It's unclear what you mean by more selective. If you're an average woman, you get more inbound interest than an average man. The tricky statistical massaging you're doing amounts to "well men are even more open-minded when it comes to more attractive woman." But that doesn't make sense: if a man has sent you a message, does it suddenly become half a message if he then messages someone else? You're penalizing men on open-mindedness for messaging more women.

That said, the blog post is weirdly presented and its graphs a bit misleading because of the skewed distributions. What would be most useful is comparing the median number of messages received by men and the median number of messages received by women (or best of all calculating the Gini indices of the respective distributions), instead of bothering to use the star ratings at all.

ETA:

> Divide the all raw numbers by 3.5, the rates will stay the same.

You're misunderstanding the data. It's not that men collectively send 3.5/4.5 of the messages; it's that each individual man sends 3.5 times as many messages as an individual woman. I.e. it already corrects for there being more men on the app. If men comprise 2/3 of users on dating apps, that means that 87.5% of all first messages are sent by men.