About a year and a half ago, I wrote an article in Nightingale about why I stopped using box plots and why I now use alternatives such as strip plots and distribution heatmaps instead. That article created quite a stir, with many readers agreeing and some disagreeing (which is great!). Among those who felt that box plots could sometimes be the best choice, a common objection was that “there are no bad chart types; each chart type has situations in which it’s the most effective choice.”
When I was developing my Practical Charts course, one of my goals was to define specifically when each of about 60 common chart types should be used, i.e., to define the specific circumstances in which each chart type would be a more effective choice than any other chart type. I managed to do that for about 50 of those chart types, but there were about a dozen that I struggled with (including box plots). Try as I might, I couldn’t come up with even hypothetical, cherry-picked scenarios in which those chart types seemed to be a more effective choice than others.
Does that mean that those chart types are “bad”? No, they’re all useful because they’re able to feature patterns and relationships within data. What I am suggesting is that, for those dozen or so chart types, alternative chart types are virtually always more informative, easier to read, less likely to misrepresent the underlying data, or some combination thereof. I don’t think those chart types are “bad”; I just think that there are better alternatives in virtually all situations, so there’s no reason to use them.
Now, it’s entirely possible that there are situations that I haven’t thought of in which those chart types would be the truly most effective choice. That’s why, for many of those chart types, I’ve asked literally thousands of workshop participants and blog post readers if they can think of situations that I might have missed. I’ve written articles about two of those chart types (box plots and bullet graphs) and will publish articles about the others over the next few months to ask as many people as I can to think of scenarios that I haven’t considered, in which a chart type that I don’t currently recommend would be a more effective choice than any other chart type.
Workshop participants and blog post readers have already sent me a fair number of example situations, but they almost always only featured the chart type that I was concerned about, without comparing that chart type to alternative chart types showing the same data in the same situation, so they weren’t particularly convincing. Until someone shows me at least one plausible situation in which a given chart type is a truly more effective choice than any other, I can’t really recommend using that chart type. How could I, if I can’t provide a single scenario (even a made-up one) in which it seems to be a more effective choice than other chart types?
Why do some people balk at the idea that some chart types might never be the most effective choice, and insist that, for every chart type, there must be situations in which that chart type would be the most effective choice? Well, I suspect that it has something to do with the fact that it’s easy to forget that chart types are just human inventions, like printing presses and electric toothbrushes; they aren’t fundamental properties of the Universe, like mathematical principles. Like any other human invention, some chart types are more useful than others, and the fact that a chart type was invented doesn’t automatically mean that there must be situations in which it’s the truly best solution. Indeed, the pantheon of human inventions that are the truly best solution in exactly zero situations is well populated.
This is probably a good time to mention that there’s a big asterisk beside anything that I say about chart types, which is that I’m talking about “everyday” charts for communicating insights as quickly and clearly as possible in reports and presentations. Charts can be used for all sorts of other purposes, however, such as for data art, entertainment, humor, advertising, experimentation, and research. For those other purposes, it’s entirely possible that the chart types that I don’t recommend for “everyday” charts would be good choices.
If you want to continue using box plots, bullet graphs, or any other chart type that I currently don’t recommend, that’s OK. I’m just offering my opinion and won’t call you out if you choose to use one of those chart types. I might not understand your choice, but I’ll respect it.
You might also decide to use one of the chart types that I don’t recommend because that’s what your audience is used to seeing. That’s also fine, but I’d suggest that your audience might have been better off if, in the early days, they’d gotten used to seeing a different chart type. “The chart type that people got used to seeing” isn’t the same thing as “the most effective chart type for a given situation.”
I will mention, however, that a concern I have with some chart types like box plots, connected scatterplots, and radar graphs is that they can make chart readers feel stupid because those chart types are, IMHO, pointlessly hard to understand compared with simpler chart types that can communicate the same insights. I’m not a fan of making people feel stupid for no reason, so I might suggest an alternative chart type when I see those particular chart types being used.
What do you think? Is there such thing as a “bad chart type”? Chime in in the comments or on LinkedIn or Twitter.
By the way…
If you’re interested in attending my Practical Charts or Practical Dashboards course, here’s a list of my upcoming open-registration workshops.