Bradley Effect or Non-Effect?

A lot of people on the left are worried about a “Bradley effect” where some people tell pollsters they’ll vote for an African American or are undecided to hide their racism and then once in the polls they’ll vote against the African American candidate.

The Bradley effect campaigns happened betwen 1982-1992 beginning with Tom Bradley’s campaign for Governor and ending with Carol Mosebly-Braun’s Senate campaign in 1992. African Americans ended up finishing far below what pre-election polls projected in a series of races for Governor, Mayor, and U.S. Senator. Will there be a Bradley effect in 2008? There are two ways to look at this.

First, take a look at recent statewide elections involving Black Candidates. In 2006, there were Black Republicans running for the Governorship of Ohio and Pennsylvania, plus the Maryland U.S. Senate race. In Tennessee, there was a Black Democrat running for U.S. Senate.

In Ohio, the final Real Clear Politics Average had Ken Blackwell down 19.4% with 37.3%. He ended up losing by 23, with 37% of the vote. In Pennsylvania, the final RCP average had Lynn Swann at 36%, and losing by 21.8%. He ended up losing by 20, with 40% of the vote. In Tennessee, the final RCP average had Harold Ford, Jr. down by 6 with 44.3%. Ford ended up losing 3 with 48%.

The one case you might make a case for a “Bradley effect” would be in Maryland where the final polls showed Lt. Governor Michael Steele down by 3.7% with 45% of the vote. He ended up losing by 10 with 44% of the vote. The authors of the Wikipedia article point out that the final polls missed the results of the Governor’s race, underestimating Democratic performance by a similar margin. So, the point here is that some situations where a person might see a “Bradley effect” may just be a case of bad polling.

So what can we say? In 2 cases, Black candidates beat the pre-election polls, in two, they didn’t. It doesn’t seem to be the case for a broad-based “Bradley effect.”

What about Obama? If there’s such a thing as a Bradley effect, we’d see it in the Democratic Primaries and there were some states where Bradley seem to be in play:

  • In California, Obama was only down an average of 1.2% and ended up losing the state by 9.6%.
  • In Arizona, the two polls before the election showed an average Hillary Clinton lead of 3.5, she won by 9.
  • In New Hampshire, pre-primary polls showed Obama by 8.3, Clinton won by 2.6, an 11 point swing.
  • In Kentucky, poll averages had Obama trailing by 29.0, and he lost by 35.6.
  • In West Virginia, he trailed by 35 and lost by 41.3.

It’s also worth nothing that Obama had some reverse Bradleys in the primary where he beat the polls:

  • Wisconsin poll averages had him up by 4.3, he won by 17.4.
  • South Carolina, he led by 11.6, and won by 28.9.
  • Virginia, he led by 17.7, and won by 28.2
  • North Carolina, he led by 8 and won by 14.7.
  • Oregon, he led by 12 and he won by 17.6.

In Southern states, Obama benefited from higher than predicted Black turnout, but in Oregon or Wisconsin, that wouldn’t seem to be a suffiicent answer. In 10 states, the polls were way off. This may be less indicative of a Bradley or reverse Bradley effect, and may instead suggest the difficulty of polling an election when you have no clue who will show up.

The four races in 2006 suggest no large “Bradley effect.” Certainly, some people voted against these candidates for racial reasons, but they weren’t trying to hide what they thought. There’s only one reason I can think of that this might be different with Obama. In the case of all four candidates in 2006, there was no major case made that not voting for Ken Blackwell, Lynn Swann, Michael Steele, or Harold Ford made you racist. The case has been made with Barack Obama.

Contrary to the opinon of liberals who are pushing Bradley, it would not even require someone being a racist to get them to lie to pollsters. Lying to pollsters is not necessarily prompted by what you feel in your heart, but you fear what the people polling will think of you if you express your opinion. No one wants to be counted as a racist. Thus, the suggestion that many votes against Obama will be based on racism may have the unintended consequence of tainting polls. For example, one need not be a racist to sense that Carol Mosebly-Braun would be a disaster in the U.S. Senate.

This seems an outside possibility, but if Obama loses, I’m sure it’s a discussion we’ll end up having.