It’s All Just A Little Bit of History Repeating
With speculation growing of a new third party in the UK, what lessons are there from the time when the centrists in Labour actually did break away?
Labour swings to the left under a leader unloved by middle Britain — it could be now or it could be three decades ago. In 1981 senior figures from Labour’s centrist old guard were sufficiently aggrieved to walk out. Four MPs — Roy Jenkins, David Owen, Shirley Williams and Bill Rodgers — issued the so-called ‘Limehouse Declaration’ and founded the Social Democratic Party.
All had sat around the Cabinet table, all had high recognition and respect in the country. 28 Labour MPs followed them. Together they formed the ‘Alliance’ with the Liberal Party’s 9 MPs and together they marched to electoral disappointment: two years after the SDP’s foundation the Alliance came third, winning many votes but few seats — just 23 out of 633 contested.
The parties eventually merged and created the Liberal Democrats, who continue to this day. This failure to break Britain’s two-party system hangs over all who have since contemplated attempting the same. But if you were Chuka Umunna or Hilary Benn or any of the high-profile Corbyn critics, what could you take from the ‘Gang of Four’s’ great misadventure?
The British Election Study contacted almost 4,000 people in 1983, chosen at random to construct a representative sample of the electorate that year. This was the golden era of surveys — almost everyone asked chose to participate. The following findings come from this dataset and one of constituency results by Prof Ivor Crewe.
Demographics: A Less Class Party, not a Classless Party
As an old saying goes “Class is the basis of British party politics: all else is embellishment and detail” and in 1983 this was still the case for Labour and Conservatives. Occupational data showed the more working-class the occupation, the more likely people were to vote Labour; the more middle-class, the more likely to vote Conservative. The relationship was pronounced. Even more pronounced was class identity — many thought of themselves as a class to which they did not strictly belong based on their job. This group was much more likely to back the party associated with their class identification than their occupational class. A professional who identified as working class would probably vote Labour, a manual worker who felt middle class would probably vote Conservative. Britain’s parties were therefore fundamentally tied to the class system. Other relationships that have grown in importance since 1983, like the impact of younger voters being more likely to vote Labour, had not yet emerged.
Alliance voters rather defied this pattern. Their voters were slightly more likely to be middle class, yes, but the relationship was weak and even weaker among those who switched to them. The Liberal Party voters from the previous election that the Alliance inherited were somewhat middle class, but their new voters appear to have rounded out their class appeal a little. In any case, knowing what class a person fit into provides very little chance of correctly guessing whether they voted Alliance or not.
Alliance vote was also linked to education, but again it was only slightly predictive. The higher a respondent’s qualifications, the higher the likelihood of them voting Alliance. Among graduates — very few in number in 1983 — the Alliance were within the margin of error of being the party with the greatest support. That said, if you had met someone and they had said ‘I have a degree!’ after thinking that they were being very oddly forward you could not necessarily assume they were probably an Alliance voter, and certainly not in the way you could more safely assume a financier who told you they were middle class was a Conservative voter. There were in fact no social characteristics that could reliably mark someone out as an Alliance voter.
Politics: Splitting the Left
Ask a Labour MP what they think of the SDP and once they’ve stopped cursing they’ll accuse the party of ‘dividing the vote and letting the Tories in’. It’s a common refrain, and its echoes can be heard in the newspaper reports of the hand-wringing over whether to stage another breakaway.
Among those who told the election study they voted Alliance in 1983, 33% recalled having voted Labour in 1979, compared to 30% for the Liberals and only 19% for the Conservatives. The remainder said they did not vote, were not eligible or could not remember. So for every 2 Labour voters that switched to the Alliance, 1 Conservative also apparently did. There is no way of telling whether these voters might have switched without the Alliance (or indeed, how accurate their memories were), but it shows that they were able to win voters with affinities to both sides, albeit at different rates.
However, the election study also asked how voters would have voted if the voting system had asked them for a second-choice party. Here 42% of Alliance voters said their second choice was the Conservatives and only 35% the Labour Party. (12% say they would not have cast the vote.) How to reconcile this? One explanation put forward in the past is that the Alliance voters drew from those who were actively disaffected with the Labour Party after 1979 and its leftward swing. This is oversold as an argument, but the underlying point still holds: just because these voters used to support Labour it did not mean they would still have been closest to Labour in the event of there being no Alliance to vote for.
What of those voters’ views? Using an index of respondents’ answers on tax, the unemployed, redistribution, private healthcare and schooling, unions and big business, it is possible to rank people from left to right more objectively than if they are simply asked to place themselves on the spectrum. Each of the five groups in the graph below comprise about a fifth of the electorate, placed relative to each other based on their political views. The graph shows how each of these groups say they voted.

Two things are clear from the parties’ support by left-right position. Firstly, the Alliance’s support (and even more so, the switcher cohort) was drawn most from people on the moderate left. While the further left a person was, the more likely they were to vote Labour, and vice versa for the Conservatives, Alliance’s support was strongest just to the left of the middle. It drops off at the far-left. This trend is even more pronounced in other, more complicated ways of presenting the data, but the graph above serves to illustrate the point.
Secondly, the Conservatives were remarkably strong across the board. They won more support from people on the left than either or both of the other parties won from people on the right. They were also the largest single party in the centre.
The result depended as much on the strength of the right as the division of the left therefore. And the best working hypothesis for why the right was strong is a Thatcher effect.
What went wrong for the SDP/Liberals 1: Thatcher
Margaret Thatcher dominated the 1983 election. 99% of Conservative voters rated her as a “very effective” or “fairly effective” Prime Minister, along with 90% of Alliance voters and 76% of Labour. Whether voters liked her or loathed her, virtually every respondent rated her. Only 29% of the population said the same about Labour’s own leader Michael Foot. Labour voters even rated Margaret Thatcher more “effective” than the leader of the party they supported. In head-to-heads both more Alliance and more Labour voters also thought she was ‘most likely to get things done’ and ‘would improve British status abroad’ than thought so of their own leaders.
Among both Labour and Alliance voters, Thatcher was more likely than their own leaders to be described with positive qualities, including “determined”, “shrewd”, “tough”, “principled” , and “decisive”. These were not small leads either. Foot, Steel of the Liberals and Jenkins of the SDP were by contrast ahead on “likeable”, “caring” and “listens to reason”.
These sorts of leads are hard to imagine today: May or even Cameron being rated more highly by Labour voters than Corbyn, Miliband or Brown? Tony Blair might be the only exception, but he is now almost as much a part of history as 1983 is.
The Alliance suffered in two ways therefore. On many measures, their supporters preferred or substantially respected the leader of the Conservative Party. If their supporters did, so presumably did many people who thought about supporting them and decided not to. Their joint leadership may also have been a problem. On every question asked, David Steel of the Liberals received more positive responses than Roy Jenkins of the SDP. While Steel was sometimes only just behind Thatcher, Jenkins was usually only just ahead of Foot. The prospect of Jenkins possibly becoming PM could not have helped their appeal.
What went wrong 2: Stakes and Stances
If the only people to vote were those who didn’t care who won, the Alliance would have been neck-and-neck with Labour, and the Tories in a distant third. Alliance voters showed little antipathy to Labour or the Tories in questions about their view of the parties they didn’t vote for. They were, by nature, moderate. This not always a motivating force.
More than that, they did not rate their party. On defence, on unemployment, on inflation, on welfare, on nationalisation, on law and order — in short, on all the big issues asked about, fewer Alliance voters felt close to the Alliance than either of the big two’s supporters felt to their chosen party. While the crossover appeal was in none of these categories high enough to make Alliance voters prefer the policy of another party (in the way they appear on some dimensions to have rated Margaret Thatcher more highly than their own leader) it still shows Alliance voters to be unrooted to the party they eventually supported. After all, based on the evidence above of the left-right positioning of Alliance voters, the party was actually close to these voters. But for some reason they did not feel this. Each of the other parties’ voters by contrast did feel their party of choice represented their views well.
Outright crossover happened on Alliance’s ability to govern. More Alliance voters trusted the Tories to keep prices down and prevent strikes and believed Labour could best reduce unemployment and properly provide welfare services than in any of these cases trusted the Alliance. On all of those questions Alliance voters were more likely to rate other parties. Again, the other parties’ voters had their evaluations of parties and their vote on balance corresponding. Only the Alliance’s voters preferred other parties.
The lack of support for Alliance’s policy and governance among even their own voters went with a general lack of support. There were few Tory or Labour voters (or non-voters) casting an eye over the Alliance and evidently thinking their team could do the best job. On every one of these policy and governance questions they came third across the whole electorate. How it happened — poor communications, the novelty of the party, lack of faith in the SDP leader — is impossible to tell, but what is clear is that Alliance’s positive appeal on competence and policy was very, very limited. That they won as many votes as they did without such appeal is remarkable. It shows they were able to appeal in other ways. But this kind of policy and competence appeal is a core part of how political parties win support in democracies. It is highly plausible to suggest that if that sense of being represented by the Alliance programme and having faith in them to deliver it had been more widespread, their share of the vote would have been much, much higher.
What they could have won: Modelling Seats
Looking back, how close did the third-party challenge come? Between them the Liberals and SDP did manage to win 7.8m votes (25.4% UK-wide). By this measure they were close behind Labour, on 8.5m (27.6% UK-wide). Labour however came first in 209 constituencies while the Alliance only did so in 23.
Britain’s voting system, using constituencies, rewards not just how many votes a party gets as a whole, but how they are spread. Those parties who have local majorities gain more seats than other parties — like the Alliance — who win plenty of votes but come second almost everywhere.
A coda to this however is that in many seats these concentrated parties stack up piles of votes more than they need: an inner city Labour MP or a Conservative MP in a leafy shire may both sit on vote tallies of 70% or more. A party with a more even spread of votes might need to get its national tally up rather higher to win 40 or 50 seats than a party whose votes are concentrated, but should require fewer votes to win sweeping majorities.
Three models are shown below for how the Alliance may have fared in terms of seats if they had won more votes. Each estimates in how many seats Alliance would have been the favourite to win, under certain assumptions. This is not exactly the same as them winning this number, but it will serve here for illustration, and is the same shortcut used by the infamous BBC swingometer over many decades.
Model 1 assumes in each constituency they won votes from Conservatives and Labour according to the two parties’ national strength. This means for every additional 5 Alliance voters, 3 would come from the Conservatives and 2 from Labour. This also follows the data in the election study that Conservative and Labour voters were approximately equally likely to rate the Alliance positively, despite not voting for them, and equally likely to give the Alliance as their second choice after the party they actually did support.
Model 2 assumes in each constituency they won votes from Conservatives and Labour according to the two parties’ local strength. This would also mean 3 Conservative converts for every 2 Labour across Great Britain, but varies the amount so that in Labour areas more votes are being won from Labour and in Conservative areas more votes are being won from Conservative.
Model 3 assumes in each constituency the Alliance won votes based on the rate given in the BES for those who switched to the party between 1979 and 1983 — i.e. around 2 ex-Labour for every 1 ex-Conservative.

Under any of these models, Alliance had a hard task of winning a majority: the swingometer suggests needing a vote share 15–18% higher. At 40–43%, this is a level that has resulted in landslides for Labour and the Conservatives in other elections. Even just pulling ahead in enough constituencies to become the largest party required an additional 14–16%. Even with the earlier methodological note about issues with the swingometer taken into account, this is a far cry from Labour in 2005, who secured a working majority of over 60 seats on just 36% of the England, Wales and Scotland vote. At such a share the Alliance would only have barely overtaken Labour, and be coming second almost everywhere they did not win.
Having said that, by the time they had reached the threshold of an overall majority every additional 1% of the vote would have be yielding 40–60 more constituencies as probable Alliance wins. At the 51% the Alliance briefly polled, they would have been favourites in around 600 of the 633 constituencies.
Conclusions
A bad general is one ready to fight the last war. Place too much emphasis on fixing the mistakes that were made last time and a leader will almost certainly make new ones. Some factors are clearly the same, chief among them the difficulties of the electoral system. A new party with a broad geographic distribution of support would be severely handicapped. It would need to either develop some root in a demographic that varied substantially by area, or to achieve extremely high vote shares in the first outing. One strategy might be to be a ‘spoiler’, denying either other party an overall majority and forming a coalition on the condition of the introduction of proportional representation. Moderate left parties have not however been doing very well in recent elections in Western Europe, and electoral reform might provide as much or more of a boost to other parties, including new entrants.
Some of the contributing factors to the Alliance failure are now however different. The Lib Dems are today still seen as untouchable by many would-be voters after their 2010–15 coalition with the Conservatives and subsequent general electoral malaise. They might not be able to play the same role in a new Alliance as the Liberals did, bringing a pre-existing bloc of votes. That said, when the SDP was founded the Liberals had just 9 seats and were still recovering in the public’s eyes from ‘A Very English Scandal’. The Alliance may have helped rehabilitate them.
More indisputably, polls today show low ratings for both May and Corbyn. Instead of one unpopular leader and one effective one, as in 1983, there are two Michael Foots. Corbyn appears to be fanatically loved by a significant segment of the population, enough to populate rallies and social media, but when the population as a whole are asked about whether they would like him to be Prime Minister they are decidedly lukewarm: almost as lukewarm as they are with Theresa May.
Brexit also provides an issue on which a new party could put themselves on a different side to both the Conservatives and Labour. Significantly, it is also seen as an important one. Whereas on every issue except proportional representation the Alliance in 1983 were in the space between the Conservatives and Labour, a party that wished to cancel Brexit would have a position supported by almost half the country and which would differentiate them from both of the two largest parties. Unlike Alliance in 1983, a new party might be seen as best placed to handle the issue that voters consider to be important. But will that still be the case if the next election is in 2022, after Brexit is a done deal and transition periods are over? And given that Alliance was not able to persuade voters that the Liberal/SDP platform was the closest to their views even though for many it was, would any new party face the same problem in conveying their message?
Looking back with hindsight, Alliance voters were a group who did not care much about the result and who chose to back a party they mostly didn’t believe were best on the issues or governing potential. That is a strange group on which to build a support base. It made for a shapeless protest rather than a paradigm shift. Labour MPs and would-be new party leaders looking for guidance in their own decision have plenty to learn from the SDP — and much they would have to avoid.
