Citing admittedly flawed Zogby poll, Novak claimed Obama result in CA “raises the specter of the ... 'Bradley effect' ”

Citing a “Zogby poll that showed a big Obama lead in California,” Robert D. Novak asserted that Sen. Barack Obama's defeat in the California presidential primary “raises the specter of the dreaded 'Bradley effect.' ” But in explaining why his poll showing Obama leading in California by 13 points did not match the actual results, John Zogby wrote: “It appears that we underestimated Hispanic turnout and overestimated the importance of younger Hispanic voters. We also overestimated turnout among African-American voters.”

In his February 11 column, conservative columnist Robert D. Novak claimed: "[T]he way [Sen. Barack] Obama [D-IL] lost California raises the specter of the dreaded 'Bradley effect.' Los Angeles Mayor Tom Bradley, an African American Democrat, unexpectedly lost his 1982 campaign for governor. His defeat came as voters told pollsters that they preferred the black candidate and then voted the other way. In California's primary last Tuesday, Obama lost by 10 percentage points despite one late survey showing him ahead by 13 points and two others giving him a one-point lead." Novak added, "[D]isbelief that their voters harbor racial prejudices leads Democrats to reject speculation that those voters lied to pollsters in claiming to support Obama. The Zogby poll that showed a big Obama lead in California, and the Suffolk and Rasmussen surveys giving him a narrow edge, it is argued, were just plain wrong. It is also claimed that the state's final tally was skewed by an unexpectedly low African American turnout."

But John Zogby, president of Zogby International, did not invoke the “Bradley effect” to explain why his poll showing Obama leading Sen. Hillary Rodham Clinton (D-NY) among likely Democratic primary voters in California by 13 points did not match the actual results. Rather, in a February 6 memo, Zogby wrote: “Some of you may have noticed our pre-election polling differed from the actual results. It appears that we underestimated Hispanic turnout and overestimated the importance of younger Hispanic voters. We also overestimated turnout among African-American voters.”

Additionally, while citing three polls in his column, Novak did not note that a SurveyUSA poll, conducted over the same period of time as the Zogby poll, February 3-4, correctly predicted that Clinton would win the California primary by 10 percentage points.

From Novak's February 11 column, “Obama's Bradley Effect?”:

Which Democrat won Super Tuesday? Thanks to the Democratic Party's proportional representation, it is not easy to say a week later. Sen. Hillary Clinton and Sen. Barack Obama ran to a virtual dead heat for the delegates at stake in 22 states that were clearly stacked in Obama's favor. But the way Obama lost California raises the specter of the dreaded “Bradley effect.”

Los Angeles Mayor Tom Bradley, an African American Democrat, unexpectedly lost his 1982 campaign for governor. His defeat came as voters told pollsters that they preferred the black candidate and then voted the other way. In California's primary last Tuesday, Obama lost by a landslide 10 percentage points despite one late survey showing him ahead by 13 points and two others giving him a one-point lead.

Was this presumed 20-point reversal caused by the Bradley effect, which has worried Democratic leaders since Obama became an obstacle to Clinton's majestic procession to the Oval Office? It is much too early for that conclusion, but the subject is on the minds and is coming up in private comments of Democratic politicians pondering the stalemate for the party's presidential nomination.

[...]

The prospect of going into a convention with the nominee unknown for the first time since 1952 upsets Democratic insiders not merely because of the uncertainty. Splitting the party along ethnic and racial lines is troubling -- especially in California, where massive Latino support for Clinton canceled out Obama's base among blacks.

However, disbelief that their voters harbor racial prejudices leads Democrats to reject speculation that those voters lied to pollsters in claiming to support Obama. The Zogby poll that showed a big Obama lead in California, and the Suffolk and Rasmussen surveys giving him a narrow edge, it is argued, were just plain wrong. It is also claimed that the state's final tally was skewed by an unexpectedly low African American turnout.

But briefings on exit polls early Tuesday evening, the product of nonpartisan technicians, cautioned listeners not to be carried away by favorable Obama numbers around the country because his actual performance often is overstated by exit polls. (Indeed, contrary to early exit poll signals of an Obama upset in New Jersey, Clinton carried the state comfortably.) No explanation was given for this aberration, but many listeners presumed it was the Bradley effect.