438916 12438916CoeCommunication Research © The Author(s) 2012

CRXXXX10.1177/00936502

Reprints and permission: http://www. sagepub.com/journalsPermissions.nav

Television News, Public Opinion, and the Iraq War: Do Wartime Rationales Matter?

Communication Research 40(4) 486­–505 © The Author(s) 2012 Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/0093650212438916 crx.sagepub.com

Kevin Coe1

Abstract The idea that the public is swayed by the rationales for war that circulate throughout the media environment is well established in popular discourse. Research on the determinants of support for war, however, has largely ignored the role that such rationales might play. This study is the first to directly test the possibility that rationales for war present in news coverage influence public support for war. Pairing a detailed computer-assisted content analysis with measures of public support for the Iraq War, this study shows that, contrary to popular wisdom, rationales for war present in television news have only a limited impact on public attitudes about U.S. military action. Keywords television news, framing, public opinion, war, Iraq Deeply ingrained in the American mythos is an image of the American public turning to media to help gauge the value of military action. We imagine people gathered around radios listening to Franklin Roosevelt, in one of his fabled “fireside chats,” promote the defense of freedom in Europe, or we think of families staring at television sets as news reports broadcast disturbing pictures from Vietnam. More recently, we recall media coverage of the Bush administration’s months-long campaign to convince the public that it was necessary to invade Iraq and displace Saddam Hussein. In popular discourse, such moments are typically viewed as deeply consequential for public attitudes about war. There is no shortage of commentary, for instance, suggesting that President Bush effectively “spun”

1

University of Arizona, Tucson, AZ, USA

Corresponding Author: Kevin Coe, Department of Communication, University of Arizona, 211 Communication Building, Tucson, AZ 85721-0025, USA. Email: [email protected]

Coe

487

news coverage in his favor and induced the public to support U.S. engagement in Iraq (e.g., Fritz, Keefer, & Nyhan, 2004; Isikoff & Corn, 2006; Rich, 2006). Despite this common perception, there is little evidence to support the idea that media content has a measurable impact on public support for war. But neither is there much evidence to support the idea that it does not. Surprisingly, the growing body of research on the determinants of public support for war (e.g., Feaver & Gelpi, 2004; Gartner & Segura, 1998; Jentleson, 1992; Mueller, 1973) has generally neglected the role that political discourse and news coverage might play in influencing the public. The few studies that have considered the role of such communication (e.g., Berinsky, 2009; Larson, 1996, 2005; Zaller, 1992) have focused narrowly on consensus or dissensus among political leaders, without precisely measuring the actual media content that might transmit these views to the public. The most extensive treatment of the subject to date (Baum & Groeling, 2010) addresses these shortcomings, but focuses primarily on the general evaluations of war policy that appear in media coverage, as opposed to the specific justifications provided in support of war. Nor do studies in this area regularly measure the amount of media use among the public to determine their level of exposure to elite perspectives, which severely constrains the interpretation of effects as media generated. The result is that we do not yet know if and how the specific rationales for war that circulate throughout the media environment during wartime actually influence public attitudes about military action. This study is the first to test this long-assumed relationship. It does so by examining the primary medium through which wartime discourse circulates: network television news. Focusing on network news is useful because its audiences, though declining, still dwarf those of other media. During the yearlong period of analysis used in this study, the average nightly audience for the three network news programs was well over 25 million people (Project for Excellence in Journalism, 2005). Further, substantial majorities—consistently in the 70% to 80% range in the decade leading up to the period of analysis—reported that television was the medium they relied on to get “most” of their national and international news (Project for Excellence in Journalism, 2005). Using framing as a theoretical framework, I employ a detailed computer-assisted content analysis procedure to measure the extent to which common rationales for war—freedom, threat, evil, and the like—were present in television news coverage during an important yearlong period of the Iraq War. Including these measures in a series of regression models with other common predictors of war support indicates that, contrary to popular wisdom, rationales for war present in television news have only a limited impact on public attitudes about U.S. military action.

News Media and the Determinants of Public Support for War A considerable and growing body of scholarship has focused on the determinants of public support for American wars (e.g., Feaver & Gelpi, 2004; Gartner & Segura, 1998; Gelpi, Feaver, & Reifler, 2005; Jentleson, 1992; Jentleson & Britton, 1998; Mueller, 1973). In general, this research has concluded that public attitudes about war are the result of a

488

Communication Research 40(4)

rational calculus. Citizens are thought to weigh the costs of the war—especially the number of U.S. casualties—against the benefits of the war and the likelihood that those benefits will ultimately be attained. To the extent that the benefits of a war are judged to outweigh its costs, the war receives public support. This calculus is conditioned by the events that take place during the war—success in a major battle might result in an uptick in support, for example—and by the level of consensus or dissensus among U.S. government officials. When political leaders are united in their support for a war, the public is more supportive. When leaders begin to split, however, citizens use partisan cues to determine whether their “side” is still supportive and adjust their views accordingly. Such elite consensus/dissensus (e.g., Berinsky, 2009; Larson, 1996, 2005; Zaller, 1992), the impact of which is heavily influenced by media framing (Baum & Groeling, 2010), is the primary place where scholarship on war support meaningfully engages the role that communication might play in shaping public attitudes about war. The puzzle that emerges, then, is this: research on the determinants of support for war focuses primarily on “real-world” factors without necessarily acknowledging that most of these factors are constructed rhetorically or transmitted to the public via the news media. For example, although actual casualties are an objective reality of war, public knowledge of casualty levels is usually gained through news coverage (Gartner, 2004). Additionally, studies that find that, citizens’ tolerance for casualties is based on the mission’s objective (e.g., Jentleson, 1992; Jentleson & Britton, 1998; Oneal, Lian, & Joyner, 1996) rely heavily on news coverage to determine that objective. After all, absent political rhetoric circulating in mass media, who would know what a particular war is aiming to achieve? As for events, they tend to mean little unless they are given importance through news coverage. Indeed, the size of the opinion rally that typically follows key events depends in part on how successful political elites are in drawing media attention to the event (Baker & Oneal, 2001). Further, without news coverage, scholars have difficulty gauging what events are significant enough to include in their studies; therefore, most use news coverage to make such decisions (e.g., Mueller, 1973; Sigelman, Lebovic, Wilcox, & Allsop, 1993). In short, the majority of the literature on public support for war focuses on real-world factors to explain public support for war. In so doing, ironically, it highlights opportunities for public communication—especially news coverage—to influence support. With this in mind, the question becomes whether the content of news coverage actually does influence the public. The few studies that have explored this question give reason to believe it might. Theoretically, new war-relevant information, transmitted to the public via news coverage, could cause citizens to alter their thinking about the war. Zaller (1993), for example, demonstrated that opinion about the Gulf War could be partially explained by the traditional Converse-McGuire model, which holds that attitude change is simply the probability of receiving a message times the probability of accepting the message given that it was received. Thus, the more exposure one has to news coverage of war, the more opportunity that person has to receive new information that might influence attitudes (see also Fan, 1993). Althaus and Kim (2006), meanwhile, found that Gulf War opinion about the president’s handling of the war responded to shifts in the quantity and tenor of news coverage. In the most extensive treatment to date, Baum and Groeling (2010) showed that public

Coe

489

attitudes are responsive to the level of favorable and unfavorable evaluations of war policy present in news coverage. However, none of these studies examined the specific justifications for war that also circulate in news coverage. In reviewing this literature, Berinsky and Druckman (2007) called for more research focusing on “the mix of messages from competing elites (of varying credibility), the number of messages, [and] the vividness of the messages,” because these factors “undoubtedly shape what people think about the war” (pp. 138-139). The present study takes a step in this direction by testing the effects of the rationales for war that circulate in news coverage. Framing theory provides a useful framework for understanding why such rationales might influence public support for war. Entman (2004) explains that framing is the act of “selecting and highlighting some facets of events or issues . . . so as to promote a particular interpretation, evaluation, and/or solution” (p. 5). News stories do precisely this when they talk about wars in certain terms as opposed to others. A war might be framed as a defense of freedom, or as a way to rid the world of evil, or as a necessity that ensures the security of the American people, or as many other things. To the extent that these frames promote a certain understanding of the war in the minds of the audience, public support might vary accordingly. Numerous studies have tracked the presence of various frames in news coverage during times of war and crisis (e.g., Bennett, Lawrence, & Livingston, 2007; Coe, 2011; Domke, 2004; Entman, 2004), but it is rare for studies in this domain to directly test the effects of these frames on public opinion. Nonetheless, evidence demonstrating the effects of framing in a wide range of other contexts, from issues of race and gender to issues of free speech, is plentiful (e.g., Kellstedt, 2000; Nelson, Clawson, & Oxley, 1997; Terkildsen & Schnell, 1997). Translating the findings of this body of research into the context of war suggests that rationales circulating in news coverage should influence public support. However, this is not the whole story. There is also reason to believe that the rationales for war present in news coverage may have very little impact on public support. In particular, several studies have shown that framing effects elicited in a laboratory might be of limited consequence in the real world (Druckman, 2004; Druckman & Nelson, 2003). This is because the effects of news frames might last for only a short period of time, and because the effects of competing frames might cancel each other out and thus generate no net change in public attitudes. Further, issues that touch on deeply held core values are less subject to framing effects than are other issues (Nelson et al., 1997). Rationales for war, which sometimes play on core values such as freedom, might therefore be hampered in their ability to move public opinion. This way of thinking about the potential for news frames to influence the wartime public is consistent with the “limited effects” tradition of mass media research (e.g., Hovland, Lumsdaine, & Sheffield, 1949; Lazarsfeld, Berelson, & Gaudet, 1944). This perspective takes seriously the fact that media are only one factor operating within a nexus of important influences in people’s lives. Consequently, news media are thought to usually act as “an agent of reinforcement rather than change” (Klapper, 1960, p. 18).

490

Communication Research 40(4)

Expectations for the Influence of News Rationales in Iraq Taken together, the literature discussed so far highlights two things: the need for a direct test of the influence of wartime rationales and the competing expectations that exist about the impact such rationales might have on public support for war. This section of the article briefly clarifies what impact we can expect the rationales that circulated in television news to have had during the yearlong period of the Iraq War considered in this study. First, we should expect rationales to influence public opinion, if only modestly. Per the discussion above, framing theory predicts that heavy news emphasis on specific justifications for the war could lead people to understand the war in these terms and adjust their support accordingly. Previous research has shown that, throughout much of U.S. history, a consistent set of rationales has been used to justify war (Coe, 2009; Ivie, 1974). America is presented as going to war to meet a direct threat to the nation’s security and values; to stamp out evil in the world; to defend freedom at home and abroad; to promote eventual peace; to support the sacrifice made by the troops; and to protect religious faith by acting in accordance with God’s wishes for the nation. Each of these six rationales frames the war in decidedly positive terms, so the net effect of each should be to increase public support. Given this, and given the popular belief that news content sways the wartime public, the following set of hypotheses is proposed: Hypothesis 1 (H1a): An increased television news emphasis on the threat rationale will predict an increase in public support. Hypothesis 1 (H1b): An increased television news emphasis on the evil rationale will predict an increase in public support. Hypothesis 1 (H1c): An increased television news emphasis on the freedom rationale will predict an increase in public support. Hypothesis 1 (H1d): An increased television news emphasis on the peace rationale will predict an increase in public support. Hypothesis 1 (H1e): An increased television news emphasis on the troops rationale will predict an increase in public support. Hypothesis 1 (H1f): An increased television news emphasis on the faith rationale will predict an increase in public support. Second, we can expect that any influence that media rationales had during this period would vary based on a host of individual-level characteristics, such as basic demographics, that typically influence support for war (Berinsky, 2009; Jacobson, 2007; Wilcox, Hewitt, & Allsop, 1996). In particular, political ideology would likely be important. Partisans have long been known to respond differently to war, and the Iraq War has been especially divided along partisan lines (Jacobson, 2007). Gaines, Kuklinski, Quirk, Peyton, and Verkuilen (2007), for instance, found that partisans shared roughly the same information

Coe

491

about Iraq, but filtered that information through their own political lens. Further, in his classic study of public support for war, Mueller (1973) noted that when asked if they support an ongoing war, “Partisans seem to see the query as one asking for the approval or disapproval of an action of the administration” (p. 116; see also Zaller, 1992). Given this connection that citizens draw between their “side” and their opinion on the war, it follows that in the case of the Iraq War—which was initiated by, and strongly associated with, a Republican administration—Republicans would be more receptive to rationales for that war. Put simply, it is easier for a message to increase support among people predisposed to support than among those predisposed to skepticism. With this in mind, we should expect Republicans to be more receptive to the rationales for war that circulated in television news during the Iraq War. Thus, the following hypothesis is proposed: Hypothesis 2 (H2): To the extent that news rationales predict increased public support, the increase will be greater among Republicans than among Democrats. Finally, we should expect news exposure to matter. In particular, it is likely that those with higher exposure would be more influenced by rationales for war circulating in news coverage because, quite simply, they would have more opportunity to encounter them (see Zaller, 1993). A recent study, for example, demonstrated that, across several different wars, levels of support changed more substantially when newspaper coverage about the war was plentiful than when it was rare (Althaus & Coe, 2011). Thus, a final hypothesis is proposed: Hypothesis 3 (H3): To the extent that news rationales predict increased public support, the increase will be greater among high news users than low news users.

Method This study focuses on television news coverage and public opinion about the Iraq War between November 3, 2003, and October 31, 2004. This period of analysis was chosen to coincide with appropriate measures of public opinion (discussed below) but has several other advantages. For one, a year is a reasonably long period of time. Although it is a fraction of the total time the U.S. spent in Iraq, it is much longer than some other U.S. military engagements (e.g., it is nearly twice as long as the entire engagement in the Persian Gulf). Further, this particular year was a transformative period in the war: “major combat operations” had concluded, the insurgency was taking full form, and several key events took place, including Saddam Hussein’s capture and the scandal at Abu Ghraib prison. During this time, public opinion about the war was still moving (varying within a range of roughly 20 points), meaning people had not yet settled their minds about the war—a key difference between this period and most of the years since. Finally, this was the period leading up to the 2004 presidential election, in which Iraq was a central issue. Thus, news coverage was devoting considerable attention to the war, making rationales more likely to surface and more likely to have serious political consequences. In sum, this period provides an excellent

492

Communication Research 40(4)

opportunity to determine if rationales for war present in news coverage influenced public opinion.

Analysis of Public Opinion The surveys that provide the standard wartime support questions used in much of the extant scholarship (e.g., Althaus & Coe, 2011; Larson, 1996, 2005; Mueller, 1973) rarely include measures of media use. Consequently, it is usually impossible to partially disaggregate these data to capture opinion dynamics among those most likely to hear media rationales. My solution to this problem is to sacrifice the breadth of analysis made possible by focusing on multiple wars for the precision of analysis made possible by finding survey data that include the necessary measures of media use. Although focusing on a particular time period during a single war limits my ability to make broad claims, it increases the likelihood that I will be able to capture those effects that do exist in this particular instance. As this study represents the first test of the influence of wartime rationales, precision over breadth seems an appropriate trade-off. With this in mind, I rely on the 2004 National Annenberg Election Survey (NAES). This survey is among the most intensive election studies ever conducted, with surveys in the field every day for more than a year: from early October 2003 to mid-November 2004. In total, the study included 81,422 respondents and had a 54% cooperation rate (for full survey details, see Romer, Kenski, Winneg, Adasiewicz, & Jamieson, 2006). For the purposes of this study, the NAES has two key advantages: (a) it includes the “worth it” question, a standard measure of war support; (b) it asks respondents about their television news use, which allows those who regularly watch television news to be analyzed separately from those who do not. To ensure I was working with the most representative part of the NAES data, I excluded a few weeks at the beginning and end of the survey period when the samples were not as large. Doing so left me with 1 year’s worth of daily opinion data: the 52 Monday through Sunday weeks leading up to Election Day. This 52-week period totaled 364 days; NAES had data for 357 of them (surveys were not completed on major holidays).1 Analysis focused on respondents who gave a valid answer to the “worth it” question (n = 35,974), which asked, “All in all, do you think the situation in Iraq was worth going to war over, or not?” Additionally, a variety of standard demographic variables that have been shown to influence public support for war were included (see Berinsky, 2009; Jacobson, 2007; Wilcox et al., 1996). These were party identification (dummy variables for Republican and Democratic identifiers), gender (males = 1, females = 0), race (Non-Whites = 1, Whites = 0), age (in years), and education (college degree or more = 1, else = 0). To analyze different groups based on the amount of network television news they consume, I used the question: “How many days in the past week did you watch the national network news on TV? By national network news, I mean Peter Jennings on ABC, Dan Rather on CBS, Tom Brokaw on NBC, and the Jim Lehrer ‘NewsHour’ on PBS.” Splitting responses at the midpoint left me with one group of low users (0-2 days) and one group of high users (3-7 days).

Coe

493

Finally, U.S. casualties were included because they have been shown to decrease support for war (Gartner & Segura, 1998; Mueller, 1973). Daily casualty data were retrieved from validated Department of Defense data compiled by the Iraq Coalition Casualty Count website at www.icasualties.org. From these counts, I calculated the 28-day marginal casualty rate; that is, the number of American deaths incurred in the past 28 days. I chose a 28-day period because it roughly parallels previous research (e.g., Althaus & Coe, 2011; Gartner & Segura, 1998) and because it mirrors the length of the cumulative news measures used in much of the analysis (discussed below). The total casualty count for the year, plus the preceding 28 days used to calculate the initial marginal rates, was 799 U.S. deaths; the weekly mean was 14.27 (SD = 9.57).

Analysis of News Coverage The sample of news coverage consists of the primary evening news programs from each of the top broadcast news networks. Thus, the programs used were ABC’s World News Tonight with Peter Jennings, CBS’s Evening News with Dan Rather, and NBC’s Nightly News with Tom Brokaw. Full-text transcripts of the three programs were accessed via the Nexis database. To identify segments that focused on the war, transcripts were searched using the string of Iraq* “anywhere in the document.” Using such a broad term ensured that all relevant segments would be recalled by the search. To increase “precision” (see Stryker, Wray, Hornik, & Yanovitzky, 2006), returned segments were included in the sample only if they mentioned substantive information about Iraq. The few excluded stories mentioned Iraq only in passing, such as to introduce or close a story about something else. Neither introductions to the news broadcasts nor previews of upcoming segments/ programs were included in the analysis. A program was randomly assigned to the first day of the period; after that the programs were alternated in a cycle such that ABC would be used one day, CBS the next, NBC the third, and so on. The news programs were sometimes preempted for other programming (usually sporting events). In these cases, the remaining available network was taken instead of the assigned network. If two networks were available, a coin flip determined which was used. This procedure yielded 1,017 segments (ABC = 266; CBS = 349; NBC = 402).2 Computer-assisted content analysis (CCA) was necessary to analyze such a large sample of texts. CCA has a long history in political communication research and is used regularly by social scientists in various disciplines (see West, 2001). Two programs were used: TextQuest (www.textquest.de/eindex.html) and Concordance (www.concordancesoftware. co.uk). To conduct the analysis, the six most common rationales for war (discussed below; see Coe, 2009) were operationalized into a series of terms and phrases drawn directly from the sample of news coverage. I began by generating a list of every distinct word that television news used during the period of analysis (15,567 distinct words; 446,611 total words). I read this list carefully, identifying any word that could potentially represent one of the six rationales for war. When necessary I looked at the words in context (facilitated by the Concordance program) to determine if their use in the text consistently represented one of

494

Communication Research 40(4)

the rationales. In cases where words were sometimes used to represent that rationale and sometimes used to mean something else, multiple-word phrases were selected to isolate the relevant uses. This procedure produced a comprehensive coding dictionary for each of the six rationales. This detailed approach ensured that the coding categories were grounded entirely in the actual wartime discourse, and that the presence of each rationale could be precisely measured. The final coding dictionaries together consisted of more than 500 terms and phrases. Brief descriptions and examples of each rationale follow, with full dictionaries available from the author on request. Eliminating a threat. This dictionary consisted of language stating that America, its allies, or civilization itself faced a threat to its security or way of life. Examples included danger, peril, tinderbox, threat, and WMD. Enemies, forces, or events that might implicitly suggest threat (e.g., Osama Bin Laden, September 11) were not included in the coding unless the speaker made explicit the threat they posed. To account for the differing degree to which the speakers emphasized threat, modifiers such as “deadly” and “grave” were counted when they preceded relevant terms. Thus, “danger” received one code, whereas “grave danger” received two. This approach was used for all of the coding categories. Confronting evil. This dictionary consisted of language stating that the enemy was evil, savage, aggressive or cowardly—in nature or behavior. Examples included barbaric, cold blooded, cowardly, evil, and sadistic. Enemies, forces, or actions that might implicitly suggest evil (e.g., Saddam Hussein, murder) were not included in the coding unless the speaker made explicit the evil they represented. Advancing freedom. This dictionary consisted of language promoting the value of freedom and democracy. Examples included democracy, freedom, liberty, liberators, and selfgovernance. Language suggesting the opposite of freedom (e.g., dictatorship, tyranny, oppression) was also included in this dictionary because speakers consistently used these points of contrast to trumpet the value of freedom. Promoting peace. This dictionary consisted of language stating that America sought to avoid war and/or achieve peace. Examples included cease-fire, détente, olive branch, peace, and truce. General statements about war being a terrible thing were not taken; speakers needed to talk specifically about avoiding or ending the war for the language to be coded. Supporting the troops. This dictionary consisted of language that referred to the troops (including Veterans) or their sacrifice. Examples included Arlington National Cemetery, fighting men, heroes, Marines, and service members. Keeping the faith. This dictionary consisted of explicit references to God and language stating the enemy was shunning God or denigrating religion. This rationale was rarely present. Examples included God, Lord, and theocracy. Two procedures were used to ensure the validity of the coding. First, numerous exclusion terms and phrases were included to further specify which terms and phrases the computer should or should not code. For example, casualties was taken as a term to represent Supporting the Troops, but “Iraqi casualties,” “no American casualties,” and the like were excluded. Second, TextQuest’s interactive coding feature was used. This feature allows each of the computer’s codes to be seen in the full context of the text, then either accepted

Coe

495

or rejected. This type of test is necessary because there will always be unexpected instances in computer coding where a certain term or phrase is used atypically and therefore does not adequately represent the coding category. I interactively coded approximately 10% of the texts and recorded the percentage of valid codes for each category (i.e., the percentage of codes that accurately captured the coding category). For each of the dictionaries, validity exceeded 90%. Specifically, the validity scores were Threat (90%), Evil (96%), Freedom (94%), Peace (95%), Troops (99%), and Faith (91%). To be clear, then, these six rationales for war—as emphasized in television news— constitute this study’s primary independent variables. Each of the six is measured by tracking the number of times it was used in television news during the period immediately preceding a survey respondent being asked if the Iraq War was “worth it.” To offer the broadest possible test, the initial regression model presented in the results section tracks each of these six variables at three different cumulative lags: 7-day, 14-day, and 28-day. So, for example, the 7-day measure for the freedom rationale is the total number of times that news stories used freedom terms in the 7 days prior to the survey being conducted. Recent research (e.g., Althaus & Coe, 2011; Althaus & Kim, 2006) has demonstrated that, outside of the laboratory, the effect of news exposure is often a cumulative process. Testing three different lag lengths will help ensure that this study is able to capture any measurable effect of television news on public support for war.

Results In the year leading up to the presidential election of 2004, the Iraq War was a regular topic in network television news coverage. Of the 364 days of news considered here, 348 contained at least one segment that mentioned Iraq. In this coverage, television news regularly presented many of the common rationales for war. Troops was far and away the dominant theme, used on average 111.23 times per week (SD = 70.22). It dwarfed even the combined presence of the other themes. Threat (M = 17.44; SD = 10.75) also had a considerable presence, whereas Freedom (M = 6.56; SD = 6.33), Peace (M = 3.08; SD = 3.46), and Evil (M = 2.08; SD = 2.52) were used less regularly. Faith (M = .63; SD = .97) was all but absent from television news during this period. The high standard deviations of all of the themes reveal the wide variation that existed in use of these rationales. Often a program would use numerous rationales on a single evening—as, for example, when the program quoted heavily from a presidential speech about the war—and then that rationale would disappear for the next few days. Given that each of the six rationales did have at least some presence in television news, we can consider the central question of interest here: Did these rationales influence public support for the war? To begin answering this question, it is useful to consider how public support for the Iraq War rose and fell during this period. Figure 1 shows, by week from the beginning of November 2003 to the end of October 2004, the percentage of respondents who said the war was “worth it.” As the figure makes clear, support for the war began and ended this period in nearly the same place, with roughly 48% of the public voicing support. Over the course of the year, though, this support varied noticeably, from a high of more than 55% in

496

Communication Research 40(4)

60

Percentage

55

50

45

40

11

11

/0

3/

0 /2 3 4/ 12 03 /1 5 01 /03 /0 5 01 /04 /2 6 02 /04 /1 6 03 /04 /0 8 03 /04 /2 9 04 /04 /1 9 05 /04 /1 0 05 /04 /3 1 06 /04 /2 1 07 /04 /1 2 08 /04 /0 2 08 /04 /2 3 09 /04 /1 3 10 /04 /0 4 10 /04 /2 5/ 04

35

Figure 1. Respondents Saying Iraq was Worth It.

mid-December immediately after the capture of Saddam Hussein, to a low of 39% in mid to late June just before the U.S. transferred power to the newly formed Iraqi government. The sharpest change that occurred is the 11-point drop in the last week of March. This was the week following the 9/11 commission hearings, which included Richard Clarke’s muchdiscussed criticism of Bush (see Clarke, 2004). In general, the trend after an early spike was downward until a midsummer reversal began a slow ascent that continued for the remainder of the period. To determine whether news rationales accounted for any of this movement, I used logistic regression to test the extent to which—after controlling for a host of variables typically thought to influence support for war—news emphasis on each rationale in the weeks preceding a respondent being surveyed predicted that the respondent would say the Iraq War was “worth it.”3 Table 1 presents the results of the initial regression analysis, using the three different cumulative lag measures discussed in the method section. All of the control variables prove to be significant predictors of war support in understandable ways. Being a Republican predicted increased support, whereas being a Democrat predicted decreased support. Men were more supportive than women. Being non-White, being older, and having a college degree all predicted decreased support. In addition, consistent with prior research, a higher marginal casualty rate predicted decreased support. H1a to H1f predicted that, even after accounting for these known determinants, news emphasis on wartime rationales would predict support. The results show that some of the media rationales did reach statistical significance. Troops was the most consistent

497

Coe Table 1. Predicting Support for the Iraq War From Television News Rationales.

Republican Democrat Male Non-White Age College degree Casualty rate Freedom Peace Troops Threat Evil Faith Constant Model χ2 Cox & Snell R2 N

7-Day

14-Day

28-Day

1.705 (0.031) –0.940 (0.029) 0.179 (0.025) –0.811 (0.037) –0.010 (0.001) –0.360 (0.025) –0.370 (0.074) 0.018 (0.256) –0.274 (0.455) 0.080 (0.024) 0.238 (0.124) –1.291 (0.583) –0.593 (1.280) 0.550 (0.061) 9556.4 .233 35,974

1.705 (.031) –0.942 (.029) 0.178 (.025) –0.810 (.037) –0.010 (.001) –0.360 (.025) –0.432 (.087) –0.168 (.172) 0.277 (.328) 0.046 (.014) 0.218 (.079) –0.670 (.414) 0.657 (.895) 0.527 (.064) 9563.4 .233 35,974

1.706 (.031) –0.943 (.029) 0.178 (.025) –0.812 (.037) –0.010 (.001) –0.361 (.025) –0.333 (.103) –0.308 (.147) –0.067 (.256) 0.025 (.008) 0.180 (.053) –0.627 (.325) 0.541 (.719) 0.510 (.074) 9578.4 .234 35,974

Note. Bold coefficients are significant at the p < .05 level. Cells contain logistic regression coefficients with standard errors in parentheses.

predictor, increasing support at all three measurement lengths. This means that the more that news coverage mentioned the troops, the higher support for the war went. Threat also predicted increased support. Freedom and Evil both predicted decreased support but showed up at only one of the three measurement lengths. Because logistic regression coefficients are difficult to interpret, the impact of these rationales can be better gauged by considering the change in predicted probability of supporting the war when use of the news rationale varies from low to high, holding the values of all other independent variables constant. This way of presenting the results is especially useful in this case because the high number of cases included in the analysis makes statistical significance fairly easy to reach. Thus, some of the differences in Table 1 might be statistically significant but not substantively meaningful. Table 2 presents the findings in this manner, showing the change in predicted probability of support when each rationale is varied from two standard deviations below the mean to two standard deviations above the mean.4 In other words, Table 2 shows the strongest effect that increasing the presence of a given rationale in television news was likely to have on a respondent saying the Iraq War was worth fighting. The results show that none of the rationales produced a substantively meaningful change in predicted support. Looking at the bold numbers in particular, which show those cases where the corresponding logistic regression coefficients reached statistical significance (see Table 1), confirms that statistical significance did not actually indicate the presence of a meaningful difference in

498

Communication Research 40(4)

Table 2. Change in Probability of Support for the Iraq War.

Freedom Peace Troops Threat Evil Faith

7-Day

14-Day

28-Day

.00 .00 .04 .02 –.02 .00

–.01 .01 .03 .03 –.01 .00

–.03 .00 .04 .03 –.02 .01

Note. Bold numbers indicate that the corresponding coefficients in the logistic regression analysis used to produce these predicted changes (see Table 1) were significant at p < .05.

predicted probability of support. In the case of troops, for instance, which was the most influential rationale of the group, moving from –2SD to +2SD increased the probability of support by only four percentage points. So far, then, these results indicate that news rationales neither uniformly influenced support for the Iraq War nor had sizable effects. Given the small magnitude of these effects and the inconsistent direction of influence, H1a to H1f are rejected. Importantly, however, different subgroups of people are likely to respond to wartime rationales in different ways. The broad analysis conducted to test H1a to H1f might therefore mask more substantive effects that occurred only among particular groups. To explore this possibility and test H2, I reran the analysis by party affiliation: self-identified Democrats, Independents, and Republicans. Further, given that high users of television news would be more likely than low users to encounter these rationales (and thus more likely to be influenced), I split out each partisan group by high or low television news use to test H3. Because the 28-day cumulative measure provided the best model fit in the previous series of regressions, I used that measure for this analysis. Table 3 presents the results. The included control variables remain generally consistent predictors of support, with two noteworthy differences: the effect of education disappears for Republicans, and casualties turn out to influence only Democrats. More importantly, the results confirm that war rationales influence different subgroups of the population in different ways. To clarify the magnitude of these effects, I again calculated the change in predicted probability of support when each rationale is varied from two standard deviations below the mean to two standard deviations above the mean. As discussed above, this way of looking at the data better illustrates meaningful impact than does statistical significance alone. Table 4 presents this analysis. The results do not support H2. Republicans proved to be the group least influenced by rationales for war, perhaps reflecting greater stability in their attitudes about the war. In partial support of H3, higher exposure to news predicted stronger effects. Four of the six significant effects were present among high news groups, as were four of the five largest effects. However, as in the broader analysis, the direction of the effects was inconsistent. Further, as above and as expected, the effects generally proved to be modest. The largest change occurred for the troops theme among high news Democrats,

499

Coe Table 3. Predicting Support for the Iraq War Among Different Groups. Democrats   Male Non-White Age College degree Casualty rate Freedom Peace Troops Threat Evil Faith Constant Model χ2 Cox & Snell R2 N

Independents

Republicans

Low

High

Low

High

0.203 (0.063) –0.748 (0.079) –0.009 (0.002) –0.957 (0.069) –0.682 (0.272) –0.360 (0.371) –0.759 (0.688) 0.064 (0.021) 0.111 (0.129) –0.496 (0.837) 2.930 (1.809) –0.055 (0.186) 324.9 .055 5,718

0.173 (0.069) –0.834 (0.095) –0.022 (0.002) –0.630 (0.074) –0.964 (0.290) –1.045 (0.417) 0.723 (0.708) 0.087 (0.022) –0.009 (0.145) 0.050 (0.911) 1.378 (2.003) 0.308 (0.211) 248.6 .041 5,972

0.223 (0.057) –0.842 (0.088) 0.004 (0.002) –0.426 (0.058) –0.418 (0.238) –0.127 (0.335) 0.402 (0.585) 0.023 (0.019) 0.420 (0.121) –0.381 (0.741) –1.665 (1.642) –0.230 (0.165) 194.2 .036 5,290

0.147 (0.063) –0.702 (0.101) –0.014 (0.002) –0.251 (0.063) –0.403 (0.265) –0.202 (0.366) –0.120 (0.661) 0.045 (0.021) 0.187 (0.136) –0.989 (0.830) –1.246 (1.866) 0.600 (0.195) 122.1 .027 4,381

Low

High

0.262 (0.069) –0.092 (0.074) –1.142 (0.110) –1.010 (0.149) 0.005 (0.002) –0.018 (0.002) 0.070 (0.071) –0.061 (0.075) –0.193 (0.284) –0.163 (0.304) 0.186 (0.425) –1.145 (0.424) 0.370 (0.707) 0.930 (0.777) –0.024 (0.022) –0.029 (0.024) 0.289 (0.155) 0.082 (0.163) –1.035 (0.927) 0.823 (0.940) 1.536 (2.025) 1.428 (2.176) 1.398 (0.205) 2.590 (0.232) 128.6 117.7 .019 .026 6,554 4,480

Note. Bold coefficients are significant at the p < .05 level. Cells contain logistic regression coefficients with standard errors in parentheses. All rationale measures are 28-day cumulative.

Table 4. Change in Probability of Support for the Iraq War Among Different Groups. Democrats

Independents

Republicans

 

Low

High

Low

High

Low

High

Freedom Peace Troops Threat Evil Faith

–.03 –.04 .10 .01 –.02 .04

–.08 .03 .11 .00 .00 .01

–.01 .03 .05 .05 –.02 –.03

–.02 –.01 .09 .02 –.04 –.02

.01 .01 –.03 .02 –.03 .01

–.09 .05 –.04 .01 .03 .02

Note. Bold numbers indicate that the corresponding coefficients in the logistic regression analysis used to produce these predicted changes (see Table 3) were significant at p < .05. All rationale measures are 28-day cumulative.

increasing the probability of support by 11 points. Freedom was the other theme that predicted noticeable changes, dropping support among high news Republicans and Democrats. With relatively small effects such as these, one must be careful to avoid overinterpretation. Still, it does seem plausible that Democrats, who in general were skeptical of the war and thus likely to reject media rationales, might find in an emphasis on troops a reason for enhanced support. It is common to hear that people support the troops but oppose the war. For some, this distinction might in fact be difficult to draw; hearing about the troops might therefore encourage increased support. As for the effect of the freedom rationale, it

500

Communication Research 40(4)

is possible that among Democrats and Republicans who were following the news closely an emphasis on freedom was a regular reminder that a truly democratic Iraq was an elusive goal. Consider that in July 2004 only 22% of the public believed that the new Iraqi government would “be a model which will help spread freedom and democracy in the Middle East” (Bowman, 2008). In general, though, the story of these data is that media rationales during this period had little effect on public support. Even when focusing strictly on narrow subgroups—the most liberal test we might undertake—the statistically significant effects are in most cases substantively modest.

Discussion This study tested a widely held assumption that had largely avoided empirical scrutiny: that rationales for war circulating in mass media influence public support for war. Pairing precise measures of television news content with daily public opinion data during an important yearlong period of the Iraq War, the study’s results run counter to popular wisdom. In a series of regression models, demographic characteristics and casualty rate predicted opinion change in ways consistent with prior research (e.g., Berinsky, 2009; Gartner & Segura, 1998; Jacobson, 2007; Wilcox et al., 1996). Importantly, however, very few of the rationales present in television news were statistically significant predictors of change in levels of public support. What is more, even the strongest predictor among these rationales—troops, which increased support among Democrats and high news-use Independents—only modestly increased the probability that respondents would support the war. Put simply, rationales for war present in television news do not appear to have the substantial impact on public attitudes that some have suggested (e.g., Fritz et al., 2004; Isikoff & Corn, 2006; Rich, 2006) In part, these findings may be the result of the chosen period of analysis. If rationales for war matter for public opinion, it may well be the case that they matter more in the early stages of the conflict than in the later stages. Once the war has progressed for some time, people’s attitudes crystallize, and they tend to interpret new information about the war as supportive of their preexisting attitudes (Gaines et al., 2007). Additionally, because people filter new information through preexisting views, rationales for war are not likely to affect the public in uniform or predictable ways. Recall that in this case an increased news emphasis on freedom, a concept which most people view favorably, decreased support rather than increased it. It is possible, then, that news rationales during Iraq mattered more than they were shown to here, but that these effects were suppressed because different people were reacting in different ways. Splitting out the analysis by party and quantity of news exposure helped control for this, but the possibility remains. Further, if public attitudes about specific rationales were changing at all over the course of the year, these changes could have limited the effects that media emphasis on the rationales was able to generate. As an example, consider the rationale of threat, the only one of the six rationales examined here for which there was consistent polling data throughout this period of the war. In December of 2003, 62% of the public believed that the war had “contributed to the long-term security of the United States.” This number steadily declined over

Coe

501

the course of the year, however, such that by December 2004 only 51% still agreed with the statement (see Bowman, 2008). Thus, the number of individuals likely to increase their level of war support based on hearing the threat rationale was steadily declining during the period of analysis. It is possible that during this time public attitudes about the other rationales were shifting as well, thereby limiting the opportunity for effects to arise. With these caveats in mind, these results must nonetheless be viewed as generally consistent with the “limited effects” paradigm (e.g., Klapper, 1960). As discussed above, the extant literature has come to mixed conclusions about the ability of news content to shift public attitudes through framing. Although it is clear that news frames are capable of influencing attitudes under certain conditions (Kellstedt, 2000; Nelson et al., 1997; Terkildsen & Schnell, 1997), these effects may also be relatively short-term outcomes that are less likely to arise outside of a laboratory (Druckman, 2004; Druckman & Nelson, 2003). This study contributed to this debate, in part, by closely tracking news content and comparing it to survey data—an approach that is still relatively uncommon in studies of media effects, which tend to favor experiments. In this case, the effects of news discourse were indeed limited. It appears that if news coverage influences public support for war, it does not do so primarily or substantially via the rationales it communicates. It may be, as Althaus and Coe (2011) suggest, that the content of wartime news coverage is less important in moving opinion than is the sheer quantity of coverage and the latent patriotism that coverage stirs among the public. It would be inaccurate to conclude from the initial test undertaken here that wartime rationales circulated in news media do not matter. Rather, what we might conclude is that observers must avoid assuming that wartime rationales inevitably influence public opinion. The better approach, which I attempted to take here, is to produce as detailed measures of content as possible and to then compare these measures to opinion data that will allow effects to emerge in those cases when they are indeed present. Baum and Potter (2008), in attempting to synthesize the complex relationships that exist between political leaders, mass media, and public opinion about foreign policy, argue that “the media influence nearly every aspect of the relationship between public opinion and foreign policy” (p. 40). This may be true, but more research that focuses close attention on news content will be needed before we are able to draw this conclusion comfortably. Importantly, because this study relied on secondary analysis of survey data, it is limited in the causal claims it can make. Even if a survey study demonstrates a correlation between two variables, it cannot rule out the possibility that a third variable might explain the relationship. For example, the regression models illustrated a relationship between media emphasis on troops and support for war among Democrats, such that as media emphasis increased so too did support. Based on framing theory, I speculated that this finding might point to a causal relationship: Exposure to media emphasis on troops stirs positive feelings about American soldiers and these feelings translate into increased support for war. This explanation is plausible, but is not demonstrated definitively by the preceding analysis. It is possible that the relationship between media emphasis on troops and increased war support is caused by another variable, such as the occurrence of major battles or the passage of time. Only experimental work can fully demonstrate causal relationships and capture some

502

Communication Research 40(4)

of the attitudinal processes that might be at work among news audiences. All of this study’s findings should be considered with this limitation in mind, and future work should devise experiments that might further illuminate the patterns identified here. Future research in this area should also make a point to carefully measure elite discourse if it hopes to understand its role in shaping public attitudes during wartime. Carefully tracking the actual content of news coverage is what provides the leverage necessary to measure the wide range of effects that may be present. Even if the specific themes present in news coverage turn out not to be a primary determinant of support for war, good measures of content will be necessary to ultimately accept this null hypothesis. Further, working with individual-level opinion data will allow for important subpopulations (for example, the partisans and heavy news viewers considered here) to be examined separately—a crucial approach given the tendency of different individuals to react differently to the same message (Gaines et al., 2007; Klapper, 1960). The tremendous amount of opinion polling done on the Iraq War, along with the growing databases of news content, should provide the data necessary for scholars to adopt these approaches and improve our understanding of the role that elite discourse plays during wartime.

Acknowledgment The author thanks Scott Althaus for sharing the casualty data used in this analysis and for his helpful suggestions.

Declaration of Conflicting Interests The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

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

Notes 1. The NAES data provide a wide range of analysis options. My approach herein is to treat the data as one massive cross-section, which provides me with a large enough sample to test hypotheses about relatively small subsets of the population (high news-use Democrats, low news-use Independents, etc.). This is a useful approach, one anticipated by those who designed the NAES (see Romer et al., 2006). Of course, such an approach is valid only if the assumed independence of the standard errors holds true in the regression model (i.e., if there is no serial correlation). In this case, that assumption does hold true. An autocorrelation plot of the residuals at 16 different lags revealed that all of them fell within the 95% confidence bands around zero. 2. News broadcasts that ran during the 4 Monday-to-Sunday weeks prior to the 52-week period analyzed here (October 6, 2003, through November 2, 2003) were also coded to facilitate measures of 4-week, 3-week, and 2-week lag news coverage used in the analysis of public opinion. Except for these measures, however, analysis of news coverage in this paper focuses just on the 52-week period from November 3, 2003, through Sunday, October 31, 2004.

Coe

503

3. The NAES also included a question asking if respondents approved of “Bush’s handling” of Iraq. Responses to this question were highly correlated with the “worth it” question (r = .70, p < .01). Running the analysis using this question instead of the “worth it” question did not substantively alter the findings. 4. To conduct this analysis I ran the regressions first with the low value and then the high value of each variable of interest, keeping all other variables constant (using the coefficients from Table 1). I then differenced the means of the predicted probabilities produced with the high and low values. This simulation procedure parallels Althaus (2003).

References Althaus, S. L. (2003). Collective preferences in democratic politics: Opinion surveys and the will of the people. New York, NY: Cambridge University Press. Althaus, S. L., & Coe, K. (2011). Priming patriots: Social identity processes and the dynamics of public support for war. Public Opinion Quarterly, 75, 65-88. Althaus, S. L., & Kim, Y. M. (2006). Priming effects in complex information environments: Reassessing the impact of news discourse on presidential approval. Journal of Politics, 68, 960-976. Baker, W. D., & Oneal, J. R. (2001). Patriotism or opinion leadership? The nature and origins of the “rally ’round the flag” effect. Journal of Conflict Resolution, 45, 661-687. Baum, M. A., & Groeling, T. J. (2010). War stories: The causes and consequences of public views of war. Princeton, NJ: Princeton University Press. Baum, M. A., & Potter, P. B. K. (2008). The relationship between mass media, public opinion, and foreign policy: Toward a theoretical synthesis. Annual Review of Political Science, 11, 39-65. Bennett, W. L., Lawrence, R. G., & Livingston, S. (2007). When the press fails: Political power and the news media from Iraq to Katrina. Chicago, IL: University of Chicago Press. Berinsky, A. J. (2009). In time of war: Understanding American public opinion from World War II to Iraq. Chicago, IL: University of Chicago Press. Berinsky, A. J., & Druckman, J. N. (2007). The polls review: Public opinion research and support for the Iraq War. Public Opinion Quarterly, 71, 126-141. Bowman, K. (2008, March 14). Public opinion on the war in Iraq. Retrieved from http://www.aei.org/ publications Clarke, R. A. (2004). Against all enemies: Inside America’s war on terror. New York, NY: Free Press. Coe, K. (2009). Words of war: Presidential rationales for military action from WWII to Iraq. Paper presented at the meeting of the National Communication Association, Chicago, IL. Coe, K. (2011). George W. Bush, television news, and rationales for the Iraq War. Journal of Broadcasting & Electronic Media, 55, 307-324. Domke, D. (2004). God willing?: Political fundamentalism in the White House, the “war on terror,” and the echoing press. London, UK: Pluto Press. Druckman, J. N. (2004). Political preference formation: Competition, deliberation, and the (ir)relevance of framing effects. American Political Science Review, 98, 671-687. Druckman, J. N., & Nelson, K. R. (2003). Framing and deliberation: How citizens’ conversations limit elite influence. American Journal of Political Science, 47, 729-745. Entman, R. M. (2004). Projections of power: Framing the news, public opinion, and U.S. foreign policy. Chicago, IL: University of Chicago Press.

504

Communication Research 40(4)

Fan, D. P. (1993). Media coverage and U.S. public opinion on the Persian Gulf. In B. S. Greenberg & W. Gantz (Eds.), Desert Storm and the mass media (pp. 125-142). Cresskill, NJ: Hampton Press. Feaver, P. D., & Gelpi, C. (2004). Choosing your battles: American civil-military relations and the use of force. Princeton, NJ: Princeton University Press. Fritz, B., Keefer, B., & Nyhan, B. (2004). All the president’s spin: George W. Bush, the media, and the truth. New York, NY: Touchstone. Gaines, B. J., Kuklinski, J. H., Quirk, P. J., Peyton, B., & Verkuilen, J. (2007). Same facts, different interpretations: Partisan motivation and opinion on Iraq. Journal of Politics, 69, 957-974. Gartner, S. S. (2004). Making the international local: The terrorist attack on the USS Cole, local casualties, and media coverage. Political Communication, 21, 139-159. Gartner, S. S., & Segura, G. M. (1998). War, casualties, and public opinion. Journal of Conflict Resolution, 42, 278-300. Gelpi, C., Feaver, P. D., & Reifler, J. (2005). Success matters: Casualty sensitivity and the war in Iraq. International Security, 30(3), 7-46. Hovland, C. I., Lumsdaine, A. A., & Sheffield, F. D. (1949). Experiments on mass communication. Princeton, NJ: Princeton University Press. Isikoff, M., & Corn, D. (2006). Hubris: The inside story of spin, scandal, and the selling of the Iraq War. New York, NY: Crown. Ivie, R. L. (1974). Presidential motives for war. Quarterly Journal of Speech, 60, 337-345. Jacobson, G. C. (2007). A divider, not a uniter: George W. Bush and the American people. New York, NY: Pearson Longman. Jentleson, B. W. (1992). The pretty prudent public: Post post-Vietnam American opinion on the use of military force. International Studies Quarterly, 36(1), 49-73. Jentleson, B. W., & Britton, R. L. (1998). Still pretty prudent: Post-Cold War American public opinion on the use of military force. Journal of Conflict Resolution, 42, 395-417. Kellstedt, P. M. (2000). Media framing and the dynamics of racial policy preferences. American Journal of Political Science, 44, 245-260. Klapper, J. T. (1960). The effects of mass communication. New York, NY: Free Press. Larson, E. V. (1996). Casualties and consensus: The historical role of casualties in domestic support for U.S. military operations. Santa Monica, CA: Rand. Larson, E. V. (2005). American public support for U.S. military operations from Mogadishu to Baghdad. Santa Monica, CA: Rand. Lazarsfeld, P. F., Berelson, B., & Gaudet, H. (1944). The people’s choice: How the voter makes up his mind in a presidential campaign. New York, NY: Duell, Sloan & Pearce. Mueller, J. E. (1973). War, presidents, and public opinion. New York, NY: John Wiley. Nelson, T. E., Clawson, R. A., & Oxley, Z. M. (1997). Media framing of a civil liberties conflict and its effect on tolerance. American Political Science Review, 91, 567-584. Oneal, J. R., Lian, B., & Joyner, J. H., Jr. (1996). Are the American people “pretty prudent”? Public responses to U.S. uses of force, 1950-1988. International Studies Quarterly, 40, 261-279. Project for Excellence in Journalism. (2005). The state of the news media 2005. Retrieved from www. stateofthenewsmedia.com Rich, F. (2006). The greatest story ever sold: The decline and fall of truth from 9/ 11 to Katrina. New York, NY: Penguin.

Coe

505

Romer, D., Kenski, K., Winneg, K., Adasiewicz, C., & Jamieson, K. H. (Eds.). (2006). Capturing campaign dynamics 2000 & 2004: The National Annenberg Election Survey. Philadelphia, PA: University of Pennsylvania Press. Sigelman, L., Lebovic, J., Wilcox, C., & Allsop, D. (1993). As time goes by: Daily opinion change during the Persian Gulf crisis. Political Communication, 10, 353-367. Stryker, J. E., Wray, R. J., Hornik, R. C., & Yanovitzky, I. (2006). Validation of database search terms for content analysis: The case of cancer news coverage. Journalism & Mass Communication Quarterly, 83, 413-430. Terkildsen, N., & Schnell, F. (1997). How media frames move public opinion: An analysis of the women’s movement. Political Research Quarterly, 50, 879-900. West, M. D. (Ed.). (2001). Applications of computer content analysis. Westport, CT: Ablex. Wilcox, C., Hewitt, L., & Allsop, D. (1996). The gender gap in attitudes toward the Gulf War: A crossnational perspective. Journal of Peace Research, 33(1), 67-82. Zaller, J. R. (1992). The nature and origins of mass opinion. New York, NY: Cambridge University Press. Zaller, J. R. (1993). The Converse-McGuire model of attitude change and the Gulf War opinion rally. Political Communication, 10, 369-388.

Author Biography Kevin Coe (PhD 2008, University of Illinois) is assistant professor of communication at the University of Arizona. His research focuses on the interaction of American political discourse, news media, and public opinion. His work has appeared in such journals as Communication Monographs, Journal of Communication, and Political Communication.