It's a since discredited number from studies on gallup polls. I think it was the freakonomics blog that had a series of posts outlining the issues with it; sinc

It's a since discredited number from studies on gallup polls. I think it was the freakonomics blog that had a series of posts outlining the issues with it; since then there were more rigorous studies concluding that the earlier conclusion was not valid.