Why you sometimes need to break the rules in data viz
It’s easy to visualise data badly. Choose the wrong chart type, omit labels or overload the graphic and you can quickly end up with something that is confusing, misleading or that fails to tell a story. To combat such crimes, a number of best practices have sprung up as the field of data visualisation has matured. From steering clear of pie charts to getting rid of “chartjunk”, these are the conventions anyone dabbling in data viz should learn first. But sometimes it’s helpful to revisit the things we learn early on with the benefit of experience. In the six years that I have been making data visualisations, I’ve found that sticking doggedly to the “rules” can sometimes impede effective visual communication. It’s easy to get bogged down in following technicalities, and lose sight of what actually works. The trouble is that best practices tend to have little nuance, and there are times when they may be irrelevant, or worth the cost of breaking. Since joining The Economist I have broken many of them myself. But if you are going to do the same, it’s helpful to understand why they became rules of thumb in the first place. As the old saying goes, you have to know the rules before you can break them. With this in mind, and with the help of The Economist’s archive, I’ve re-examined five things I was taught not to do when I first started out in data visualisation.The Economist occasionally gets complaints from readers about charts where the numerical axis doesn’t start at zero. It’s no wonder they are treated with suspicion. A truncated axis can cause small differences to appear larger, or make a rise or fall look more dramatic. The misleading use of broken scales — particularly by political campaigns — has given axis-breaking a bad reputation. But sometimes small variations between data are important, and you need to limit the range of the chart in order to make the story visible:
0 Comments