The 2017 General Election: Volatile Voting, Random Results

Electoral Reform Society,

Posted on the 21st August 2017


From Darren Hughes, Chief Executive

Darren HughesThe 2017 General Election was the third strike for the First Past the Post voting system.

From producing a hung Parliament in 2010 – something not meant to happen under Westminster’s winner-takes-all voting system – to a slim majority in 2015, the way we elect our House of Commons isn’t doing the one thing it was claimed to be good for – delivering decisive results.

This June’s election outcome throws up yet more questions about the legitimacy of our voting system. Our report shows how far from being ‘strong and stable’, First Past the Post is failing to deliver for the public. With one in five voters trying to second-guess each other by opting for ‘lesser evils’, we are left with a lottery election where casting a ballot is like casting a die.

Not only have the last three elections either produced hung parliaments or results so unrepresentative they demean the electoral process (2015 was the most disproportionate in British history), the last two have seen the highest ‘voter volatility’ since 1931.

Our voting system is failing to keep up and is undermining the faith voters have that seats in Parliament will reflect the votes they cast. This lottery approach to running elections means we have no idea what will happen or how votes will be reflected in our elected Commons.

In the nations of the UK there’s much to reflect on. As in 2015, the first-placed party was different in every nation. For both the Conservatives and Labour in Scotland, their revivals are still not being reflected properly in seats. And these revivals follow difficult years, where both parties were kept alive by proportional voting systems at Holyrood and for Scottish local elections.

In Wales, Labour continue to be over-represented to the detriment of other parties, from the Conservatives to the Liberal Democrats. And in Northern Ireland, voters familiar with being able to vote for a wide range of parties ended up forced into two party camps.

This report gives pause for thought for all sides of politics. For the Conservatives there is the dubious distinction of having put on substantially more votes but actually losing seat share and with it their majority in the House of Commons. Indeed, the Conservatives have not been delivered a strong majority under FPTP for thirty years. One has to look back to the 1987 General Election for such a majority, despite strong vote shares in 1992 and this year.

For Labour, it is the brutal reality that so many of their votes did not contribute in any way to the size of the Parliamentary Labour Party – indeed it is only Wales that is keeping Labour ‘fairly’ represented in Westminster. This is because in many of the seats Labour won, MPs have a majority of tens of thousands – when only a single vote is needed. In many safe Conservative seats, a Labour surge led to zero increase in representation.

In the end, we have a system that recognises the geographical location of a voter and nothing else. It is where voters are – rather than their choices – that matters. This must change if we are to restore legitimacy to our political institutions.

But the real question for our politicians is this: if the two main parties can gain over 80 percent of the vote for the first time in decades, in a system designed for two parties, and yet both still lose – when will they show the leadership the country so desperately needs and fix our voting system?


With thanks to Doug Cowan, Charley Jarrett, Emma Levin, Josiah Mortimer and Jessica Blair for their contributions to the report.

The third strike for First Past the Post

Coalition, slim majority, minority

By Jess Garland and Chris Terry

On 26th June 2017, a full 17 days after the results of the general election were declared, a deal was finally agreed between the minority Conservative government and the Democratic Unionist Party to enable the government, on a case by case basis, to get its legislation through parliament.

The results of the 2017 general election, which saw the Conservatives reduced to 318 seats despite a 5.5 percent increase in their vote share, were realised under a system designed to deliver stable, single-party governments. (We include the Speaker of the House in this figure to be consistent with previous reports and general usage. The Speaker in the House of Commons renounces their party affiliation on taking the post and only votes in the case of a tie-break.)

On 42.4 percent, the Conservatives had not only increased their vote share (up from 36.9 percent in 2015), they had achieved the same vote share as in 1983 – a year which saw a landslide 397 Conservative MPs elected.

And yet, the Prime Minister returned to parliament having lost her majority whilst the Labour opposition drafted an alternative Queen’s Speech. First Past the Post had delivered the country neither a decisive outcome nor a stable government.

The volatility of this supposedly ‘strong and stable’ electoral system has been exposed in the last three general elections. In 2010 First Past the Post delivered us a coalition government, the first since 1945, under a system designed to produce single-party majorities. In 2015, First Past the Post gave us the most disproportionate election to date with a majority government secured by under 37 percent of the vote share.

Now, in 2017, despite over 80 percent of votes going to just two parties (the highest combined vote share since 1970), First Past the Post could not deliver a majority government. The 2017 general election was the third strike for First Past the Post – it’s out.

Figure 1: 2017 General Election results

Party Vote % Vote % change Seats Seats
Seats % Seat % change
Conservative 42.4 5.5 318 -13 48.9 -2
Labour 40 9.5 262 30 40.3 4.6
SNP 3 -1.7 35 -21 5.4 -3.2
Lib Dem 7.4 -0.5 12 4 1.8 0.6
DUP 0.9 0.3 10 2 1.5 0.3
Sinn Féin 0.7 0.2 7 3 1.1 0.5
Plaid Cymru 0.5 -0.1 4 1 0.6 0.2
Green Party 1.6 -2.1 1 0.2
UKIP 1.8 -10.8 0 -1 -0.2
SDLP 0.3 0 -3 -0.5
UUP 0.3 -0.1 0 -2 -0.3
Others 1 n/a 1 0.2 n/a

(We include the Speaker of the House in the Conservative figure to be consistent with previous reports and general usage. Lady Sylvia Hermon was reelected in North Down as an Independent Unionist)

While Labour achieved a nearly proportional result, getting just over 40 percent of the seats on 40 percent of the vote, the Conservatives took a skewed seat-share, with just under 49 percent of the seats on 42 percent of the vote.

Yet other parties that have traditionally been marginalised in this system continued to be underrepresented. The Liberal Democrats’ 7.4 percent of the vote translated into just 12 seats (less than 2 percent of seat share) and the Greens only retained their one seat despite attracting over half a million votes – the largest votes per MP ratio. UKIP also attracted over half a million votes but no MPs in return.

Figure 2: Vote and seat percentage change

The great divider

As with the 2015 general election, the system has thrown up electoral injustices and not just for the smaller parties. Looking at the results by nation we can see the geographical anomalies.

Whilst the two-party squeeze led to a broadly ‘representative’ result UK-wide, Labour lost out in Scotland with just seven MPs for over a quarter of the vote (27.1%) whilst the Conservatives on a similar vote share (28.6%) returned 13 Scottish MPs to Westminster. In Wales, Labour’s 48.9 percent of the vote delivered them 70 percent of the available seats; the Conservatives just eight seats (20%) for their 33.6 percent.

Not only does First Past the Post over-represent parties whose vote is geographically concentrated within constituencies, it exaggerates our regional and national differences. The 2015 general election saw, for the first time, different parties gaining the most seats in each of our four nations. This trend has continued with the Conservatives in England, SNP in Scotland, Labour in Wales and Democratic Unionist Party in Northern Ireland gaining the most seats in 2017.

Most of all however, this lottery election was marked out by the unpredictability of results. In the space of two years, parties’ fortunes have fluctuated hugely. In Glasgow North East for example, a 43.9 percent increase in vote share for the SNP in 2015 switched to a 9.2 percent increase in vote share for Labour in 2017. We are witnessing huge changes in partisan alignment and our system is struggling to keep up. This election saw the second highest aggregate level volatility (the movement of votes between the parties, aggregate volatility is measured as the combined change in vote shares for each of the parties divided by two.) since 1931. The most volatile was 2015.

The 2017 general election also saw an increase in very marginal seats. Eleven seats were won by less than 100 votes. North East Fife was held by the SNP by just two votes. Such are the vagaries of the system that the Conservatives could have won an absolute majority on the basis of just 533 extra votes in the nine most marginal constituencies. A working majority could have been achieved on just 75 additional votes in the right places. Two very different outcomes based on less than 0.0017 percent of voters choosing differently.

And yet despite the increase in very marginal seats, for many the election was business as usual. Though several high profile seats changed hands (Sheffield Hallam, Moray, Gordon) only 99 of 650 seats actually elected a new representative (12 of whom were former MPs), and only 70 seats (10.8%) changed party hands in spite of significant volatility. Many seats saw massively increased majorities for the incumbent, meaning for many voters the place in which they vote is ever more unlikely to be represented by a different party.

Voters in key marginals experience a very different election to the rest of the country. Election spending is always forced by the system into a handful of marginal constituencies but this year saw this activity move more significantly online. Micro-targeting strategies were employed by parties this election to ensure even greater focus on key seats.

Parties have always targeted voters in key marginals with increasingly sophisticated information, but this election saw voter targeting move into the social media age. Thirty-five percent of 18 to 25 year olds in our survey had been contacted about the election on Facebook (BMG Polling on behalf of Electoral Reform Society, Fieldwork: 26th – 29th May 2017, Sample: 2,016 GB adults aged 18+) with less than two weeks to go before polling day, while only eight percent of this age group recalled having been canvassed on the doorstep, and six percent were canvassed in the street. Whilst social media is becoming an increasingly important battleground for elections, First Past the Post is ensuring that the battle is still being played out in only the most marginal constituencies.

By placing electoral outcomes in the hands of a small number of voters in a few select places, the electoral system is creating an ever more unpredictable electoral environment. With a volatile and fragmenting electorate, First Past the Post (an electoral system designed to produce single-party majorities in a two-party system) is failing on its own terms.

Westminster Archipelago

No return to two-party politics

Tactical voting

Despite Labour and the Conservatives gaining over 80 percent (82.4%) of the vote share between them, a look under the surface shows this was no return of two-party politics. Our research suggests that voters did not flock back to the two largest parties with enthusiasm. Millions of voters planned to vote tactically this election, with twenty percent saying they would be choosing the candidate that was most likely to beat the one they disliked (ibid). This is over double the proportion who said they would do so in 2015. Projecting this onto actual turnout would equate to nearly 6,500,000 people voting tactically.

Votes for the Liberal Democrats, Green Party and UKIP fell off significantly at this election; an election held just two years after the last. In 2015 the Lib Dems came second place in 45 of the 57 constituencies that elected a Liberal Democrat MP in 2010. In 2017 despite winning 11 of those 57 seats, they came second place in only 26 (House of Commons research note (2017) ‘General Election 2017: full results and analysis’, 13 July 2017.).

Overall, the number of seats in which the Liberal Democrats came second dropped to 38 in 2017, from 63 in 2015. This suggests the Liberal Democrat vote was not only used efficiently but that large numbers of traditionally Lib Dem voters were voting tactically.

In Bristol West where the Labour Party increased their vote share by over 30 percent, the Green Party saw a 14 percent decrease in vote share, despite this being their number one target seat and having come a close second place in 2015. The Liberal Democrat vote share in this seat dropped by 11.6 percent. In York Central, a seat which saw the second highest increase in vote share for Labour (22.8%), no UKIP or Green candidate stood despite these parties coming third and fourth in 2015. And in Clacton, the Conservatives’ third highest increase in vote share (24.6%) saw UKIP lose its one seat with a 36.8 percent drop in vote share.

These huge swings in support towards Labour and the Conservatives in seats where just two years ago voters came out in significant numbers for other parties suggests a considerable amount of tactical voting (Green Party, 2017. Greens announce final candidate numbers).

This election also saw the return of vote swapping and a growth in tactical voting sites: an entire industry trying to give voters a sense of worth in a system that routinely wastes their democratic choices. The SwapMyVote site – a matchmaking site for disillusioned electors – again offered voters the chance to be paired up with voters of a different hue in another constituency (one where their choice would actually matter).

In addition to vote swapping, there were a range of online tools to assist with tactical vote choices, some based around progressive alliances, some advising on tactical votes along Remain and Leave lines.

Taking the choice away

It was not only voters who were corseted by the system – parties also felt the restrictions of First Past the Post. The Greens, Liberal Democrats and UKIP entered electoral pacts in order to achieve their goals. The Green party didn’t stand candidates in 183 of 650 constituencies (up from 77 contests that did not feature a Green Party candidate in 2015). The party confirmed that at least 22 candidates had stood aside ‘to increase the chance of a progressive candidate beating the Conservatives’.

The Liberal Democrats also entered into ‘progressive alliance’ arrangements with the Green Party; 42 seats featured ‘progressive alliance’ arrangements in which one or other party stood down.

UKIP didn’t stand candidates in 272 seats compared to only 26 seats that were not contested by a UKIP candidate in 2015. UKIP leader Paul Nuttall said he would put ‘country before party’ in not opposing pro-Brexit MPs.

A vote-wasting machine

Under First Past the Post, one person takes all. This means votes for candidates that don’t come first, and votes for the winning candidate over and above what they need to win, go to waste.

Over 22 million votes (68%) were wasted this election. Five constituencies saw over 90 percent of the vote making no difference to the outcome (Manchester Gorton, Liverpool Walton, Knowsley, Liverpool Riverside, Liverpool West Derby).

Figure 3: Top ten largest number of wasted votes

The increasing geographical concentration of votes means many more votes going to waste. This is particularly an issue for the Labour Party who hold 34 of the 35 safest seats. Overall, 3,515,872 Labour votes in Labour winning constituencies did not go towards electing the Parliamentary Labour Party (an average of 13,419 wasted Labour votes per seat).

The five constituencies with the largest number of wasted votes (Bristol West, Isle of Wight, Hornsey & Wood Green, Bethnal Green & Bow, West Ham) wasted in excess of 50,000 votes per constituency.
Another way to look at the waste of votes in First Past the Post elections is to measure votes for non-elected candidates. These votes are completely disregarded in a winner-takes-all contest.

Figure 4: Wasted votes by region and nation

Across all 650 constituencies 44.12 percent of votes went on non-elected candidates – that’s over 14 million voters whose choice was not reflected in the outcome.

Figure 5: Top ten largest number of votes going to non-elected candidates

Divisive system, divided country

New divides

The UK entered the general election looking very much like a country divided. Following the referendum on the UK’s membership of the European Union, the Remain and Leave sides of the argument seemed irreparably split and this divide was set to shape the 2017 general election. In Scotland the issue of independence, so important during the 2015 campaign, was still shaping the debate and now had a new EU dimension.

The EU and Scottish independence referendums have cast British politics in a new frame. Though the referendums have in many ways simply highlighted existing polarised opinions on these issues, they have raised these divisions to the fore. These emerging political divisions suggest a realignment of British politics. Cultural and identity issues are shaping voting preferences and parties are struggling to hold together the internal contradictions thrown up by these new political cleavages.

According to the British Election Study panel data, only 56 percent of Remainers and 56 percent of Leavers intended to vote for the same party as they did in 2015. On issues such as globalisation and immigration, Remainers and Leavers from different parties are more closely aligned with each other than they are within their respective parties. Going into the ‘Brexit election’ these new divides were always going to have an impact but the question was whether the general election could find a way through, healing the sense of division. What emerged was an electorate apparently defined by new and more deep divides.

Age is becoming a significant divide in British politics. Amongst first time voters aged 18 to 19, Labour are 47 percentage points ahead of the Conservatives. Amongst the over seventies, the Conservatives are 50 percentage points ahead of Labour. Alongside this, as would be expected, a divide in terms of employment status with retirees and students following a similar pattern (Conservatives ahead by 39 points amongst retired voters and Labour 45 points ahead amongst students) (ibid). In addition to age and employment, educational attainment represents a significant political divide with the Conservatives leading amongst voters with lower educational qualifications (22 percentage points) and Labour leading amongst those with higher educational qualifications (17 percentage points) (ibid). Geography is also key with Labour advancing in urban metropolitan areas and the Conservatives in suburban, post-industrial and coastal towns; a geographical division accentuated by cultural divides (See Will Jennings and Gerry Stoker (2016) The Bifurcation of Politics: Two Englands, The Political Quarterly, Vol. 87 (3).).

These new dividing lines are creating unexpected outcomes. The Conservatives increased their vote share in traditional Labour seats such as North East Derbyshire, Stoke-on-Trent South and Mansfield (which had not elected a Conservative MP at any point since its creation in 1885). Whilst Labour advanced in wealthier city constituencies such as Kensington (held now for the first time ever by Labour) and Battersea. In Canterbury, Labour’s election win ended a record 185-year Conservative run in the seat. The old electoral rules no longer apply.

Our politics is being shaped by new political cleavages, new political divides that reflect the concerns, cultures and identities of 21st century voters. Yet our 19th century voting system cannot fairly represent and give voice to this range of voter preferences.

National divides

Our 2015 General Election report highlighted the geographic polarisation of politics in the UK. We argued that these divides, exaggerated by our voting system, made the UK appear more politically divided than it actually is. Many of these trends have continued, and some new dynamics have emerged at this election.

The 2017 election has played out differently in each nation. In England and Northern Ireland multi-party politics is being forced into a two-party shape. Yet in Scotland multi-party politics is breaking through single party hegemony, while in Wales, one party continues to dominate. Few of these outcomes reflect how people actually want to be represented.


Two-party polarisation is most evident in England with parties other than the Conservatives and Labour sharing just 1.7 percent of the available English constituency seats. Yet these parties gained a total of 12.6 percent of the vote in England.

The discrepancy between votes and seats largely benefitted the Conservatives whose 45.6 percent vote share translated into 55.7 percent of the seats. UKIP’s vote share fell the most in England, down 12 percentage points.

Figure 6: Election results in England

Party Vote % Vote % change Seats Seat change Seats % Seat % change
Conservative 45.60% 4.6 297 -22 55.70% -4.1
Labour 41.90% 10.3 227 21 42.60% 3.9
Lib Dem 7.80% -0.4 8 2 1.50% 0.4
Green Party 1.90% -2.3 1 0.20%
UKIP 2.10% -12 0 -1 -0.2
Others 0.80% n/a 0 n/a n/a n/a

Despite a shift in political divides across the UK, with parties making gains in unexpected areas, voters’ choices are still being marginalised because of where they live.

In South West England, nearly 15 percent of the vote share returned the Liberal Democrats only one MP out of the 55 representing that region (less than 2% seat share). Labour’s historical under-representation in the South East continued despite gaining 4 seats this election. The party won 28.6 percent of votes in the South East and just 8 seats (9.5% seat share). In the North East the Conservatives increased their vote share by 9.1 percent (up to 34.4%) and yet only retained the three seats they held in 2015 (10.3% of seat share).

Figure 7: English Regional Conservative and Labour vote share vs seat share

Northern Ireland

Northern Ireland’s multi-party politics was squeezed wholly into a two-party shape in 2017. Over a third of voters in Northern Ireland voted for parties other than the Democratic Unionist Party and Sinn Féin – and yet all but one of the seats went to these two parties. Comparing these results to Assembly elections in Northern Ireland it is clear that multi-party politics is being forced into a two-party competition under First Past the Post.

Northern Ireland Assembly elections held earlier this year using the Single Transferable Vote (STV) saw no fewer than eight parties (and one independent) elected. All multi-member constituencies are represented by more than one party, with three or four different parties elected in most constituencies. Belfast South elected five Assembly Members from five different parties; a range of views now represented by just one MP in Westminster. The spread of voting preferences in this constituency saw the previous MP elected on the lowest winning vote share in history (24.5%) in 2015.

Figure 8: Election results in Northern Ireland

Party Vote % Vote % change Seats Seat change Seats % Seat % change
DUP 36.00% 10.3 10 2 55.60% 11.1
Sinn Féin 29.40% 4.9 7 3 38.90% 16.7
SDLP 11.70% -2.2 0 -3 -16.7
UUP 10.30% -5.8 0 -2 -11.1
Alliance 7.90% -0.6 0
Others 4.60% n/a 1 5.60%

This range of voting preferences cannot be adequately represented under First Past the Post and leads to anomalies; one of the most pro-Remain constituencies in Northern Ireland has elected an MP from a eurosceptic party on only 30.4 percent vote share.

Figure 9: Vote share vs seat share in Northern Ireland


From single-party predominance, Scotland is shifting back towards multi-party politics whilst England goes the other way. Huge swings in Scotland saw 21 of the 59 constituencies change hands, more than any other region or nation.

37 of the 50 seats with the lowest winning vote share UK wide were in Scotland. These small vote shares (typically winners command less than 40% of the vote) suggest voters are choosing to spread their vote around a range of parties.

Voters in Scotland appear to have turned in large number to tactical voting strategies in order to break single-party rule. Huge swings in Scotland saw Scottish constituencies deliver all five of the top five largest decreases in Labour vote share, three of the five largest decreases in Liberal Democrat vote share and four of the top five increases in Conservative vote share UK wide. Four seats saw SNP vote share drop over 18 percentage points.

Figure 10: Election results in Scotland

Party Vote % Vote % change Seats Seat change Seats % Seats % change
SNP 36.90% -13.1 35 -21 59.30% -35.6
Conservative 28.60% 13.7 13 12 22.00% 20.3
Labour 27.10% 2.8 7 6 11.90% 10.2
Lib Dem 6.80% -0.8 4 3 6.80% 5.1
Scottish Greens 0.20% -1.1 0
UKIP 0.20% -1.4 0
Others 0.30% n/a n/a n/a n/a n/a

Nine of the ten largest overturned majorities were in Scotland, including Banff and Buchan where a majority of over 14,000 for the SNP turned into a majority for the Conservative party of 3,600. An example of voter volatility and how all parties fortunes can fluctuate even in a short space of time. Scotland also has four of the top ten smallest majorities (North East Fife, Perth and North Perthshire, Glasgow South West, Glasgow East) demonstrating just how precarious victory can be under First Past the Post.

Figure 11: Scottish vote share vs seat share


Labour continues to dominate in Wales despite suggestion of other parties breaking through. A 12 percentage point increase in vote share saw the Labour Party in Wales take 70 percent of the available seats on 48.9 percent of the vote. Plaid Cymru achieved a broadly proportional 10 percent of seats for 10.4 percent of vote share but the Liberal Democrats lost their one MP in Wales. The Liberal Democrats now have no representation in Wales for the first time. The Conservatives lost three seats despite increasing their vote share by 6.3 percent while the UKIP vote fell 11.6 percent.

Figure 11: Election results in Wales

Party Vote % Vote % change Seats Seat change Seats % Seats % change
Labour 48.90% 12.1 28 3 70% 7.5
Conservative 33.60% 6.3 8 -3 20% -7.5
Plaid Cymru 10.40% -1.7 4 1 10% 2.5
Lib Dem 4.50% -2 0 -1 -2.5
UKIP 2% -11.6 0
Green Party 0.30% -2.2 0
Others 0.20% n/a n/a n/a n/a n/a

Our pre-election polling in Wales found nearly one in four voters were planning to vote tactically this election (YouGov polling for ERS Cymru, Fieldwork: 29th – 31st May 2017, Sample Size: 1014 Welsh Adults)

. Twenty-four percent of Welsh voters polled said they would be voting for the candidate most likely to beat the candidate they disliked. Significant declines in vote shares for parties other than Conservatives and Labour suggest this was the case.

Arfon constituency had the eleventh smallest majority UK wide with just 92 votes deciding the contest. Ceredigion beat Belfast South to the lowest winning vote share at 29.23 percent. This constituency saw the second highest increase in turnout (up 6.2%), Merthyr Tydfil & Rhymney saw turnout increase 7.5 percent.

Figure 12: Welsh percentage Vote change vs percentage seat change


Introduction to the options

Amongst the polarising issues and wasted votes, tactical voting and electoral pacts, there was good news too. Turnout was, at 68.7 percent, the highest it’s been since 1997 and turnout in the 18 to 24 age group is estimated at around 16 percentage points higher than it was in 2015 (though still lower than turnout in older age groups). In some constituencies turnout increased over 12 percent (such as Newcastle upon Tyne East, Ilford South).

Electoral registration grew too with 2.9 million people applying to register to vote in the month from the announcement of the election up to the deadline to register (18 April – 22 May); 612,000 of those registered on the day of the deadline. With 46.8 million people registered, this was the largest ever electorate for a UK wide poll.

However, many of the new registrations were duplicates. Electoral Registration Officers have reported duplicate registrations ranging from 30 to 70 percent. It is more important than ever that electoral registration is made efficient, accessible and fit for the 21st century.

The number of women MPs increased, with 208 women MPs elected in 2017 – up from 191 in 2015. Yet at just 32 percent, women MPs are still outnumbered two to one by male MPs. The UK is now only 40th in the world ranking of women’s parliamentary representation.

Yet these small improvements in turnout and representation do not detract from the vast problems that exist with Britain’s electoral system: from widespread tactical voting and forcing a diverse electorate into two camps, to safe seats and divisive regional disparities.

The odd one out

Britain is the only democracy in Europe to use First Past the Post (FPTP) to elect its MPs (The electoral system of Belarus may be argued to be FPTP, but it is widely considered to be an autocracy). Many alternative electoral systems exist. Using exclusive YouGov research for the ERS, we have projected the election under three electoral systems – the Alternative Vote (AV), the Additional Member System (AMS, sometimes also known as Mixed-Member Proportional Representation or MMP) and the Single Transferable Vote (STV). Our methodologies for these projections can be found in the appendix of the report.

These systems all deal with deficits in the FPTP voting system in different ways, and have been selected because while they do this they also retain a constituency link. All these systems have also been suggested as changes for the UK. It is worth noting that the projections are just that, rather than predictions. It is impossible to know for sure what elections under these systems would look like given that a change in the electoral system will be followed by people becoming more used to the electoral system and its functioning.

A new electoral system is also likely to see changes in the way parties campaign and perhaps even to the structure of the party system itself. Hence these should be taken as a guide, rather than a certainty. Nonetheless, they offer highly detailed insights into how the result could be different under fairer voting systems.

The Alternative Vote

The Alternative Vote electoral system is known for being the electoral system put to a referendum in 2011. It is also used in lower house elections in Australia. The Alternative Vote solves several issues with First Past the Post and retains a constituency link.

AV uses single-member constituencies like Britain’s current electoral system. The principle difference with FPTP is, rather than cast a single ‘x’ against their most favoured candidate, AV asks voters to rank candidates in order of preference by placing a ‘1’ next to their favourite candidate, a ‘2’ next to their second favourite candidate, a ‘3’ next to their third and so on. Voters may rank as many or as few candidates as they wish.

When the counting begins the first preferences (the ‘1s’) are counted first. If a candidate receives more than half the first preferences, then they are deemed elected. If no candidate achieves this then the worst performing candidate is eliminated, and their votes redistributed based on their second preferences (the ‘2s’). Candidates continue being eliminated, with their highest remaining preference acting as an instruction to the returning officer on where their vote should be distributed, until a candidate receives more than 50 percent of the votes.

AV therefore deals with several issues with FPTP – it makes it much more difficult to ‘split’ a vote, and hence voters do not need to vote tactically. They can be safe in the knowledge that if they opt for their most favoured party, their vote will not ‘let in’ a party they dislike – as they can also preference the party best placed to defeat them.

Yet AV does not deal with perhaps FPTP’s greatest flaws – its disproportionality and its safe seat culture. AV can, in the right circumstances, be even more disproportional than FPTP. AV also changes little in safe seats, which are likely to remain safe. In this respect AV is an improvement on FPTP, but a minor one.

Figure 13: 2017 AV projection

GB (not including NI and Speaker) AV Difference from actual result
Conservatives 304 -13
Labour 286 +24
Liberal Democrats 11 -1
UKIP 0 0
Green Party 1 0
SNP 27 -8
Plaid Cymru 2 -2

Labour are the chief beneficiaries of our 2017 AV projection. This is a marked contrast to our 2015 report which found that the Conservative majority would have doubled from 12 to 24. This is because, in this case, preferences from the Lib Dems and Greens subsequently going to Labour tend to outweigh preference flows (The movement of votes from eliminated candidates to candidates still in the race) from UKIP towards the Conservatives, whereas in 2015 the opposite was true. Electoral systems can benefit either party in different circumstances, and as the party system continues to change it should not be expected that AV’s benefits will remain static.

As we can see the AV result is very close, and indeed still results in a hung parliament. It is not necessarily the case that AV would produce hung parliaments in every election – but as in the FPTP election, the closeness of this election is reflected in the result.

The Additional Member System

The Additional Member System, sometimes known as Mixed Member Proportional Representation, is used to elect the Scottish Parliament and the Welsh and London Assemblies. It is also used for elections in Germany and, since a successful electoral reform referendum in 1993, in New Zealand.

AMS is a mixed system, bringing together two different kinds of representation. Voters get two ballots: the first is for a constituency representative elected by First Past the Post, while the second is to vote for a chosen party in an area, on a proportional basis. These seats are compensatory – that is to say that list seats are assigned taking into account how many FPTP seats have already been won by a party.

This means if a region has ten seats, five of which are FPTP and five of which are list, and a party wins half the list votes and three of the of the FPTP seats, it should win two of the list seats. In the version of AMS used in the UK it is possible to have ‘overhangs’ in which a party may win more FPTP seats than its proportional share. British AMS does not correct for this difference as it does in Germany and New Zealand.

The design of AMS systems can differ widely, and its proportionality can differ widely depending on details. The Scottish Parliament and Welsh Assembly versions of AMS are very similar, for instance, but the Scottish Parliament tends to be much more proportionate because the ratio of list members to constituency members favours the constituency members less and because the regions have more list seats, which tends to bolster proportionality. The assumptions we have made for our version of AMS tend to create a much more proportionate version than the version used in Wales (details in the appendix).

AMS is often viewed as a compromise system, bringing together aspects of FPTP and list PR, a ‘best of both worlds’. AMS, it is argued, delivers a single constituency representative for voters and strong proportionality. While constituencies would have to be larger, the regional MPs can act as a secondary level of representation and a check on individual MPs which is not available in a single-member system.

However, it may also be the ‘worst of both worlds’. While a few of the difficulties of safe seats are alleviated – there is reason to campaign in safe seats to win list votes – safe seats still remain a component of the system. Additionally, AMS’s use of closed lists opens issues regarding party insiders selecting representatives rather than the voters.

Figure 14: 2017 AMS projection

GB (not including NI and Speaker) AMS Difference from actual result
Conservatives 274 -43
Labour 274 +12
Liberal Democrats 39 +27
UKIP 11 +11
Green Party 8 +7
SNP 21 -14
Plaid Cymru 4 0

As we can see, in our AMS projection Labour and the Conservatives are tied, this is due to the different way in which people responded to our poll (Updated 29 Aug, the methodology of our projections can be found in the appendix).

AMS we can see produces a relatively proportionate result, but this should not blind us to its other issues. Whilst proportionality is a vitally important element of an electoral system, it is not the be all and end all. While AMS improves on FPTP in many ways, people’s link to their representative is multifaceted, and these other aspects of representation must not be overlooked.

The Single Transferable Vote

STV 2017 results map
A projection of the 2017 General Election under the Single Transferable Vote

The Single Transferable Vote has long been the ERS’ preferred electoral system. STV operates in small multi-member constituencies, generally of around three to six MPs. It is used in elections in Ireland, Malta and to the Australian Senate. It is used in the UK to elect Scottish local councils, and all representatives in Northern Ireland except Westminster MPs.

STV has many advantages. Firstly, it tends to produce broadly proportional election results. But it combines this with powerful constituency representation and ties. Voters’ ability to influence who represents them, both in terms of parties and candidates, is incredibly strong.

Due to this strong link, representatives are incentivised towards a high level of constituency service. A 1997 study found that Irish TDs were far more active in their constituencies than British MPs, while a recent ERS report showed how election campaigns in Ireland are highly localised partly as a result of the voting system.

Figure 15: 2017 STV projection

GB (not including NI and Speaker) STV Difference from actual result
Conservatives 282 -35
Labour 297 +35
Liberal Democrats 29 +17
UKIP 1 +1
Green Party 1 0
SNP 18 -17
Plaid Cymru 3 -1

As we can see from this table, Labour outperforms the Conservatives in our STV projection. This is largely because of stronger second and third preferences from the Liberal Democrats and Greens, similar to Labour’s better performance in the AV projection. However, it should not be assumed that Labour would always benefit from STV more than the Conservatives – in our 2015 projection the Conservatives and UKIP, together, would have won a majority.

Summary of projections

Figure 16: 2017 Alternative projections

Conservative 317 304 274 282
Labour 262 286 274 297
Liberal Democrats 12 11 39 29
UKIP 0 0 11 1
Green Party 1 1 8 1
SNP 35 27 21 18
Plaid Cymru 4 2 4 3


The third strike

The 2017 General Election was the ‘third strike’ for First Past the Post. Far from being a system that guarantees stability, it is a system based on the false premise that stability means single-party majority government. Westminster’s voting system has failed to deliver even this flawed concept of stability.

From delivering a government majority on less than 37 percent of the vote in 2015 to failing to deliver a majority on over 42 percent, an electoral system ostensibly built on delivering decisive majorities has failed spectacularly in the last three elections.

Politics in the UK is changing rapidly. But the voting system is unable to cope with this change. In these diverse and volatile times, we need a system that can adequately represent voters’ choices – not one that creates geographically random results.

Wasting votes, forcing tactical decisions and removing choice from the contest can only exacerbate the sense that politics is failing people. The notion that stability can be achieved by suppressing minority voices, by forming a majority at all costs, has been shown up. A country divided cannot be healed by a winner-takes-all mentality and the winner-takes-all system that promotes it.

Our analysis shows that First Past the Post is failing the UK electorate – but also that there is an alternative. The rise in turnout at this election demonstrates that despite the system, people are engaged and care about the outcomes of our elections. Now we need to give people an electoral system that respects and reflects their actual choices – wherever they are.

We’ve laid out how that can be done. Now it’s time for parties to act.


Projections methodological note

Projecting election results under different systems is tricky work. Electoral behaviour may change. We commissioned YouGov to run a post-election poll with a sizeable sample (13,273). In that poll we asked three questions designed to understand how voters would vote under AV, AMS and STV.

For AV and STV we asked a question as follows:

Please indicate how you would have voted in the General Election on Thursday 7th May, if you had been asked to rank the parties in your order of preference.

Put 1 for your most preferred party, then 2 for your second party, 3 for your third choice etc.

You may rank as many or as few choices as you wish. If you would not vote, or do not know how you would vote, tick the boxes below.

For AMS we asked the following two questions:

Now, imagine that the voting system used for the General Election gave you two votes, one for a constituency MP and one to elect additional MPs to represent your region (in a similar way to the system already used in the Scottish Parliament and Welsh Assembly).

How would you have cast your vote for your CONSTITUENCY MP in the General Election on Thursday 8th June under this system?

And how would you have cast your REGIONAL vote in the General Election on Thursday 8th June under this system?

For AV the projection was relatively simple. First votes in constituencies were recalculated based on first preference results in the poll by region. So if, for instance, 90 percent of those who voted Conservative in London said they would rank them ‘1’ on their ballot then 90 percent of Conservative votes in each area were assumed to be a first preference. We then ran an AV vote in each constituency based on regional results.

For AMS and STV it was necessary to create new constituencies boundaries. For STV, constituencies were created by aggregating existing constituencies into three- to six-member seats. In the vast majority of cases these constituencies are the same as in our 2010 and 2015 reports.

Once this was done we ran an STV vote in each new multi-member constituency in the same way as above. For AMS it was necessary to create both new single-member constituencies and regions.

The single member constituencies were created by aggregating together, generally, two constituencies into one. In some cases existing constituencies were kept because of odd numbers in regions, tricky geography or special exemptions (such as the Isle of Wight). It was attempted to create these constituencies with regard to areas of growth since the last boundary review.

For regions we created regional areas smaller than an official government region designed to be locally representative, often covering a county or two put together. The smallest regions had 3 list representatives (for instance Cornwall) whereas the largest, Greater Manchester, had 13. A 5% threshold was applied to each region.

As above, votes were redistributed based on our AMS question in statistical regions. As with elections to the Scottish Parliament and to the Welsh and London Assemblies, list calculations used the d’Hondt method.


We have made a number of minor corrections to the report after printing. The original result of the AMS calculations failed to include the list results for the Yorkshire and Humber region, the STV result mistakenly included the speaker and the Conservative’s result under AV in Figure 16 was one too low. The STV results map for East Sussex has been corrected to two Conservative, one Labour and one Liberal Democrat.


Produced with the generous support of Jonny Wineberg, Roy Pomfret, Alison Williams, Chris Game, Graham Sowter, Geoffrey Mead, David Lambourn, Peter Richell, Allan Mackenzie, John Beveridge, Samantha Davies, David Beach, John Headon, Edrich Adigun-Harris, Glyn Davies, Tim Baynes, David Nolan, Mike Williams, Stephen Flynn, Raymond Adams, Maggy Daniell, David Bailey, Rowland Carson, Alexander Lowe, Eddy Bryan, Andrew Patrick, Joanna Huckvale, Peter Watsham, Barbara Saville, Yusef Alabbadi, Miki Jablkowska, John Gwyther, Jeremy Broadribb, Nigel Fordham, Kate Allen, Laura Watts, James Russell, John Saunders, Geoffrey Bennetts, Stephan Judge, Pat Fletcher, Sydney Norris, Samuel Priestman, Alan Hedges, David Moon, Thomas Harker, Robin Shawyer, Jeffrey Demuth, Peter Prime, Howard Linsley, Catherine Lack, Paul Skinner, John Simon Hopper, Mark Cummins, Barry Hopewell, Judith Brown, Marc Charlier, Martyn Baldwin, Phillida Shipp, Gill Solnick, Bill Suett, Norman Ridley, Alan Underwood, Michael Gardner, John de Carteret, Diane Sider, James Kellow, Roger Searle, Alan Marsden, Ronald Goldberg, Helen Dimond, Patricia Lacy, Alex Salvaris, Robert Maitland, Peter Fisher, Shaun Coster, Julia Brown, Peter Craven, Gordon Morris, Nicholas Tatam, Peter Hammond, Geoff Mills, Dudley Simms, Stewart Ware, Robert Brice, Roy Palmer, Mike Kirby, John Salt, Stuart Thomas, Michael Borman, Roderick Passant, Mark Child, Nigel Baldwin, Sam Hale, Ann Joice, Peter McGinty, David Rogers, John Calder, Colin Cheesman, Noeline Clayfield, John Marshall, Alan Hutton, Margaret Green, Stephanie Flude, David Royle, John Mayhew, Richard Wilson, Val Harvey, Steven Lindsay, David Brear, Caroline Whitaker, Charlotte Cornell, Peter Waller, Michael de Selincourt, Jason Trask, Louise Mccafferty, Dorothy Forsyth, Ivan Beavis, Brian Salter, Ian Magrath, Fiona Sawyer, Niall Martin, Peter Jackson, Michael Pillidge, Peter Robert Adamczyk-Haswell, Alex Wardle, Ellen Meredith, Guy Johnson, Susan Francis, David Hayes, Patrick Hislop, William Bootiman, Tim Hogg, Lynn Andrews, Florence Edwards, Bob Troup, Randolphe Palmer, Alan Picton, Colin Olcot, Nick Senior, Sarah Mumford, Liz Reason, Chris Dorling, Derek Chandler, Alastair Lang, Andrew Thomas, Andrew Mock, Stephen Dixon, John Gibbons, Lucy Bremah-Andoh, Gareth Renshaw, Iain Cundy, Roger Marshall, Duncan Lyons, Helga Warzecha, Sandra Allsopp Phillips, Helen Sullivan, Anders Wang-Rask, Richard Murray, Terrence Walters, Carmel O’Dell, Donald J C Cameron, David Mcdowell, David von Dadelszen, Kris Page, Laurie Calverley, Robert Palgrave, Linda Ewles, Robert Toft, David Mellenchip, Jacqueline Van Oosterwijk, Peter Willett, Brian Axe, Ian East, Sue Fuller, Adrian Nelson, Ronald Williamson, Rosalind Gauci, Judith Cooper, Tim Coldicott, Louise Birley, Terence Cartwright, Alec Oattes, Derek Wyatt, Darrin Geen, Jeffrey Poulter, Margaret Foggie, Stephen Hunter, Susan Hannis, Peter Coote, Thomas Stopp, Noel Hamel, Patrick McKeown, Edward Pearson, Harold Baxter, Erik Williams, Neil Sloan, Anthony Wren, Robin Goodfellow, Eileen Cox, Peter Anderton, Ed Whelan, Ralph Tyler, Joanna Eley, Elizabeth Timothy, Graham Law, Michael Wall, Norman Matthews, Tim Ashby, Maura Richards, Mark Finch, John Lubbock, George Gwynn, George Penaluna, Auriol Stephenson, Peter Hendrix, Darren Kelliher, Anne Grzybowski, Paul Brunger, Edward Skinner, Joe White, Jean Hackett, Meg Hunter, David Stedman, James Killbery, Marnie Anderson, Iain M Maclean, Robert Ingram, Ian Lebeau, Imran Qureshi, Stuart Danton, Martina Hedley, Jamie Wrench, Tony Price, Harout Manougian, Peter vaughan, Mark Williams, Christopher Burrows, Michael Rozee, Pat Yates, David Appleyard, Gerald Pryke, Ursula Canton, Marc Beacham, Carol Keefe, Nicholas Hay, Bob Young, Amber Donebauer, Stuart Elms, Patrick Keavney, Joseph Dilworth, Susan Kyme, Charles Chichester, Joanne Bennett, Christopher Sampson, Claire Stonehouse, Toby Atley, Robert Brennan, Ian Douglas, David Hoare, Eric Ollington, Richard Young, Will Shepherd, Margaret Clamp, Carol Jones, David Geall and David Chapman.

Read more publications...