accurate about current voter intent, one must still try to figure out which of those voters will actually go to vote on November 4th. Obama’s supporters are significantly more likely than McCain supporters to be “very enthusiastic” about their candidate. Obviously, higher enthusiasm correlates with greater likelihood of actually voting for the guy you say you like. Helps: Obama
But not so fast. Older voters, who tend to support McCain, are historically more likely to vote than younger citizens, among whom Obama has such a big lead. Will the youth vote disappoint again, as usual? Helps: McCain
What does it all mean? Who knows. It’s impossible to tell even how much any one of this factors will come into play, let alone what the aggregate effect will be. Maybe we should pay less attention to polls and more attention to the political futures markets? Intrade speculators are moving towards Obama; market prices indicate that speculators think there is about a 64% chance that Obama will win. They also favor him in all of the battleground states, predicting that he has a 53.5% chance to win Ohio, 54.5% chance to win Virginia, and 54.5% chance to win Florida, which is virtually must-win for McCain. Obama is also up big in Colorado (68% chance to win) and has pulled close in North Carolina (48% chance to win).