One of most fatuous bits of reporting I've seen recently on television "news" was a couple of days ago on Diana Sawyer's evening news.
I'm sorry I've been unable to locate the piece in ABC-TV's online archives. (Maybe they withdrew it out of shame.)
Anyway, the report purported to demonstrate how even today, despite People Magazine's recent naming of a black woman for its "Most Beautiful" distinction, young children still demonstrate a bias toward white skin.
So our intrepid reported (someone named Roberts) tried reproducing a famous experiment from decades ago, in which little girls were shown dolls with different skin color and asked to identify the one they preferred.
Sure enough, nearly all of the girls shown, no matter whether their skin was white, black, or in between, selected the "white" doll. But curiously, I thought, each of them in turn said she liked it because of the blonde hair. Not one of these kids said she picked the white doll because of its skin!
Well, sorry Ms. Roberts, but your whole little "experiment" proved absolutely nothing. It's possible your premise is correct, but there's no way to prove it from this little bit of fluff you produced. It's a basic principle in such tests to ensure that you isolate the variable you want to study, and make sure other variables are controlled. In this case, that would have meant a choice among a black doll with dark hair; a brown doll with dark hair; and (DUH!) a white doll also with dark hair. Then you might be able to draw some conclusion about skin color. Otherwise, better to stay away from stuff (e.g. scientific experiments) you don't understand. But don't fret, I'm sure about 80% of your audience bought the premise, despite the voodoo science involved.
P.S. And what about a black doll with blonde hair? Which would they pick then?