"I'm Not a Racist or Sexist, But These Other People . . ."

February 13, 2008
Posted by Jay Livingston

After the New Hampshire Democratic primary, there was some talk of the Bradley Effect: the black candidate gets a smaller percentage of the actual vote than predicted in the polls. Apparently, norms of political correctness keep people from telling a stranger (i.e., the pollster) they’re not going to vote for the black guy; the privacy of the polling booth removes that social pressure.

You can come at this problem from a second angle by asking not just, “What do you think?” but “What do other people around here think?”

Do you think most people you know would vote for a presidential candidate who is black, or not?
		Would	Would Not	Unsure
		%	%	%
		65	21	14
Would you personally vote for a presidential candidate who is black, or not?
		Would	Would Not	Unsure
		%	%	%
		90	6	4

When you get a 25-point discrepancy like this (30 points in some polls), it’s hard to know which is more accurate.

The same effect holds for women candidates.

Do you think most people you know would vote for a presidential candidate who is a woman, or not?
		Would	Would Not	Unsure
		%	%	%
		56	34	10
Would you personally vote for a presidential candidate who is a woman, or not?
		Would	Would Not	Unsure
		%	%	%
		81	15	4

Note that 2 ½ times as many people (15% to 6%) said they wouldn’t vote for a woman as said they wouldn’t vote for a black. The same discrepancy holds for how they thought others would vote. About one in five thought their fellow Americans wouldn't vote for a black, but a third of the sample thought that their acquaintances wouldn't vote for a woman.

Data are from a CBS News/New York Times Poll. Jan. 9-12, 2008. N=995 registered voters nationwide.