The basic base rate question is: what should I expect in situations like this? Most often we have looked at base rates through the lens of projects. We have an optimistic tendency to think, “yeah but…”. Sometimes it is! Sometimes it’s not.
The general advice for using base rates has been to start with them, rather than our impressions, and then adapt from there.
Another way to think about base rates is as sampling. It’s important to get the “situations like this” part right, right? This is tricky, and this came up during the summer of 2021 as more and more covid vaccinated people became infected with the covid virus. At one point 67% of Iceland’s cases were among the vaccinated.
“When you look at Iceland and graph out (cases) by who is vaccinated, who is not, and where the cases are, you can see that there are more cases in the vaccinated group than the unvaccinated group.” – Dr. Kat, NPR Planet Money, August 2021
That sounds like the vaccine doesn’t work, or doesn’t work as well, or never-worked?! Maybe, but maybe our conclusions are muddied by an initial assumption that’s wrong.
Rather than jump right to Iceland, let’s pull a Zeckhauser and simplify everything. Imagine in Indiana there is a group of 100 people, half are vaccinated and half are not. In the vaccinated group there are five infections and in the unvaccinated group there are five infections. Putting aside “long-infection”, hospitalization, and death, it-looks-like, in-this-case, that the vaccine is meh.
Okay, now in Nevada there is another group of 100 people. This time there are 90 which are vaccinated and 10 are not. In the vaccinated group there are five infections and in the unvaccinated group there are five infections. Putting aside the same other-factors, in this case the vaccine is doing a lot of work! This was the case in Iceland too. Six of every thousand vaccinated people caught covid while fifteen of every thousand unvaccinated people caught covid. And all of the other-factors were much worse for the unvaccinated group. Vaccination reduced someone’s risk by more than half.
This idea is known as the “base rate fallacy” but really it’s comparing apples to apples which will make the idea stick better anyway(another bit of Zeckhauser advice is to keep explanations simple). BRF is good for talking with economists and behavioral scientists but for implementing this idea it’s an apple-to-apples question a day that will keep the bad decisions at bay.
The Waiter's Pad 