Reading The Irish Times online this morning we were pleased to see evidence being displayed to support the contention “Economy ‘vulnerable’ due to reliance on few multinationals”. Data from the OECD, CSO, Department of Business, Enterprise and Innovation, and Eurostat were visualised to illustrate some of the points being made. The problem was, we weren’t sure which points, or what was being illustrated.
It got us thinking – wouldn’t the Irish Times readers benefit from clearer visualisations, and what might they look like?
Let’s take just one example, the chart showing Corporation Tax:
Donut charts are often ridiculed by visualisation experts but we see them used widely in contexts where the aesthetics of the visualisation are held to be extremely important – like in publicly consumed newspaper articles. There is something to be said for a visually appealing visualisation and this one seems pleasingly fractal-like, almost like a nautilus. But what is it telling us?
Well, from the label we can tell it’s a distribution. And from the same label we know it’s depicting corporation tax receipts in 2017. And from the label on the largest section we can see that 66.2% is related to >10M, … but we don’t know what that means. Even after careful consideration we can’t tell.
- Is it that 66.2% of corporation tax receipts represent more than 10 million euro?
- Is it that 66.2% of corporations paying tax submitted more than 10 million euro?
- Is it that 66.2% of corporations paying tax submitted on average more than 10 million euro?
- What is the total amount represented by the entire graph.
At a glance, we don’t know
- Is >10 million the largest amount that can be submitted, or is there a higher bracket?
- What is the natural pattern that seems to be emerging from the gradually diminishing proportions.
You might expect that the chart is ordered from highest to lowest amount, but more careful examination proves that hypothesis incorrect. It’s actually ordered by distribution amount – creating the false impression that as returns decrease so do their slice of donut.
- When creating a visualisation, always start off with the question you are trying to answer. When the visualisation is completed, check that it actually answer that question – clearly.
- Humans find it easier to judge length rather than area – distributions are often best displayed using histograms or bar charts, for continuous and categorical data respectively. There are other options, and rules are made to be broken – but you need to know the rules!
- Avoid introducing non-natural ordering as it can confuse your viewers.
- Pie and donut charts only make sense if you are displaying parts of a whole: in this case, we didn’t know what the whole figure was.
Cliff Taylor is an eminent journalist, well-respected in his field, and we are sure that the figures here have been correctly interpreted by him and his team – it’s just that using these visualisations, we can’t actually verify it.
In fairness, it can be hard to produce visualisations to journalistic standards and to journalistic deadlines if you haven’t been taught how to do so. The same can be true of any business reporting role, that’s why we designed our two-day “Effective Data Visualisation” course, running in September. We can show you well-established best practice, and introduce you to some excellent free software to allow you to develop intuitive, precise, meaningful visualisations, good enough for the Irish Times. You might like to come too: https://theanalyticsstore.ie/effective-data-visualisation/
The Analytics Store is the leading data visualisation training and consultancy firm in Ireland. We assist companies in developing and implementing data visualisation and analytics strategies in Tableau, R, D3 and SAS. Our client base includes the leading consultancy firms, banking groups, and many public organisations. Depending on your data visualisation maturity, we offer tailored services that bring you to the next level. If you’d like to talk, you can reach Nina on email@example.com