I asked her at dinner night before last, at the biennial IRIS conference.

She said they tried to get earthquakes in a range of tectonic settings, also they took earthquakes with known triggering (but this is for triggered events much later than the passage of the Love and Rayleigh waves).

It would have been better to not select but rather do all quakes above some magnitude, however, I think the results are still quite significant, and doing all the events would have been a chore.