It is surprisingly hard to draw clear conclusions about voters’ intentions from electoral results
Posted by hkarner - 11. Juli 2016
Source: The Economist: Free exchange
Subject: X marks the knot
THE voters have spoken, runs an old gag, but what on earth did they mean? In the wake of Britain’s vote to leave the European Union, bereft Remain campaigners have sought to explain away the result. Some argue that Britons would never have voted for Brexit if they had known what it involves. Others claim voters were not really expressing an opinion about the EU, but simply protesting about the state of the country. The outcome, the revanchists insist, was not an accurate reflection of voters’ desires, but an electoral malfunction.
To most, that sounds like sour grapes. But there is a body of academic work that supports the idea that elections often misfire. For one thing, voters can be capricious. In a recent book, “Democracy for Realists”, Christopher Achen and Larry Bartels recount how people in New Jersey were significantly less likely to vote to re-elect President Woodrow Wilson in 1916 if they lived near the sites of recent shark attacks. By the same token, voters seem to punish politicians for floods and droughts, but instead of seeking candidates who plan to spend more time or money preparing for such calamities, they simply unseat the incumbent. They are also myopic, judging politicians’ economic management on the basis of only the very recent past. Their opinions can fluctuate wildly, depending on how questions are asked. Before the Gulf war of 1991, almost two-thirds of Americans said they were willing to “use military force”, but less than 30% wanted to “go to war”.
Messrs Achen and Bartels also show that many people neither follow politics closely nor scrutinise policy carefully. Voters do not always understand the politics of different parties: in Germany, only half of them can place “Die Linke” (“the Left”) on the left-right scale. Many do not even know who represents them: in 1985 only 59% of American voters could say whether the governor of their state was a Democrat or a Republican. The authors present compelling evidence that voters tend to pick a candidate first, then bring their policy views into line with their choice. That is true, they argue, of both educated and uneducated voters.
And then there is the question of how well voting systems distil voters’ intentions. There are mathematical difficulties in aggregating their desires. Most of the time, individuals’ preferences are at least vaguely consistent: if you prefer apples to oranges, say, and oranges to bananas, then you probably also prefer apples to bananas. But this is not true of groups. Even if each individual has consistent preferences, they can vary among voters in such a way that, in two-way races, candidate A would beat candidate B, candidate B would beat candidate C, and yet candidate C would beat candidate A. What, then, is the “will of the people”?
Another problem is that adding extra options may change a group’s choice. Al Gore might have won a straight vote against George W. Bush in America’s knife-edge presidential election in 2000. But the candidacy of Ralph Nader, a left-wing challenger, may have split the left-wing vote, allowing Mr Bush to win.
Devising electoral systems to get around such problems is far from straightforward. Take three fairly simple-seeming principles. First, if voters prefer A to B in a straight run-off, that preference should not change when you introduce C to the list. Second, if voters unanimously prefer pizza to pasta, then pizza must always outrank pasta. Finally, assume voters can at most rank the available options: they can say that they prefer lower taxes to higher taxes, but not how much they care. Yet in 1951 Kenneth Arrow, an economist, proved that it is impossible to devise a voting system that satisfies all these assumptions.
For example, French presidential elections aim to minimise vote-splitting by whittling the field down to two candidates, who then face each other in a run-off. Yet in the first round, vote-splitting remains a problem. This can lead to tactical voting. A communist, say, who supports a radical candidate might nonetheless vote for the mainstream Socialist Party to ensure that a left-winger of some description makes it into the run-off. Building on Arrow’s work, Allan Gibbard and Mark Satterthwaite proved in the 1970s that every common electoral system, even those in which voters can rank their preferences, is distorted by tactical voting.
In principle, referendums are simpler. British voters were given a straight choice between leaving the EU and remaining; there was no complicated menu of options. However, it was not clear what a vote to leave meant. One option is the “Norwegian model”, under which Britain retains access to the single market but must continue to allow unfettered immigration from the EU. Yet many Brits voted out because they wanted to reduce immigration. In effect, “half-in” and “out” votes were lumped together. The result may well have been different had they been separated.
Don’t ask, don’t tell
There is a long intellectual tradition which argues that voters should never be presented with such questions. Indeed, representative democracy is predicated on the idea that many have neither the time nor the inclination to wrestle with the details of policymaking. James Madison, one of America’s Founding Fathers, and Edmund Burke, his philosophising contemporary, argued for a “trustee” model, whereby voters elect politicians to make difficult decisions for them. In the 20th century Joseph Schumpeter argued, more bluntly, that policy should be left to those with the time and skill to get it right. The role of voters is to throw the rascals out if they sense things are going wrong.
What, then, would Schumpeter have said about the EU referendum? On the one hand, he would probably have argued that it should never have been held. On the other, he might well have interpreted the outcome as an angry public, sensing that something had gone wrong, giving the elites a good kicking—just what voting is for.