[herrvogel-]

This short video[1] explains it quite well imo. They also make the interesting point that efficacy rates are not comparable because the studies were done at different times/locations with different spread of the virus in the population.

[1]https://youtu.be/K3odScka55A

[amichal]

Reading the more technically detailed comments I think really reinforces throwaway894345's point. I understood what "91%" means (much better than non-techies in my circle) but assumed a lot incorrectly about the controls and the actual "end point" they are measuring. It seems lots of folks are saying the different studies use different sets of symptoms etc and not routine testing at intervals to identify those who have the disease.

I suspect most of the public doesn't even know what 91% means. Combine that with being bad at evaluating risk in general. It mostly represents a 10x improvement in cases per capita per time, but I think telling folks that their risk went from 1 in 10,000 to 1 in 100,000 (or whatever) probably doesn't mean a lot. Folks just want "safe" or "not-safe" :(