Politico has an interesting story this morning on the many summer surprises (in other words, the known unknowns) that could impact November's elections. Not to pick on the author, but this is a funny sentence:
The rule is almost ironclad: If unemployment hits double digits, it's difficult to see the party in power having a successful election year.
Not only is that rule caveat-ed to death (almost ironclad... difficult to see...), it's also not really a rule. The only midterm election in the last fifty years with unemployment anywhere near double digits was in 1982, when President Ronald Reagan and the Republicans lost 26 House seats. But that's a data point, not a law of politics. And the president's party has suffered worse House losses six times in the last half century, with unemployment as low as 4 percent.
As political scientist Seth Masket has pointed out, personal disposable income has a more statistically significant impact on midterm elections than unemployment levels.