I agree. The R² is a small thing — a sign that the model isn’t a good fit, but definitely not a dealbreaker on its own.

The small sample divided up into many subcategories renders the study scientifically meaningless. It’s effectively a collection of anecdotes, with all the problems that entails.