Public Health England (PHE) are producing regular updates on the state of play with the Covid-19 pandemic.
They have published charts showing the weekly rate of cases per 100,000 of the population in England, broken down by local authority. The last two weeks at the time of writing (9th October 2020) are as follows – this is the chart for the week ending 27th September:
This is the chart for the week ending 4th October:
The alert reader will immediately spot that the more recent chart presents a much rosier picture than the previous week – a far lower concentration of areas with a high level of cases being reported, which is great news!
Unfortunately not, as the even more alert reader will spot that the scale of the two charts has changed dramatically, by an average factor of 7, in fact. The rate of new cases required to trigger the darker red shading has gone from 45 to 335.
In principle changing the scale is fair enough, as clearly you can’t apply the same break points to two charts when the overall level of cases is rising or falling. If you left the upper ceiling at 45 cases per 100,000 an increasing number of areas will sneak into the dark red and the chart will be meaningless.
However, if you look at the overall number of cases reported in England for the two weeks in question, you find that they have risen from 39,876 to 69,698. This is a factor of 1.75.
Therefore (if we round the figures to sensible numbers) the lowest level on the second chart should have moved from 5 to 10 as opposed to 35, and the upper level should have gone from 45 to 80, and not 335!
That should give you a more meaningful distribution, and enable a valid comparison to be made between the two charts. I have already seen on social media cases where people have shared this data as proof that things are not as bad as some claim.
Fortunately, a good number of people have also pointed out the above inconsistency, but not before some damage is done.
We, as social media users, need to be more aware of these things and public or official bodies, who should be supplying meaningful data without misrepresentation, should be more careful.