Analyses of media content and public opinion show the massive shifts in NDP support that occurred during the campaign, but also point to relative stability where the Conservatives were concerned.
The 2011 election may have lacked substance, but it certainly did not lack drama. Commentators and partisans alike have dissected the result to account for the success of both the Conservative Party and the NDP, along with the decline of the Liberal Party and the Bloc Québécois. The explanations for each are linked, at least in southern Ontario where vote splitting on the left — a consequence of NDP success — seems to have helped the Conservatives by drawing votes from the Liberal Party. But the Conservatives also made gains on their own. And the success of the NDP, especially in Quebec, clearly requires some additional explanation as well.
We take a first step here toward an account of the shifts in and perhaps explanations for the 2011 results. We draw on a combination of media content analysis and a careful campaign-long survey. Specifically, we start by looking at results from the 2011 Canadian Federal Election Newspaper Content Analysis, conducted at the McGill Institute for the Study of Canada in conjunction with the Canadian Election Study (CES). We then look at opinion data collected by the Canadian Election Study itself. We will present first a general view of trends over the campaign and then a preliminary analysis of some subgroups that seem to have been highlighted throughout this election: visible minorities, women and youth.
Overall, the results suggest that in spite of rather limited substantive campaign content, the 2011 campaign did matter to voting decisions, albeit in a way that few observers anticipated. The Quebec media seem to have played an early role in the rise of the NDP, most likely together with a sense of ennui toward the other major federal parties. Outside Quebec, the basic demographics of vote intentions were not fundamentally different in 2011 from what they were in 2008. Rather, minor shifts in vote shares produced more striking shifts in seat shares — striking in terms of producing not just Conservative and NDP gains, of course, but also a Conservative majority.
The 2011 Canadian Federal Election Newspaper Content Analysis tracked coverage in English-language dailies starting on March 14, roughly two weeks before the campaign began. The study included 4,876 stories from the Calgary Herald, Montreal Gazette, National Post, Regina Leader-Post, The Globe and Mail, Toronto Star, Vancouver Sun and Winnipeg Free Press. In addition, we included 734 articles from La Presse and Le Devoir.
Media stories are analyzed for content using automated software developed at McGill University. (Details are available at lexicoder.com.) Automation allows us to examine a huge volume of data, but some of the topic and tone dictionaries operate in English only. We thus have complete data for some coding, and English-only data for others. Even so, the data offer a valuable snapshot of what happened during the campaign.
Take, for instance, figure 1, which reveals party “first mentions” in news stories throughout the campaign. “First mentions” capture the party or leader mentioned first in an article — they capture something about the prominence of competitors and the framing of articles. An article about the Conservatives tends to mention the Conservatives first, for instance. And an article about the Liberals, in the context of an election in which the Conservatives are leading, tends also to start with the Conservatives — “in light of what the Conservatives are saying, the Liberals today…” These first mentions are a simple and reliable means of tracking the relative prominence of parties in campaign media coverage.
In figure 1, we see trends in first mentions that mimic the trends in the polls released during the campaign. The Conservatives clearly led throughout the campaign, though they lost some ground in the final weeks. The NDP surge, alongside the Liberal decline, kicked in over the final week and a half before election day.
The effect of shifts in party prominence, particularly for the NDP, was augmented by shifts in party “net tone” (figure 2). Net tone is the number of positive minus the number of negative words co-occurring alongside leader and party names, based on the Lexicoder Sentiment Dictionary (lexicoder.com). Note that the measure is based on English-language papers only. The Liberals got a small bump as the campaign started, but that dissipated quickly. In spite of the steadiness in the Conservative’s vote shares, the tone of their coverage actually worsened gradually over the campaign. The NDP, in contrast, started high and continued even higher. By the end of the campaign, then, the NDP not only was getting more coverage, it was getting more positive coverage.
Why this is, however, is not readily evident in media data. Table 1 shows the proportion of English-language articles with various topic keywords at the start and at the end of the campaign. There were no major shifts in issue salience over the campaign. So the rise of the NDP was certainly not a product of shifting attention to issues. There are perhaps hints here that the Conservative success was a product of coverage focused mainly on issues of the right — the economy, foreign affairs and crime. But coverage of these issues was high throughout the campaign.
That said, there is one shift in media coverage that is abundantly clear: mid-campaign, Quebec media started paying much more attention to the NDP. Figure 3 shows trends in NDP first mentions in Quebec and newspapers in the rest of Canada (ROC). The shock in Quebec media began on April 16 and reached an early peak three days later. After a brief drop around April 25, coverage of the NDP then increased steadily until the end of the campaign, both inside and outside Quebec. The same was true of vote intentions, and it is to those intentions that we now turn.
We directed the CES, the flagship academic research tool for understanding federal elections dating back to 1965. The study interviewed an average of 130 voters per day over the campaign, using what is called a rolling cross-section design. These telephone interviews lasted about 20 minutes and tapped voters’ intentions as well as political attitudes and socio-demographic characteristics. Data at this stage are preliminary — they are unweighted, and do not include the battery of questions asked in a follow-up post-campaign survey. Even so, analysis of these data offers some valuable early insights into the election campaign. We use the data here to show how the campaign evolved, and how these attitudes and characteristics are related to the choices voters were making.
The study captured the striking campaign dynamics portrayed by commercial polls and media coverage. Figure 4 shows the path of party support and clearly picks up the stunning NDP surge. Some of the NDP rise came from undecided voters making up their minds, but the declines in the Bloc Québécois and Liberal lines indicate that some voters also deserted these parties and switched to the New Democrats. The Conservatives, in contrast, did not make significant gains during the campaign, but started with a sizable lead.
The place to start, then, is to try to account for the dynamics that culminated in the greatest electoral upheaval in nearly two decades. A prime suspect is the leaders. Did Gilles Duceppe and Michael Ignatieff somehow fall afoul of voters; or did Layton-mania sweep Quebec and perhaps leak into the rest of Canada? Figure 5 shows how voters felt about leaders on a 0-to-100 “thermometer” scale. Ignatieff certainly started behind the other leaders, 15 points back of Harper and 20 behind Layton and Duceppe. His fortunes improved over the first week but eventually regressed to where he started, at about 35 on the scale, and no better than his predecessor, Stéphane Dion. Jack Layton started high and only got higher, with a significant uptick starting right after the debates in mid-campaign (April 12 and 14), sending him rapidly from readings around 50 to nearly 60. But even Layton settled back to about 55 by campaign’s end, putting him 10 points clear of his nearest competitors. Harper and Duceppe more or less treaded water a few points below the 50 mark.
Outside Quebec (and again mirroring trends in media content), the NDP did not take off until after the midpoint of the campaign. It is true that this follows the debate, but it also follows the rumblings from Quebec that the NDP was making gains. The NDP’s Quebec surge was reflected in national polls, and media outside Quebec followed suit.
Figure 6 shows trends in vote intentions in Quebec; most notable is the two-stage NDP liftoff. The debates likely played a role in changing things for the NDP, though it seems unlikely that they were the only pivotal moment. A generally well-received appearance on Tout le monde en parle before the debates likely played as big a role in NDP fortunes in Quebec. Note that Layton’s ratings rose from 50 to 60 in Quebec before the debates. This paralleled vote intentions, which started where the NDP left off in Quebec in 2008 at 12 percent support, then rose to nearly 25 percent before the end of the second week, overtaking both the Liberals and Conservatives. The second half of the campaign then saw a second shift in support from all of the other parties to the NDP after the debates. And note that their rise came at the expense of all of the other parties.
Outside Quebec (and again mirroring trends in media content), the NDP did not take off until after the midpoint of the campaign. It is true that this follows the debate, but it also follows the rumblings from Quebec that the NDP was making gains. The NDP’s Quebec surge was reflected in national polls, and media outside Quebec followed suit. Figure 7 shows vote intentions in the rest of Canada, with the NDP rise over the last two weeks starting from undecided voters and perhaps Green supporters, and then only in the last week eating into Liberal support.
Understanding this election requires a clear view of which voters moved to other parties, and which stayed, and this has to be done separately inside and outside Quebec. At this stage we can rely only on vote intentions, gathered during the campaign, and not yet on reports of actual 2011 votes (though the latter are being gathered now in a postelection wave of the CES). Even so, results are telling. So we try here to use data that are somewhat closer to votes by looking at vote intentions from the second half of the campaign only.
Table 2 shows voter transitions across the country, with a person’s report of their 2008 choice in the first column and the rest of the row indicating percentage of those 2008 voters that chose each party in 2011. The bold numbers show the percentage of a party’s 2008 voters who stayed with that party in 2011.
Usually, a party’s strength in an election is reflected in most of its prior supporters sticking with the party. Inside Quebec, the most successful in this respect was the NDP, holding nearly 90 percent of its 2008 voters. Outside Quebec, the Conservatives did the same. In Quebec, the NDP attracted support from all parties. Our data show that federalists and sovereignists came over to the NDP. In the second half of the campaign in Quebec, the Liberals were holding just over half of their 2008 voters, and the Conservatives and Bloc Québécois only two-thirds of theirs. The NDP was the beneficiary, taking about three-quarters of those leaving the three other main parties.
Outside Quebec, there was more loyalty to 2008 choices, and we detected no significant change in support after midcampaign, except in support for the Liberals, who, in the last week, could hold onto only 70 percent of their voters, with deserters going two to one in favour of the NDP (not shown in table 2). Tellingly for the Liberals, while the NDP held onto 82 percent of its voters after mid-campaign outside Quebec, the other NDP voters were going two-to-one to the Conservatives. (That is, 11 percent of 2008 NDP voters intended to vote Conservative, while just 5 percent of 2008 NDP voters intended to vote Liberal.)
Overall, we see relatively little movement in the campaign, or since 2008, among those outside Quebec describing themselves as partisans (those who say they feel closer to one party or another). The place to look for clues to Conservative success is among nonpartisans. Fifty-six percent of them described the Conservatives as the party best able to manage the economy, while the NDP and Liberals attracted only 20 percent each on this score. This gives the Conservatives a huge advantage on what is typically an influential issue in nonpartisans’ decisions.
The Conservatives were indeed in trouble on their relationship with Parliament, as 50 percent of nonpartisans agreed that the “Conservatives had been disrespectful towards Parliament,” while only 30 percent disagreed. But this issue hardly stung them, as they still attracted 44 percent of the vote among those who “somewhat agree” with this statement. Overall, more than one Conservative voter in four agreed that the Conservatives had been disrespectful toward Parliament.
Commentators inferred from the Conservatives’ majority that the party, and Stephen Harper in particular, have finally convinced Canadians that they have no “hidden agenda” and will run a mainstream government. Yet Canadian voters remain unconvinced. Half of the electorate agrees with the statement that “Stephen Harper is just too extreme,” unchanged from his first election as leader seven years ago.
The stability of views about Harper is largely echoed in voting trends among women, youth and immigrants. There was no doubt, certainly among the parties, that this election would be shaped by these demographic subgroups. Several parties vowed to extend their voter base among women, for example, and pundits talked a good deal about the gender gap during the campaign. We know from previous work that in recent elections women have tended to vote increasingly for parties of the left. Indeed, the NDP has picked up the women’s vote disproportionately over time. (See figure 8, where the “vote gap” is the percentage-point difference between vote choice by women and men; though note again that 2011 data must rely on vote intentions.) How did this effect play out in the most recent election? To what extent were the successes of the NDP and the Conservatives shored up by the women’s vote?
Our preliminary analysis shows that in spite of previous trends, the support for the NDP nationwide was evident among both women and men. Indeed, the gender gap shrank to just 2 points, meaning that only a negligibly higher proportion of women than men intended to vote NDP (see figure 9). By contrast, the Conservatives were unable to cut into their deficit among women, which continues to hover around 4.5 points.
Our preliminary analysis shows that in spite of previous trends, the support for the NDP nationwide was evident among both women and men. Indeed, the gender gap shrank to just 2 points, meaning that only a negligibly higher proportion of women than men intended to vote NDP.
The story is largely the same in Quebec. There, men were more likely to vote Conservative (an 8-point gap), and — in contrast to the rest of Canada — women were more likely to vote Liberal (again 8 points) and slightly more likely to vote Bloc (3 points) than men. In sum, nationwide, no single party carried the women’s vote disproportionately, and the Conservative success has still not found an equal foothold among women.
There was also a lot of talk in this election about the youth vote, particularly given dismal turnout rates among the youngest Canadians in recent elections. Indeed, many have commented on the strong online presence of campaign-related information that youth potentially consume. Examples include Web sites with funny pictures of political leaders when they were young; nonpartisan, entertaining videos with minimal preaching content encouraging young people to vote; awareness campaigns such as ShitHarperDid.ca; and voteswapping sites that were intended to help people bypass the first-past-the-post system, among others. About 78,000 young adults pledged to vote on the Apathy Is Boring Web site, and 35 vote mobs took place across Canadian university campuses. We cannot yet say if these mobilization attempts actually encouraged youth turnout; but we can use the CES to look at whether younger people’s vote intentions varied from those of their older counterparts.
Previous studies find that in 2004 young voters slightly preferred the NDP, as compared with older voters; though younger voters also tend to have less stable party identifications. In 2011 there were much greater age gaps in vote intentions. By far the largest gap can be found in the Conservative vote (see figure 10), where support is distinctly lower among those aged 40 and under. Younger people were more likely to prefer the NDP, and in Quebec the Bloc (by around 4 points more each); to a smaller degree youth were also more likely to plan to vote Green.
The Liberals were shunned this time by those aged 39 and under and those aged 40 years and over alike. Additional analyses of vote shifts show that the NDP voter base in 2011 was made up at least in part of younger voters who voted Liberal in 2008 and in part of older voters who voted for the Bloc three years ago. By the same token, older Liberal voters and younger Bloc voters from 2008 were more faithful to their parties in this election.
A separate online study of McGill University students (1,000 respondents) seems to confirm some of these insights. Of all the students who usually identify as Liberals, about 15 percent intended to vote NDP, another 15 percent were undecided one week before the election, and only 65 percent intended to vote for the Liberals again. But young Green Party identifiers also contributed to the NDP surge — about 18 percent intended to vote for the NDP this time around.
What about immigrants? Given the leaked party message on Jason Kenney’s efforts to woo the “ethnic” vote, the immigrant vote was a major theme in media content even before the campaign started. Election-night media coverage, spurred on by Conservative gains in southern Ontario in particular, continued the trend. Early results from the CES do not support the hypothesis that the Conservative success in 2011 was a product of making headway with immigrants, however.
Figure 11 shows the immigrant vote gap in 2011. Immigrants in this case are first-generation Canadians from countries other than the US and Europe — for the most part, then, visible minority immigrants. There is a tendency for immigrants to vote slightly less for Conservatives in 2011, though note that this gap is not markedly smaller than it was three years earlier. (In 2008, the gap was roughly -5.5.) We should not minimize the pre-2008 shift — from 2006 to 2008, there was a marked increase in the proportion of first-generation immigrants voting for Conservatives (in line with other Canadians as well, of course). And the vote gap for Liberals, about +10 in 2008, had shrunk to just over +5 points by 2011. But the Liberal decline among immigrants does not seem to have been matched by a Conservative gain. Overall, the distributions of immigrant and nonimmigrant votes are very similar.
In sum, the Conservative success of 2011 does not appear to have been a product of the party diversifying its appeal, at least not among any of the three groups investigated here. We should perhaps be cautious in our interpretations here — our analyses focus at this stage on vote intentions expressed during the campaign, and decisions may have shifted at the ballot box. But based on our data thus far, the NDP clearly broadened its support — not just in terms of picking up a huge proportion of the Quebec francophone vote, but also in terms of picking up, proportionally speaking, more men, more youth and somewhat more new Canadians. And the Conservative success, in contrast, seems to have been more about keeping the base happy and appealing to a greater proportion of sympathetic voters on the centre right.
The 2011 election produced a massive shift in the composition of the Canadian Parliament, and it may well be the first step in a permanent and profound change in the Canadian party system. Media content captures an important part of the story: a midcampaign shift in attention to the NDP, gradual declining interest in the Liberals and steady support for the Conservatives. In Quebec, the NDP rise was particularly clear, and it found an echo in the rest of Canada. Indeed, it is the results in Quebec that point most clearly to the possibility of real and permanent change in the federal political scene.
That said, while the NDP gained, so too did the Conservatives, and it is the latter gain that matters most in the short term. Our survey results, preliminary though they are, point to the possibility that the Conservative success in 2011 was not a product of massive shifts during the campaign, however. Where the Conservatives are concerned, 2011 seems to have been in large part a continuation of trends that began in 2008. The Conservative base is not much broader — in terms of proportion of vote, or in where those votes came from — than it was in 2008. The electoral results are fundamentally different, however; and where that will lead us in 2015 is not at all clear.