Virtual Politics 1: Hearts and Minds Poll

The last opinion poll was carried out by Hearts and Minds in October 2006, its findings on party support are applied to the possible March Assembly election. Full poll details here (pdf file).The party scores and projected number of seats were:
UUP 17.7% – 19 seats (down 5)
SDLP 16.8% – 18 seats (no change)
DUP 30.6% – 33 seats (no change)
SF 20.1% – 21 seats (down 3)
Alliance 3.5% – 4 seats (down 2)
PUP 1.6% – 2 seats (up 1)
UKUP 1.8% – 2 seats (up 1)
NIUP 1.4% – 2 seats (up 2)
RSF 2.4% – 3 seats (up 3)
32 CSM 0.6% – 0 seats
Cons 0.5% – 0 seats
Greens 2.1% – 2 seats (up 2)
2 remaining seats could not be determined but would most likely go to the largest two parties, the DUP and SF.

Apart from a drop in support for Sinn Fein and Alliance and growth for the smaller parties, the Hearts and Minds poll closely mirrors the party support levels in the 2005 local government elections, reinforcing the case that it should be the indicator of choice for the upcoming elections. The drop in SF’s vote seems to be going to RSF and 32 CSM and the drop in Alliance seems to be going to other small parties in particular the Greens. 30.6% is a drop from the DUP’s score of 33.7% in the Westminster election but this 3% can be found in the UKUP and NIUP scores. If a fair reflection, it would indicate that the large shifts in voting patterns have occurred and we will see little movement in coming elections.

The combined UKUP and NIUP score is interesting. They are essentially the same political platform and the poll has them in the race for 4. This would be comparable with the the UKUP’s 1998 performance and we now that this vote is concentrated in east Ulster. However, the figures indicate their gains would be at the UUP’s expense not the DUP’s but they would act as a barrier to DUP growth.

In the past two Assembly elections, a party scoring more than 0.7% of a vote have gained a seat (with the exception of the UDP in 1998) so even at such levels a party can be competitive. A barrier to the smallest parties is they need a concentration of their vote to be in contention for seats, such support levels spread evenly across 18 constituencies will not deliver seats.

Important notes on this and other polls:
1. The similarity between the names of Sinn Fein and Republican Sinn Fein could have caused respondent confusion so overestimating RSF support. I am not aware of any other opinion poll giving RSF such a rating before so it seems to be an unusual outcome. However, there is a some correspondence between RSF and 32 CSM combined ratings and the % of SF voters implacably opposed on policing.
2. Any shifts in party support from previous elections are pretty much within the error of margin in the poll.
3. Seats were allocated on a round-up/round down basis with the exception of parties scoring less that 0.7% (32CSM and Conservatives). No party with a score less than 0.7% has gained a seat in the NI Assembly.
4. The lack of opinion polls since SF and DUP became the dominant parties in their respective communities mean it is difficult to assess whether or not their historical under-polling continues or not.

Poll Details
1. 1,100 persons in total responded to the survey.
2. Interviews took place across all 18 Constituencies and 26 District Council Areas of Northern Ireland (43 sampling points in total).
3. Fully representative results of Northern Ireland adult population in terms of age, gender, religion, geography and social class.
4. Sex breakdown – Male (49%), Female (51%).
5. Age breakdown – 18-24 (19.7%), 25-34 (20.2%), 35-44 (20.1%), 45-59 (20.7%), 60+ (19.1%).
6. Socio-economic groupings – ABC1 (38.8%), C2DE (60.7%), refused (0.5%).
7. Religion breakdown – Protestant (56.2%), Catholic (43.8%).
8. Polling took place during the 27th – 30th October 2006.
9. Margin of Error ± 2.95 %.

  • slug

    I think you will have to bear in mind that this is a poll of 1100 people so anything under 5% is not statistically reliable at all – the sample is just too small. The figures for the larger parties (the big four) are more reliable (with a margin of error of a few %).

  • Ulick

    “I think you will have to bear in mind that this is a poll of 1100 people so anything under 5% is not statistically reliable at all – the sample is just too small.”

    Not true. So long as the sample is random, 1100 is more than adequate. In fact according to Krejcie and Morgan (1970) “as the population increases the sample size increases at diminishing rate and remains constant at slightly more than 380 cases”. This view is supported by most eminent statisticians including Borg and Gall (1979), Oppenheim (1992) and Cohen, Manion and Morrison (2005) who sate that a sample of 384 is sufficient for a population of 1,000,000 to provide the possibility of 5% sampling error of 5% with a confidence level of 95%. So it would seem that the 2.95% margin of error using a sample of 1100 is about right.

  • slug


    Yes I know that. What I mean is that the figures for the political parties with less than 5% of the vote aren’t reliable, as they are statistically insignificntly different from zero. Some of them could easily double or triple within the standard of error.

  • The Devil


    Not quite so, the stats you are quoting are based on people telling the truth in another wise normal society, the quota is far too small for an accurate poll of any society emerging from civil unrest, armed conflict or sectarianism.
    Just far too many people avoid the truth at these street polls as has been proved over the years but the downplaying of the S/F PUP and DUP vote, as was the case with previously the IIP or Republican clubs

  • seabhac siulach

    “The similarity between the names of Sinn Fein and Republican Sinn Fein could have caused respondent confusion so overestimating RSF support.”

    This is giving the respondents very little credit…we might equally say that the similarity between the names DUP and UUP caused similar confusion. As others have commented, the results of the smaller parties are essentially in the ‘noise’ and cannot be trusted, statistically. The apparent increase in the smaller republican parties is, however, interesting seeing as they get practically zero media coverage (and none that is positive).

    While your numbers show RSF picking up 3 seats the truth is that they cannot stand in these elections (even by running on an abstentionist ticket), as they rule themselves out by the political test oath:
    “I declare that, if elected, I will not by word or deed express support for or approval of —
    (a) any organisation that is for the time being a proscribed organisation specified in Schedule 2 to the Northern Ireland (Emergency Provisions) Act 1978: or
    (b) acts of terrorism (that is to say, violence for political ends) connected with the affairs of Northern Ireland.”
    · The text of the oath from the published British “Elected Authorities (Northern Ireland) Act”.

    They refuse to take this oath, believing political violence to be justified (unlike,ahem, the Provos) so the three Provo seats are safe…unless the percentage of RSF votes were to transfer en-masse to ‘dissident’ anti-Provisional Sinn Fein (PSF) republicans, who may or may not stand and certainly not in every constituency. It is more likely that the percentage of abstentions, in the event of an election, will just increase. PSF can probably win 20+ seats with a reduced vote in this case. It is also likely that many potential RSF voters will just vote for PSF, holding their noses, as the least worst option on many occasions…

  • Frustrated Democrat

    There has been a lot of water under the bridge since then….powersharing, policing etc… it would be interesting to see another poll soon.

    I think the UKUP would be higher and the DUP lower and soon the PUP and other small unionist parties will disappear, leaving 3 or 4 seats up for grabs.

    As everyone knows a split party loses votes and the DUP and SF are split so who knows what will happen before elction day as HW said ‘a week…….

  • Irish Aussie

    There is nothing wrong with the sample size but I believe there is a problem with the sample its self
    The poll states that the sample was made up of 584 unionists 53.1%, 439 nationalist 39.9% and unaligned 66 6.0%.
    This in my opinion slightly over states the unionist vote and under states the nationalist vote
    The net effect of this is that the total unionist vote Dup,Uup,Pup etc comes in at 53.1% which means the poll is suggesting that the total vote of the Alliance party the Green party the Tories etc is comeing exclusively from the nationalist and unaligned groups.
    A pretty unlikely senario in my opinion a much more likely senario is that the unionist vote is somewhat overstated.
    As someone who did this sort of work for nearly a decade I find it “interesting” that you would go to the trouble of identifing which groups the sample came from but then not weighting them accurately

  • Greenflag

    The figures don’t matter anyway . Why would they ? Election to a powerless Assembly will just continue the farce of a failed political and economic entity . Northern Ireland’s Nationalists and Republicans need to look at a fair Repartition of Northern Ireland by a neutral international organisation as the solution.

    Leave Paisley and his ilk to a 2 county or so sized Unionist theocracy and be done with it . We don’t need another 40 years of politcal mayhem to prove that Unionists are genuinely opposed to a UI?

  • fair_deal

    Irish Aussie

    “I find it “interesting” that you would go to the trouble of identifing which groups the sample came from but then not weighting them accurately”

    1. I did not conduct nor was I involved in the management of the poll so save the conspiracy theories. In previous threads when polls have been mentioned people usually ask about the sample hence the inclusion of sample info.
    2. The poll says it got a representative sample of the population based on religious background, socio-economic standing and geography.
    3. If as you say you were ‘involved’ in this stuff you would know you can’t get an accurate picture of party political support if you predetermine the levels of support.
    3. The differences you highlight are within the margin of error of the poll. The SF/SDLP vote in 2005 local government was 40.6% and 2003 Assembly 40.5%. The average SDLP/SF % over the past four elections is 41.3%. The error of margin in the poll is 2.95%.

  • Rubicon

    Ulick – I think you are placing too much confidence in the survey (The Devil’s point aside). A crucial factor is the distribution within the 18 constituencies. Do you know if the survey was stratified by Parliamentary Constituency? If not, you can put little reliance on translating the results in to seats.

    Take for example the returns from the last Assembly election where SF and the UUP (allowing for the Donaldson & co. defection) ended out with exactly the same number of seats but with the UUP have far fewer 1st preference votes.

    You’d be well advised not to read too much (seats wise) from surveys like this. Where they an be useful is tracking support shifts over time (particularly for the big 4).

  • Benn

    For what it’s worth, there’s a newer poll, November 2006. As for accuracy, these things can be a guide, but they’re never “THE TRUTH.” When a pollster gets ahold of me I spin the damn thing as best I can. Sure, only one voice, but I imagine there are others like me who want to manipulate some, just as we are being manipulated. Some of my answers are on the mark, others are intended to sway the results. Is that wrong of me, or just politics? Cheers, Ben

  • Valenciano

    The NIUP on 2 seats?! Do they even exist any more? Their website has been inactive for some time and their party spokesmen seem to have retired. With not even the one man band of a McCartney to keep them afloat I’d be amazed if we ever hear of them again.