Originally posted by Bryn
View Post
There appears to be a problem with both slightly cloudy presentation of the data by Zoe here and the subsequent (mis)interpretation and invalid comparison by the tweeter.
With careful reference to the wording of:
it can be seen that the Zoe figure of 33,118 "cases" comprises a mean 17,501 symptomatic daily cases in the unvaccinated plus a mean 15,537 symptomatic cases in the vaccinated. 17,501+15,537 = 33,118.
This is completely consistent with the official gov.uk "cases" being 40k+ as that will include many asymptomatic cases as well as not including many symptomatic cases that people don't bother to get tested for. That's before further taking into account the fact that the Zoe data is automated symptomatic self-reporting and is measuring a different thing and yielding results on a smaller time lag.
These things will all have correlations which are no doubt carefully calibrated to get R, prevalence and all the rest of it. Zoe are (as I understand it) calibrating their symptomatic self-reporting numbers to extrapolate prevalence etc. My understanding is that their outputs have been widely acknowledged to be very well correlated with other more laggy "official" measures. Zoe output can be said to give trend signals earlier, and with a fairly reasonable degree of confidence.
Regardless of all this, in both Zoe and e.g. gov.uk datasets, for some while:
i) First differential of positivity has been consistently falling for some time - i.e. rate of increase of daily cases is falling
ii) Second differential is small however - rate of increase of rate of decrease is 0 or worse
None of that contradicts anything Spector is saying, though ii) is less comforting than it might be. It is not inconsistent with his claim that we're heading for a plateau and then a much slower tailing off than has been seen before. Which isn't the same thing as saying that will happen - which nobody really can.
[Like Spector and many others I'm personally in favour of continued mask wearing in many situations for as many people who can as both a courtesy and as it appears to have a minor but non-negligible effect on spread, albeit the balance of evidence remains vague at best]
Comment