Opinion Polls in the North: Sheer incompetence, or something more sinister?

It has long been accepted that the track record of ‘professional’ polling firms in the political sphere in the north of Ireland is one that leaves much to be desired. However, the consistency of their incompetence begs the question: are these organisations professionally incompetent? Or is predicting election results in the north of Ireland simply beyond the capacities of these pollsters? Or, given the recurrence of a theme indicating a reduced level of support for Sinn Fein and the DUP, is there a more sinister reason for the continued publication of such wide-of-the-mark polls only days/ weeks prior to elections?The evidence of the collective incompetence of the pollsters is overwhelming, and is outlined below from each of the last four major elections in the six counties (2001 Westminster, 2003 Assembly, 2005 Westminster and 2007 Assembly.)

The pciture within unionism is similar to that within nationalism. DUP/ Sinn Fein continuously fare poorly in the opinion polls, whilst UUP/ SDLP fare much better in the ‘make-believe’ world of the pollsters than the real world of elections.

But whereas the trend within polls in recent elections has been one which has at least recognised the real supremacy of the DUP within unionism, this has not been the case within nationalism, which has lead to many false dawns for the SDLP when the real polling begins and ends.

The recurrence of a theme indicating a revival in fortunes for the SDLP is unmistakeable: in 2001, the pollsters put the SDLP 9% points AHEAD of Sinn Fein going into the election: they ended up behind Sinn Fein in first preference votes, suggesting the poll was nearly a full 10% points wide of the mark in relation to levels of support within nationalism.

In 2003, the pollsters put the SDLP 2% points ahead of Sinn Fein: once again, Sinn Fein emerged victorious by 6.5%, suggesting the pollsters were 8.5% off in their predictions of the margin between the nationalist parties.

In 2005, the pollsters had the two nationalist parties level pegging going into the election: once again, Sinn Fein emerged 6.6% ahead of the SDLP.

And in 2007, the pollsters put Sinn Fein 2% points above the SDLP going into the election. Unsurprisingly, they got it wrong- again. The margin between the two nationalist parties had grown in real terms to 11.3%, meaning the pollsters were out by more than 9% points.

All of which is important when we consider the ability of political spin doctors (particularly, though not exclusively, from the SDLP) to ‘persuade’ elements within the media to run with stories suggesting several of their candidates are in line to make electoral breakthroughs. Opinion polls making seriously flawed predictions contribute to the false hype around specific candidates, which ultimately culminates in expressions of shock and surprise when the electorate deliver a completely different result.

Examine the UMS/ Belfast Telegraph poll published on 31 May 2001. In this poll, the party’s share of the vote was UUP 25%, DUP 14%, SDLP 25%, Sinn Fein 16 % and Alliance 5%. The margin of error of 3% was ridiculously breached in the case of 3 of the 4 main parties when the ballots cast returned the following results:
UUP 26.8%
DUP 22.5%
SF 21.7%
SDLP 21%
Alliance 3.6%

Now, move forward to the November 2003 Assembly election. The UMS- (now going by the name Millward Brown)/ Belfast Telegraph poll found the following levels of party support only weeks prior to the election:
UUP 26%
DUP 20%
SF 20%
SDLP 22%
Alliance 6%

And, once again, the poll proved to be significantly wide of the mark when the results came in:
DUP 25.6%
UUP 22.7%
SF 23.5%
SDLP 17%
Alliance 3.7%

This time, all of the four main parties came in outside of the traditional 3%margin of error for polls, with the same trend as in 2001: namely, Sinn Fein and the DUP massively under-represented and the UUP and SDLP over-represented.

Step forward in time to March 2005, when Millward Brown/ Belfast Telegraph joined forces once again to publish a pre-election poll with the following figures for each party:
DUP 28%
SDLP 20%
Sinn Fein 20%
UUP 16%
Alliance 5%

Of course, when the votes were counted, the following results came in, once again bearing little resemblance to the MB/Belfast Telegraph poll:
DUP 33/7%
Sinn Fein 24.3%
UUP 17.7%
SDLP 17.5%
Alliance 3.9%

As you can see, the pattern of under-estimating the Sinn Fein and DUP vote was once again confirmed, as was that of over-estimating the SDLP/ UUP vote.

Finally, to the 2007 Assembly election predictions, published in the Belfast Telegraph, this time by MORI/ IPSOS:
DUP 25%
SF 22%
SDLP 20%
UUP 16%
Alliance 9%

The results in the past week indicate yet another abysmal failure on behalf of the pollsters:
DUP 30.1%
SF 26.2%
SDLP 15.2%
UUP 14.9%
Alliance 5.2%

Once again, Sinn Fein and the DUP reached levels of support considerably beyond those predicted by the Belfast Telegraph poll, whilst Alliance and the SDLP in particular came in well below the predicted figures.

  • It couldn’t just be the simple fact that the pollsters are actually going out and asking people which way they’d vote (passive) as distinct from election day when voters actually have to go and make the effort to vote (active)?

    Either that or people are embarrassed to admit to voting for the nutjobs in DUP/SF.

  • slug

    The poll was not bad for the non-nationalist vote; it got it right in fact, by pointing to a pick-up for Alliance and a slump for the UUP.

  • mickhall

    Chris,

    You may be right, who knows, but I am not sure why you have raised the issue in the manner you have, for if the polls had been fixed, surly the main beneficiaries would be the DUP and SF. For it would in all probability get those people who normally vote for these two parties, but had doubts this time around, to put to one side their apathy or disillusionment with the party they normally support at the polls, and get along to the polling station to caste their vote.

    It has become clear for some time the UK State has been placing their hopes [In this election] on the DUP and SF, one only has to read the headlines in the London papers after the results to understand this. I am not suggesting there is anything particularly underhanded about this, as government always prefer to deal with a solid block rather than a group of smaller parties who act like eels in a sack.

  • SuperSoupy

    As a pollster for the Belfast Telegraph admitted they were instructed to only poll Greater Belfast and Derry ignoring the rural West and other areas that return well for SF – I assume it is a mixture of the sinister and incompetence. Sinister from those making the request, incompetence by the organisation complying.

  • Cock up rather than conspiracy here, Chris. People are more reluctant to tell pollsters that they are voting for Sinn Féin or the DUP than other parties. It’s the old spiral of silence scenario. Could be mitigated by all sorts of means, and if any NI polling company wants to retain me as a consultant, I’m available…

    And anyway, you can spot the general trend. When I saw a narrow SF lead over the SDLP in polling, I thought that meant a hefty SF lead over the SDLP in ballots cast and I suspect so did you and most others.

    As I said when the poll came out, based on previous polling inaccuracies:

    In a totally unscientific and deprecable way, I will apply those modifiers to today’s poll to get:

    DUP 31%, SF 26%, SDLP 15%, UUP 13%, Alliance 7%.

    Not so far off, eh?

  • SuperSoupy

    Sammy,

    You were shocked at the Belfast Telegraph pollster’s claims at the time. Y

    Polling in those restricted areas was bound to produce an imbalanced result. The Belfast Telegraph knew it – asking was sinister. The polling organisation complying was incompetent.

  • Polling in those restricted areas was bound to produce an imbalanced result.

    If they did and our poster didn’t just have his facts wrong… I didn’t get around to checking with Mori last week (like all of us, kinda busy) but I’ll try and chase this week.

  • GavBelfast

    Or maybe don’t want to admit that they will vote for / have voted for SF and (to a lesser extent) the DUP?

  • graduate

    Is it just me or do people always look slightly guilty and shifty when going into/coming out of polling stations? Certainly last week I got the impression from some voters that they were doing somenthing like sneaking into Anne Summers rather than doing their civc duty- slightly dodgy and furtive. Maybe that’s why the pollsters are so far off, people just ain’t telling truth, especially in leafy suburbs ’cause they dion’t want the neighbours to think they vote for anybody other than alliance/uup/sdlp

  • SuperSoupy

    If the contribution we had to Slugger’s was correct and Sammy is going to check, the reason this poll was so off was the Belfast Telegraph asked MORI not to poll in areas where the Sinn Féin vote is strong, MORI complied and made no reference to this anywhere.

    Both sinister and incompetent if true.

  • Pat Mc Larnon

    ‘People are more reluctant to tell pollsters that they are voting for Sinn Féin or the DUP than other parties.’

    Is there any evidence to support this theory?

  • wild turkey

    Sinister? If the sponsoring organisations, the Belfast Telegraphy or whoever, is telling pollsters where to sample; ie include and/or exclude specific areas, and then goes on to report the results as ‘representative’, yes it is deliberately misleading, more propaganda than fact and hence sinister.

    anyway, with respect to the Belfast Tele I defer to the wit and wisdom of lou reed, don’t believe half of what you see and none of what you’re told

    Incompetence? If the pollsters go along with this crap and allow the results to published in such a way as to mislead the public into believing the results are representative, then the pollsters are incompetent. and possibly greedy.

    below is extract from wikpedia on random sampling.
    remember kids if it ain’t a random sample, then there are severe limitations on the extent to which the sample can said to representative of the wider population
    A sample is a subset chosen from a population for investigation. A random sample is one chosen by a method involving an unpredictable component. Random sampling can also refer to taking a number of independent observations from the same probability distribution, without involving any real population. A probability sample is one in which each item has a known probability of being in the sample.
    The sample will usually not be completely representative of the population from which it was drawn— this random variation in the results is known as sampling error. In the case of random samples, mathematical theory is available to assess the sampling error. Thus, estimates obtained from random samples can be accompanied by measures of the uncertainty associated with the estimate. This can take the form of a standard error, or if the sample is large enough for the central limit theorem to take effect, confidence intervals may be calculated.

  • SuperSoupy

    http://sluggerotoole.com/index.php/weblog/comments/dup-get-mori-poll-boost/P75/

    as a regular Slugger reader, and someone who actually works for IPSOS-MORI and carried out this poll over the last 2 weeks, perhaps i can answer a few questions raised, if anyone has anymore then fire away…
    please bear in mind that all guidelines and questions and quotas were supplied to us by the Belfast Telegraph themselves.

    kensei,
    “This poll only counts those who have stated they intend to vote.”

    incorrect, the views of everyone interviewed, whether they intended to vote or not, were counted.

    SuperSoupy,

    “methodolgy? questions? questioners? sample size? etc…

    Firstly I’d like to know the name of the polling organisation standing over these claims.”

    1. Phone interviews with various population demographs supplied by the BT, eg. X amount of males 18-35, X amount of females who work part-time, X amount of males whose home is rented from the NI Housing Executive, etc.

    2.
    – Personal demographics (age, working status, home ownership status, religion)
    – What issues interviewees felt most important from a pre-defined list, as well as the option to tell us more important issues which weren’t on this list.
    -How interviewees felt towards various NI/British/Irish politicians on a scale of 1-5, in terms of favourability/unfavourability.
    -What party they were likely to vote for.
    -Their reasons for voting for said party.
    -The likelihood of March leading to a working Executive/Assembly.
    -If Assembly not up and running by March, whether Assembly members should be paid indefinitely/limited time longer to reach a deal/stop paid imeediately.
    -Likelihood of Paisely/McGuiness being able to work together.
    -Whether policing should be responsibilty of British Government or local politicians.
    -Likelihood of voting in election.

    3. Telephone interviewers from call centre in Dublin, who were told to use our Belfast office address and phone number if company details were asked, aswell as being told to say that were calling from a company based in belfast, on behalf of the Belfast Telegraph.

    4. i can’t remember exactly, approx. 1200 if my memory serves me, definitely 1000+ anyway…

    you should have bothered reading the article to find out the company name.

    also, the polling was done in random households in 3 quotas of:
    Belfast City
    Greater Belfast Area
    Derry

    Posted by James on Mar 02, 2007 @ 03:05 AM

    If the above is true. All those that carried the poll would be guilty of incompetence. That includes for starts BBC, UTV, RTE and the Irish News amongst many others.

  • Is there any evidence to support this theory?

    Only that every opinion poll ever carried out in NI has grossly understated the SF vote, and the DUP tend to fair badly. Ours is almost always overstated, except for when we flirted with extinction a few years back.

    If the above is true. All those that carried the poll would be guilty of incompetence. That includes for starts BBC, UTV, RTE and the Irish News amongst many others.

    Yup. That’s why I want to be sure before I get the boot in. It’s actually a bit of a Slugger/blogosphere coup if that poll turns out to be as methodologically shite as indicated.

    Right, I’m away into the garden, it’s 16 degrees here and far too nice to be on the computer.

  • Aaron McDaid

    If there’s any conspiracy, it’s to the benefit of SF and the DUP. It helps to get their vote out if their voters think they aren’t going to do very well.

  • Pat Mc Larnon

    ‘Only that every opinion poll ever carried out in NI has grossly understated the SF vote, and the DUP tend to fair badly. Ours is almost always overstated, except for when we flirted with extinction a few years back.’

    Well that is no then. You were quite definite in your assertion that voters were reluctant to give their opinion to pollsters.

  • heck

    I am not a statistician but if the pollsters are not getting reliable answers and this is a consistent trend couldn’t it be corrected for in the data reduction?

    I have always been suspicious about the way the Belfast Bellylaugh trumpets polls and I tend to put in down to dishonest journalism rather than dishonest pollsters-but I don’t know. Maybe some Bellylaugh staffer can post the answer.

  • GavBelfast

    Does there have to be a murky, conspiracy theory behind everything?

    It is entirely possible that people prefer not to admit they have voted for the extremes, and either lie about who they voted for/are going to vote for, or more SF/DUP voters refuse to say.

    Why does there have to be evidence for that? It’s just a suggestion that COULD be plausible.

    Everything’s such a big deal here.

  • Roisin

    [i]Only that every opinion poll ever carried out in NI has grossly understated the SF vote, and the DUP tend to fair badly.[/i]

    A random thought on the SF opinion poll results: Perhaps the old ‘whatever you say, say nothing when you talk about you know what’ still holds for a significant number in the nationalist community. One would expect that should have faded with the new political landscape, but old habits may die hard. Or SF and DUP voters may be more inclined to take the piss out of pollsters. Or the pollsters are simply incompetent. Whatever the reason, it’s a problem for the credibility of the polling agencies and the sponsoring journals.

  • SuperSoupy

    Gav,

    Both I and Sammy find one element of the suugested methodology highly questionable, if true – the sample locations.

    Concentrating on “Belfast City, Greater Belfast Area and Derry” would lead to a clearly imbalanced result.

    A paper insisting on this and a polling organisation not stating it happened with it’s data return are, if true, very suspect.

    That’s why Sammy is seeking confirmation.

    If the Belfast Telegraph did this and others covered it, this could be a deliberately biased and completely illegitimate poll that deceived many other media outlets who would then be utterly incompetent by carrying it without examining the methodology.

  • GavBelfast

    SuperSoupy, the Telegraph is an amaterish set-up in many ways anyway – sure their front page feature the other day was based upon asking 100 people who left one polling station in Strangford what they had done, and that (to them) told the tale of the whole election.

    It reminded me of Les Dennis on Family Fortunes.

    I know if I was asked my opinion in one of these polls, I would take the piss – it’s none of their business and does little other than fill the pollsters coffers.

  • Roisin

    How much of an imbalance? It would account for the SF/SDLP imbalance and overinflated APNI percentage, but what of the DUP/UUP?

  • John Lowe

    If the polls are consistently getting things wrong there are a number of possible explanations.

    If could be down to people not telling the truth to pollsters. This certainly happened in England for the Westminster election in 1992.

    It could be indicative of systematic electoral fraud although I assume that security measures will prevent this.

    A more plausible explanation is differential turnout. That is SF and the DUP are much better at getting their vote out.

    Finally it could be that the pollsters methodology is faulty. They should be adjusting the outcome to allow for factors such as differential turnout and less than truthful responses. It’s a matter of fine-tuning the opinion poll results to fit the actual results.

  • SuperSoupy

    John,

    Sampling a spread of locations that are more representative would be the easiest correction.

    Asking for locations that clearly aren’t representative is always going to give incorrect results.

  • A random thought on the SF opinion poll results: Perhaps the old ‘whatever you say, say nothing when you talk about you know what’ still holds for a significant number in the nationalist community.

    I certainly think so, and I don’t think it’s restricted to the nationalist community.

    It would account for the SF/SDLP imbalance and overinflated APNI percentage, but what of the DUP/UUP?

    There’s been a parallel inflation of the UUP vote and understatement of the DUP vote in most polls. Certainly in the one in the Tele a few weeks ago.

    I disagree that it’s conspiracy; most journalists don’t really understand polling, and certainly not Northern Ireland ones. The NIO aren’t the sharpest tools in that particular box either. I’m a firm believer in Occam’s Razor – the simplest possible explanation is almost certainly the right one, and “people talking crap because they don’t know what they’re talking about” is the simplest one here.

    If it were a grand conspiracy against Sinn Féin, it would hardly be a very clever one given that it allows Shinners to have a good laugh at how wrong the polls are after every election.

  • John Farrell

    I hate to say I told you so…but…I told you so.

    If an opinion poll asks how many people brush their teeth more than once a day, you will get a 100% response that we all do. But dental nurses would say different.
    Likewise DUP and SF support is understated. That much is a generally known fact.
    The AP on 9% is just plain laughable and when this was pointed out by many the standard response from the pretty straight middle of the road guys was to talk AP support down to around 7% and the actual result is 5.2%

    But what do the Alliance Part actually know about the West. Well not much.
    West Tyrone……well obviously they dont even know where that is cos they couldnt find the electoral office
    Mid Ulster….Margaret Marshall (221 votes) is from Carryduff.
    Fermanagh South Tyrone…Alan Leonard (521 votes) is from Newtownards
    Newry Armagh …Marie Hendron (278 votes) is from East Belfast
    The only local candidate was Foyles Yvonne Boyle (224 votes) is from Portstewart, in East Derry.

    Well of course being a blow in is no bar to knowing a place but isnt the inability to actually find a local candidate at least a clue to the lack of local support. Yet the middle of the road types saw only the stats that MORI presented in the Belfast Telegraph.
    There are lies
    There are damned lies
    There are statistics
    And there are Alliance Party opinion poll results.

    Yet the “experts” on this site relied only on the figures and exposed themselves as not understanding them and worse having no feel for the political mood music in Norn Iron beyond the statistics that suited their Alliance world view.
    It was well known that the poll (which coincidently suited Belfast Telegraph editorial stance) was about Belfast and Derry and yet the sites “experts” bemoan that their expertise was undermined by the methodology of the poll.
    You gotta see the funny side.
    MORI polls eh
    M =Money for
    O=Old
    R =Rope from
    I =Idiots in Telegraph

    Conspiracy or Stupidity?
    Well, The Telegraph likes this kind of result. It suited them.
    MORI wants the Telegraph contract.
    Alliance Party types on this site got a buzz out of the results.
    And the rest of us got to have a laugh at Telegraph, MORI, AP and their supporters expense. No fatalaties.

  • Gonzo

    It may not be representative, but I know one person who was phone-polled in south Down.

    Hardly an urban centre.

  • Benn

    It ain’t over ’til it’s over, and maybe it ain’t over even then. If I can remember back that far, Florida had been showing Gore was ahead, then it was called for him based on exit polling, then it got too close to call, then the Republicans sent James Baker down there to steal it and the Supreme Court sealed the deal by stopping the recount. When the ballots were actually counted and recounted with integrity it was Gore by some margin, but by then it was too late to “count.” Since that time television networks have been more cautious on election day, but the whole US media has done their best to aggregate scant evidence in order to influence the masses with “observations” of trends prior to election day. If the Telegraph stacks the deck by concentrating the polling area but does not provide that information in a prominent way it’s fraud, no more and no less.

    Beyond that, people often lie in the poll and tell the truth in the ballot box, and sometimes the other way around. Call me a sinner, but when a pollster asks me anything I feel no compunction about lying to advance my cause since it’s all about spin. My goal is just to figure out which way to lie, and then hope for the best. Polling is an art and a game, dressed up by those who profit from portraying it as a science. Woe to those (the public) that buy in to such shite. Ben

  • Henry94

    Benn

    Fir enough in so far as it goes but the polls in the north are consistently less reliable than the polls in the south and in Britain. The methodology must be similar.

    So who is skewing who and why?

  • Cahal

    People should remember that as recently as 15 years ago, being a known SF supporter was enough to make you a target of loyalist death squads; a group which is still armed to the teeth.

    Those aiming for the normalization of politics here should maybe concentrate on trying to disarm loyalist militias/murderers.

    Although I get the feeling this isn’t top of any unionist politicians list of priorities.

  • John Farrell

    I love the way that the “experts”who actually wanted to believe these polls are now using the flawed methodology of the polls as a fig leaf to cover their own inability to read what was really going on.

  • John Lowe

    Wild Turkey (12),

    Opinion polls never use random sampling. It is far too expensive. You would have to select electors at random (perhaps using their electoral roll numbers). If one of the sampled electors was not in when the pollster called or telephoned they would have to follow up again and again until all the original sample had been contacted. To ignore them and choose someone else would bias the sample (against shift workers etc).

    Opinion polls use a technique to try and replicate the electorate as a whole by selecting a (non random) sample to give the correct proportion by religion, class, occupational group, age, and location.

    If they get the sample wrong in any of the above then it should be no surprise that the predicted outcome is also wrong.

  • blogrus

    People are embarrassed at their own bigotry. That is the key reason why polls are wrong. – maybe they should be asking questions such as:
    “What way do you think your nationalist / unionist friends/neighbours will vote?” At least that way, people may be able to hide behind their sectarionism and bigotry by passing it on to to others.

    What a society we live in. Bigots rule the day, and they are recruiting more and more supporters who are too embarrassed to admit it, but in the secrecy of a ballot paper, show their true sectarian colours.

    No Protestants (except for the odd one or two) would vote for Sinn Fein and no Catholic would vote DUP – ergo, two sectarian giants within NI politics. Is that what we really want?

    People out there, please tell the truth. And, pleas try to avoid the risk that having a sectarian political assembly will only serve to polarise our people and run the risk of a future conflagration which could be worse than the one we have just left.

    I know it is too late for this election, but let’s not be cowards. Let’s stand up and be counted.

    Blogrus

  • Aaron McDaid

    Blogrus, you said “No Protestants (except for the odd one or two) would vote for Sinn Fein and no Catholic would vote DUP

    There are Protestants in SF, not just voting for them, including at least one of their candidates on Wednesday.

    I understand the inhabitants of Rathlin are mostly Catholic and vote DUP because of Paisley’s excellent constituency work such as getting them electricity.

    You say Protestants shouldn’t/won’t vote SF because they are sectarian, and then say SF are sectarian simply because most of their voters are Catholic. But that’s a circular argument!

  • Sean Fear

    Political polling only produces a fraction of pollsters’ income (c.2-5%). Mostly, they do market research. But their political polling is the one that is most measurable against actual outcomes. They have a very strong vested interest in getting it right, and no pollster is going to rig a poll in the interests of the newspaper or government department that commissioned it.

    Pollsters make mistakes, and are very interested to find out why they get it wrong. I imagine that in Northern Ireland there is a very strong reluctance in many hard line Republican and Loyalist areas either to speak to pollster, or to be honest with them.

    I expect MORI only sampled in Greater Belfast and Foyle because these are the areas where it’s easiest to get people to talk to you. What I suspect they did (although it would be interesting to hear from them) is to extropolate shifts in voting intention in these areas to the whole Province. Hence, their poll picked up a very big shift from Unionist to Alliance, and non-aligned. That shift did occur in the Greater Belfast area, but not elsewhere.

    What would be interesting is if internet pollsters, like Yougov were commissioned to survey Northern Ireland. People are much less shy with internet pollsters, and you get a much higher response rate, and Yougov might be much more accurate in recording SF and DUP figures.

  • PaddyReilly

    You often find with religions that the more demanding ones stay afloat and the liberal ones go under. You need a little bit of eternal damnation to get the congregation to turn up. Politics may be similar: perhaps DUP and SF voters are better at keeping their promise to turn up and vote, and SDLP UUP and Alliance ones more likely not to bother. Note that there was a surge of voters for the GFA, and this surge was in favour of a peaceful settlement.

    And maybe less DUP and Sinn Féin voters have telephones.

    Whatever the case, the MORI pollsters did quite well, given that they confined themselves to Belfast and Derry. The Green Party did win a seat: Alliance did increase their representation, and the Unionist/Nationalist proportions were just about correct.

    However there is one reliable predictor which I have yet to see proved wrong: the observation that the Unionist share of the vote goes down by 5% per decade, or 1% every two years.

    I calculated that the Unionist vote would be at 50.6% in March 2007, and in fact the Unionist/Loyalist block did win 50% of the seats plus one more, exactly right. However the (first preference) Unionist vote was even lower: about 47% for the four major parties, 49% if you include the various ‘broken biscuits.’ I take it that this was caused by the rise of the Green Party, and that Green Party voters reverted to Unionism in their second etc choices.

    So the prediction for the future for the (1st preference) Unionist vote is as follows. First the higher figures, which now signify the sum total of the Unionist vote plus transfers from Green Party voters, and thus the likely percentage of Unionist seats in any Assembly :-

    2007 50.5
    2008 50
    2009 49.5
    2010 49
    2011 48.5
    2012 48
    2013 47.5
    2014 47
    2015 46.5
    2016 46

    Next, the revised figures for the sum total of 1st prefs going to any candidate who could be considered Unionist:

    2007 49
    2008 48.5
    2009 48
    2010 47.5
    2011 47
    2012 46.5
    2013 46
    2014 45.5
    2015 45
    2016 44.5

    Finally, the expected vote for the Big Four Unionist Parties: DUP, UUP, PUP, UKUP.

    2007 47
    2008 46.5
    2009 46
    2010 45.5
    2011 45
    2012 44.5
    2013 44
    2014 43.5
    2015 43
    2016 42.2

    Presumably transfers from Alliance etc (currently around 4%) will keep the actual vote in favour of the status quo (i.e. partition and the Union) above the 50% mark for much of this time, but by the end of this period this will have become untenable.

  • Benn

    That’s a very noble vision of polling Sean. Another way of viewing it is that pollsters have political customers and commercial customers, and the two are not the same. Sure, they’d like to get it right, but what is right? The party commissioning the poll wants a product, provides specifications, and sets terms. In the ignoble world of politics, information is power, and sometimes you get what you pay for. The fact that we’re all still puzzling about this suggests that either the outfit doing the polling is doing a poor job of it and doesn’t want to confirm that fact, or that they’re doing a good job of it the people commissioning the poll don’t want the terms of the contract known. Either way, consistent under-reporting shows a pattern that should be compensated for in the methodology, or maybe that’s the point. Hard to know which it is, but after a while it stinks, no? Ben

  • Sean Fear

    There’s nothing noble about it. If your polling is demonstrably wrong, you’ll find that a declining number of people want you to carry out market research. Sacrificing commercial clients, in order to satisfy a political party or newspaper would be a very stupid thing for a pollster to do.

    For about ten years from 1992, opinion pollsters (with the exception of ICM, and at the end Yougov) consistently underrrecorded support for the Conservative Party. This wasn’t down to any deliberate bias on the part of pollsters, but was down to the fact that some Conservatives were unwilling to admit their allegiance, and for a number of reasons, pollsters now have a far higher response rate among those sections of the population that tend to vote Labour than among those that tend to vote Conservative.

    Eventually, they devised techniques to overcome this problem.

    You’ll find a great many of the relevant arguments on Anthony Wells’ excellent website, UK Polling Report.

  • Gonzo

    The poll wasn’t confined to Greater Belfast or Derry.