A Picture Can Say 1000 (Misunderstood) Words
Image by Charles Minard (1781-1870) - see upload log, Public Domain, https://commons.wikimedia.org/w/index.php?curid=297925
Warning, lots of 'graphic' content…
Raise your hand if you know who Edward Tufte is.
Raise both hands if you have one or more of his books on your shelf and have attended one of his seminars.
High-five your neighbor if you also have the famous graphic shown above (Mindard's Map of Napoleon's March to Moscow) framed and hanging in your office.
If you're not familiar with E.T.'s work, it is definitely worth some exploration time. This article won't come anywhere close to the eloquence and informational value of one of his books or seminars, but then this is meant to be a short blog post for busy people. And, I need to get back to running my business and learning about coding algorithms in Q# for quantum computing…
Next time you look at a graph, ask yourself if it really answers your questions, or raises more. Assume that you are missing something important about it. Imagine how different it might look if the Y axis was in a different scale, or started with a number other than zero. Determine whether you are looking at values, or changes in values, or cumulative values. If it's a 3D graph, consider what it might look like in 2D and know that 3D can distort proportions. If there are two series, are they shown on the same or different Y axes, and are both Y axes shown as log? If you're looking at a time series, what are the scale and scope of the X axis?
Since this post is about visualization, here's a series of eight graphs I created to demonstrate the point without using 8,000 words. Note: they are all depicting the same S&P500 and Consumer Price Index historical data, in different ways.
My personal favorite is the pair of bar graphs, where the Y axis starts at different values on each one.
I'll leave you with one final point: distorted graphs are not always created with malintent. On the contrary, many are made that way simply because the person making the graph thinks it is more visually appealing one way vs. another. So yes, assume you are missing something. But no, don't assume that a bad visualization indicates malintent.
Take some time to learn more about Edward Tufte's work. And reach out to Cybele Data Advisory for help improving your organization's graphing skills!