COVID-19 Outbreak (Update: More than 2.9M cases and 132,313 deaths in US)
They do not offer a link to the actual study.
They actually do, it's the word 'modelling', which goes
here. Not very obvious.
But off the top of my head I'm deeply skeptical of the findings. They've made some serious assumptions and those are amplified by what appears to be a fairly simple model.
Look at it this way: the UK is testing people with symptoms, mostly hospital admissions at this point. If we take the last five days, around 18% of those have been positive. That
really doesn't seem to fit with the notion of 50% of the UK having been exposed to it, does it? Especially since that's a group that should be biased towards testing positive compared to the average member of the population.
Even allowing for the possibility that some of that group would have had it but no longer be testing positive for it, as that would be the minority of the group (as with the nature of exponential spread far more people would have had it recently than would have had it earlier) that wouldn't seem to account for the discrepancy.
Which means either the testing is missing large numbers of people, or the model is flawed. I don't think there's anything which supports the testing missing that many cases, so I'm going with flawed model.