Correlation and causality
Stress that correlation does not prove causation because it meets only one of
the criteria necessary
to establish causation (covariation). Because it does not establish temporal
precedence, correlational evidence may lead people to confuse effects for causes
(as indicated by the saying, "Insanity is inherited; I got it from my children"
and the question "Why were so many Civil War battles fought at national
parks?"). Because it does not rule out extraneous factors, side
effects may be also be confused for causes. Going over table 7.2 will make the point
that correlation does not establish causation clear.
Another way to establish the correlation does not establish causality is to present a relationship and
then have students come up with causal explanations for those relationships. To make this activity more fun, argue on behalf of incorrect causal explanations. Relationships you could use include
- The positive correlation between ice cream consumption and number of assaults.
- The correlation between divorce and death (this correlation and some interesting interpretations of correlational evidence are available from the following web site: http://christianparty.net/mortdiv.htm).
- Alzheimer's disease and neurofibrillary tangles
- barometer readings and bad weather; mud and rain
- wine consumption and health
- smoking and low grades in college (Darrell Huff describes this example in his book "The Complete How to Figure It"
)
- number of fire engines at a fire and amount of fire damage (William C. Burns suggested this example)
- milk consumption and cancer
- cars and life expectancy
- vaccinations and autism (this
ad might convince some people that correlation is causation ).
Thus, you could show that ad, and then (as calmly as you can), present the
scientific side of the debate (this
article does a good job of summarizing the scientific perspective).
- weight and vocabulary of elementary school students
- women's skirt length and stock prices (Mabry, 1971)
- an argument, based on correlational evidence, that
sports causes crime (given that the argument goes against popular
wisdom, this link should stimulate discussion
You could have students work in groups to fill out a nice
worksheet (by Julie Wright) that helps students understand (a)
that correlation
doesn't mean causation but (b) that, sometimes, for ethical reasons, we are stuck with correlational data.
A humorous way of making the point is to show this short (124 second) clip.
Note that there is also a link to this clip in the bonus powerpoint for Chapter 7.
Back to Chapter 7 Main Menu