For 2012, the unemployment rate doesn’t matter. The change in it does.

How fundamental is the state of the nation’s economy to election outcomes? So fundamental that 75 percent of the time forecasters can correctly predict which party will win the general election without even knowing the candidates’ names.

In fact, they can do this months in advance of the election, before the party conventions. But they’re not looking at what many pundits and politicians think they’re looking at. The conventional wisdom in Washington is that the unemployment rate will likely decide the election. This even appears to be what Mitt Romney thinks. But it’s wrong.

Romney’s speeches are mainly about jobs. “If the president wants to talk about jobs, and I hope he does, we’ll be comparing my record with his record and he comes up very, very short,” he likes to say.

But does it matter that unemployment is at high levels going in to 2012? Intuition leads many people to believe it must matter, but in fact, the unemployment rate as measured in the third quarter of an election year has very little relationship to the election outcome. See the figure atop this post.

The two series are correlated at -.09 and the mainly flat line in the figure shows the direction (or lack thereof) of the linear trend between unemployment rates in election years and election outcomes. The unemployment rate is both unrelated and a poor fit to election outcomes.

Which isn’t to say the unemployment rate is meaningless. But what matters is not how high it is, but how fast it’s changing. While the unemployment rate itself has little relationship to incumbent party vote shares, the change in this rate between the fourth to second quarters is a good predictor of incumbent fortunes. A 1 percent drop in the unemployment rate yields a half-a-point increase in incumbent vote share. That suggests the latest jobs numbers are, just as you would think, very good news for Obama.

But the big question for the Obama campaign isn’t jobs so much as it’s growth. Increasing growth by 1 percent returns 400 percent more (in terms of vote share) than decreasing unemployment by 1 percent. Compare the numbers on the unemployment rate to the trend and fit of economic growth, as measured by the change in the U.S. gross domestic product (GDP) from the fourth quarter of the year before the election to the second quarter of the election year. This relationship is stronger and tighter.

The data suggest that for every 1 percent growth in GDP from the fourth to the second quarter of an election year, an incumbent party can expect to pick up an additional 2.5 percentage points on Election Day. Two-and-a-half points in vote share may sound small, but consider that of the 16 elections in these graphs, five elections (1960, 1968, 1976, 2000, and 2004) were decided by this small (or smaller) a margin of victory.

Lynn Vavreck is an associate professor of political science and communications at UCLA.