It's wrong to call our media innumerate. They're perfectly capable of handling, at the least, simple numbers in a relatively accurate fashion. They just don't want to. And since "race tightening!" is a more interesting headline than "race much the same as it's been in recent weeks," we're getting a lot of coverage about a single Zogby tracking poll showing a McCain surge, despite the fact that six of the seven non-Zogby tracking polls showed an Obama gain during the same period. But it's worth noting that Zogby has a history of making headlines with this sort of thing. Days before the California primary, for instance, Zogby made news with a poll showing Obama ahead by 13 points in California. He ended up losing the state by nine percent. Days before the 2004 election, Zogby got a lot of play in liberal circles for predicting a Kerry win, despite the fact that his own battleground polling showed Kerry trailing. Crucially, Zogby's large mistakes are not partisan in their nature, but sensationalist. Sometimes they show unlikely Democratic surges, other times they show unlikely Republican comebacks. If a reporter had a source who kept giving him spectacular scoops that turned out to be wrong, the reporter would have to stop using the source or risk damaging his own career. But if the news networks have a pollster who keeps giving them spectacular results that turn out to be wrong, they keep using him, because if they stopped their network would be less interesting than that of their competitors and that would damage their careers.
