I was delighted to have my article about the future of the RPI published in the Daily Telegraph today.
In summary: the demise of the RPI will hit pensioners and investors; renaming the CPIH as RPI is statistical fraud; the lack of transparency and failure of the regulation system means we do not know if the ONS is out of control or political pressure is being applied – or both; the UK will forever tether itself to the EU’s definitions of inflation: and, the sneaky and unjustified change raises questions about the UK’s financial probity.
Continue reading The RPI – Government-sponsored statistical vandalism
The new/latest Public Health England map of Covid-19 cases is not very different from its immediate predecessor. Yet this map has issues: not only does it give misleading messages (exaggerating the extent of the virus) but it fails to publish the data that is plotted (and explain how it’s derived). It also seems to have a category of rate (the yellow, 0-10 category) which is impossible for an area to have (there are none on the map). There are also lots of communities that are blank (labelled as “missing data”, though now changed to “suppressed”) and there is poor labelling of some geographies. One day, I hope, someone will ask why after six months is PHE still unable to produce an online map that that is not riddled with basic errors?
Continue reading The latest PHE map of Covid-19
Who can you trust? The presentation of data at the coronavirus briefing on 13 October (by Prof Jonathan Van-Tam, Deputy Chief Medical Officer, and Prof Stephen Powis, NHSE Medical Director) failed to meet a number of basic standards of statistical neutrality. We are left wondering whether the public servants were deliberately being non-neutral (ie overtly political or self-serving with a one-sided message) or simply using the information in a poor way. Critically, do they believe what they say? If the presentation was political, it should have been made by a politician. If the aim was to use the data in a professional, analytical, balanced and useful way, and the health officials felt unable to do it, the presentation should have been given by an independent statistician. When public servants present data in a one-sided way it does not serve the public well.
I have written to the chair of the UK Statistics Authority to ask the questions below. I will update the blog when a response is received.
Continue reading Coronavirus data briefing – a statistical critique
The DCMO says “What I would give to have had the level of data, testing and medical insight we have now back in February and March this year“. Well, the data has got better – going from barely anything to a still-inadequate something – so why does he appear not be using it to diagnose the situation and plot a way out of the crisis?
In the run-up to today’s announcements on lockdowns, he’s probably leaning heavily on the latest incarnation of the (turgid) weekly note from PHE. It is full of charts starting from the lows in July and rising “exponentially” since. They are perfect fodder for the scare-mongering medics and press but as they are mostly national figures and free of context, they are of little analytical value.
He might like to give his thoughts on the detailed local area data on infection testing which seems to give a very clear message about where Covid-19 is thriving – among students at university. If that is the conclusion to be drawn from the charts below, we should ask the advisors why there have been no student-specific policies in the last few weeks and – although it seems to be too late for the communities in the wider Liverpool – whether such targeted policies would help other cities avoid the worse of more widespread lockdown. In short, the people of Doncaster, Barnsley, Wakefield, Mansfield and many other places whose behaviours have not led to a material rise in infections will rightly feel aggrieved if they are shut down due to student infections causing Covid-19 hotspots in entirely separate social circles.
Continue reading Covid-19: Students are the second wave
Sir Ian Diamond and Ed Humpherson appear in front of the Public Administration Committee on Wednesday. Here are 25 questions I’d ask them. The statisticians seem upbeat: Sir Ian must be enjoying his new-found stardom (having been on our TV screens more than any of his predecessors) and Ed produced a “celebration” blog. There’s nothing wrong with the blog but it should be expected from the bosses not the regulator! That’s because there are questions to ask. Many think that the data response to the crisis from government as a whole has been incoherent, the daily slides need transforming, and so much data we want and deserve has not been published. This can be turned around. Importantly, the statisticians must now help the population navigate the loosening of the lockdown restrictions with clear information about risks. UKSA needs to serve the public good as required in legislation. Continue reading PACAC: 25 Covid-19 questions for UKSA
OK, the country wasn’t. But this is the implication of a report published last year – the UK was top-ranked in Europe for preparedness. I don’t draw attention to this report because I have any particular faith in its conclusions, or to poke fun at a conclusion that now looks questionable, but simply to show that some respected voices were taking a different view from the much-discussed Exercise Cygnus, which has been used relentlessly to bash the government. This, along with any other decisions about forward planning for disasters, is a very complicated topic not least because experts can differ in good faith in their views. Continue reading UK: well prepared for the pandemic
This blog looks at the different measures of the death toll due directly or indirectly to the Covid-19 epidemic. The measures vary considerably and at the end of April the death toll could plausibly be anywhere between 4,000 and 50,000 depending on the definition chosen. None of the three familiar data sets published at least weekly are likely to be the one that will be used for the final ‘best estimate’ when the history is written. (This blog originally appeared on the Royal Statistical Society website as an output of their Covid-19 Task Force.) Continue reading The many definitions of a Covid-19 death toll
Along with many a statistician, I was delighted to see numbers being right at the heart of the Prime Minister Question Time today. It was a good exchange between the leader of the opposition, Sir Keir Starmer, and the Prime Minister. It would be fair to say that the former one this, their first joust at PMQs, by a very skilful use of data. He won on style and substance. At the same time, it’s only fair to say that the Prime Minister was let down by civil servants who once again have not done their political masters any favours, and failed to present the government’s case in a fair light. Continue reading PMQs: Starmer uses statistics to edge the win
The optimistic interpretation of the facts underlying the discussion on the BBC’s Today programme about Covid in Northern Ireland did not ring true. Northern Ireland looks as if it’s several weeks behind England and Wales and the trend is not yet obviously on a downward path. The number of deaths to date is also higher than the 263 quoted. Continue reading Northern Ireland Covid deaths
Germany is the out-and-out star performer in Europe’s Covid-19 battle to keep deaths low. Or is it? Without reassurance we should doubt that the numbers published by Germany are either broadly correct or comparable with other countries. Germany does not seem to publish weekly deaths data (or even monthly, on a timely basis) so we cannot get a handle on excess deaths. Please help me Germany – I have failed to get a straight answer from anyone! Just how can a public health system be so good and the stats so poor? Or to put it another way, how can you know that the response to coronavirus was great if you don’t have the evidence to support the claim? Continue reading German Covid-19 death toll – for real?