The 2004 election campaign presented a very nar- row set of issues to Canadian voters. Opposition parties hammered Canadians with the following message: think about the sponsorship scandal, and pun- ish this Liberal Party for it. The Liberals, of course, coun- tered with a different reading of the scandal, and warned Canadians what Conservatives might do with social poli- cies (particularly same-sex marriage and abortion) and health care. With only a few fleeting exceptions, the parties’ campaigns were made up entirely of these few issues and that was reflected in media coverage. News content focused on sponsorship, social policy and health care, and " in the absence of any other policy content " the horse race.

Eighteen months later, on the cusp of winter tire sea- son and with an eight-week federal election campaign to look forward to, Canadians must have wondered about their prospects for something new. It came as a sur- prise to many that they got it. The 2006 election will certainly be remembered for the resulting change in Ottawa, but the campaign itself was also a memorable one. It was a campaign of ideas and issues. Policy platforms, above and beyond the government accountability issue, were not just on the table during the 2006 campaign, they actually took up most of the space there. Indeed, from a news media perspective, the anato- my of this Canadian election was very different from the last one. Canadians read much more about issues, and they were exposed to a much wider conversation about sub- stantive policy options.

How did this play out this time around? Once again, McGill University’s Observatory on Media and Public Policy took on the chal- lenge of analyzing print media cover- age of the election campaign. Our 2006 content analysis was directly comparable with the one gathered in 2004 (see Policy Options, September 2004). It relied on all campaign con- tent " news, editorials and opinion pieces " in seven major daily news- papers across Canada: the Vancouver Sun, Calgary Herald, Toronto Star, Le Devoir, La Presse, National Post and Globe and Mail. Beginning on Tuesday, November 29, a team of 12 coders were randomly assigned and rotated through newspapers during the campaign. In total, over 3,500 articles, drawn from the main news sections of each newspaper, were ana- lyzed in our study: 2,431 news stories and 1,319 editorial and opinion pieces. (Detailed data are available at ompp.mcgill.ca)

The media content analysis tracks the volume and tone of cover- age for parties and leaders, as well as trends in issue salience, and the study was designed in a way that most codes are purely objective " that is, not dependent on subjective decisions by coders. The study recorded, for instance, the first three parties, leaders and issues mentioned in an article, in the order in which they appear, as an indication of prominence. ”œFirst mentions” are a particularly valuable indicator of the issue agenda and the prominence of parties and lead- ers in the campaign. We also recorded the general focus of articles: primarily campaign-focused (”œhorse race”), or primarily issue- focused. This provides an indication of how much coverage is focused on polling results and stories about the campaign trail, versus policy issues.

We also included one set of subjective codes for tone " positive, negative or neutral " for parties and leaders. To ensure as much reliability as possible, all coders went through a practice ses- sion, and there were weekly meetings and communication throughout the campaign. Stories were also randomly selected for double-coding to check ”œinter-coder reliability” " the consis- tency with which different coders come up with identical codes. For the objective material, this is of course relatively easy; for the subjective ”œtone” codes, we had to be very care- ful. Our instructions to coders were as follows: the default ”œtone” for all mentions is neutral; a leader or party mention has to be very clearly positive or negative to be coded as such. We may have missed some of the more subtle tone in articles, but then so do regular readers of the newspapers. Predictably, most of the clearly posi- tive or negative tone for parties and leaders occurs in the editorial and opinion pages.

That is the study, then; what were the results? First was the (rather unexpected) focus on issues. Figures 1 and 2 show the weekly proportion of policy versus horse-race coverage in 2004 and 2006, and illustrate the principal difference between the two campaigns. For practically the entire 2006 campaign, the volume of articles focused on policy dis- cussion was greater than, or at least equal to, the number of articles devoted to the election horse race. During the 2004 election, policy issues had played a much more limit- ed role in media coverage, a role increasingly constrained as the cam- paign progressed. In 2006, coverage of parties’ policy proposals was not eclipsed by polls and campaign strat- egy. The majority of this election campaign’s media coverage was poli- cy-focused.

Figure 3 shows trends in the prominence of individual policy issues over the campaign. National unity was a relatively prominent issue in the campaign, not just because the Liberal Party attempted to portray itself as the only ”œnation- al” party, but also because of the Conservative Party’s ”œopen federal- ism,” which proposed a role for Quebec at international conferences and a solution to the so-called fiscal imbalance between provinces and the federal government. Indeed, national unity became the most salient issue in week 4, at the very moment that the Conservatives revealed their policy. National unity regained momentum during the final three weeks of the campaign as alle- gations surfaced about spending irregularities by Option Canada dur- ing the 1995 Quebec referendum.

There were a number of other important issues over the campaign. Taxes and crime led media coverage during week 2; this followed the Conservative proposal to cut the GST and the Liberal plan to ban hand- guns. International affairs also received considerable coverage after Martin challenged the US on environ- mental policy, as did the US ambas- sador’s response. And health care and social issues also received some cover- age, but they did not dominate the policy discussion as they had in the 2004 campaign.

By far the most salient issue of this campaign was, as in 2004, government accountability and cor- ruption. Figure 3 shows that corrup- tion was the most prominent issue in the news media at the start of the campaign, and then re-emerged as the most salient issue of the cam- paign’s second half, aided by news that the RCMP was launching an investigation into a Liberal govern- ment decision on income trusts. Indeed, the overall salience of accountability and corruption in the past two election campaigns is difficult to overstate. It was the most discussed item in the press for four of five weeks in the 2004 cam- paign, and it was the leading issue for five of the eight campaign weeks during the 2006 election. In effect, government process rather than spe- cific government policies has domi- nated recent election news in Canada.

The prominence of accountability and corruption issues in media coverage is therefore not surprising. What is striking is that the shift in voter preferences captured in polls did not coincide with heightened corruption coverage, but occurred during and immediately following weeks in which the focus was on other policy domains. That is, it appears as though the early shift in Conservative support occurred fol- lowing three weeks of substantive policy discussion in which govern- ment accountability appeared in only about 10 percent of the stories on the election. The implication is that, more so than negative stories about Liberal corruption, substantive policy coverage was critical to the Conservatives’ electoral success. What about coverage of parties and leaders over the campaign?

Where tone is concerned, it is worth noting that the tone of news cover- age in this election was predomi- nantly neutral, and " as in 2004 " any clearly positive or negative tone was mainly reserved for columns and editorials. And, when an opinion was cast about a party or a leader, it was much more likely to be negative than positive.

The main difference in 2006 was the evolution of party and leader tone over the course of the cam- paign. Figures 4 and 5 illustrate trends in ”œnet tone,” measured as the percent positive mentions minus percent negative mentions, and thus indicating the relative weight of pos- itive over negative coverage. The fig- ures show the gains made by Stephen Harper and the Conservative Party during the sec- ond, third and fourth weeks of the campaign; Harper’s net tone jumped from -16 to +10 and Conservative coverage improved from -6 to +7. These shifts show the extent to which the Conservative Party bene- fited most from these early weeks of sustained policy discussion. Proposals relating to the GST, child care and national unity not only seem to shift the issue agenda away from accountability, but also coin- cide with significant improvements in the tone of coverage for the Conservatives and their leader.

While Harper and his party gained momentum during the early stages of the campaign, Paul Martin lost the most ground, at least where tone of news coverage is the measure. Martin’s net coverage decreased seven percentage points from -5 in the first week to -12 by week 4. Meanwhile, the net tone for Bloc Québécois and NDP leaders remained relatively stable; Jack Layton dropped four percentage points while Gilles Duceppe gained four percentage points.

In the final weeks, net tone for all parties and leaders converged somewhat. By week 8, coverage of the Conservatives, Bloc Québécois and NDP was almost identical in terms of tone. The tone of coverage about Stephen Harper and Jack Layton was also about the same at the end of the campaign, though Gilles Duceppe lost more ground during the final week of the campaign, sliding from -2 to -6. Even the Liberals and Paul Martin received more positive cover- age during the second half of the campaign. In fact, the final week of the 2006 campaign was the most pos- itive of all for Paul Martin. At the end, however, he and the Liberal Party were still well below the aver- age tone received by all other parties and leaders.

Coverage of parties and leaders also has to do with their promi- nence in media reports. To capture this, we relied on ”œfirst mentions” " the party or leader mentioned first in each article. First mentions indicate who is framing the election cam- paign " whether the article is about Martin, for instance, and others reacting to Martin, or, alternatively, about Harper, and others reacting to Harper. Figure 6 shows first mentions over the campaign.

As in 2004, Paul Martin led first mentions for most of the 2006 cam- paign. It’s not clear this benefited his cause since the coverage was almost entirely negative. Indeed, we found that the net tone for Liberals was improving, just as their prominence in stories was decreasing. Significantly, the opposite was true for Conservatives " their net tone was improving alongside their increasing prominence in coverage. The last two weeks of the campaign were all about Stephen Harper. During the final days, more than half of the articles involving party leaders mentioned Harper first.

In addition to campaign trends, our analysis of media coverage also sheds light on differences between newspapers. How did that coverage differ? Election campaigns tend to re- invigorate concerns about ”œmedia bias,” among party leaders, political strategists and journalists alike. In the last campaign, we found evidence of partisan biases in newspaper edito- rials and opinion pieces, but relative- ly neutral news coverage. The analysis of this campaign reinforces that conclusion.

Figure 7 shows mean-adjusted ”œnet tone” (that is, percent positive mentions minus percent negative mentions of a party) ratings for the two leading parties in the 2006 campaign, in each of our seven newspapers. To mean-adjust net tone ratings, we take the average net tone for a party in all newspapers " listed in table 1 " and look at the difference between that and the net tone for that party in each indi- vidual newspaper. For example: table 1 shows that the average net tone for Liberals in editorial and opinion pieces was nearly -28; net tone in the Calgary Herald was -43; so the mean- adjusted net tone " the difference between the Calgary Herald and the average " was -15. That is, net tone for the Liberals in Calgary Herald edi- torial and opinion was 15 points below the average.

Mean-adjusted measures provide a good sense, then, for whether coverage in one newspaper tended to be more negative or positive than coverage in other newspa- pers. For each, we distinguish between (1) editorial and opinion, and (2) news. The distinction is criti- cal. The function of opinion and edi- torial pieces is to convey opinion, and we should reasonably expect there to be ”œbias” reflected in these pieces. News stories are supposed to be relatively neutral, however. This is the traditional ”œwall of separation” between editorial and news content. Separating our results into these two categories allows us to test just how thick this wall is.

We note, first of all, the tenden- cy for media to present mainly neg- ative information. That is, on virtually any issue we should expect news content to tend toward the negative " this is the function of a critical ”œfourth estate,” after all. This negative tendency is reflected in fig- ures 4 and 5, which lump together news and editorial content to show weekly trends. In both figures, rat- ings are much more often below the zero line than they are above it. The tendency is also shown in table 1, which shows the average net tone for parties over the entire campaign, separating news and editorial con- tent. Four values hover very close to neutral; the other four are clearly negative.

It is also the case that net tone varies much more in editorial than in news content, exactly as we should expect. This is true in table 1; it is also the case when we look at news- papers separately in figure 7. In every case where net tone is significantly different from zero, tone in editorial content is much greater than tone in news coverage. Indeed, it is rare for news coverage to show any signifi- cant tone at all. The traditional wall of separation is certainly evident here, although it also true that where there is tone in news content, it tends to be in the same direction as the tone in editorial content. Put dif- ferently, where editorial and opinion pieces tend to support the Conservatives, news content on the Conservatives is more likely to lean " albeit slightly " in a positive direction than in a negative direc- tion. It is notable that two of the few cases in which this was not the case involved coverage of the Conservative Party in the Globe and La Presse. Both papers showed nega- tive editorial and opinion coverage, but mildly positive news coverage. And in the end, both papers endorsed the Conservatives.

Our analysis has tried to paint, in rather broad strokes, a pic- ture of the 2006 campaign as it appeared in newspaper content. On several fronts, the 2006 picture is similar to that in 2004. Corruption remained a central issue in the 2006 campaign, and clearly did damage to the Liberal brand. The overall net tone for parties and leaders tended toward the negative, particularly for Martin and the Liberals, who started well below the other parties and leaders and remained there for the duration of the campaign. And edi- torial and opinion content showed clear partisan preferences, generally in line with widely accepted beliefs about the newspapers in our study, while news content remained rela- tively neutral.

There are important differences on other fronts, however, and these are worth highlighting. Indeed, these differences may explain how and why Stephen Harper became prime minister. First is the rather extraordinary focus on policy issues in this campaign " extraordinary in the sense that many expected a vac- uous and vicious election campaign, and this was not the case. There was much more policy discussion in 2006 than in 2004, and this discus- sion dealt with a more diverse range of issues. This issue focus was in large part due to the Conservative campaign, and apparently with good reason: the Conservatives appear to have benefited greatly from policy discussion. The shift in Conservative ratings occurred pre- cisely when policy discussion was at its most active.

One lesson of the 2006 campaign may be that policy discussion works " that both journalists and the Canadian electorate may appreci- ate the opportunity to explore sub- stantive issues. Negative coverage of Liberals, and ongoing attention to corruption and scandal, certainly didn’t hurt the Conservatives. But the shift in Conservative fortunes seems here to have been driven in large part by policy discussion. It is too early to tell which policies may have been specifically endorsed by Canadians, and it seems probable that there is no single answer " Quebecers are likely not supporting the same Conservative policy propos- als as Albertans, or Ontarians. Just the existence of policy proposals seems to have helped, and this is of fundamental importance. In the 2006 election campaign, unlike 2004 and in contrast to most predictions, policy mattered.