Who’s Citing Whom?

Abstract

In a news environment that falls short in covering public priorities, debates play an important role in bringing electoral politics closer to constituents. Given that political elites routinely influence campaign news through the gatekeeping process of source selection and that journalists facilitate debates, it is worth examining the indexing of elite sources in these campaign events. This study uniquely explores the routines of news professionals in debates by examining source selection. This study finds that debate questions citing exogenous sources tap political elites far more frequently than members of the public and identifies media characteristics predictive of elite indexing.

Keywords

press performance content analysis news coverage and reporting political communication

News professionals seldom operate in a vacuum. News content is uniquely influenced by various institutional norms and routines: processes that bind the profession, build credibility, and save time. Through the gatekeeping process of source selection, exogenous influences—including political elites and the public—can play a role in framing news content (Cook, 1998); however, not all political players are equally influential in shaping the news. News content is shaped by the routines of the trade, including the indexing of elite sources. Given that news professionals are trained to rely on “official” sources to build news credibility, political elites play a more active role in framing news relative to members of the public (Entman, 2004).

The talking points of political elites routinely attract news attention during elections (Just et al., 1996; Kaid, McKinney, & Tedesco, 2000; Kiousis, Kim, McDevitt, & Ostrowski, 2009). Despite amplifying the messages of elites, election coverage often diverges from the priorities of the public (Just et al., 1996; Patterson, 1994). This pattern is also observed in electoral debates. A key function of these campaign events is to make elections more responsive and accountable to public information needs (Jamieson & Birdsell, 1988); however, research has shown that debates often neglect public interests (Benoit & Hansen, 2001; Jackson-Beeck & Meadow, 1979; McKinney, 2005). These patterns are important given that debates have the power to shift the election priorities of the public (Swanson & Swanson, 1978).

Because debates are presumed to reflect information needs of the electorate, it is important to understand which stakeholders are framing the debates. Although research has explored the issue topics shaping debate agendas (see Carlin, Vigil, Buehler, & McDonald, 2009; Jackson-Beeck & Meadow, 1979) and the sources of frames in campaign news, we know far less about the forces influencing electoral debates. Of particular interest is the extent to which indexing routines of news professionals influence elections beyond campaign coverage. Considering that news professionals facilitate electoral debates and that journalists function according to norms and routines of the trade, just how prevalent is the news media’s indexing bias toward elites during the gatekeeping process of source selection?

This study explores the exogenous influences framing electoral debates by examining the sources referenced in the questions posed to the candidates. In content analyzing the nature of the sources cited in debate questions across electoral contexts, I find that debate questions cite political elites far more frequently than members of the public, and that the indexing of elites is conditioned by variables specific to media characteristics of the participating journalists.

Gatekeeping Functions of the Press

In an information-rich society, journalists function as gatekeepers determining which issues reach the public domain. At its core, gatekeeping refers to news judgment, which White (1950) found to be influenced by individual journalists’ attitudes and predispositions. The gatekeeping process goes beyond decisions that determine which matters make the news agenda to also include news judgments influencing content as well, including the sources journalists choose to cite. The gatekeeping literature has developed considerably since White’s seminal study; scholars have since demonstrated that gatekeepers of news include both news professionals and various news norms and routines (Shoemaker, Eichholz, Kim, & Wrigley, 2001; Shoemaker & Reese, 1996; Shoemaker, Vos, & Reese, 2009).

News norms and routines provide a theoretical framework for explaining the production of news. This framework suggests that the gatekeeping process is highly influenced by institutional norms, processes, and expectations of the business. A prominent news routine is the emphasis of “official” sources that supply news content and establish story credibility. Such news routines, as Ryfe (2006) explains, function as tacitly understood imperatives. Given that many news routines are not explicitly expressed, these embedded practices are learned through training and socialization. The implicit nature of these routines may also explain why the early gatekeeping literature focused on individual-level influences of news.

The work of Shoemaker and colleagues (2001) discovered that individual journalist characteristics were poor predictors of which topics advanced through the gatekeeping process. Research instead points to news routines as a more reliable predictor of the gatekeeping process (Ross, 2007; Shoemaker et al., 2001; Shoemaker & Reese, 1996). News routines essentially act as “surrogates or shortcuts” for the gatekeeping decisions of individual journalists (Shoemaker et al., 2001, p. 235). News is not the product of a single gatekeeper but, rather, passes through multiple gates within a news organization. Editorial staff are the arbiters of news-making decisions; however, exogenous influences can also shape this process. Although news routines save journalists time and simplify news production, these routines are not without their (unintended) consequences.

Gatekeeping decisions ultimately frame news stories. In emphasizing officials as primary sources, the gatekeeping process limits political discourse to political elites (Page, 1996). Given that news professionals tap official sources to establish story credibility, frames reflected in the news often reflect the frames of elites (Bennett, Lawrence, & Livingston, 2006; Entman, 2004). Although the gatekeeping process is well documented, research has not fully explored this process in non-news contexts such as electoral debates. Moreover, Shoemaker et al. (2009) suggest that the existing literature too often emphasizes market influences of news decisions in lieu of institutional norms and routines. Thus, this study explores the gatekeeping process of source selection through the theoretical framework of news norms and routines to understand influences of debates. To demonstrate the importance of examining exogenous debate influences, I discuss the implications of the gatekeeping process of source selection relative to election frames.

Indexing: Elites, Frames, and Campaign News

Framing theory makes sense of the communication environment by defining events, issues, and problems through attribute emphasis. As Entman (1993) explains, framing emphasizes some aspects of an event over others and renders those aspects more salient to the public. Framing concerns not only the presence of attributes but also the concealment of attributes (Goffman, 1974). Although political elites utilize framing intentionally and strategically, journalists subconsciously frame news to simplify events and issues (Entman, 2004). These frames not only organize complex matters for news audiences but also inform public opinion (Entman, 2004; Iyengar, 1991). Entman, Matthes, and Pellicano (2009) contend that research has not devoted enough attention to frame formation. Moreover, the routines and processes of the news trade have unintended effects over the framing of news stories.

Although objectivity norms discourage journalists from actively framing news, the sources they rely on function as a key level of framing (see Pan & Kosicki, 1993; Scheufele, 1999). Thus, the gatekeeping decisions related to source selection holds serious implications for how elections are framed. Although news professionals avoid intentional framing of news stories, their quest for credibility in the form of citing “official” sources enables elites to engage in frame formation. This news routine is best understood through the indexing literature.

Cook’s (1998) theory of an institutional press suggests some degree of uniformity in news gathering processes. One of these uniform processes includes the routine emphasis of “official” sources, or the indexing of elite perspectives (Bennett, 1990; Bennett, Lawrence, & Livingston, 2007; Cook, 1998; Zaller & Chiu, 1996). This news routine, as Bennett (1990) noted, provides a megaphone for political elites and often limits the scope of viewpoints to “official” sources. Given that the media depend on official sources to supply news and establish story credibility, political elites play key roles in framing elections. The indexing process ultimately empowers political elites to frame campaigns. Studies (see Berkowitz & Adams, 1990; Gilbert, Eyal, McCombs, & Nicholas, 1980; Weaver & Elliott, 1985) have found that political elites are highly influential in shaping campaign coverage, and the emphasis on elite sources results in news professionals neglecting counter perspectives and frames (Bennett et al., 2006). Although this pattern of elite indexing does not hold for all types of news coverage (see Althaus, 2003; Salwen, 1995), it is a pattern commonly observed in the context of campaigns and elections. In fact, elites dominate source influence even in democratizing news platforms, as evident from research exploring 2008 election news delivered through YouTube (Dylko, Beam, Landreville, & Geidner, 2011).

Scholarship has demonstrated that during elections, news professionals are highly attentive to the talking points of political elites as a result of the gatekeeping process and indexing norms. Because debates are moderated by news professionals, such routines may also influence related campaign events. Given that research found that debate and public agendas rarely align (Benoit & Hansen, 2001; Carlin et al., 2009), assessing the exogenous sources cited in debate questions may shed light on this discrepancy. Cook’s (2006) work outlining institutional news routines suggests that routines may vary among news organizations, and recent work has shown that media characteristics influence the frames found in election news including organizational variables and the type of media outlet (Dimitrova & Kostadinova, 2013; Stromback & van Aelst, 2010). Thus, one might expect that the attentiveness to political elites in debates is contingent on various media characteristics of the participating news outlets and journalists.

Although the influence of political elites is evident in the sources cited by the press, perspectives from the public are absent from campaign coverage (Berkowitz, 1987; Entman, 2004). Since the first town hall debate in 1992, formats involving public input have increased. Research (Benoit & Hansen, 2001; McKinney, 2005) has indicated that town hall formats are more adept at addressing public concerns; this finding, coupled with the fact that citizens are not trained to behave according to news norms and routines, suggests that the public should be less concerned about the talking points of elites (and less likely to cite elites when questioning the candidates). For formats solely tapping news professionals to set the agenda, questions citing political elites should be more prevalent than what is observed for the town hall format that largely solicits questions from the public. Therefore, this study hypothesizes as follows:

H1a: Questions in town hall formats are less likely to cite political elites.

H1b: Questions in town hall formats are more likely to cite the public.

H1c: Questions in media-controlled formats are more likely to cite political elites.

H1d: Questions in media-controlled formats are less likely to cite the public.

Although indexing elite sources is a common news routine across media, commercial pressures should also play a role in echoing political elites because doing so saves time and costs. Prior research (Dunaway, 2008; Hamilton, 2004) has shown that commercial incentives influence news content. News organizations beholden to corporate shareholders provide more strategic narratives of campaigns, often emphasizing the agenda of political elites by covering candidate statements and campaign ads. Thus, nonprofit journalists should have less incentive to cite elites in their debate questions due to the absence of profit-driven motivations.

The work of Weaver and Elliott (1985) also found differences between local and national news in the ability of elites to shape the news; however, this variable has not been addressed in the debate literature. The work of Farnsworth and Lichter (2011) suggests that campaign coverage is less substantive for national “celebrity” journalists. Relatedly, local journalists who do not contend with the same competitive pressures as news professionals on the national stage may operate under varied news norms—and that may be reflected in source selection. As more local moderators are replaced by national news professionals (Schroeder, 2000), it is important to understand whether these journalists function similar to one another in the debate context. Because local journalists hold greater proximity to their news audiences, their reliance on elite sources should be less frequent than those working for national media who, as Fallows (1996) suggests, have more in common with their elite sources than the public they serve. Inversely, cable journalists with greater news holes to fill must frequently look to “official” sources to supply content; these journalists should be more likely to cite political elites when questioning the candidates. Thus, a moderator’s media background may predict which exogenous sources frame the debates; therefore, this study hypothesizes as follows:

H2a: Questions in debates with nonprofit moderators are less likely to cite political elites.

H2b: Questions in debates with nonprofit moderators are more likely to cite the public.

H2c: Questions in debates with local moderators are less likely to cite political elites.

H2d: Questions in debates with local moderators are more likely to cite the public.

H2e: Questions in debates with cable moderators are more likely to cite political elites.

H2f: Questions in debates with cable moderators are less likely to cite the public.

Because moderating journalists cannot dictate total control over debates in press panel formats and members of the public are no longer the lone questioners in town hall debates, it is also important to isolate stakeholder influence at the source level.¹ Applying the same rationale guiding the moderator variables, this study also hypothesizes the following:

H3a: Questions from the public are less likely to cite political elites.

H3b: Questions from the public are more likely to cite the public.

H3c: Questions from nonprofit journalists are less likely to cite political elites.

H3d: Questions from nonprofit journalists are more likely to cite the public.

H3e: Questions from local journalists are less likely to cite political elites.

H3f: Questions from local journalists are more likely to cite the public.

H3g: Questions from cable journalists are more likely to cite political elites.

H3h: Questions from cable journalists are less likely to cite the public.

Research has also suggested that medium affects the success elites have in influencing news content. Ridout and Mellen (2007) found that print news agendas are more likely than broadcast news agendas to converge with candidate agendas. Thus, one should expect the medium of the participating journalists to affect the gatekeeping function of source selection in debate questions. Therefore the study hypothesizes as follows:

H4: Questions from print journalists are more likely to cite political elites.

Although underdeveloped in the literature, sponsorship could hold underlying influence over debates considering profit-driven motivations are known to influence campaign news (see Dunaway, 2008; Hamilton, 2004). To explore this variable, this study asks the following:

RQ1: Does commercial media sponsorship influence the exogenous sources cited in debate questions?

Data and Method

To test these hypotheses, a data set of televised debate questions is developed from transcripts across three electoral contexts: (a) U.S. general election debates, (b) primary presidential debates, and (3) state-level debates. By including every question (N = 1,950) from the selected transcripts, sampling bias is minimized by preventing the possibility of selecting questions that best fit expected outcomes. Bias is also minimized by limiting the sample to debates in which complete transcripts are available. Thus, each unique question asked in the transcripts is included in the data set.²

The data set includes every question from televised general election presidential and vice presidential debates (n = 658), 1960-2012. Transcripts were obtained through the Commission on Presidential Debates’ online archive. Given that primary debates have become more prolific in recent years, the data set also includes a subsample of primary debate questions (n = 981). The availability of full transcripts for primary debates is limited, and they were collected through University of California (UC), Santa Barbara’s American Presidency Project; the subsample spans from 2000 to 2012. This sample draws from a mix of debates featuring both parties and encompasses debates throughout the primary process—those held early with numerous candidates and those falling closer to the general election when parties have reduced potential nominees to two candidates. This subsample is also inclusive of press panels, single moderator, and town hall formats.

To generalize findings to state-level elections, a subsample of U.S. Senate and gubernatorial debates is also included. Again, availability of complete transcripts from older election years is sparse. Thus, the debates in this subset (n = 311) span from 2000 to 2012. The transcripts include debates held in both midterm and presidential years and offer balance among format diversity and geographic location. One sampling limitation results from the fact that published Senate and gubernatorial debate transcripts are more accessible for high profile, competitive races—a variable controlled for in the models. State-level transcripts were obtained from reputable news outlets and Purdue University’s Cable-Satellite Public Affairs Network (C-SPAN) Archives.

Dependent Variables

Debate questions were content analyzed for embedded citations. To measure the dependent variables, three research assistants were trained to assess whether the question contained a citation of exogenous influence and assess the nature of the source cited.³ This methodological approach follows research exploring the news influence of elites as measured by the sources cited by news professionals in their stories (see Berkowitz, 1987). Research assistants coded each question using a mutually exclusive categorical citation variable in which they assigned values based on the exogenous source cited. Coders assigned a 0 for those without an exogenous reference and distinct values for the following citations: candidate statements, campaign staffers, political ads, public officials, polls, members of the public, and “other” sources (see Appendix).⁴

Before analyzing the data, 66 questions posed by candidates to other candidates were eliminated from the data set. The data showed that 35.2% of questions (n = 607) cited an exogenous influence. Next, questions citing candidates, campaign communications, political ads, and other political elites were collapsed into a political elite dummy variable assigning 1 for questions citing political elite and 0 for all else. After collapsing these categories, 29% of the data set (n = 547) reflected questions citing political elites. A second public dummy variable parsing out questions citing the public (1) from those that did not (0)⁵ was generated from the categorical citation variable. For questions citing members of the public, frequencies were markedly low, accounting for 3.2% (n = 60) of the total sample. This suggests that source selection in debates follow patterns of elite indexing present in campaign news.

Independent Variables

To examine the relationship between the media and elite indexing, the study first includes coding for debate format to allow for more systemic comparisons between press questions and questions generated from citizens. In drawing a distinction between press panels (incorporating several participating journalists) and single moderator formats, comparison of two formats with varying levels of press input is possible. Questions appearing in press panel debates were assigned a 1; those in single moderator debates a 2; and those in town halls a 3.

Variables addressing the professional background of question sources and moderators are important considering scholars have found that campaign news varies according to profit pressures. Two additional measures of influence are also accounted for: moderator characteristics and question source. These variables offer important distinctions of attribution. The question source variable labels each question according to who asked it. Questions were assigned a 1 when asked by the public, a 2 when asked by a network journalist, a 3 for cable journalist, a 4 for local broadcast journalist, a 5 for a national print journalist, a 6 for local print journalist, a 7 for nonprofit journalist, an 8 for persons from a non-media nonprofit, a 9 for wire service, a 10 for radio journalist, an 11 for online journalist/blogger, and a 12 for candidate source.⁶ Given that print news agendas are more likely to converge with candidate agendas (Ridout & Mellen, 2007), the model tests medium at the moderator and source level by accounting for whether the journalist works in print or broadcast news. These dichotomous variables are derived from the question source coding parameters outlined above, and all moderator variables follow an identical coding scheme.

To address RQ1, questions were coded according to whether they came from a debate sponsored by commercial media. This dichotomous media sponsorship variable treats any debate with sole sponsorship or co-sponsorship by commercial media as 1 whereas questions coming from debates with strictly nonprofit sponsorship are coded as 0.⁷

Control Variables

To better isolate predictors of exogenous influences, the models also control for debate and electoral context. Because campaign coverage has grown more infotainment driven over time, drifting further from the election priorities of the public (Delli Carpini & Williams, 2001), debate year is included as a continuous variable. The model also controls for debate type by indicating whether the question came from a presidential (1), vice presidential (2), Republican primary (3), Democratic primary (4), Senate (5) or gubernatorial (6) debate. (Senate and gubernatorial cases were collapsed into a single state-level variable). State-level debate agendas may be more likely to cite public concerns based on the sheer proximity of these localized races to the public.

A debate topic variable categorizes questions according to whether they were asked in a domestic policy debate (1), mixed-agenda debate (2), or foreign policy debate (3). Because news professionals are especially reliant on official sources when covering foreign affairs (Bennett et al., 2007; Zaller & Chiu, 1996), these designated topics should also influence the propensity to cite political elites. Second, two continuous variables measuring debate length (in minutes) and question response time (in seconds) were included because lengthier debates and response times may influence the substance of debate questions.

Research (Walgrave & Van Aelst, 2006; Wanta, 1991) has shown that the ability of political elites to influence the news is also conditioned by political context; thus, electoral controls are established by including candidate and contest-specific variables. For example, the number of participating candidates in a debate may increase opportunities to cite political elites. Thus, a continuous candidate number variable records the number of candidates participating in the debate. This variable also serves as a proxy for the life cycle of a campaign, with higher numbers of candidates indicative of questions asked earlier in the campaign process.⁸

Election news coverage is also conditioned on the incumbency factor (Arnold, 2004; Cohen, 2008), and research has shown that sitting candidates—particularly those with strong popularity—hold considerable influence over news content (Wanta, 1991). This is controlled for as a dichotomous (1/0) variable indicating the presence of a participating incumbent. Finally, a competition variable is introduced to measure the tightness of the race. As Hayes (2010) observed, news professionals are more likely to diverge from elite talking points in competitive races. Thus, competition is measured as a continuous variable based on the spread of final vote count percentage.⁹

Results

Debate Format and Question Cites

In first running correlations, the analysis showed a negative relationship between the town hall format and questions citing political elites, r(1,882) = −.15, p < .01. This suggests that debates primarily driven by the public are less likely to regurgitate the frames of political elites. Although the town hall format moved in a positive direction with questions citing the public, the relationship fell short of reaching statistical significance. For both formats in which journalists control the agenda, a positive relationship is observed with questions citing political elites, albeit the relationship is only significant for the press panel format. The data showed a positive relationship between press panels and questions citing political elites, r(1,882) = .13, p < .01. However, significant relationships were not found for the press panel or single moderator format and public cites. In short, press-influenced formats differ in the extent to which they follow elite indexing news norms.

Moderator Characteristics and Question Cites

For the moderator variables, a statistically significant relationship is not found for nonprofit moderators and cites for political elites. The analysis shows a negative relationship for the moderating nonprofit journalist variable and public cites, r(1,882) = −.05, p < .05, but the direction counters expectations. As expected, a negative relationship is observed between debates with moderating local journalists and questions citing political elites, r(1,882) = −.11, p < .01. Inversely, the data show a positive relationship between moderating local journalists and questions citing the public, r(1,882) = .21, p < .01. This finding falls in line with expectations based on proximity of local journalists to their news audiences. Surprisingly, the data show a negative relationship between cable news moderators and questions citing political elites, r(1,882) = −.09, p < .01. Nonetheless, a significant relationship between cable moderators and questions citing the public is not observed. These patterns are somewhat surprising and warrant closer examination in the regression models.

Source and Question Cites

The analysis also reveals a negative relationship between questions asked by the public and cases citing political elites, r(1,882) = −.15, p < .01, whereas questions asked by the public are more likely to cite members of the public, r(1,882) = .07, p < .01. At the source level, the data surprisingly show a positive relationship between nonprofit journalists and questions citing political elites, r(1,882) = .05, p < .05. Paralleling the relationship for the nonprofit moderator variable, a negative relationship between nonprofit journalist sources and questions citing the public is observed, r(1,882) = –.05, p < .01. These findings suggest that profit motivations are less influential in shaping debate content than they are found to be in shaping campaign news.

The data also show a negative relationship between local press sources and questions citing political elites, r(1,882) = −.10, p < .01, reinforcing the negative relationship found between local moderating journalists and cases citing elites. For local journalist sources, the data reveal a positive relationship with questions citing the public, r(1,882) = .09, p < .01. The relationship between cable news sources and questions citing political elites did not reach statistical significance; however, the data show a negative correlation between cable sources and questions citing the public, r(1,882) = –.07, p < .01. Finally, print journalist sources were not significantly correlated with questions citing political elites; however, broadcast sources held a positive and significant relationship with questions citing political elites, r(1,882) = .12, p < .01. Thus, the proximity of the journalists to their news audiences and the nature of the news outlet seem to play a role in influencing source selection.

Sponsorship

Although commercial media sponsorship did not correlate with questions citing the public, the data indicate that sponsorship plays a role in political elite influence. Questions asked in debates with a commercial sponsor held a positive relationship with citations from political elites, r(1,882) = .05, p < .05. In other words, pressures related to generating mass audiences may also shape the gatekeeping process of source selection. Next, these relationships are assessed after introducing control variables in a regression model.

Regression Model: Political Elite Cites

Because the dependent variables are dichotomous, these relationships are tested in a logistic regression model controlling for electoral context and debate rules. First, the study found that the sponsorship variable demonstrates that questions asked in debates with commercial media sponsorship are 1.6 times as likely to cite political elites as those asked in forums without a media sponsor. This finding reinforces justification for RQ1 exploring media sponsorship as an influencer of debates. The study suggests that news professionals participating in commercially sponsored debates are more responsive to political elites during the source selection process. The finding holds implications for the ability of elites to frame debates, as these campaign events continue trending toward commercial sponsorship models (see Schroeder, 2000).

For the format variables, one of two relationships holds (see Table 1). Although the town hall format expectedly yields a negative relationship with questions citing political elites, the relationship is insignificant. Thus, H1a is not supported. The null finding could be explained by journalists interjecting in town hall formats with their own questions, which threatens the authenticity of these town hall style forums. The press panel variable performs as expected, with questions in this format being more than 1.6 times as likely to cite political elites as questions from other debate formats. Therefore, partial support for H1c holds after controlling for other influences, meaning that formats with the greatest level of press influence (panels featuring multiple journalists) are most likely to follow elite indexing norms observed in campaign coverage. This finding supports the notion that news routines are often learned through socialization: When in the presence of other news professionals, journalists are especially keen on tapping “official” sources. Nonetheless, format is less influential in predicting elite indexing norms than moderator and source variables.

Table 1.

Logistic Regression Model: Predictors of Political Elite Cites (N = 1,884).

Predictor	B	SE B	Wald	Exp(B)
Media variables
Press panel format	0.47**	0.14	10.85	1.60
Town hall format	−0.28	0.29	0.97	0.75
Print moderator	−0.22	0.33	0.44	0.80
Local moderator	−0.87*	0.36	5.75	0.42
Cable moderator	−0.11	0.15	0.52	0.89
Media sponsor	0.46*	0.20	4.94	1.58
Nonprofit source	2.09**	0.32	42.62	8.11
Public source	−1.76**	0.32	30.71	0.17
Broadcast source	0.30	0.16	3.39	1.35
Local source	−0.83**	0.22	14.70	0.44
Electoral controls
Election year	−0.01	0.01	2.94	0.99
Candidate No.	−0.03	0.04	0.44	0.97
Competitiveness	−0.01	0.01	1.01	0.99
Incumbent	0.17	0.17	0.93	1.18
State-level	−0.22	0.26	0.71	0.80
Debate context
Length	0.00	0.00	0.05	1.00
Response time	0.00	0.00	0.59	1.00
Domestic policy	−0.28	0.15	3.17	0.76
Constant	26.45	16.13

Note. χ² = 209.69, df = 18, log likelihood = −1,030.18; Nagelkerke R² = .15.

p < .05. **p < .01.

Although the model found that cable moderators yield a slight negative relationship with political elite cites, the relationship does not reach statistical significance (therefore, H2e is not supported); however, local moderators hold a significant negative relationship with questions citing elites. From the model, all other variables are held at their mean levels to show that the local moderator variable yields a predicted probability of .133, whereas questions stemming from debates moderated by national journalists have a predicted probability of .268 (see Figure 1). This reflects a change of .135, or an overall increase in political elite cites of 50.4% when national journalists are at the helm of a debate. Thus, the expectation outlined in H2c is supported. Moreover, this relationship is corroborated at the source level. Questions asked by local journalists hold a negative relationship with political elite cites; H3e is also supported. This implies that proximity to news audiences may increase the spectrum of perspectives indexed in debates, and that local journalists relax elite indexing norms relative to national journalists.

Figure 1.

Predicted probabilities: Local moderators and question cites.

Another source that diminishes the likelihood of elite indexing includes questions asked by members of the public. Public sources hold a negative relationship with political elite cites, demonstrating support for H3a. The predicted probability of the public source variable is computed by holding all other variables in the model at their mean values: Questions asked by the public yield a predicted probability of .093, whereas those from all other sources hold a predicted probability of .372 (see Figure 2). These values reflect a difference of .279 or an overall increase in elite cites of 75% for questions asked by news professionals. This finding is unsurprising given that members of the public have not been socialized to news norms and routines. When it comes to predictors of elite indexing, the strongest comes from an unlikely source: nonprofit journalists.

Figure 2.

Predicted probabilities: Public source and question cites.

Questions from nonprofit journalists were more than 8 times as likely to cite political elites as those from other sources. Therefore, H3c is not supported. To better explain this relationship, all other variables in the model are held at their mean levels to generate a predicted probability of .678 for nonprofit sources whereas those asked by all other sources hold a predicted probability of .206 (see Figure 3). This shows a difference of .472 or an overall increase in political elite cites for questions asked by nonprofit sources of 69.6%. Albeit surprising, the effect is both statistically significant and sizable. Again, this finding seems to suggest the profit incentives shaping campaign news matter less in the context of electoral debates. The emphasis on elite sources from nonprofit journalists may result from a possible disconnect with news audiences given that they are national news professionals.

Figure 3.

Predicted probabilities: Nonprofit source and question cites.

Finally, the model showed a lack of support for H4, meaning print journalist sources did not correlate with questions citing political elites; after introducing controls, a lack of relationship was also noted for broadcast journalists. In other words, elite influence is not affected by the medium as observed in patterns of campaign news.

Regression Model: Public Cites

In a second model testing predictors of public cites, the study found that election year holds a positive and significant relationship, demonstrating that with each one-unit change in election year, there is a modest increase in questions citing the public (see Table 2). This finding is unsurprising considering that the advent of the town hall format in 1992 transferred some agenda-setting functions of the press to the public. For the independent variables of interest, the model demonstrates that several relationships hold.

Table 2.

Logistic Regression Model: Predictors of Public Cites (N = 1,884).

Predictor	B	SE B	Wald	Exp(B)
Media variables
Print moderator	0.58	0.72	0.65	1.79
Nonprofit moderator	−0.44	0.55	0.63	0.65
Local moderator	2.39**	0.60	15.86	11.01
Nonprofit source	−2.73**	0.82	11.23	0.07
Public source	1.95**	0.62	9.92	7.01
Broadcast source	−1.06	0.59	3.16	0.35
Cable source	−1.92	1.05	3.34	0.15
Local source	0.19	0.49	0.16	1.22
Electoral controls
Election year	0.08*	0.03	6.04	1.08
Candidate No.	−0.06	0.11	0.36	0.94
Competitiveness	0.03	0.02	2.94	1.03
Incumbent	0.46	0.49	0.88	1.58
State-level	0.51	0.77	0.44	1.67
Debate context
Length	0.00	0.01	0.08	1.00
Response time	0.00	0.01	0.00	1.00
Domestic policy	0.46	0.48	0.93	1.58
Constant	−165.37*	65.42

Note. χ² = 95.52, df = 16; log likelihood = −218.08, Nagelkerke R² = .20.

p < .05. **p < .01.

First, the local moderator variable not only performs as expected but also the effect is strong and statistically significant. Questions from forums moderated by local journalists are more than 11 times as likely to cite members of the public as questions from debates facilitated by other moderators. From the model, all other variables are held at their mean values demonstrating that questions in debates with local moderators have a predicted probability of .108 for citing the public; questions in debates with other moderator types hold a predicted probability of .011 (see Figure 1). This reflects a difference in −.097 or an overall decline of approximately 89.8%. With local journalists having greater proximity to their news audiences, they may be more in tune with public priorities and, consequently, less reliant on elite indexing. In sum, local moderators are a strong predictor of questions citing the public; therefore, H2d is supported. The stronger measure of source attribution for local journalists also moves in the expected direction, but the relationship falls short of reaching the .05-level of statistical significance. Therefore, H3f is only marginally supported, and future research must continue to explore the relationship between local journalists and indexing norms before these findings are fully attributed to news audience proximity.

As expected, the cable source variable decreases the likelihood of questions citing the public, but this negative relationship merely approaches statistical significance (p < .10). Thus, the data do not fully support H3h; however, nonprofit press sources hold a significant negative relationship with questions citing the public. When holding all other variables in the model at their mean levels, the data showed that nonprofit sources have a predicted probability of .000, whereas all other question sources hold a predicted probability of .018 (see Figure 3). This demonstrates a difference of nearly 100%, suggesting that nonprofit sources seldom cite the public. Thus, H3d is not supported. Because nonprofit journalists were a predictor of elite indexing, it seems that opportunities for citing the public are reduced, and this neglect of non-elites could be a by-product of reduced proximity to news audiences.

Although questions citing the public are chronically low, they are often driven by members of the public participating in the debate process. Unsurprisingly, questions from the public are 7 times more likely to cite the public than questions from news professionals. Holding all other variables at their mean values, the data show a predicted probability of .047 for the public source variable whereas all other sources generate a predicted probability of .001 (see Figure 2). This reflects a change of −.046 or an overall difference of 97.9%. Therefore, H3b is supported, and public question sources are much stronger predictors of non-elite source selection than town hall formats that remain susceptible to influence from moderating news professionals. The implications of these findings are discussed below.

Discussion

When comparing sources of exogenous influence, the gatekeeping process of source selection parallels the process in campaign news insofar that political elites are more commonly cited than the public. This suggests that debates are rather susceptible to elite frames and are shaped more by elites than the electorate. In fact, questions citing the public account for fewer than 5% of cases in the data set. This is unsurprising considering that news professionals are conditioned to seek out “official” sources (Bennett, 1990; Cook, 1998). It is worth noting that the evolution of debate format (e.g., town halls) has had some success in making the process more responsive to the public, as the electoral year variable shows debate questions are more likely to cite the public over time. It is also worth noting that the Commission on Presidential Debates eliminated the press panel format from general election debates in 1996—the format found to be most predictive of questions citing political elites.

This study suggests that debate questions are more likely to echo talking points of political elites than members of the public. This reinforces prior research noting routine agenda convergence between political elites and the press (Hayes, 2010; Tedesco, 2001). Moreover, this study suggests that indexing routines embedded in the gatekeeping process render elites far more equipped to frame debates than the public that these campaign events were designed to serve. Nonetheless, not all news professionals enable elite framing of debates with the same level of emphasis. For example, local journalists with greater proximity to their news audiences seem more attentive to the public. Local moderators and local journalist question sources reduce the likelihood of cases citing political elites; local moderators are the greatest predictor of questions citing the public. It is also clear that variables holding a positive relationship with questions citing the public often hold an inverse relationship with questions citing elites. Simply put, debate agendas—such as news agendas—are finite. Relying on one exogenous source tends to crowd out another.

For example, local press variables perform as predictors of public cites while also diminishing the frequency of questions citing political elites. Simply put, local moderators are less likely to provide a megaphone to political elites. This is an important finding given that scholarship has found particularly strong indexing routines for national news media, which results in an over-reliance on official sources and, at times, an allegiance to official sources (Bennett, 1990; Zaller & Chiu, 1996). This study suggests that such an allegiance to elite sources may be mitigated by proximity to news audiences. Public debate participants also hold a positive relationship with questions citing the public while their participation diminishes the likelihood of cases citing political elites. Thus, the public seems less interested in echoing the talking points of political insiders.

Peculiarly, nonprofit press sources nearly never cite the public but hold a predicted probability of nearly 70% in citing political elites—far exceeding all other predictors of political elite source selection. One explanation for this finding, given the results found for local news professionals, is that the nonprofit press corps hold closer proximity to the Washington establishment and are further removed from their national news audiences. In other words, proximity to news audiences may shape the gatekeeping process of source selection more than commercial influences.

One limitation worth noting is that a question citing the public does not necessarily signal a question that addresses an issue of high priority for the public, and questions citing political elites may simultaneously address an issue of public concern. Results must also be tempered by the fact that other political actors can influence debate content—including special interests or the business community (see Berkowitz & Adams, 1990; Manheim & Albritton, 1984). Nonetheless, this study provides a preliminary look at the exogenous influences of electoral debates and compares relative influence between two key stakeholders: the public and political elites. Given that debates were designed to be responsive to public information needs (Jamieson & Birdsell, 1988), these campaign events should minimize elite indexing. This study demonstrates that when exogenous sources shape debates, those sources are far more likely to reflect political elites over members of the public. If these forums are truly representing the public, we should see more diversity in source selection from the gatekeeping process than what is shown herein.

Footnotes

Appendix

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biographies

Jason Turcotte (PhD, Louisiana State University) is an assistant professor in Communication at Cal Poly Pomona. His research interests include political communication, news, and public opinion.

References

Althaus

S. L.

(2003). When news norms collide, follow the lead: New evidence for press independence. Political Communication, 20, 381-414.

Arnold

R. D.

(2004). Congress, the press, and political accountability. Princeton, NJ: Russell Sage.

Bennett

W. L.

(1990). Toward a theory of press-state relations. Journal of Communication, 40, 103-125.

Bennett

W. L.

Lawrence

R. G.

Livingston

(2006). None dare call it torture: Indexing and the limits of press independence in the Abu Ghraib scandal. Journal of Communication, 56, 467-485.

Bennett

W. L.

Lawrence

R. G.

Livingston

(2007). When the press fails: Political power and the news media from Iraq to Katrina. Chicago, IL: The University of Chicago Press.

Benoit

W. L.

Hansen

G. J.

(2001). Presidential debate questions and the public agenda. Communication Quarterly, 49, 130-141.

Berkowitz

(1987). TV news sources and news channels: A study in agenda building. Journalism Quarterly, 64, 508-513.

Berkowitz

Adams

D. B.

(1990). Information subsidy and agenda-building in local television news. Journalism & Mass Communication Quarterly, 67, 723-731.

Carlin

D. B.

Vigil

Buehler

McDonald

(2009). The third agenda in U.S. presidential debates: Debate watch and viewer reactions, 1996-2004. Westport, CT: Praeger.

10.

Cohen

J. E.

(2008). The presidency in the era of 24-hour news. Princeton, NJ: Princeton University Press.

11.

Cook

T. E.

(1998). Governing with the news: News media as a political institution. Chicago, IL: Chicago University Press.

12.

Cook

T. E.

(2006). The news media as a political institution: Looking backward and looking forward. Political Communication, 23, 159-171.

13.

Delli Carpini

M. X.

Williams

B. A.

(2001). Let us infotain you: Politics in the new media environment. In Bennett

W. L.

Entman

R. M.

(Eds.), Mediated politics, Communication in the future of democracy (pp. 160-181). New York, NY: Cambridge University Press.

14.

Dimitrova

D. V.

Kostadinova

(2013). Identifying antecedents of the strategic game frame: A longitudinal analysis. Journalism & Mass Communication Quarterly, 90, 75-88.

15.

Dunaway

(2008). Markets, ownership, and the quality of campaign news coverage. The Journal of Politics, 70, 1193-1202.

16.

Dunaway

Stein

R. M.

(2013). Campaign news coverage and early voting. Political Communication, 30, 278-296.

17.

Dylko

I. B.

Beam

M. A.

Landreville

K. D.

Geidner

(2011). Filtering 2008 US presidential election news on YouTube by elites and nonelites: An examination of the democratizing potential of the internet. New Media & Society, 14, 832-849.

18.

Entman

(1993). Framing: Toward a clarification of a fractured paradigm. Journal of Communication, 43, 51-58.

19.

Entman

(2004). Projections of power: Framing news, public opinion, and U.S. foreign policy. Chicago, IL: University of Chicago Press.

20.

Entman

Matthes

Pellicano

(2009). Nature, sources, and effects of news framing. In Wahl-Jorgensen

Hanitzsch

(Eds.), The handbook of journalism (pp. 175-190). New York, NY: Routledge.

21.

Fallows

(1996). Breaking the news: How the media undermine American democracy. New York, NY: Vintage.

22.

Farnsworth

S. J.

Lichter

S. R.

(2011). The nightly news nightmare: Media coverage of U.S. presidential elections, 1988-2008. Lanham, MD: Rowman & Littlefield.

23.

Gilbert

Eyal

McCombs

Nicholas

(1980). The state of the union address and the press agenda. Journalism Quarterly, 57, 584-588.

24.

Goffman

(1974). Frame analysis: An essay on the organization of experience. Boston, MA: Northeastern University Press.

25.

Hamilton

J. T.

(2004). All the news that’s fit to sell: How the market transforms information into news. Princeton, NJ: Princeton University Press.

26.

Hayes

(2010). The dynamics of agenda convergence and the paradox of competitiveness in presidential campaigns. Political Research Quarterly, 63, 594-611.

27.

Iyengar

(1991). Is anyone responsible? How television news frames political issues. Chicago, IL: University of Chicago Press.

28.

Jackson-Beeck

Meadow

R. G.

(1979). The triple agenda of presidential debates. Public Opinion Quarterly, 43, 173-180.

29.

Jamieson

K. H.

Birdsell

D. S.

(1988). Presidential debates: The challenge of creating an informed electorate. New York, NY: Oxford University Press.

30.

Just

M. R.

Crigler

A. N.

Alger

D. E.

Cook

T. E.

Kern

West

D. M.

(1996). Crosstalk: Citizens, candidates, and the media in a presidential campaign. Chicago, IL: University of Chicago Press.

31.

Kaid

L. L.

McKinney

M. S.

Tedesco

J. C.

(2000). Civic dialogue in the 1996 presidential campaign: Candidates, media, and public voices. Cresskill, NJ: Hampton Press.

32.

Kiousis

Kim

McDevitt

Ostrowski

(2009). Competing for attention: Information subsidy influence in agenda building during election campaigns. Journalism & Mass Communication Quarterly, 86, 545-562.

33.

Manheim

J. B.

Albritton

R. B.

(1984). Changing national images: International public relations and media agenda setting. American Political Science Review, 78, 641-657.

34.

McKinney

M. S.

(2005). Let the people speak: The public’s agenda and presidential town hall debates. American Behavioral Scientist, 49, 198-212.

35.

Page

B. I.

(1996). Mass media as political actors. PS: Political Science & Politics, 29, 20-24.

36.

Pan

Kosicki

G. M.

(1993). Framing analysis: An approach to news discourse. Political Communication, 10, 55-75.

37.

Patterson

T. E.

(1994). Out of order. New York, NY: Vintage.

38.

Ridout

T. N.

Mellen

(2007). Does the media agenda reflect the candidates’ agenda? International Journal of Press/Politics, 12, 44-62.

39.

Ross

(2007). The journalist, the housewife, the citizen, and the press: Women and men as sources in local news narratives. Journalism, 8, 449-473.

40.

Ryfe

D. M.

(2006). The nature of news rules. Political Communication, 23, 203-214.

41.

Salwen

M. B.

(1995). News of Hurricane Andrew: The agenda of sources and the sources’ agendas. Journalism & Mass Communication Quarterly, 72, 826-840.

42.

Scheufele

D. A.

(1999). Framing as a theory of media effects. Journal of Communication, 49, 103-122.

43.

Schroeder

(2000). Media sponsorship of the 2000 primary debates. International Journal of Press/Politics, 5, 104-111.

44.

Shoemaker

P. J.

Eichholz

Kim

Wrigley

(2001). Individual and routines forces in gatekeeping. Journalism & Mass Communication Quarterly, 78, 233-246.

45.

Shoemaker

P. J.

Reese

S. D.

(1996). Mediating the message: Theories of influences on mass media content. White Plains, NY: Longman.

46.

Shoemaker

P. J.

Vos

T. P.

Reese

S. D.

(2009). Journalists as gatekeepers. In Wahl-Jorgensen

Hanitzsch

(Eds.), The handbook of journalism (pp. 73-87). New York, NY: Routledge.

47.

Stromback

van Aelst

(2010). Exploring some antecedents of the media framing of election news: A comparison of Swedish and Belgian election news. International Journal of Press/Politics, 15, 41-59.

48.

Swanson

L. L.

Swanson

D. L.

(1978). The agenda-setting function of the first Ford-Carter debate. Communication Monographs, 45, 347-353.

49.

Tedesco

J. C.

(2001). Issue and strategy agenda-setting in the 2000 presidential primaries. American Behavioral Scientist, 44, 2048-2067.

50.

Walgrave

Van Aelst

(2006). The contingency of the mass media’s political agenda setting power: Toward a preliminary theory. Journal of Communication, 56(1), 88-109.

51.

Wanta

(1991). Presidential approval ratings as a variable in the agenda-building process. Journalism & Mass Communication Quarterly, 68, 672-679.

52.

Weaver

Elliott

S. N.

(1985). Who sets the agenda for the media? A study of local agenda-building. Journalism Quarterly, 62, 87-94.

53.

White

D. M.

(1950). The gatekeeper: A case study in the selection of news. Journalism Quarterly, 27, 383-391.

54.

Zaller

J. R.

Chiu

(1996). Government’s little helper: U.S. press coverage of foreign policy crises, 1945-1991. Political Communication, 13, 385-405.