Hi Adam, I’m working on an article related to this one about a fallacy I see often related to confirmation bias, where people use “correlation does not imply causation” to disregard information about subjects like climate change. I love this article but I wonder if the statement “there is no correlation without causation” is a little too strong. I find three possible alternatives:
- There’s not enough data. (you can usually mitigate this by collecting more data until coincidences disappear). This shows up in examples like the following: https://www.fastcodesign.com/3030529/infographic-of-the-day/hilarious-graphs-prove-that-correlation-isnt-causation
- There’s selection bias. This too can be mitigated by reviewing your methodology thoroughly. (this may also partly explain the examples in the above link)
- X may be correlated to Y not because X causes Y but because, in a sense, X *is* Y. This case is usually pretty easy to eliminate. Your alarm is not traffic in any sense. It’s just not. A silly example is that harry chins are highly correlated with beards or that there’s a high correlation between how many kilowatt hours someone uses and how many joules they use. A less silly, real-world (although imperfect) example relates to people who believe HIV doesn’t cause AIDS when it’s perfectly obvious that a Human Immuno-Virus causes an Immune-Deficiency almost by definition! (of course, it’s not quite correct to say HIV *is* AIDS, I’m at a loss for a much better example, though).
Anyway, I believe you cover the first one in your article, but that comes far after you make the statement that there is no correlation w/o causation. It makes me uneasy not to see the caveats to that statement given upfront.