The alt accounts thing is contradicted by the numbers. In this case, it's not merely that attendees create a few accounts, they are responsible for the vast majority of onboarding across the whole platform.

However, yes it is not possible in this case to reliably split out cause and effect for this data generally. I would lean towards it being a two way thing, that cohesion of the community is reinforced by in person rituals, HiveFest being the biggest and most impactful, despite a major self selection element. There is more rigorous research that supports this from other open source communities.