How to Forecast an Election

In a long and detailed post, Nate Silver argues that “fundamentals”-based models—which rely on information about the economy and foreign affairs—are mostly inaccurate when it comes to forecasting elections. It’s hard to excerpt the post, which you should read, but here is a key passage:

The “fundamentals” models, in fact, have had almost no predictive power at all. Over this 16-year period, there has been no relationship between the vote they forecast for the incumbent candidate and how well he actually did — even though some of them claimed to explain as much as 90 percent of voting results.

His broader argument—which, presumably, will be detailed in his forthcoming book—is that accurate election forecasting requires modelers to take “horse race” polls into account. When you include presidential approval ratings and other related data, the odds of an accurate prediction increase dramatically.

The question of how to forecast an election is fascinating, and I look forward to what Silver finds in his work and research. That said, I can see why some writers might lean heavily on the economic determinism that he critiques; the opposing view—that campaigns are all that matter in elections—is extremely prevalent, and far more impactful when it comes to how elections are covered. You can go too far with it, but there is value in emphasizing the huge role the economy plays in affecting outcomes—it can help turn attention away from frivolous campaign coverage, and toward the broader trends that matter more for how people vote.