Very interesting article! I too have noticed this phenomenon. Feynman was noted for playing bongo drums. And it's worth pointing out that Edward Witten (generally considered to be the smartest living physicist by other physicists) started out with a BA in history and a minor in linguistics. He wrote magazine articles, worked for a presidential campaign, tried economics for a while, switched to applied math, and then finally gave theoretical physics a shot, at which point he made numerous contributions to string theory, quantum gravity, and quantum field theory. He's also the only physicist to win the Fields Medal.
And his father is a theoretical physicist specializing in gravitation and general relativity, it's not like one day he got off his bed and "gave theoretical physics a shot".
Firstly, it relies on two different sources of information about people's hobbies: detailed biographies for Nobel laureates and something called the Sigma Xi society for the general scientific population. I'm not sure what this society is or how much information it has on its members' hobbies, but I envisage some kind of form that invites you to name your hobbies. You would expect perhaps not the same level of detail as you'd get from a biography or obituary.
The second problem is taking members of the Sigma Xi society as representative of the scientific population at large.
This is an interesting idea, but one source of bias that troubles me is that after people have received a nobel prize, it might bias how they are characterized by themselves or others. Maybe being a "famous person" makes us characterize a person as more interesting, and upgrade their interests to hobbies or something more serious.
For reference, according to the article the data about the hobbies of members of the Royal Society, National Academy, and of Nobel Prize winners was gathered from biographies, memoirs, obituaries, and other research.
The type of science that's pursued by a person with a penchant for making things that can be showcased to others is perhaps the sort of science that's good material for a Nobel prize.
