The Alternative Vote explained

With the Alternative Vote referendum fast approaching, here’s an attempt to demystify the process from my guest blogger, Professor John Maloney, an economics colleague and election expert.

Have you had the official leaflet describing the alternative vote system? In one way it’s doing a good job for first-past-the post, by making the counting procedure under AV sound endless. The bottom candidate drops out, their second preferences are allocated, there’s another count, the new bottom candidate drops out, and so on until one candidate has more then half the votes. Well, in Exeter, there were seven candidates last year. That sounds like anything up to five counts (when you get down to two, one of them must have more than half the vote – unless it’s a tie!). Goodbye to election night and welcome to counting week?  In fact it wouldn’t have been as cumbersome as that. Look at the actual Exeter result:

Ben Bradshaw (Labour) 19,942

Hannah Foster (Conservative) 17,221

Graham Oakes (LD) 10,581

Keith Crawford (UKIP) 1,930

Chris Gale (Liberal) 1,108

Paula Black (Green) 792

Robert Farmer (BNP) 673

What’s the first thing that strikes you about these figures, if you put your mind into AV mode? It should be the fact that, even had Mr Oakes received all the second-preference votes of all the bottom four candidates, he would have remained inexorably in third place and the next candidate to be eliminated. In fact, only two counts would have been needed; the original one and then a recount with the second preferences of all five bottom candidates given to either Labour or Conservative.

Who would have won? Ben Bradshaw beyond a doubt. The UKIP vote would have broken heavily for the Tories, but the Green vote would have been largely Labour.  As for the Liberal vote it’s hard to say; perhaps even harder to judge is the redistribution of the BNP vote, which might have split quite evenly between the allegedly right-wing Conservative party and the allegedly working-class Labour Party (and also the alleged protest voters’ party, the Lib Dems, had they still been in the frame.)  But, unless every opinion poll in the year before the election was wrong, a majority of the Lib Dems themselves preferred Labour to the Conservatives.  Quite possibly Mr Bradshaw would have ended up with a bigger majority than he actually got.

But the big imponderable is how far people will use their second, third and fourth choices. Obviously, if no one at all bothered with second preferences, AV would cease to be AV and first-past-the-post would stay in place. But how many people will put down a second choice? And how many people in Exeter would have ranked candidates all the way from one to seven?  An opinion poll last summer found that a large minority of voters would not bother to endorse more than one candidate.  But an opinion poll taken before the argument had even started may not be a good guide to how people behave if AV becomes the reality. Will candidates woo the voters along the lines of ‘if you can’t put me first please put me second, or failing that third – or even fourth, please?’  What they will probably stop doing is to say ‘please put me first, even if I’m your second choice, because only I can beat party X.’  Indeed supporters of AV have claimed that tactical voting  — backing the strongest challenger to the candidate you want to keep out – will disappear, and all first preference votes will now have the additional virtue of sincerity.

There will certainly be less tactical voting. But it’s not obvious there will be no tactical voting. Suppose your order of preference is Tory, Lib Dem, Labour, but you think that the Tory can’t win and that Labour will get the most first-preference votes.  Assuming the Lib Dem second preferences are more pro-Labour than the Tory second preferences, you will want the Tory second-preferences to count. How do you achieve that? By making sure that the Tory candidate, your actual favourite, drops out in the first round. Your (tactical) first-preference vote will be for the Lib Dems.

Is this a likely scenario or a university lecturer being tortuous? Nearer the second than the first, no doubt. But some voters might think like that, and it’s another factor which makes the consequences of AV as unpredictable as the result of the referendum itself.

2 comments to The Alternative Vote explained

  • Robin,
    I am a bit late to the party on this, but a conversation over dinner recently got me thinking. Much of the fanfare around AV is concerned with the elimination of tactical voting—even though, as you rightly say, it does no such thing. Indeed, from the Gibbard–Satterthwaite theorem we know that there is no reasonable voting rule that can eliminate the incentive for tactical voting. This was the point that I made at the dinner table, to an immediate response of “so what?” But I think this is an important issue.

    It’s pretty well-known in economics circles that—when the social ideal is unattainable—the second-best is not always attained by setting all policy parameters as close to their first best counterparts as possible. It is therefore not obvious to me that, absent a way to rule-out tactical voting, we should be trying to minimise such behaviour—especially given that there are other dimensions to a voting system’s performance, such as the probability of a hung parliament. In my view, this argument has a particular resonance in light of the fact (Arrow’s theorem) that, even if people could be forced to vote sincerely, there is no way to turn all of those honestly revealed preferences into a decision rule that fairly represents the decentralised views in society.

  • John Maloney

    Greg, I entirely agree. The AV crowd were making out that a) ‘sincere’ voting is more virtuous than tactical voting b) tactical voting will be eliminated with AV. In my view a) is just as incorrect as b).

Leave a Reply

 

 

 

You can use these HTML tags

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

*