"There are three kinds of lies: lies, damned lies and statistics."
I hate that quote. Or, to be more accurate, I love its humour, but I hate how it gets brandished around like a club.
I was reading about possible lottery fraud, and this comment popped out at me:
Statistics can be shown to mean whatever they want.
Whenever someone wants to dispute a finding that relies on numbers, Twain's quote is never far from the surface. People bring it out as though it were some logical proof or eternal verité, instead of being a punch line.
For the non-scientist, numbers are slippery, unpredictable and untrustworthy things. You can't look them in the eye. You can't question their motives.
For scientists, numbers are the real deal. You dispute the interpretation, the analysis, the experimental design, what have you... but it is highly unusual to dispute the numbers. One saying goes, "Interpretations may change, but the data should be valid forever."
Fortuitously, this article in New Scientist makes a similar point. When climate scientists say "Very likely," they are expressing a confidence that the probability is more than 90%. In comparison, nearly half of non-scientists thought the probability was less than 66%.
Such things tap into of important differences between how non-scientists and scientists approach problems.