08 April 2011

A switch to AV will make a bigger difference than most people realise.

A lot is made of the fact that AV is not a proportional system (although in practise it does generally produce more proportional results than our present system). Yet AV could have a massive impact on voting patterns in the UK.

In fact I think the No team realise the impact AV will have more than the Yes team, which is why MPs in safe seats and councillors are pumping money into full page ads in the press and dominating the ads on the internet to try to persuade people to vote no.

In Australia they elect only 150 MPs compared to our 650 MPs at Westminster (changed to 600 for next election - an anti-democratic change). This means Australian constituencies are huge, 3 times the number of electors and geographically can cover huge areas (some of their constituencies are bigger than the entire UK!).

This has a disproportional effect, as the bigger the constituencies the less proportional the result (local elections in the UK are more proportional for this reason as wards are around a tenth the size of constituencies). This has kept Australian politics largely a two party affair as smaller parties need more resources to reach larger numbers of voters and more time and elections to overhaul the huge voting leads of the bigger parties.

The larger constituencies in Australia also explain why they have had less coalition government than the UK despite using AV.

So the proportional effects of AV are masked in Australia, where smaller parties are still in their fledging development.

In the multi-party UK, AV could have a massive proportional effect since there is enormous appetite to vote other than the big two. We can see this from the fact that even under FPtP a third of voters do this, most in the knowledge that their vote won't count towards the result, either their vote is completely wasted or they are playing the long game hoping a better result gradually over many years of elections will bring their smaller party into contention (it can be a long wait - many decades and fixing 5 year terms is another anti-democratic move of this government). In the latest PR Euro elections over 50% (55% to be precise) voted other than the big two.

Over time and elections, as people get used to AV, they will realise they are now free to vote as they please without having to make a choice between wasting their vote and having it count towards the final result. This is bound to encourage even more to venture away from the big two parties. No wonder the establishment MPs are so scared!


  1. "although in practise it does generally produce more proportional results than our present system"

    Seriously? I think that is wrong. Last year I took the whole data set of elections in Australia from the second world war onwards, and the whole data set of elections in the UK from the second world war onwards and did some analysis. I calculated the Gallagher Index for each election (a measure of disproportionality), and then plotted the Gallagher Index against the proportion of the vote going against the two large parties. So in the case of UK elections it would be all non-Tory or Labour votes, in Australia it would be all non Liberal/National votes or Labour votes.

    There is a clear linear relationship in both systems. As more people vote for neither of the top two parties, the level of disproportionality decreases for both systems. So both are more proportional when there is a more two party system.

    The slope of the FPTP elections is steeper, indicating that it becomes more disproportional more quickly than AV, as less people vote for the top two parties.

    But FPTP is MUCH more proportional in a two party election than AV is. In a close two party race, AV is MUCH more likely to give the election to the party that comes second than FPTP is.

    So although AV is more proportional in elections with fewer first and second party voters, it is still NOT proportional, and disproportionality still increases with increased "other" vote. And it is LESS proportional in elections where there is a preference for voting for the top two parties.

    The other salient observation is that AV is also much less predictable, there is much more variation about the regression line, meaning that the level of proportionality AV gives is extremely unpredictabel, so although in GENERAL AV is marginally more proportional for elections with high "other" (including LD) votes, it is still possible for AV to be more unproportional even in these sorts of elections.

    In the end, the only real conclusion that we can draw is that AV is NOT more proportional, only that it is different. Sometimes it will produce more proportional results, and sometimes it will produce less proportional results.

    It WILL tend to produce much more unpredictable results, and will tend to give majorities to parties that come second in first preference votes much more often the FPTP does.

    It does help to actually do some analysis and have some evidence.

    This analysis is simple to do.

  2. "This has a disproportional effect, as the bigger the constituencies the less proportional the result"

    That is true.

    "AV could have a massive proportional effect"

    That is not true. For sure, the smaller the constituencies, the more disproportional the result for both FPTP and AV. But I see no evidence that AV will increase proportionality. Quite the reverse.

    There is a flaw in your logic, and it goes like this.

    Let's say more people are likely to put smaller parties as their first preferences, that makes sense, because those votes won't be lost. For the sake of argument let's assume that first preference choices go down by 5% for each of the two big parties, up by 3% for the LDs and up by 7% for "others".

    Most seats will not change hands. In 2010 this would mean that the national vote would look something like

    Tory 31%
    Labour 25%
    LD 26%
    Other 18%

    BUT, would this mean that the LDs get MORE seats than Labour? Probably not, those 5% that didn't vote Labour would certainly have put Labour as their final preference, those votes would come straight back to Labour from "other", and so would the Tory votes come straight back to the Tories.

    We would probably elect a handful of independents (AV makes it easier for independents to get elected), and most analysis shows that LDs might get something like 30 more MPs.

    But the overall result is LESS proportional, because AV will ENCOURAGE voters to put non-mainstream parties as their first candidates, while transferring those first preferences to mainstream parties over successive rounds of counting.

    So we could still be in the position where a party gets 20% of first preferences, but has a parliamentary MAJORITY.

  3. Alan, you admit that 'in general' AV is more proportional when a sizeable part of the electorate vote for other than the big two parties. This is exactly what I was saying whan I said that AV is generally more proportional. Where's the disagreement. The UK is no longer a two party system and the clear trend is for more people to vote other than the big two, this will accelerate under AV so more proportional it is then!

    As for your comparison with Australia - there are other factors that affect proportionality or otherwise other there, that mask just how proportional AV can be. For example, seats are bigger over there, both in terms of size of electorate and geography - both these have disproportional effects, and also until recently in Australian elections there was some serious malapportionment / gerrymandering going on that distorted results (rural seats had much smaller electorates in W and S.Australia and Queensland).

  4. "you admit that 'in general' AV is more proportional when a sizeable part of the electorate vote for other than the big two parties."

    For sure, as long as the non first+second party vote share is greater than about 10-15%, then AV is in general more proportional than FPTP, lower than 10% and FPTP is more proportional. But that does not mean that it cannot throw up MORE disproportional results than FPTP, even when this criterion is met. Because it is more UNPREDICTABLE. It has a MUCH greater variation, and will certainly often produce a less proportional result quite often, even when non first/second party votes are greater than 15%. further, the more multi-party the system is, the more disproportional the result AV gives, this is a feature it shares with FPTP, and is a product of single member districts, rather than how votes are cast, or counted.

    When we plot disproportionality against non first+second party vote share, for both AV and FPTP, there is a clear correlation, or if you prefer, proportionality is NOT independent of the level of votes that go to non first+second placed parties.

    But under Irish STV, German MMP and Finish open list PR we see NO such correlation, proportionality is totally independent of the share of the vote that does NOT go to the two biggest parties.

    That is how a system can be judged to be proportional.

    AV is NOT proportional, it is NOT even semi-proportional. It is NOT even "more" proportional than FPTP. What it is is "sometimes more proportional than FPTP, and sometimes less proportional than FPTP", and whether it is more or less proportional is as much down to chance as anything else.

    So the disagreement is in the claim you make that AV could have a "massive proportional effect". That is simply wrong. The only way to get a "massive proportional effect" is to introduce PR. Claiming "massive proportionality" for AV is the sort of exaggeration we get again and again from pro-AV campaigners.

    "that mask just how proportional AV can be"

    That is totally incorrect. AV is NOT a proportional system. NO electoral system that relies on single member districts can EVER be a proportional system. And to keep claiming that AV Can be proportional is to continue to be totally wrong.

    There is only ONE way to have a proportional electoral system, and that is by having MULTI-MEMBER CONSTITUENCIES.

    That is what EVERY academic political scientist says, that is what every book ever written about electoral systems says, that is a MATHEMATICAL fact.

    You're arguing in the face of all of the facts.

    AV and FPTP share the same major defect, they rely on single member districts. Live with it, it's a fact. THAT is what makes BOTH so unproportional.

  5. Alun, AV is not a proportional system but it does in general give more proportional results than FPtP. You seem to think it has an equal chance of being more or less proportional, yet the FACTS do show it is on average slightly more proportional. Of course this is all relative, as compared to PR AV is a poor substitute, as you correctly point out we need PR and I agree with you.