Constituency-level analyses of national opinion polls carried out during 2013: A review.

Adrian Kavanagh, 31st December 2013

During the past year (2013), I carried out constituency-level analyses of thirty four different national opinion polls in order to produce Dail-seat estimates based on those support levels. These analyses involved thirteen Red C (Sunday Business Post-Red C and Paddy Power-Red C) polls, twelve Sunday Independent-Millward Browne polls, four Irish Times-Ipsos MRBI polls and five Sunday Times-Behaviour & Attitudes polls. Averaging the different Dail seat-level estimates for the different political parties and political groupings across these polls produces the following estimate: Fianna Fail 46, Fine Gael 52, Labour Party 11, Sinn Fein 25, Independents and Others 24. But notably different seat-estimates were produced depending on the time of year that these polls were held in, but especially depending on what polling companies were involved in the carrying out of these different polls, as will be illustrated below.

Constituency support estimates for different parties and groupings form the basis of this general approach asking the question in relation to different opinion poll results – what do these poll figures mean in terms of the likely number of Dail seats won by the different parties and groupings? Although the Irish electoral system is classified as a proportional electoral system, the proportion of seats won by parties will not measure up exactly to their actual share of the first preference vote, mainly because these first preference votes need to be filtered through the system of Irish electoral constituencies. In order to address this question, I estimate what the party first preference votes would be in the different constituencies, assuming similar (proportional) change in party vote shares in all constituencies. This of course is a very rough model and it  cannot take appropriate account of the fact that changing support levels between elections tend to vary geographically, while it also fails to take account of the local particularities of the different regions in cases where no regional figures are produced in association with different national opinion polls meaning that there is no scope to carry out separate regional analyses based on these poll figures. Thus constituency support estimates for different parties will be over-estimated in some constituencies and under-estimated in others, but the expectation would be that the overall national seat figures figures estimated will be relatively close to the true level, given that over-estimates in certain constituencies will be offset by under-estimates in other constituencies. Based on these estimated constituency support figures, I proceed to estimate the destination of seats in the different constituencies. The constituency level analysis involves the assigning seat levels to different parties and political groupings on the basis of constituency support estimates and simply using a d’Hondt method to determine which party wins the seats, while also taking account of the factors of vote transfers and vote splitting/management (based on vote transfer/management patterns observed in the February 2011 election). Based on such an analysis and using the new constituency units (as defined in the 2012 Constituency Commission report), these analyses estimates what party seat levels would be, should such national support trends be replicated in an actual general election. (Analyses on the adriankavanaghelections.org elections commentary site suggests that Fianna Fail would seem to be the party most likely to be positively effected by the redrawing of the constituency boundaries, with the Labour Party being the party likely to be the most adversely effected by these changes.)

Due to unusually high/low support levels for some parties or political groupings in certain constituencies in the previous election, the model may throw up occasional constituency predictions that are unlikely to pan out in a “real election”, but the estimates here cannot be seen as highly accurate estimates of support levels at the constituency level as in a “real election” party support changes will vary significantly across constituency given uneven geographical shifts in support levels. But the ultimate aim of the models are to get an overall, national-level, estimate of seat numbers and these are based on the proviso that an over-prediction in one constituency may be offset by an under-prediction in another constituency. I have made some further corrections to the base support figures for the different parties for this analysis to take better account of the impacts on support of the 2012 Constituency Commission report boundary changes with especial reference to the Dublin constituencies. For instance, these figures better reflect the weaker positions of Fine Gael and Fianna Fail in Dublin Central after the moving out of the Ashtown area to Dublin West and the Botanic/Drumcondra area to Dublin North West, but also their stronger positions in Dublin West and Dublin North West. Fine Gael are assigned an extra seat in Dun Laoghaire on the basis that the Ceann Comhairle, Sean Barrett, will be automatically returned at the next general election (unless he decides to retire from politics before this) and this constituency will effectively be rendered a three-seat contest at the next general election.

The following table outlines the party Dail-seat level estimates produced for each of the thirty four national opinion polls analysed during the 2013 calendar year:

Poll FF FG LB SF IND/OTH
ST-BA Dec 22 39 59 15 20 25
IT-MRBI Dec 12 38 58 8 32 22
SBP-RC Nov 24 40 60 16 20 22
SI-MB Nov 17 43 52 9 30 24
PP-RC Nov 7 48 50 9 22 29
SBP-RC Oct 27 45 57 9 22 25
SI-MB Oct 13 51 50 8 25 24
IT-MRBI Oct 1 39 54 0 38 27
SI-MB Sep 29 51 53 11 26 17
ST-BA Sep 22 39 51 13 26 29
SBP-RC Sep 15 44 53 12 22 27
SI-MB Sep 1 45 56 3 34 20
SI-MB Aug 18 53 58 4 24 19
PP-RC Aug 8 38 56 13 20 31
SI-MB Jul 7 55 48 4 27 24
SBP-RC Jun 30 39 56 15 24 24
ST-BA Jun 30 47 57 2 23 29
IT-MRBI Jun 14 51 42 10 33 22
PP-RC Jun 13 44 59 11 21 22
SI-MB Jun 2 51 49 13 22 23
SBP-RC May 26 50 48 13 23 24
SI-MB May 19 53 40 16 26 22
SBP-RC Apr 28 44 57 13 22 22
SI-MB Apr 14 55 43 15 22 23
ST-BA Mar 31 43 55 3 22 35
SBP-RC Mar 24 43 58 16 19 22
SI-MB Mar 17 56 44 10 28 20
SI-MB Mar 3 43 45 14 31 24
SB-RC Feb 24 46 55 14 21 22
SI-MB Feb 17 53 45 16 26 18
IT-MRBI Feb 9 54 45 13 24 22
SB-RC Jan 27 38 56 16 25 23
ST-BA Jan 27 42 51 15 26 24
PP-RC Jan 10 37 56 18 23 24

The following table shows how these figures can be averaged out according to which polling company was involved in the carrying out of the different polls, as well as the time of year (different quarters) in which these polls took place:

Poll FF FG LB SF OTH
Average 2013 46 52 11 25 24
Red C 43 55 13 22 24
Ipsos MRBI 46 50 8 32 23
Millward Browne 51 49 10 27 22
B&A 42 55 10 23 28
January-March 46 51 14 25 23
April-June 48 50 12 24 23
July-September 46 54 9 26 24
October-December 43 55 9 26 25
Maximum seat estimate 56 60 18 38 35
Minimum seat estimate 37 40 0 19 17

Based on the figures, the best polls from Fianna Fail’s perspective would appear to have been those taken during the spring and summer of 2013, while their worst figures came at the beginning and the end of that year. The best poll for Fianna Fail was the Sunday Independent-Millward Browne poll of March 17th, while the very first poll taken during the year – the Paddy Power-Red C poll of 10th January – was the party’s worst in terms of seat estimates figures. In short, Fianna Fail’s improving fortunes in opinion polls in the latter part of 2012 continued into the first half of 2013, but the latter part of the year saw this resurgence in Fianna Fail support levels tailing off somewhat.  

The best polls from a Fine Gael perspective would appear to have been those taken at the end of 2013, while their worst figures came during the spring and early summer. The best poll for Fine Gael was the Sunday Business Post-Red C poll of November 24th, while the Sunday Independent-Millward Browne poll of 19th May was the party’s worst in terms of seat estimates figures.  

The worst polls from a Labour Party perspective would appear to have been those taken during the summer and autumn of 2013, while their best figures came at the beginning and the end of 2013. The best poll for the Labour Party was the Paddy Power-Red C poll of 10th January (the very first poll of 2013) , while the worst poll from a Labour Party perspective was quite obviously the Irish Times-Ipsos MRBI poll of 1st October.  

There does not seem (based on the quarterly averages noted above) appear to have been any period of the year in which Sinn Fein were performing significantly better or worst across the different opinion polls in terms of the Dail-seat number estimates produced for this party. The best poll for Sinn Fein was the Irish Times-Ipsos MRBI poll of 1st October, while the worst poll from that party’s perspective was quite obviously the Sunday Business Post-Red C poll of 24th March.  

Similarly, there does not seem (based on the quarterly averages noted above) appear to have been any period of the year in which the Independents and Others grouping performed significantly better or worst across the different opinion polls in terms of the Dail-seat number estimates produced for this grouping.  The best poll for the Independents and Others grouping was the Sunday Times-Behaviour & Attitudes poll of March 31st, while the Sunday Independent-Millward Browne poll of 29th September was the worst in terms of seat estimates figures for this grouping.

On average, this was a disappointing year for the government parties in terms of support levels in the different national opinion polls, with this being reflected in the Dail-seat number estimates for these parties arising from the different constituency-level analyses of these polls. None of the opinion polls carried out during 2013 produced a scenario in which these parties were predicted to win a sufficient number of seats combined (79 for the next Dail, assuming the Ceann Comhairle position is filled by someone else from another political party/grouping) to command a majority in the (158-seat) Dail. On a more positive note however, the last few polls pointed to improving fortunes for Fine Gael and Labour. The Red C polls generally tended to be consistently associated with more positive support estimates (and hence Dail-seat estimates) for the government parties, while the estimates associated with the Ipsos-MRBI and Millward-Browne polls were decidedly less favourable for these parties. The best party support and Dail-seat number estimates for the Labour Party were associated with the Red C polls, while the party support/seat estimates levels for Fine Gael were also highest in the Red C polls but also the Behaviour & Attitudes polls.

Combined seat-level estimates for Fianna Fail and Sinn Fein generally exceeded those for the government parties drawing from the analyses of the majority of the polls (24 of the 34) carried out during 2013, including all of the Ipsos-MRBI and Millward-Browne opinion polls. The gap between the combined Dail-seat levels for potential Fianna Fail-Sinn Fein and Fine Gael-Labour alliances  was most evident in the polls carried out in the middle of the year, but this gap did narrow significantly with the polls carried out as 2013 drew to a close. The Millward Browne polls generally tended to be consistently associated with more positive support estimates (and hence Dail-seat estimates) for Fianna Fail, while the estimates associated with the Behaviour & Attitudes and Red C polls were decidedly less favourable for that party. In the case of Sinn Fein, it was the Ipsos-MRBI polls which generally tended to be consistently associated with more positive support estimates (and hence Dail-seat estimates) for that party, while the estimates associated with the Behaviour & Attitudes and Red C polls were decidedly less favourable for Sinn Fein, as was the case with Fianna Fail. While the Dail-seat level estimates, on average, for the Independent and Others grouping did not vary significantly between the Red C, Millward-Browne and Ipsos-MRBI polls, this grouping did fare significantly better with the Behaviour & Attitudes series of polls.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s