COULD the colour of Barack Obama鈥檚 skin scupper his bid for the presidency? It鈥檚 a sensitive question, and one that pollsters and commentators are desperate to answer. What they are all trying to unpick is the 鈥淏radley effect鈥�.
Named after an African American called , it encapsulates the electoral phenomenon that led Bradley to lose the 1982 race for the governorship of California to a white candidate, despite having been comfortably ahead in opinion polls. The discrepancy was put down to two things: prejudiced voters who were unwilling to tell pollsters they would not vote for a black man; and the possibility that those who refused to be polled were more likely to be racially prejudiced.
This effect was seen again in 1989 when another African American, Doug Wilder, was elected governor of Virginia by a much smaller margin than predicted. And it was apparent during the 2008 Democratic primaries between Obama and Hillary Clinton. by Anthony Greenwald and Bethany Albertson of the University of Washington, Seattle, showed Obama lost in New Hampshire despite being ahead in the polls, and lost California by a bigger margin than expected. In both states Obama鈥檚 votes were about 6 percentage points lower than predicted. So some pundits argue he needs to lead John McCain by 5 points or more to overcome covert racism.
Advertisement
However, the effect is not as clear-cut as these findings suggest. During the Democratic primaries, the Bradley effect did not seem to be present in all states. Wisconsin, whose racial make-up resembles New Hampshire鈥檚, voted overwhelmingly for Obama. More telling was a 鈥渞everse Bradley effect鈥� that Greenwald and Albertson found in south-eastern states such as Georgia, South Carolina and Mississippi. 鈥淭hese are former slave states, where there is a long tradition of white dominance and black subservience,鈥� says Greenwald. 鈥淚f you are either a black or a white poll respondent in those states, you may feel some pressure to say that you support a white candidate.鈥�
Their study suggests that polls in such states underestimated support for Obama by 5 to 15 percentage points. If this persists in the election, Obama could steal hardcore Republican states like North Carolina and Georgia, where polls show him either tied, or losing by less than 10 percentage points. This may be enough to offset any states lost due to the Bradley effect.
If it still exists, that is. Scott Keeter, director of survey research at the in Washington DC, studied 2006 senate and gubernatorial races in Ohio, Maryland, Pennsylvania, Tennessee and Massachusetts 鈥� all of which featured a race between a black and a white candidate 鈥� and found that the black candidates lost or won by expected margins. 鈥淚 came away thinking that we really don鈥檛 have a Bradley effect any more,鈥� says Keeter. He thinks racial prejudices have softened significantly since the 1980s. 鈥淥lder, more conservative voters have passed on, and been replaced by younger, more liberal voters who are more comfortable with ethnic diversity.鈥�
鈥淥lder voters have been replaced by younger ones who are more comfortable with ethnic diversity鈥�
More importantly, Obama鈥檚 relative inexperience, especially in matters of foreign policy, means those questioned by pollsters can be honest about their support for McCain and not worry about appearing racist, Keeter says. 鈥淭here鈥檚 ample reason for conservative Democrats or independents to say that they are going to vote for McCain and not be ashamed of it.鈥�
Other factors could help Obama, despite any lingering racism. Many whites could vote for Obama as a symbol, 鈥渢o help move the country past a very bad legacy with respect to slavery and racism鈥�, Keeter says. Then there is what is being called the 鈥淥bama effect鈥�. His support among African Americans is at a stratospheric 90 per cent. And young voters support Obama by a factor of 2 to 1 over McCain. A high turnout among African Americans could swing some southern states, especially North Carolina, towards Obama, while a high youth turnout could impact every state. Pollsters find it hard to incorporate these sub-populations into their polls, because of a lack of previous data on which to base their models.
Finally, the Bradley effect could be lessened by the cellphone effect. Most state polls use automated voice-response systems, which are prohibited from calling mobile phone users. Nearly 1 in 5 Americans are now unreachable by landlines, and they tend to be young and more likely to support Obama. done since June by the Pew Research Center, which uses mobile phones and landlines, showed that missing out cellphone users can underestimate support for Obama by nearly 2 percentage points.
Early voting has already begun in the US, and exit polls conducted by in five states show that , a swing of anywhere from 15 to 30 percentage points in his favour compared with pre-election polls. In the last presidential election, these states chose George Bush by an average of 6.5 points. If such trends continue, the 2008 presidential election could see the Bradley effect dead and buried.
US Election 2008 鈥� Science and technology are at the heart of many of the issues facing the candidates. Find out more in our special report.
