Yeah I saw someone talking this up on Twitter but nobody linked to it. Thanks for linking to it; it's a shit study. They do absolutely nothing to filter out DoorDash, Tasqrabbit, Chewy or Uber Eats drivers and count "exposure" as "they were in the restaurant at the same time" as if there was some sort of "cross-class encounter" between the family in a booth on one side of the Olive Garden and the family in a booth on the other side of the Olive Garden. The only actual time there might be a "cross-class encounter" is if your family is one of the three that actually fits in the foyer in front of reception because otherwise you're probably waiting in your car.

We use SafeGraph Patterns data for the months December 2021 to July 2022. The SafeGraph data pools mobile location information from different sources to estimate the number of visitors to different places of interest.

Absolutely no investigation into the quality or quantity of interaction was made; libraries (where kids sit around and listen to storytellers) and churches (where everyone is invited to sing together) have the exact same baseline "interaction" as Dollar General (where everyone is head-down grabbing toilet paper as quickly as they can). This is Carl Sagan's "we can't find intelligent life on earth" study minus the irony.