Working to make government more effective


Poor chart rating for the government’s coronavirus communications strategy

Gavin Freeguard is unimpressed with the government's data visualisation attempts

Good data visualisation is key to good communication, and graphs and statistics were prominent in the prime minister’s statement on easing the lockdown. However, Gavin Freeguard is unimpressed with the government's data visualisation attempts

The coronavirus outbreak has brought the government’s communications strategy to the fore, with its use of numbers and statistics key to its messaging.

There are many criticisms to be made of the numbers themselves. Statistician David Spiegelhalter, who has politely asked the government to stop quoting his articles, has also criticised the ‘number theatre’ of parading figures that are difficult to measure and are almost certainly underestimates – although the public’s desire for certainty in an uncertain time may also be driving some of this approach.

But charts and graphs, and how numbers are communicated through data visualisation, have also been prominent. There has been some powerful presentation – think ‘flattening the curve’ or the Financial Times’ tracking of mortality numbers. But there have also been some less successful efforts. That includes the slides which accompanied the prime minister’s Sunday night broadcast.

The equation 

The criticism here is an easy one to make. The use of the = and the + makes it look like an equation. It isn’t.

The dial

Dials have finite scales – the top speed of a car, an empty-to-full petrol tank. R – the reproduction rate, or basically, how many other people somebody with the disease will pass it on to – does not really have a finite scale. Indeed, the R for coronavirus was believed to be between 2 and 3 before restrictions were introduced. Those numbers are off the dial. And why is this dial out of 1.67? Why is ‘1’ positioned where it is on the dial?

If you were subdividing ‘1’ into smaller units, you would go for 0.1, or 0.2, or 0.25, or 0.5. For some reason, it is divided into six here. This also highlights the lack of connection between the graphics and what the prime minister was saying – he explicitly references R being between 0.5 and 0.9, which is not something you could show easily on this chart.

You could also criticise the chosen colour scale. RAG (red-amber-green) ratings are difficult for those with colour blindness, and overcomplicates what could be a binary differentiation: all that matters is showing what is below 1 (good) and above (bad).

The alert scale

The newly-announced alert scale has been criticised for using the same colour scale as the one deployed on the R dial, as this implies some connection between the two. This is a fair criticism, since threading of colours through a presentation can be a useful technique.

It is not clear what 2, 3 and 4 represent, while having a five point scale and then placing the country somewhere between two of the points is not exactly best practice. While there are often good reasons for having a five point RAG scale, there is also a suspicion that it creates two categories that are merely there as a way to guard against criticism (5) and complacency (1) and will never be used. 

The map

There is a small risk that people could read this as meaning there is already a flare-up in their local area, should that be where one of the red bubbles is placed on the map. Showing an R number below 1 (which is good) alongside threatening looking coronavirus outbreaks (bad) could also be confusing.

The curve

Some criticism of the curve has focused on the R=1/red area at the start of the chart. What does the red area represent? Is it the journey to the peak or the time that has already elapsed? And why does R=1 at the peak, when we know it was higher?

On the other hand, it is a reasonable representation of the fact that R=1 is a crucial number, and reducing that is a good thing. But it also looks like a visualisation of the actual R level/real data, when it isn’t. Directly linking R to the triggering of adjustments also seems to ignore the other tests the government set out at the start of the presentation, and there could be an implication that measures will be adjusted when R falls below specific values.

Then there are the icons. It may have been useful to use some text to say what they represent. It may have been more useful to simply use text instead.

How government communicates during a crisis matters – and that goes for charts, graphs and infographics as much as the spoken and written word.

Related content