YouGov - Labour would do worse under AV
There is a new YouGov poll for Channel 4 News asking how people would vote under AV. It projects that Labour would suffer the most under AV, with the Lib Dems gaining the most and the Conservatives largely unchanged - their losses would be cancelled out by gains.
On a normal uniform swing, the standard FPTP voting intention figures in the poll would give us 255 Conservative seats, 355 Labour, 16 Lib Dem and 24 Others - so a Labour majority of 60. YouGov's projection of how the seats would pan out under people's AV voting intentions are 255 Conservative seats (so no difference, though there will be churn in which seats!), 342 Labour (down 13 compared to FPTP) and 29 Liberal Democrats (up 13 compared to FPTP). Labour would instead only have a majority of 34.
At the previous election (and at elections before that) polls have repeatedly suggested that AV would favour the Liberal Democrats and Labour and disadvantage the Conservatives, or in some cases that both Labour and the Conservatives would both suffer, but the Conservatives would suffer more.
This was largely because Labour and Liberal Democrat voters told pollsters they would be very likely to give their second preferences to one another, while Conservative voters favoured the Liberal Democrats for second preferences, albeit less enthusiastically. Hence in Con v Lab marginals Liberal Democrat second preferences helped Labour, in Con vs LD marginals Labour second preferences helped the Lib Dems, and in the relatively small number of Lab v LD marginals Conservative second preferences helped the Lib Dems a bit.
Last Summer, when the coalition was still young and Lib Dem support was still in the mid teens, YouGov asked how people would vote under AV and found this pattern beginning to change. The remaining Liberal Democrat voters were as likely to give their second preferences to the Conservatives as to give it to Labour, and Labour voters had become significantly less likely to give their second preferences to the Liberal Democrats. At the time AV was still better for Labour than the Tories, but if the trends apprent then continued it could easily have changed.
This week YouGov & Channel 4 did a fresh exercise asking people once again to say how they'd vote under AV. This time we repeated the question used by the British Election Study in 2010, giving people a picture of a ballot paper and asking them to actually enter numbers next to as many or as few candidates as they wanted to. This allowed us to better project the actual effect than was possible last Summer - for example, Labour second preferences disproportionately go to the Green party, but given that the Green party will normally have already been eliminated in a count before Labour is, it's actually their third or fourth preferences that count.
Conservative voters are now most likely to give second preferences to the Lib Dems (41%), followed by UKIP (27%). UKIP are, of course, unlikely to actually benefit from many Conservative second preferences - what will actually matter in the course of most election counts is how Conservative voters' lower preferences divide between Labour and Liberal Democrats - here 29% of Conservative voters do not give Labour or the Liberal Democrats any preference, 8% put Labour higher on their ballot, 63% put the Liberal Democrats higher on their ballot.
When the BES ran the same question in May 2010 over half of Labour voters gave their second preferences to the Liberal Democrats. This has collapsed - 30% of Labour voters would now give second preferences to the Greens, 19% wouldn't give one at all, 18% would give it to UKIP and only 16% to the Lib Dems. Taking into account preferences further down the ballot though, if it was between Lib Dems and Conservatives 46% of Labour ballots would end up being transferred to the Lib Dems, 12% to the Conservatives, 42% neither.
Turning to the Liberal Democrats, their second preferences now split fairly evenly, but with the Conservatives just ahead:
31% to the Conservatives, 24% to the Greens and 24% to Labour. Again, looking at what would ultimately happen to Lib Dem votes if they had to be transfered to Con or Lab, 46% would end up in the Conservative pile, 39% in the Labour pile, 15% neither.
The minor parties's transfers end up much where you'd expect them - Green voters favour Lab or LD strongly over the Conservatives, and Labour slightly over the Lib Dems. UKIP voters tend to favour Conservatives over Lib Dems, Conservatives slightly over Labour. Around a third of BNP voters don't give any preferences to any of the three main parties, but those that do tend to favour the Conservatives and Labour over the Lib Dems.
What this all boils down to is that in Con v Lab marginals the lower preferences of Lib Dems would help the Conservatives win seats from Labour, in Lab v LD seats Conservative lower preferences will help the Lib Dems win seats from Labour, in Con vs LD seats Labour lower preferences will help the Lib Dems... but Con losses there will be cancelled out by Con gains against Labour.
Of course, this all needs a lot of caveats - it assumes both a uniform swing, and that each parties second preferences split in the same proportions across the country. It also cannot take into account what effect an election campaign fought under AV would be - this poll shows people's AV first preferences being largely the same as their FPTP vote, but I suspect in practice smaller parties would effectively campaign to get some people to use their first vote to send a message. All that aside though, it is looking as though AV would work against Labour, as Conservative and Lib Dem voters seem increasingly likely to give preferences to one another.
See also Gary Gibbon's post here
