Basically all of my Master's and PhD work involved scientific programming, but it was definitely not data scientist work. You're making a big assumption that science is statistics. There's scientific simulations (and deriving the models) and scientific visualization as well. Those are only two out of many possibilities.