Hello Skeptics.SE!

I've been wondering -- do you know a catch phrase illustrating the "correlation != causation" concept?

I find that some people have trouble understanding it, while a funny example could explain it better.

Something along the lines of "statistics have demonstrated that cancer rates and global warming both have gotten worse in the past 20 years, therefore we can conclude that cancer causes global warming".

(That's not very funny, but it is the kind of absurd example I'm looking for.)

Pictures speak a thousand words: xkcd.com/552

@Jamiec Haha

Yes, that's pretty nice on the internet, but what about an oral conversation?

Wikipedia has a bunch of examples: en.wikipedia.org/wiki/Correlation_does_not_imply_causation

As ice cream sales increase, the rate of drowning deaths increases sharply.
Therefore, ice cream consumption causes drowning.

@Jamiec Oh, I didn't think about Wikipedia. This one is good!

@ChrisW Yes, that's reverse causation. Wikipedia suggests "windmills cause wind" for this.

But I think that the "third factor" one is most common.

@Aeronth I think my example is "third factor" too: the third factor factor being the underlying heart condition, which is the "cause" of the medication and of the heart attack. Reverse causation would be if you were taking medication because of the heart attack.

@ChrisW I see. The distinction between "bad heart condition" and "heart attack" did not seem obvious to me, but I guess your point is valid.

@Sancho: Very good question. I would answer "neither"!

If you try to answer to a level that Deepak Chopra and Mehmet Oz will be blown away by your crystal clear logic, and never again make such claims, you are on a fool's errand.

If you try to answer with in-references, Latin phrases and assumptions that people instantly get the science to an undergraduate level, you may get a few votes from the regulars, and you may convince a few hardened skeptics who had never heard of the claim before, but they probably would have reached the same conclusion anyway, had they ever given it any thought. (Hypocrisy alert: I think I fall for this trap more than I should.)

I think the ideal answer is to address the person who isn't a self-described skeptic, but who heard this claim and was a little interested (or, even better, a little doubtful) and wanted to learn more about it - i.e. hit them before they are committed to either side.

I would love it if each of our answers introduced a new concept to the naive reader, that gave them just one more tool in the skeptics toolkit for the next doubtful claim they heard, rather than presented just a fact for them to memorise for this one situation. But I see this is hard, hard work, and wishful thinking.

@Oddthinking Personally I'd say we should write our answers with the assumption that most readers will be the random person who finds the link from Google based upon whatever search they were doing. So maybe assume a high school education as a bar?

en.wikipedia.org/wiki/Legal_burden_of_proof describes different standards of evidence. Write as if the OP who challenges the claim need prove nothing, and that your answer therefore has the burden of proof. Give the best evidence you can: ideally you can provide at least "preponderance of evidence". Don't strive for "beyond reasonable doubt" until or unless someone cross-examines your answer with an actual reasonable doubt to be answered.

There's a difference between a person being "biased" (having some preconceived ideas or opinions) and being "prejudiced" (having already made their decision). Expect your audience to be biased, but not prejudiced (prejudiced people aren't the 'target market').