First I would like to start out this blog by making a distinction between “data art,” “infographics,” and “data visualization.” I will only focus on data visualization and my suggestions only apply to that. Recently several data visualization practitioners tweeted/re-tweeted this video entitled “Shut up about the y-axis. It shouldn’t always start at zero.” I’d like to issue a formal rebuttal against this video, and the ideas proposed. First of all it’s title is arrogant and worded in such a way as to and shut down any conversation or criticism against it. This is foolish. Anyone who wants to learn more should always welcome criticism and an open dialog should always be supported. A video that starts with “Shut Up…” doesn’t rank high on my list of educational material so naturally I was already very skeptical.
In the video they rightly excoriate Faux News for one of their infamous bar-charts which do not start at zero.
The reason is that when our eyes see a bar chart, our brain begins to automatically compare the vertical lengths without effort. So a bar that shows a 35% tax rate and a 39.6% tax rate should start at zero because our brains cannot help but automatically interpret a line of 35 units and 39.6 units as being ~4 units different. When we see a line-graph our eye follows the line and does not make automatic value comparisons.
The video then goes on to show us (~0:55) a graphic that appears to be some flavor of US employment. The video does tell us what data it is or where it can be found, but it generally it appears to be US employment ages 25-54. This data can be found here: https://research.stlouisfed.org/fred2/series/LNS12300060 There are of course many variations of employment data, (not seasonally adjusted, seasonally adjusted, part time, are people in school or the military included?) but the Vox video does not care to explain any of this. The video then says “Now the share of US adults who have a job has fallen to a generation low, it’s a really big deal. But if you chart this data with the Y axis at zero you can’t see the change at all.” I have significant problems with this statement. First of all, you can see the change, it’s just not very significant! In their own video graphic they show only a ~2.3% decline from 1995 to 2015 from 79.7% to ~77.4% !
You’d never know from this chart that US employment has been fairly stable for a long time.