One of the fundamental ideas of scientific thinking can be summed up in the phrase: “Correlation does not imply causation.”A simpler way to say that is: “this follows that” does not mean “this follows from that.” Even in science, it’s easy to confuse the two.
Let’s look at an example. Suppose you notice that sometimes the tar in the pavement outside your house is soft and pliable, while other times it’s hard and brittle. This is an observation of the world, what a scientist would call “a datum”.
By checking the tar every day for a reasonable period of time, you may discover a pattern: whenever the kids next door play baseball, the tar goes soft. Repeat the experiment to be sure: yes, you will find, there is a statistically significant linkage between these events, no matter what method of data analysis you use. Have we just proven that the baseball affects tar?
No, we haven’t; and the reason is that the connection here is still one of correlation. Our observation of the tar and the kids is entirely valid scientific practice, as far as it goes; however, the leap to causation — that is, to the conclusion that baseball causes tar to soften — is invalid.
You’ve probably guessed already what the real link in our scenario is: both these things happen when the sun comes out. But the distinction between correlation and causation in less obvious situations is often a tricky one — and something scientists have to keep in mind all the time.