If you thought this blog was dead, you were almost right. Lately, lots of things have felt much more important than expressing my opinions on visual journalism – and they still do. However, a couple of weeks ago I realized that without having me and likeminded persons to defend them, piecharts seem to be going through a veritable shitstorm – so here I am, coming to the rescue, even though I am afraid it may already be too late.
From a recent Twitter conversation (emanating from an admittedly terrible pie chart, shown below), I understood that according to conventional wisdom, no one is able to read a pie chart correctly because angles are supposed to be so much harder to judge than lengths. Pie charts suck; bar charts are simply better, seemed to be the general conclusion.
Quite a fun read for someone old enough to remember when everybody hated bar charts. Someone like me, for instance.
One of the books that really influenced the way I have been working with, and taught, data visualization over the years is Say It With Charts by Gene Zelazny, the Director of Visual Communications for McKinsey & Company. I was totally convinced by his approach and have been using it ever since.
Instead of using different chart types as his theoretical starting point, Zelazny lists five different types of comparisons. Each type of comparison can be visualized with at least two possible chart types (often three or more). When comparing components, Zelazny says, your options are 1) a pie chart or 2) a stacked bar chart, and one is not basically superior to the other. It depends on which relations you want your audience to notice.
Comparing items is a whole different ball game. When comparing items, what you want your audience to see are the exact values and the relations between them – not their relation to a whole, as with components. Trying to use a pie chart to represent this kind of comparison would be plain stupid; a bar chart, horizontal or vertical (the latter should rightfully be named a column chart) is almost always the right choice.
Trying to define which one is ”better” – a pie chart or a bar chart – would be like discussing whether a hammer is better than a screwdriver. It all depends.
And to those who insist that most people find it hard to judge which slices of a pie chart are bigger vs smaller, allow me to quote another one of my gurus – Dan Roam, in his modern classic The Back of the Napkin: ”If you’ve ever been to a kindergartner’s birthday party, you’ll have seen that six-year-olds have no problem picking the biggest piece of the pie. If they can figure it out, so can we”.
The object of the Twitter discussion I mentioned was this pie chart from Berlingske. And true enough, this chart fails in so many ways. For a start, the pieces do not add up to 100 percent. And there’s too many slices. And so on. Almost everything’s wrong here.
However, writing off an entire chart category because someone has been misusing it would be like giving up cars because of one bad driver. So please, could we try to make this discussion just a tiny bit more nuanced?