News coverage gets me irritated whenever statistics are mentioned.
This morning’s example (paraphrasing): “South African gays are upset about discrimination in blood donation; their donations are being refused, despite the fact that the majority of HIV cases in the country are in heterosexual women.”
I’m not wondering if there really is or isn’t a case of discrimination going on there; what irritated me is that there’s actually no valid statistical link between the two clauses of the sentence.
That’s one example; it’s on NPR, so I’d have thought they’d take a moment to get it right. Here’s another example from New Scientist magazine, who should know better (I’m paraphrasing again):
“A survey of accident victims indicates that more of them have cellphones than the general population, demonstrating that people who have cellphones are more likely to have accidents.”
No, no, no. It demonstrates that people who have accidents are more likely to have cellphones; maybe cellphones appeal to people who drive aggressively!
Notice that neither of my examples are cases where the statistic actually disproves the claim made in the accompanying article; only that they do not lend any valid credence to the claim.
What’s your favourite example of – not a maliciously misused statistic, so much – a statistic that, to the educated reader/listener/viewer, does nothing to prove the claim being made?