With the off-year Election Day now upon us, press outlets as diverse as Fox News and The New York Times are continuing to overemphasize the broader significance of today’s contests. (The lead headline at NYTimes.com all day has read, “3 Contests on Election Day Could Signal Political Winds”—with, of course, a caution deep in the text that it is “probably not wise to draw broad lessons from Tuesday’s results.”) But, encouragingly, there’s also been quite a bit of pushback against that narrative from some influential Web sites and, in at least one case, from a leading newspaper.

One of the Web sites is FiveThirtyEight.com, where Nate Silver notes that gubernatorial elections are poor stand-ins for people’s feelings about national politics because, well, people vote differently in state and national campaigns. Here’s Silver:

The correlation between gubernatorial elections and elections to the House, Senate and Presidency has been very weak, at least recently. In fact, if you compare the share of the vote that the Democratic candidate got in the most recent gubernatorial election in each state to the share that Barack Obama got last November, [the correlation] is almost literally zero.

Silver suggests that some modicum of insight might be gleaned from the results of the congressional race in upstate New York’s twenty-third district, though that campaign has been so topsy-turvy that a variety of competing interpretations will seem plausible. And as for the state races, he says:

New Jersey and Virginia don’t have particularly much informational value — we won’t become very much smarter about the future based on what happens there. To the extent that we do learn something, it will probably be hints about turnout, motivation and enthusiasm, rather than something about the electorate’s policy preferences.

Over at The Atlantic, Josh Green takes a different route to a similar conclusion:

The pointless thing about prognostications, even ones as vague as these, is that they can’t factor in any of the changes that could occur in the interim. And there are bound to be plenty of them: major health-care reform seems likely; a weak economy and high unemployment seem possible; and the introduction of a compelling Republican agenda is at least feasible. So let’s take a deep breath and remember to view Tuesday night’s elections for what they really are: A landslide (Virginia), a circus (New York), and a Lions-vs.-Rams battle of bottom-dwellers between two of New Jersey’s least popular politicians.

While it’s good to see Silver and Green doing their best to correct the narrative, they’re writing for political junkies. Are there any mainstream, broad-audience publications that are doing the same? Happily, yes. Here’s the lede of a story by Naftali Bendavid and Anton Troianovski in today’s Wall Street Journal:

Republicans appear positioned for strong results in three hard-fought elections Tuesday. But isolated, off-year contests aren’t always reliable indicators of what will happen in the wider federal and state races held in even-numbered years.

To arrive at this conclusion, the Journal reporters did the obvious thing: they looked back over the past twenty years to see how well off-year results correlated with future outcomes. The finding: sometimes an off-year sweep was followed by a win in the next federal election; sometimes it wasn’t. The way this was characterized in a graphic—“Off-year races: Sometimes predictive, sometimes not”—is, if anything, overly generous; knowing that something is “sometimes predictive” is of little use unless we have some way to divine when “sometimes” is. (That last point was not made by Chris Cillizza of The Washington Post, who writes today that “at least twice in modern memory, off-year governors election in Virginia and New Jersey have served as leading indicators of a wave to come”—while overlooking those times when the election didn’t serve as an indicator.)

• 1
• 2