Eminent statistician Sir David Spiegelhalter suggested Number 10 ‘frightened’ the public into complying with lockdown rules
Number 10 cherry-picked ‘spurious’ coronavirus data to justify England’s second lockdown and may have intended to frightened the public, according to one of Britain’s top experts.
Eminent statistician Sir David Spiegelhalter suggested ministers had eroded public trust by choosing only to show worst-case scenarios, which were often based on out of date data.
The Cambridge professor told MPs today: ‘I don’t want to ascribe motivation to anyone of course. But if someone was really trying to manipulate the audience and frighten them and persuade them that what was being done was correct, rather than genuinely inform them, then this is the kind of thing they might do.’
Doom-mongering graphs which predicted 50,000 cases by mid-October and 4,000 deaths a day by late November were used by Downing Street to justify England’s second lockdown.
Those fantastical charts were widely ridiculed because the country was recording 14,000 daily infections last month and daily deaths currently average 441.
Speaking at a House of Commons select committee today, Professor Spiegelhalter added: ‘Those projections were made by one team early in October under certain very pessimistic assumptions.
‘They’d already been revised twice by the time they were shown to the public so it was completely inappropriate to present them to the public.
‘I’m not saying the judgment [to decide to go into lockdown] was wrong, I’m not making any comment about that.
‘What I’m objecting to strongly is the fact such spurious data and graphs were being presented to the public as a justification for the decisions that were being made.
‘You didn’t need that graph, you just needed quite short-term projections to tell something needed to be done or we could be in real trouble very quickly.
‘There is good data available and yet at some point the need to persuade people, to instill a certain emotional reaction in people seems to take over at really quite a high level of decision making. I think it’s quite unfortunate.’
Even Tory MPs compared the doomsday data used to justify the second lockdown to the controversial dossier that sent Britain to war with Iraq.
A graph projecting deaths hitting 4,000 per day by the end of December (blue line) was wheeled out at a Downing Street press conference last month announcing the second lockdown
The 4,000 deaths per day scenario was based on the assumption that there would be 1,000 per day by the start of November. There are on average 441 occurring across the UK as of November 23
The statistician was giving evidence to the House of Commons Public Administration and Constitutional Affairs Committee today.
He told MPs: ‘I’m on record as having complained about what I call “the number theatre” of briefings in which big numbers were being thrown out, which seemed to be intended to give an effect favourable to politicians rather than genuinely trying to inform the public, who are sacrificing so much over this whole period.
‘So just to come back to the important issue of trust… trustworthiness is the first pillar of the code of practice of statistics in this country.’
Professor Spiegelhalter was asked about the use of graphs and data in the hastily organised Downing Street press conference last month to announce the second lockdown.
Covid deaths rose steadily in November to take all-cause fatalities 19% above average – but the number is well below England’s spring first wave which saw 21,000 people dying each week
Covid deaths in England continued to rise in mid-November to 2,274 in the second week of the month, meaning deaths from all causes were 19 per cent higher than usual for the time of year.
Office for National Statistics data published today shows there were 1,833 excess deaths in the week that ended November 13, with a total of 11,495 people dying.
Although higher than average, the figure is only just over half the 21,157 deaths from all causes in England’s worst week in April during the peak of the first wave — when Covid claimed around 8,000 lives.
The number of people dying with the viral infection is now on par with those succumbing to flu or pneumonia for the first time since June.
The coronavirus deaths recorded in the week that ended November 13 were a rise of 503 on the 1,771 recorded in the first week of the month. The data won’t take into account any effects of lockdown because it can take infected patients several weeks to fall severely ill.
Figures show that numbers of people dying of any cause is higher than average in all regions, with the northern parts of England still worst hit by Covid. In the North West, deaths are 38 per cent higher than usual, while in the South East they are only marginally higher at two per cent.
Hospitals, care homes and private homes are all seeing more fatalities than they would expect to at this time of year, and one in five deaths are now linked to Covid-19, compared to one in six a week earlier. But the ONS data is backdated and known to be too old to reflect the current situation in the country.
Department of Health death counts, which take in daily totals from hospitals and care homes for the whole UK, suggest that the number of people being killed by Covid-19 has started to level off after a surge in October.
In the last three weeks of October, from the 10th to the 31st, the average number of people dying with coronavirus each day soared four-fold from 63 to 259, while in the first three weeks of November it rose far more slowly – by 57 per cent from 260 to 409.
There are signs of a flattening in the trend, although experts have warned deaths will likely stay in the hundreds per day for weeks to come as the effects of huge numbers of infections in October and early November continue to trickle through.
He added: ‘The communication on that Saturday was particularly poor, they were rushed into the briefing because somebody had leaked the main graph to the BBC the day before.
‘This graph which claimed a projection of up to 4,000 deaths a day was widely ridiculed, and quite rightly.
‘That was never intended for public communications it was out of date even at the time when it was released. It just shouldn’t have been shown.
‘I think what this shows is a real problem we’ve seen throughout the pandemic of the use of reasonable “worst-case scenarios” for public communication.’
Scientists modelling the pandemic to inform the Government produce three forecasts to give ministers a clearer idea of how the virus is spreading.
They use a best-case scenario and a worst-case scenario. Experts then take an average of the two to produce the third forecast.
Professor Spiegelhalter added: ‘There is a role for those [worst-case scenario forecasts] – essentially for planning, how many body bags to buy etcetera – but three times I can think of [when they were misused].
‘At the beginning we were told up to 500,000 deaths could occur, yeah, okay, if we did absolutely nothing.
‘Then we had that graph, the famous one where cases were going to double every week and we’d end up with 49,000 by the end of October. In fact we had 14,000 by the end of October reported.
‘That is not trustworthy communications, these are based on extreme assumptions that we just don’t do anything.
‘I don’t want to ascribe motivation to anyone of course but if someone were really trying to manipulate the audience and frighten them and persuade them that what was being done was correct, rather than genuinely inform them, then this is the kind of thing they might do.
‘Because no matter what you say about scenarios they will be interpreted as predictions.’
He said: ‘Scenarios are fine, but when they’re used in proper planning you have multiple scenarios you don’t just produce one, you don’t just produce the worst possible one, you show people the reasonable ones, optimistic ones, pessimistic ones, so you get an idea of the fan of possibilities and what sort of decisions might lead us into those different scenarios.
‘And to just choose one to show it to people it breaks pretty much all the rules that I laid out earlier on.’
The data row comes as Royal Society scientists warned the UK does not have enough access to real-time statistics and by default relies on out of date figures.
The Data Evaluation and Learning for Viral Epidemics (DELVE) said the Government should make better use of data gathered by private companies, as is done in other countries, rather than sticking with its centralised data gathering approach.
The report, Data Readiness: lessons from an Emergency, highlights how data such as aggregated and anonymised mobility and payment transaction data could be used to give a more accurate picture of the pandemic at national and local levels.
That could in turn lead to improvements in evaluation and better targeting of intervention, the group says, rather than crude approaches.
During the pandemic, counts of the daily flow of people from one place to another between more than 3,000 districts in Spain have been available at the click of a button.
This was based on a collaboration between the country’s three main mobile phone operators.
It has allowed Spanish officials to have a deeper understanding how the movement of people contributes to the spread of the virus, DELVE claim.
Spain is one of the few European nations which avoided a second national lockdown, though transmission of the disease is still high.
Neil Lawrence, professor of machine learning at the University of Cambridge and an author of the report said: ‘The UK has talked about making better use of data for the public good, but we have had statements of good intent, rather than action.
‘We need to plan better for national emergencies. We need to look at the National Risk Register through the lens of what data would help us to respond more effectively.
‘We have to learn our lessons from experiences in this pandemic and be better prepared for future crises. That means doing the work now to ensure that companies, the public sector and researchers have pathfinder projects up and running to share and analyse data and help the government to make better informed decisions.’
Professor Lawrence added: ‘Mobile phone companies might provide a huge amount of anonymised and aggregated data that would allow us a much greater understanding of how people move around, potentially spreading the virus as they go.
‘And there is a wealth of other data, such as from transport systems. The more we understand about this pandemic, the better we can tackle it.
‘We should be able to work together, the private and the public sectors, to harness big data for massive positive social good and do that safely and responsibly.’