Economists Do It With Models

Warning: “graphic” content…

Bookmark and Share
Correlation Versus Causation Again, In Cartoon Form…

September 8th, 2010 · 3 Comments
Fun With Data

Many of you are familiar with my favorite correlation versus causation cartoon. I have to admit that this one is a sick and twisted close second:

(From Saturday Morning Breakfast Cereal)

The general correlation versus causation problem is that when we see events A and B occurring together or one after the other, we cannot conclude that event A caused event B. We can’t do this because there are two other possibilities for an explanation- it could be that event B caused event A instead (admittedly difficult when event B happens later), or it could be the case that some renegade event C caused both A and B. This last case is particularly important because it highlights the possibility that events A and B have no direct connection.

In the example above, event A is the lights being turned off and event B is the guy getting shot. (Apparently my theory that people don’t like being lectured to has some merit.) It should be obvious which of the causal explanations is most likely in this case. (If it’s not, allow me to directly advise you to not enter a dark room with a serial killer.) This reminds of the time a friend of mine pointed out that she’s not scared of water, she’s scared of drowning- her argument was that it’s not like she looks at a glass of water and runs away screaming. This makes perfect sense, and yet how many people claim to be afraid of the dark???

Tags: Fun With Data

3 responses so far ↓

  • 1 Tim // Sep 10, 2010 at 4:38 am

    This cartoon reminds me of something that happened a number of years ago. I was working in Chicago that summer and the city was experiencing a terribly hot summer. The city was doing what it could by opening cooling centers, asking people to check on elderly neighbors, etc.

    Several weeks into the heat there was a news story about a report, I believe by the coroner’s office, regarding the number of “heat-related deaths.” It gave the impression that deaths were counted as heat-related as long as they occurred during the heat wave. If it was hot out when one got hit by a car, it was heat-related. If it was hot out when someone was shot, it was heat-related.

    I suspect it wasn’t intended that way by the coroner’s office, but that’s the way it was reported.

  • 2 Joao Pedro Afonso // Sep 12, 2010 at 12:04 pm

    I must thank you Jodi for stressing this “Correlation Versus Causation” thing. Not that I wasn’t aware of it, but the way you stress it, empowers me to be more attentive.

    For instance, one of the hot cakes in the newspapers in my country is a study from IMF and another from the world bank, where they found correlations between public debt levels and impoverishing, particularly some turning points where the rising of the debt is followed by decreasing in the potential rising of the GDP (whatever that word “potential” means). And of course, the title of the news were “Debt causes impoverishing”!

  • 3 Kristian // Jan 15, 2015 at 10:01 pm

    Wonderful post! We are linking to this great content on our site.

    Keep up the great writing.

    website; Kristian,

Leave a Comment