Here’s something that I can’t unlearn after years of statistical training: when you’re comparing samples of data along some metric, it’s not sufficient to report their averages. You also need to report how much your data varies around those averages. Without information about variability, you might misinterpret your data’s overall patterns. This makes it particularly important to visually mark variability on the plots that we use to communicate findings in a memorable way — for example, with error bars.