Also, it's a much noiser data-set, especially since puberty seems to be starting earlier. You can kind of get by if you ensure you're population is all pre-pub

Also, it's a much noiser data-set, especially since puberty seems to be starting earlier. You can kind of get by if you ensure you're population is all pre-pubescent or all post-pubescent and similar in age. But anything in between is a mess of confounders.