The Bradley Effect is a term that has been prominent in this campaign that suggests that poll numbers (and exit polls) may not be indicative of actual voting results in the presidential election tomorrow. Its origin is from African American Tom Bradley’s failed 1982 bid for California governor, in which poll numbers suggested a decisive win. Bradley went on to lose that election, and race is presumed to have been a factor. Barack Obama, who has inherited his Kenyan father’s complexion, is ahead in most nationwide polls of likely voters, but pundits have been expressing caution due to this potential Bradley effect.
I’ve heard that the difference could be as many as 6 points, which easily puts Obama’s leads within striking distance in contests nationwide. However, I expect that 26 years later, the racial climate has sufficiently changed, and a new generation of voters are going to be less fickle in this regard.
Finally, the notion of a Bradley effect had me thinking about the reverse of this situation. Consider in some parts rural America (not all by any means), in communities that are predominantly white, you have a social group of persons who may harbor understated (or even overt) racial biases. A person in such a community may feel uncomfortable publicly voicing his or her preference for an African American candidate, for fear of social exclusion or backlash. Even when on the phone with pollsters, in earshot of others, they may be inclined to support John McCain, though they plan to mark Obama in a secret ballot.
It’s not a huge leap of logic to imagine that such a situation probably does exist somewhere in this country. However, it is certainly unclear to me whether or not such an effect might have any kind of significant impact.
Of course, the hope is that neither effect is either significant or necessary, but the reality is that probably both exist. Unfortunately, getting data to examine this effect would be very difficult.