Over on Left Foot Forward there is a paper by a Dr Matt Qvortrup arguing that the introduction of AV would help Labour and hurt the Conservatives, and therefore it would be in the Labour party’s partisan interest to support it.

The paper is almost all made up of polling evidence from elections between 1997 and 2010 which demonstrate that the Labour party would have gained more seats under AV (or at least, lost out less than the Tories!), and reference to the 2008 London mayoral election where Labour performed better under AV (Ken Livingstone still lost, but the re-allocated second preferences split in his favour, so AV was a plus for Labour)*

I wouldn’t make any argument with any of this, throughout this period AV would certainly have helped the Liberal Democrats and been kinder to Labour than to the Conservatives.

Where I would depart from the argument is the assumption that the same pattern would apply now. Normally, in the absence of other information it is a fair assumption that people will keep on doing what they have done in the past. In this case though, there has been a massive shift in British politics and it is fair to question whether that assumption is safe.

AV voting would have helped Labour between 1997 and 2010 because Labour voters were likely to give second preferences to the Lib Dems and vice-versa. To put it simplisticly the left-of-centre, anti-Conservative vote was split between two-parties, and AV would have effectively united that vote behind the better performing of the two parties in each seat.

The formation of the coalition will likely have changed that. People who voted Liberal Democrat seeing it as a left-of-centre, anti-Conservative party probably aren’t voting Lib Dem anymore. The rump of remaining Liberal Democrat supporters are likely to be more positively inclined to the Conservatives, and Conservative supporters themselves will likely see the Lib Dems more positively.

In his paper Dr Qvortrup acknowledges this with reference to an article by John Curtice floating the same possibilities, but handwaves it away by saying there is no evidence of it and Curtice is just making assumptions. As Carl Sagan once said, “absence of evidence is not evidence of absence”.

As it happens, while John Curtice didn’t cite it, there is evidence to back up Curtice’s assumptions. YouGov asked a “how would you vote under AV?” question at the time of the general election and found a similar pattern of second preference distribution as everyone else – Labour voters were more likely than Conservative voters to give second preferences to the Lib Dems, and Lib Dem second preferences broke in favour of the Labour party. However, YouGov then repeated the exercise in July 2010 after the formation of the coalition to test the hypothesis that pattern of second preferences would have shifted.

Indeed it had – Conservative voters had become more likely to give second preferences to the Liberal Democrats, Labour voters much less likely to give second preferences to the Lib Dems, and Lib Dem voters’ second preferences now split evenly between Labour and the Conservatives.

On a crude projection, these splits would still been slightly better for Labour than the Conservatives, but the more important finding was the change in the pattern of second preferences. That evidence comes from July 2010 when the Lib Dems were still in the high teens – my guess is that these trends would be even more pronounced now as the Lib Dems are ever more reduced to a rump of those supporters happy with the Conservative coalitions and Labour supporters become ever more antagonistic towards the Liberal Democrats.

Of course, that is still an assumption. While there are no more recent polls that specifically asked about second preferences under AV, we can at least look at what Liberal Democrat voters say in other questions that might act as a proxy for whether they prefer Labour or the Conservatives. Looking at the most recent YouGov polling this week:

Remaining Lib Dem voters think David Cameron would make a better PM than Ed Miliband by 31% to 1%.
Remaining Lib Dem voters would prefer a Conservative led government to a Labour led one by 60% to 26%
Remaining Lib Dem voters approval rating of Cameron is plus 44, of Miliband minus 31.

The assumption that these voters would still be more likely to give their second preferences to Labour than the Conservatives is, quite frankly, fanciful.

Of course, the main reason for this shift isn’t that Lib Dem voters have suddenly become more right-wing or more pro-Conservative. They haven’t – the opinions of people who voted Lib Dem in 2010 are still tilted towards Labour. The reason is that most Liberal Democrat supporters from 2010 don’t support them anymore! Many or most of those Liberal Democrat voters who in May 2010 would have told pollsters that their second preferences would go to Labour would now give their first preferences to Labour. Those that remain are a more pro-Conservative rump.

And there, as they say, is the rub. AV is not a system that automatically favours right wing parties or left wing parties, nor one that will always favour Labour or always favour the Conservatives. It depends entirely upon the circumstances – during recent decades it will have tended to have been better for the left because we’ve had a political landscape where there were two left-of-centre parties with supporters who were comparatively comfortable with voting tactically or lending their votes to each other.

If at the next election the landscape is instead a Lib Dem party whose support has been reduced to mainly those voters who are reasonably well disposed to the Conservatives, where many Conservatives view the Lib Dems as allies rather than enemies, and where many Labour voters see the Lib Dems as turncoats and Tory stooges, the pattern of second preference voting may be utterly different.

(*As an aside, I’m confused by the rather odd claim that Ken Livingstone would have won the London mayoralty had he secured an extra 11,182 votes. After second preferences had been re-allocated Boris Johnson won the election by 1,167,738 votes to Ken’s 1,028,966, a majority of 139,772.)

YouGov have released their final Welsh poll of the year for ITV Wales (the final daily poll of the year for the Sun, incidentally, comes tonight). Welsh assembly voting intention figures with changes from last month are:

Constituency: CON 23%(+2), LAB 44%(nc), LDEM 6%(-3), Plaid 21%(nc)
Regional: CON 22%(+2), LAB 42%(+1), LDEM 5%(-4), Plaid 21%(+1)

On a uniform swing (and making the fairly safe assumption that Labour will reclaim Blaenau Gwent) this would give Labour 30 seats in the Welsh Assembly, the Conservatives 15, Plaid 13 and the Liberal Democrats 2. The Lib Dems, incidentally, would no longer pick up any top-up seats, once you get down to 5% or so support the maths just doesn’t stack up. Amongst the others, UKIP were equal to the Lib Dems on 5% in the regional vote, but they wouldn’t get any seats either (on these figures they’d be closest to getting one in South Wales East).


Tonight’s YouGov poll for the Sun has topline figures of CON 40%, LAB 42%, LDEM 9%. The small Labour lead in YouGov’s daily polling looks like it’s being consolidated. Meanwhile net government approval is down to minus 19, the lowest the coalition government have recorded so far.

UPDATE: There is also a new Angus Reid poll, topline figures are CON 35%(nc), LAB 41%(+1), LDEM 9%(-4). I think that six point lead is the biggest Labour have recorded so far this Parliament, and it’s only the second pollster to show the Lib Dems in single figures. Others are presumably up to around 15%, which I think is also the highest any company has shown so far this Parliament.

ICM’s final Guardian poll of the year has topline figures of CON 37%(+1), LAB 39%(+1), LDEM 13%(-1). Changes are from a month ago, but there is no significant change.

The small Labour lead over the Conservatives is now consistent across all the pollsters – the differences between the different companies remains the Lib Dems and, in some cases the others. ICM normally show the highest level of Lib Dem support, YouGov the lowest – a big chunk of the difference is normally down to the reallocation of don’t knows. ICM and Populus reallocate some of the people who say they don’t know how they’d vote in an election tomorrow to the party they voted for last time, currently this tends to increase the reported level of Lib Dem support by a couple of points since, as you’d expect, there are a significant chunk of Lib Dem voters out there who say they aren’t sure what they’d do in an election.

ICM also asked about voting intention in the AV referendum and found a 6 point lead for AV, 44% to 38%. This is the same margin as the ICM/ERS poll reported on Friday. As I explored in a post earlier today, the exact wording of these things makes a significant difference, though Julian Glover’s report in the Guardian suggests the approach they used in this survey was pretty much the same as in the ICM/ERS survey.

Last week there were a couple of polls on AV. YouGov’s regular tracker on how people would vote in an AV referendum had AV on 33% and FPTP on 39%. Meanwhile an online ICM poll for the ERS had AV on 35% and FPTP on 22%.

I posted on it here, and suggested that the difference was probably the wording, suggesting three possible reasons the wording might produce different results. First, YouGov gave respondents summaries of what AV and FPTP actually were. Secondly, ICM only mentioned AV in the question, while YouGov mentioned both AV and FPTP. Thirdly YouGov presented the referendum as a coalition government policy, ICM just as something that was happening in 2011.

It was the last of these differences that seemed to attract the most comment – did mentioning the government make people more likely to vote no? On this at least, I can give a firm answer. YouGov reasked the question the day after their latest regular tracker, but changing the wording to remove the mention of the government – the results were virtually identical (32% for AV, 40% for FPTP). Mentioning the coalition in the question does not make any significant difference to the result.

Since the polls last week, we have some more AV polling to play with. Firstly there is a second ICM poll for the ERS, once again conducted online. This one found 36% in favour of AV, 30% in favour of FPTP and 34% don’t knows.

There was no option of people saying they would not vote, indeed, the question was not presented as being how people would vote in a referendum, but just yes, no or don’t know to the actual question that will be on the ballot paper. Normally I’d be a bit wary of that – how people vote in a referendum does not necessarily match with their views on the issue actually being voted upon. In this particular case, the question came after a question about how likely people would be to vote in the referendum, so respondents should at least have been answering the question in the context of the referendum.

Comparing this ICM question and ICM’s question from a week before, mentioning both AV and FPTP in the question rather than just AV seems to have increased support for FPTP by about 8%, although some of that difference will also be the lack of a “won’t vote” option (most people who told ICM they definitely wouldn’t vote went on to say don’t know to the AV or FPTP question, but those who did express an opinion were more likely to prefer FPTP.)

The second bit of “new” polling (both this and the “new” ICM poll were conducted a couple of weeks ago) is YouGov for the British Election Study. As well as the BES’s main studies before and after elections, they do rolling monthly surveys between elections. One of the regular trackers they include is voting intention in the AV referendum. They use the referendum question as initially proposed (they may change it to match the amended version in time, I don’t know) and ask how people will vote in the May 2011 referendum.

“How will you vote in the May 2011 referendum on a proposed change in the UK electoral system? Do you want the United Kingdom to adopt the ‘alternative vote’ system instead of the current ‘first past the post’ system for electing Members of Parliament to the House of Commons?”

The most recent results for the YouGov/BES AV question at the beginning of this month were AV 34%, FPTP 34%, Wouldn’t vote 5% and Don’t know 27% – so AV and FPTP neck and neck. Compared to the normal YouGov/Sun AV question with the summary of the systems, this has more don’t knows and fewer FPTP supporters, suggesting that telling people what the rival systems are boosts support for FPTP by roughly 5%.

It has a slightly higher level of support for FPTP than the second ICM/ERS poll, but in the same ballpark. It could be down to asking it in the context of voting in a referendum, or it could be sampling differences, or the difference could be just good old random variation (if the “true” level of support for FPTP was 32% both polls would be within the margin of error).

So for anyone who is still following, the leads shown using the different question wording

No systems summary, referendum context, prompted by AV only: AV +13 (ICM/ERS #1)
No systems summary, no referendum context, prompted by AV & FPTP: AV +6 (ICM/ERS #2)
No systems summary, referendum context, prompted by AV & FPTP: even (YG/BES)
Summary of systems, referendum context, prompted by AV & FPTP: FPTP +6 (YG/Sun)