I know that not a lot of people read the Herald but the ones at Pacific Quay do and, in this time of disease, they’re getting big fearful audiences again (500 000?).
So, it’s all wrong, once again, and it’s wrong in a way to scare us badly when we don’t really need any further scaring to know how to behave sensibly, as our NHS and Government work to save us.
Even before we get to that graph which has been scaring folk across our MSM, the Herald opens with this arbitrary arithmetic to exaggerate our plight as evidence of reduced risk and greater resilience emerges:
‘The 29 deaths in two days eclipses the 25 death reported in the previous five days. The number of reported positive Covid-19 cases over the past two days stands at 747 -75 more than in the previous four days put together.’
Eclipses? Really? 29 eclipses 25? Like the Sun and the Moon?
It’s true but kind of pointless. We know we’re heading toward a peak in the number of cases and of deaths but what this doesn’t tell us, crucially, is how bad this is. That’s because it has no context such as:
- NHS Scotland is saving many patients with a consistent mortality rate, even as the number of cases grows, of around 3%.
- The mortality rate in England is 8% and in Italy it’s 11%.
- There are clear signs from several sources, even the Herald’s own Helen McArdle, that there will be fewer deaths in Scotland, as social distancing kicks in at an earlier stage in our infection.
- The 50 coronavirus hubs initiative in Scotland looks very promising.
All of that is before we get to the graph:
I used to teach 11 and 12 year-olds how to select the right kind of graph for the data you have. If you’re going to just add how many deaths you have each day, to the total you had the previous day, then a bar graph is appropriate. A line graph is for a trend where the line shows the level that day and NOT the cumulative total. The line graph above suggests, wrongly, that there were 76 deaths on April 1st alone when there were actually 16.
This does not show you how the mortality rate is changing, it just piles them up and it doesn’t put them in the wider of context of what might have been, so that you can judge for yourself, how bad this is. Here is the correct trend represented in a line graph where we, arbitrarily I know, represent where we are, if the total of deaths was to rise to 100:
Or what about 50?
The important thing is that even where you play with the vertical axis, the line graph used correctly, does not distort reality as badly as the one being widely used.