[The SNP’s] Covid prevalence stats are just not true
The Guido Fawkes blog has published an article claiming that the Scottish National Party (SNP) used data on the prevalence of Covid-19 that isn’t true.
In fact, both sides correctly cited their chosen figures, but the Guido Fawkes article used less reliable data when claiming to contradict the SNP.
What the SNP said
In a tweet on 6 January, the SNP account said: “Latest ONS [Office for National Statistics] data shows Covid is less prevalent in Scotland than England.”
It also included a clip of Stewart Hosie MP on the BBC show Politics Live on the same day [starting 23mins 30 seconds] saying: “The ONS numbers that came out [in] the last day or two show that the prevalence [of Covid-19] in Scotland is markedly lower [than in England].”
This is correct. The ONS had published new data the day before showing estimated prevalence of one person in 20 testing positive for Covid in Scotland the week ending 31 December, compared with one person in 15 in England. (And one in 20 in Wales, and one in 25 in Northern Ireland.)
What Guido Fawkes said
In an article on 7 January, the Guido Fawkes blog said that this claim by Mr Hosie and the SNP was “not true”.
It also posted a tweet saying that it was a “falsehood”.
In support of its point, the article said: “While the SNP cite ONS statistics, the government’s Covid dashboard spells a different picture for cases by nation for the last 7 days.
“Not only does England have a lower prevalence than Scotland, it has the lowest prevalence of all four UK nations despite having almost no legal restrictions. The lockdown lovers always say they’re following the science… except when they aren’t.”
The article included a screenshot from the dashboard showing a lower seven-day rate of cases in England compared to the other nations.
However, Mr Hosie and the SNP specifically said that their claim was based on ONS infection survey data, not on the number of cases on the Covid Dashboard.
Their comments are not false just because the dashboard offers different evidence.
Comparing the data
Guido Fawkes used less reliable data to talk about the true level of recent Covid prevalence than the SNP or Mr Hosie.
Strictly speaking, the dashboard case figures quoted by Guido Fawkes are not a measure of “prevalence” (the share of people who have Covid at a given time). They are a measure of incidence (the number of new cases occurring at a given time).
Case data from the dashboard is also more likely to show a biased picture of Covid throughout the country, because it is affected by the number of tests being done and by decisions about who gets tested. It also currently does not include some reinfections, as it does not include people testing positive more than once.
The ONS survey, on the other hand, covers a representative sample of people in private households.
As Kevin McConway, Emeritus Professor of Applied Statistics at the Open University, told the Science Media Centre last month: “To a considerable extent, [case numbers] can be affected by who is choosing to be tested. If the numbers of people being tested, or the types of people being tested, change over time, then that can introduce potential biases in the trends in numbers of confirmed cases.
“In other words, they might not accurately represent the real trends in infections in the country.”
On the other hand, he said the ONS survey “isn’t affected by biases arising from the number and type of people who choose to be tested in the standard testing programmes.”
Although Guido Fawkes says the government Covid dashboard shows cases by nation over the last seven days, it actually presents the figures for the most recent seven days that complete data is available for, which is how the dashboard publishes it.
The figures therefore exclude data from the four most recent days and refer only to “the 7 days ending 5 days before the date when the website was last updated.” The Guido Fawkes data came from an update on 6 January, so it refers to tests taken between 26 December and 1 January.
The ONS data being used by the SNP and Mr Hosie came from the tests taken at almost the same time, between 25 December and 31 December.