Swing Low Sweet William

Because of the "first-past-the-post" electoral system that we use in Britain, it is not sufficient for a party to win votes: it has to concentrate those votes in the right places so that it can win seats. This is one of the problems facing William Hague. At the last election, Conservative votes were distributed around the country far less efficiently than were Labour votes; if there is a uniform swing across the country to the Conservatives at the next election, the Tories will need to win far more votes than Labour to become even the largest party in a hung Parliament, let alone to win an overall majority.

But is that assumption of uniform swing a reasonable one? At the election of 1992 and again in 1997, Labour did much better in the marginal seats than it did in the rest of the country. In other words, the swing was not uniform, and Labour gained many more seats than it might have expected under uniform swing.

To find out, we can re-analyse MORI's most recent political polls, to isolate the voting intentions of electors living in marginal constituencies. To deprive Labour of its overall majority, the Conservatives need to gain 89 of Labour's seats; to win a majority of their own, they need a further 75 gains. If we analyse the results from these seats separately - the 89 most vulnerable seats, and the next 75 "pivotal" seats (which are the seats that both parties need to win for an overall majority - we see that the swing over the Dome and petrol crisis period was much smaller. We need to treat the figures with a little caution, as the sample size within these key seats in any single survey is relatively small (around 250 in each of the two categories), but we can get at least a rough approximation of what is happening.

In the period between the three MORI Omnibus polls carried out over 3 August - 4 September and the MORI political poll for The Times carried out on 21-26 September, there was a swing from Labour to the Conservatives across the country of 7%. Before this period, which included the announcement of further funding to bale out the Millennium Dome and then the petrol crisis, taking the aggregate of the three polls, the Conservatives had 32% of voting intentions and Labour had 48%, a lead of 16% (higher than in 1997); afterwards, the lead had been cut to just two points, with the Tories up 3 on 35% and Labour down 11 on 37%.

But the bad news for the Tories is that this gain was not taking place in the key constituencies where they need to win the next election. In fact, not only did Labour support fall less in these constituencies than in the rest of the country, but Tory support fell here too! The entire collapse in the Labour vote in the marginals went to the Liberal Democrats and to other parties.

Up to the start of September, although the Conservatives were well behind in the polls, at least the pattern of swings was in their favour: there had been a 3.5% swing to them since the election in the most vulnerable seats, and a 3% swing in the pivotal seats, even though the country as a whole had swung against them. The vote shares were an under-indication of the number of seats they could expect to win. No more! In the marginal seats the swing since the election is now almost exactly the same as in the country as a whole. All that gain of support has been concentrated where it can't do them any good. Of course, they would still gain a few seats because in some constituencies Labour's share would have fallen past theirs; but not many. Unless this poll is a misleading one-off (and it must be emphasised again that the sub-sample sizes are small and the margin of error correspondingly large), uniform swing has re-established itself; and that is very bad news for the Tories.