Evil Plots

At part of an experimental General Education class, this past fall I taught a two-week course module to non-science majors entitled “Evil Plots” all about the ways graphs can be used to misrepresent data. The idea was to teach students graphics by bad example, and to introduce a healthy level of skepticism to their media consumption (I’m looking at you, Fox News).

The module covered three ways plots can go bad:

  1. The plots itself is faulty – such as distorted axes.
  2. The data is questionable – such as non-representative samples.
  3. The interpretation is suspect – such as spurious correlations.

In this post, I will focus on item three. There is a fun book on spurious correlations  that I recommend. But there is nothing like working with your own data to drive home the point, so I started the class by asking students to answer the following series of pseudorandom questions:

  1. How many hours a week do you spend on school assignments and studying?
  2. How much do you love math on a scale of 1-10 (1=would rather have my teeth drilled, 10=math problems are better than ice cream and kittens)
  3. What day of the month were you born?
  4. What is your height in inches to the nearest inch?
  5. How many days did you spend at the beach this year?
  6. What is the most miles you’ve driven a car in a single day, ever?
  7. How many songs do you listen to in a day?
  8. How many slices of pizza did you eat in the past month (best estimate).
  9. How many states have you visited in your life? (Driving though or stopped in an airport count)
  10. How many letters are in your first and last name combined?

I then hunted for correlation between all the possible combinations of variables. Naturally, most were pure scatter, but sure enough, some exhibited reasonable correlation. For example:


It is possible to predict your height based on the day of the month in which you were born. Who knew!?


This entry was posted in Uncategorized. Bookmark the permalink.

Leave a Reply