This article is a great example of how correlation might be suggestive of causation, but is not itself causation.

Are pickup trucks inherently more dangerous than other vehicles? Or are pickup truck drivers more dangerous than other drivers? Or are pickup trucks more likely to be driving on more-dangerous roads than on less-dangerous roads?

Unfortunately I wasn't able to read the article due to the paywall, but I didn't see any mention of controlling for confounders in the abstract.