People behaving badly

Ben Goldacre's "Bad Science" is always fun to read, but this week he has a particularly relevant message for editors:

In areas of moral and political conflict people will always behave badly with evidence, so the war on drugs is a consistent source of entertainment.

Picking on any flavor of War On is always a good way to win a round on the house around here, and Dr. Ben does a particularly fine job on this one ("At the point where mild cocaine use was described in positive tones the Americans presumably blew some kind of outrage fuse"). But for us the takeaway point is the first one: "In areas of moral and political conflict, people will always behave badly with evidence."

Yes, that's George Will and his followers inventing assertions about first-person pronouns, or Brent Bozell making up things about other people's data, or Fox News stoking fear from nonprobability samples. But it's also the AP warning you that Twitter makes you stupid or the wrong kind of music will turn your kid into a sleaze. People do bad stuff with data for evil reasons and for kinder ones; our job is to cut down on that sort of misbehavior, no matter who does it or how pure their motives.

So when a story that lands in your basket starts dragging out evidence to back up its lede, you need to ask it a few things:
What kind of data would support that assertion?
Is that the kind you have?
Well ... does it?*

If a story can't answer these three questions to your satisfaction, kick it off the bridge.

* "Do they," if you insist. Doesn't help to have the right stats if they don't say what you proclaimed they do.