Monday, June 16, 2008

The Bradley Effect Analyzed

See this New York Post article in which the author has interviewed leading political pollsters concerning the Bradley Effect. For those readers who aren't addicted to political arcana, the Bradley Effect is the tendency for opinion polls in advance of an election to overstate the size of the subsequent vote for African-American candidates.

Named for Mayor Tom Bradley of Los Angeles who ran unsuccessfully for governor of California, all the polls incorrectly predicted he would win. The effect has also been observed in other political contests featuring a black candidate.

Two reasons are put forward for the effect. The most common explanation is that when voters are asked for whom they plan to vote, a significant number believe they should say they will vote for the black candidate when they actually plan to vote for the white opponent.

The article reports Andrew Kohut, head of the Pew Research Center, has a different view:
Kohut's contention is not that white voters lie to pollsters, but that pollsters fail to fully capture the views of lower-income, working-class and elderly white voters who disproportionately refuse to participate in telephone surveys, even though they vote on Election Day.

Whoever is right, you will be hearing more about the Bradley Effect in the months between now and November.