Electoral System Oddities- First Past the Post Systems Do Not Just Distort Electoral Outcomes, but also the way that Leaders are Evaluated

The beginning of May was a good time for election junkies. One election, in Alberta, led to the province’s first ever NDP majority government. Two days later, the British election produced an un-expected Conservative majority for David Cameron. Both elections also saw the collapse of parties that had been important in the previous election. In Alberta the Progressive Conservatives fell to third place after spending 44 years in government. In the United Kingdom the Liberal Democrats went from 56 seats (and membership in a coalition government) to 8 seats a fourth place in parliament. The success of party leaders is often measured by their electoral results. In the UK David Cameron was praised for winning the Conservatives a majority while Labour leader Ed Miliband and Liberal Democrat leader Nick Clegg resigned losing significant numbers of seats for their party. In Alberta the NDP’s Rachel Notley and the Wildrose Alliance’s Brian Jean were praised while Jim Prentice took the blame for significant PC losses. Pundits should be careful about judging leaders by the seats they win in first past the post electoral systems. Just as seat shares can distort electoral outcomes, they can also distort the way that people view leaders. Leaders can pick up votes, successfully convincing voters to move over to their party, but appear to be failures because those votes do not translate into an increased number of seats. Conversely, leaders who lose voters between elections can have those losses masked by seat gains. In Alberta and the UK, increases in seats shares made Wildrose Alliance leader Brian Jean and British Conservative leader David Cameron look more successful than they actually were, while Labour leader Ed Miliband’s drop in seats hides the fact that he actually increased Labour’s overall share of the vote.

One of the key purposes of a party’s election campaign is to convince voters to move their support over to the party. While there are certainly many factors that influence voters’ decision, party leaders’ tours play an important role in the success of parties’ campaigns. As the most visible members of their party, leaders are often the focus of news coverage, election advertising, and have a unique opportunity to engage with voters across the country (instead of just in their constituency) throughout the campaign. While leaders might be able to convince voters to switch their support over to their party, the ability to convince voters in the right places to switch their support is a much more difficult task. Parties can certainly target particular districts, spending a greater number of resources in districts that are likely to be close as opposed to those that party will is either almost certain to win or almost certain to lose. Many aspects of a party’s campaign, though, are received equally by people across the country. Even if advertising campaigns can be concentrated on particular cities, they can rarely be focused on just those districts that are likely to be swing districts. National media coverage of leaders will be received with as equal frequency in districts that are highly supportive of one party as it will be in districts that are likely to swing between parties. Beyond focusing more spending in certain ridings and making an extra appearances in particular ridings, the best that leaders can try to do in to increase their seat share is convince voters across the country to support their party. The ability of leaders to control the distribution of their vote share is much weaker than their ability to control the size of their vote share.

The difficulty that parties have translating votes into seats is compounded by the lack of control that parties often have over the success of their rivals. Both the emergence and the collapse of parties can have significant consequences for the number of seats that a party wins even if their overall share of the vote does not change. The collapse of a key rival can mean that seats that were close races in a previous election become clear wins for the party. The rise of a new rival party can make seats that were once safe for a party hotly contested. Both the Alberta and the UK election featured these effects. The collapse of the Conservative vote in Alberta meant that Wildrose Alliance candidates were often competing against much weaker candidates than they were in the previous election in 2012. The collapse of the Liberal Democrats in the United Kingdom had similar effects, while the rise of the Scottish National Party meant that the Labour party ended up fighting close races (and indeed losing) in seats that had previously been Labour strongholds.

The success of David Cameron’s Conservative party in the United Kingdom is a prime example of how parties can increase their share of the seats without any change in their popular vote. Cameron’s Conservatives managed to move from a coalition government (reliant on the Liberal Democrats for support) to a majority government. They gained a total of 28 seats, despite increasing their overall share of the vote by less than 1 percentage point. Cameron was able to win a majority because of the collapse of the Liberal Democrats. 36 seats switched to the Conservatives between in the 2015 election (8 Conservative seats switched from the Conservatives to other parties in order to result in a net gain of 28 seats). Of these 36 seats that switched, 27 came from the Liberal Democrats and only 9 from the Labour party. In only one of two of these seats (including those the Conservatives won from the Labour party) did the Liberal Democrats lose less than 10 percentage points of the vote compared to their 2010 total. In eight of those seats the Liberal Democrat vote losses exceeded 20 percentage points. The average Conservative gain over all of the seats they won was just 4 percentage points while Labour lost only an average of 0.5 of a percentage point in the seats they lost to the Conservatives (the Conservative gain in seats they took from Labour was an average of 3.7 percentage points). The Labour party actually gained votes in all but one of the seats that the Conservatives took from the Liberal Democrats. The Conservatives won their majority not because they were able to convince voters to abandon Labour, but because they were running against severely weakened Liberal Democrat candidates. In only one seat that Labour lost did their losses exceed Conservative gains, suggesting that even in those seats the Conservatives were able to benefit from taking votes from the Liberal Democrats in addition to any support they may have won from Labour. The collapse of the Liberal Democrats played a larger role in the Conservative victory in the UK than the increases in vote share that the Conservatives managed during the campaign.

If David Cameron gaining a majority government without gaining much in the way of additional vote support seems odd, Brian Jean’s success in the Alberta election is even stranger. Jean’s Wildrose Alliance dropped 10 percentage points compared to the 2012 election. Despite this, they managed to gain 4 seats, holding on to their position as leader of the official opposition (despite finishing third in popular vote share). Like the British Conservatives, the Wildrose Alliance benefited from the collapse of a major rival. The Progressive Conservative share fell by 16.2 percentage points. Of the 7 seats that went PC in 2012 but Wildrose Alliance in 2015 (Wildrose Alliance losses account for the net gain of only 4), in only 2 did they exceed their 2012 vote share. 5 Wildrose Alliance candidates managed to pick up seats previously won by the PCs despite winning less support than the Wildrose Alliance won in their riding in 2012. Those results are not evidence of a successful campaign. Rather they suggest that the Wildrose Alliance benefited from against weakened Progressive Conservatives. Had the Progressive Conservatives been as strong as they were in 2012, the Wildrose Alliance likely would have lost a substantial number of seats.

Where the British Conservatives and Wildrose Alliance were benefited from the collapse of their competitors and the way those collapses interacted with the electoral system, the British Labour party was hurt by the rise of a rival. Labour suffered a net loss of 24 seats in 2015, and did particularly poorly in Scotland. There, Labour lost 40 seats to the Scottish National Party (SNP). Labour’s losses in Scotland were not just significant with respect to seats, they also dropped 18 percentage point there (while the SNP picked up 30). Popular vote wise, though, Labour was able to make up for their losses in Scotland by increasing their support in the rest of the UK. Across the entire United Kingdom, the Labour vote share actually increased by 1.5 percentage points, a gain greater than the Conservatives’. Despite increasing their vote by more percentage points than the Conservatives, Miliband’s Labour party actually lost seats. This is because, while the Conservatives were fighting a weaker rival than they were in 2012 in the Liberal Democrats, in many seats Labour was fighting a stronger rival in the SNP. When it comes to convincing the public to support Labour, Miliband did a better job than previous Labour leader Gordon Brown did in 2012. Yet, because the Liberal Democrats were weaker and the SNP was stronger, Miliband ended up losing seats. This is not to suggest that Miliband’s campaign was a success. He failed to win over the voters that abandoned the Liberal Democrats and failed to stop the rise of the SNP in Scotland. An increase in of 1.5 percentage points is not the kind of popular vote increase that can be expected to take one from opposition to government. At the same time, Miliband should not be seen as doing worse for Labour than they did in 2012.

None of this is to argue that leaders should be judged entirely by their ability to increase or decrease their share of the popular vote. The circumstances in which leaders fight elections are often different. Leaders in opposition facing an unpopular government might rightly be criticized for failing to win government even if they manage to increase their party’s share of the popular vote. Leaders of governing parties might be rewarded for maintaining their share of the popular vote if the polls in the years preceding an election show a sharp decline in support for the governing party. Finally, election results should not be fully attributed to the success or failure of party leaders. Economic success or failure, the rise of new issues and political cleavages (such as Scottish Nationalism), and the good or poor performance of a government can affect public opinion in ways that leaders are unable to compensate for over the course of an election campaign. To the extent that leaders are judged on their results though, in first past the post systems it is important to look beyond the number of seats that the leader’s party wins or loses. The way that the electoral systems translates seats into votes in different elections, depending on changes in strength of a party’s rivals, can make their campaign look either more or less successful that it actually was.


How Alberta Became an NDP Province: Electoral Institutions, Party System Change, and Rachel Notley’s NDP victory

Growing up in Alberta I never thought I would see what happened in this provincial election. The NDP have won a majority government, 44 years of Progressive Conservative government. Rachel Notley and the NDP ran an incredibly strong campaign and deserve a lot of credit for winning a large amount of support in a province that, compared to most, is quite conservative. The success of the NDP, however, cannot be fully understood without an examination of the changes in party politics that have occurred over the last two elections. Changes in the party system over those two elections produced consolidated support for the NDP on the centre-left of the political spectrum and a split vote on the right. The first past the post system, which has long punished Alberta’s centre-left for vote splitting, this time hurt the right for doing the same. The NDP appear to have gained the support of a small but significant number of right party supporters in this election, but the major reason that they have won a majority government has to do with the rise of the Wildrose Alliance and decline of the Liberal Party.

The start point for an explanation of what happened in Alberta this election should be the previous election in 2012. What is notable in that election is the surge in support for the Wildrose Alliance, and that that surge in support did not translate into a decline in PC support. The popular vote for the Wildrose grew by 27 percentage points in that election, yet the PC vote only fell by 8.8 percentage points. It is unlikely that this was a result of Liberal and NDP voters moving to the Wildrose Alliance. The ideological distance between the Wildrose Alliance and the other opposition parties as well as the fact that Wildrose Alliance picked up rural seats that were former PC strongholds suggests that the Wildrose gains largely came at the expense of the PCs. The PCs were able to hang on to power by moving to the centre and casting themselves as the best chance centrist and left voters had of keeping the Wildrose Alliance out of power. This resulted in a jump in combined right party support in that election. The total right party share of the vote in the 2012 election was 78%. Alberta is a conservative province, but it is not that conservative. The total right party vote in most election hovers around 55%-60%. The only way the right gets to 78% is if one of the right party attracts a significant number of strategic voters from the centre-left. It is far more likely that those strategic votes went to the Allison Redford led PCs than the Wildrose Alliance. The PC party that emerged from the 2012 election was thus significantly weaker than it appeared. It had lost a large portion of its right wing base and was now reliant on strategic votes who were probably closer ideologically to the NDP and Liberal parties. These were not voters who could be expected to remain PC over the long term, and thus the shift in the vote on the right of political spectrum in 2012 left the PCs vulnerable to a surge in support for either the Wildrose Alliance or a strong Liberal or NDP party.

It is worth noting here that the surge in Wildrose Alliance support was in 2012, not 2015. The Wildrose Alliance vote share in 2015 actually dropped 10 percentage points compared to the 2012 election. Though some of these voters may have moved to the NDP as anti-PC voters, the larger proportion probably returned to their closest ideological competitor, the PCs. The PCs thus won some of their base right wing base back in 2015, but not enough to make up for the strategic voters that they had lost. The Wildrose Alliance picked up an additional 4 seats compared to the 2012 election, but not because they had convinced more voters to support them. The fact that they were facing a weakened PC party allowed them to win seats that they would have lost last election to PC candidates.

The problems in consolidating the right vote were exacerbated by how close the PCs and Wildrose Alliance were in the polls. The final 308 projections (based on aggregations of polling data) had the Wildrose Alliance and PCs separated by only about 2%. This is a problem for strategic voters on the right, because it made it unclear as to which party presented the best opportunity to beat the NDP. If one of the two parties had taken a clear lead in the polls, right voters could have consolidated around that party and kept a significant number of ridings from falling to the NDP. As it turned out, the combined vote for the right parties was substantially higher than the NDP’s share of the vote (52% to the NDP’s 40.6%), and split pretty evenly between the two right parties. The Wildrose Alliance took the greater number seats (21 with 24.2% of the vote) but the PCs took a greater number of votes (27.8%). The NDP won 19 of their 53 ridings with less than 40% of the vote. Those 19 ridings would have been difficult for the NDP to win without significant vote splitting on the right.

On the other side of the political spectrum, the demise of the Liberal party made an important contribution to success of the NDP. Throughout the 1990s and 2000s vote splitting was a problem for the left and the centre of Alberta politics. Rarely did the centre and left have a majority of voters, but prior to 2012 the NDP and Liberals would take between 35% and 40% of the popular vote. In 2015 the combined NDP/Liberal/Alberta Party vote share was 47%. This is substantially higher than past elections, suggesting that the NDP did take some PC votes, but it is also not the radical shift that the change in seats suggests. Indeed, the NDP/Liberal vote share has been higher than it was in 2015. In 1989 the combined vote share for these parties was 55% while in 1993 it was 51%. Over the past three decades vote splitting on the centre and left has made the Alberta centre-left look a lot smaller than it actually is. After taking most of the centre and left vote from the Liberals (who fell to 4% and 1 seat in 2015), the NDP did still need to take some right wing voters in order to win government, but they did not need that many. A shift of 7 to 10 percentage points from the right to left is not inconceivable when a right government is unpopular (as both Prentice was). The unpopularity of the PC government led by Don Getty in the late 1980s and early 1990s actually caused a greater shift from right to left than occurred in Alberta in this election. The difference between this election and the 1989 and 1993 elections was that this time, the centre-left was consolidated behind one party. Liberal voters moved in large numbers from the Liberals to the NDP, and the right vote that shifted to the left ended up shifting to the NDP instead of being split between the NDP and the Liberals.

As a final note, it is worth pointing out that the anti-incumbency bias in polling was once again present in this election (though it was not large enough to significantly affect the result). 308’s final projection for the PC vote was 24% with a predicted high of 26%. The actual PC vote of 27.8% fell only within 308’s most optimistic projections for the PCs. This suggests that whatever caused polls to under-predict incumbent vote share in past elections (in Alberta, BC, and federally) was again present in the 2015 Alberta election, though this bias was not as large as in the BC or previous Alberta election. The NDP vote share was also low compared to 308’s aggregation of polls. The NDP took 40.6% of the popular vote, just below the 40.9% that 308 had predicted as the low score for the NDP and well below the 44.5% in the middle of 308’s predicted NDP range (though, again, the NDP score was within the most pessimistic of 308’s projections). The lesson to take from this is that polls showing shifts from an incumbent party to a challenger should be taken with a grain of salt. These shifts are probably smaller than polls project, but that does not necessarily mean that they are not there at all. A significant gap in polls favouring a challenger may still be real, it just may not be as large as the polls predict.

There should be two major take-aways from this Alberta election. The first is that there was a shift away from the right to the left, but that shift was small. On its own, it was probably not enough to put the NDP in power or to end the hold that right parties have over Alberta’s politics. The second, is that a large part of the NDP victory was a result of changes to the party system and the way that those changes interacted with the electoral system. The demise of the Liberals and rise of the Wildrose Alliance meant that Alberta politics went from having a dominant right party competing with two much smaller centre and left parties to a dominant centre-left party competing with two mid-sized right parties. The first past the post electoral system then punished the mid-sized right parties for splitting their votes and rewarded the NDP for consolidating the centre and left vote. With respect to the party and electoral systems these were the perfect conditions for an NDP victory in Alberta. If the right parties and voters can coordinate to a greater degree in the next election, the NDP will face a much tougher challenge in the next election.


Undecideds Matter: Recent Polls in Alberta May Be Overstating NDP Support By Ignoring Undecideds and Voter Volatility

If someone had predicted two months ago that a week before Alberta’s May 5th election multiple polls would give the NDP a 8-10 point lead they would have have gotten some strange looks. It was reasonable to suggest that in the wake of the Progressive Conservative unpopularity that Alberta might finally see a change in government, but few would have predicted that that change would be to an NDP government. Indeed, the last time the NDP won more than 4 seats was in 1989. The likelihood that the NDP forms a government may be overstated in a number of recent articles on the election. Polls that have included a measure of undecided voters or voters willing to switch votes show the potential for substantial volatility. This volatility is likely to hurt the NDP in three ways. Polls showing large numbers of undecided voters are likely to be understating the strength of both the PCs and the Wildrose Alliance, while polls showing a high level of uncertain voters could see the NDP hurt if either the right vote coalesces behind a single party or if voters move away from the NDP towards the incumbent PCs.

A couple of polls suggest that there may be a substantial number of undecided voters who simply are being counted in the numbers being reported in most polls. In a poll conducted immediately after the debate, Mainstreet Technology reported that 21% of those surveyed were undecided. Including undecideds brought the NDP down to 24% from the 31% of decided voters only. On April 30th, CBC reported on a poll they had commissioned with Return On Insight that had the number of unsure/non-voters at 26%. Including these respondents in the sample brought the NDP down to 28% from a 38% share of decided voters. CTV reported on a Think HQ poll that has the NDP at 39% but has an undecided percent at 13%. The Return on Insight and Think HQ polls probably do not have enough undecideds for either the PCs or the Wildrose Alliance to catch the NDP if the undecided voters break disproportionately towards the right parties. The election would however be a lot closer if that happens. A 10 point lead for the NDP could be reduced to a 4 or 5 point lead if undecideds break disproportionately towards a single right party. That is a lot closer to the margin of error, and a lead that it is not inconceivable to see the PCs or the Wildrose close over the last few days of the campaign.

A scenario where undecideds move towards one of the right parties is fairly likely. Partisan competition on the right of the political spectrum is much closer than on the centre/left. The PCs and the Wildrose Alliance have maintained base levels of support above 20%, suggesting that they both stand a reasonable chance of winning seats in the election. Because of this, it has been unclear as to which of the right parties strategic right wing voters should support. For right voters casting a vote on principle the Wildrose Alliance and the PCs both have significant advantages but also significant drawbacks. The Wildrose Alliance offer an alternative on the right to conservative voters who are frustrated by a combination of PC mismanagement of the economy and scandals within a PC-led government that has been in power for too long. That the Wildrose Alliance is further to the right of the PCs may, however, make more moderate PC voters nervous about switching their votes over to the Wildrose Alliance.

On the centre and left of the Alberta political spectrum the choice is clearer. The Liberal party has had a poor election. They have run a weak campaign compared to a very strong one for the NDP. That in and of itself may be enough to convince many on the centre and left to move their support to the NDP. On top of that, the high NDP poll numbers and low Liberal numbers mean that the NDP are the best choice for any strategic voters looking to support whichever party offers the best chance at defeating one of Alberta’s right parties. The decision over who to support is more difficult voters on the right of the political spectrum than it is for voters on the right. There should thus be more undecided voters on that side of the spectrum. As they make up their minds between the Wildrose Alliance and the PCs, the percentages for both those parties should increase while the percentages of support for the NDP and Liberals should drop. The Liberals and NDP do not need to lose any of their existing supporters for this to happen, they only need to win over fewer undecideds than the Wildrose Alliance and PCs do.

Not all polls have large numbers of undecided voters. Global news reported on a poll conducted by Ipsos that had only 8% of respondents undecided. This poll, however, showed a great deal of volatility in the opinions of decided voters. In this poll, 48% of respondents who gave a vote choice also said that they could change their mind and vote for a different party. This presents two concerns for the NDP. The first is that right wing voters could coalesce behind the right wing party they see most likely to defeat the NDP. The ideological distance between both the PCs and the Wildrose Alliance and the NDP means that most right party voters are likely to prefer one of those parties to the NDP. If either the PCs or the Wildrose Alliance emerge as the clear challenger to the NDP, defections from the weaker right party to the stronger could easily cause the stronger right party to catch the NDP. Three Hundred Eight’s aggregate poll numbers have consistently shown support for right parties that are consistent with the past elections resultes I noted in my previous post. Even though the NDP is leading the polls, a substantial majority of voters in Alberta are still supporting right wing parties. The Ipsos poll shows that many of these supporters, 36% of Wildrose Alliance supporters and 46% of PC voters, are willing to change their votes. The fact that right party support remains high and substantial numbers of right party supporters report a willingness to change their votes make a shift towards the strongest right party reasonably likely. Such a shift could lead the strongest right party to surpass the NDP in support.

The other concern for the NDP regarding voter volatility lies in their ability to hold on to their own supporters. Polls in Canadian elections have consistently under-predicted incumbent vote shares (see the BC provincial election, the previous Alberta election, and the 2011 federal election). This may be a result of individuals who are unwilling to tell pollsters they are voting for an unpopular incumbent, voters who are persuaded to switch their support back to the incumbent as opposition parties leading polls receive increased media scrutiny, voters who get cold feet over voting out a government at the last minute, or some other factor not apparent to pollsters. The Ipsos poll notes that 49% of respondents who said that they supported the NDP said that they were willing to change their mind. To the extent that the NDP has won over PC voters, and it is not clear that they have won many consistent PC voters (a significant number of the individuals who voted for the PCs in the last election but are voting NDP now are likely to have been voters the strategically voted PC in 2012 to keep the Wildrose Alliance out of power), they are at significant risk of losing those voters in the days leading up to the election.

To this point the NDP have had a very good campaign. They are likely to take a large number of seats, far exceeding the number of seats and votes that they have won in any of the elections in the past two decades. The number of undecideds and the volatility should make one skeptical of any predictions of an NDP government. At this point in the campaign the number of undecideds and voter volatility are both working against the NDP. The election should still see a large NDP vote share and should make a major breakthrough in Alberta politics. The NDP vote share in this election looks like it is going to big (at least by Albertan standards), but it may not be as big as polls are predicting.