Thread:
This @nytimes piece from @apoorva_nyc is great science writing, but I worry it is an example of the streetlight effect.
https://www.nytimes.com/2020/11/23/health/coronavirus-holiday-gatherings.html?action=click&module=Top%20Stories&pgtype=Homepage

The argument is that there is little data to suggest small gatherings drive the surge in covid cases, and that's true but...

That may be because it's hard to capture data from small gatherings. If you have an outbreak in a jail, you are pretty damn sure it's from the jail.
How do you pin down that the source was from a backyard barbecue? Without really good contact tracing, you can't.

The streetlight effect is a form of observational bias that occurs when you look for something where it's easiest to find (the typical example is a drunk looking for his keys under a streetlight. he lost them in the park, but "that's where the light is".)
https://en.wikipedia.org/wiki/Streetlight_effect

Take Colorado, for example. The article notes that there are only 81 cases attributed to social gatherings, but a ton more attributed to other places:

But look at the numbers:
-81 from "social gatherings"
-4000 from jails
-3300 from colleges
-2400 from assisted living
-450 restaurants / bars.
That's 10,231 cases with attribution.

Colorado has had 200,000+ cases of coronavirus. We don't know the source for the VAST majority.

I worry that this article may be taken to convince people that small gatherings are safe, or at least less risky then, I don't know, being a prisoner or something. This is obviously not the intention, but the internet will internet on this.

Bottom line: We don't know how the vast majority of people got coronavirus. I wish we did.

It's tricky to tell if people got it from small gatherings. That does NOT mean the risk of small gatherings is low.

It's the streetlight effect.

Stay home for Thanksgiving.
/Thread