No to AV wins!

So the referendum has resulted in a resounding No to the Alternative Vote (AV) proposal. Although not surprised, I am very dissapointed. I felt that AV was superior to First Past The Post (FPTP) for a number of reasons. In any constituency were a candidate is supported by a majority of those who voted, the result would be the same under AV or FPTP. However, in any constituency were the vote is divided amongst 3 or more candidates, none of whom have a clear majority (of those who voted), AV is more likely to result in a winner who is at least liked (when considering the preferences of the voters) by a majority of those who voted, while FPTP can easily result in a winner were this is not the case.

Something I didn’t know about, until I started reading about AV and FPTP, is the Condorcet condition. Essentially a voting method satisfies the Condorcet condition if the winner is someone who would beat any of the other candidates in a one-on-one fight. Neither FPTP nor AV strictly satisfy this condition, but it is my understanding that AV is much more likely to satisfy it than FPTP. I think (although I may be wrong) that AV satisfies the Condorcet condition if there are only 3 candidates while FPTP regularly would not. If there are 3 candidates, it is quite likely that 2 of them will split a particular section of the vote (i.e., left-leaning for example) and – under FPTP – the other could win even though they would lose if they were competing against either one of the other two candidates in a one-on-one contest (I keep using terms like “contest” and “fight” which I don’t really like, as elections are not a game or a battle, but I can’t think of other suitable terms). Under AV, however, the second preferences of those who’s first votes went to the third placed candidate would determine the winner (if there isn’t an outright winner already) and would indicate that the winner is at least “liked” by more than 50% of those who’s votes count.

The only example I could find where AV does not satisfy the Condorcet condition is from a comment on Gower’s Weblog and considers the following situation

27 % vote A,B,C,D (order of preference)
26 % vote B,C,D,A
24 % vote C,D,A,B
23 % vote D,A,B,C.

Under AV, candidate D would be eliminated first and the second votes of those who put D first would go to A who would then have 50% of the votes counted. Candidate C would then be eliminated and since D is already eliminated, the third preference votes of those who put C first would also go to A who would then win with 73 % of the votes counted. However, if you consider first and second votes only, B has 53%, A has 50%, C has 50% and D has 47%. The Condorcet winner might then be B as they would presumably beat any of the other candidates in a one-on-one battle. However, the differences are so small that any one of these candidates represents approximately the same fraction of the constituency and, from a democratic perspective, any one of them would be a reasonable winner. If this a typical example (which it may not be) where AV violates the Condorcet condition, then it doesn’t seem like a particularly significant problem.

Although disappointed about the result of the referendum, I’m also disappointed by the manner in which the referendum was lost and what it implies about our democratic leaders. The Conservatives (in particular the right wing of Conservative part) seem extremely pleased by this “victory”. Is this because they truly believe that FPTP allows us to select politicians who genuinely represent the majority view in the UK? I don’t think so. I suspect it is because the Conservatives know that the UK is a slightly left-leaning country and anything that is even slightly more proportional will be to their disadvantage. I should acknowledge that AV isn’t actually proportional, but it does at least allow the constituency results to more closely represent the views of the constituents. What does this imply? In my opinion, this implies that the party that currently essentially runs the country knows that they are not (and probably never will be) supported by a majority of the voters and rather than reconsidering their ideology, they’ve managed to maintain a voting system that suits them and allows them to have a better chance of remaining in power than they would have had were the voting system to change to AV. In fairness, it wasn’t only the Conservatives who campaigned against AV. Large parts of Labour did the same, probably for similar reasons.

It’s possible that the Conservatives genuinely believe that they are the best party to lead this country to greater and better things. The whole point of democracy, however, is that it is not up to them to make this decision, it is meant to be up to the electorate. Supporting a voting system that regularly results in winners who are not supported by even a majority of those who voted is a cynical and selfish attitude. Some might argue that we should not expect a party to support something that is likely to be to their disadvantage (compared to the current system at least). It’s like asking Turkeys to vote for Christmas. I would argue that if they truly believe that their ideology is “best” and believe that they could convince the electorate of this (which is partly the job of a politician) then they should feel comfortable with any reasonably fair voting system. Maybe their objection to AV suggests that even they don’t have confidence in the value of their own ideology.

AV and minority parties

It seems like one of the main objections to the Alternative Vote (AV) is that the second (or higher) votes of those who voted for minority parties could be decisive. The main objection seems to be “why should the second, third or fourth, etc., preference votes of people who’s first vote goes to the likes of the BNP or UKIP get to – in some cases – determine who wins the election”. Firstly, I think this attitude is wrong. I object to the BNP and I think UKIP is a silly party that also has some rather objectionable views, but they are currently legitimate parties and those who vote for them are entitled to make that choice without prejudice. Arguing that AV would give these voters some say and that this would therefore be bad is wrong, in my opinion. I know we’re not disenfranchising these voters, as the AV system doesn’t exist yet, but this argument does feel equivalent to an argument for disenfranchising a section of the community. Maybe the views of these voters would change if they felt their votes were having some influence. It could be a positive step.

Secondly, I don’t actually see the logic in this argument against AV. I looked up some numbers and at least 80% of UKIP and BNP candidates lost their deposits. This means that they received less than 5% of the vote in their constituency. Only a total of 2 or 3 candidates in each of these two party received more than 10% of the vote and the average for UKIP was 3.1% and for the BNP was 1.9% (since neither party had a candidate in every seat, the average per candidate is somewhat higher). These are minority parties, which by definition means they receive very few votes. If the second, third or fourth preference votes of voters who’s first votes went to UKIP or the BNP makes a difference, it implies that there must have been two other candidates both of whom had close to 50% of the votes counted and hence were similarly liked (or disliked) by the voters in that constituency. If the UKIP and/or BNP higher preference votes pushes one of these candidates over the 50% threshold, surely this is a reasonable result. You have two candidates who are similar (since not all people vote, a few percent difference may mean that the popularity of these two candidates is statistically the same) and a sensible method had been used to differentiate between them.

There are probably some reasonable arguments against what I’ve said above. There will be some constituencies where the UKIP/BNP share of the vote is above 10% and this could make quite a big difference. Again, why should the views of these voters not be taken into account. Also there are planned to be 600 constituencies and the number in which the UKIP/BNP share exceeds 10% is something like 5 (so 1% or less). Another argument might be that higher preference votes from UKIP/BNP may preferentially support the Conservatives rather than Labour and the Lib Dems. However, there are other minority parties (Greens for example) who’s voters are more likely to support Labour and the Lib Dems, so again this doesn’t seem like a reasonable argument either. As I’ve said in a previous post AV appears to produce results that – compared to FPTP – will more reasonably reflect the views of the electorate. Arguing that it will give undue power to minority party voters is not only wrong as these voters should have the right to express their views (as long as they’re not breaking the law), but these are minority parties and so the only time that they will make a difference is when two candidates are very close anyway, in which case why does it matter.

AV vs FPTP

I haven’t really been following the Alternative Vote (AV) versus First Past The Post (FPTP) debate in much detail. Yesterday, however, I read Tim Gower’s post Is AV better than FPTP and found it very interesting and informative (on another note, I found David Broomhead’s article A formula for fair voting very poor).

I don’t want to re-explain all the details about AV but, it seems to me at least, that FPTP is only really fair if there are only two candidates. If there are more than two, then there is a reasonable chance that the vote will be split and that the winner will be someone that more than 50% of the voters would rather had not won. With AV, it seems that if all the voters were to vote their preference honestly (including voting for only one candidate if that is their choice) the result would at least reasonably reflect the views of the electorate. I should add that my definition of fair here is not that it is fair to the candidates but that it fairly reflects the views of the voters.

What’s been most disappointing about what I’ve seen of the debate is how it hasn’t included much discussion of which system would better allow the views of the country to be represented. Certainly the No to AV group seems to be arguing that we need a system that will produce clear winners and that the AV system will result in more coalitions. Even if the latter is true, it seems fine to me if this actually reflects the distribution of current political views in the country. Our voting system shouldn’t be based on the make-up of the political parties, the political parties should adapt if the voting systems tells them that their policies are not sufficiently popular. In short, we should come up with a voting system that will allow for the views of the electorate to be determined and let the political parties adapt accordingly, not the other way around.

I have, however, found this “it’s the result that matters” attitude quite common. My understanding of what my university does when it decides who should be offered a place is that it uses all sorts of information, not simply the A-level grades that the students get or are likely to get. I think this is quite correct and if done properly (and there is a lot of work going into determining how to do this properly) the students who are most likely to succeed will be offered places. To me, this is what’s important. I have, however, heard others say that it’s not fair if someone’s parents spend a lot of money sending them to a good school and then someone with lower grades is offered a place ahead of them. If grades were a perfect indicator of how someone would perform at university, this view would be quite correct, but they’re not (or at least I have seen studies showing that they’re not). If someone has had extensive help in how to cope with exams and how to achieve good grades, their results may suggest that they will perform better at university than they actually will. The reverse is also probably true.

Similarly, it seems at university that students and staff alike have forgotten that exams are simply forms of assessment. We spend four years teaching students various things about a subject and we need to assess how much they have learned and how capable they are likely to be once they graduate. It’s not a competition to see who can get the highest marks. If it any stage we feel that students are achieving grades that do not reasonably reflect their abilities we should change the system. This is not to disadvantage students but simply because it should – in my opinion – be designed to fairly assess students’ abilities. The voting system should be the same. By and large, the voting system should be designed to produce a result that largely represents the views of those who voted. Proportional representation would be ideal – in my opinion – but given that we value constituency politics, AV seems like a reasonable option. Despite David Broomhead’s article in the Guardian, tactical voting is difficult in the AV system, and if most voters were to indicate their preferences, the winner should be someone preferred by more than 50% of the voters which seems much better than the current system where the winner might get less than 50% of the votes, and more than 50% of the voters would have preferred someone else to have won .