Low Turnout: Who Loses?

In last week's column I set out some of the evidence that suggests that the turnout in the forthcoming election may be even lower than in 1997. However, one aspect that I didn't address was the political impact of a low turnout.

The received wisdom, which seems to be accepted almost universally, is that a low turnout this time will harm Labour. The Tory vote is less soft than Labour's, it is argued, partly because the Tories have been reduced to their core supporters while all the floating vote is in the Labour column. Conversely, other commentators suggest that it is Labour's 'core' voters who are less likely to turn out, because the party's heartlands are disillusioned with a government that has made too many compromises of principles in its search for the political centre ground. Whatever the reason, it has certainly been the case that the Conservatives have made substantial gains at both council and European Parliament level over the last couple of years in elections which have plumbed new depths of poor turnout in British elections.

Surveys last year and earlier this year offered some support to the thesis that Labour had most to lose from a low turnout: our survey for The Times in January, for example, found 63% of Tories and 62% of Lib Dems but only 52% of Labour supporters saying they were certain to vote; the difference in voting intention between all those who named a party and those who said they were certain to vote was the equivalent of a 3.5% swing to the Tories. The corresponding poll in March found the same. However, as the table shows, the differential was much smaller in February, and again at the end of last month, in two polls for The Times and The Sun. Indeed in the last of these, the difference was in Labour's favour, though it was not statistically significant.

Voting Intention by certainty of voting, 2001

	Con %	Lab %	Lib Dem %	Others %	Con lead %
18-22 Jan
All	31	50	14	5	-19
Certain to vote	34	46	15	5	-12
Difference	+3	-4	+1	0	+7
15-20 Feb
All	30	50	14	6	-20
Certain to vote	30	49	16	5	-19
Difference	0	-1	+2	-1	+1
22-27 Mar
All	31	50	14	5	-19
Certain to vote	34	46	16	4	-12
Difference	+3	-4	+2	-1	+7
19-24 Apr
All	30	50	13	7	-20
Certain to vote	31	48	13	8	-17
Difference	+1	-2	0	+1	+3
30 Apr-1 May
All	32	50	13	5	-18
Certain to vote	32	51	12	5	-19
Difference	0	+1	-1	0	-1

Source: MORI/The Times/The Sun

Even if Labour's vote is disproportionately reduced by a low turnout, will this cost the party many seats? Naturally, with the first-past-the-post electoral system, that all depends in which constituencies the Labour voters fail to turn out. All the poll indications have been that the fall in turnout will be worst in Labour's safe seats: in our April Times poll, for example, only 41% of adults in Labour's super-safe heartland seats (with a majority of 35% or more) were certain to vote, while in the Labour marginals (majority 10% or less) the figure was 55%. This is, after all, natural enough: the incentive to vote in a constituency where there is no realistic doubt about the result is clearly much lower than in a knife-edge marginal. If this is the case, then even supposing that Labour's vote share is cut by a differential turnout, the party would lose many fewer seats than a simple projection of the vote shares would suggest.

One element is so far missing from this analysis. We have been able to measure the differential effect by party support, and the differential effect by constituency type, but within the limits of a single poll it is not possible to combine the two because the sample size is too low. However, we have now been able to aggregate the four MORI/Times polls conducted this year, giving a combined sample size of over 4,000, and to look at the differential party effects of certainty to vote in different classifications of constituency. The results are fascinating, and suggest that even if the turnout is as low as current indications suggest, Labour may not suffer by it.

	Con %	Lab %	Lib Dem %	Others %	Con lead %
Conservative-held seats
All	40	40	18	2	0
Certain to vote	46	33	20	1	+13
Difference	+6	-7	+2	-1	+13
Labour marginals (majority 10%)
All	31	53	12	4	-22
Certain to vote	31	52	15	2	-21
Difference	0	-1	+3	-2	+1
Labour semi-safe (majority 10-35%)
All	27	58	8	7	-31
Certain to vote	26	60	8	6	-34
Difference	-1	+2	0	-1	-3
Labour super-safe (majority 35%+)
All	16	66	12	6	-50
Certain to vote	19	63	12	6	-44
Difference	+3	-3	0	0	+6

A word or two of caution. The sample sizes used here are still small, and the four surveys that were aggregated cover a three-month period, with two pointing to a significant party differential and two pointing to none. But accepting these reservations the pattern itself is clear enough. In the seats the Tories already hold, a low turnout will tighten their grip (many of these, of course, are rural seats that have suffered from Foot and Mouth Disease in recent weeks). In Labour's heartland seats, the Tories will gain ground in votes, but not nearly enough to make any unlikely gains. But in the marginals all parties will suffer equally from low turnout ? those who are most certain to vote are no different in their allegiances to those who are not. There are no sweeping gains to be had. This evidence is too weak to be conclusive, but it may be that the entire assumption which has been so widely accepted without question is, in fact, completely mistaken.

A low turnout would, of course, be bad for democracy; it is healthy for as many citizens as possible to participate in choosing their government. It would presumably be bad, too, for the government's prestige: a low turnout would imply a weakened popular mandate. But it now seems conceivable that it will not, after all, do any direct harm to Mr Blair's majority ? more bad news in a bad week for Mr Hague.