Earlier in the week we also saw the publication of this article by Martin Boon and John Curtice, their take on why the polls overestimated the Lib Dems at the last election, based upon a call back survey of 1,200 of the respondents to their final survey.

The first reason Boon & Curtice suggest is a late swing, though they see this as explaining only part of the error. 95% of people who told ICM before the election they were going to vote Conservative reported actually having done so after the election, for Labour the figure was 93%, for the Lib Dems it was somewhat lower at 87%. That said, they say that the number of people switching towards the Lib Dems was almost as much as they lost, so this can only be a small factor.

Secondly, Boon & Curtice do not think differential turnout was a big factor. People who said they were going to vote Lib Dem before the election were not significantly more likely to tell ICM in the post election poll that they didn’t vote.

Thirdly, ICM’s don’t knows disproportionately brokein favour of Labour, backing ICM’s decision to use their “spiral of silence adjustment” (Boon & Curtice still refer to this as “shy tory syndrome” in their article, though in this case it was “shy” Labour voters). One of the conclusions they draw in the article is that it may be better to make the adjustment even stronger.

Finally, Boon & Curtice say ICM may have weighted the Lib Dems too highly in their past vote weighting, and that this will be high on the list of their investigations in coming months.

It is also worth reading Roger Mortimore of Ipsos MORI’s reply in the comments to the article, where he makes a very valid point that underlines just how difficult it is to work out for certain what went wrong with the polls – many factors that could have skewed the final polls could also have skewed any call back survey. To take Roger’s example, if the problem was people saying they would vote Lib Dem but not actually doing so on the day, those people might also have claimed to have voted when they didn’t in a call back survey.

In that specific case, we will eventually have concrete evidence of whether differential turnout was a problem or not (the British Election Study recontact interviewees after the election and ask if they voted, but also check them on the marked electoral register to see if they really voted), but there are similar problems we will never be able to rule in or out for sure. If, for example, some people told pollsters they would support the Lib Dems but actually voted for someone else (for whatever reason – being the fashionable thing to say perhaps), they may have said Lib Dem in the ringback survey for that same reason.

UPDATE: And on more topical matters – tonight’s YouGov voting intentions are CON 42%, LAB 34%, LDEM 17%.

A quick couple of posts to catch up with a couple of things I should have posted about during the week but didn’t get round to.

First, the Speccy this week has the results of a YouGov poll on voting intention in a general election fought under AV. This is a repeat of the questions asked by YouGov for Channel 4 at the start of the week, but carried out this week (the Channel Four polling was conducted about a fortnight ago). The big picture remains the same – Labour voters have become far less likely to give their second prefereces to the Lib Dems than they said they were at the time of the last election, Lib Dem voters are now as likely to give their second preferences to the Conservatives as Labour.

As a crude rule of thumb, how Lib Dem voters split determines how AV affects Lab -v- Con seats, and how likely Labour or Conservative voters are to back the Lib Dems determines whether the Lib Dems gain more from AV in their Con -v- LD margins or their Lab -v- LD marginals.

In this commentary on the YouGov website I’ve made some projections of what the result would be in a general election now with FPTP or with AV. On first past the post (using vote shares of CON 41%, LAB 35%, LDEM 16% – the normal YouGov voting intention that day), the Conservatives would win 320 seats, Labour 280 seats and the Lib Dems just 24. In contrast, with AV the result would be Conservative 309 (11 fewer), Labour 273 (7 fewer), Lib Dem 43% (19 more).

As I mention in the article, this is a very crude projection at the moment because we don’t have the data to factor in minor parties. In particular, it won’t be much good in Scotland or Wales because we don’t have the data to judge how votes will transfer to and from the SNP and Plaid. Still, it’s a start, and it suggests that there will not be a vast difference between the number of seats Labour and the Conservatives loose under AV now that Conservative and Lib Dem voters have become more likely to exchange second preferences (if that tendency grew stronger, we could even reach a point where AV actually helped the Conservatives).


The latest YouGov voting intentions are now up here, and show CON 41%, LAB 36%, LDEM 15%, Others 9%.

YouGov have been continuing daily polling since the election (and will continue doing so), but from this week have started putting out voting intention and government approval figures daily again. In fact, we’ve been asking them both daily for several weeks, but only putting them out in chunks now and again.

I’m not going to make a big fuss of the new figures each day – not least because one can only say “no significant change from yesterday” in a limited number of ways without repeating yourself. Voting intention only changes slowly over time, and outside an election campaign there no point pretending that every new figure is going to show something exciting. That doesn’t mean it isn’t important or interesting though – just that doing voting intention every day requires us to read it and use it somewhat differently.

There is a metaphor that Peter Kellner used to use many years ago when YouGov was relatively young, of how polling used to be regarded as a fine wine, or aged whisky – brought out only for special occassions and greatly revered by those who had paid a fortune for it. It would be much better if polling was like running water, cheap, easy to get and already ready and available to dip into whenever you wanted to know what the public thought.

That’s how you should treat the YouGov daily voting intention figures. Don’t pay attention to the daily movements, anything dramatic is probably sample error anyway – rather you should look at the bigger picture and watch how they develop over time. Don’t get excited over one day’s figures – Labour might be up 4 today or down 3 tomorrow, but the next election is 5 years away. What is really matters is the trend, the slow (or sometimes fast) tectonic movements in party support. With a week or a month’s worth of data we can watch a party’s support going up or down with confidence, rather than making guess from a once-a-month peek at public opinion.

What it is there for is context, background and analysis, whenever you want to know what the levels of party support are… you’ll be able to open it and see. Want to know what effect something had on party support – the figures will be there quietly ticking away. Want to look for a correlation between party support and something else, the back data is all there for you to analyse. Prefer to look at a 5 day rolling average, the figures are there for you to work out. Running water rather than Whisky :)

UPDATE: Changed this a bit, since several people were leaving commetns saying they disagreed…and that they thought what I had intended to write myself. Hopefully I’ve got it across more clearly now!

600 seats

I’ve written a lot about AV over recent days, what about the boundary review. Now we know the new target number of seats upon which the quota will be set (600), the tolerance that will be allowed either side of that quota (5%), and the exceptions that will be allowed (the Western Isles, Orkney & Shetland and a cap by area), we can take some guesses at what the overall impact will be.

The North East is rather tricky to fit into the new quotas. Northumberland only qualifies for 3 seats (while Berwick-upon-Tweed is a large, underpopulated seat, it doesn’t come close to the geographical limit!), but they would be grossly overpopulated so would need to be paired with one or more Tyne and Wear Boroughs. Durham could be divided into 6 seats, but the Cleveland Boroughs need to be paired with it if not to produce oversized seats. We’d end up with 14 seats in Northumberland and Tyne and Wear, down 2, and 12 seats in Cleveland and Durham, down 1.

In Yorkshire North Yorkshire would not lose anything, and would presumably have only minor changes. Humberside would lose 1 seat, as would both South and West Yorkshire.

The North West is also relatively straightforward on paper, Merseyside would lose 2 seats, Cheshire would lose 1, Lancashire would lose 1, Manchester would lose 1 and so would Cumbria. In practice there are probably some tricky problems to solve. The Wirral would currently get three seats, but they would be just above the 5% limit, so unless the quota has risen by December 2010 (or the population of the Wirral fallen), the spectre of a cross-Mersey seat would rise again. Cumbria is also probably also going to be tricky to divide into 5 neat seats.

In the East Midlands, Leicestershire and Lincolnshire would retain 10 and 7 seats, so would probably have only minor changes. Derbyshire and Nottinghamshire would both lose a seat. Northamptonshire would qualify for 7 seats, but they would be too small to be within 5% of the new quota, so it would need to be paired with a neighbouring county. The most obvious candidate would be Bedfordshire to the South, which also needs to be paired to avoid undersized seats. Between them they would have 12 quota sized seats, compared to 13 currently.

In the rest of the East of England Hertfordshire and Suffolk would have only minor changes. Cambridgeshire could also be treated alone, but Norfolk needs to be paired in order to produce seats within the quota limits, and a pairing with Cambridgeshire would produce seats closest to the quota – between them the two counties would retain 16 seats. Finally for the East, Essex would need to lose 1 seat.

The West Midlands are another tricky region. Worcestershire, the West Midlands (down 3) and Staffordshire (down 1) can all be divided into seats within 5% of quota (though dividing Birmingham’s huge wards into seats within the 5% tolerance will be fun!). Shropshire and Herefordshire would need to be paired, but putting them together doesn’t help, so they would need to be dealt with together with Worcestershire (between them losing one seat). But this leaves Warwickshire too large to result in 5 seats inside the 5% limit. It could be paired with some of the Metropolitan boroughs, but a neater solution may be pairing Warwickshire with Oxfordshire, which would otherwise be oversized – together the two seats would retain their existing number of seats.

The rest of the South East should have very little disruption from the review. Berkshire, Buckinghamshire, East and West Sussex, Surrey could all retain the same number of seats and hit the new quota. Hampshire would lose a seat based on its own electorate, but unless an extra exception is made it will need to be paired with the Isle of Wight creating a cross-Solent seat. Between them the Isle of Wight and Hampshire will retain the same number of seats. Kent therefore becomes the only county in the South East to lose a seat.

In the South-West Cornwall will probably be upset about being paired with another county, but it is unavoidable. With an entitlement of almost exactly 5.5 seats it will need to be paired with Devon, between them having 17 seats, one down on currently. The former county of Avon will lose 1 seat, Gloucestershire will be largely unchanged. This leaves Dorset and Wiltshire where the average seat sizes will be too small, and Somerset where they will be too large. To me, the most sensible solution is pairing Wiltshire and Dorset, with Somerset paired with one or both of the parts of Avon originally drawn from Somerset. The result will be that Avon/Somerset lose one seat between them, and Dorset/Wiltshire lose one seat between them.

London as a whole will have 70 seats, down from 73. There are obviously a large number of possible pairings of Boroughs to get to this point.

Northern Ireland will lose 3 seats.

Wales will suffer the harshest reduction in seats, down from 40 to 30 as its quota comes into line with the quota elsewhere in the country. Once again, there will be some tricky decisions for the boundary commission. My guess is Gwynedd will need to be linked with Clwyd (losing 3 seats between them), Powys will need to be linked to some other county – perhaps Gwent. The ERS’s stab at what sort of result boundary changes might produce had a rather odd link between Powys and Dyfed, which looks unlikely, but does make the maths work nicely. Either way, most of the rest of Wales will need to be linked up and there are various ways it might pan out.

Finally, Scotland would have a quota of 51 seats, down from 59. However, we know there are exceptions to the rules for the Highlands and Islands. These mean that the Western Isles and Orkney and Shetland retain their current undersized seats. The Highlands are entitled to 2 seats based on the quota (though they would be more than 5% from the quota, so it would need to be paired.) In practice, I think it would be impossible to come up with a solution that didn’t involve a seat larger than the current Ross, Skye and Lochaber, which is to be the statutory geographical limit on size, so the Highlands will probably retain three seats (one possible solution that kept all the seats within 5% of the quota and under the geographical size of RS&L would be to put the south of the current RS&L with the undersized Argyll and Bute, then splitting the remainder of RS&L between the other two highland seats – I think one would still end up being too large geographically though. With the Highlands and Islands taken care of, the rest of Scotland would be entitled to 48 seats, producing a total of 52 or 53, down 6 or 7.

YouGov have some more AV polling, this time for Channel 4. In the past the assumption has been that AV would help Labour and the Liberal Democrats, and indeed this was backed up by polling evidence from past elections. In their final poll before the 2010 election YouGov asked respondents how they would have cast their second preference votes if they had been voting under AV.

Amongst Conservatives voters 45% would have given their second preferences to the Lib Dems, 5% for Labour, with the rest not sure, not casting a second vote, or casting one for minor parties. Amongst Labour voters, 6% would have given their second preference to the Conservatives, 64% to the Lib Dems. Lib Dem voters would have split their second preferences in favour of Labour by 42% to 27% for the Tories. Peter Kellner’s estimate based on those splits is that this would have cost the Conservatives about 30 seats, with Labour gaining 11 and the Lib Dems 19.

However, AV does not by definition help Labour and hurt the Tories. If Lib Dem voters split in favour of the Tories, and Labour voters were less willing to transfer their support to the Lib Dems there would be a different result.

YouGov repeated the same experience at the end of June. Second preferences now break differently. Conservative voters are much the same, but Labour voters are now much less likely to transfer to the Lib Dems, from 62% at the election, now only 33% of Labour voters would give their second preference to the Lib Dems. Lib Dems now break in favour of the Conservatives rather than Labour, though not by very much (38% to 33%).

None of this should come as a surprise of course – Labour voters are obviously less likely to give second preferences to the Lib Dems if they see them as Conservative-allies, and those Lib Dem voters who preferred Labour over the Tories at the election may no longer have the Lib Dems as their first preference in the first place! The impact, however, is that vote transfers from AV would now help the Conservatives more than Labour. Peter’s calculation is that had these transfers applied in the general election (admittedly a rather false scenario!), the Conservatives would have lost only 2 seats, while Labour would have lost 13.

Precisely predicting how AV votes and transfers will translate into seats is a complicated matter (though one we’ll have to tackle should the AV referendum be successful), but the point is that the assumption it is damaging to the Conservatives is based on Labour and Lib Dem supporters disproportionately swapping their second preferences between one another. If that changes, as the polling suggests, and Conservative and Lib Dem supporters instead tend to second preference each other’s party, AV would end up disproportionately hurting Labour.