The reason for this distinction is that if you only divide the population into other-employed and self-employed, they have similar profiles in almost every way, except that the self-employed earn less. That doesn't square with the theory that entrepreneurs are somehow different from the general population -- smarter, more charming, whatever it might be.
They use data to show that the population of unincorporated and incorporated businesses closely tracks two groups in the underlying population of job titles, those which are more manual and those which are less. They also note that business types rarely change, so what you start with is usually what you wind up with.
When you make that distinction, the paradox of entrepreneurship disappears. The populations have distinct characteristics.
If you want entrepreneurs to include folks with small businesses (fine by me, regular small business is much harder than being handed millions of dollars by a VC) then the conflict with Schumpeter's notion reappears. If you go with the narrower definition, this paper has useful contributions.
An argument over the exact scope of "entrepreneur" as a word doesn't really change the shape of the data.