Until I went back and compared the numbers I took your description to mean that their dislikes here were much stronger than the ones for other clusters.

I'm afraid that during this comment you are being hilarious, but during that part of your post you were being unintentionally misleading.

And here I just thought you had really strong priors for perfectly equal distribution.

It totally flew over my head then (and I'm probably not the only one). I guess I'm too used to popular science journalism that takes a study with a small correlation or a small effect or a small sample size, and rounds it to the nearest cliche to put a triumphant headline like "SMOKING CAUSES HOMOSEXUALITY", whereas the original study was that in a group of 20 rats, filling their cage with smoke made male-female rat interaction frequency decrease by 7%.