2009_dunn_candidate and media agenda setting

Journal of Communication ISSN 0021-9916

ORIGINAL ARTICLE

Candidate and Media Agenda Setting in the2005 Virginia Gubernatorial Election

Scott W. Dunn

School of Communication, Radford University, Radford, VA 24142

This study examines relationships among candidate and media agendas during the 2005Virginia gubernatorial election. Candidate press releases and newspaper articles aboutthe election were analyzed using the computer program VBPro. High agenda correlationsindicated that the 2 major-party candidates, Democrat Tim Kaine and Republican JerryKilgore, and the 4 major newspapers serving the state discussed similar issues during thecampaign. Cross-lagged correlations indicated both candidates’ agendas shared reciprocalrelationships with the agendas of 2 papers, whereas the candidates set the agendas of2 others. Intermedia agenda-setting effects were also observed. The candidates’ agendasinfluenced each other during the campaign’s hot phase, although post hoc analysis indicatedthat Kilgore earlier set Kaine’s agenda. Implications for future agenda-setting research arediscussed.

doi:10.1111/j.1460-2466.2009.01442.x

For more than 30 years, agenda-setting theory has been one of the dominanttheoretical approaches used to study mass media effects, particularly in politicalcontexts. Over the last 3 decades, agenda setting has evolved from a hypothesis aboutmedia effects on the public to a research paradigm that examines the transfer ofsalience at all levels of political discourse. This study builds on previous researchto examine agenda setting in the 2005 Virginia gubernatorial election, specificallyexamining the relationship between candidate agendas and media agendas.

Corresponding author: Scott W. Dunn; e-mail: [email protected] paper is based on research conducted for the author’s master’s thesis in the Department ofCommunication at Virginia Tech. The author thanks his thesis advisor, Dr. John C. Tedesco,and his thesis committee members, Dr. Robert E. Denton, Jr. and Dr. Andrew Paul Williams,for their invaluable guidance on this project. Portions of this paper were presented at the2006 meeting of the Eastern Communication Association in Philadelphia, PA and at the 2006meeting of the National Communication Association in San Antonio, TX.

Journal of Communication 59 (2009) 635–652 c© 2009 International Communication Association 635

Candidate and Media Agenda Setting S. W. Dunn

Election context

In 2005, the Commonwealth of Virginia’s gubernatorial election earned much mediaattention as the more hotly contested of only two gubernatorial elections occurringin the United States (see Dao, 2005; Nguyen, Rowlands, & Starr, 2005; Siegel,2005). Observers of American politics considered Virginia’s gubernatorial election,a microcosm of the national political landscape, and believed its outcome couldforeshadow election trends in 2006 and 2008.

The major candidates in the 2005 Virginia gubernatorial race were Democrat TimKaine, Republican Jerry Kilgore, and independent Russ Potts. Kaine, who ultimatelywon the race, was previously the state’s Lieutenant Governor and a former Richmondmayor (The Almanac of Virginia Politics, 2002). Kilgore is a former Virginia AttorneyGeneral who resigned the office in order to concentrate on the gubernatorial race(Shear, 2005). Potts is a Republican state senator who ran as an independent (Laris,2005). The major-party candidates offered a significant political contrast. Kaine ranas the natural successor to popular Democratic incumbent Governor Mark Warner.As Lieutenant Governor, Kaine helped Warner enact budget reforms that revivedthe state’s struggling economy and got Warner named one of ‘‘America’s Five BestGovernors’’ by Time magazine (Tumulty, 2005). Kilgore countered by arguing thatKaine was too liberal on economic and social issues and dismissed Kaine’s role inWarner’s successful administration, as well as associating Kaine with liberal positionson issues such as gun control and the death penalty in an attempt to lessen his supportamong socially conservative voters.

Carsey (2000) pointed out that state-level elections have received significantlyless scholarly attention than presidential elections. He called for more research ongubernatorial campaigns for several reasons: the increased responsibilities placed onstate governments since the 1980s, the need for a larger sample of campaigns in whichto test theories, and the unique characteristics of various gubernatorial campaignsthat allow researchers to study a wide range of theories. A few studies have used stategovernment contexts to study media agenda setting (Atwater, Fico, & Pizante, 1987;Roberts & McCombs, 1994; Tipton, Haney, & Baseheart, 1975). This study furthersthis research program by systematically studying agenda setting at both the mediaand the candidate levels.

This race offered an ideal context for observing agenda-setting effects for severalreasons. First, the mainstream media in Virginia are made up of a manageable numberof outlets (see Rozell, 1991; Wilson, 1991). There are enough media sources to studyvariety and fluctuation in the media agenda, but not so many as to make it difficult toget a sufficiently complete picture of that agenda. Second, the recent electoral historyof the commonwealth indicates that, while Republican candidates usually win thestate in presidential elections, in statewide elections Virginia is a ‘‘two-party state’’(Morris & Sabato, 1998, p. 40), in which the Republican and Democratic candidatesboth have a legitimate chance of victory. Therefore, the issues that became salient tothe public may have had a significant impact on the election results, in contrast to

636 Journal of Communication 59 (2009) 635–652 c© 2009 International Communication Association

S. W. Dunn Candidate and Media Agenda Setting

races in which voting decisions are primarily based on predetermined ideological orparty loyalties. Finally, there were no national elections on the ballot along with thecandidates for governor, so it is possible to study the race without needing to accountfor ‘‘coattail’’ effects provided by national candidates (see Vermeer, 1982) or havingto distinguish between coverage of national and state races.

Literature review

In its most basic form, the concept of agenda setting is defined as the process throughwhich certain issues become more relevant than others, as originally demonstratedby McCombs and Shaw (1972). Subsequent research mostly strengthened the casefor effects of the media agenda on the public agenda (see especially Funkhouser,1973; Iyengar, Peters, & Kinder, 1982; Weaver, Graber, McCombs, & Eyal, 1981). Thebasic agenda-setting hypothesis treated the public agenda as a dependent variableinfluenced by the independent variable of media agendas. However, subsequentstudies have treated media or policy agendas as dependent variables, affected by anumber of different independent variables (Dearing & Rogers, 1996). A few studies(Tedesco 2001, 2005a, 2005b) have also examined candidate agendas. The presentstudy builds on those agenda-setting studies that treat media and candidate agendasas both dependent and independent variables.

Agenda-setting researchers have often assumed a uniform media agenda, perhapsbecause those studies that have compared agendas among various media, includingthe original Chapel Hill study, have tended to find high intermedia correlations.However, as early as 1976, McCombs and Shaw encouraged researchers to examineinfluences on media agendas. McCombs (2004) identified three layers of influenceson media agendas: news sources, other news media, and news norms. The presentstudy focuses on the first two: news sources (specifically, information subsidiesprovided by campaigns) and other media.

Candidate press releases as agenda settersGandy (1982) showed that journalistic sources use ‘‘information subsidies,’’ orattempts to make favorable information ‘‘available at something less than the costa user would face in the absence of the subsidy’’ (p. 61). Research has shown thatinformation subsidies can effectively help set the media agenda (e.g., Sigal, 1973;Turk, 1986) even though journalistic norms dictate that news editors avoid relyingtoo much on information subsidies for content (Curtin, 1999; Vermeer, 1982).Turk (1986) found that about half of the press releases issued by Louisiana stateagencies were used by newspapers in the state, and that about half of the newspaperarticles about those agencies used their press releases. More importantly, she found asignificant level of correlation between the issue agendas of the information subsidiesand the newspapers.

Particularly of interest to the present study is research on the influence of pressreleases provided by political campaigns. Bolden (1970) and Kaid (1976) did early



studies of this type, analyzing the success rates of press release placement for a Texasgubernatorial candidate and an Illinois state senate candidate, respectively. Bothresearchers found that the candidates got the content of more than half of theirpress releases printed in local newspapers, with a substantial number of statementscopied from the releases verbatim. These two early studies offered initial evidencethat campaigns can influence media coverage with information subsidies, but theyalso established that this influence is far from complete or universal.

Vermeer (1982) analyzed how often candidate press releases were usedby newspapers covering four different candidates (two gubernatorial and twoCongressional). Between 70 and 80% of the candidates’ press releases were pickedup by newspapers. Vermeer observed that the images of the candidates seen inthe totality of news stories about the campaign roughly reflected the images thecandidates conveyed in their press releases. The newspapers’ rewrites substantiallychanged the content of many individual press releases, but over the course of thecampaign the positive and negative traits each candidate conveyed of himself, hisopponent, and the campaign’s major issues generally survived the treatment of mediagatekeepers. Thus, Vermeer’s study offered further support for the utility of candidatepress releases for disseminating campaign agendas to voters.

One limitation of these early studies was their focus on the number of pressreleases that could clearly be identified as sources for specific news stories. As Ver-meer (1982) observed, press releases that do not explicitly inspire a newspaper storymay be used for background information. It is logical to assume that the influenceof campaign press releases goes beyond those cases in which their content is directlyreprinted by newspapers. Atwood (1980) broadened the study of candidate infor-mation subsidies by using agenda-setting theory to determine the overall influenceof candidate press releases on news agendas. His analysis of candidate press releasesand newspaper coverage of a Congressional race showed strong correlations betweencandidate and newspaper agendas. Cross-lagged correlations indicated that the can-didates and newspapers exerted reciprocal influence on each other. Similar reciprocalrelationships were found between candidate and media agendas in 1994 Texas guber-natorial race (Evatt & Bell, 2001), the 2000 U.S. presidential primaries (Tedesco,2001), and the 2004 U.S. presidential election (Tedesco, 2005b). However, Ku, Kaid,and Pfau (2003) found that news releases posted on candidate websites exerted anagenda-setting influence on newspapers and television news during portions of the2000 general election for U.S. president.

Overall, research on information subsidies provided by political campaigns tomedia outlets confirms that this strategy is successful, although the extent of thatsuccess varies across different election contexts. Generally, past research indicatesthat a fairly high number (at least half) of candidate information subsides will directlyinspire stories in the news media, and that there tends to be a high correlationbetween candidate and media agendas. Studies such as the present one are necessaryto explain why certain campaigns are more successful than others at influencingmedia agendas.



Intermedia agenda settingSeveral studies have demonstrated agenda-setting effects among various mediaoutlets, a process known as intermedia agenda setting. Of most relevance to thisproject, early studies of agenda setting found high levels of agreement among theagendas used by various newspapers (but not necessarily other media) in coverageof national (McCombs & Shaw, 1972) and statewide (Tipton, Haney, & Baseheart,1975) elections. Although these early studies did not address the direction of influenceamong these publications, more recent intermedia agenda-setting studies indicatethat certain media outlets, especially The New York Times at the national level, haveparticularly strong agenda-setting power (Danielian & Reese, 1989; Mazur, 1987;Reese & Danielian, 1989). The present study adds to this research by examiningagenda-setting effects among local newspapers during a statewide race.

Intercandidate agenda settingIn addition to studying how campaigns shape the media agenda, this study examinesthe effects that campaigns have on each other through information subsidies.Tedesco (2005b) examined influences between the major-party campaigns in the2004 presidential race and found that the two campaigns mostly exerted reciprocalinfluence on each other (see also Kiousis & Shields, 2006). However, there was aclear agenda-setting effect between May and June, when John Kerry’s agenda seemedto influence George W. Bush’s agenda. Although this intercandidate agenda-settingeffect may not have made a significant difference in the results of the election, it islikely that a similar effect closer to Election Day could have shifted the outcome.

Tedesco (2005a) found stronger evidence of intercandidate agenda setting in the2004 Democratic presidential primary. Throughout the race, the major candidatesinfluenced each others’ agendas as they worked to find the right issues to connectwith Democratic voters. Most notably, candidates Wesley Clark, Howard Dean, andJohn Edwards displayed agenda-setting effects on the campaign of eventual nomineeJohn Kerry between November and December of 2003. This followed an unexplainedshift in Kerry’s agenda during the month of November. This is important becausein late 2003, Kerry was not considered a frontrunner for the nomination. However,after his win in the Iowa caucuses in January 2004, Kerry quickly sealed his primaryvictory. Kerry’s shift back to an issue agenda consistent with those of the rest of thecandidates during the month before the Iowa caucuses may be one explanation forhis sudden reversal of fortune.

The study of intercandidate agenda setting is still a new research program, but thetwo studies cited above indicate that it is a fertile area of inquiry for future elections.Considering the evidence that campaign agendas influence media agendas, which inturn influence public agendas, it is worthwhile to investigate how candidate agendasare formed.

Research questions

Past research has shown that candidates and newspapers tend to influence eachothers’ issue agendas during the course of a campaign, but the exact path of that



influence has not been consistent across studies. This study attempts to identifythe path of that influence in the 2005 Virginia gubernatorial election by asking thefollowing research questions:

RQ1: Did the candidates’ press releases influence the agendas of Virginia newspapers?

RQ2: Did any Virginia newspaper exert an influence on the agenda of any other Virginianewspaper?

RQ3: Did either candidate’s press releases influence the other candidate’s agenda?

Methods

To assess the candidate agendas, the author collected every press release issuedby Kaine and Kilgore’s campaigns from January 1, 2005 through Election Day,November 8, 2005. These press releases were downloaded from the candidate’s Websites. Previous researchers (Miller, Andsager, & Reichert, 1998; Tedesco, 2001, 2005a,2005b) have used press releases posted on candidate websites as representations ofcandidate agendas.

To assess the media agenda, the author collected all articles mentioning anyof the three candidates in The Washington Post, Richmond Times-Dispatch, TheVirginian-Pilot (Norfolk), and The Roanoke Times during the months of Septemberand October, the hot phase of the campaign. These are the newspapers in Virginiathat are archived in the LexisNexis database, plus the Post, located in Washington,DC, which frequently covers Virginia politics. Past research on media coverage ofVirginia elections has limited its scope to one (Carsey, 2000; Wilson, 1991) or two(Rozell, 1991) newspapers, apparently under the assumption that The WashingtonPost and the Richmond Times-Dispatch exemplify the press for the entire state. Theaddition of two other newspapers in different parts of the state allowed for the testingof this assumption.

The researcher assessed candidate and media agendas during the months ofSeptember and October1 via the VBPro content analysis software. The author createda list of 31 issues and search terms related to each issue based on past agenda-settingstudies and on preliminary observation of the gubernatorial campaign’s discourse.The VBPro software was set up to calculate the number of times the references toeach issue, strategy, or audience appeared in the press releases and news articles.

Data were analyzed using the cross-lagged correlation method developed byRozelle and Campbell (1969) and adapted for agenda-setting research by a varietyof scholars (e.g., Atwood, 1980; Evatt & Bell, 2001; Roberts & McCombs, 1994;Tedesco, 2001, 2005a, 2005b). In this method, Pearson product moment correlationcoefficients between two variables measured concurrently and within each variableacross time are used to calculate a baseline coefficient2 that serves as a pointof comparison for correlations between the variables across time. If both of the‘‘diagonal’’ correlation coefficients (the correlation between Variable One at Time



One and Variable Two at Time Two and the correlation between Variable Two atTime One and Variable One at Time Two) are above the baseline, the variablesare said to influence each other reciprocally. If, however, one diagonal correlationis above the baseline while the other is below it, then an agenda-setting effect hasoccurred. This indication is especially strong if the within-variable correlation forthe influenced agenda is below the baseline, because that condition indicates that theagenda did not just shift incrementally in order to conform with another agenda, butactually shifted noticeably during the time lag under study.

The cross-lagged correlation approach allows researchers to establish the level ofcorrelation that exists between two agendas and the temporal order in which thoseagendas formed. In addition, since the baseline coefficient takes into account changesin each variable over time, this method offers some control for outside influences thataffect both variables similarly. Although it does not entirely control for all outsideinfluences, the cross-lagged correlation technique does serve as a useful alternative inresearch contexts (such as most agenda-setting studies) in which outside influencescannot be measured or controlled.

Results

Influence between candidates and newspapersRQ1 asked whether the agendas conveyed by the candidate’s press releases influencedthe agendas of newspapers covering the gubernatorial race. Figure 1 shows the cross-lagged correlations between the months of September and October for Kaine’s issueagenda and the agendas of the four newspapers.3 All of the correlations reportedhere are robust, ranging from .71 to .99, indicating that Kaine’s press releases andthe newspapers tended to strongly agree on the issues important in the election. Thecross-lagged correlation between Kaine’s agenda and that of the Richmond Times-Dispatch indicates that the two agendas shared a reciprocal relationship, since boththe cross lags (r = .99 and .97) are above the baseline value of .95. The same is trueof Kaine and The Washington Post. Kaine’s September agenda has a .99 correlationwith the newspaper’s October agenda, which is well above the baseline of .90. Thenewspaper’s September agenda has a .90 correlation with Kaine’s October agenda.Although this coefficient is noticeably lower than .99, it is equal to the baseline valueof .90, indicating that the influence is reciprocal.

In contrast to the Richmond and Washington newspapers, Kaine’s press releasesappear to have influenced the agendas of The Roanoke Times and The Virginian-Pilot. Kaine’s September agenda and The Roanoke Times’s October agenda have a.99 correlation, well above the baseline of .84, whereas the reverse correlation fallsbelow the baseline at .83. Similarly, the September Kaine agenda and the OctoberVirginian-Pilot agenda have a .97 correlation, well above the .77 baseline, whereasthe reverse correlation is .71. Notably, both newspapers exhibited high correlationsbetween their September and October agendas: .86 for The Roanoke Times and .84 forThe Virginian-Pilot, both above their respective baselines. These coefficients indicate



Kaine

RichmondTimes-Dispatch

.99

.95

.97

.99.97 .99

September October

Baseline = .95


Kaine Kaine

Roanoke Times

.99

.86

.83

.99.83 .99

September October

Roanoke Times

Kaine

Baseline = .84

Kaine

Virginian-Pilot

.99

.84

.71

.97.70 .98

September October

Baseline = .77

Virginian-Pilot

Kaine Kaine

WashingtonPost

.99

.90

.90

.99.90 .99

September October

Kaine

Baseline = .90

WashingtonPost

Figure 1 Cross-lagged correlations between Tim Kaine’s agenda and newspaper agendas.

that Kaine influenced these two newspapers during the hot phase of the campaign,but not to the extent that the newspapers demonstrated drastic agenda shifts.

Figure 2 shows the cross-lagged correlations between the agendas of Kilgore’spress releases and the four newspapers. As with Kaine, Kilgore’s campaign hada reciprocal relationship with the Richmond Times-Dispatch. Kilgore’s Septemberagenda and the newspaper’s October agenda have a .99 correlation, well abovethe baseline of .94, whereas the opposite cross-lagged correlation is .94, equal tothe baseline. The same pattern held for The Washington Post. Kilgore’s Septemberagenda and the newspaper’s October agenda have a correlation of .99 and the reversecorrelation is .87, both above or equal to the baseline of .87. However, the other twonewspapers were significantly influenced by Kilgore’s agenda. Kilgore’s Septemberagenda correlates very highly with the October newspaper agendas: .99 for TheRoanoke Times and .96 with The Virginian-Pilot. However, the correlations betweenthe September newspaper agendas and Kilgore’s October agenda are much lower:.78 for The Roanoke Times and .65 for The Virginian-Pilot. Both of these correlationsare below their baselines of .83 and .75, respectively. Again, the newspapers maintaincorrelations above their respective baselines across the 2 months, so the agenda shiftswere not drastic.

Intermedia agenda setting among virginia newspapersRQ2 asked if any of the four newspapers included in the analysis influenced any othernewspaper’s agenda. Figure 3 shows the cross-lagged correlations between each of



Kilgore


.99

.95

.94

.99.95 .99

September October

Baseline = .94


Kilgore Kilgore

Roanoke Times

.99

.86

.78

.99.80 .99

September October

Roanoke Times

Kilgore

Baseline = .83

Kilgore

Virginian-Pilot

.99

.84

.65

.96.68 .96

September October

Baseline = .75

Virginian-Pilot

Kilgore Kilgore

WashingtonPost

.99

.90

.87

.99.88 .87

September October

Kilgore

Baseline = .83

WashingtonPost

Figure 2 Cross-lagged correlations between Jerry Kilgore’s agenda and newspaper agendas.

the four newspapers. Every comparison shown indicates the presence of intermediainfluence. During the period under investigation, the Richmond Times-Dispatchinfluenced all three of the other newspapers. The correlation between the SeptemberTimes-Dispatch agenda and the October Virginian-Pilot agenda is .99, well above thebaseline of .82, whereas the reverse correlation is .67, well below the baseline. Thecorrelation between The Virginian-Pilot’s September and October agendas is above thebaseline (r = .84), indicating that the shift in the newspaper’s agenda was not drastic.

The correlation of the Richmond Times-Dispatch’s September agenda with TheRoanoke Times’ October agenda (r = .98) and The Washington Post’s October agenda(r = .97) are also above their respective baselines (.87 and .92) while the reversecorrelations (r = .80 and .88) are not. In addition, these two newspapers’ agendasduring the 2 months have correlation coefficients below their baselines (r = .86for The Roanoke Times; r = .90 for The Washington Post). Therefore, the RichmondTimes-Dispatch exerted an agenda-setting effect on these two newspapers that resultedin significant shifts in their agendas.

The Washington Post influenced the agendas of both The Roanoke Times and TheVirginian-Pilot. The September agenda of The Washington Post has a correlation withthe October agenda of The Roanoke Times (r = .93) that is above the baseline of.87, whereas the reverse correlation is below the baseline (r = .82). The Septemberand October agendas of The Roanoke Times correlate with each other at .86, belowthe baseline, indicating a significant agenda-setting effect. Similarly, The WashingtonPost’s September agenda has a .97 correlation with The Virginian-Pilot’s October




Roanoke Times

.95

.86

.80

.98.93 .99

September October

Baseline = .87

Roanoke Times

RichmondTimes-Dispatch Roanoke Times

Virginian-Pilot

.86

.84

.76

.92.96 .99

September October

Virginian-Pilot

Roanoke Times

Baseline = .83


Virginian-Pilot

.95

.84

.67

.99.86 .97

September October

Baseline = .82

Virginian-Pilot

RichmondTimes-Dispatch Roanoke Times

WashingtonPost

.86

.90

.93

.82.98 .99

September October

Roanoke Times

Baseline = .87

WashingtonPost


WashingtonPost

.95

.90

.88

.97.98 .99

September October

Baseline = .92

WashingtonPost

RichmondTimes-Dispatch Virginian-Pilot

WashingtonPost

.84

.90

.97

.71.94 .98

September October

Virginian-Pilot

Baseline = .84

WashingtonPost

Figure 3 Cross-lagged correlations between the agendas of the four newspapers.

agenda, well above the baseline of .84, whereas the reverse correlation is below thebaseline at .71. However, in this case, The Virginian-Pilot’s agendas of each monthare correlated at .84, equal to the baseline.

Finally, The Roanoke Times influenced the agenda of The Virginian-Pilot duringthe time lag studied. The Roanoke Time’s September agenda has a correlation of.92 with The Virginian-Pilot’s October agenda, which was above the baseline of .83,whereas the reverse correlation is below the baseline at .76. The correlation betweenThe Virginian-Pilot’s agendas for the 2 months is above the baseline at .84, indicatingthat the Norfolk newspaper’s agenda shift was not drastic.



Kilgore

.78

.84

.86

.77.98 .99

March April

Baseline = .80

Kilgore

Kaine Kaine

Figure 4 Cross-lagged correlations between Kaine and Kilgore’s agendas.

Overall, a clear hierarchy on influence emerged. The Richmond Times-Dispatchinfluenced the three other newspapers, The Washington Post influenced the tworemaining papers, and The Roanoke Times influenced The Virginian-Pilot.

Intercandidate influenceRQ3 asked if either candidate’s agenda influenced the other’s agenda during the courseof the campaign. Between the months used for comparisons with newspaper agendas(September and October), both candidates maintained agendas that are nearlyperfectly correlated (r = .99) with their own agenda and with their opponents agenda(within months and across months). On the basis of these data, it appears that the twocandidates had established consistent issue agendas by the beginning of Septemberand neither candidate deviated from them in the final months of the campaign.

Because there was clearly no intercandidate agenda setting occurring during thehot phase of the campaign, candidate press releases from the rest of the year wereanalyzed. During all 2-month periods from January through October, the influencebetween the two candidate’s agendas was reciprocal with the exception of the Marchto April time lag. As seen in Figure 4, Kilgore’s March agenda and Kaine’s Aprilagenda have a correlation of .86, above the baseline of .80. Kaine’s March agenda andKilgore’s April agenda have a correlation of .77, below the baseline. Moreover, Kaine’sMarch and April agendas have a correlation coefficient that is below the baseline(r = .78), indicating that Kilgore’s agenda-setting influence caused a significant shiftin Kaine’s agenda. Thus, for this single time period, Kilgore set Kaine’s agenda.

Discussion

This study examined the agenda-setting effects among candidates and newspapersduring the 2005 Virginia gubernatorial campaign. Findings based on one electioncannot necessarily be generalized to other campaigns. However, the findings of thisstudy contribute to scholarly understanding of agenda setting by lending furthersupport to some previous findings and raising questions about others.

One overall trend that is evident in the data is that the two candidates and fournewspapers maintained highly consistent issue agendas with each other and with



their own agendas. This observation supports the findings of several previous studies(Tedesco, 2001, 2005a, 2005b). The correlations exhibited here are not quite asstrong as those observed by Tedesco (2005b) in the 2004 presidential race, but theyare sufficiently strong, especially in subsequent months, to support the assumptionthat candidates will generally exhibit consistent agendas throughout the course of acampaign. The candidates appeared to run their campaigns consistently with Sheaand Burton’s (2001) advice to pick campaign themes early and focus on themconsistently throughout the race and the newspapers followed the candidates’ lead inmaintaining consistent agendas.

Interestingly, the data indicate that the Virginia candidates and newspapersmaintained more consistent agendas than in the Texas campaigns studied by Robertsand McCombs (1994) and Evatt and Bell (2001). Evatt and Bell (2001) found thatthe candidates maintained a combined correlation of only .53 and the newspapershad a correlation of only .67 between the two halves of their campaign’s hot phase.Even more remarkable, Roberts and McCombs (1994) found a correlation of only.33 between the newspaper agendas in the two halves of the hot phase. The Virginiacandidates and newspapers, by contrast, exhibited very high issue agenda correlationsthroughout the campaign’s hot phase.

This study indicated that both candidate agendas shared reciprocal relationshipswith the agendas of the Richmond Times-Dispatch and The Washington Post whileexerting unidirectional influence on The Roanoke Times and The Virginian-Pilot.This finding differs somewhat from most previous studies, which have tended tofind reciprocal relationships between candidate and media agendas (Atwood, 1980;Evatt & Bell, 2001; Roberts & McCombs, 1994; Tedesco, 2001). Only Ku et al. (2003)and Tedesco (2005b) found evidence of candidate information subsidies influencingmedia agendas.

The candidate influence on media agendas found in this study is not necessarilyinconsistent with the findings of previous studies when the specific media analyzedare examined more closely. The Richmond Times-Dispatch and The Washington Post,the two newspapers that shared reciprocal influence with the candidates, more closelyresemble the newspapers used in previous studies than The Roanoke Times or TheVirginian-Pilot do. The Richmond Times-Dispatch is the major daily newspaper in thestate capitol, comparable to Texas’s Austin American-Statesman, used by Roberts andMcCombs (1994), and the Austin, Dallas, and Houston newspapers used by Evattand Bell (2001). The Washington Post was among the nationally circulated mediaexamined by Ku et al. (2003) and Tedesco (2005b) and comparable to the nationaltelevision network newscasts used by Tedesco (2001). The Roanoke and Norfolkpapers, on the other hand, are not located in state or federal government centers, andtherefore cover state politics from a greater distance.

This study indicates that media outlets that maintain a distance from statepolitics are more influenced by candidate agendas than those located in centers ofgovernment. Apparently, distance from the government does not necessarily translateinto journalistic independence from the candidates’ agendas. Rather, it appears that



newspapers like the Richmond Times-Dispatch and The Washington Post develop amore independent agenda, perhaps because they have greater access to alternativegovernment sources that present different agendas from those presented by thecandidates. This is an intriguing finding that warrants greater attention from futureresearchers.

The finding that several large newspapers influenced the issue agendas of smallernewspapers is consistent with findings from past intermedia agenda-setting studies,which have consistently shown the ability of the influential The New York Times toinfluence other newspapers’ agendas (Danielian & Reese, 1989; Mazur, 1987; Reese& Danielian, 1989). A period of 2 months is far from sufficient for demonstratinggeneralizability, but this study does offer preliminary evidence that the RichmondTimes-Dispatch tends to influence other Virginia newspapers’ issue agendas, followedby The Washington Post, The Roanoke Times, and The Virginian-Pilot. As a preliminaryfinding, the path of issue agenda influence from the Richmond Times-Dispatch toThe Washington Post to The Roanoke Times to The Virginian-Pilot is important tounderstanding Virginia’s media landscape.

Interestingly, the observed path of influence among the newspapers does notcorrespond directly to the newspapers’ rankings by circulation (Audit Bureau ofCirculations, 2007). However, it does correspond to the number of news storiesabout the gubernatorial race published by each paper. The Richmond Times-Dispatchpublished more stories about the race than did The Washington Post, which publishedmore than The Roanoke Times, which published more than The Virginian-Pilot.The number of stories each newspaper published about the race corresponds togeographical considerations. As the only daily newspaper in the state’s capital, theRichmond Times-Dispatch can be expected to take the lead on issues covered instate elections, since it has the most immediate access to state government officialsand candidates. The Washington Post does not have the same access to the state’sgovernment, and its news coverage is divided between national politics and the localpolitics of Virginia, Maryland, and Washington, DC. However, The Washington Postis the biggest of the four newspapers in terms of readership and it is the only one thatis considered among the elite national newspapers. Therefore, it is not surprising thatThe Washington Post would exert an influence on the agendas of The Roanoke Timesand The Virginian-Pilot.

More interesting is the relationship between the two smaller newspapers. TheRoanoke Times published more articles about the race and showed a clear influenceon The Virginian-Pilot, which has a larger readership and serves a larger metropolitanarea. Both are owned by Landmark Communications, but The Virginian-Pilot isconsidered the company’s ‘‘flagship newspaper’’ (Landmark Communications, n.d.,para 3). One explanation for The Roanoke Times’ influence on The Virginian-Pilotand its greater interest in the race is that Kilgore is a native of southwest Virginia, thepart of the state served by The Roanoke Times. Since neither candidate came from theTidewater region served by The Virginian-Pilot, the candidates may not have madeas strong of an effort to seek coverage in that newspaper.



Only one incidence of intercandidate agenda setting was observed, during theMarch to April time lag. It is noteworthy that this instance involved Kilgore causinga shift in Kaine’s agenda. It is also important to note that this shift occurred earlyin the campaign, indicating that Kaine was more flexible and willing to experimentwith possible campaign themes early in the year. Although it may not be the case theKaine was conforming to the issue agenda of Kilgore, but merely repositioning hiscampaign agenda to issues that were playing more strongly to his base constituency,the evidence shows the shift occurred in Kilgore’s favor.

Kaine’s ultimate victory in the race indicates that it is not an inherently badstrategy for a candidate to shift his or her agenda to conform to that of his or heropponent. Tedesco’s (2005a) study of the 2004 Democratic presidential primaryfurther supports this conclusion, as ultimate winner John Kerry shifted his agendaseveral times to conform to the agendas of his opponents. By talking about the sameissues as your opponents it is easier to get into the campaign dialogue and debate. Inthe 2004 general election, Tedesco (2005b) showed that Kerry set Bush’s agenda forthe month of June. Again, in this case it was Bush, the ultimate winner of the election,whose agenda was set by his opponent. Future researchers can further investigate thisconclusion that ultimate election winners are just as likely as unsuccessful candidates,perhaps even more likely, to have their agendas set by their opponents. What may beat play in these shifts of agendas is a candidate’s recognition of the need to defenda stance on an issue that is under attack or that has been elevated in salience by theopponent. A skilled campaigner and an organized campaign are able to shift issueagendas and still maintain interpretive dominance (Smith, 1995) on issues.

The present study assessed the candidates’ agendas based on analysis of candidatepress releases. Although these releases should represent the candidates’ agendas fairlywell, future researchers should examine additional campaign materials that may influ-ence media agendas and the agendas of opposing campaigns. Information subsidiesprovided by campaigns come in many forms other than traditional press releases. Forexample, in the election examined in this study, the candidates and representatives oftheir campaigns maintained frequent contact with reporters via e-mail and conferencecalls (R. E. Denton, Jr., personal communication, November 15, 2005). In addition,political advertising has been shown to influence media agendas during elections(Boyle, 2001; Roberts & McCombs, 1994). Candidate speeches and platforms arealso promising focuses for agenda-setting research (Kiousis & Shields, 2006). Futurestudies could use interviews with journalists or field observations of newsrooms toget a clearer picture of how various information sources contribute to media agendas.

This study was limited by the fact that news articles were collected only duringthe last 2 months of the campaign. The high number of articles published about thecampaign in Virginia newspapers made it necessary to limit the sample in this way.The presence of intercandidate agenda setting early in the campaign raises the issue ofwhether candidates might exert stronger media agenda-setting influences before thebeginning of the campaign’s hot phase. Perhaps the reciprocal relationships foundbetween the candidates and two of the newspapers are artifacts of the newspapers



maintaining agendas that were set by the candidates earlier in the year. A study thatincluded news articles from throughout the year might detect these early influences.

Conclusion

As McCombs (2004) pointed out, the advent of new communication technologiesnecessitates reexamination of accepted communication theories such as agenda set-ting. The assumption of this project is that, far from delegitimizing the study of agendasetting, technologies such as the Internet open up possibilities for studying influenceson more than just the traditional media, policy, and public agendas. This study usedpress releases and media content available on the Internet to replicate elements of pre-vious studies examining the influences of candidate information subsidies on mediaagendas and on the agendas of other candidates. Numerous studies have demon-strated the influence of candidate and media agendas on public agendas, so the lineof research continued in this study is important to understanding how those agendasare formed. A better understanding of this process is essential to developing a broaderunderstanding of how communicative processes function in political elections.

This study found that the 2005 Virginia gubernatorial candidates successfullyused information subsidies to influence two newspaper agendas that Kilgore exertedan agenda-setting influence on Kaine early in the campaign, and that Virginianewspapers exhibited intermedia agenda-setting effects on each other. This studyoffers an important contribution to the study of candidate and media agenda setting byextending that line of research to a state electoral context besides Texas gubernatorialelections. This emphasis on state-level politics represents an underexplored area ofpolitical communication research. Future scholars should expand this research intomore electoral contexts, including other states as well as local elections. Concurrentanalyses of elections occurring in different places at the same time or in subsequentyears would allow researchers to examine trends over time in how candidate andmedia agendas interact. In addition, future research should examine these phenomenain local and statewide primary elections, to see how those contexts differ from thepresidential primaries studied by Tedesco (2001, 2005a).

As candidates and media outlets compete to be heard in an increasingly crowdedcommunication environment, political communication researchers must focusgreater attention on how candidates and media interact in campaigns. As Trentand Friedenberg (2004) argue, ‘‘Mass communication has become the center stagefor all major political events’’ (p. 118). If this assumption is true, then communicationscholars need to examine not only how mediated political communication affects thepublic, but also who shapes the content that appears in that communication. Thisstudy represents one of several preliminary attempts to increase that understanding,and it should serve as both a contribution to that line of inquiry and a call to scholarsto focus more attention on the full spectrum of agenda-setting processes.



Notes

1 Data for November 1 through November 8 were also collected and compared with thedata from October using the same cross-lagged correlation method. In everycomparison, the relationship between October and November was either identical to therelationship between September and October or it demonstrated reciprocal influencewhere a one-way influence existed over the previous time lag (see Dunn, 2006). Thus, theOctober to November comparisons were omitted from this article in the interest of space.

2 The formula for calculating the baseline value is

Baseline = RX1Y1 + RX2Y2

2

√(RX1X2)2 + (RY1Y2)2

2

3 All correlation coefficients reported in Figures 1 through 4 are statistically significant atthe p < .01 level based on two-tailed tests.

References

Almanac of Virginia Politics. (2002). Ashburn, VA: World-Wide Publishing Services.Atwater, T., Fico, F., & Pizante, G. (1987). Reporting on the state legislature: A case study of

inter-media agenda-setting. Newspaper Research Journal, 8, 53–61.Atwood, E. L. (1980). From press releases to voting reasons: Tracing the agenda in a

congressional campaign. Communication yearbook, 4, 467–482.Audit Bureau of Circulations. (2007). Top 200 newspapers by largest reported circulation.

Retrieved July 22, 2007, from http://www.accessabc.com/products/top200.htmBolden, J. H. (1970). Daily newspaper coverage of the 1970 Texas gubernatorial campaign.

Unpublished master’s thesis, University of Texas, Austin, TX.Boyle, T. P. (2001). Intermedia agenda setting in the 1996 presidential election. Journalism

and Mass Communication Quarterly, 78, 26–44.Carsey, T. M. (2000). Campaign dynamics: The race for governor. Ann Arbor, MI: University

of Michigan Press.Curtin, P. A. (1999). Reevaluating public relations information subsidies: Market-driven

journalism and agenda-building theory and practice. Journal of Public Relations Research,11, 53–90.

Danielian, L. H., & Reese, S. D. (1989). A closer look at intermedia influences on agendasetting: The cocaine issue of 1986. In P. J. Shoemaker (Ed.), Communication campaignsabout drugs: Government, media, and the public (pp. 47–66). Hillsdale, NJ: Erlbaum.

Dao, J. (2005, October 20). 2 stump in popular governor’s shadow. The New York Times,A16. Retrieved December 7, 2005, from LexisNexis database.

Dearing, J. W., & Rogers, E. M. (1996). Agenda-setting. Thousand Oaks, CA: Sage.Dunn, S. W. (2006). Candidate and media agenda setting in the 2005 Virginia gubernatorial

election. Unpublished master’s thesis, Virginia Tech, Blacksburg, VA. Retrieved October14, 2006, from http://scholar.lib.vt.edu/theses/available/etd-04212006-115449/

Evatt, D. S., & Bell, T. (2001). Up stream influences: The early press releases, agenda-setting,& politics of a future President. Southwestern Mass Communication Journal, 16, 70–81.

Funkhouser, R. (1973). The issue of the 60’s: An exploratory study in the dynamics of publicopinion. Public Opinion Quarterly, 37, 62–75.



Gandy, O. H. (1982). Beyond agenda setting: Information subsidies and public policy.Norwood, NJ: Ablex.

Iyengar, S, Peters, M., & Kinder, D. (1982). Experimental demonstrations of the‘‘not-so-minimal’’ consequences of television news programs. American Political ScienceReview, 76, 848–858.

Kaid, L. L. (1976). Newspaper treatment of a candidate’s news releases. Journalism Quarterly,58, 93–96.

Kiousis, S., & Shields, A. (2006, June). Intercandidate agenda-setting in presidential elections:Issue and attribute agendas in the 2004 campaign. Paper presented at the annualconference of the International Communication Association, Dresden, Germany.Retrieved from the Communication and Mass Media Complete database.

Ku, G., Kaid, L. L., & Pfau, M. (2003). The impact of web site campaigning on traditionalnews media and public information processing. Journalism and Mass CommunicationQuarterly, 80, 528–547.

Landmark Communications. (n.d.) Our businesses. Retrieved July 22, 2007, fromhttp://www.landmarkcom.com/businesses/index.php

Laris, M. (2005, November 4). Doggedness is a lifelong trait for underdog Potts; State senatoris far behind in the race for Va. Governor but is pressing on with a sense of destiny. TheWashington Post, B04. Retrieved from LexisNexis database.

Mazur, A. (1987). Putting radon on the public’s risk agenda. Science, Technology, and HumanValues, 12(3/4), 86–93.

McCombs, M. (2004). Setting the agenda: The mass media and public opinion. Malden, MA:Blackwell.

McCombs, M. E., & Shaw, D. L. (1972). The agenda-setting function of mass media. PublicOpinion Quarterly, 36, 176–187.

Miller, M. M., Andsager, J. L., & Reichert, B. P. (1998). Framing the candidates inpresidential primaries: Issues and images in press releases and news coverage. Journalismand Mass Communication Quarterly, 75, 312–324.

Morris, T. R., & Sabato, L. J. (1998). Virginia government and politics: Readings andcomments. Charlottesville, VA: University of Virginia.

Nguyen, B., Rowlands, T., & Starr, B. (2005, November 8). Election day 2005: The write stuff[Television broadcast]. Atlanta, GA: Cable News Network. Retrieved from LexisNexisdatabase.

Reese, S. D., & Danielian, L. H. (1989). Intermedia influence and the drug issue: Convergingon cocaine. In P. J. Shoemaker (Ed.), Communication campaigns about drugs:Government, media, and the public (pp. 29–46). Hillsdale, NJ: Erlbaum.

Roberts, M., & McCombs, M. (1994). Agenda setting and political advertising: Origins of thenews agenda. Political Communication, 11, 249–262.

Rozell, M. J. (1991). Local v. national press assessments of Virginia’s 1989 gubernatorialcampaign. Polity, 24, 69–89.

Rozelle, R. M., & Campbell, D. T. (1969). More plausible rival hypotheses in the cross-laggedpanel correlation technique. Psychological Bulletin, 71, 74–80.

Shea, D. M., & Burton, M. J. (2001). Campaign craft: The strategies, tactics, and art of politicalcampaign management. Westport, CT: Praeger.

Shear, M. D. (2005, January 18). Va. Attorney General to resign; Va. Governor’s race to getfull attention. The Washington Post, B01. Retrieved April 5, 2006, from the LexisNexisdatabase.



Siegel, R. (2005, November 4). All Things Considered [Radio broadcast]. National PublicRadio. Retrieved December 7, 2005, from LexisNexis database.

Sigal, L. V. (1973). Reporters and officials: The organization and politics of newsmaking.Lexington, MA: D. C. Heath.

Smith, C. A. (1995). The struggle for interpretive dominance. In K. K. Kendall (Ed.),Presidential campaign discourse (pp. 293–303). Albany, NY: State University of New YorkPress.

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

Tedesco, J. C. (2005a). Intercandidate agenda setting in the 2004 Democratic presidentialprimary. American Behavioral Scientist, 49, 92–113.

Tedesco, J. C. (2005b). Issue and strategy agenda setting in the presidential election:Exploring the candidate–journalist relationship. Journalism Studies, 6, 187–201.

Tipton, L., Haney, R. D., & Baseheart, J. R. (1975). Media agenda-setting in city and stateelection campaigns. Journalism Quarterly, 52, 15–22.

Trent, J. S., & Friedenberg, R. V. (2004). Political campaign communication: Principles &practices. Lanham, MD: Rowman & Littlefield.

Tumulty, K. (2005, November 21). Mark Warner, Virginia: A cell-phone king’s finereception. Time. Retrieved December 7, 2005, fromhttp://www.time.com/time/archive/preview/0,10987,1129563,00.html

Turk, J. V. (1986). Information subsidies and media content: A study of public relationsinfluence on the news. Journalism Monographs, Serial No. 100.

Vermeer, J. P. (1982). ‘‘For immediate release’’: Candidate press releases in American politicalcampaigns. Westport, CT: Greenwood Press.

Weaver, D., Graber, D. A., McCombs, M., & Eyal, C. H. (1981). Media agenda-setting in apresidential election: Issues, images, and interests. New York: Praeger.

Wilson, H. L. (1991). Media treatment of black candidates: The 1989 Virginia gubernatorialcampaign. Virginia Social Science Journal, 26, 82–90.


L’établissement de l’ordre du jour des candidats et des médias lors des élections pour le poste de gouverneur de la Virginie en 2005 Scott W. Dunn Résumé Cette étude examine les relations entre les ordres du jour des candidats et des médias lors des élections pour le poste de gouverneur de la Virginie en 2005. Les communiqués de presse des candidats et les articles de journaux portant sur l’élection ont été analysés avec le programme informatique VBPro. De fortes corrélations de l’ordre du jour indiquent que les deux candidats liés aux grands partis (le démocrate Tim Kaine et le républicain Jerry Kilgore) et les quatre journaux principaux de l’État ont discuté d’enjeux similaires lors de la campagne. Des corrélations avec décalage indiquent que les ordres du jour des deux candidats partageaient des relations réciproques avec les ordres du jour de deux de ces journaux, alors que les candidats établissaient l’ordre du jour des deux autres journaux. Les ordres du jour des candidats s’influençaient mutuellement lors de la phase active de la campagne, bien qu’une analyse ultérieure indique que plus tôt, Kilgore établissait l’ordre du jour de Kaine. Les conséquences pour la recherche future sur l’ordre du jour sont discutées.

Candidate and Media Agenda Setting in the 2005 Virginia Gubernatorial

Election

2005년 버지니아 주지사 선거에서의 후보와 미디어 아젠다 세팅에 관하여

Scott W. Dunn

School of Journalism and Mass Communication, University of North Carolina at

Chapel Hill, Chapel Hill,

NC 27599

요약

본 연구는 2005년 버지니아주의 주지사 선거기간동안 후보자와 미디어

아젠다들과의 관계들을 연구한 것이다. 컴퓨터 프로그램인 VBPro를 사용하여

선거에 관한 후보자들의 언론보도문과 신문기사들을 분석하였다. 높은 정도의

아젠더 상관관계들은 두 주요정당의 후보들 (민주당의 Tim Kaine와 공화당의

Jerry Kilgore)과 버지니아주의 4개 주요신문들이 캠페인기간동안 비슷한

의제들을 토론하였음을 보여주고 있다. 교차지연 상관관계들은 두후보자들의

아젠더들은 두 신문의 아젠다와는 상호적 관계를 가지는반면, 다른 두 신문의

아젠다를 형성하였다는 것을 보여주고 있다. 매체간 의제설정 효과들 역시

조사되었다. 후보자들의 아젠더들에 있어, 비록 사후검정분석은 Kilgore가

초기에는 Tim Kaine의 아젠더를 형성한 것으로 나타났으나, 캠페인이 절정에

다다른 기간동안에는 서로간 영향을 준 것으로 나타났다. 미래 아젠더 세팅

연구들에 있어서의 함의들이 논의되었다.

2005年弗吉尼亚州州长选举中的候选人和媒体的议程设置

Scott W. Dunn

北卡罗莱纳大学教堂山分校新闻与大众传播学院

【摘要：】

本论文研究了在2005年弗吉尼亚州州长选举中的候选人和媒体议程之

间的关系。我们运用计算机程序VBPro分析了有关候选人的新闻稿和有关

选举的报刊文章。议程高度的相关性表明，两大党的候选人——

民主党人Tim Kaine和共和党人Jerry

Kilgore，以及该州四家主流报纸都在选举过程中讨论了相似的议题。议程

的交叉滞后相关性表明，两位候选人的议程与其中两家报纸的议程相互作

用，而候选人为其他两家报纸设置议程。我们也发现了跨媒体的议程设置

的效果。虽然事后分析表明Kilgore较早地为Kaine设置议程，但是在竞选

的激烈阶段，候选人的议程相互影响。此外，本文讨论了对未来的议程设

置研究的意义。

2009_dunn_candidate and media agenda setting

Documents