For reasons beyond those outlined by Bryan Caplan in his Myth of the Rational Voter, its apparent that most voters use irrational voting strategies. Caplan’s analysis is largely centered on how voters make decisions on policies. E.g. voters are biased against foreigners and so lean against liberal immigration even when liberal immigration policies would benefit them. But more importantly, standard voting strategy is so irrational that voters don’t even choose a strategy that aligns with their irrational choices.
The big WHY is that voters, year after year, choose features over benefits, or this case biased perceptions over expected results. Think about the US presidential election in 2000. George W. Bush campaigned on a platform of “humble foreign policy” opposed to intervention and state building. The American people bit, or at least a plurality did, and a new era of humble foreign policy began in which the United States invaded two countries in attempts to set up new American-aligned democracies.
This would be funny if a) it didn’t involve wars and b) it didn’t make so much sense. Its easy to look at the administration with which Bush began office and grant a greater likelihood for military action than the policy Bush peddled on the campaign trail would admit. His administration was full of neoconservative hawks. Now there’s a little hindsight bias at work here, as the world didn’t know prior to the actual Iraq war to what extent the neoconservatives in the administration would support such a war and invasion. Further, Bush didn’t appoint his cabinet until after the election. But he did staff his administration with people who had, only two years earlier, recommended to President Clinton that he remove Saddam Hussein from power. Impeachment was unlikely a viable strategy for said removal.
Strictly with respect to Iraq, Al Gore was probably less likely to invade and mount a full scale occupation. He might have sent troops to all kinds of smaller campaigns much like the Clinton administration did. He might have even bombed an African pharmaceutical factory or two.
Voters ignore the distribution when judging candidates. If Joe Public really likes outcome X, and party A is twice as likely as party B to support outcome X or pursue policies that result in outcome X, then Joe Public should demand an extraordinary amount of evidence from party B’s candidate that he or she will pursue outcome X. Campaign palaver does not constitute evidence.
Voters also ignore the game that informs politicians viable strategies. If Joe Public really likes fiscal restraint, and split government is twice as likely to result in fiscal restraint, he should really vote for split government, regardless of how much he “likes” any of the candidates. You might think all kinds of things about Ron Paul, but if he were elected the results of a Ron Paul administration would probably be a lot more palatable to the electorate than the menu of policies and ideas he espouses. As President he could veto bills but Congress could still override his veto, and given the expected propensity of a Ron Paul White House to veto Congressional bills, Congress would have to achieve a great deal of consensus to pass legislation. That might result in more logrolling, but it would probably result in fewer overly grandiose programs (the kind replete with unintended consequences, like foreign invasions or national testing schemes that bloat administrative costs).
As it stands, the process of developing these expected outcomes and the information required are probably too much for even the “above-average” voter. But keeping these things in mind would do a great deal to temper voter dissatisfaction post-election.