Recently, there has been quite a few people (notably some MPs) criticising the quality of COVID-19 models that have been produced by scientists to try and predict how the virus will spread. Given this, I decided to investigate and see whether there is any tangible evidence to suggest that the models used during the pandemic have been as ‘poor’ as many individuals are suggesting. In doing so, I came across a really interesting blog entry by the London School of Economics which discusses, among a few other things, the quality of the Imperial model used at the start of the pandemic, and does so by considering the arguments set out by a paper criticising such models (https://muse.jhu.edu/article/773103/pdf). It concludes that the paper ‘fails to show that the ICL model relied upon bad data, or that it generated poor and overblown predictions’. One of the key ideas to take away from the blog entry is that considering the complexities involved in judging the quality of COVID models may mean that some of these models have been more accurate than they seem.
It’s an intriguing read for anyone interested: https://blogs.lse.ac.uk/covid19/2021/05/06/bad-data-and-flawed-models-fact-checking-a-case-against-lockdowns/
