It is important to understand that this study, and other observational analyses like it, onlyidentifycorrelations: these relationships are not necessarilycausal. That is, one cannot read thisstudy, or others like it, to imply that changing one of the variables in our model would changedeath rates; we can only say how the death rates and the variables analyzed move together. Wetake much care in stressing this throughout the note.

I feel like study 1 missed that memo - he jumps from 'we have a significant result of one of our 5 parameters, controlling for the rest' to 'so therefore PT is the driving cause', paying lip service to but not actually addressing the fact that there's a plethora of things that public transit use could be a proxy of.

So ignoring that one and focusing on the much better second study; it's interesting that it seems to be such a significant factor. I'd be very curious if these results could be repeated in non-US metropoles like Singapore or Tokyo.

You've been the one who pointed out to me that PT in the US is, effectively, poor people transport. So while I'm not surprised that PT increases risk, I'd also not be surprised if there's simply an unaccounted poverty variable that would explain most if not all of this correlation away.