Two variables can move together in lockstep — r = ±1.0 — yet have nothing to do with each other. Raise and lower the correlation strength and watch the causal story stay fixed.
Sometimes two things line up by pure chance, with no link at all. From 2000 to 2009, how much cheese Americans ate per person and how many people died tangled in their bedsheets rose and fell almost in step (a famous example collected by Tyler Vigen). Cheese does not cause bedsheet deaths, bedsheet deaths do not cause cheese eating, and no third variable drives both: with enough pairs of numbers, some will match by luck. Tests love this trap because the numbers look convincing, but the causal story is fiction. Always ask: is there any plausible link, or could this be chance?