[jiggy2011]

But does a gradient really provide a clear picture if ~90% of your data points stick to one side or the other? To me this seems to be just as clumsy as a gay straight/gay/bi classification system. Either way risks warping your thinking as described in the article.

[btipling]

We are talking about people. So I think this kind of research and distinction makes for a powerful reminder of why tolerance for diversity is so important. Were those young people who do not happen to fit so nicely into these categories encouraged to be themselves rather than conform with 90% of the other data points, maybe they could live happier, stress free lives.

[jiggy2011]

Perhaps, but when you have large populations that can be mapped to discrete points then it does make sense to label and observe these.