Fat, Money, Correlation, Causality
Yet another tool for teachers
Many students (and many former students) are confused by statistics. Happily, once they've grasped the basic idea that "correlation does not imply causality," the rest comes more easily. Here is a new, fun tool to help students think about correlation and causality.
Clever and Delicious
The tool is cleverly described as being a study performed at the University of Michigan. The university issued a press release in which they describe it this way:
In the study, one of the first to examine the economic effects of obesity on mature men and women, the researchers examined a wide range of demographic, physical and mental health characteristics to see whether these factors explained the economic differences between obese and non-obese women.
I may be taking a liberty here, but I'm pretty sure that this study is really a deliciously clever teaching tool. It presents a conclusion that is both eye-catching and absurd. It is thus a splendid thing to pique students' curiosity.
How to Use This Study
First, point out what the researchers said they set out to do: "to examine the economic effects of obesity on mature men and women..."
Then read aloud the beginning of the press release:
Economic penalty of extra pounds to middle-aged women
WASHINGTON, D.C.---Extra pounds can be expensive for middle-aged women, according to University of Michigan researchers analyzing data on more than 7,000 men and women in their 50s and 60s.
The researchers, who are at the University of Michigan Institute for Social Research, set this up cleverly. Ask students to read the entire press release. (They can find it on-line at <http://www.umich.edu/~newsinfo/Releases/2000/Nov00/r111700c.html>.) Then point out to them a few key pieces of data:
Average Adjusted Individual Net Worth [Women]
Normal to overweight --- $225,973
Mildly obese --- $247,140
Moderately to severely obese --- $90,303
And then read aloud the clear, simple conclusion:
[O]besity is economically burdensome for women. This may be due to cultural norms of attractiveness, which stigmatize obese women in a variety of ways," the researchers conclude.
Some students, who have never had much reason to think about it before, believe that correlation and causality are pretty much the same thing. This study shows a clear correlation between fatness and wealth. And of course it demonstrates no causality at all. If any of your students have difficulty seeing this, merely ask them some of the basic questions from the AIR Teachers' Guide:
Can you think of even one different explanation that works as well or better?
Did the test really, really, truly, unquestionably, completely test what the author thought he/she was testing?
Is the scientist ruthlessly honest with him/herself about how well his/her idea explains everything, or could he/she be suffering from wishful thinking?
Even the dullest student will enjoy coming up with lots of alternative explanations why "individual net worth" seems to correlate with fatness. And even the dullest student will enjoy seeing how the study ignores all but one possible explanation.
Too Too, Too (Two)
Let me repeat that I am sure the authors of this study did the whole thing as a clever, amusing tool to help students see how easy it is to assume -- perhaps very mistakenly -- that if two things are correlated, one MUST be caused by the other.
If you'd like to further amuse your students, and perhaps spur their interest in history, too, then ask them to write a short essay. In this essay, they should: (A) describe the conclusion reached by the Michigan researchers; and then (B) compare and contrast it with a statement made long ago by Wallis Warfield Simpson, the Duchess of Windsor: ''you can't be too rich or too thin.''
For even more amusement, you can add a little extra to the assignment: Ask students to: (C) track down press reports about this study, and then (D) notice how many of those press reports say that being fat causes people to be poor.
© Copyright 2000 Annals of Improbable Research (AIR)
This is a HotAIR exclusive feature. For a complete
list of the others, see What's