Avoiding data manipulation

- You have likely heard the famous Mark Twain quote, "Facts are stubborn things, but statistics are pliable." When I teach students how to be more persuasive in their presentations, I spend a lot of time on how to present data and how to use statistics and all the ways that data can be manipulated to say what you want it to say. In this lesson, I'll point out a few common data manipulation methods so that you can keep an eye out for them and know how to look past them. The first is related to false causation, but it goes one step further by using charts to show how two things correlate over time. You're likely to see right through my claim that increases in Nicholas Cage movies cause more accidents in home pools, but the correlation's there. Since 1999, these two statistics have moved really closely. Coincidence? Sure. Causation? Highly unlikely. Another common statistics, manipulation, is to visualize data using a truncated y-axis, which is also called omitting the baseline. This is often used when comparing two or more things where the difference is really not that great. I had a student group conduct market validation activity in which they were comparing three options. Option one had 62 votes. Option two had 54 and option three had 52. On their presentation chart, they omitted the baseline. They set 50 as the baseline rather than zero, which made the difference between the three seem much bigger than it really was. Similarly, you can manipulate the y-axis so that changes look smaller than they are. I once helped a restaurant try to get their costs in line. I knew that they were jumping all over the place so I met with the manager in charge to discuss the issue. He really didn't want attention brought to the recent shifts so he showed me a chart of his cost over time. What he had done was take the y-axis and have it represent a much larger span of costs making the jumps much less noticeable. Also, watch out for cherry picked data that manipulates the x-axis to highlight only areas that you want to see. For example, if sales increased in the first quarter of a year, but A, had been decreasing for two years before, and B, had started to decrease again afterwards, you could have a chart that only showed January through March, thus making it look like sales were doing really well. Cherry picking doesn't have to just be in a graph. Most commonly you'll see it when people are only sharing data that supports their claims. Ask for comparative analysis for a full picture. There are so many other ways that data can be manipulated, often without negative intent. You may even do it yourself. So you should consume all data with care. Look for ways it could be misinterpreted before you use it to make important decisions. Encourage your team to always provide full, accurate data so you can make truly informed decisions and move forward with confidence.