According to conventional political science literature, the state of the economy is the primary variable for explaining campaign outcomes. Gaffes and debates and primary speeches gather a lot of attention by pundits, but they ultimately don't matter on Election Day. 

The most recent issue of PS compiled 13 different articles by political scientists who used different models for predicting the November election. A recent WaPo article summarized their findings. 8 found that Obama would win; 5 predicted a Romney win. The 8 that picked Obama said it would be a very close win. Their findings are widely different from Nate Silver, who has been consistently picking Obama.

If the economy is the primary variable for picking an election, why are there such different predictions between Silver and political scientists? Why are there differences even among the political science community? Everybody is using different variables to describe the economy.

We had a nice chat about the economy and the election yesterday. Today, Derek Thompson explains that the economy doesn't seem to be slowing down Obama, because things are slowly, very slowly, improving. 

This is a crummy economy. And it's nothing to be satisfied with. Our unemployment crisis is tragic. Personal income growth is pathetic. But perhaps the reason the economy has faded as an indicator of doom for the president is that the overall pace of its recovery has been thermostatically mediocre and only-just-barely acceptable to only-just-barely enough voters. As the models predict: Things getting better, slowly puts us in a mood of frustration … but not quite a mood of change. 

Will tonight's debates make a difference? Most of the time, debates don't matter. People have already made up their minds, and only die-hard political junkies really pay attention.

Well, this die-hard political junkie will be paying very close attention. And twittering through the whole thing.