In a recent interview (linked by Todd), Fred Thompson astutely pointed out that people often vote for presidential candidates on the basis of issues that the president has no control over. This is absolutely correct. For example, short term economic conditions often have a decisive impact on the outcome of presidential elections even though presidents have little or no ability to prevent recessions. No modern incumbent president has ever won reelection in a recession year, and no modern incumbent has ever been denied reelection in a time of strong economic growth. Yet short term growth rates are almost certainly caused by factors that presidents have little or no control over.

The problem is not confined to presidential elections. Candidates for other offices also often win or lose elections on the basis of issues that they can't control. For example, a recent study finds that farm state voters routinely punish the incumbent party whenever agriculture is hurt by bad weather - even though state officials obviously can't control the weather.

Why does this happen? After my last few posts, it probably comes as no surprise that widespread political ignorance is a big part of the answer. Because each individual vote makes so little difference to the outcome of an election, voters have very little incentive to acquire even basic information about politics and public policy. Not surprisingly, extensive evidence shows that most citizens have very low political knowledge levels.

As a result, they often rely on crude "information shortcuts" to choose who to vote for. One of the most common shortcuts is what scholars call "retrospective voting" - punishing incumbents when things seem to be going badly. Retrospective voting is not a stupid or irrational strategy. Unfortunately, however, it breaks down when voters punish incumbents for events that are beyond their control - or reward them for positive events that they didn't cause. And highly ignorant voters often find it difficult to tell the difference between those events incumbents have the power to influence and those they don't. They also often can't tell the difference between a bad outcome that could have been mitigated with improved policies and one that would have been even worse if the incumbents hadn't adopted the best policies they could. To take the recession example, they often can't tell the difference between the following three scenarios:

1. There is a recession, but the president can't affect it in any way, positive or negative.

2. There is a recession, but it would have been even worse if not for the incumbent president's good policies.

3. There is a recession, and the president helped cause it or made it worse than it otherwise might have been by adopting suboptimal policies.

Whenever some visible bad event happens, voters tend to assume it is a case of 3, discounting the possibility that it's really a case of 1 or 2.

For more discussion of retrospective voting and its flaws, see this paper I wrote for the Cato Institute.