5-6 pages reflective

10.AndrewFournierSoroka.pdf

Home >Political Science homework help >5-6 pages reflective

8 The Canadian Party System

Trends in Election Campaign Reporting, 1980-2008

BL AKE ANDRE W, PATRICK FOURNIER, AND STUART SOROKA

There is little doubt that there have been significant changes in the Can- adian party system over the last twenty years. The 1993 election saw the collapse of the Progressive Conservative (PC) Party, in seats if not also in votes, and the rise of a right populist party that has in one form or another – Reform Party or Canadian Alliance – replaced the PCs as the main party of the right. In 2004, the right reunited under the banner of the Conservative Party of Canada, an entity significantly dissimilar from the PCs. The Liberals suffered a string of failures during the last decade, falling from a majority government in 2000, to a minority government in 2004, to op- position in 2006, to even more losses in 2008, and finally to the most dev- astating defeat of its entire history in 2011. While the New Democratic Party (NDP) vote share climbed steadily between 1993 and 2008 (from 7 percent to 18 percent), its breakthrough in terms of seats and votes oc- curred in 2011. NDP success in that election coincided with the collapse of the Bloc Québécois (BQ), a party that had maintained strong regional support in the province of Quebec between 1993 and 2008. In 2011, Canada experienced a sea change in the distribution of seats in the House of Com- mons, as the Conservative Party formed a majority government with the NDP as official opposition, while the Liberal Party and the Bloc Québécois were nearly decimated. Clearly, the Canadian party system in place today is fundamentally different from the one that existed throughout the 1960s, ’70s, and ’80s.

Parties, Elections, and the Future of Canadian Politics, edited by Amanda Bittner, and Royce Koop, UBC Press, 2013. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/templeuniv-ebooks/detail.action?docID=3412833. Created from templeuniv-ebooks on 2018-01-21 17:57:56.

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

162 Blake Andrew, Patrick Fournier, and Stuart Soroka

That the distribution of party support has shifted is beyond doubt – both within constituencies and across regions. The Trudeau and Mulroney ma- jorities garnered votes from across the country. Indeed, this pan-Canadian support is the defining feature of what scholars have labelled the “third party system” (Carty 1988; Carty et al. 2000; Johnston et al. 1992). This sys- tem collapsed in 1993, however, when two regionally based and predomin- antly ideological parties broke through in both public support and seats. In short, 1993 saw the end of pan-Canadian politics. The newly minted “fourth party system” – one that Carty and colleagues (2000) defined in terms of regionalization of party support and campaigning, as well as growing ideo- logical polarization across parties – has been in place ever since. Arguably, the electoral upset of 2011 has continued this trend.

With regard to seat and vote shares, the regional dimension of contem- porary Canadian politics is fairly obvious. The largest parties all rely heav- ily on regional strongholds. Since 1993, the West has been dominated by the Reform Party/Canadian Alliance/Conservative Party of Canada, Ontario has tended to be a secure realm for the Liberals, and the BQ has systematic- ally collected the plurality of seats in Quebec. But the regionalization pic- ture is not static. Most notably, the Conservative Party has begun gaining support east of the Prairies, particularly in Ontario, where in 2008 it sur- passed the Liberal Party in both seats won and vote share, marking the first time in two decades that this province has slipped out of the hands of Canada’s “natural governing party.” The 2011 election consolidated this change in the base of Conservative support, while Quebec swung dramat- ically from supporting the BQ to supporting the NDP.

Until the dramatic shift in the distribution of seats in the House of Commons after the 2011 election, ideological polarization across parties had been less evident in election results. It has nevertheless been a popular thesis among many political pundits, who liken trends in Canada to the polarization that has occurred between the Republicans and Democrats in the United States during the same time frame. Recent studies of Can adian voting behaviour do find some evidence of growing ideological differen- tiation (Blais et al. 2002; Gidengil et al. 2012; Nevitte et al. 2000). Values and fundamental outlooks have played an increasingly important role in explaining voters’ preferences under the fourth party system.

These two trends – regionalization and ideological polarization – can be analyzed in various ways. Generally, they have been examined with evi- dence relating to voting behaviour or party organizations. We believe that campaign-period political communication may also offer useful means

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

163The Canadian Party System

of capturing – and perhaps even explaining – these shifts. In this chapter, we first ask whether evidence of regionalization is found in news coverage of elections. Is newspaper coverage of parties increasingly divergent across Canadian regions? We then ask whether polarization is evident in news- paper coverage. We explore the possibility that more polarized party politics is reflected in the relationship between parties and issues in campaign- period news content.

This chapter represents a first attempt to use news content from recent election campaigns to explore the relationships between parties, issues, coverage, and voters, and to examine the ways in which party competition and the party system have changed during the last three decades. We analyze all campaign stories published in five major Canadian newspapers over the course of the six federal elections held between 1993 and 2008, as well as all stories from the Globe and Mail’s coverage of federal campaigns during the 1980s. We explore differing coverage of parties and leaders across regions, and trends in the reporting of issues. Whether this type of analysis tells a story of continuity or change about the fourth party system is the focus of the work described in this chapter. Our evidence points towards continuity.

Background

Regionalization Regional protest parties have been a staple in Canadian politics (Johnston et al. 1992), and include the agrarian Progressives during the 1920s, Social Credit during the middle third of the twentieth century, and the western Reform Party of the 1990s, to name just a few. There have also been some durable regional divides in the traditional support bases of the major par- ties throughout Canadian history (Johnston et al. 1992). For instance, from Confederation until the 1960s, the Liberals – the party of Catholics and those less adamant about the connection to Britain – were generally stronger in Quebec and the Maritimes than in other parts of Canada. Indeed, the forty-year period from the 1920s to the 1950s was defined large- ly by regional parties and regional support. As Carty and colleagues (2000, 19) put it, “regional brokerage seemed so much the essence of Canadian politics that it was difficult to imagine that this party system might end.” But that is exactly what happened. Compared with the second party sys- tem, both the Progressive Conservatives and the New Democrats broad- ened their support base during the period that began in the 1960s. The Liberals, though, actually exhibited the reverse pattern: they contracted

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

164 Blake Andrew, Patrick Fournier, and Stuart Soroka

geographically in relation to the previous system, in which they had domin- ated nationally (Johnston et al. 1992). Nonetheless, the third system was marked by nation building and pan-Canadian politics. This was most nota- bly the nature of the Trudeau governments, and it culminated with the Mulroney majorities of the 1980s.

With the explosion of the third party system – and the birth of a fourth one – in 1993, regionalization of party support again became the norm. The Reform Party, the Canadian Alliance, and the Bloc Québécois all de- pended heavily on geographic pockets of support. The major, formerly pan- Canadian parties do so as well, with the Conservatives based strongly in the West and the Liberals relying on Ontario and Atlantic Canada. It is also important to note that regional divides are not simply the result of differing population profiles (Gidengil et al. 1999). Even after controlling for a myriad of socio-demographic characteristics, the significance of region as a pre- dictor of vote choice remains very strong (Nevitte et al. 2000; Blais et al. 2002; Gidengil et al. 2012). Indeed, the regional cleavages in Canada have been noted as the greatest in all Western democracies except Belgium (Dalton 1996, 325).

Regionalization is not limited to the basis of party support. It also ex- tends to the patterns of campaigning (Cross 2004). More and more, Can- adian political parties are running regionally targeted campaigns. Based on internal polling and analysis of past election trends, campaign organizers are highly informed about where their support base resides (and where it does not). This knowledge has meant that electoral strategies, directions, and communications are becoming increasingly regionalized.

In the analyses that follow, our goal is to further test the regionalization hypothesis through the lens of media coverage. That voting is regionalized is indisputable. Whether media representations of party competition are too is another matter, however, and one that has been barely explored thus far. We aim to determine whether or not the media’s treatment of the 1993- 2008 campaigns reflects growing regional divides and perspectives.

Parties, Issues, and Polarization Why does support for each political party fluctuate across elections? Clearly, many factors could be at play – notably socio-demographic characteristics, fundamental beliefs, partisan identification, leader evaluations, economic perceptions, government performance, and strategic considerations – either because the values of these factors or their effects on vote choice change from one election to another. Policy issues are another potential cause.

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

165The Canadian Party System

Dynamics of party support could be a consequence of some combination of (1) the shifting priorities and policy preferences of Can adian voters, and (2) the shifting priorities and policies of Canadian parties. The movements in vote shares of Canadian parties have received a good deal of attention, much more than the extent to which either voters or parties have changed in terms of policy objectives and priorities.

Where voters are concerned, we can glean some information from re- search on public opinion. Two strands of the literature are useful in this regard. First, there are studies of agenda setting. They have documented that the ebb and flow of the importance attached by the public to different issues stems from the amount of media attention and government activity devoted to the topics (Iyengar and Kinder 1987; McCombs and Shaw 1972; Soroka 2002). Second, there is work on issue ownership. Political parties are perceived to be more competent than others in certain policy domains (Budge and Farlie 1983; Petrocik 1996; Bélanger 2003). While issue owner- ship is generally quite stable over time, reversals do occur. Fluctuations in issue competence can be explained by variations in party popularity and government performance, as well as by party system change – namely, the appearance of new parties (Bélanger 2003).

Surveys can only go so far, however. Election questionnaires typically include a small number of policy attitudes. Also, issues come up unexpect- edly during campaigns, and they are not always captured by survey instru- ments. Even with a perfectly policy-focused questionnaire, the link between issues and parties cannot be adequately explored using survey data alone. Surveys measure those issues that are most important to voters, but insofar as these issues do not mirror exactly parties’ priorities, they miss part of the story. Party system change, where party policy is concerned, is reflected partly by shifts in what citizens want to buy, and also by shifts in what par- ties want to sell.

Media coverage can inform us about the variations in the overall sali- ence of policy issues. The main advantage of news content is that it will likely reflect some combination of voters’ (and journalists’) interests and parties’ campaign foci. The main disadvantage is that the two cannot be easily disentangled. In short, the methods we use below cannot distinguish between content that reflects what voters might think about when they think about the Liberals, and what the Liberals want voters to think about when they think about the Liberals. Nevertheless, campaign media content tells us what elections were about and what they are remembered for. These are the stories told during the campaign.

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

166 Blake Andrew, Patrick Fournier, and Stuart Soroka

The analyses below explore the connections between policy issues and political parties as they are captured in news content. Results provide infor- mation not only on the shifting attentiveness to issues over time but also on the varying links between parties and issues, both within a given election and from one election to the next. Results may also speak to the potential polarization of party politics in Canada. Are parties increasingly linked to different policy domains? More specifically, are Conservatives now more clearly linked to issues on the right, while other parties are more clearly linked to issues on the left? Ultimately, we are interested in whether the post-1993 period should be viewed as a watershed moment in terms of party polarization in Canada.

A Note: A Changing Media Environment? Our data reflect not just changes in party politics and election campaigns but also changes in the media environment. Indeed, the dynamics outlined above could be strengthened, or weakened, by transformations in the media industry.

During the period under study, most mainstream media, particularly newspapers, faced a rapid and sizable decrease in both audiences and rev- enues. This may have influenced the quantity and perhaps quality of cam- paign coverage. Both newspapers and television news outlets (public and private) had fewer reporters on the campaign trail, and many turned more and more to newswire services as a source of information. As a result, vari- ation across newspapers may be limited, and/or media coverage may in- creasingly focus on topics that are easier to treat: horserace and polls, party leaders, regional heavyweights, and long-standing issue reputations. Indeed, consumer demand may also be pushing for this type of coverage (Iyengar et al. 2004). There is already ample evidence of a large focus on horse race coverage in recent Canadian elections (Soroka and Andrew 2010), but there is no empirical research focusing more broadly on the ways in which the nature of the news industry has affected news content in Canada. The latter will not be our focus either, though we should not discount the pos- sibility that some of what we find (or do not find) is a consequence not just of shifts in electoral competition and party politics but also of the nature of the media industry.1

Methods Our analysis is based on an automated content analysis of a database con- taining over twenty-seven thousand stories from the 1980 to 2008 election

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

167The Canadian Party System

campaigns. The database includes all stories on the last six federal elections published in five English-language newspapers: Globe and Mail, Calgary Herald, Mont real Gazette, Toronto Star, and Vancouver Sun. The database also includes Globe and Mail election coverage of the 1980, 1984, and 1988 federal campaigns. Stories were extracted from a combination of the Nexis and Factiva full-text news indices.

For the sake of simplicity, particularly in terms of automated content analysis but also where the discussion of the party system is concerned, we focus here on English-language print media content. With that (significant) limitation in mind, the selection of newspapers was intended to capture some regional variation. The Globe and Mail is the largest nationally circu- lated newspaper in Canada, while the Herald, Gazette, Star, and Sun are widely distributed dailies in Canada’s four most populated regional mar- kets. All papers produced exclusive daily reports from the campaign trail during each of the campaigns included in the analysis.

The reason for the decision to use newspapers rather than television is partly substantive and partly pragmatic. Substantively speaking, although television was undoubtedly the main source of campaign information for many Canadians during these campaigns, about 80 percent of Canadian voters typically pay some attention to print media during campaigns, and newspapers tend to be the main source of news for about 30 percent of voters on average (Gidengil et al. 2004), so newspapers are a significant source of campaign news. They also serve an agenda-setting function for television, and recent analyses find few significant differences in content, at least in terms of subject focus, between newspapers and television (Soroka 2002). More pragmatically, television transcripts are not available for the entire twenty-eight-year period surveyed here, and when they are, they capture only the audio content. Video content is obviously a critical com- ponent of television news and has a strong influence on the way audiences process the audio component of a given TV news story (Graber 2001). Even when it is available, video content is notoriously difficult to code, and, because much of the story information is communicated through video, television content tends to have far fewer words than newspaper content.

The volume of words in news stories is especially important to our re- search here, since words are the unit of analysis for our automated content analysis. Computer automation has become a mainstay of empirical re- search in the study of political communication. Since the 1950s, scholars have been developing computer-assisted methods to analyze textual infor- mation in new and interesting ways. Given the ever-increasing volume of

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

168 Blake Andrew, Patrick Fournier, and Stuart Soroka

political information, automation enables the processing of enormous bodies of data efficiently and effectively. Reliability – the degree to which all stories are coded using exactly the same decision rules – is a key advantage of computer-automated content analysis. And validity – the extent to which automated analyses produce results that are in line with what we would find with trained human coders – is typically much better for text-oriented newspaper stories than for video-oriented television reports.

There are of course numerous approaches to automation. The dictionary- based approach used here to identify parties and leaders and to assign topics is the oldest and simplest type of computer automation, and has been widely used in the social sciences. The method, often referred to as fre- quency analysis, involves calculating a simple word count of definitive key- words in a predefined categorical dictionary. The relative frequency of words from dictionary categories can then be analyzed to classify various texts. With a well-defined dictionary, a basic word count provides a surprisingly powerful analysis of the topical composition of a text. There are many dif- ferent programs available to implement computer automation. Here we use Lexicoder, our own Java-based multi-platform software that implements frequency analysis. (The program, available at http://www.lexicoder.com, can be adapted for any number of categorical dictionaries.)

Of course, the validity of frequency analysis depends on the quality of the coding dictionary. The real challenge is to develop a conceptually sound, comprehensive lexicon that is valid over time and across diverse corpora. Numerous machine-readable dictionaries are available for automation. Many of the most well-established content-analytic lexicons focus on psychological or cognitive categories (Martindale 1975, 1990; Namenwirth and Weber 1987; Stone et al. 1966). There are a number of sentiment and affect lexicons to automate the tone of texts, but far fewer to automate topics.

The first dictionary used here is very straightforward: a list of all major party and major party leader names at the time of each federal campaign between 1993 and 2008. The full list includes party names (and derivatives) of the following major party labels during the twenty-eight-year period: Liberals, Progressive Conservatives, Conservatives, Reform, Canadian Al- liance, NDP, and Bloc Québécois. The dictionary also includes all party leaders in each of the six campaigns: (1980) Pierre Trudeau, Joe Clark, and Ed Broadbent; (1984 and 1988) John Turner, Brian Mulroney, and Ed Broadbent; (1993) Jean Chrétien, Lucien Bouchard, Preston Manning,

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

169The Canadian Party System

Audrey McLaughlin, and Kim Campbell; (1997) Jean Chrétien, Preston Manning, Gilles Duceppe, Alexa McDonough, and Jean Charest; (2000) Jean Chrétien, Stockwell Day, Gilles Duceppe, Alexa McDonough, and Joe Clark; (2004) Paul Martin, Stephen Harper, Gilles Duceppe, and Jack Layton; (2006) Stephen Harper, Paul Martin, Gilles Duceppe, and Jack Layton; (2008) Stephen Harper, Stéphane Dion, Gilles Duceppe, and Jack Layton. In most cases, the last name of the party leader alone captures lead- er mentions; this is more difficult for Stockwell Day, for instance, although even here we can use a capital “D” to distinguish leader mentions using last names only.

The topic dictionary used here is the most recent version of the Lexicoder Topic Dictionary (LTD). It is based on a “seed list” of five to fifteen paradig- matic words for each topic, reflecting the core policy focus for that domain, and generated manually by expert coders with reference to comprehensive policy indexes, particularly the Policy Agendas Project coding scheme (see http://www.policyagendas.org). The seed list is then expanded using WordNet, a comprehensive lexical database of the English language (Miller et al. 1990; see http://wordnet.princeton.edu). The topic dictionary used here is freely available at http://www.lexicoder.com. It is worth stressing that the number of keywords varies by topic, and therefore cross-topic comparisons for a single campaign should be viewed with caution. (Simply put, topics with greater numbers of identifiable keywords will tend to show greater frequencies.) We will make some cross-sectional inferences below, but will focus on longitudinal analyses, for which our data are better suited.

For the time being, the important point is that, with such a dictionary, frequency scores can be generated for parties, leaders, and policy domains. For our cross-analysis of parties and issues, we rely on a new proximity feature in Lexicoder – namely, we look at the co-occurrence of topic key- words and party/leader names within a single story and also within the same sentence. These co-occurrences can be easily and reliably (indeed, as reliability is defined above, perfectly reliably) tracked across news content from the past six federal election campaigns.

For the first set of tests regarding regionalization, analyses are very straightforward comparisons of the relative frequency of party names. Basic dataset descriptives are shown in Table 8.1, which indicates the num- ber of stories collected from each newspaper in the sample. All stories published in the respective papers during the official campaign period

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

170 Blake Andrew, Patrick Fournier, and Stuart Soroka

have been included. Note that campaigns varied somewhat in length dur- ing the period of our study. The longest campaign occurred in 2006, while the shortest occurred in 1997, 2000, and 2004.

Results Tables 8.2 and 8.3 present some initial trends in party coverage over time. Table 8.2 shows how frequently each party’s name (or its leader’s name) oc- curred across all of the newspapers included in the analysis, during each campaign. Party/leader mentions for each party are expressed as a percent- age of all party/leader mentions in that campaign. For instance, in 1980 the Liberal Party’s share of coverage (47.4 percent) is calculated by dividing the number of times that party was mentioned by the total mentions of all main parties/leaders during that campaign period.

TABLE 8.1 Dataset descriptives, number of stories collected from newspapers during national election campaigns

Newspaper 1980 1984 1988 1993 1997 2000 2004 2006 2008

Calgary Herald 908 516 653 656 940 611 Globe and Mail 1,053 1,390 1,061 506 672 785 968 1,455 828 Montreal Gazette 941 612 672 632 833 635 Toronto Star 1,443 841 921 945 1,152 1,040 Vancouver Sun 904 507 563 595 737 354

Total 1,053 1,390 1,061 4,702 3,148 3,594 3,796 5,117 3,468

TABLE 8.2 Party mentions in different newspapers, by election

Party 1980a 1984a 1988a 1993 1997 2000 2004 2006 2008

Liberal 47.4 47.2 40.0 30.0 31.4 42.7 44.1 41.3 34.5 PC/Conservative 40.5 39.4 41.2 33.8 25.9 18.9 37.6 40.9 49.3 Reform/CA – – – 17.7 23.7 23.9 – – – NDP 12.1 13.4 18.8 10.1 8.7 8.6 14.0 13.2 12.8 BQ – – – 8.4 10.2 5.9 4.2 4.7 3.4

N 11,153 18,319 15,093 60,384 44,994 44,559 52,597 62,815 34,252

Note: Cells contain party mentions as a percentage of all party and leader mentions in an election. a Globe and Mail only.

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

171The Canadian Party System

Two key points are illustrated in Table 8.2. First, 1993 was clearly a water- shed for party visibility. All parties holding seats in the House of Com mons during the 1980s suffered a significant decline in commentary directed to- wards them during the 1993 campaign. In particular, Liberal and NDP share of mentions dropped by 33 percent or more from their respective 1988 levels.2 The 1990s (including the 2000 election) marked a significant period of decline in relative visibility for the popular parties of the 1980s, a trend clearly precipitated by the arrival of the Reform Party/Canadian Alliance and the Bloc Québécois.

That being said, viewed in the context of the almost thirty-year span of our dataset, it is also clear that more recent elections (2004, 2006, and 2008) bear a much stronger resemblance to campaigns of the 1980s than to those of the 1990s. Indeed, by 2004 all of the major players in Canada’s political party landscape were, for the most part, fully restored to 1980s standards in terms of media coverage. This return to pre-1993 levels fits with the restorative patterns identified by Richard Johnston in Chapter 13: periods of flux and volatility followed by a return to earlier, more stable patterns.

Table 8.3 reinforces these points, showing coverage of “main” parties by article, where “main” is the party/leader name mentioned most in an arti- cle. Whereas Table 8.2 treats all party/leader mentions equally and ignores articles as a unit of analysis entirely, Table 8.3 focuses on articles and as- signs each a main party. There are some advantages to thinking about arti- cles that focus most on one party or the other: the main party as measured here is very likely the party a typical newspaper reader would recall if asked to explain what a given story was about; it is the party name he or she would

TABLE 8.3 Main party mentions in different newspapers, by election

Party 1980a 1984a 1988a 1993 1997 2000 2004 2006 2008

Liberal 46.6 55.2 40.6 29.5 38.9 50.6 49.1 46.3 26.4 PC/Conservative 44.1 36.5 48.1 40.0 21.9 14.7 38.2 42.2 64.1 Reform/CA – – – 16.5 21.9 24.1 – – – NDP 9.3 8.3 11.3 7.1 7.6 6.2 10.6 8.5 8.1 BQ – – – 7.0 9.8 4.4 2.2 3.1 1.4

N 850 1,092 833 3,671 2,458 2,820 2,978 3,962 2,399

Note: Cells contain “main” party mentions as a percentage of all articles mentioning at least one party or one leader. a Globe and Mail only.

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

172 Blake Andrew, Patrick Fournier, and Stuart Soroka

come across first and be reminded about most frequently in the course of reading through the full story.

That said, the results in Table 8.3 are very similar to those in Table 8.2. Our measure of main party points to major changes during the 1990s, but also to a pattern, in recent years, of continuity with campaigns of the 1980s. The most striking indication of continuity is media coverage allotments for the Liberal and Conservative parties. During the 1980s, the combined share of Conservative and Liberal main party coverage was about nine articles out of ten. Comparatively, for the three elections since 2000, the combined main party share for those two parties has been 8.9 out of 10. Clearly the party merger on the right prior to the 2004 election is the significant driver of this result. But it is also noteworthy that since 1997 the Bloc Québécois has virtually disappeared from English-language media coverage insofar as its being the main party of election news stories.3

Tables 8.2 and 8.3 suggest that, when viewed in the aggregate, recent media coverage of Canadian politics shares much in common with cover- age patterns of the 1980s. The key outlier is the Conservative advantage in main party mentions in 2008. Nearly two-thirds of stories about the 2008 campaign featured the incumbent Conservatives. The next closest party was featured in about one of every four stories (26 percent). Where main party mentions are concerned, 2008 clearly stands alone – not just from campaigns of the 1980s and ’90s but also from the campaigns of 2004 and 2006. The gap between incumbent and main rival was much wider (38 per- cent) and the intensity of focus on the incumbent party (64 percent of stor- ies) was never stronger than in 2008.

What about regionalization in Canadian media coverage of election campaigns? The prominence of party coverage should correspond with that party’s regional support base; it should also, relatedly, mirror to some de- gree the attention each party pays to a given region. Most importantly, we expect to find evidence of growing regionalization since 1993. Have con- sumers of regionally based newspapers been exposed to increasingly dis- tinct narratives of Canadian party politics over the past fifteen years? Four of the five papers in our analysis are regionally based, and not widely dis- tributed outside their respective markets. We explore cross-sectional dif- ferences between these papers – the Sun, Herald, Star, and Gazette – here.

Table 8.4 shows, above all, that variance in party coverage is driven more by party strength in national elections than by party strength in regions. When a party’s prominence in one regional newspaper went up (or down), it also tended to go up (or down) in other regional newspapers at the same

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

173The Canadian Party System

time. Consider, for instance, the Liberals, who consistently trended upward in all regional papers from 1993 to 2000, and then trended downward in all papers from 2004 to the present. It is no small coincidence, we suggest, that the pattern of Liberal media attention in all regional papers follows the tra- jectory of Liberal Party power in Canadian politics over the past fifteen years. Conservative coverage tells a similar story (see Figure 8.1).

TABLE 8.4 Main party mentions, by regional newspaper

Vancouver Calgary Toronto Montreal Election Party Sun Herald Star Gazette

1993 Liberal 25.6 24.3 29.8 32.5 PC 42.3 36.1 44.0 37.3 NDP 9.7 3.6 8.8 5.5 Reform 17.9 31.3 13.1 8.5 BQ 4.4 4.8 4.4 16.1

1997 Liberal 41.5 36.3 43.5 41.8 PC 20.7 20.8 21.8 17.7 NDP 9.9 6.0 8.4 4.9 Reform 24.7 31.5 19.0 14.0 BQ 3.2 5.5 7.4 21.6

2000 Liberal 62.2 53.3 52.3 54.8 PC 11.2 15.4 13.7 12.8 NDP 5.5 4.2 6.9 5.3 Canadian Alliance 20.1 25.9 24.8 14.2 BQ 1.0 1.1 2.3 12.9

2004 Liberal 46.1 49.9 49.4 52.6 PC 38.2 42.2 39.4 30.5 NDP 14.9 7.0 9.7 10.5 BQ 0.9 1.0 1.6 6.3

2006 Liberal 45.9 45.2 50.2 47.6 PC 41.8 46.4 38.1 36.9 NDP 10.1 7.0 10.1 6.8 BQ 2.2 1.4 1.6 8.7

2008 Liberal 31.9 20.3 28.7 25.2 PC 55.9 73.7 61.0 62.6 NDP 11.4 5.5 9.5 7.8 BQ 0.9 0.5 0.8 4.4

Note: Cells contain main party mentions by party, as a percentage of all articles mentioning at least one party.

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

174 Blake Andrew, Patrick Fournier, and Stuart Soroka

Note, however, that consistency in the ebb and flow of party prominence does not mean that party visibility was always identical in these papers. Relative to other regional papers, the Calgary Herald has focused more on parties of the right since 1993. Similarly, the Vancouver Sun has tended to

P e rc

e n t

m a in

p a rt

y m

e n ti

o n s

(a ll

s to

ri e s )

1993 1997 2000 2004 2006 2008

P e rc

e n t

m a in

p a rt

y m

e n ti

o n s

(a ll

s to

ri e s )

1993 1997 2000 2004 2006 2008

Liberal

PC/Conservative

Calgary Herald Montreal Gazette

Toronto Star Vancouver Sun

Year

FIGURE 8.1 Main party election coverage of the Liberal and PC/Conservative Parties in regional newspapers

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

175The Canadian Party System

print more coverage focused on the NDP than others during the same time period. The key point is that while regional differences are apparent in these papers, there has been virtually no sign that regional variance has been in- creasing (or decreasing) since 1993. Any changes in party prominence that occurred since 1993 have, for the most part, co-occurred in all papers and in the same direction.

The major caveat in our argument pertains to coverage of the Bloc Québécois. It is clear from Table 8.4 that English-language media have in- creasingly relegated the BQ to the sidelines of Canadian election coverage. Even the Montreal Gazette has dramatically reduced the attention paid to the BQ over the past fifteen years. In 1993-97, about one in five election stories published by the Gazette focused on the Bloc, compared with less than one in twenty stories by 2008. This decline is both striking and some- what prescient, considering that the BQ’s electoral success had, until its dramatic collapse in 2011, been the most consistent of any party in Canadian federal politics since 1993. We should also stress that the BQ’s salience in French-language media has not declined significantly in recent years (see Andrew et al. 2006, 2008; Goodyear-Grant et al. 2004).

In short, and with the important caveat regarding Bloc Québécois cover- age in mind, our results point to a clear conclusion with respect to region- alization. It is true that those who read the Herald, Star, Sun, or Gazette during the past fifteen years were exposed to a unique dose of coverage about the major parties, but it is also true that changes in party visibility tended to occur simultaneously across the various geographic regions of Canada. Save for coverage of the Bloc Québécois, our data do not support the hypothesis that regionalization of party coverage has been growing since 1993.

As an introduction to the polarization analysis, Table 8.5 provides an overview of issue salience in election campaigns since 1980. It reports the mean number of topic keywords per article. A score of 1 indicates that a word relating to that topic appears once in every election article on average, 0.5 indicates that a word relating to that topic appears in one of every two articles, and so on. Trends in issue salience are important to the polariza- tion story, because they indicate whether issue ownership in media cover- age might have translated into electoral advantages. It is worth reiterating that issue ownership, as we conceive of it in campaign media, may not ne- cessarily produce electoral benefit for parties or leaders. That depends on a host of other factors, including the credibility of the party on the issue, the importance voters place on the issue, and the priority of the issue on the

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

176 Blake Andrew, Patrick Fournier, and Stuart Soroka

party’s agenda. What we document below is but one piece of a bigger puz- zle, but it is nevertheless a piece of the puzzle about which little is known, and one that likely matters for how both voters and parties prioritize issues in the first place.

Note, first of all, that there are several particular elections in which a specific issue played a larger than normal role. For example, the volume of keywords related to international trade spiked during the free trade election campaign of 1988. Similarly, energy played a key role in the 1980 campaign, as did employment during the mid-1990s. To anyone familiar with Can- adian politics over this period, many of the spikes in salience are highly in- tuitive given the context of political discourse leading up to each campaign.

TABLE 8.5 Issue coverage in different newspapers, by election

Issue 1980a 1984a 1988a 1993 1997 2000 2004 2006 2008

Agriculture/ forestry/fishing 0.08 0.10 0.11 0.05 0.05 0.05 0.06 0.04 0.05

Civil rights 0.10 0.12 0.11 0.12 0.12 0.12 0.21 0.18 0.08

Crime and justice 0.17 0.22 0.25 0.24 0.33 0.34 0.28 0.49 0.31

Economy 0.50 0.36 0.40 0.44 0.37 0.36 0.34 0.30 0.43

Education 0.11 0.08 0.10 0.16 0.15 0.16 0.14 0.15 0.15

Employment/ labour 0.12 0.22 0.15 0.32 0.25 0.11 0.09 0.09 0.14

Energy 0.49 0.07 0.10 0.03 0.03 0.05 0.09 0.07 0.15

Environment 0.03 0.01 0.10 0.03 0.03 0.05 0.04 0.04 0.07

Finance/ commerce 0.15 0.17 0.24 0.16 0.12 0.15 0.11 0.11 0.23

Foreign affairs/ defence 0.76 0.67 0.73 0.61 0.65 0.54 0.67 0.70 0.69

Health 0.04 0.04 0.05 0.12 0.17 0.26 0.28 0.17 0.10

Immigration 0.02 0.01 0.02 0.04 0.02 0.03 0.02 0.03 0.03

Social welfare 0.08 0.07 0.07 0.12 0.11 0.11 0.12 0.12 0.12

Trade (international) 0.12 0.10 0.41 0.13 0.08 0.07 0.08 0.09 0.07

Transportation 0.07 0.06 0.04 0.04 0.03 0.03 0.04 0.03 0.05

Note: Cells contain the mean number of topic keywords per article. a Globe and Mail only.

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

177The Canadian Party System

What is perhaps less predictable about issue coverage over time is that there is little evidence of any consistent trends in issue salience. No issue represented in this table has consistently grown in prominence over time, just as no issue has consistently faded from election discourse. There have been ebbs and flows over the course of two or three campaigns, but in no case has there been a consistent pattern. Indeed, from the perspective of this aggregate snapshot, it is clear that no issue, and certainly no set of issues (left or right), has systematically risen (or fallen) in salience from 1993 onward.

Our tests of issue ownership at the party level are included in Tables 8.6 to 8.9. Recall that the conventional wisdom is that parties became not only more regionally entrenched in 1993 but also more ideological in nature. If parties are truly increasingly differentiating themselves along ideological fault lines, then party ownership of specific issues should show signs of strengthening over the span of our study. We address the polarization de- bate here by looking at whether the occurrences of issue keywords in elec- tion media content have indeed become more (or less) associated with the various parties over time.

Tables 8.6 to 8.9 present data on within-sentence co-occurrence of party/ leader names and issue keywords. To be clear: the tables show the average number of topic keywords co-occurring in sentences with party/leader names within each article. A value of 1 would suggest that, on average, arti- cles include at least one topic keyword in the same sentence as, for instance, a Liberal Party/leader mention. Values can of course be much higher – in an article focusing on the NDP and health care, there may be 25 health care words appearing in sentences that also include the NDP party or leader name. Averages across all articles tend to produce much smaller values, since articles cover a wide range of topics. So, for instance, the top-left cell in Table 8.6 indicates the average number of health care keywords in sen- tences that also mention the Liberals, across all articles mentioning the Liberals in the 1980 campaign: 0.025, which means 0.025 health care key- words in all sentences mentioning the Liberals, or roughly 1 keyword in sentences mentioning the Liberals for every 40 articles. (That’s not a lot of health care coverage.)

Note that these statistics capture in part the varying volume of coverage of the different parties – if a party is mentioned more, then there are a greater number of sentences in which issues may be mentioned. That prom- inence is included here intentionally – the data should reflect the promin- ence of co-occurring party and issue mentions. Note also that these

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

178 Blake Andrew, Patrick Fournier, and Stuart Soroka

statistics are not entirely independent of any shift over time in the volume of horserace coverage (Soroka and Andrew 2010). If horserace-oriented coverage increases (or decreases) consistently for parties/leaders, and men- tions of any party/leader in that frame tend to occur in sentences where policy mentions are minimal, then our data will show diminishing (or ris- ing) figures for issue/party co-occurrences. Again, this is intentional – our data should reflect the prominence of co-occurring party and issue mentions.

We focus in these tables on four prominent issues in Canadian politics: health care, foreign affairs/defence, crime/justice, and the economy. These are also issues that vary from election to election in terms of salience, as shown in Table 8.5. Broadly speaking, our expectation is that issue owner- ship will show signs of strengthening over time. For instance, we expect gaps between the parties most and least associated with a given topic to widen over the period of the study. This would indicate, from a news media vantage point at least, that parties have become increasingly polarized.

For health care, as shown in Table 8.6, there is no evidence that differ- ences between parties have increased over time. The gap between the par- ties most and least associated with health care in 1980 (NDP versus Liberals) is quite similar to the gap in 2008 between the Conservatives and the BQ. Although there is some indication that health care ownership was occur- ring between 1997 and 2004 for the Liberals, differences between parties in terms of association with this issue have almost completely faded in the two most recent campaigns.

TABLE 8.6 Health care coverage in different newspapers, by party

Election Liberal PC/Conservative Reform/CA NDP BQ

1980a 0.025 0.026 – 0.010 – 1984a 0.035 0.023 – 0.012 – 1988a 0.043 0.041 – 0.033 – 1993 0.040 0.057 0.073 0.037 0.004 1997 0.183 0.088 0.080 0.054 0.017 2000 0.289 0.071 0.196 0.092 0.013 2004 0.380 0.238 – 0.079 0.010 2006 0.125 0.134 – 0.091 0.004 2008 0.059 0.062 – 0.046 0.001

Note: Cells contain the mean number of topic keywords in sentences that also mention party/leader names. a Globe and Mail only.

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

179The Canadian Party System

The story is similar for foreign affairs/defence, crime/justice, and the economy, as shown in Tables 8.7 to 8.9. There is no indication that the major shakeup in the party system that occurred in 1993 coincided with a trend towards issue ownership in any of these domains. Simply consider the tra- jectory of ownership in these areas for the Liberals and Conservatives. Since 1980, neither party could claim to have clearly seized control of any issue represented here. Indeed, although both parties have, at one point or another, laid claim to each of these issues, there is relatively little evidence for enduring ownership, or increasing ownership, in any area by either the

TABLE 8.7 Foreign affairs/defence coverage in different newspapers, by party

Election Liberal PC/Conservative Reform/CA NDP BQ

1980a 0.775 0.701 – 0.221 – 1984a 1.001 0.766 – 0.191 – 1988a 0.872 0.850 – 0.336 – 1993 0.494 0.622 0.296 0.155 0.125 1997 0.598 0.475 0.487 0.188 0.186 2000 0.629 0.242 0.323 0.107 0.065 2004 0.829 0.835 – 0.228 0.067 2006 0.777 0.777 – 0.192 0.071 2008 0.480 0.814 – 0.138 0.037

Note: Cells contain the mean number of topic keywords in sentences that also mention party/leader names. a Globe and Mail only.

TABLE 8.8 Crime/justice coverage in different newspapers, by party

Election Liberal PC/Conservative Reform/CA NDP BQ

1980a 0.117 0.130 – 0.029 – 1984a 0.271 0.166 – 0.045 – 1988a 0.161 0.174 – 0.059 – 1993 0.095 0.138 0.097 0.036 0.028 1997 0.235 0.193 0.193 0.033 0.036 2000 0.277 0.126 0.215 0.036 0.043 2004 0.197 0.301 – 0.043 0.018 2006 0.425 0.363 – 0.099 0.021 2008 0.158 0.364 – 0.045 0.015

Note: Cells contain the mean number of topic keywords in sentences that also mention party/leader names. a Globe and Mail only.

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

180 Blake Andrew, Patrick Fournier, and Stuart Soroka

Liberals or Conservatives. Ultimately, parties tend not to be clearly associ- ated with issues at all, and ideological polarization is simply not a reality, at least not in terms of media coverage.

There is evidence of Conservative strength in the 2008 election, to be sure. During that election, the incumbent governing party surged ahead in terms of association with foreign affairs/defence, crime/justice, and the economy. Liberals showed no advantage (indeed, a slight disadvantage) in the one “left” issue investigated here, health care. The 2008 Conservative campaign appears to have worked where media content is concerned, but this was not part of a long-term trend that began years before, and it is too soon to tell whether it is a shift that will last well into the future.

Conclusion This chapter began by making the argument that we might learn something useful about party politics and electoral competition in Canada by looking at campaign-period media coverage. The analyses here have covered the past nine federal elections before the most recent one, in 2011, and have, we believe, confirmed this hypothesis. Content analysis of media data – particularly automated analyses of large bodies of media data – offer an opportunity to examine a number of critical questions with somewhat more detail, or at least different detail, than the more traditional data sources permit.

The critical questions we have explored here focus on regionalization and connections between parties and issues. Changes in the number of

TABLE 8.9 Coverage of the economy in different newspapers, by party

Election Liberal PC/Conservative Reform/CA NDP BQ

1980a 0.479 0.550 – 0.135 – 1984a 0.352 0.398 – 0.135 – 1988a 0.357 0.416 – 0.177 – 1993 0.373 0.335 0.165 0.052 0.037 1997 0.403 0.330 0.240 0.101 0.048 2000 0.441 0.131 0.247 0.084 0.013 2004 0.460 0.447 – 0.015 0.013 2006 0.316 0.370 – 0.083 0.014 2008 0.474 0.626 – 0.130 0.020

Note: Cells contain the mean number of topic keywords in sentences that also mention party/leader names. a Globe and Mail only.

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

181The Canadian Party System

parties make it difficult to directly compare the levels of regionalization that existed in the third and the fourth party systems, but regionalization does not appear to have increased in the six elections since 1993. To be clear: there are certainly differences in party coverage across regions, but those differences are no greater now than they were twenty years ago. Issue/ party linkages are somewhat similar; linkages between parties and issues vary over time, but there is little evidence of increasing (or increasingly di- vergent) issue ownership by parties in the recent past. If issue/party con- nections are an indication of polarization, then on this front there are no striking differences between the current system and the one it replaced.

Overall, then, our results point to a story of continuity rather than change in media coverage of the party system. Our findings are at odds with some previous work (and work in this volume) finding evidence of more profound changes in the distribution and sources of electoral support for Canadian parties. They also conflict with current, popularized accounts of an increas- ingly ideologically divided political process.

What accounts for this difference in results? One possibility is that the more common view – focusing on profound change in the early 2000s – is simply incorrect. There has been change in the Canadian party system, but that change has generally been overstated. We have shifted to a party sys- tem that combines features of the party system both pre- and post-1993: essentially, a united right but with the addition of one strongly regional party in Quebec. This changes the nature of party competition, to be sure, as is readily apparent in other chapters in this volume, but the basic struc- ture of politics in Canada has actually changed somewhat less than is com- monly believed.

There are of course different possibilities. Others may be wrong, but we may be as well. It is possible that that media content has not accurately reflected the dynamics of the party system in Canada. So perhaps politics is increasingly regionalized and divisive, but media coverage (1) is not being drawn in the same direction, or (2) was always regionalized and divisive. Intuitively, the latter seems the more likely possibility. We are not inclined to believe either option, however. Past work shows that media coverage tends to capture the state of public affairs rather well (e.g., issue attentive- ness, party support, and the state of the economy). We expect that it does here also.

That said, there are some weaknesses in the way we have used media data to capture the nature of party competition. We do not deal with the tone of media content here, just frequencies of coverage. Shifts in regionalization

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

182 Blake Andrew, Patrick Fournier, and Stuart Soroka

may be more apparent when we consider both the frequency and the tone of party/leader mentions. So too might polarization. When references to issues are identical in quantity across parties (as was the case with foreign affairs in 2006, for instance), perhaps one party is discussed in a more posi- tive light than another. In fact, an American study argued that parties’ issue reputations influence the favourability of news treatments: candidates are covered more positively when reports focus on their own issues rather than their opponents’ issues (Hayes 2008). We intend to examine this more closely in forthcoming work. We also want to consider coding that cap- tures the possibility that attention to an issue remains roughly equal across parties but that parties have increasingly divergent positions on the issue.

Accurately capturing tone and party positions in media content is of course no easy matter, and for the time being we are satisfied with the sim- pler but more reliable measures used here. These measures suggest that Canadian party politics is in some critical ways not fundamentally differ- ent now from the 1980s. The Conservatives seem to have been particularly dominant on several “right” issues in 2008. This may well be the beginning of a new party system; it is too soon to tell. But it is not part of a general trend over the past thirty years. The media’s recent portrait of party com- petition in Canada is markedly similar to that of the past two or three decades.

Notes 1 There is of course a body of work that focuses on links between party and electoral

politics and the structure of mass media. Innis’s work (1951) was among the first (and most profound) linking the structure of politics to the structure of communi- cation. See also Abramson et al. (1988) on the parallel evolution of mass media and party politics in the United States.

2 The Conservatives were not immune, losing almost a fifth of their coverage during the same period.

3 This timing coincides with the end of the party’s stint as Her Majesty’s Loyal Opposition.

References Abramson, Jeffrey B., F. Christopher Arterton, and Gary R. Orren. 1988. The Elec

tronic Commonwealth: The Impact of New Media Technologies on Democratic Politics. New York: Basic Books.

Andrew, Blake C., Antonia Maioni, and Stuart Soroka. 2006. “Just When You Thought It Was Out, Policy Is Pulled Back In.” Policy Options 27 (3): 74-79.

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

183The Canadian Party System

Andrew, Blake, Lori Young, and Stuart Soroka. 2008. “Back to the Future: Press Coverage of the 2008 Canadian Election Campaign Strikes Both Familiar and Unfamiliar Notes.” Policy Options 29 (10): 79-84.

Bélanger, Éric. 2003. “Issue Ownership by Canadian Political Parties 1953-2001.” Canadian Journal of Political Science 36 (3): 539-58.

Blais, André, Elisabeth Gidengil, Richard Nadeau, and Neil Nevitte. 2002. Anatomy of a Liberal Victory: Making Sense of the Vote in the 2000 Canadian Election. Peterborough, ON: Broadview Press.

Budge, Ian, and Dennis J. Farlie. 1983. Explaining and Predicting Elections: Issue Effects and Party Strategies in TwentyThree Democracies. London: Allen and Unwin.

Carty, R. Kenneth. 1988. “Three Canadian Party Systems: An Interpretation of the Development of National Politics.” In Party Democracy in Canada: The Pol itics of National Party Conventions, edited by George C. Perlin, 15-30. Scarborough, ON: Prentice Hall.

Carty, R. Kenneth, Lisa Young, and William P. Cross. 2000. Rebuilding Canadian Party Politics. Vancouver: UBC Press.

Cross, William. 2004. Political Parties. Vancouver: UBC Press. Dalton, Russell. 1996. “Political Cleavages, Issues, and Electoral Change.” In

Comparing Democracies: Elections and Voting in Global Perspectives, edited by Lawrence LeDuc, Richard G. Niemi, and Pippa Norris, 319-42. Thousand Oaks, CA: Sage.

Gidengil, Elisabeth, André Blais, Joanna Everitt, Patrick Fournier, and Neil Nevitte. 2012. Dominance and Decline: Making Sense of Recent Canadian Elections. Toronto: University of Toronto Press.

Gidengil, Elisabeth, André Blais, Richard Nadeau, and Neil Nevitte. 1999. “Making Sense of Regional Voting in the 1997 Canadian Federal Election: Liberal and Reform Support outside Quebec.” Canadian Journal of Political Science 32 (2): 247-72.

Gidengil, Elisabeth, André Blais, Neil Nevitte, and Richard Nadeau. 2004. Citizens. Vancouver: UBC Press.

Goodyear-Grant, Elizabeth, Stuart Soroka, and Antonia Maioni. 2004. “The Role of the Media: A Campaign Saved by a Horserace.” Policy Options 25 (8): 86-91.

Graber, Doris A. 2001. Processing Politics: Learning from Television in the Internet Age. Chicago: University of Chicago Press.

Hayes, Danny. 2008. “Party Reputations, Journalistic Expectations: How Issue Ownership Influences Election News.” Political Communication 25 (4): 377-400.

Innis, Harold A. 1951. The Bias of Communication. Toronto: University of Toronto Press.

Iyengar, Shanto, and Donald R. Kinder. 1987. News that Matters: Television and American Opinion. Chicago: University of Chicago Press.

Iyengar, Shanto, Helmut Norpoth, and Kyu S. Hahn. 2004. “Consumer Demand for Election News: The Horserace Sells.” Journal of Politics 66 (1): 157-75.

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

184 Blake Andrew, Patrick Fournier, and Stuart Soroka

Johnston, Richard, André Blais, Henry E. Brady, and Jean Crête. 1992. Letting the People Decide: Dynamics of a Canadian Election. Montreal and Kingston: McGill-Queen’s University Press.

Martindale, Colin. 1975. Romantic Progression: The Psychology of Literary History. Washington, DC: Hemisphere.

–. 1990. The Clockwork Muse: The Predictability of Artistic Change. New York: Basic Books.

McCombs, Maxwell E., and Donald L. Shaw. 1972. “The Agenda-Setting Function of Mass Media.” Public Opinion Quarterly 36 (2): 176-87.

Miller, George A., Richard Beckwith, Christiane Fellbaum, Derek Gross, and Katherine Miller. 1990. “Introduction to WordNet: An On-line Lexical Database.” International Journal of Lexicography 3 (4): 235-312.

Namenwirth, J. Zvi, and Robert Philip Weber. 1987. Dynamics of Culture. Boston: Allen and Unwin.

Nevitte, Neil, André Blais, Elisabeth Gidengil, and Richard Nadeau. 2000. Unsteady State: The 1997 Canadian Federal Election. Don Mills, ON: Oxford University Press.

Petrocik, John R. 1996. “Issue Ownership in Presidential Elections, with a 1980 Case Study.” American Journal of Political Science 40 (3): 825-50.

Soroka, Stuart N. 2002. AgendaSetting Dynamics in Canada. Vancouver: UBC Press.

Soroka, Stuart, and Blake Andrew. 2010. “Reporting Canadian National Elections.” In Mediating Canadian Politics, edited by Linda Trimble and Shannon Sampert, 113-28. Toronto: Pearson.

Stone, Philip J., Dexter C. Dunphy, Marshall S. Smith, and Daniel M. Ogilvie. 1966. The General Inquirer: A Computer Approach to Content Analysis. Cambridge, MA: MIT Press.

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .