Canadian Journal of Communication Vol 35 (2010) 413-429
©2010 Canadian Journal of Communication Corporation

Experts in the News: The Differential Use of Sources in Election Television News

Kathleen Ann Cross
Simon Fraser University

Kathleen Ann Cross is a lecturer in the School of Communication at Simon Fraser University, K9671-8888 University Drive, Burnaby, BC. V5A 1S6. Email: .

ABSTRACT  This article examines the use of news sources in mainstream television coverage of a Canadian election and considers these findings within the context of broader studies on the significance and role of sources in election news. Using content analysis, it considers the type, nature, and functions of sources in election reporting and pays particular attention to the competing theories of the use of sources in news discourse. Through a review of the type, length, and frequency of sources, a more media-centric theory of source use emerges. It is suggested that rather than “primary definers” or “political entrepreneurs,” sources often function as mere “symbolic representations” of an already framed story in election discourse.

KEYWORDS  Content analysis; Political communication; Elections

RÉSUMÉ  Cet article examine l’utilisation de sources à la télévision grand public lors d’une élection canadienne, puis considère les résultats de cet examen dans le contexte plus large d’une étude de la signification et du rôle des sources dans les reportages électoraux. L’article recourt à l’analyse de contenu pour recenser le type, la nature et les fonctions des sources dans les nouvelles et porte une attention particulière à certaines théories contrastantes sur le recours aux sources dans les discours de l’actualité. En passant en revue le type, la durée et la fréquence des sources, l’article formule une théorie centrée sur les médias. Il suggère que, plutôt que « définisseurs primaires » ou « entrepreneurs politiques », les sources sont souvent de simples « représentations symboliques » d’une histoire déjà encadrée dans les discours électoraux.

MOTS CLÉS  Analyse de contenu; Communication politique; Élections


In media studies, as well as in newsrooms, those who provide information and context to the news narrative, offer official or unofficial opinions, and give eye-witness accounts of events are referred to as news sources. Research on the social production of news, journalistic practices, and influences on news content has contributed to an understanding of the use of news sources, the journalist-source relationship, and the potential effect selective source use has on the framing of news stories. The manifest nature and discursive use of all types of news sources is an integral part of understanding news reports about parliamentary elections. This article examines the use of news sources in the mainstream television coverage of a Canadian election and considers these findings within the context of broader studies on the significance and role of sources in election news. In particular, it considers the type, nature, and function of the use of sources in this election coverage and pays particular attention to the competing theories of source use in news discourse.

Do sources matter?

Decades of research into the role of news sources in the social construction of news have shown how the choice and framing of news sources has significant impact on news content. Many of these research studies have further identified that while news sources form a substantive amount of news content, not all news sources enjoy the same degree of access to the media (Gans, 1979; Hackett, 1985; Hall, Clark, Critcher, Jefferson, & Roberts, 1978). Stories in the news tend to be weighted toward elites, giving disproportionate attention to those who are known public figures or who have political and/or economic power. Similarly, news sources are also disproportionately powerful figures. This discernible pattern in the use of news sources has an effect on both the content and the interpretation of news discourse. Studies into the day-to-day practices of news work, for example, have suggested that the pattern of source use in news is an indication that news is a representation of conversations between the powerful in society, while the general public are relegated to a position of mere observers (Fishman, 1980; Tuchman, 1978).

One of the foundational studies in news analysis demonstrated news media to have a preferred orientation toward “primary definers”—those official sources associated with powerful positions or institutions, such as law enforcement, government, and representative interest groups. Hall et al. (1978) argued that the relationship between sources and journalists resulted in those with the most power in society setting the terms of the debate, or the “primary definition” of an issue or event, to which all other interpretations must conform. This view notes that the news media are grounded in “objective” statements, quoted from “authoritative” institutions represented by “accredited” sources. Hackett (1985) found similar patterns in television news and noted that not only do sources from powerful institutions define the terms of debate, they are also afforded “higher status modes,” such as being interviewed in authoritative office settings rather than on the street. The recurring theme in much of this analysis supports the view that the structured hierarchy of access by elite sources to news discourse results in framing of news content that favours the powerful. In this way, the news media’s use of sources contributes to the media’s ideological role in the power structures in our society. As Curran (2000) noted:

The media’s role is never solely confined to imparting information, but necessarily involves arbitrating between the discursive frameworks of rival groups. Which frameworks are included or excluded matters, because over time it can affect collective opinion and indirectly, the allocation of social resources in society. (p. 138)

Although the “primary definer” view of the role of sources has been a dominant one in critical perspectives, it has not gone without challenges and corrections. Gans (1979) suggested that rather than a linear process of journalists using information from sources and transmitting it to audiences, the process is more circular. He outlined a “large number of feedback loops” that inform the journalist-source relationship, including journalists’ choice of the sources they consider “suitable” for their news organization’s audiences, resulting in what Gans suggests is “closer to being a tug of war than a functionally interrelated” system (p. 80). Others have argued that news is primarily a product of the socially constructed interaction between reporters and their sources, and as a result a “politics of knowledge” emerges from this relationship that socially constructs certain information as knowledge (Ericson, Baranek, & Chan, 1989, p. 377).

Perhaps the most influential challenge to Hall et al.’s 1978 model was raised by Phillip Schlesinger (1990). He argued that the “primary definer” view of news sources is far too simplistic a model since it does not recognize, among other things, the conflicts between competing elites, the role of “off the record” sources, the news media’s challenges to the primary definers, and news workers’ role in creating their own stories (albeit a small number of stories). Two of Schlesinger’s criticisms are especially relevant to this discussion. He argued that the “primary definer” theory of source roles in news does not recognize the possibility that different “primary definers” may enjoy a differential access to news media discourse. In addition, he criticized the model for its atemporal or ahistorical perspective that assumes a permanence of source roles without accounting for changes in source credibility and function over time. He noted how the major empirical studies on the sociology of journalism—those researchers who have looked at the practices and routines of news work, such as Tuchman (1978) and Fishman (1980)—“shied away” from the primary definer model and recognized that some groups have better access than others (p. 76). Schlesinger (1990) suggested that news sources were not structurally determined primary definers but rather political entrepreneurs who seek to achieve the position of defining the news frame for strategic purposes, but do not always succeed.

In the case of election news, there is significant evidence for both the “primary definer” and the “political entrepreneur” roles of news sources. Media strategists from all political parties use sophisticated media tactics to influence news content and “get their message out.” Indeed, the importance of the mass media as the primary source of political information for voters requires nothing less. In addition, groups with social, political, and economic interests attempt to capture the media’s and the public’s attention to their cause and to influence the election debate. Some of these interest groups are successful in capturing the media agenda and some are not.

Yet neither primary definer nor political entrepreneur theories can fully explain the diverse and disparate use of sources in the news. For example, how do we explain the use of so-called independent experts for commentary or the journalistic tendency to seek out people on the street for what is defined as “public opinion?” Certainly many of the experts are themselves members of powerful elites, and some are seeking political influence, but some powerful sources fail to influence the election discourse altogether. When are different types of sources used and under what circumstances? What are the patterns of source use, and how do they fit within these two dominant theories of sources? Studies of election news discourse have often failed to recognize the media’s own role in choosing, differentiating, and framing source access and source use during election narratives. This article will explore this gap.

Source use in Canadian election studies

Much of the recent literature on election news in Canada agrees with the view that, at least in the case of elections, the news media are secondary definers in setting the election agenda compared to the political parties (see, for example, Attallah, 2004; Attallah & Burton, 2001; Dornan, 1997; Johnston, Blais, Brady, & Crete, 1992). Rather than the media using political sources, it is argued that the political sources use the news media. Political journalists, it appears, are both manipulated by political actors and simultaneously struggle against such manipulation in a negotiation of control that has been described as a “dance of the dialectic” between journalists and politicians (Gilsdorf & Bernier, 1991, p. 15). According to these views, both the political actors and the political journalists try to control the issue agenda and messages of the campaign news, with the political actors having the upper hand.

Yet little research has explored the use of sources who were not political elites during election campaigns. Further, research regarding the use and frequency of television news sources during Canadian election campaigns is sparse. Most of the election news research has focused on party leaders at the expense of other types of sources, the amount of news space occupied by the party leaders, or the amount of coverage where the party leaders were “principal actors,” rather than being quoted as sources (Crete, 1991; Dornan & Pyman, 2001; Frizzell & Westell, 1989; Johnston et al., 1992; Soderlund, Romanow, Briggs, & Wagenberg, 1984). All of these studies found that party leaders were the primary focus of news reports about the elections and accounted for the majority of news space, hovering around 60% of the television or print coverage. One study noted that in total, political actors (leaders, other candidates, and party spokespersons) accounted for 83% of the news focus (Crete, 1991), while another commented that regardless of the actual numbers, the most prominent news stories, and the lead items on television newscasts, are generated by the leader’s tour at the expense of all other potential sources. Indeed, “other actors are quoted in reaction to the statements of the party leaders … [and] local candidates and interest group leaders tend to be virtually absent from high-profile coverage” (Fletcher & Everett, 1991, pp. 199-200, emphasis added). A survey of studies about the news media during elections in Québec found that “most journalists’ sources were agents of the economic, political or social establishments” in the 1980 and 1984 federal elections (Charron, 1991, p. 130). Further, the researcher concluded that

collective, non-institutional entities (nation, people, society, citizens) were less represented in stories than institutional collectives, such as countries, regions, provinces and institutions … [thus] … the national question, in this kind of discourse, is more the business of the official political institutions than the people of Quebec. (p. 130)

Similarly, Gans found that the “unknowns” (protestors, victims, voters, participants in events) appeared in American television news only about one-fifth of the available news time, while “knowns” (the elites, such as politicians and government or institutional officials) made up the remaining four-fifths (Gans, 1979, p. 13). This “hierarchy of access” was supported five years later by a published study on Canadian television news indicating that those who are interviewed most often are male politicians, business leaders, and group spokespeople (Hackett, 1985). However, while a 1991 study of local television coverage (vs. national coverage) upheld the findings that group spokespeople and politicians received the bulk of television coverage (about 45% altogether), ordinary people were still quoted almost 25% of the time (Papas, 1991). Papas further suggested that the slightly higher number of public voices indicated that, contrary to previous studies, the news is indeed “accessible” to the average person and that such access bodes well for democratic society. Since the literature on the use of vox populi and individuals in television news stories was contradictory, this offered an interesting area to explore in the data.

The literature on media coverage of Canadian elections suggests that at least two trends could be explored in this study: 1) the particular source use of non-political actors such as interest group representatives, “experts,” and the public; and 2) the implications of source use when compared to the topics on which sources were quoted.

Study methodology

The data for this study was drawn from the television news coverage of a Canadian provincial election in 2001. At the time of the election (April 17 to May 16, 2001), four major television news stations were reporting in the province of British Columbia: BCTV, CBC, Global, and VTV. The six o’clock dinner news hour program of all four networks was videotaped from the day before the election call until the final campaign day of the election—for a total of 29 days of newscasts.1 In total, 102 broadcasts were videotaped in the 29-day period, resulting in 341 election stories or “units” in the 102 broadcasts. The full study tracked who was quoted, the length of quotes, and which topics sources spoke about. Story topics were determined by trained coders with the help of a coding protocol, pilot study, and training sessions. Inter-coder reliability for topics was 92%.

For this analysis, a person was considered a source if they were directly shown and heard making a statement—or directly seen and heard on air, commonly referred to as an “actuality” clip. Coders recorded all sources who were quoted, including their name (if available) and their affiliation. All sources were coded in the order in which they first appeared in that news item and were counted as one source, regardless of how many times they were quoted in that news item. Thus, the “source counts” represent each separate person quoted in each news item, not the number of quotes themselves. Further, each source was labelled as a source type—a descriptor that reflected how the newscast identified and used the source (such as student, Liberal party candidate, or hospital representative). After all of the sources were recorded, a post-coding review of the sources further grouped them into broader source categories.

In addition, two other calculations were used to further evaluate the use of sources: the length of quote and the type of story about which they were quoted. All quotes were timed (to the nearest second) to calculate the total amount of source time and the average quote length for each source and each source category. Since the focus of this study was on news stories only, units identified by the coders as feature interviews or panel discussions were removed from the calculations. Finally, the source use was cross-tabulated with the story type. To determine the topics in the television news, all election stories in the data set were coded to identify the primary topic of the story and any other major topic mentioned. The story topic categories were later collapsed into two broad categories of “issue” stories and “campaign” stories. A story was coded as an “issue” story if the primary or most dominant topic was a policy issue of interest in the campaign, such as forestry or healthcare. An issue was coded as a “campaign” story if the primary focus was the various party campaign activities or strategies, the actions of the leaders, poll results, or scandals or gaffes—rather than a policy issue.

Table 1Study findings

The research found four general categories of sources—political actors, individuals, interest group representatives, and experts, (plus a category designated “other” for a minor number of sources who did not fit into the other types, such as debate moderators). These general findings are illustrated in Table 1 and the categories discussed in more detail below.

Political actors

Not surprisingly, of the five categories of sources, “political actors” were most commonly used as sources in news stories, at 58.6% or over half of all sources used. The category of political actor included all candidates and leaders, ex-candidates and ex-leaders, sitting political representatives at any level of government (civic, provincial, or federal), and political workers. The category also included supporters of a particular party, but only if they were in a political rally or similar situation, indicating they had been identified by the party itself as a supporter. It did not include “people on the street” quotes that indicated support for one party or another, since these were random sources and thus were coded in the “individual” category as vox pop or “person on the street” quotes. It also did not include former party affiliates (political strategists or ex-politicians) who were used by the media as political experts in panel discussions and interviews, since these sources were not active in the current election and were primarily identified by the media as experts on the party, rather than active party affiliates themselves. Generally, this finding confirms that the media focus during election campaigns remains on the leaders’ tours, and the leader is the primary representative of the party.

More surprising in this case study is the finding that party leaders accounted for only about half of the quotes from political actors and only about 30% of quotes from the total number of sources. As noted earlier, the previous studies had found that as a representation of news space and focus the leaders accounted for about 60% of all coverage (see, for example, Crete, 1991; Frizzell & Westall, 1989; Johnston et al., 1992; Soderlund et al., 1984). The dominance of news focus on the leaders certainly remains true of this data. Yet while party leaders were the majority focus of the news accounts, they were not the majority of quoted sources, representing only about one-third of the total source quotes in the election stories. Instead, it is the combined quotes of all “political actors,” rather than just the leaders, that accounts for approximately 60% of the source quotes in the coverage. Almost one-third of the quotes from political actors came from other candidates, although only a small number of candidates were repeatedly used by the news media for comments. Thus, while previous research on news focus might suggest leaders are the primary focus, a more precise methodology that takes into account the actual quotes from leaders, as used here, indicates that this tendency for news media/outlets to focus on leaders may have been overstated.


Those coded as individuals were used approximately 26% of the time as news sources. This category included, for example, employees, business owners and managers who were speaking for themselves and their business only, landlords, tenants, unaffiliated protesters, patients and their families, and members of an audience and people on the street (vox pop). Regardless of whether or not these sources were identified by name, they were not identified as having any affiliation to an organization or institution and were not used to provide specific expertise. Almost two-thirds (64.6%) of the individual sources were streeters or vox pop, where individuals were flagged down in a public place (on the street, in a shopping mall) for an off-the-cuff, candid reaction to a question asked by the reporter. In addition, the data indicated that there were twice as many male vox pop as there were female (66% male to 34% female), suggesting that the “man on the street” is, on average, still literally true.

It is an intriguing finding that in this data the use of individuals in television reports represented almost as much frequency as the political leaders. Although early studies on sources in the media noted the lack of access to the media by the ordinary or average person (Gans, 1979; Tuchman, 1978), this research suggests that individuals who are unaffiliated with institutional power have significant profile in the news. How significant this access is can be at least partially determined by examining what topics these individuals were quoted on and for how long. Importantly, this category of individuals is often easily accessed by the media and is commonly used to convey a kind of voter populism. Coverage of this category has the added bonus of being inexpensive to produce. In comparison, political actors are often eager to talk to the media and “get their message out,” but reporting on political actors often involves days on a leader’s tour, travelling by bus or airplane at the media organization’s expense. The use of individuals as sources, however, often requires only that the reporter walk out onto a street or into a restaurant to mingle with the public. This may partially explain why the number of vox pop sources far outstrips all other types of individual sources put together.

Representatives and experts

The next two categories of sources, the “representatives” and the “experts,” were particularly interesting. Sources were coded as representatives (7.2%) if they were identified by the news story as affiliated with an organization or institution and were seen to be speaking as representatives of that organization. Examples of representatives included First Nations leaders, a nurse speaking on behalf of the nurses union, an affiliated member of a lobby group, or members of community organizations (such as a resident’s group) or non-government organizations (NGOs). The representative category also included “institutional representatives,” such as provincial or city government officials, school officials, and public health authorities. The distinction was made between institutional and non-institutional representatives to indicate interests that were attached to public or government organizations, as opposed to those organizations that were non-governmental.

In contrast, sources were coded as experts (6.8%) only if they were identified by the media as having a particular expertise or used as knowledgeable experts about a particular topic due to their institutional affiliation or their experience. For example, university professors, researchers, and political commentators were all used by the media as “experts.” In some cases these experts were presented as “independent” (as in the case of university marketing professors or public opinion pollsters), while in other cases their expertise was as partisan political commentators whose past position in election campaigns gave them insider expert knowledge of campaign tactics and party dynamics. These “party pundits” were used as experts about the political parties and their strategies, as opposed to spokespersons for the party, and sometimes were critical of the official party line or strategy.

In addition, the news media used research institutes (“think tanks”) as experts, even though they often have a socio-economic and political orientation. Two examples of research institutes included in the category of experts in this case study were the Fraser Institute, which was coded as a “right-wing think tank,” and the Canadian Centre for Policy Alternatives (CCPA), which was coded as a “left-wing think tank.” Thus, the difference between sources who were coded as experts and those coded as representatives lies in the type of affiliation as well as the use and framing of the source by the news media. Table 2 indicates the source use breakdown in the types of sources for both representatives and experts.

Table 2

The largest type of representative used as a source were union spokespersons, likely due to the two major labour actions conducted during the election, one an overtime ban by nurses due to stalled contract negotiation and the other a strike by bus drivers. Studies on media coverage of labour disputes have noted that there is an “over-representation” of union sources compared to management sources during such disruptions linked to the framing of labour disputes as “caused” primarily by the unions, leaving the employers to “manage” the disruption (Beharrell & Philo, 1977; Davis & Walton, 1983; Hackett, 1983; Hartley, 1982; Silva, 1995; Walton & Davis, 1977). Given this, it is not surprising that union representatives received more than one-quarter of the quotes from representatives, even during election stories.

Public officials were the second-most-common representative quoted, comprising just over 23% of the total sources used. The next most frequent representatives quoted were speaking on behalf of non-governmental interest groups, including First Nations representatives (10.3%), a group concerned with “leaky condos” (10.3%), doctors’ organizations and business organizations (at 9% each), environmental representatives (5.1%), and community or neighbourhood representatives (3.8%). Together, all non-governmental interest groups, outside of union representatives, totaled 47.5% of the representatives used as sources. If union spokespeople are included, the total NGO representative count is 74.4%. However, the total use of NGOs accounts for only 5.5% of the total number of all sources quoted (at 60 out of 1,087 sources). Thus, it cannot be said that the news favoured interest groups as sources when reporting on the election.

As for the experts, almost 42% of all experts used as sources were political pundits affiliated with a party. Every television station had its own political panel or group of analysts comprised of two or three commentators associated with political parties. Less common as an expert source was the “neutral pundit,” such as a political historian, who commented on the campaign. University professors were used about 18% of the time as experts, and pollsters were given 8% of the time allotted to experts, representing less than 1% of the total sources quoted (although the results of their polls were the basis of news reports far more often). Indeed, experts outside of partisan politics were used as sources only 43 times, representing more than half of the experts quoted but only 4% of all the sources quoted in the election news. Therefore, it can be safely concluded that neither experts nor representatives of organizations or public institutions were a high priority for the news media to talk to during this election. What is interesting, then, is when and under what conditions did the media use these two categories as sources? This will be explored further.

Sound bites and sound bite news

Gathering evidence as to who was quoted provides an impression of the dominant sources used in this television coverage of an election. However, it does not reveal whether these quotes represented a small or large part of each story and whether the use of the source made a substantive contribution to the story. Since television coverage is highly constrained by time, two indications of the importance of source use is the amount of airtime devoted to source quotes, plus the length of the quotes or “sound bites.” Studies have consistently shown that the length of the average sound bite on television news has been steadily shrinking for the past three decades. Drawing on 20 years of presidential elections in the United States from 1968 to 1988, for example, Daniel Hallin (1992) found that the average sound bite had decreased from 43 seconds to less than 9 seconds. The Canadian research on the shrinking length of election news sound bites echoes the American findings. Gilsdorf & Bernier (1991) noted that the “30 second clip” of the 1979 Canadian federal election had been reduced to the “12 second sound bite” in 1988 television coverage. Overall, studies have indicated that the length of all sound bites in Canada fell from approximately 60 seconds in the 1960s to under 10 seconds in 1993 (Hackett & Hissey, 1996). Given the previous research, it was anticipated that the length of sound bites in this case study would be consistent with previous findings and be less than 10 seconds in length. Indeed, as the following discussion outlines, this hypothesis was supported.

Figure 1The total time all sources quoted in these news stories (excluding feature interviews and panel discussions) represented over one-third (32.8%) of the total time of election stories in the corpus, with an average quote length of slightly less than 9.08 seconds (see Figure 1). While the expectation that the average sound bite would remain at less than 10 seconds is supported, it is interesting to note that the sound bites did not fall further below the 10-second mark, as the trajectory of previous studies of the “shrinking sound bite” might have indicated. This finding suggests that there might be a limit to the diminishing length of a quote as it is used in television news, possibly indicating the average length a quote needs to be in order for it to be comprehensible to a television audience. Nevertheless, it would be safe to suggest that the emergence of the shorter sound bite has affected both the content and the tone of election news. Hallin (1992) has argued that television sound bites have shifted the focus of news reports on election campaigns from what the candidate is saying to what the journalist is saying about the candidate. Indeed, Hallin further argues that the shortened sound bite could be one of the key reasons for increased focus on political analysis and campaign stories and less media attention on the policy issues of an election. This strong connection between “horse-race journalism” and “sound bite journalism” is supported by his findings that “stories with high percentages of time devoted to horse-race themes tended to have short sound bites, while more issue-oriented stories tended to have longer ones” (Hallin, 1992, p. 21).

In addition, there is evidence that “sound bite news” has affected the journalist-source relationship in election campaigns. The messages that a politician delivers to the public are covered in very short and simplified sound bites, sometimes consisting of only a single quoted sentence. This “message” is then used in television news to symbolically represent the whole campaign strategy or platform of the candidate. The reliance on sound bites reduces the ability of candidates to communicate any real policy content to voters and instead limits the impact or focus to one of image, rather than issues. Hence reporter commentary and “wrap-ups” of what was said have replaced the political campaign messages, and election news coverage has shifted to a more mediated, journalist-centred content. Some communication scholars have suggested that this tendency has driven political candidates to seek other ways of communicating that allow them to speak directly to the voters (Hallin, 1992). In particular, politicians and their advisors have had to adjust their campaign strategies to adopt shorter, snappier quotes that will be easier for the television to air, and they have begun to rely more on website and campaign material to publicize their policy platforms.

As stated above, in this study the average length of quotes was less than 10 seconds. However, the differences in length of quotes between the categories of sources may suggest the relative importance attributed to source types by the news organization. For example, Figure 1 shows that the average length of a sound bite for the two most frequently quoted categories of sources was somewhat different. As noted previously, political actors were quoted almost 60% of the time (58.6%), and had an average sound bite of 8.9 seconds. Individual voters were quoted about one-quarter of the time (26%) but had an average sound bite of only 6.34 seconds, about two-thirds that of political actors. The average length of sound bites for an expert was the longest, at 13.4 seconds, and representatives were quoted on average for about 11.07 seconds. However, while their sound bites were longer, both experts and representatives were only used as sources in news stories at a frequency of about 7%.

Although the overall sound bite has shrunk, it is important to note that there are differences between the length of a quote from a member of an “elite” and the length of a quote from an ordinary person. The average sound bite for elites (candidates and other institutionally attached sources, including experts and representatives) in this study was 11.12 seconds, whereas the average sound bite for an individual voter was 6.34 seconds. Therefore, although the news media may be increasingly mediating the messages of politicians and candidates as well as increasing the use of ordinary citizens in their coverage, this does not result in greater voter input into the news stories, as some journalists and researchers have suggested. This study demonstrates that while the amount of “ordinary people” in the television news of the election represents about one-quarter of all sources, most quotes from voters are relegated to mere representations of a viewpoint or an issue, primarily to reinforce a storyline, rather than being part of the actual debate on issues. Indeed, on average, ordinary people are not quoted long enough to provide any significant or useful comment on the election or the election issues. They are, in fact, used as populace “fillers” of an already constructed news story. In this way, vox pop, or the “voice of the people,” appears to be used merely as “evidence of the public gaze” (Langer, 1998, p. 63) rather than as evidence of active citizen participation in the election discourse.

Sources and topics: What were they quoted about?

If, as Gans noted more than 25 years ago, “the most salient characteristic of sources is that they provide information as members of representatives of organized and unorganized interest groups” (1979, p. 80, emphasis added), about which topics did the different categories of news sources provide information? As noted above, numerous studies on the role of news sources have suggested that news topics and sources have a mutually constitutive nature. Hall et al. (1978) noted that it is sources who act as “primary definers” of the meanings of events and issues. Sources may also be understood as functioning as a “filter” in the process of news selection and production (see Herman & Chomsky, 1988). In this way official and elite sources help to define what is objective, credible, and authoritative information in the news. Thus, it is instructive to explore the topics on which sources were quoted.

This study showed that fully 75% of the election story topics were about the election campaign, and only about 25% of the stories were devoted to election issues or policy topics. Consequently, it is not surprising to find that most of the sources were quoted in stories primarily about the campaign activities and strategies and not about policy issues. For example, in stories where the primary topic was coded as an issue, political actors were used as sources almost 18% of the time (114 occurrences), compared to 82% of the time (523 occurrences) when they were used as sources in campaign stories (see Table 3). Political actors and their activities were themselves the focus of most of the election stories, and most political actors were used as symbolic indications of these activities (e.g., shown addressing a crowd of supporters) rather than to make substantive quotes about the election issues.


As well, expert sources were quoted primarily for campaign-related stories. For example, experts were used as sources only 17.6% of the time in stories coded as primarily about an issue, compared to more than 82.4% of the time in stories about the campaign. This suggests that the vast majority of experts used as sources by the media were experts on election campaigning. Reviewing examples of the use of experts supports this suggestion. The experts used by the media were primarily political pundits. Other experts included pollsters, marketing professors, political science professors, and advertising professionals, who were asked to comment on various aspects of the campaign. Political analysts or “pundits” were repeatedly interviewed to explain the latest poll findings and make predictions about the election results. Advertising professionals were asked their opinion about the parties’ political advertising. A political science professor was asked to compare the party campaign strategies of the leading parties and to compare strategies of different elections. Hence, the dominant use of “expertise” was not to discuss policy or governing issues, such as the state of healthcare or the party platforms on taxation, but to discuss the campaign itself.

Individuals used as sources were also used primarily in campaign stories. However, those coded as a “representative” of an organization were the only type of source used more often in issue stories than in campaign stories. These findings suggest particularly interesting implications about both the role of experts and the role of representatives in election news. Gans (1979) suggested that the “experts” in the news were not there as members of organized and unorganized groups, but as the most informed in a field of study or activity and therefore able to provide insight to the journalist and the audience. He outlined how their function in the news report, then, was to contextualize events and tell the audience what it “really means” or predict what will happen, while at the same time claiming neutrality and dispassion. Yet in this case study the overwhelming use of experts was to comment on political campaigning as opposed to issues. Although some commentary about the campaign from political “experts” can be expected (after all, the strategic process of electioneering is fascinating to many people, including journalists), what is surprising is that very few experts were used to discuss policy issues or substantive topics.

For example, the public’s highest-ranked issue topic for this election was healthcare—according to public opinion polls. Yet of the first three sources used in all of the stories that were primarily about healthcare (51 sources in total), only two were coded as “experts.” To be fair, there were also a small number of sources from the health profession: doctors’, nurses’, and union representatives were coded in these stories. However, due to the fact that the nurses were conducting job action at the time due to a breakdown in collective bargaining, these quotes were primarily the result of the labour dispute making it into the election discourse, and consisted of comments about how the nurses’ job actions or bargaining positions would affect the campaign, not about the issue of healthcare per se. In addition, within these healthcare stories, almost 60% of the quotes (30) were from political candidates, and a very few (5) were from patients and their families. This suggests that although one of the most prominent issue topics in the election news was healthcare, health stories consisted of the political parties outlining their positions about healthcare or their position on the nursing dispute, but provided very little investigative research into the validity or implications of claims and comments. Few independent researchers, scholars, or practitioners were brought in to discuss the veracity of claims or provide anything more than superficial commentary.

However, and as noted above, although both experts and representatives were used almost equally as sources, the majority of representatives were used to comment in stories primarily about policy issues. Since most of this category was comprised of individuals who represented interest groups, one could presume that their views were representative of their interest group and could not be taken as an “independent” analysis of an issue. This is not to say that the views of interest groups are not valid, only that the news media appear to have relied primarily on “interested” representatives to provide information on issues—representatives who may have their own political agenda in discussing an issue—rather than independent experts. Indeed, the relative abandonment of issue discussions altogether, combined with the lack of expertise within the limited issue discussions, demonstrates how the television news media prefer to quote from organized interest groups when reporting on policy issues in an election, yet fail to provide information for voters to evaluate the various claims. Ironically, it is the news media who identify the experts as having some claim to greater knowledge of an issue and less of a political or social agenda (except in the obvious case, here, of party pundits). Yet while the claim of an expert’s “objectivity” may itself be suspect, it is noteworthy that the media does not use this expertise as much as they use representatives when exploring policy issues.

Sources as symbolic representations and story fillers

The evidence of this study suggests the need to develop a further understanding of the role of sources in the news media, beyond what the “primary definer” and “political entrepreneur” theories can accommodate. While the “primary definer” thesis sees the media as a “secondary definer” in the discourses of the powerful, and the “political entrepreneur” thesis suggests sources compete for media attention, there is evidence for further complexities to the theories of source use in the news. The evidence in this study suggests that news organizations have a substantial role in selecting and preferring certain types of sources over others and a significant role in the length of source quotes and use of source types during elections—leading us to consider adding a much more media-centric theory of source use to the list of possibilities.

For example, we could assume that one type of primary definer in an election study would be political leaders. Yet while 60% of the people quoted in the election news studied here were defined as “political actors,” only about half of the quotes from political actors were from the party leaders. Thus, while party leaders may have been the focus of the majority of news attention, they received significantly less air time than this focus might suggest. Combined with the evidence of an average quote of less than nine seconds from any political actor, the amount of significant information that can be conveyed is clearly restrictive, if not trivial. To be sure, the frequency and length of direct quotes may not indicate the influence of political actors on news discourse as a whole, yet the fact that more than three-quarters of the election stories focused on the campaign rather than policy issues suggests a media fascination with election strategies and activities. Such evidence further reinforces the suggestion that little actual information is translated by political actors during election news coverage. How much of this is due to the actions and speech of political entrepreneurs and how much is due to the choices and conventions of the news organization becomes a question for further investigation.

In addition, the increasing tendency of news coverage to use unaffiliated individuals and “streeters” in election reports indicates a media-centric preference for source types. The individual category accounted for the second-most-frequent use of sources in this election coverage, at 26% frequency. Yet while the overall average length of sound bite for all categories was just under 10 seconds, the average sound bite from the individual category was only about 6½ seconds. Such fragments of the public voice cannot reasonably be expected to offer anything more than impromptu exclamations in response to a simplified question posed by a reporter. And, as with political actors, a large majority of individuals (73%) were quoted in stories about the campaign, not about policy issues. Thus, we can safely presume that unaffiliated individuals are used by the news media as signifiers of the voting public—mere story fillers—rather than as active agents in the political process.

Interestingly, while both representatives of organizations (those speaking for their organization) and those who were coded as experts (used by the media because of their specific expertise) were quoted minimally (each about 7% of all sources), their quotes were lengthier than average—11 seconds and 13 seconds, respectively. Although still short, these lengthier quotes might suggest that the news discourse provides experts and representatives a longer time to voice their views. However, upon further investigation we find evidence for a different conclusion. For example, experts were used a staggering 82% of the time in stories about the campaign and only 18% of the time in stories about issues. Indeed, when one looks more closely at the type of expertise that the media considered important, the list is dominated by political pundits and marketing professors talking about party strategies, party advertisements, poll results, and campaign activities. Experts were rarely used to talk about policy issues. In contrast, the only source category to be used more in stories about issues was representatives—58% of the time representatives were quoted in issue stories, and 42% of the time they were quoted in campaign stories. It appears that when covering issue stories, the news media preferred to quote from interest groups and organizational representatives far more than from experts. Arguably, this tendency in source use on the part of the media provides viewers with knowledge of interest group positions on a policy issue but not necessarily information about the issue itself—and certainly not independent expertise about the policy issue.

Together, the findings in this study point to a much more media-centric view of source use in election news. Political actors and unaffiliated individuals form the largest categories of source use in the news, but both groups are highly mediated by journalistic discourse, as evidenced by the limited length of their quotes. Further, the news topics are dominated by campaign-related stories rather than policy-related stories, suggesting that these sources were used as “symbolic representations” of their role in the election “race” rather than as substantive definers of the news content. It could be argued that political actors and individuals were used primarily as story fillers to pad or punctuate an already framed news story, and experts were sought out by the media to comment on campaign strategy and political marketing rather than policy issues. Thus, as stated earlier, it may be that journalists rely on sources, at least in part, to symbolically (rather than substantively) represent the story they have already chosen to tell. Such a suggestion raises significant questions about the intentions of news organizations and the institutional and social function of the news media in democratic societies. To what extent do the news media themselves become the primary definers of political news? What stories do the news media choose to tell about elections, and why? An exploration of these questions would require a broader sociological and institutional analysis of news organizations that is clearly beyond the scope and methodology of the present article. Nonetheless, this study provides an evidence-based addition to the dominant theories of source use in news media.


1. Technical difficulties resulted in some data being unavailable for this study. One broadcast was unavailable from Global (April 17), and two from BCTV (May 9 and May 15). Of the 22 potential broadcasts from CBC (they did not run 6 p.m. newscasts on weekends), three were unavailable. Thus, the total number of CBC broadcasts used for this data was 19. And finally, technical difficulty also resulted in taping of the 11 p.m. newscast rather than the 6 p.m. broadcast on VTV (May 9). The data in this study, then, represents most but not all of the television newscasts during the election. Because of this, care was taken in making overall volume assumptions and per-station conclusions.


Attallah, Paul. (2004). Television and the Canadian federal election of 2004. In Jon H. Pammett & Christopher Dornan (Eds.), The Canadian general election of 2004 (pp. 264-289). Toronto, ON: Dundurn Group.

Attallah, Paul, & Burton, Angela. (2001). Television, the Internet, and the Canadian federal election of 2000. In Christopher Dornan & Jon H. Pammett (Eds.), The Canadian general election of 2000 (pp. 215-242). Toronto, ON: Dundurn Group.

Beharrell, Paul, & Philo, Greg (Eds.). (1977). Trade unions and the media. New York, NY: Macmillan.

Charron, J. (1991). Relations between political parties and the media in Quebec election campaigns. In Frederick J. Fletcher (Ed.), Reporting the campaign: Election coverage in Canada (pp. 79-145). Toronto, ON: Dundurn Press.

Crete, J. (1991). Television, advertising and Canadian elections. In F. J. Fletcher (Ed.), Media and voters in Canadian election campaigns (pp. 3-44). Toronto, ON: Dundurn Press.

Curran, James. (2000). Rethinking media and democracy. In James Curran & Michael Gurevitch (Eds.), Mass media and society (3rd ed.) (pp. 120-154). London: Arnold.

Davis, Howard H., & Walton, Paul (Eds.). (1983). Language, image and media. Oxford: Basil Blackwell.

Dornan, Christopher. (1997). The television coverage: A history of the election in 65 seconds. In Alan Frizzell & Jon H. Pammett (Eds.), The Canadian general election of 1997 (pp. 149-166). Toronto, ON: Dundurn Press.

Dornan, Christopher, & Pyman, Heather. (2001). Facts and arguments: Newspaper coverage of the campaign. In Christopher Dornan & Jon H. Pammett (Eds.), The Canadian general election of 2000 (pp. 191-214). Toronto, ON: Dundurn Group.

Ericson, Richard, Baranek, Patriciam, & Chan, Janet. (1989). Negotiating control: A study of news sources. Toronto, ON: University of Toronto Press.

Fishman, Mark. (1980). Manufacturing the news. Austin, TX: University of Texas Press.

Fletcher, Fred J., & Everett, Robert. (1991). Mass media and elections in Canada. In F. J. Fletcher (Ed.), Media, elections and democracy (pp. 179-222). Toronto, ON: Dundurn Press.

Frizzell, Alan, & Westell, Anthony. (1989). The media and the campaign. In Alan Frizzell, Anthony Westell, & Jon H. Pammett (Eds.), The Canadian general election of 1988 (pp. 75-90). Don Mills, ON: Carleton University Press.

Gans, Herbert J. (1979). Deciding what’s news: A study of CBS Evening News, NBC Nightly News, Newsweek and Time. New York, NY: Vintage Books.

Gilsdorf, W. O., & Bernier, R. (1991). Journalistic practice in covering federal election campaigns in Canada. In F. J. Fletcher (Ed.), Reporting the campaign: Election coverage in Canada (pp. 3-77). Toronto, ON: Dundurn Press.

Hackett, Robert A. (1983). The depiction of labour and business on national television news. Canadian Journal of Communication, 10(1), 5-50.

Hackett, Robert A. (1985). A hierarchy of access: Aspects of source bias in Canadian TV news. Journalism Quarterly, 62(2), 256-265, 277.

Hackett, Robert A., & Hissey, L. (1995). News media and party politics in Canada: Perspectives on a shifting relationship. In H. Thornburn (Ed.), Party politics in Canada (7th ed.) (pp. 34-53). Scarborough, ON: Prentice Hall Canada.

Hall, Stuart, Clark, John N., Critcher, Chas, Jefferson, Tony, & Roberts, Brian. (1978). Policing the crisis: Mugging, the state, and law and order. London: Macmillan.

Hallin, Daniel C. (1992). Sound bite news: Television coverage of elections, 1968-1988. Journal of Communication, 42(2), 5-24.

Hartley, John. (1982). Understanding news. New York, NY: Routledge.

Herman, Edward S., & Chomsky, Noam. (1988). Manufacturing consent: The political economy of the mass media. New York, NY: Pantheon Books.

Johnston, Richard, Blais, André, Brady, Henry E., & Crete, Jean. (1992). Letting the people decide: Dynamics of a Canadian election. Montréal, QC: McGill-Queen’s University Press.

Langer, John. (1998). Tabloid television: Popular journalism and the ‘other news.’ New York, NY: Routledge.

Papas, Lisa Katherine. (1991). Notable quotables: A study of source selection in local television news. Unpublished master’s thesis. Kingston, ON: Queen’s University.

Schlesinger, Phillip. (1990). Rethinking the sociology of journalism: Source strategies and the limits of media centrism. In M. Ferguson (Ed.), Public communication: The new imperatives (pp. 61-83). London: Sage.

Silva, Edward. (1995). More perishable than lettuce or tomatoes: Labour law reform and Toronto’s newspapers. Halifax, NS: Fernwood Publishing.

Soderlund, Walter C., Romanow, Walter I., Briggs, E. Donald, & Wagenberg, Ronald H. (1984). Media and elections in Canada. Toronto, ON: Holt, Rinehart & Winston of Canada.

Tuchman, Gaye. (1978). Making news: A study in the construction of reality. New York, NY: Free Press.

Walton, Paul, & Davis, Howard. (1977). Bad news for trade unionists. In Peter Beharrell & Greg Philo (Eds.), Trade unions and the media (pp. 118-137). New York, NY: Macmillan.

  •  Announcements
    Atom logo
    RSS2 logo
    RSS1 logo
  •  Current Issue
    Atom logo
    RSS2 logo
    RSS1 logo
  •  Thesis Abstracts
    Atom logo
    RSS2 logo
    RSS1 logo

We wish to acknowledge the financial support of the Social Sciences and Humanities Research Council for their financial support through theAid to Scholarly Journals Program.