The 2010 general election result was considerably biased in Labour’s favour: if they and Conservatives had won equal shares of the vote total, Labour could have obtained as many as 54 more seats than their Tory opponents. This bias partly reflected unequal electorates across the country’s constituencies. Recently published data show that the number of registered electors nationally has since declined. But is Labour’s advantage still there? Ron Johnston, Charles Pattie and David Rossiter analyse those data and show that, unless the Conservatives win a lot of seats from Labour on 7 May, if the two parties are roughly equal in their number of votes Labour could again benefit from the inherent biases in the electoral system, perhaps by as many as 30 seats.
All UK general election results since the 1970s have been biased, favouring Labour over the Conservatives – bias being defined as the difference in the number of seats each would have gained if they had equal shares of the votes cast. If that had occurred in 2010 – with votes distributed across Britain’s constituencies in the same proportions as the votes actually cast – Labour would have obtained 54 more seats than the Conservatives.
Several factors create this pro-Labour bias; the most consistent have been differences between constituencies won by the two parties in their average electorates and turnout rates. Small constituencies can be won by fewer votes than large ones; so can those with low turnouts compared to those with high. The mean electorate in Conservative-won seats was 72,304 in 2010, but 68,672 in those won by Labour; average turnout in those two groups of seats was 68.2 and 61.2% respectively. The former difference was worth 18 seats to Labour in the total bias of 54; the latter was worth 31 seats.
The Conservatives tried to remove the impact of differences in average electorates: the 2011 Parliamentary Voting System and Constituencies Act required all constituencies to have electorates within 5% of the national average by the time of the 2015 general election, and the Boundary Commissions’ revised recommendations for new seats applying this rule would have removed any pro-Labour bias. But the redistribution was aborted, the Liberal Democrats voting with Labour and against their coalition partners to delay the redistribution until 2016, in retaliation for the lack of progress on House of Lords reform.
But has that difference in mean electorates been reduced, if not eliminated, by changes since 2010 in the distribution of the electorate across Britain’s 650 constituencies? Labour’s advantage over the Conservatives was a consequence of:
- Smaller constituencies on average in Scotland and Wales (65,234 and 58,627 electors respectively) – where Labour won 67 seats and the Conservatives only 9 – compared to England (average 71,918), where the Conservatives won 297 seats to Labour’s 191;
- A decline since the constituency boundaries were defined – using data for 2000 in England and Wales, and 2004 in Scotland – in the average electorate in seats won by Labour (most of which are in urban areas) compared to those won by the Conservatives.
In general, Labour won the smaller constituencies and those with declining electorates: they needed fewer votes to win there than did the Conservatives in the larger constituencies and those with expanding electorates.
As the 2015 election is to be fought in the same constituencies as 2010, these differences presumably remain in place – and might even be exaggerated, thereby enhancing Labour’s advantage – which could be crucial in determining the largest party in a close-run election. But has there been any clear pattern of change over the five years? The Office of National Statistics recently published the number of registered electors in each constituency in December 2014 (except that the Scottish data will not be available until May 2015). These will not be the final figures at the 2015 election, because enrolment is open until mid-April, but comparing them with those for December 2009 (before the 2010 election) provides insights on trends since then. (For Scotland, we have had to use the 2013 data.)
Across Britain, despite overall population growth in recent years, the average constituency electorate declined by 228 individuals – in part because a large number of people have moved home but not registered at their new address (especially young people who were registered as students but have since graduated and moved away): others qualify to vote but have not registered (again, many of these are probably young people). The Electoral Commission estimates that because of these patterns there as many as 1 million new ‘missing voters’, joining the several million who were not registered before 2010.
The first graph shows a very strong correlation between each constituency’s electorate in 2009 and 2014 – the overall pattern of constituency sizes has not changed – but with one very clear variation: average electorates declined in both England and Wales (by 558 and 888 respectively) but increased by 2,669 electors in Scotland (no doubt reflecting Scots’ keenness to vote in the 2014 Independence Referendum).
There were considerable variations around these averages, however: 286 constituencies experienced an increase, 158 of them by more than 1,000 electors; 346 experienced a decline – 213 of them by more than 1,000 electors and 96 by more than 2,500. Have the declines been concentrated in Labour-held seats, thus increasing their advantage over the Conservatives? Or has the recent population growth in many UK cities diluted the pro-Labour bias?
The answers – as illustrated in the second diagram – are yes, but only slightly to the first question, and thus no to the second. Only constituencies won by the SNP in 2010 have, on average, increased in size. The mean electorate in Conservative-held seats declined by 224 between 2009 and 2014, compared to 1,179 in Labour-held seats (despite the growth in Scotland where it holds 41 seats). The difference between the two parties’ mean electorates was 4,016 in 2009; in 2010 it was 4,101. Thus if the Conservatives and Labour each won the same seats in 2015 as 2010, Labour could anticipate a favourable bias of some 18-20 seats if the parties have near-equal vote shares because of this factor alone.
That is an unlikely outcome, of course. Labour’s initial strategy for 2015 targeted 106 seats. If it won them all, and all other seats stayed with their 2010 winner, the average Labour constituency in 2014 would have 68,098 electors and the average Conservative constituency 72,810 – the gap would be 4,712 electors, and the pro-Labour bias probably larger than five years ago. (The 106 seats that Labour would win – 89 of them from the Conservatives, 12 from the Liberal Democrats, 4 from Nationalist parties and one from the Greens – had an average electorate in December 2014 of 68,682.) On the other hand, the 40 seats that the Conservatives have targeted as potential gains – 32 from Labour and 8 from the Liberal Democrats – averaged electorates of 67,475 in 2014. If all were won, the average electorate in Labour-won seats would be 68,112, whereas in Conservative-won seats it would be 71,442, a slightly smaller gap between the two of only 3,330: there would still be a pro-Labour bias, but reduced because some smaller constituencies had crossed into the Tory camp.
The marginal seats on average have smaller electorates than those that are relatively safe for the two parties, therefore. The more of them that the Conservatives win, the smaller the gap between each party’s mean electorate and the smaller the likely pro-Labour bias in the outcome.
One other scenario worth exploration concerns Scotland, where the average electorate increased after 2009. In 2010 Labour won 41 seats there, the Liberal Democrats 11, the SNP 6 and the Conservatives 1. Some commentators suggest that the SNP might win most of the Scottish seats. If, to take the extreme case, the SNP won all 59, the average electorate in England and Wales would be 67,381 for Labour and 71,795 for the Conservatives. Labour would still have an advantage over the Conservatives in the translation of votes into seats should the two parties get approximately the same number of votes overall.
How about turnout variations? The average in 2010 was 61.2 and 68.2% in Labour- and Conservative-held seats respectively. In Labour’s 106 target seats it was 66.3 whereas in the Conservatives’ it was 64.9; if Labour won all of its targets, turnout in 2015 – if the 2010 pattern is replicated – where it won would average 62.7% whereas in the remaining seats in Conservative hands it would be 68.8%. If the Conservatives won all of their targets, turnout in all of its seats would average 67.8%, whereas in those retained by Labour it would be 60.9.
Once again, the conclusion is clear – Labour would be advantaged by the same pattern of turnout differentials across the constituencies in 2015 as in 2010 (even if the SNP won all of Scotland’s seats, when the average turnout would be 60.9% and 68.2% in Labour- and Conservative-held seats respectively in England and Wales).
Labour had a considerable advantage over the Conservatives in 2010 – as at previous elections – because its seats had fewer electors on average. (Which is not to deny that some Labour-held seats have large electorates: two of the biggest in 2014 were Manchester Central and Ilford South.) That situation will not change markedly in 2015, unless the Conservatives win a large number of Labour-held marginals. Turnout differences gave Labour a further – and more substantial – advantage over its main rival in 2010, and that too is unlikely to change markedly in 2015.
In conclusion if, as all the opinion polls suggest, the two parties are close in their vote shares on 7 May, Labour could get as many as 30 more seats than the Conservatives (with the size of that gap dependent on the outcome in Scotland). This could be sufficient to make Labour the largest party, giving Ed Miliband the first attempt to form a government – even if Labour came only second in the vote tally. Such an outcome is almost certain because of the lower turnout in Labour seats. The Conservatives’ failure to get the differences in constituency size changed, because the creation of new constituencies was aborted in 2013, makes Labour’s advantage even more certain.
Ron Johnston is Professor of Geography at the University of Bristol.
Charles Pattie is Professor of Geography at the University of Sheffield, specialising in electoral geography.
David Rossiter has worked in a research capacity at the Universities of Sheffield, Oxford, Bristol, Leeds and Essex. He has been involved in the redistricting process both as academic observer (for example The Boundary Commissions, MUP, 1999) and as advisor to the Liberal Democrats at the time of the Fourth Periodic Review.
There are implications for electoral reform and the choice of a new voting system.
Among other factors, proportionality should not depend on variations in constituency size or turnout, or for the geographical concentration, or otherwise, of any one party’s supporters.
The parliamentary votes that a Party has must be proportional to the total number of votes it polls across the nation. Every vote must count, and must be of equal significance. We should not choose a system that could allow local circumstances to result in national anomalies, or depends on frequent and contentious boundary revision.