Race, Polling, Fibbing

In an article in the Wall Street Journal of 8/2/08, Ellen Gamerman writes of efforts researchers make to determine the extent of bias in response to survey questions. To determine whether people commonly lie, various techniques have been employed: use of a computer; use of a computer avatar named Victoria (two varieties: one with various gestures, one who stares blankly); placement of responses at various points on a computer screen (testing their "Good Is Up" theory); asking a control question; asking the respondent, at the end of the interview, what race he/she believes the interviewer belongs to.

Such studies are particularly relevant in the context of this presidential election, given that one of the two major candidates is black. Gamerman reports

Peter Hart, a Democrat on a bipartisan team conducting the Wall Street Journal/NBC News poll, estimates that 10% of current Democrats and independents who say they support presumed Democratic Party nominee Barack Obama may not be giving a fully honest answer, at least based on their responses to broader questions about race.

I suppose Republicans are not even pretending they're going to vote for the Democratic nominee. Anyway, this suggests the "Bradley Effect," a (untested) theory developed to explain the loss of Thomas Bradley in the 1982 California gubernatorial race after pre-election, and exit, polls showed him ahead.

I'm sure the Obama and McCain teams have pondered this factor and its possible effect (which I don't believe is primarily due to racism). But.... there are two other critieria which may work in the opposite direction, possibly giving Obama a greater percentage of the vote than polls prior to the election would indicate. They are:

- the likely unprecedented turnout of blacks;
- the turnout of the increasing number of individuals, primarily the young, who use only cell phones, eschewing land-line phones, and thus not reachable in polling done by phone.

Blacks, obviously, will be voting overwhelmingly for Obama; youth generally, less so, but still solidly. And I suspect the number of young people who rely on cell phones who vote for McCain, who only recently was introduced to the Internet and gives something of the appearance of a technological troglodyte, will be tiny.

So ask yourself these questions: 1) How often have reports of surveys explained to what extent the researchers are accounting for the greater percentage of blacks in this electorate than ever before in a presidential election? (answer: not often); 2) How often have reports of surveys explained whether the researcher is accounting for individuals who have no land-line phone? (answer: less often). Which helps explain why these polls measuring support for McCain v. Obama probably explain very little, and perhaps why their results so radically differ from each other.