The Swing Voters Curse: The reason the Lib Dem vote didn’t live up to the polls at the General Election?
1 April 2011 4 Comments
The final poll before the general election had the Lib Dems on 27% but the final results were disappointing. Re-polling after the election has found that those who said they would vote Lib Dem continue to say they voted Lib Dem. So either all the polling companies got it wrong or something else was going on. This experiment gives us a new way of looking at the 2010 general election for the Lib Dems.
The conundrum of why so many people vote in safe seats has left some to postulate that many people vote because they think that their vote may count; they see themselves more as a ‘swing voter’. However, many people go to the voting booths and still choose not to vote and so this left researchers looking at why.
Researchers at the Center for Experimental Social Science at New York University looked into this by testing what is known as the ‘swing voters curse’. The swing voters curse theory was taken from an economic model of human behaviour: The swing voter knows that their vote could decide an outcome. But they also know that information is “asymmetric” in elections, meaning that other participants might have slightly better or slightly worse information about the candidate. The swing voter doesn’t want to pay the post-election cost of picking an ill-suited candidate based on incorrect information. So the swing voter is “cursed,” stuck trying to get the best result without paying too much in the end. The voter can’t lower how much his vote counts and so the swing voter simply chooses not to vote, and instead delegates it to those he thinks might be better informed.
Using volunteers in a computer-based election their results showed that voters with no inside information about the ‘candidates’ chose to abstain 91 percent of the time, essentially delegating their vote. If you lack information, you are less likely to vote.
However, one unexpected conclusion that the Curse creators derived from their theory is that the hypothetical swing voter, who would normally refrain because he lacks information, will vote when he knows partisans are present. Partisans are those who will always vote for one particular candidate, regardless of what information is available.
In subsequent rounds of the computer-based elections, the group was told that one ‘candidate’ had a slightly higher chance of ‘winning’ and that a certain number of partisans would vote for a ‘candidate’ no matter what. This time, those volunteers lacking inside information chose to vote against the partisans as well as the ‘candidate’ that was most likely to ‘win’ – up to 58 percent of the time. Essentially, the involvement of partisans increased the likelihood that uninformed voters would participate in the election.
You vote because you know there are people that will vote against the correct thing no matter what, so you seek to balance them out … the voter will do this even if his information about the best choice would normally lead him the other way
If we re-examine the 2010 general election we can indeed see that there was an increase in the number of people who voted as would be predicted by this experiment – a closer predicted result means more people believe they are swing voters and therefore more likely to vote. They were given more information through the TV debates and so this may have contributes to the increase too.
Party 2005 Total 2005 % 2010 Total 2010 % Vote change
Conservative 8,784,915 32.4 10,703,754 36.1 +1,918,839
Labour 9,552,436 35.2 8,609,527 29.0 -942,909
Liberal Democrat 5,985,454 22.1 6,836,824 23.0 +851,370
Total 27,148,510 61.3 29,691,780 65.1 +2,543,270
However, as people thought there were more people voting for the Lib Dems, many may have voted against them with this belief in mind. In swing seats this mattered as the party lost 13 seats (and gained 8). Some of the extra people who voted in the election cancelled out what would have been gains as they voted against the belief that the Lib Dems may win in that seat. Clearly this is a preliminary study and is not refined and so will not translate directly to all elections, but maybe in tight elections this effect has more prominence?
The lessons from this experiment are that in tight elections the more information a voter has about a candidate and party, the more likely they are to vote. But also the more they think other people will vote for a party, the more likely there will be people wanting to vote against them.
Maybe the 27% ‘believed’ support the Lib Dems received in the election is what people would have voted if the ‘swing voters curse’ did not exist.