you might ask the reason in an email to Kris Pankow. I suspect it was a practical consideration, maybe even explained in the supplementary material that I couldn't get to, such as shallow events for big surface waves, or events at otherwise quiet times, or just a subset chosen at random times to avoid an unwieldy dataset. Maybe they used a different magnitude than Roger did.

I'd only be concerned about event selection if they picked events for which they had already identified triggered seismicity, and known triggering regions accounted for a significant part of the observed triggering, which possible.