But lockdown timing exhibits a much stronger relationship.

Countries that locked down earlier in their outbreaks subsequently had much lower daily death tolls than those that locked down later (accounting for when outbreaks began)

Lockdowns, and their timings, matter. Who knew?

Notes:
• For US lockdown I’m using Mar 20 (CA, NY, CT, IL etc had locked down)
• I’ve kept same basket of countries for all charts to make sure strength of relationship isn’t affected by sample
• I’ll keep updating these as more countries cross "10 days since 50th" threshold

A few pointing out poor data quality in e.g Brazil, India, China could be playing a role in weak relationships we see for population & urban centres.

I agree (got a chart coming tomorrow on how much we might trust any given country’s data) but for now this is best data we have.