Author of study which projected UK deaths could come in under 6000 says modelling overtaken by reality; 20k-30k is much more likely if the measures being taken are effective in flattening curve. (So 5700 figure should be dropped entirely by commentators)
The trajectory for < 6000 had a peak of 260 deaths on April 5th. Sadly 260 was hit yesterday; there is no suggestion that this is likely to be the peak
The governments' chief medical and scientific advisers are also saying this weekend that containing to 20,000 UK deaths would now count as an (obviously tragic) relief/escape from where we appear to be now. Hence PM letter preparing people for worst is to come in weeks ahead
Much interest and debate in projections and models.. Most people want control more when we don't have it. And we want to understand new & scary phenomena. Most experts producing models are clear that there is so much uncertainty (despite our desire to know what *will* happen)
Two problems in media reporting/culture
* dramatic outlier claims get much more attention if v.high or v.low deaths > middle
* Complaints that "experts" can't make their mind up, can be exacerbated by misunderstanding who has said what (this confuses different people),
* dramatic outlier claims get much more attention if v.high or v.low deaths > middle
* Complaints that "experts" can't make their mind up, can be exacerbated by misunderstanding who has said what (this confuses different people),
- Tom Pike is an engineer, rather than an epidemologist, so his study was a statistical model. Some have confused it with other Imperial studies. (His own view was that the hopeful stats made the case for the lockdown, rather than the case against it)
Overall media culture some unhelpful biases re science
- drama (headlines/prefer outliers),
- dislike uncertainty (narratives > numbers)
- conflict (who's right? why do [useless] experts disagree?)
But many exceptions. @evanhd a vg communicator of numeracy, nuance & complexity
- drama (headlines/prefer outliers),
- dislike uncertainty (narratives > numbers)
- conflict (who's right? why do [useless] experts disagree?)
But many exceptions. @evanhd a vg communicator of numeracy, nuance & complexity
The 5700 headline was never much of a "model" at all. (Just a v.simple statistical projection - what would happen if UK got onto China's [reported] trajectory).
Media confused by "Imperial College" brand into over-projecting it.
Media confused by "Imperial College" brand into over-projecting it.
This thread explains accessibly what the simple statistical projection in this paper showed & how
I think we can go too far in under-estimating the value of this skill - to communicate, which means simplifying, while conveying nuances. ("Its complicated/uncertain" doesn't help, if can't convey the gist of the expertise).
