The Choking Game: A Lesson in Correlation and Causation


There is a striking correlation between the decline of pirates and increasing global temperatures. Obviously the solution to combat global warming is not to replenish the population of pirates, but this is a great example illustrating that correlation does not always imply causation. Sometimes a correlation is so ridiculous that no one would even consider equating it with causality (another equally ludicrous example is the correlation between ice cream sales and crime rates). Sometimes it’s a bit more subtle, like the correlation between the use of marijuana and other drugs, and the accompanying theory that marijuana is a gateway drug bound to turn all users into ravenous heroin addicts (while this theory is not supported by science, those responsible for making and enforcing drug policies can’t seem to leave it behind). But either way, it’s always important to try and tease out correlation from causation.

Photo courtesy of:

Some studies are intended to look only at correlations, and these studies can certainly provide useful information. Other studies are designed to test causality, and these usually involve control and treatment groups where the variable in question can be manipulated. One recent example of a purely correlative study is the story on the “choking game” that’s been all over the news lately. For those unfamiliar with this game, there is a trend—not new, but one I somehow missed out on in my youth—where kids choke each other with the goal of achieving a temporary high due to lack of oxygen. Obviously when strangulation is involved, there are dangers—there were 82 related deaths reported in the media between 1995 and 2007.

Realizing how dangerous this game can be, some public health researchers decided to look at information from a survey given to 8th graders in Oregon to try and figure out what types of kids were playing the choking game. The survey included questions about whether and how often the students had engaged in various risky behaviors, including the choking game. The researchers wanted to identify other characteristics that could help them predict which children would be more likely to play the choking game, so they could better target public health messages about the dangers of the choking game to the kids most likely to try it.

They found that kids who had tried the choking game were also more likely to have experimented with sex or drugs compared to kids who had never played the choking game. This is useful information to help identify kids likely to try the choking game and inform them of its dangers before they do, but it unfortunately tells us nothing about whether playing the choking game actually puts someone at risk for other risky behaviors in the future. Nevertheless, this is the sensationalistic spin that major news sources thought would attract the most readers. Most likely, a kid who experiments with the choking game is the same kind of kid who is going to experiment with other things like drugs or sex. There is no reason to think that preventing kids from playing the choking game will make them any less likely to use drugs or have sex, and an idea like this is exactly what you cannot prove in a study designed to measure only correlations.

Playing a game involving choking is obviously fraught with risks, and I don’t wish to downplay the dangers inherent in such an activity. Accidental and completely preventable deaths are a tragedy in their own right, but we shouldn’t pretend that eliminating one risky behavior will eliminate them all. Next time you read about a correlation in the news, just think of the pirates.



Add new comment

Filtered HTML

  • Web page addresses and e-mail addresses turn into links automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <blockquote> <code> <ul> <ol> <li> <dl> <dt> <dd> <p> <div> <br> <sup> <sub>
  • Lines and paragraphs break automatically.

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.