Bill Bush answers your questions about polls...

Question from Samara Shah, elu94141@srv0.law.ed.ac.uk

What prevents a student, registered on their term time and home town electoral register, voting twice?

The law. Get found out and they'll take legal action against you, so even if tempted it's wise to resist. When you go the polling station they cross off your name as having been given a voting slip. Ballot records are kept for weeks after the election in case someone challenges the result, and wrongly cast ballots is one of the things that can be the basis of the challenge. It's rare, but that's because most elections give clear-cut results.

Time to go - many questions unanswered, mainly because they were too difficult!

Thanks for your interest, and your patience,

Bill

Given the huge amount of regional variation in voting patterns across the country (look at the South West compared to the Midlands, as well as the different political situations in Scotland, Wales and Northern Ireland), surely the polling organisations would do better to publish regional polls. Would the BBC consider publishing regional results from its Exit poll on election night?

regional polls are expensive - 1,000 respondents are usually needed for a GB poll, and by an annoying statistical truth, you also need 1,000 for each region. so to do a full GB regional spread you probably need a sample size of 13 or 14,000. But it is absolutely right that regional variations exist and are important. Some polling companies add together several polls to get decent sized sub-samples - so watch out for Gallup and Mori in particular.

Looking into the future, say 10 years or so, can you envisage the break up of the two main parties, and the creation of a right wing party, with say Redwood, Portillo etc., a left wing party, with say Cook, Prescott etc. and a center party, with say Major, Blair, Clarke etc. Or will they continue to have two incredibly split parties that stay together just for the sake of electiblity rather than beliefs ?

I think party break-ups would only follow a change in the voting system - it happened in New Zealand when they introduced PR. The UK system is tough on small and new parties, and most politicians remember just how hard it was for the SDP to get established despite all their advantages.

I have a couple of questions: 1. Why hasn't the BBC commissioned any polls during this camnpaign? Channel 4 have done so, and do not appear to have been attacked for partiality as a result. I assume there will be an exit poll forThursday evening? 2. Why is there such scant coverage of Northern Ireland? This is an area where a BBC-sponsored poll would fill a gaping hole left by other parts of the media. It would be extremely interesting to see what the balance is between the UUP and the DUP, the SDLP and SF, and whether the Alliance Party is being squeezed by the extremism of both sides. 3. Why are Don't Knows reported so rarely? If you have a series of polls showing, say a 20% lead for one party or other, but there are on average 25% don't knows, who won't necessarily split in the same proportions as those who do express a preference, then surely the headlines figures that are reported are misleading? thank you for your answers

1. There are so many other polls that we thought we'd save your money and report other people's. Anyway their record has been less than perfect, and there is always a tendency to highlight your own poll's findings, so it is easier to be objective if you don't commission any. We did do a poll of young voters, but we were more interested in their attitudes than in their voting intention (ie what they are interested in, rather than how they would vote).

2. Polling in Northern Ireland is not easy. Respondents are often shy about revealing heir voting intention, especially if they think that someone might disapprove of their answer.

3. I agree, the don't knows (and the "might change my minds") should be given greater prominence. We try, but we'll try harder in the next few days.

Bill, I'm trying to work out what the party situation is in Bristol West (currently held by William Waldegrave) - tricky at the best of times but seemingly very complicated this time. As pointed out on the BBCs web site a traditionally strongly Conservative ward (Westbury) has been added to the constituency... so we could say that the constituency is now divided as follows: South/West - LibDem, East - Lab, North - Cons. One also has to consider the fact that there are nearly 20,000 students from both UWE and Bristol are 'in town' (Isn't it unusual to have a general election when students are at University?). But the parties campaigning strategies are very odd... the Conservatives have not been seen at all in the constituency (I literally have not met anyone yet who has seen hide or hair of them) - Labour have made a small effort at the beginning + 1 constituency-wide leaflet (and thats it) - The LibDems have been spamming the constituency with seemingly every bit of paper in the country - and really interestingly the Referendum party have been holding public meetings and making a lot of noise in the North of the constituency where traditionally a lot of the conservatives reside. Whereas there is a disinct possibility that LibDem vote = Labour vote and the Conservatives will take it... there seems to be a lot of take up amongst the Cons vote of the Referendum Party message! You hinted yesterday that Labour/LibDem might divide effort so that they weren't fighting each other, so much as the Conservatives. Although they vehemently deny any Lib/Lab pact - does the BBC's Political Unit have any intelligence on whether Labour have let the LibDems have a clear run at Bristol West? Is this a usual occurence - where are the LibDems copitulating to Labour... Isn't this odd given that I seem to remember that Bristol West is in Labour's Top 50 target seats. Any comments on the above?

Bristol west one of the few "3 ways" in England. Where Lab and LD do fight they often tear lumps off each other - see Rochdale and Oldham for some gems. But generally the try to keep out of each other's way. The LibDems often do look highly visible, because in many areas there is only one obvious target for them. Eg in Birmingham they are putting most effort into Yardley, but in the other seats they'll be largely absent, and will be trying to encoursge their activists to go to Yardley. The 3 ways are very hard to call, and pollsters don't help much at that local a level. I'd be interested to know how the students are taking the campaign - are they all children of the '80s hoping to be accountants and bankers, or are they more radical, or just ignoring the whole thing? If there really are 20,000 of them they could have a big say in the result - if they are registered in Bristol and if they bother to vote.

Great site, I was just wondering, using historic data and assuming say Labour polling 47% to a Conservative 37%, what time would you think Labour would win the crucial seat. Essential information for those planning a party!

My view of parties is that they should be over by about 10.30. Unfortunately the first result won't be in until 11.00 or so (probably sunderland South) and then we hit a real delay. The English counties and a few other local authorities have elections on the same day and this really slows down the count. So the winning post moment won't be until say 2.30, or even 3 am. By then everyone should be tucked up in bed listening to the radio (because that's where I'm working on the night) .

Where / when are the various polls conducted? In many years of reading poll figures, I have never been asked my opinion, nor do I know of anyone who has been asked. Given the vast number of polls which have been conducted since I became entitled to vote, I am somewhat disturbed by this.

Polls usually are face to face, in home or telephone. sometimes they are taken from the electoral register. However there are over 35,000,000 voters and only 1,000 or so needed for a poll, so the chances of being polled are pretty small.

You say that the campaign for 'best pollster' is just as savage as that for no.10, but ICM's effort suggests a labour win of ~40 seats, somewhat more believable than todays poll in the Telegraph which suggests a landslide in excess off 200 seats. Are ICM right afterall?

Of course the Tories can do better than the polls and the pundits predict. They've been around a long time and won a lot of elections, and in 1992 they won by nearly 8% even though the polls said that they were behind. However the polls can only report what they find, and with some suggesting big Labour leads, and only ICM getting down to a Lab lead of 5% it would seem that the difference is methodological, rather than ICM simply being better at getting the "truth".

IN A COUNTRY WHERE OUR CONSTITUTIONAL ARRANGEMENTS LEAVE THE MAJORITY OF THE POPULATION UNREPRESENTED, DO POLLS REFLECT THE TRUE VIEWS OF THE ELECTORATE OR DO THEY CONSIDER THE POLITICAL ENGINEERING, THAT WE HAVE COME TO ACCEPT AND REFLECT THE INEVITITABLY BIASED OUTCOME?

Interesting view that our constitutional arrangements leave the majority unrepresented. All voting systems have flaws, and there is no unique "best" system. It just depends what weaknesses you are prepared to live with. The Israelis have a pure PR system, yet some criticise it for giving too much influence to tiny minorities. Disillusion with politics and Governments seems to be pretty widespread regardless of the electoral system, so may be it's got more to do with modern times than simply a problem with the UK. Polls can only ask questions about the real choices immediately available, anything else tends to be so hypothetical that respondents get lost in trying to work out an answer. So it is true that there may well be thousands of different opinions out there that get pushed into the limits of the existing debate, and many issues get lost. But at the risk of sounding pious, democracy shouldn't be passive - if voters think politicians are ignoring them then they should give them a hard time. Politicians can be rude and evasive with media interviewers, but find it much tougher when "real people" have a go.

When we see the headline figures of support for parties, what, if any account is taken of 'Tactical' voting. For example, someone who may be a Labour supporter would probably give Labour as the party he/she will vote for. However, he may want to see the Conservative candidate voted out and vote for the Liberal Democrats. This being the case, how would this change the true picture?

The LibDems are hoping (and planning) that local factors and tactical voting/tactical campaigning will be their saviour. They are putting a huge amount of effort into say 50 seats, which they hold or come a good second. Most of these are straight Con-LibDem fights and so they hope that Labour voters in those areas will switch to keep the Tories out. This can work - in 1992 they won almost as many seats as the Alliance did in 1983, but with a third fewer votes. And in local elections since 1992 the LibDems have run consistently third, behind Labour and the Tories (and fourth in Scotland, behind the SNP as well) but have managed to come out second in terms of seats, because they are learning how to concentrate their vote. If they can repeat this next week they may win several more seats than their simple national vote share might suggest.

Sir, With the new polls out this morning many are considering that the ICM poll yesterday was simply a rogue result. However, would you agree that this could well be a closer representation of the true intentions of the British electorate than the polls which show a twenty percentage point lead for Labour. Most of these polls have also shown that voters class the caring issues such as Health, education and benefits as the issues they will be voting on next week, however, would you not also agree that the last election demonstrated once again that these issues, which are classed as important by respondants of pollster`s qustions (partly to look as if they care), is replaced by a more `selfish` view once they enter the polling station. As the Tories have been close to Labour or beating them in polls of who the voters trust over tax and the economy in recent weeks, this combined with the results of the ICM poll (the margin of error of which could mean just a 2% lead for Labour, would appar to point as they did at this point in the campaign five years ago, to a tight election, possibly a Tory win rather than the landslide some are still predicting I would be intrigued to see whether you would agree with this assesment. Yours, R Avis

I agree that the Tories will be pleased with the findings on tax and the economy. They are usually the real driving forces behind voter attitudes. But something odd may be happening this time - Labour's changes of economic policy might have reduced the sense of risk,so voters feel that whoever wins the economy would be little changed. It's interesting that a MORI question found that even though 48% thought that the Govt had produced an economic recovery, most of those 48% felt that it was still time for a change. Could be that sleaze, splits on Europe, or just boredom, are making some voters think that a change of Govt is desirable. If so then we'll have to stop using the Clinton cliche - "It's not the economy, stupid"?

I understand that 25% of the electorate are undecided, and that these are the crucial 'floating voters' who will decide the election. If we do not know their opinion then what is the point of all these irrelevant polls ?

The point is surely that we only know that c 25% are undecided because of the polls - so we can then judge that there is a lot to play for. We can also try and find out why people are undecided, and what might help them make their minds up. Polls don't tell us about the future, they give us a glimpse of what attitudes were a few days ago. Thuings may change, that's why elections are such fun (as well as being important etc)

Loads of questions - great response. I want to get round to more of these, so although I have to disappear now, I'll come back tomorrow (Friday) at 13.00 to finish off. I'm head of the BBC's Political Research Unit so don't just restrict your questions to polls. Hope to hear from you,

bye

Bill

We hear a lot about the 'private' polls that the parties use, and also about their polling based on canvas returns. How different are these from the polls published by the newspapers, and what should we read into the claims that these polls show a different result?

Don't trust any party's private poll figures unless they reveal the methodology, the full questionaire etc. Otherwise even more unreliable than the British weather.

Do any of the polls take account of our constituency system or do they simply take an across the country sweep?

They can't take local factors into account - not without a huge increase in sample size. So they assume that if they pick up a change (Con up 5% say) then the Con vote goes up by 5% everywhere. This is called uniform swing, which you'll see referred to in some of the more pointy-headed articles in the election supplements. In real life things are more varied - so for example the LibDems might do better in the South West of England than their national vote share might suggest because of local and regional factors that the polls don't pick up.

How do the polling companies take account of ex-directory and, perhaps more importantly today, cable telephone numbers which aren't listed?

They use some variant of random dialling. this means that they don't take numbers out the phone book, but select from all potentially available numbers. In theory, if your phone number falls inside the range used by phone companies you have an equal chance of being called. Anyone ever been phone-polled? apart from someone selling double-glazing?

Why don't pollsters adopt the tactic of inviting interviewees (shy or otherwise) in face-to-face surveys to adopt some anonymous strategy for indicating their voting intentions - such as the selection and depositing of small counters to and from bags, say? Psychologists have used more sophisticated variants of this strategy for many years. I acknowledge that in a busy high street there are various practical problems with this but as you know in face-to-face surveys 30-36% of respondents say 'don't know' and maybe half of those are 'won't says'.

Absolutely right. In our exit poll we ask people to fill in aballot paper rather than tell an interviewer. we find that repeating what they have just done in the polling booth, with a guarantee of anonymity, helps lower the refusal rate. In 1992 we got down to a 15% refusal rate - and still got the Tory vote too low by 2%

The polls only indicate voting preference for those who are prepared to state what their intentions are. What is the proportion of 'don't knows' or 'refuse to answer', and how can this be accomodated into the weighting of the polls?

Polls can get huge refusal/don't know rates (30% if they are unlucky). They often only report the people who refused the voting question, but who did answer other questions - so what happens to all the people who don't take part at all? Or who in telephone polls are never home to answer the phone. Dealing with the refusers is the biggest test the pollsters have.

In 1992 the polls underestimated the Tory vote, and this time around the polls are attempting to compensate for Tory shyness. Do you think that this might be going to far, could the polls be seriously underestimating the amount of tactical voting by people who have no love for Labour or the Lib Dems but just could not bring themselves to vote Tory? How many points are tactical votes worth in a target seat?

The pollsters are using such different methods that they can't all be right. The competition for "best pollster" is at least as nasty as the scrap for number 10. Tactical voting is one of the great mysteries of voting. It clearly happens, but not much, except for by-elections. Most voters only follow politics in a generalised way, so don't make detailed local calculations. Often what we call "tactical voting" is really the result of "tactical campaigning" ie a party pulls out all the stops in one area, and doesn't bother elsewhere. As a result of all the effort, voters in the target seat come round to support the hard-working party and vote accordingly - but they are voting for their first choice so it isn't really tactical voting.

Concerning the answer you gave to John Evans concerning possible reasons for why the ICM Poll is so different from the others Don't the other polling organisations also use the same methods that ICM use, Telephone Polling, Quota Samples etc ?

ICM really are different. Their "shy Tory" routine is unique. For example Mori use face to face interviews rather than telephone, and don't adjust for shy Tories at all.

I have heard representatives of polling organisations say that their job is not to predict a final election result but to report the answers - truthful or otherwise - that the public give to their questions. Yet they also adjust their findings, which implies that they are actually trying to predict a result. Do you think the pollsters are trying to have it both ways?

Polls can only tell you about the past - it's how people said they felt a few days ago. But the newspapers who report the polls only want to predict the result, so pollsters often see their work exaggerated, but as it's the client who pays, it's hard to resist sometimes.

Channel 4 news last night highlited the large amount of don't knows and voters who could still change their minds. Do the polls from the last 2 days give any indication to which party (if any) is most likely to benefit?

Conventionally the "don't knows" are often the "don't votes". The turnout will be 75% or less, so the polls have to make a guess about who will actually go to the polling station. Usually the late-switchers only go for one party rather than another if there's some big campaign disaster. Otherwise they divide themselves across the party spectrum much like everyone else. So numerically they could make all the difference, but a party has to pull out all the stops to get them (or hope that someone else messes up bigtime)

If the Conservatives were to win this election, would political opinion polls be finished?

If the Tories win, and the polls don't pick up a big late movement, we'll all have to say "what's the point". If they do pick up a big late swing, then we might say, what's the point of polls before the last few days - after all, the Labour leadd has been big since 1992, and if the Tories win then it will prove that voters have two mindsets - one when an election is happening (3 weeks every 5 years) and another when an election is irrelevant.

Is there any justification for the dominant role opinion polls play in our electoral process? They don't simply predict how people are going to vote; they are a major influence on how people DO vote - particularly in constituencies where tactical voting is likely to be a factor. In the absence of a sensible electoral system (i.e. one in which the voter's best policy is to vote for the candidate or party they want to win), should not the publication of opinion polls during the run-up to an election be made illegal?

Good point. I think that increasingly voters watch the polls and react accordingly. But you could say that it is information that they need to make their decision - what if you want Labour to win, but not with a landslide? Or if you are an-anti-Labour voter and want to get a sense of how best to use your vote to beat the red (or purple) menace?

anyway, if you were to ban them you'd have to ban carrying them out, otherwise you'd get "secret polls" being leaked selectively, or published in Ireland or something.

Is there any difference in the polling method used by the ICM which explains why they are so out of line with the other polls?

ICM use very different methods:

first they have switched to telephone polling

second they use random not quota samples

third, they compensate for "shy Tories"

John Curtice, Strathclyde University and all-round good thing, has got a terrific article in today's Guardian on just this point. Anyone who reads the article will wonder how a supposedly intelligent newspaper could have gone mad with its headlines yesterday.

This week in the Cambrian News which covers the constituency of Ceredigion, the Labour ccandidate has leapfrogged from fourth to first place, which suggests an overturning of the Plaid Cymru ex-MP Cynog Dafis. The opinion poll which puts Labour on 52% and Plaid on 32% is very significant as it was highly accurate at the last election when Plaid leapfrogged from 4th to 1st place beating the Lib Dem sitting Mp. This suggests an upset in a fairly Plaid strong area and the effect of Universities on election results in particular constituencies. I wonder whether you would like to comment on this? JJKnox

Single constituency polls have an even worse record than national polls. And Ceredigion is especially hard to call, because it's the kind of rural seat where voters really care about the individual, and so the party label is less important. It's also got a University in it, and university seats are also likely to be unusual. I don't think Labour are building their plans around a win here, but who knows? At least the poll makes the campaign more interesting

Is the 5% lead given to the Labour Party by the recent ICM Poll just a blip, or does it indicate that the Tories might still win on the 1st of May?

Hi, running a bit late, so apologies for that. Looks like quite a few of you want to talk about the ICM poll that showed Labour at just 5% ahead. So here goes:

So far, it looks like a blip. There's no law that says polls are right, but the ICM poll is so out of line that it feels unreal compared to the others. But if the others come in line by the start of next week we'll all be saying how great ICM were to spot the change first.