General Election 2015: Northern Ireland forecast
Just to get the build-up going to the upcoming Westminster election on 7th May we'll be running our Opinion Panel polling every month from February to May. We will be feeding this data into our prediction models, and then projecting the results for each of Northern Irelands 18 Westminster constituencies, and tracking any trends up and down.
This will be similar to the prediction models we ran for last May's European election when three weeks before that election (in the Belfast Telegraph) we e.g. predicted Sinn Fein for a 26.2% vote share with the actual result coming in at 25.5%; and the DUP prediction was 20.8% with the actual result turning out to be 20.9%. So (in our humble opinion!) our Euro election prediction models weren't that bad!
In terms of our Westminster election prediction model this is how it works:
On a monthly basis, from now until election day on May 7th, we will be polling the LucidTalk Northern Ireland (NI) Opinion Panel which is made-up of 440 members and is carefully constructed to provide an accurate representation of Northern Ireland opinion - via gender, area of residence, age-group, community background, socio-economic group, and employment group. The opinion panel has a pool of approximately 1,200 hundred members who regularly take part in LucidTalk poll projects, and for this project a representative sample of 440 opinions will be collected on a monthly basis, up until the 7th May election day.
Our Opinion Panel members were asked, and will be asked:
(a) Likelihood of voting in May 2015, and
(b) What party they currently plan to vote for, and
(c) Their current opinion regarding who would win in each of the 'non-100%' certainty seats (currently nine - see table).
Results will be presented as a % likelihood of the named political parties winning each of the 18 Westminster Parliamentary seats. Our forecasts will not predict vote share, size of vote, and/or size of majority etc., as to do this would require a much higher sample and more comprehensive poll. In this context, it should be noted that if the % scores go up or down for the main prediction, then it's the alternative party (right-hand column in the table) that is retreating or advancing by that same score.
Just to provide a comparison of trends and changes moving forward, we polled our Opinion Panel in January, and we show any changes from January in our Opinion Panel results table. Similarly we will be able to track any changes and trends on a month by month basis until election day in May.
As far as our predictions are concerned, it should always be remembered that the Westminster election is a winner take all, first-past-the-post, type election. For example, unionists keep reminding us that in 2010 Rodney Connor, the agreed unionist candidate for Fermanagh and South Tyrone lost to Sinn Fein's Michelle Gildernew by only four votes. However if it had been 400, or 4,000 votes, it would still have been the same result - Michelle Gildernew MP. Taking this into account, when looking at our forecast it is probably fair to say that the 'Alternative Parties' need to get their opponents (i.e. our main prediction) down to a 65-60% (or less) probability score, in order for these 'Alternative Parties' to be in with a realistic chance of winning on May 7th.
We can see from our Forecast table that out of the eighteen Westminster seats there are only nine that are 'in play' i.e. have wining probabilities for a particular party at less than 100%. Even then, there are only two of these that currently have winning probabilities of less than 75% i.e. East Belfast and South Belfast.
So as you can see from January the SDLP have slipped down a bit in South Belfast going from 65% probability of holding that seat to 60%. Perhaps this reflects how the sitting South Belfast MP Alasdair McDonnell expressed and explained his views, and his parties views, regarding abortion? However he still remains a 60% favourite to hold his seat. But in terms of our models, 60% isn't a certainty.
Likewise the DUP have gone down 5% in North Belfast but still remain firm favourites to hold this seat with an 85% probability.
Interestingly one of our 100% seats in our January Opinion Panel poll has changed - Newry and Armagh. Here Sinn Fein were on a certainty to hold this seat but members of our Opinion Panel from that area, and others, have noted that the SDLP candidate ex-GAA star Justin McNulty may put in a strong performance. This may also perhaps be assisted by some Unionists who may chose to vote tactically for McNulty in a seat that Unionists have no hope of winning. Interestingly, if the 'Alternative Vote' system had passed the 2011 referendum then almost certainly Sinn Fein would lose this seat to the SDLP, as the combined SDLP/UUP/DUP vote is substantially more than the Sinn Fein vote. However, with the first-past-the-post system it would require a substantial part of the Unionist vote in this constituency to switch on this tactical basis, and vote for McNulty, to oust Sinn Fein. This is unlikely, and as such the new Sinn Fein candidate Mickey Brady is still on a high 90% probability score, saying that Sinn Fein are still strong favourites to hold this seat formerly held by Conor Murphy.
In a similar situation to Newry and Armagh, our Opinion Panel has shown that the UUP candidate in Upper Bann Jo-Anne Dobson seems to making an impact causing the DUP score for holding this seat to drop 5% from 80% to 75%. However, as with all these analyses, it should be noted that 75% probability is still a strong score for the incumbent DUP MP David Simpson. As such, he currently remains a firm favourite to hold this seat on May 7th.
Then we have the seat that everyone is talking about - well the politicos anyway - East Belfast. This is a constituency were we carried out a full constituency poll in January with the results published in the Belfast Telegraph a few weeks back. This showed that the DUP were narrowly ahead in terms of re-taking this seat from the Alliance party's Naomi Long. Our Opinion Panel has re-confirmed this with the DUP on 55% probability of retaking East Belfast. However, this is obviously a very narrow lead for the DUP, and can easily be overtaken by the Alliance Party between now and the 7th May. Interesting stuff!
We would be very surprised if any of the other nine '100%' predictions were to change between now and 7th May. All of these nine seem nailed-on certainties as the saying goes. However, there were ten of these 'certainties' following our January poll, and then Newry and Armagh changed in our February poll, so who knows!
So let's see how it goes over the next nine weeks to election day. For your diaries the next Opinion Panel results will be published at the start of April and start of May. So our last Opinion Panel forecast, will be only a few days before the election, and we'll be able to see how accurate this turns out to be when compared to the actual election results!
LucidTalk's opinion panel forecast (poll period: 19 - 27 February 2015):
Methodology and Background
The LucidTalk Northern Ireland (NI) Opinion Panel is made-up of 440 members and is carefully constructed to provide an accurate representation of Northern Ireland opinion - via gender, area of residence, age-group, community background, socio-economic group, and employment group. The opinion panel has a pool of approximately 1,200 hundred members who regularly take part in LucidTalk poll projects, and for this project a representative sample of 440 opinions will be collected on a monthly basis, up until the 7th May 2015 - election day. NB The LucidTalk Northern Ireland Opinion Panel is constructed and managed to British Polling Council (BPC) standards.
Opinion Panel members will be asked:
(a) Likelihood of voting at the 7th May 2015 Westminster election?, and
(b) What party they currently plan to vote for?, and
(c) Their current opinion regarding who would win in each of the 'non-100%' certainty seats (currently eight - see table).
- Alternative forecast for seats is only provided were the probability of the main forecast is less than 100%.
- All % probabilities are shown in '5%' blocks, e.g. 50%, 55%, 60% etc..
- % forecast is calculated via LucidTalk computer-based prediction models, taking into account:
- Current LucidTalk 'Opinion Panel' poll (440 participants): February 2015. 80%
- Current and recent LucidTalk polls for 2015 and 2014 (September): both results and trends. 10%
- Election results and patterns from 2010 - 2014 (European, UK, NI Assembly): both results and trends. 10%
- Above forecast table is representative of NI opinion, and specific constituency opinion, to an error of +/-4%.
- It should be noted that the forecast table provides the % likelihood of the named political parties winning the specific parliamentary seat. It does not predict vote share, size of vote, and/or size of majority etc.
- Projections currently assume no 'agreed candidate' arrangements between e.g. Unionist parties. When and if any 'agreed candidate' pacts are announced these will be built-in to the computer prediction models.
- Re. the one independent MP as included in the forecast table - In North Down it is assumed that Sylvia Hermon will run again, and hold that seat.
This forecast was collated using a UK election prediction model based on a blend of the traditional uniform national Northern Ireland Swing, and proportional loss forecasting techniques. It also takes into account recent LucidTalk local constituency opinion polls. All the results and data were then analysed using multiple iterations to achieve the correct forecast percentages.
Bill White is Managing Director of Belfast polling and market research company LucidTalk, polling partners to the Belfast Telegraph. You can follow LucidTalk on Twitter: @LucidTalk.
Belfast Telegraph Digital