Post hoc, ergo cut it out

Every now and then, students wonder why an editing class veers into stuff like informal logical fallacies at the expense of another round or three of grammar calisthenics. There are two reasons. First, because news sources like to deceive journalists. Second, and equally important, because journalists are prone to deceiving themselves. And when we don't hold our own conclusions to the same standards we set for others, we also risk deceiving the public -- or, perhaps worse, giving the public reason to think we're stacking the deck.

Here's a real-life example from today's primary coverage. The topic is Clinton's use of attack ads on national security:

The ads had an impact:
• A majority of voters in both Ohio and Texas thought Clinton was more qualified to be commander in chief, according to exit polls.
• A majority of the people in both states who made up their minds in the last three days -- after the ad started airing -- went for Clinton.

Step forth all who said "post hoc, ergo propter hoc." Right. We have no way of knowing whether the ads had any such impact at all. These are correlational relationships, not causal relationships. For the first bullet point: Did the exit poll ask whether voters saw the ad? Did it ask whether the ad changed their view about who's more qualified? If not, on either count, there's no ground for stating a cause-effect relationship. For the second, same two questions (did you see it, and did it make you vote for Clinton), with an added caution. Late deciders have been important at a number of points during the primary process. Why would it be valid to conclude that these late deciders were turned into zombies by one ad? What if a voter settled on Clinton on Sunday, then saw the ad Monday? What about the coven of witches in Austin who drank the blood of a birth-strangled babe, then danced widdershins around an effigy of Obama, mocking his lack of international experience in their strange unearthly croaks?

Post hoc reasoning isn't something you can do if you're the New York Times, rather than some lesser breed without the law. It isn't something you can get away with in an "analysis," rather than a news report. (NPR made the same error in a news report this morning, and it was equally wrong.) The correlation is interesting, but it's the extent of what we know. When we pretend otherwise, we're guessing or opinionating (or both). And that makes us look as if we don't know the rules -- or, worse, that we're willing to bend them for one candidate or another.

By the time it gets to the copydesk, of course, such a logical fallacy is hard to stop. It's been written by a power figure, approved at the center, and approved by at least one authority (likely more) in-house. But an error that's been approved by a lot of people doesn't magically become correct. All it becomes is an error with a lot of powerful friends.

Funny, the McClatchy Washburo (to give credit where it's due, the old K-R side of the office) was justifiably honored as the Iraq debacle grew worse for having withstood exactly that problem: An error of judgment that came to seem so much like conventional wisdom that very few people (or organizations) bothered to point out that it was entirely unsubstantiated. Asking about the emperor's fashion tastes is always in style. Could we try it for campaign reporting too?