This is a decent start. However, the data set is statistically insignificant. Only 52 founders? HBR can do much better than this.
I once emailed Reid Hoffman (a friend knows him) proposing to do an analysis of the life-cycle of an entrepreneur (age, what jobs they held before founding, repeat founder success, etc.) I never heard back from him; I'm sure he's got better things to do. But I wish LinkedIn would set up a data-mining department, like OKCupid's. They could do some really interesting stuff.