A new ABC News story on a recent survey of 1,125 adults (with black oversample; 3 percent margin/error) includes this conclusion in its headline: “Poll Finds Four in 10 Think Obama’s Candidacy Will Improve Race Relations.” I would have titled it, “Poll Finds Many Whites Think in Racist Terms, Will Not Vote for Obama.

According to the news report 40 percent of Americans think Senator Obama’s candidacy will “improve race relations” in the U.S. The story then adds:

Three in 10 whites express less racially sensitive views, such as having some feelings of prejudice or believing that blacks in their communities do not experience discrimination; they hold generally critical views of Obama and favor John McCain for president by a 26-point margin. But an additional two in 10 whites are at the high end of racial sensitivity — and they favor Obama by 19 points.

By “less racially sensitive” whites, as Mordy pointed out in a previous post today, they mean “racist” whites, but somehow cannot bring themselves to talk candidly about racism from their own poll. And notice that for these white folks who admit to racist views, McCain has a huge advantage. The smaller group of whites at the other end of the spectrum are viewed as counterbalancing the more racist whites:

Yet the views of these whites are counterbalanced by the preferences of the most racially sensitive whites, a smaller group (21 percent of whites) but broadly pro-Obama. These are whites who have a black friend, who think blacks in their area suffer discrimination and report no personal feelings of prejudice; they support Obama by 55-36 percent.

Somehow 30 percent is balanced by 21 percent? Is this a type of new math that works only on matters of racism?

And the write-up makes a big deal of the findings on whites reporting black friends. One would think they might check out the research literature on this matter. If they did, they would find studies showing that whites often exaggerate when it comes to “black friends,” who often do not exist or are just fleeting acquaintances at work. Then the ABC report continues:

All told, just 21 percent of whites say the race of the candidate is very or somewhat important in their vote — and Obama’s support is essentially the same among those who say race matters and those who say not…. Obama’s race shows little if any net effect on vote choices overall.

At no point do they seem to realize that many whites misrepresent their racial views in surveys. Like most pundits in the television media, they seem to take such polls as true readings of white sentiments. And then, of course, they can conclude that race will have no net effect on white voter choices in this election. Oh, Really?