I can't believe this is even up for debate. But it's nice to have the data.

Edit: regression discontinuity design is a new one for me. It looks interesting. The closest parallel I've used is a cutpoint regression

As I said on FB, it's only rational that any sort of physical barrier would help, even if not infallible.

RD is a standard example of a "natural experiment," i.e. a situation where events, or in this case a policy choice, randomized people into treated and control groups. This is essential in demonstrating a causal connection beyond mere correlation.

Yep, we used to do that with things like latent variable modeling but TBH it's been decades since I've built one from scratch and would have to go back and hit the books again if I had to do one now.

I think I told this story but I walked into my PCP for a medication refill request and the woman behind the counter asked me if I had COVID!

I had to explain to the nurse that I was still wearing masks!

Sigh.