Steve Levitt often quips that "People lie, numbers don't." I say that both people and numbers lie some of the time. The tough question is figuring out how much trust you should give people or numbers in any particular case.
Example: I suspect that many parents really do like some of their kids more than others. How couldn't they?... unless they've got identical twins, of course! Even so, I would expect people to say that they feel the same about all of their kids. At least in our society, you're not supposed to play favorites.
It turns out that I'm wrong. In her latest book, Judith Harris discusses two fascinating studies in which parents admitted the very thing I would have expected them to lie through their teeth about:
In two separate studies, British and American parents of two small children were asked whether they felt more affection for one than the other. More than half admitted that they did. The overwhelming majority of these parents - 87 percent of the mothers and 85 percent of the fathers in the American study - said they favored the younger child.
Why do I believe the parents who admitted to favoritism? As they say on Law and Order, it's a "statement against interest." Of course, the results are so surprising that we should wonder a bit about where the numbers came from. But Harris is an unusually careful reader of research, so I'm not too worried.
P.S. If you don't know much about Harris, here's an interview where she answers ten pointed questions.