Coincidence, or what?

Now that it's too late to go back and change your answers on the content analysis exam, here's a way of getting at that question about whether and when you should recommend some slightly more outre flavor -- like discourse analysis -- over old-fashioned quantitative bean-counting: How interested are you in rare cases?

Quantitative analysis isn't always hospitable to outliers; indeed, because outliers can do the same thing to your data set that Bill Gates can do to a salary distribution, we have ways of hunting them down and dispatching them. That makes for smoother, and  more reliable, patterns of aggregate data. If you want to catch year-on-year changes in how the War On Terror® looks to most people, you can't put too much weight on something that shows up every few months. The point of content analysis is to catch it when it moves from every few months to, oh, once a month or so.

If you're watching Jon Stewart interview Judith Miller even as we speak, you probably agree that that's a good and valuable trend to monitor. But smoothing out the occasional glitch in the Times, or ignoring the odd foil-helmeted rant, also means overlooking rare cases that do mean something. Hence, when Drudge reported Wednesday that the "Clinton Cash" author had brought some fresh security on board, it's nice to have just moved a pile of stuff into the "random paranoia" folder -- including the February suggestion that CEOs were likely to start disappearing if they crossed the wrong Kenyan. 

You have to admit it's hard to set up a quantitative sampling frame to capture media discourse about state dinners. But -- spelling aside -- you also have to admit it's good to have a way to account for the every-now-and-then "Obamas feast" story.