...I suggest that the fall in nationalist turnout to levels approximating to Unionist turnout is a more important factor...I posted a comment that turnout differential between the communities may not tell us all that much, but agreeing that we might learn more from some more analysis. This post is that analysis.
...It seems to me that nationalist/republican exuberence with the GFA, and a latter realisation that they don’t need ‘every vote’ to win most western constituencies caused an exceptional voter-turnout in early post GFA elections. Whyte shows that nationalist vote is constant despite latter lower voter turnout down in nat-majority areas...
...do you agree that in the Westminster election there was a narrowing of the differential in the community background turnout with the Nationlaist turnout falling and the unionist turnout either stabilsing or growing?...
Now, the question itself needs careful unpacking. Many (myself admittedly sometimes included) fall into the lazy trap of assuming that all Catholics are default Nationalist voters, and all Protestant voters are default Unionist voters. This of course isn't so. A voter can only be defined as Nationalist or Unionist if they actually go into the polling booth and mark the ballot paper for a Nationalist or Unionist candidate. A fall in support for Nationalist parties doesn't mean that Nationalist voters did not turn out, it means that voters who were Nationalist voters last time either didn't vote or opted for someone else.
Nationalist votes in 2010 were down from 300,156 to 282,912 (but up from 41.8% to 42.0%); Unionist votes down from 369,704 to 340,602 (decrease from 51.8% to 50.5%). That’s a 5.7% drop in absolute voting numbers (17,244) for Nationalists and a 7.9% drop (29,084) for Unionists. If the Unionist vote had dropped by the same 5.7% as the Nationalist vote in 2010, it would have been 348,464 rather than 340,602, a difference of 7,844 between the counterfactual situation and the reality. The Alliance Party got 14,471 extra votes in reality. So my initial broad-brush statement that "The decline from 51.8 per cent of votes for unionist parties in 2005 to 50.5 per cent in 2010 is more than accounted for by the gains made by Alliance and smaller centre parties" is at least accurate though possibly could be more precise.
(Health warning: all figures for individual seats below are calculated from notional 2005 results rather than actual results. For some of the seats of course this makes no difference but it means that for those that have changed the precision of my figures should be taken with a pinch of salt.)
Anyway, my conclusion from staring at the figures in a jet-lagged state for some time is that one can't really draw much more from them than that turnout was down overall and particularly in the West and in boring contests. I will, however, refine my previous statement that the Unionist decline is more than accounted for by Alliance growth. In fact, the 7,844 gap between votes for Unionist parties as cast in reality, and as they would have been had their party support declined at the same rate as Nationalist support, is not only less than the 14,471 votes gained by Alliance, it is less than the 9,019 extra votes gained for her party by Naomi Long in East Belfast. The bigger picture is messier. There was no other seat where Alliance gains exceeded Unionist losses. So the picture is clearly much more complicated. (I should clarify that Alliance support is of course much more broadly based than picking up disaffected Unionists; the context of my original remark was an article looking at what had happened to Unionist votes since 1921.)
The four seats with a more than 10% drop in turnout are all safely held by Sinn Fein - West Tyrone (-11.7%), Mid Ulster (-10.3%), West Belfast (-10.3%), and Newry and Armagh (-10.2%). We could add also Foyle, safely won by the SDLP with a turnout change of -8.8%, to this category. However, it is not only Nationalist parties whose vote was falling there. Nationalist vote share actually increased in West Tyrone, Mid Ulster and West Belfast, and dropped by only 1.1% in Newry and Armagh and 0.9% in Foyle; in West Tyrone there were special circumstance because of Kieran Deeny's candidacy in 2005, but in Mid-Ulster the Unionist vote fell by 2100 and in West Belfast by 1500, proportionately rather more than the Nationalist decrease (respectively 2500 and 3300). So I think this disproves the theory that post-Agreement Nationalists are now sitting back from the polling station, comfortable in the glow of victory, unless we add to that picture post-Agreement Unionists morosely abstaining in the gloom of defeat (ie we would have to believe in people being both less likely to vote if their side wins, and also less likely to vote if their side loses, which rather removes the usefulness of the analysis).
One thing does seem clear to me though. Turnout dropped least where there was an interesting contest (with one exception which I'll get to later). In the two seats won by moderate women challenging the Unionist establishment, North Down and East Belfast, turnout actually increased (though not by much - 0.8% and 0.9% respectively). In North Belfast, where there was a possibility of a sufficiently split Unionist vote for SF to gain a seat, and Strangford, where the UUP had put up one of its more credible candidates and the DUP were defending the dubious legacy of Iris Robinson, turnout dropped by only 1.5%. In South Belfast, where the SDLP had what would have been a tough defence in a different year, and South Antrim, where the UUP leader was doing his ineffective best to unseat the DUP, turnout dropped by 3.2% and 3.5% respectively. Everywhere else was within the 4-7% window, apart from the five safe Nationalist seats mentioned in the previous paragraph. (The TUV challenge doesn't seem to have made much difference to turnout.)
One anomaly really jumps out. If people are more likely to keep voting where there is an interesting contest, why did turnout in Fermanagh and South Tyrone drop by 4.5%, 1500 fewer votes cast in an election where the electorate had increased by 1200, and where the result was determined by a margin of only 4? One could of course argue that we're simply seeing the same geographically based turnout drop of around 10% in other western seats, masked by a bonus of 6% or so extra because of the closeness of the race. In comparison with 2005, the total Nationalist vote decreased by 990, the total Unionist vote by 1625 (and we can be exact because the boundaries were the same). This is the one seat where a local differential in the change in community turnout surely did determine the outcome. Had the two sides lost proportionately the same number of votes - had indeed the Unionist and Nationalist vote totals decreased by the NI-wide average of 7.9% and 5.7%, rather than the actual figures of 7.0% and 4.0% - the Unionist candidate would have defeated Sinn Féin by 175 votes. But my own gut feeling is that the numbers bear out the reports of serious internal problems in the Unionist campaign which were discussed on various blog posts after the election (and, frankly, indicated between the lines of the court judgement affirming the election result).
So, my conclusions are that i) the relatively greater drop in Unionist turnout is very marginal and more than accounted for by the East Belfast result alone; ii) we're now seeing a lower turnout from both Catholic and Protestant voters in the rural areas where a high turnout was once accepted as inevitable; and iii) voters, like commentators, will pay more attention if the race appears to be a close one.