Conventionally, the implications of the opinion polls are interpreted on the assumption of uniform swing. At present, this implies that Labour retain a massive advantage, under which they can trail in numbers of votes while still keeping an overall majority of seats. Meanwhile Michael Howard would need the biggest swing achieved by any party since the war merely to reach Downing Street with a wafer-thin majority.
Probably that message is broadly true. But is uniform swing a safe assumption? In both 1997 and 2001, Labour far outperformed its national result in the marginal seats, winning more seats than uniform swing would have predicted for a given vote share and so setting up the apparently biased starting position that holds today. Is this a permanent shift in the distribution of votes, or simply a result of factors specific to those two elections? Is this a temporary and unnatural Labour advantage in the marginals – an ever-more tightly wound spring that might be released to drive voting patterns in the opposite direction at any moment?
Suppose the Labour advantage built up in the marginals over the last two elections were to melt away; then the pattern of vote distribution would not be that of 2001 but that of 1992. In that case, the Tories would need only around a four percent swing to wipe out the Labour overall majority, equating to a one percent Labour lead nationally – well within the scope of their current poll showing.
Two factors, at least, contributed to Labour’s over-performing in marginal seats. Low Tory morale and effective Labour organisation have ensured that the latter has campaigned far more effectively on the ground, a difference that bites most in the marginals, where both parties’ efforts are most concentrated. Labour has also certainly been preferred to the Conservatives by most Lib Dems – which was not true in the 1980s – allowing a squeeze and switch of tactical votes.
What happens to the tactical vote may well depend on the Liberal Democrats, assuming their supporters will take advice from their leaders. The dilemma facing the Lib Dems is that, while they are far better placed to pick up seats from the Tories than from Labour, this is of no help in their primary aim: to gain a share of power. Only a hung parliament can give them this, and only Tory gains in Labour marginals are likely to contribute significantly towards that – therefore their case may be best-served by Lib Dems voting tactically for the Tories in Labour seats.
Of course, this would force them to abandon the argument, given momentum by by-election successes, that they can leapfrog the Tories from third place to capture Labour seats thanks to disillusionment of Labour’s core voters with Blair over Iraq. But, realistically, swings of Brent or Hodge Hill dimensions will not recur across the country at a general election and without them, it is only Conservative seats that the Lib Dems can take in any numbers (of the 100 most marginal Conservative seats, the Lib Dems stand second in 31; in the 100 most marginal Labour seats they are second in only seven). Yet the ambitions the by-elections have aroused may be enough to persuade Lib Dems to strive for unlikely victory rather than voting tactically in marginals, and so to buttress Labour MPs in seats they could not otherwise hold – unless somebody tells them otherwise. Which way will Charles Kennedy jump?
Nor will the campaign in the constituencies necessarily follow the example of 1997 and 2001. Though perhaps impossible to measure, the morale of the ‘troops’ on the ground can be a crucial factor in an election, especially when the task is to turn grudging, latent support into votes at the ballot box. In both 1997 and 2001, Tory morale at local level was at rock bottom; all the signs are that it is more healthy now, and this year’s local elections will have helped cement that improvement in many parts of the country, though the possibility of splintering and defections to UKIP are a threat.
At the same time, Labour has suffered. Policy differences are corrosive in themselves, but the more direct human factor should not be discounted. Here the council election defeats this year, when Labour lost a net 479 seats, though almost forgotten by the media, may be of real import. Hundreds of Labour party members, many senior and influential within their local parties, were councillors at the end of May and are no longer; others have kept their seats but lost council leaderships, committee chairmanships or memberships and perhaps even their turn as mayor. Most, if not all, will be blaming Tony Blair and his government for their personal disasters.
But these are the very men and women on whom the organisation of the constituency election campaign ultimately depends, the grassroots membership who knock on doors and stuff leaflets through letterboxes. Will they work as hard for Labour in 2005 as in 1997 and 2001? If Labour’s superior organisation crumbles as Tory morale improves, it is by no means impossible that it is the Tories who will over-perform in the marginals, even without the tactical voting factor.
Neither of these factors can be measured at this stage – few voters make tactical voting calculations in mid-parliament, and the effectiveness of constituency campaigning will only make its contribution in the final weeks. Realistically, they are unlikely to turn Labour victory into Labour defeat, but they add to the uncertainties.
But one startling instance of a non-uniform swing, demographic rather than geographic, has already emerged. Traditionally, and fairly consistently since women first won the vote in 1918, Tory support has been higher among women than among men. The gap has been smaller in recent years, and dwindled briefly to zero at the 1987 election, but for women to be less likely than men to support the Conservatives is unprecedented; but, suddenly, it has been happening. Throughout 2004, up to the end of August, MORI’s monthly polls have been finding the Tory share lower among women than men. Aggregating all MORI’s polls from January to August (a total of more than 30,000 interviews), we find of those who are absolutely certain they would vote, men would vote 35 percent Labour, 35 percent Conservative and 19 percent Lib Dem, while women would vote 36 percent Labour, 32 percent Conservative and 24 percent Lib Dem.
The reasons for this switch are not yet clear, but it may well be, as the Lib Dems are the beneficiaries, that Conservative support for the government’s policy in Iraq is one contributory factor.
However, MORI’s September poll shows a break in the pattern, with the Tories once more performing as well among women as men (better if we consider only those certain to vote) while the Lib Dems perform worse. Is this just a blip, or have the old certainties reasserted themselves? Only time will tell, but if the former it means the profile of Conservative supporters will be significantly different from at any election in the recent past, and to offset their lower relative support among women the party will need to do much better among men than before.
Likely to be most significant of all, though, is turnout. MORI’s September poll gave the Tories a one percentage-point lead assuming a 50 percent turnout, however Labour would have an eight-point lead on an 80 percent turnout. Lib Dem share of the vote, by contrast, is constant at all likely levels of turnout. Clearly lower is likelier, with the incentive to vote not helped by the diminishing numbers who can distinguish between their preference for the parties’ policies on key issues. (Labour’s lead on healthcare, for instance, has fallen from 40 points before the 1997 election to thirteen today). Any party that can drag back to the polls the supporters who voted in 1997 and stayed at home in 2001 will have no need to worry about non-uniform swings or any other vicissitudes of the electoral system.