It's important to note (the article does, but it's worth emphasizing) that the correlation for coordinated dance moves is not with present testosterone levels,

It's important to note (the article does, but it's worth emphasizing) that the correlation for coordinated dance moves is not with present testosterone levels, it's with pre-natal testosterone exposure. The first is measured with blood tests, the second with finger lengths. Present testosterone levels correlate strongly with height, muscle definition, and beard thickness. Prenatal testosterone levels correlate with "male-typical" personality traits (increased visuospatial abilities, impaired empathy and language skills), taste bud insensitivity, autism incidence, homosexuality incidence, and decreased autoacoustic emissions. In short, the best dancers aren't necessarily jocks, they're geeks and homosexuals. Honestly, I'd like to see this same study performed with self-identifying homosexuals screened out from the sample pool, as they tend to have much higher 4d:2d finger-length ratios and could drastically affect the results. I'd bet money that the correlation still persists, but the social dynamics are so obviously different between straight and gay men at a dance club that it's worth controlling for.