A word of warning.

Always expect things to be added slowly to this blog.

Thursday, May 20, 2010

AV - Alternative Vision - Alternative Votes

This is a brief review of the impact alternative electoral systems might have had on the last election in NI for Westminister on 6th May 2010. In this, the first of two reviews, I consider what impact the Alternative Vote (or AV) might have had. After the May 6th election the DUP held 8 seats, Sinn Féin had 5, the SDLP had 3, the Alliance party had 1 and there was one independent. By my count, under AV, the final tally would have been DUP 8, Sinn Féin 5, the SDLP 2, UCUNF 2 plus one independent.

As a system, AV has been widely discussed by various electoral commissions in the UK, so this provides a rough test case for the type of result it might have generated. Given the complex electoral mathematics at Westminister, this is no little story. Whilst there are numerous flavours of AV, most suggest that, in effect, a successful candidate must get to over 50% to be elected. The method uses either a number one preference and an alternate (for the vote transfer), or applying a STV type system for single seats (i.e. to reach a quota of 50%+). In effect, I've fudged that below to carry the calculations through to the end.

In terms of the broader narrative of the election, both Peter Robinson and Reg Empey would probably have won their seats, whilst the failure of the unity candidate in Fermanagh/South Tyrone would perhaps have been off-set by the victory of Paula Bradshaw in South Belfast. The SDLP would have lost a seat and the Alliance (somewhat ironically given the LibDem link-up) might not have gained. So the post-election mood amongst the various parties would probably be very different (apart from SF, seemingly).

To do this - I simply ran the stats of the NI seats to see which (if any) might have changed hands. To date I've done some numbers for two forms of voting system - the alternative vote system (AV) where you nominate candidates in order with your vote transfered to the next candidate when yours is eliminated - and the single transferable vote system (STV) for multi-seat constituencies. I'll include that as a future blog. Today's concentrates on AV itself. For vote transfers, I've used the data provided on the 2007 Assembly election to assist in transfers from one party to another as a broad guide. Again, all these things have in-built health warnings and this should be largely inhaled as an exercise in navel gazing, anyway.

A quick demonstration of the niceties of the current first past the post principle is given by the DUP's David Simpson. He won Upper Bann with 33.8% of the vote, a lower vote share than Rodney Connor on 45.5% in the constituency of Fermanagh/South Tyrone and Gerry Kelly on 34% in North Belfast. Obviously, one seat does not provide a critique of the whole methodology, but a considered view of each seat shows that the overall result would have differed in a number of cases, some perhaps significant.

Basically, you need to get over 50% to take a seat under AV, so the likes of Martin McGuinness, Lady Sylvia Hermon and Gerry Adams would be elected immediately. Jeffrey Donaldson, on 49.9% wouldn't have much of an issue on getting to the 50%. Others that would be so close to the line that a loss would be very unlikely (factoring in sectional transfers) would be Margaret Ritchie (48.5%), Pat Doherty (48.4%), Ian Paisley Jr (46.4%), Sammy Wilson (45.9%) and Jim Shannon (45.90%). None would appear to under any real threat from an AV type voting system.

The first seat where AV might have produced a different result would be Fermanagh/South Tyrone, although a review of the electoral numbers suggests it is very unlikely. As Michelle Gildernew and Rodney Connor both polled 45.5%, whilst the third place candidate (Fearghal McKinney) only polled 7.6%, any elimination system would probably remove all candidates other than Gildernew and Connor since none could catch either of them up. While an apparent reluctance of SDLP voters to transfer to SF has been a feature of some previous elections, I wonder how much that would extend to this particular contest, following the polarisation of the election with a unionist unity candidate. That would probably have prompted enough SDLP voters (maybe 60%) to give their transfers to SF. In this case, if 4.5% of McKinney's vote transfered to SF with 3.1% to Connor, it would be enough to keep Michelle Gildernew in the seat. Although, in any event, it would still be very tight but SF would hold on.

A number of other seats seem relatively safe, such as Foyle, where Mark Durkan's 44.7% would have been enough to hold off the threat of SF, since even the first elimination would probably see the distribution of the votes of all candidates up to Maurice Devenney (on 11.9% for the DUP). Even in the (somewhat) unlikely event that all the 11.6% of votes transfered to Martina Anderson, it would still only have her on 43.8% and behind Durkan with the DUP vote to be distributed. So it wouldn't have any impact on Foyle either.

Similarly, for Conor Murphy who won on 42%. While the order of elimination may be an issue here, the three lowest candidates would be distributed first as the highest (William Irwin on 12.8% for the DUP) would not be able to pass out the next highest candidate. The distribution of those votes would possibly lift Danny Kennedy (UCUNF) from third on 19.1%, past Dominic Bradley (on 23.4%) in second, unless sufficient numbers of the Alliance (1.2%), Frazer (1.5%) and DUP transfer (12.8%) swung in behind Bradley to keep him ahead. Presuming that Frazers 1.5% would go solidly to Kennedy, and assuming (on previous transfers) that more Alliance votes might go to Kennedy than Bradley, it would require at least 4.15%, or one third of the DUP's 12.8% to go to Bradley. Since 80-90% of DUP transfers have gone to the UUP, this seems inherently improbable. The underlying issue here would be whether Bradley could be kept in the field against Murphy, assuming that an anybody-but-SF vote would combine to keep him out. It seems unlikely that such a high percentage of DUP transfers would be in-built to aid Bradley (over transfering to another unionist). So, the elimination of Irwin would probably benefit Kennedy enough to push Bradley back to third. In the event of Bradley's 23.4% (plus add-ons) being distributed, it would seem unlikely that less than one third would go to Murphy over a unionist (normally SDLP-SF transfers are in the region of 60%), so this, despite a little drama, this would probably be a SF hold.

Now South Belfast would, anecdotally, have been regarded as a much more likely seat for a change. However, while Alasdair McDonnell's 41% was matched by a combined DUP-UCUNF vote of 41%, the fate of the seat would likely be subject to the distribution of both Adam McGibbon's 3% (Green) and Anna Lo's 15% (Alliance) of the vote. Based on the 2007 Assembly transfers, the Green vote seems to go 50-50 to unionist and nationalists, whilst the Alliance vote goes 60-40, favouring unionists, generally UUP (about 85%). Applying this, I'd have Bradshaw up by 1.1% of McGibbons vote and 7.6% of Anna Lo's, taking her to 26%, while Spratt would have moved up to 25.5% slightly behind Bradshaw going into the last distribution of votes, with McDonnell on 48.5%. This order of the field would radically alter the last distribution of votes under AV. In the last Assembly election, general transfers from DUP to UUP/SDLP split 94%/6%, which amounts to an additional 1.53% for McDonnell and 23.97% to Bradshaw. It feels unlikely that, faced with a choice between a unionist and a non-unionist, that Spratt's voters would give a preference that might aid McDonnell, so AV would seemingly have given the seat to Paula Bradshaw.

Nigel Dodds would seem to be pretty safe under AV, as well. His 40% would have risen considerably on the elimination of the four lowest votes (since none could overtake either himself or Gerry Kelly on 34%). Again, using Assembly transfer figures (based on the ratio of DUP/SF from a particular party and a 50-50 split of McAuleys 1.1%) Dodds would attract a further 11.95% seeing him safely over the 50%.
East Belfast would present an additional challenge, given the circumstances of the election, probably more so than most other seats. Naomi Long's 37.2% is well short of the 50%, but the immediate elimination of all but her and Peter Robinson (on 32.8%) in the first redistribution would normally have put her into serious trouble, as one would expect significant numbers of Ringland's 21.2% and Vance's 5.4% to favour an openly pro-union candidate. While the context of the election may arguably suggest otherwise, if the UCUNF to DUP/Alliance transfers even partially followed the last Assembly election (94%/6%), Peter Robinson would have held onto his seat under AV as he'd probably have got upwards of the 17.2% he'd needed from Ringland and Vance who had a combined 26.6%.

David Simpson, despite the low winning take of 33.8%, would no doubt have won easily enough under AV. The elimination of Heading (Alliance) and Kelly (SDLP) would probably have pushed O'Dowd (SF) from 3rd to second, eliminating Harry Hamilton (UCUNF), whose vote would then have got Simpson over the line with ease. So no change there.

Gregory Campbell, despite winning on a modest 34.6%, would have been under relatively little pressure from the rest of the field. The elimination of Fitzpatrick (Alliance) would mainly benefitted the SDLP and UCUNF, whilst the elimination of Willie Ross (TUV) would mainly have transfered to Campbell, keeping him in the very low 40s. In this case, Ross and Fitzpatrick would probably have contributed enough transfers to Macaulay (UCUNF) to lift him above Cathal Ó hOisín (who may have gone up to 20% at most) into second, probably on around 22%. The distribution of Thomas Conway's SDLP vote (following the last Assembly transfers) would give maybe 6.5% to Macaulay and 8% to Ó hOisín, but perhaps not enough to displace Macaulay from second place. At the final re-distribution (of Ó hOisín's vote), there would be the odd spectacle of SF transfers deciding between a DUP and UCUNF candidate. Given a 50-50 split in the SF to DUP/UUP transfers in the last Assembly election, only a general Campbell-out campaign would have unseated him. But more than likely a straightforward DUP hold.
Lastly, South Antrim where William McCrea won on 33.9%. Close behind him was Sir Reg Empey on 30.4% leading the local branch of the Unionist-Conservative ticket which would, on policy terms, not support the likes of AV. The first eliminations would have included the TUV (5.4%), Alliance (7.7%), SDLP (8.7%) and SF (13.9%). The TUV vote would have probably have transfered strongly to McCrea, whilst the balance of the 16.4% of Alliance and SDLP transfers would have heavily favoured Sir Reg Empey with more than 14% going to Empey, taking him up to 44.4% or higher. SF transfers would split 50-50 between Empey and McCrea (although you'd feel that reaction to McCrea would be strong enough for that  percentage to lean more to Empey). Splitting McLaughlin's vote in two would bring Empey to over 50%. So the paradox is that the candidate fronting an alliance that opposes AV did not win a seat that he would probably have won under AV. There is probably a lesson there!

No comments:

Post a Comment