I’ve seen various people pick up this article by Julian Glover today, so it probably needs some background. On Wednesday there was a joint conference with the BPC and the National Centre for Research Methods and at the end of the day the chairman put all the speakers on the spot and forced them to make a prediction of the election result (actually he asked them to predict the Conservative lead at the general election, but everyone apart from Nick Moon gave predicted shares).

The title of this article is a line that Bob Worcester likes to roll out at election time when he is asked the question (and in fact, Martin Boon from ICM prefaced his answer on Wednesday with it). Opinion polls ask what would happen in a general election tomorrow – apart from the final eve-of-poll figures when there actually is an election tomorrow, they don’t try to predict what will happen in several months or weeks time. Wednesday’s predictions are just the personal opinions of the particular pollsters who were there.

For the record though, as Julian reports, Nick Moon of NOP predicted an 8 point lead. Simon Atkinson of MORI predicted a tight 4 point lead. Everyone else clustered closer to a 10 point lead, the highest specific lead predicted was Andrew Cooper of Populus who plumped for CON 41% to LAB 29% (though Martin Boon went for a Tory score of “one point more than the next highest”, so I suppose he actually gave the most Tory answer).

Martin’s answer probably gives you the hint that these were not the most serious of predictions! As Peter Snow says, it was just a bit of fun. Before someone asks, I wasn’t speaking so didn’t get put on the spot for a prediction ;)

Anyway – for those anxiously waiting for the next poll, I’m expecting at least one this coming Sunday.


99 Responses to “Polls don’t make predictions, but pollsters sometimes do”

1 2
  1. I don’t really know what basis there is for a 4% lead one poll went as low as 6% but that as far as I can see a rogue poll. I think that one was just put in there to sound exciting.

    I think it’s way to early to know what the results will be. If Labour want a chance of winning the next two months are crucial as they need to cut the tories lead to about 5% in a few polls by the end of march if they want any chance of winning.

    The crucial things that could change things will be, Iraq enquiry, The economy, unemployment, manifestos, Election posters and the 3 debates.

    I think Labour’s polls of Feb-March will be there highest but may well fall nearer to an election especially if there is a double dip figure in April.

  2. 8% lead is in-line with the model I’ve been using (click on the link in my name for my site)… And 4% is in line with the short-term regression trend.

    That should give pause to those boosting for a Clear Conservative Win… And that the rest are only around 10% is a worrying sign that any seat majority might be a slight one.

    I doubt this will sway those who insist that the Conservatives will still gain a healthy majority of seats because of “The Marginals” or “The Nation won’t tolerate it” or “Insert Argument Here”.

  3. @JAY BLANC
    There are 2 out of what, 6 or 7 pollsters who agree with you.
    This means 4 or 5 do not, and go on to say the Tories can expect a “reasonable” majority. Further, you should read Mike Smithson today. Mike has an interesting article on his site regarding overboosting Labour figures.

  4. Re: Paul B
    Another crucial thing is Camerons almost daily speeches and anouncements of “draft manifesto” that “may change” before election. This could lead him to make another error. Listening to todays court case about those two young boys and their abusive father – the father was a product of Thatcher and Majors Britain with the massive unemployment. His character was formed before Blair was elected. The effects of unemployment can be diificult to nullify it can take years. Trying to link labour with such events could be dangereous. Reacting too quickly to events for a headline can be a mistake.

  5. The tory polling figures are weak and the trend seems to show their lead falling for some time.
    Polls will show a “protest against” vote until the election is actually called, but IMO are not very reliable.
    I will be VERY interested to see what the first few polls show once an election is actually called.
    I personally would have thought 4% could be quite close – the problem for Labour could be in the marginals where it really matters as the Tories will be putting so much more resource into them.

  6. @Paul:

    I suspect he thinks that the trend is likely to narrow leading up to the election, and that perhaps he thinks Labour will campaign better in the official campaign. As Anthony says, the polls show what would happen now, now what will happen in June. He may be wrong, I don’t know. But I don’t think it’s unreasonable to predict a narrowing of the lead to a razor thin majority.

  7. It has been said before but it was always going to be difficult for the Conservatives to gain a 6.5% swing and the 130 or so gains they need to get any sort of majority. Upping that to a worthwhile majority of course increases the difficulty. It would take circumstances that are fairly unusual in historic terms (swings of that size don’t often happen) .

    However, I think the situation is such that this event may well occur (ie, Conservatives getting a usable majority): I think if we somehow had another 5 years of Labour government we’d see a 1997-style swing at the end.

    As above I think the 4% lead is unlikely, the 6% more like it.

  8. @Sue Marsh – I’m not sure we can say the Tory polling figures are weak, or in particular decline at the moment. They lost momentum just prior to Christmas and dipped below 40% in a few polls, while Labour has and still does in most polls appear to be at the top of their recent range, which is historically a very poor range. However, the Tories have returned to consistent 40% polling and while they would I’m sure like to be further ahead, I don’t see many independent posters regarding this as ‘weak’. Yes, there are potential weaknesses in the Tory position, but that’s a different point altogether.

    @Percy Holmes – I would agree that it’s decision time on Cameron and the risk of alienation of potential supporters increases with increased policy specifics. Probably like you, I found Cameron’s reference to one specific criminal case today opportunistic and distasteful – I personally see it as poor politics and if Labour had any nouse they would remind people of yesterday’s crime figures. However, you then seem to make a similar error in blaming Thatcher for this poor families circumstances.

    On the issue of poll predictions, it is interesting watching these posts and how they change over time. Its clear to me that party supporters post more when their party has some good news in the polls. In the middle of last year the majority view seemed to be that the Tory lead would expand up to the GE. For whatever reason there does appear to have been a shift in expectations. I have said for some time I expect a tightening of the lead and would be surprised with some polls at 5-6% lead before too long. However, there are many issues that could derail Labour again so nothing is set in stone here.

  9. I think that assessment could be about right, Keith.

    This election may well be one too early for a Tory landslide.

    They are very likely to win; but they could easily find themselves with a small(ish) majority. Much as Labour would have done, had they won in 1992.

    It would almost be better for them if they lost this one – narrowly – as Labour did in 1992. After all, even with a recovery, times are going to be hard over the next few years and the Conservatives will win few friends with their rigourous cutting program.

    (Obama is a case in point; after 12 months the honeymoon is over. The US electorate seem to have forgotten that he inherited a dreadful situation from 8 years of Republican government. The Conservatives in the UK may find themselves similarily disregarded in 12 months time).

  10. @David in France,

    I think the Conservatives are in a slightly different position to Obama. The “Golden Boy” gloss has already worn off Cameron and his party have been quite deliberately reducing expectations about what they can achieve in the current climate. This may cost them a couple of percentage points but will help them when it actually comes to governing.

    I agree that their majority will probably be small, however. I am guessing under 40. But as has been discussed before, with abstinence by nationalists and cooperation on some issues from the DUP a small majority will probably be manageable. Of course it would leave them vulnerable to poor performances in by-elections whittling away at their majority, but I suspect that the new parliament, with its massive influx of young blood, will throw up less by-elections than normal.

  11. I think that april 26th will be the nail in labours coffin i am betting on negative growth 1st quarter. I think that people on here underestimate the amount labour is now hated

  12. @NEIL A
    You sum things up with accuracy in my opinion.
    @PERCY HOLMES
    Your pathetic excuse for the father of those two monsters just about sums up the state of mind of the left. I was made redundant in 86, 96, and twice in one year in 97. Like Norman Tebbits dad I got alternative jobs. I have 2 children, the girl is a mother and a science teacher, the boy is a Captain in the Parachute Regiment. Both are graduates. Should I have turned to drink and let them go to hell?

  13. I’m with Keith and David.

    Also, if you look at the broad picture it seems that there has never been any enthusiasm for Conservative beliefs or their favourite solutions but there is also a feeling that Labour are tired and the electorate are tired of Labour. The small majority or hung result actually reflects what people think of the main parties and Con-hung would seem to be just deserts.

    Is it difficult to accept that even FPTP can produce the result that voters want?

  14. @SUE MARSH
    Your comment is effectively saying the polls are wrong and Anthony Wells does’nt know what he is talking about. Why do you think the headline figures on this site give the Conservative party a 44 seat majority? I fully realise things can change either way, but for heavens sake can Labour supporters leave Walter Mitty at home for a change.

  15. @John:

    “Is it difficult to accept that even FPTP can produce the result that voters want?”

    Given it’s historical accuracy, I feel it is! FPTP is meant to have a winner’s bonus (personally I quite like that aspect), for it to not have one is a surprise.

  16. @ ALEC

    “I found Cameron’s reference to one specific criminal case today opportunistic and distasteful ”

    I heard Cameron-he specifically said that the problems he was talking about today featured under both parties.He referred to many other cases-including Bulger.He went out of his way to say that thse factors were not specific to one place “like Doncaster”, and were not nationwide.
    He did his level best to address the very predictable portrayal which Byrn & Ball would make-and still made despite his best efforts.

    I think however that timetabling that speech today was a misstake.

  17. @David in France

    “It would almost be better for them if they lost this one”

    The premise being I suppose that things will be so tough they will become mighty unpopular.
    But thats only half the story by the time the following election comes round the Tories will be able to point at lower unemployment, more overtime, lower inflation, a growing economy. And thats not necessarily because they are any good, thats purely the economic cycle. Add to this a 10% cut in MPs (and it will be labour MPs) entirley self serving we need 25% less, but hey ho. And if thats not enough then there is always the option of stopping Welsh and Scottish MPS voting on English affairs.
    NO this is a very good election for the Conservatives to win.

  18. Anthony expects a poll on Sunday. I think this one will be massively important for Labour. They need a poll that shows ‘don’t knows’ choosing Lab/ Libdem because they cannot expect to make a dent in the Cons 40%; that appears to be set in stone.

    I do think that Cameron’s personal ratings will begin to align with the 40% instead of being much higher.

  19. BRYAN:-

    I think you could be right-but I wish I was as confident about “lower inflation”.

    I think there might be a rise in inflation to cope with.

    Also , the “opposition” will not be at Westminster-it will be the Unions-probably on the streets.

    Implementing supply side reforms in Education will be a priority and NUT willfight it every inch of the way.
    Pubic sector jobs will go & UNISON will resist strongly.

  20. @Colin – it was largely the timing and general presentation of the theme where I found his planning wanting. It’s a symptom of bandwagon jumping, which he doesn’t need to do. It should have been easy to foresee the way it would be reported and there are already plenty of media references to this that are not very supportive. He also goofed on violent crime figures – he really does need to be careful when claiming a factual basis for what he saying, especially with TV debates in the offing. In the prism of an election campaign such slips are the kind of thing that destroy credibility.

  21. Bryan

    I see what you are saying. And you could well be right.

    Regardless of who wins the GE in 2010, by the time of the next GE (say, 2014) the economy and so on should be in better shape.

    But will the party who steers the rough seas of 2010 – 2014 get the credit?

    As you say, possibly so.

    But not necessarily so.

    History shows that anything can happen…

    In 1992, for example, the Conservatives presided over a massive recession but won the election. Then, ironically, when the economy and most other indicators were doing well – in 1997 – they were kicked out en masse.

    I guess we’ll have to wait and see.

  22. @ Sue Marsh

    “The tory polling figures are weak and the trend seems to show their lead falling for some time.”

    For your benefit as you seem to be either unable to read figures and/or accept them staring out from a screen.

    Since the beginning of December, the Tory vote has averaged 40.3%.

    The last 10 of those 20 polls average 40.3%, and, behold, since the beginning of September, its been 40.3%

    I can only assume then there is a defficiency somewhere in your understanding of simple maths.

    Would you like to elaborate on your opening sentence a little?

  23. Oh, and Anthony, for the record, I’d like to complain about some of the comments made on this thread by Roland Haines and, just, by Barry P.

    None of these are aimed at me but both are unneccessarily personal and abusive in tone.

  24. @ David in France

    “None of these are aimed at me but both are unneccessarily personal and abusive in tone”

    If there are figures clearly displayed, which there are, and, accurate analysis of them takes but 60 seconds with the simplest of calculations and, as a result of that ‘analysis’ a sentence such as:

    “The tory polling figures are weak and the trend seems to show their lead falling for some time.”

    is posted; to my mind, and I hope most, there is either one, or a combination of two or more of the defficiencies I mention present. It is in no way personal or abusive, it’s factual, nothing less.

  25. @Barry P

    You are, however, being very selective in what you state from your figures. And you are presenting them in a way that is tilted in your favour. And stating that anyone who disputes your presentation of them is “unable to read figures”.

    As an alternative, someone could present a scatter graph of polling, with a moving average, and show a trend… This would produce a substantial presentation of figures that may well show a different story than your one.

  26. The guardian support labour but hate brown. Its always interesting where each article comes from it showed that two pollster thought it would be close the rest thought it over ten %

  27. I’m struggling to understand Simon Atkinson’s numbers but at least he added some variety to the various predictions.

  28. @ Jay Blanc

    “You are, however, being very selective in what you state from your figures”

    OK, lets go back to the beginning of July, is that unselective enough for you. The average from there is 40.4%. Or, would you like me to go back to the beginning of May, 40.03%, that shows an increase in the Tory lead. Or, how about, the beginning of last year, 40.6%

    Which date up to present would you like me to cover, considering Sue Marsh’s comment “..their lead falling for some time.”

    And it’s not my presentation, it is the pure and simple mathematical average analysis of different sets of known data over differing, extending periods.

    If you can show different sets of average leads by the Tory party by other mathematical, logical methods, please let us all see how you would present it. Until you can do, Sue Marsh’s comment is 100% incorrect.
    And, I don’t think for one Sue actually analysed any figures, it’s probably just a partisan statement based on what Sue would like to see, not what is actually the actual lead stability.

  29. I haven’t analysed the figure myself, but strictly speaking your figures don’t rebut Sue’s point. she was talking about Tory lead, you spoke about Tory polling, not compared to Labour.

  30. @Barry P

    Um… Please click on my name on this comment. You’ll find it links to my site, which has a nice graph showing that scatter graph of polling, moving average and trend plot I was talking about.

    You’ll see that your ‘average polling of 40’ is a little bit of an illusion. The Conservatives had a period from the end of July to the start of October of all their polling being above 40, plotting somewhere around 43. In fact, from the period from July to November, there were only two polls that put them below 40.

    Going from plotting loosely around 43, to plotting tight to 40, *is* a drop in polling for the Conservatives.

  31. @Quincel

    The Conservative Lead figure has dropped five points since October, in a sort of ‘bouncing down the stairs’ pattern.

  32. OK, I did crunch the numbers, because I like facts and figures. Feel free to double check them, I did write them down fairly fast.

    Tory Lead:

    Jan 2010: 11.71%
    Dec 2009: 11.46%
    Nov 2009: 12.20%
    Oct 2009: 13.78%
    Sept 2009:12.82%
    Aug 2009: 15.86%
    July 2009: 15.13%
    Jun 2009: 14.62%

    It’s not a constant trend, and most of it happened from September to October, but over the past 9 months or so the Tory lead has definitely dropped from around 15% to around 12% or perhaps just under.

  33. By the way, those % are just the average Tory lead from polls in that month.

  34. And for what it’s worth Jay, I am checking out your blog.

  35. My average ‘lags’ a little behind yours, having the 15.8 in October. This is just due to using a Modified Moving Average rather than Periodical Average.

  36. There is no statistically significant trend of a reduced WMA CLead over the last 1 or 2 months. Unless something significant changes, we should expect a CLead of about 11%.

  37. @ Jay Blanc

    I may be mistaken here but, do your graphical displays only include selected pollsters.

    I may be wrong again here but, do you intentionally omit one of the pollstsers used on this blog because you disagree with their results?

    As I said, I may be wrong here so would you mind solving my dilemma please.

  38. There’s a poll exclusion policy on my site. I exclude polls that don’t ask a formal voter intention question, which excludes polls by AR since they only ask a party support question. I also exclude polls commissioned by The Sun as they’ve been playing silly games with selective reporting, and pressured YouGov into dubious practices. I still accept YouGov polling commissioned by anyone else.

  39. Explain this to me….

    If the Tory poll lead has declined by 3% to 12% in 9 months, and there are 4 months to the election, what is the rationale for believing that this trend would take the lead down below, say, 10% by the general election?

    I think on current facts the safest assumption would be that the lead would be somewhere a couple of points either side of 10%, subject to the various factors at play.
    These could include the possibility of an improving economy helping Labour, the belief that the governing party picks up votes in the campaign, the belief that the Tories are outperforming the UNS in their target seats, the belief that Ashcroft money will help them take some of those seats, the possibility that Brown will be hurt by his Chilcott appearance, the possibility that Tory confusion over tax policy will have legs, the possibility that one of the leaders will emerge head and shoulders ahead in the debates, differential turnout on the day (bad weather, whatever) etc..

    Personally I think people’s views about whether the lead will be 4%, 6%, 10% or 14% are entirely based on their own subjective take on what effect these factors will have. Noone is “right” or “wrong”. Everyone is working from their own hunches, and any one of us could be right or wrong.

  40. @ Jay Blanc

    “There’s a poll exclusion policy on my site. I exclude polls that don’t ask a formal voter intention question, which excludes polls by AR since they only ask a party support question. I also exclude polls commissioned by The Sun as they’ve been playing silly games with selective reporting, and pressured YouGov into dubious practices. I still accept YouGov polling commissioned by anyone else.”

    AKA, only use selected data which you agree with, trash the rest.

    Anthony includes these polls for good reason on this blog, if you wish to exclude them, that is of course your choice. However, when comments are being made within a thread on a blog, who’s data is clearly visible and the reasoning behind it’s inclusion evident, you are not singing from the same hymn sheet as most others.

    “You’ll see that your ‘average polling of 40? is a little bit of an illusion”

    Well, do you consider these figures an illusion? I have used the figures on this blog and averaged them every 10 polls for the last 140. i.e. the first figure is an average of the last 10 polls, the second figure an average of the last 20 polls, the third figure an average of the last 30 polls and so on.

    40.3
    40.3
    39.9
    40.2
    40.6
    40.3
    40.3
    40.4
    40.3
    40
    40
    40.2
    40.3
    40.5

    Pretty good illusion don’t you think?!

  41. “I also exclude polls commissioned by The Sun as they’ve been playing silly games with selective reporting, and pressured YouGov into dubious practices”

    So YouGov engage in dubious practices when pressured by their paymasters? That’s quite an accusation to make about a BPC member. I’d be tempeted to issue a retraction if I were you.

  42. @ Shopkeeperman

    “I’d be tempeted to issue a retraction if I were you.”

    So would I but that will ruin his argument

  43. WOW glad I opted out of taxonomomics and statistics at school.
    You guys are arguing at best about 1% to 1.5% swings. Nit picking and obsessive are the words that come to mind. Sorry !!!!

    Please do not bombard me that its really about 1.2% to 2.3%……

  44. Barry P

    ** I have used the figures on this blog and averaged them every 10 polls for the last 140. i.e. the first figure is an average of the last 10 polls, the second figure an average of the last 20 polls, the third figure an average of the last 30 polls and so on.**

    I hope you don’t mind me saying -your method is flawed. Why are you using the older samples in all your averages? They will be out of date and distorting the recent polls and their averages.

    It would work if you drop the older polls as you use new one’s as AW does. Will be interesting to see what happens if you do it that way ;-)

  45. @ AL J

    I am merely pointing out that a statement such as ““You’ll see that your ‘average polling of 40? is a little bit of an illusion” is incorrect. I agree that to use perhaps a moving average would be better but, the point I am making can easily be checked by anyone.
    It shows that current figures of Tory support are virtually identical to an average Tory support figure for the last 140 polls etc. But thanks, your point is accepted.

  46. @BARRY P

    “It shows that current figures of Tory support are virtually identical to an average Tory support figure for the last 140 polls etc. But thanks, your point is accepted.”

    er, um, check the scattergraph for the last 6 months on ‘electoral trends’ before you make such an incorrect statement

  47. @Barry P

    “I may be wrong again here but, do you intentionally “…

    …include in *your* ‘best gloss on polls in favour of Tories I can scrape from the bottom of the barrel’ comments, a pollster consistently miles off the main (and reputable) polling firms ? Not very partisan huh genosse !

    Furthermore do you count the polls commissioned by that wholly neutral organ ‘het zon’ with its wholly neutral deployment of data and ‘polling message’ ? Coverage and usage that brings polling utterly into disrepute ?

    If you do, oh boy are you in for a shock later this year as you will be anticipating a majority of 30 or more for Camerborne.

    @JAY BLANC

    I’ve found your blog over the last couple of months both educational and valuable and bearing much similarity to my anecdotal evidence from 2005 voters across the political spectrum about where their thoughts are leading them.

    Keep up the good work mate

  48. To everyone – forget the statistics! It’s about what people think, and guess what – people can change their minds. NEIL A has got it right – lots of stuff happening, and any predictions on what will happen have more to do with personal belief of the predictor than anything else. As soon as a poll is published it’s out of date. I don’t know for certain what will happen, but in general I believe that things will either get worse, get better, or stay the same.

  49. @ Rob Sheffield

    Sorry, based on the polls on this blog, my figures are correct.

  50. I trust the most accurate pollsters – Angus Reid – were not invited?

1 2