The issue at hand was whether or not African-American children (negros back then - this is 1938 we're talking about, concurrent with Adolf Hitler arguing you could tell a Jew by the size of his nose) felt themselves inferior to white children.

Then, once again, they should introduce other alternatives. Perhaps it's simply "different" that they like. Not necessarily white skin. In terms of how they felt about themselves, the choice was either "yes, like me" or "no, not like me".

So even if that's what you were looking for (though it's fairly obvious that they are trying to show racial bias), it's still a flawed experiment.

The fact you want to see whether or not kids feel inferior or superior to green children illustrates that you're either trolling or clueless.

Why? Do you have a problem with green skin? Do you feel white/black skin is superior? This is exactly the reason the options should be included.

But I'm frankly astonished by your steadfast insistence on the outright invalidity of the data.

People are astonished by me often. My point still stands. It's a shit experiment. I see what they are trying to get at, but they are doing it horribly.

Though I am curious, has this test been done in predominantly african communities? Or simply in the US and other white-skinned areas?

There's a lot of factors to consider, and this experiment doesn't isolate any of them. Perhaps it's simply the fact that it's 'different' that they like. The environment they grew up in is definitely a factor. Is the preference social or genetic? Is it really a skin tone issue, or just a case of color association? Have they tried the experiment substituting real kids/babies for the dolls? Was the result the same?

What about a dark haired and light skinned doll? Perhaps it's the hair color they are connecting to. Are we sure that the kids weren't instructed to give certain answers? How can we be sure?

As I said though, if the only difference is skin color, the selection MUST be about skin color. And since the question is phrased as a matter of picking only one, one must necessarily be picked as "better". It's fundamentally flawed.

What's the ideal/desired result, exactly? Honestly, it looks exactly like what you'd expect. Roughly half would pick one doll, and the other half would pick the other doll. Naturally there's a skewed result (not perfectly equal), but that's to be expected.