As an analyst, you spend time meticulously collecting, parsing, and interpreting data, but if you can’t properly communicate your interpretations to decision makers, all of your effort may fall on deaf ears. To help you avoid making some common yet preventable mistakes, I’m going to walk you through the top three difficult-to-see sins of data visualization.
Visualizations are a translation device – they allow technical-minded analysts to harness incomprehensible trends and figures and transform them into palatable images. The easiest way for key decision makers not fluent in the language of analysts, to easily digest complex and contextual data is in image form, such as a bar chart. This means that misleading visualizations can do more harm than good for an organization if important decisions are based on faulty depictions of data.
Unfortunate Ways to Visualize Data:
1. Pie Charts
The bursting of the dot-com bubble is generally attributed to the proliferation of pie charts. The human eye has a difficult time perceiving relative area in a circle; a quarter is distinguishable from a half, but 30% and 40% slices are generally indistinguishable from one another. The only thing worse than the pie chart is the “pie of pie chart,” which is the visualization equivalent of The Eagles. You can see a pie of pie chart below, but it is NSFW(Not Safe For Work) which means you should never use it for work purposes.
I’m sorry you had to see that, but some lessons need to be learned the hard way.
What to use instead: Stacked Bar Chart!
Usually, pie charts are chosen to display a distribution of data at a single point in time. A stacked bar chart can serve this purpose, while also improving legibility. The fundamental superiority of bar graphs over pie charts is rooted in a fact of visual perception: we can compare the 2-D positions of objects (such as the ends of bars) more easily and accurately than we can compare the sizes or angles of pie slices.
2. 3D Graphs
Excel is singlehandedly responsible for one of the most egregious trends in corporate culture in the last decade: visualization for its own sake. The 3D graph options in Excel have allowed analysts to really take their reports to a whole new dimension… for absolutely no reason at all. I’m not talking about visualizing three-dimensional data – the use of the Z axis is defensible in certain situations, such as science perhaps.
Clean, legible, and consistent visualizations are aesthetically pleasing and they also efficiently and correctly communicate insights to decision makers. If you find yourself using 3D graphs to “really spice things up”, please find other ways to make the data interesting.
What to use instead: 2D Graphs!
The great thing about the first two dimensions is that they are perfectly sufficient for data visualization. Just because Excel allows you to be that guy doesn’t mean you should be. If you truly have a legitimate reason to breach the third dimension in order to visualize some complicated math or physics material, please proceed!
3. The Overachiever Graphs
Hey, I just had a sweet idea – what if we made a new type of graph that looked like a bonfire and the height of each flame corresponded to how dumb these graphs are? Using these unfamiliar, newly-concocted visualizations falls under the same category of offense as the use of the 3D Graph: it is unnecessary. If you put a radar chart in front of someone, you’ll have to explain what a radar chart is before you even start explaining how to interpret it. The general rule of thumb is as follows: if it doesn’t actively enhance understanding of the analysis, don’t waste everyone’s time by including it.
What to use instead: Normal graphs!
If you’re illustrating data at a single point in time, use a bar chart. If you’re showing change over time, use a line graph. If you’re comparing timeframes or other qualitative differences, use a column chart. There is no shortage of information out there on how best to visualize data in different circumstances. As you can see, there is also an abundance of ways to poorly display data. Instead of pie charts, 3D graphs, and who-the-heck-knows-what-type graphs, be considerate and display your data in a coherent, efficient, and ‘pleasing to the eyes’ manner. Remember, it’s more about the analysis and the actions you take from that analysis, so don’t trip yourself up with a bad visualization that derails the true meaning for the presentation in the first place.
Data Visualization Wrap-Up
To wrap things up, I leave you with what is widely regarded as the “Best Visualization of All Time,” a depiction of Napoleon’s march to Moscow created by Charles Joseph Minard in 1861. It displays SIX datasets in TWO dimensions: the number of Napoleon's troops, the distance traveled, temperature, latitude and longitude, direction of travel, and location relative to specific dates. As impressive as this is, it’s a perfect example of what your reports should not look like. Aim higher.
Comments? Questions? Leave your thoughts below…
For further information or assistance with data visualization, I can be reached at firstname.lastname@example.org