So, I found this thread on Twitter:
If you dig into it, the whole thread shows how you can start with a headline-grabbing yet misleading bit of data, then gradually burrow down into the reasons why it’s misleading, how choosing slightly different data tells a a different story, and why the outcomes based on that data are also not what they appear.
It’s a great use of visualisation, that just manages to dodge the notorious ft paywall as well.