Did the Office for National Statistics really produce ‘false data’ on coronavirus infections?

Were new cases actually plateauing in October?

Had a look at this question for @indypremium ... https://www.independent.co.uk/independentpremium/news-analysis/ons-false-data-coronavirus-b1768714.html

In short, no.

But it's an understandable confusion.

The latest estimates of new daily cases in England produced by ONS modelling of data from its weekly coronavirus survey HAS been revised down dramatically and DOES show much flatter profile in October than previous estimates..

...As many have pointed out (such as @SarahDRasmussen & @ChrisGiles_ ) the ONS modelling estimates are very volatile.

The time series gets revised a lot as new info gets fed into the model.

Some argue it swings too much, and that this is a deficiency of the model...

....Perhaps.

But let's park that issue for the moment.

The ONS, for its part, stresses we shouldn't, in any case, look at it's modelled estimates for a sense of what's been happening in recent weeks.

We should, it says, look at it's UN-modelled estimates...

...Essentially, the raw data from the survey.

And this raw data still shows an October spike in new cases...

....Moreover, this corresponds with the rising trends identified by other surveys such as Imperial College London's REACT survey and the government's own UK testing data...

...So the big revisions in the ONS modelling data (probably) shouldn't cause us to conclude that new cases were flat or that there was clearly no need for the second lockdown...

....But this, I think it's fair to say, has been a bit of a data farce.

And, personally, I think the ONS (which by common consent has been generally excellent in the pandemic) ought to be more careful in how it presents its MODELLING data in its survey releases...

...Anyway, thanks to @Kit_Yates_Maths and @toxvaerd1 and @ONS (and the many experts on here who regularly share their analysis) for assistance.