Communication of Statistics

Matt Triggs discusses the government's handling of the Covid-19 statistics. 

The Government communication strategy throughout the Coronavirus pandemic has been, on the whole, poor. Although there’s much to be said about the messaging, rollout of policy, and format of communication, they are not the subject of this blog, which will instead focus on the communication of statistics.

The UK Statistics Authority (UKSA), the Government statistics watchdog, is responsible for the Code of Practice for Statistics, which the Government and Government departments must follow at all times.  On 11th May, Sir David Norgrove, Chair of UKSA, wrote to Matt Hancock in his capacity as Secretary of State for Health asking for clarity on the definition of testing targets. Although the UKSA didn’t comment when the 100,000 daily test target for the end of April was announced, once it became clear that the Government was not being entirely open with the public (e.g. counting mailed rather than completed tests, including multiple tests from the same people), it became a matter of interest.

Following this, on 2nd June, Norgrove again wrote to Hancock seeking further clarity and reminding the Secretary of State of his obligation under the Code of Practice. This addressed issues with continued confusion between completed and mailed tests, the volume of repeated tests, the proportion of positive tests, and the lack of clear notes and data. In a strongly worded statement, Norgrove wrote that testing statistics “...fall well short of expectations [of the code of practice]. It is not surprising that given their inadequacy data on testing are so widely criticised and often mistrusted.”

In addition to the official watchdog, other notable statisticians have been critical. Speaking on 10th May to Andrew Marr, Professor Sir David Spiegelhalter, former President of the Royal Statistical Society and chairman of the Winton Centre for Risk and Evidence Communication described the daily press conferences as “number theatre” and “not trustworthy statistics” - citing (for example) the daily reported numbers of hospital deaths. 

Indeed, on 8th June, Hancock claimed that there were “no deaths in London hospitals” which, due to the reporting lag, is simply not true. When working with imperfect data (which this definitely is), it is especially important to bear in mind the assumptions and conditions on which they have been collected and to ensure that these are communicated clearly. The experts that were typically present at the daily briefings made a reasonable attempt to express these when voicing over the figures, but Ministers have not been as careful. In fact, there were four deaths - each of whose family would have heard the Secretary of State for Health not just treat them as a statistic, but forget they exist at all.

Even before the current COVID-19 crisis, the Government had a questionable record in the communication of statistics. The 50,000 additional nurses for the NHS pledged in the 2019 General Election illustrates another example of the willingness of the Johnson government to play “fast and loose” with figures. Although not egregious enough to incur comments from the UKSA, it certainly was not a clear and transparent communication of policy and statistics. Although the argument was sound (more nurses would be working in the NHS if you could convince people not to leave), expressing this nuance was seemingly not as important as the snappy one-liner and served only to anger and confuse the voters and the press.

The general public should not have to have a high degree of statistical literacy in order to understand complex official data (with an appreciation of the necessary caveats and assumptions) put in front of them, but the onus should be on politicians and leaders to communicate statistics fairly and clearly.

Statistics should serve good governance and policymaking, and not be used to mislead the public. It might be too much to ask that politicians don’t cherry-pick statistics to support their arguments at election time, but using a smokescreen of data to hide government failings during a pandemic is not acceptable.

Matt is a statistician and data scientist working in financial services. Currently based in Nottingham, his current interests are not going out, remaining socially distanced, and tearing out the longest hair he’s had in 15 years watching daily press conferences.

He tweets at @mjtriggs

Do you like this post?