Our paper on how well experts and laypeople forecast the size of the COVID-19 pandemic in the UK (back in April) is out: journals.plos.org/plosone/articl… @d_spiegel @alex_freeman Short answer: Not very, though experts were much better than nonexperts. Both groups underestimated it 1/8 
Everyone was asked to make four predictions for all of 2020 - total UK deaths, UK infections, world infection fatality rate, UK infection fatality rate. And to pick two numbers where they were 75% sure the true value would fall in between. 2/8
True values were inside experts' '75% sure' ranges for only 44% of experts' predictions. True values fell inside nonexperts' '75% sure' ranges only 16% of the time 3/8
Experts predicted median 30K deaths, 4M infections. Laypeople, 25K deaths, 0.8M infections. True values more like ~75K deaths, ~6M infections. (talking all infections here, not just confirmed cases) 4/8
Laypeople also thought people who got COVID-19 would be much more likely to die than they really were (but still underestimated the overall impact) 5/8
One caveat: 'Expert' predictions came from statisticians, mathematical modelers, virologists, clinicians, and epidemiologists recruited via social media - doesn't tell us a lot about the experts advising gov't 6/8
...and laypeople were recruited through @respondi (proportional to UK on age & gender): they weren't practiced predictors like the impressive folks at @metaculus & @superforecaster 7/8
Even so, suggests we all ought to be a little less sure that we know how things are going to turn out - experts included! 8/8
• • •
Missing some Tweet in this thread? You can try to
force a refresh
