17 The longest war story

Transcription

1 The longest war story Elite rhetoric, news coverage, and the war in Afghanistan Tim Groeling and Matthew A. Baum At the time of this writing, the ongoing war in Afghanistan is entering its fourteenth year. This war s unprecedented length makes it a particularly interesting case for scholars studying media coverage of war. In addition to providing tremendous variation in conduct, costs, and outcomes, it sheds unique light on to how wars are fought, as well as sold to citizens both at home and in allied nations in our more partisan twenty- first century political environment. That partisan environment encompasses not only a shift partisan control of government, but also increasingly has applied to the channels through which the public learns about such wars. In this chapter, we examine the actions of the press and political elites with respect to the Afghanistan war in order to assess the competing strategic narratives used by different political actors to describe and justify the conflict. Specifically, we examine () the relative prevalence in the media of narratives supportive of and in conflict with the Bush and Obama administrations narratives regarding the war, () whether more partisan news outlets systematically differed from more traditional outlets in their patterns of coverage, and () how party rhetoric and news coverage regarding the war shifted with the change in presidential administrations in. To answer these questions, we rely on data tracking civilian and military casualties in Afghanistan and other objective measures of the conflict s progress and cost, as well as original content analyses of presidential, congressional, and media narratives regarding the war and news coverage of that rhetoric by various news outlets, including CBS Evening News, Fox News Channel, MSNBC, and the New York Times. Public support for war, and for Afghanistan There are multiple paths through which news coverage could potentially influence policy outcomes. The most widely studied, particularly among democracies, is via its influence on public opinion. Indeed, some scholars (Betz ; Dimitriu ) have argued that the war effort in Afghanistan has been hampered by the inability of the United States to construct a Strategic ch.indd // ::

2 The U.S. s narrative on Afghanistan compelling strategic narrative that could persuade the various key audiences the U.S. public, NATO allies, and Afghan elites to support the mission. Others (Dommersnes ) have noted that different coalition members developed strategic narratives emphasizing different aspects of the mission in order to appeal to the preferences of their domestic populations. Hence, for instance, Norway s preferred strategic narrative characterized their effort as supporting humanitarian initiatives in Afghanistan while Denmark emphasized its role in the fight against terrorism. In order to understand the role of strategic narratives in selling the war, we thus begin by briefly assessing the link between news coverage and public attitudes. The causes and consequences of public support for the overseas application of military force are a subject of long- standing scholarly debate (e.g., Lippmann ; Almond ; Rosenau ; Baum ; Holsti ; Eichenberg ; Howell and Pevehouse ). Some scholars have argued that citizens are fundamentally rational, supporting conflicts they believe are achieving their stated goals at a reasonable cost (Kull and Ramsay ; Feaver and Gelpi ; Eichenberg ; Gelpi, Feaver, and Reifler /), especially in terms of casualties (Mueller ). Others counter that public support for a foreign policy initiative depends upon the mix of elite rhetoric about the policy to which citizens are exposed (Brody ; Baum and Groeling ; Baker and Oneal ; Hetherington and Nelson ), predominantly via the mass media. When citizens observe elites expressing bipartisan support for a policy, they typically respond favorably (Larson, ). In contrast, when citizens observe elites engaging in partisan bickering in the media about the merits of a policy, they tend to choose sides, largely albeit not perfectly along partisan lines. In addition, the indexing school of political communication holds that a mostly passive news media index their coverage of foreign policy events to reflect criticism of the president by partisan elites: if elites stand united behind the president, the press will follow with positive coverage (e.g., Hallin ; Bennett ; Page and Entman ; Zaller and Chiu ). The implication is that the media are, at least most of the time, largely passive and non- strategic (Zaller and Chiu ; Bennett, Lawrence, and Livingston ). Our approach All of these perspectives share in common an assumption that citizens can accurately assess the state of rhetoric or reality, because the media will faithfully report what is happening on the ground and what elites are saying about it. Yet, research on both the traditional (Baum and Groeling ; Groeling and Baum ) and partisan (Baum and Groeling ) media show that this is not necessarily the case. Indeed, building on our Strategic ch.indd // ::

3 T. Groeling and M.A. Baum prior work on news, foreign policy, and public opinion (see also Baum and Groeling ), we argue that in the traditional, or so- called mainstream news media, the news values of journalists combine to accentuate news with certain characteristics (novelty, balance, conflict, and authority) in a way that disproportionately mutes support for the president among his own partisans and rewards dissenters within his own party. The model also predicts that reporters will tend to respond relatively slowly to changes in objective conditions on the ground in a conflict. This effect, in turn, is magnified by the rise of partisan news outlets, which tend to systematically vary from traditional news outlets in their content (Groeling and Kernell ; Baum and Groeling ; Groeling, ). Such outlets are inclined to emphasize the preferred strategic narratives of an ideologically preferred administration, while challenging them when an administration perceived as ideologically hostile is in power. Indexing theories have tended to overlook the fact that journalists are not solely reporters; they are also interpreters. Their interpretations regarding the newsworthiness of different pieces of information in turn color the representation of politics to which citizens are ultimately exposed. To the extent this representation is distorted, so too most likely will be the conclusions citizens draw from it by indexing their opinions to media coverage of elite debate. The implications of such inconsistency are potentially quite consequential for democratic representation. Senator Arthur Vandenberg famously opined that when it comes to foreign policy, politics stops at the water s edge. Our research (see for instance Baum and Groeling ) reveals little evidence supporting this view. Rather, partisan politics has long crossed the water s edge, even during the Cold War. Moreover, the qualities that journalists prefer in news stories result in a strong tendency to overrepresent negative, critical coverage of the president, particularly when it originates within his own party, as well as of support for the president from the opposition party (on the relatively infrequent occasions it is offered). We argue that, at least in the traditional media, these overrepresentations stem not from any partisan preferences of the news media, but rather from pervasive institutional and professional incentives that shape journalists standards of newsworthiness. Unsurprisingly, we further argue that journalists from partisan media outlets differ from their counterparts in the mainstream media, preferring bad news about the other party and good news about their preferred party and generally deemphasizing balance as a news value. The effect is that in partisan media, when the president is from the favored party, journalistic preferences for conflict are subsumed by their goal of supporting a fellow partisan president. This means a prevalence of one- sided good news stories. Conversely, when the other party is in power, partisan media outlets will tend to exceed even traditional journalists in their reporting of bad news about the president and his party, with little Strategic ch.indd // ::

4 The U.S. s narrative on Afghanistan concern for balance (Baum and Groeling ). The implication is that the news preferences we ascribe to the mainstream media are selectively, if at all, applied in the partisan media, thereby accentuating the distortion between the true nature of elite debate and events on the ground and their representation in such media outlets. From these three assumptions, we conclude that journalists preferences shape the representation of elite discourse available to citizens in times of foreign crises. Indeed, they suggest that the media may systematically distort public perceptions of policy debates in Washington by presenting to the public an unrepresentative sample of elite rhetoric. This in turn seems likely to influence the public s propensity to support presidents foreign policies. Nevertheless, policy- makers informational advantage vis- à-vis the public in the realm of foreign affairs nearly always affords them at least some leeway in promoting their preferred strategic narratives regarding foreign policy events. This is particularly true for the president, who, as Commander- in-chief, is the most authoritative and informed information source in matters of foreign policy. It is thus no coincidence that American media primarily index their coverage of foreign affairs to the president and his spokespeople (Cook ; Baum and Groeling ). In contrast, members of Congress, regardless of party, tend to be highly risk averse, typically eschewing public pronouncements on foreign policy issues, at least until the success or failure of a policy is evident (Zaller ). The president s leeway arises to some extent independent of the true nature of such events (see Baum and Groeling for a discussion of this tendency, which we label the Elasticity of Reality ). Yet the extent of this discretion varies over time and with circumstances, typically contracting as the public and the media gather more information. We therefore model media coverage of Afghanistan dynamically, observing trends over time as events unfolded in America s longest war. Data and methods Afghanistan news prominence To track the overall prominence of news coverage regarding Afghanistan, we searched New York Times front pages on Lexis- Nexis for Afghanistan mentions using Nexis geographic tag search. Because the front page of the New York Times is a limited and reasonably consistent unit, the monthly count of these stories should provide a good measure of the centrality of Afghanistan in the national news media. For context, we also provide data over the same time period for front- page stories mentioning Iraq. Figure. shows the distribution of these stories over time. Figure. shows that after the / attacks, Afghanistan dominated the front pages of the New York Times (NYT). At the peak of coverage in Strategic ch.indd // ::

5 Afghanistan Iraq September February July December May October March August January June November April September February July December May October March August January June November April Figure. New York Times front-page stories related to Iraq and Afghanistan, by month. Strategic ch.indd // ::

6 The U.S. s narrative on Afghanistan October, the NYT ran different front- page stories about Afghanistan more than three per day, on average. Over the next six months, the NYT averaged more than one front- page Afghanistan story per day. However, in April, there were only front- page Afghanistan stories, and the NYT would then average less than one per day until November. In contrast to the relatively brief six- month focus on Afghanistan, Figure. shows massive and sustained front- page attention to Iraq. Beginning in September, Iraq dominated the front page of the NYT, peaking with fully front- page stories in April. Next, we examine the degree to which NYT coverage of the two conflicts appeared to be responsive to casualties. In this case, we modified our search to examine coverage throughout the newspaper, instead of just on the front page, but restricted our results to cases where the stories mentioned casualties within words of Iraq or Afghanistan. Figure. shows that the NYT coverage of the Afghanistan conflict mentioned casualties with great frequency in the earliest stages of the conflict (indeed, peaking before any casualties were sustained). As with overall story volume, in the months leading up to the Iraq War, stories addressing Iraq casualties dominates coverage in the NYT, with peaks in coverage seeming to correlate strongly to mass casualty events in Iraq. In Figure., we plot the relative proportion of casualty mentions in Iraq vs. Afghanistan stories in the NYT, showing the degree to which Monthly casualties ( Iraq; Afghanistan) Top is actual; bottom is NYT casualty mentions Figure. Monthly US casualties in Iraq and Afghanistan compared to New York Times stories on Iraq and Afghanistan mentioning casualties. Strategic ch.indd // ::

7 T. Groeling and M.A. Baum Percent September December March June September December March June September December March June September December March June September December March June September December March June September December March June September December September December March Figure. Relative proportion of New York Times mentions of casualties from Iraq (top) vs. Afghanistan (bottom) stories. casualty coverage was dominated by the Iraq War beginning in late and continuing for nearly six years. Only in the lead- up to the presidential election did Afghanistan casualties become comparatively more salient. Partisan rhetoric in the news Of course, while these New York Times searches give a sense of the volume of information potentially available to citizens about Afghanistan at any given time, it does little to indicate the persuasive content or impact of such news. To get a sense of the sort of strategic narratives Americans were seeing about Afghanistan over the course of the conflict, we conducted two content analyses: the first examines elite evaluations of the conflict, while the second tracks the use of competing strategic narratives across media outlets throughout the conflict. Elite evaluations Locating and analyzing every possible evaluation related to the conflict would be impossible. To narrow the task to an achievable dataset, we limited our search along two dimensions. First, we chose to only code the Afghanistan evaluations of sitting U.S. senators. While other prominent March June June September December March June Strategic ch.indd // ::

8 The U.S. s narrative on Afghanistan figures are commonly cited in news stories and rate highly in the public mind, senators tend to be among the most prominent and well- known partisan figures outside the executive branch, especially in the realm of foreign policy. Second, we limit our search for these evaluations to news and public affairs programs aired by CBS News, Fox News, or MSNBC. We chose these outlets because all three had verbatim transcript archives on Lexis- Nexis that covered the full period of the conflict, and because the outlets are commonly perceived as being mainstream, conservative-, and liberalleaning, respectively. We began our searches by locating every story archived by these networks after August that cited a senator and referenced Afghanistan. Coders then read each story to determine whether a senator made a statement related to Afghanistan. If the coder flagged the story as containing a statement related to Afghanistan, the statement was then coded for its content by that coder and blindly by two other coders. A fourth coder then reviewed a mechanically generated comparison of the prior coders work and arbitrated any disputes. Although our research assistants coded a variety of information about each statement, our three key variables were: () [Is the] Afghanistan war going well? (coding choices: no opinion expressed, good/getting better, or bad/getting worse); () Administration/US handling of Afghanistan? (coding choices: no opinion expressed, praise/support, or criticism/ opposition); and () Reduce/withdraw or stay the course/increase [troops]? (coding choices: no opinion expressed, withdraw from Afghanistan or reduce troop levels there, or continue our presence or increase it). Competing strategic narratives In addition to examining specific elite evaluations of the war s progress over time, we analyze the aggregate use of terms fitting competing strategic narratives during the conflict. To do so, we first examined the narratives advanced by supporters and opponents of the war (particularly documents like President Bush s speech to West Point, the NSC Strategy Document, President Obama s speech to West Point, etc.) and then derived search terms that signified the use of these narratives over time. President Bush summarized some of the main parts of the narrative favoring intervention in Afghanistan in his West Point Speech, saying, Our Nation s cause has always been larger than our Nation s defense. We fight, as we always fight, for a just peace a peace that favors liberty. We will defend the peace against the threats from terrorists and tyrants. We will preserve the peace by building good relations among the great powers. And we will extend the peace by encouraging free and open societies on every continent. Strategic ch.indd // ::

9 T. Groeling and M.A. Baum Bush s speech thus placed a heavy emphasis on spreading freedom and liberty while defending against terrorism and other threats. Critics of the administration s narrative advanced their own arguments opposing the initial or continuing U.S. involvement in Afghanistan. These narratives questioned the legitimacy, effectiveness, unintended consequences, costs, or other perils of the war for the United States and the international community. Measures of reality Our analysis also tracks relatively objective measures of the progress of the war in Afghanistan, including measures of total U.S. troops deployed in Afghanistan, improvised explosive devices discovered or detonated in Afghanistan (a proxy for insurgent activity), coalition casualties, and U.S. casualties in Afghanistan. We also gathered overall Gallup and Afghanistan- specific polling data, Afghanistan- related items in the White House Bulletin, and simple controls for congressional majority status, presidential election years, midterm elections, and presidential or vice presidential candidates who were sitting senators. Analysis of elite evaluations To help give the reader a basic understanding of the elite evaluations data, we begin our analysis with Table., which provides descriptive statistics for our key variables. As Table. shows, Fox News aired more senatorial evaluations than CBS and MSNBC combined during the period of our study (September, to June, ). Only percent of the evaluations originated with members of the president s own party (see Groeling for a discussion of how the president tends to shift media attention away from his own party in Congress toward the opposing party). More than half of the evaluations occur when the president s party is in the minority in the Senate. Figure. presents the monthly counts of these evaluations over time. These tallies show an ebb and flow to the senatorial evaluations that strongly parallels that of the New York Times attention to the conflict. The most common senatorial evaluation is support for maintaining or increasing the U.S. force in Afghanistan; in contrast, the least common type of evaluation was a call for force reduction or withdrawal. Criticism of the administration s handling of the war ( percent of all evaluations) was the next most common type of evaluation, while positive evaluations of the administration s handling, and positive or negative evaluations of how well things are going in the conflict each occurred in about percent of evaluations. Strategic ch.indd // ::

10 Table. Descriptive statistics, key variables (n = ) The U.S. s narrative on Afghanistan Variable Mean Std. Dev. Min. Max. CBS (%).. Fox (%).. MSNBC (%).. President s party is source (%).. President s party in Senate minority (%).. Negative evaluation of administration s.. handling of Afghan War (%) Positive evaluation of administration s.. handling of Afghan War (%) Negative evaluation of how well things are.. going in Afghanistan (%) Positive evaluation of how well things are.. going in Afghanistan (%) Evaluation calls for maintaining or.. increasing troop levels in Afghanistan (%) Evaluation calls for reducing or withdrawing.. troops from Afghanistan Afghan IEDs discovered or detonated per.., month Consumer sentiment.... US troops deployed to Afghanistan,.,., Gallup presidential approval.. Monthly coalition casualties in Afghanistan.. Monthly total Afghan-related stories, NYT.. front page Monthly Afghan mentions, White House.. Bulletin Monthly US casualties in Iraq.. Next, Figure. presents the type, tone, and amount of the evaluations occurring during each presidential term. Figure. shows how that in the middle of the conflict, very few evaluations appeared on the news programs. The evaluations that did occur, however, tended to be more critical of the direction of the conflict and the administration s handling of it (although it should be noted that relatively few senators were calling for a troop reduction or withdrawal until Barack Obama assumed office). In Figure., we examine the source of each network s senatorial evaluations. It is immediately obvious from Figure. that with the exception of MSNBC in the Obama administration the presidential party (PP) evaluates the war far less than the non- presidential party (NPP). MSNBC also has a somewhat unusual pattern of sources in Bush s second term, in which they appear to have only featured a single Republican evaluation of the war (versus from Democrats). Strategic ch.indd // ::

11 September February July December May October March August January June November April September February July December May October March August January June November April September February Figure. Total monthly Afghanistan evaluations by senators appearing on CBS, Fox, and MSNBC news shows. Strategic ch.indd // ::

12 Number of senatorial evaluations (%) Going well or getting better, n = Going badly or getting worse, n = Number of senatorial evaluations (%) Admin handling well, n = Admin handling badly, n = Number of senatorial evaluations (%) Troops withdraw or decrease, n = Troops same or increase, n = Senatorial Afghanistan evaluations, by presidential term and getting better/worse Bush Bush Obama Presidential term Bush Bush Obama Senatorial Afghanistan evaluations, by presidential term and handling +/ Presidential term Bush Bush Obama Senatorial Afghanistan evaluations, by presidential term and stay/go Presidential term Figure. Evaluations, by presidential term. Strategic ch.indd // ::

13 Number of senatorial evaluations (%) Fox SrcNPP, n = Fox SrcPP, n = Number of senatorial evaluations (%) CBS SrcNPP, n = CBS SrcPP, n = Number of senatorial evaluations (%) MSNBC SrcNPP, n = MSNBC SrcPP, n = Bush Bush Obama Senatorial Afghanistan evaluations, by presidential term and party Presidential term CBS senatorial Afghanistan evaluations, by presidential term and party Bush Bush Obama Presidential term MSNBC senatorial Afghanistan evaluations, by presidential term and party Bush Bush Obama Presidential term Figure. Party source of Afghan evaluations, by news outlet and presidential term. Strategic ch.indd // ::

14 The U.S. s narrative on Afghanistan Next, in Figure., we start to get a sense of any broad variations in tone across outlets and presidential administrations. The chart shows the relative prevalence of favorable vs. unfavorable evaluations of either the is the situation going well question or the administrative handling question. Fox and CBS show similar patterns (although Fox again has about twice as many evaluations) both show basically even amounts of positive and negative evaluations during Bush s first term, with a big drop in favorability in Bush s second term, followed by an increase in Obama s first term. Somewhat surprisingly, MSNBC is far more favorable in all three presidential terms and appears to have a less positive ratio of evaluations in Obama s first term. In contrast, Figure. shows a much more rapid decline in support for maintaining or increasing troop levels among senators appearing on MSNBC (around percent apparently favor withdrawal in Obama s first term), relative to CBS, or especially Fox, where such support still commands strong majorities. Of course, while this analysis is suggestive, to really understand the impact of editorial choices on which stories are selected, it is important to control for other potentially confounding events or circumstances that might otherwise explain the observed differences in news content. Table. begins this process by testing whether the party differences in the sources each outlet chooses to feature are statistically significant. The unit of analysis for these logits is the individual evaluation, and the dependent variable takes a value of one if the source of that evaluation is of the same party as the president, and zero otherwise. The first two models shown in Table. basically replicate the simple no- control examination we did earlier with our descriptive charts and tables. While the Base model for Bush suggests that all three outlets chose their sources similarly, the Base results for Obama suggest significant variation (with Fox significantly less likely and MSNBC significantly more likely than CBS (the excluded category) to have a Democratic senator delivering their evaluations). When we add a series of controls in columns three and four, the results do not appear to change much, if at all, for either presidential administration. In Table., we conduct a similar analysis for the likelihood each network will air an evaluation praising the administration s handling of the Afghan conflict. In this case, however, we have a couple of different ways to specify our dependent variable. The first assumes that if a senator made any sort of evaluation of the Afghan conflict, they could have used that opportunity to instead praise the administration s handling of the war. In other words, that formulation takes as its population all evaluations, and tries to predict the probability that a particular observation will give a positive assessment of the administration s handling of the war. Models and use this modeling, while models and instead assume Strategic ch.indd // ::

15 Number of senatorial evaluations (%) Fox GoingHandleNeg, n = Fox GoingHandlePos, n = Number of senatorial evaluations (%) CBS GoingHandleNeg, n = CBS GoingHandlePos, n = Number of senatorial evaluations (%) MSNBC GoingHandleNeg, n = MSNBC GoingHandlePos, n = Fox senatorial Afghanistan evaluations, by presidential term and positive and negative Bush Bush Obama Presidential term CBS senatorial Afghanistan evaluations, by presidential term and positive and negative Bush Bush Obama Presidential term MSNBC senatorial Afghanistan evaluations, by presidential term and positive and negative Bush Bush Obama Presidential term Figure. Valence of going well or administration handling evaluations, by news outlet and presidential term. Strategic ch.indd // ::

16 Number of senatorial evaluations (%) Fox TroopDec, n = Fox TroopInc, n = Number of senatorial evaluations (%) CBS TroopDec, n = CBS TroopInc, n = MSNBC TroopDec, n = MSNBC TroopInc, n = Number of senatorial evaluations (%) Fox senatorial Afghanistan evaluations, by presidential term and troops Bush Bush Obama Presidential term CBS senatorial Afghanistan evaluations, by presidential term and troops Bush Bush Obama Presidential term MSNBC senatorial Afghanistan evaluations, by presidential term and troops Bush Bush Obama Presidential term Figure. Senatorial support for troop increases or decreases, by news outlet and presidential term. Strategic ch.indd // ::

17 Table. Logit analysis of probability of selecting presidential (versus non-presidential) senators to evaluate Afghanistan Base model Controlled model Bush Obama Bush Obama n = Pseudo r =. n = Pseudo r =. n = Pseudo r =. n = Pseudo r =. Fox. (.). (.)**. (.). (.)* MSNBC. (.). (.)**. (.). (.)** IEDs. (.)****. (.) Consumer sentiment. (.)*. (.) US troops deployed to Afghanistan. (.). (.)**** Gallup presidential approval. (.)****. (.) Monthly coalition casualties in Afghanistan. (.). (.). (.). (.) Monthly total Afghanistan-related stories, NYT front page President s party in Senate minority. (.)****. (.)**** Monthly Afghan mentions, White House Bulletin. (.)**. (.) Monthly US casualties in Iraq. (.). (.) Constant. (.)**. (.). (.)**. (.) Notes * p.. ** p.. *** p.. **** p.. Strategic ch.indd // ::

18 Table. Logit analysis of the probability of airing a senator praising the administration s handling of the Afghan conflict Model (population = all evaluations) Model (population = any handling evaluation) Bush Obama Bush Obama n = Pseudo r =. n = Pseudo r = n = Pseudo r =. n = Pseudo r = Fox. (.). (.). (.). (.) MSNBC. (.). (.)*. (.). (.)* Constant. (.)***. (.)***. (.). (.)**** Model (population = all evaluations) Model (population = any handling evaluation) Bush Obama Bush Obama n = Pseudo r =. n = Pseudo r =. n = Pseudo r =. n = Pseudo r =. Fox. (.)^. (.). (.)*. (.) MSNBC. (.). (.). (.). (.)^ IEDs. (.)****. (.)**. (.)****. (.)** Consumer sentiment. (.)****. (.). (.). (.) US troops deployed to Afghanistan. (.). (.)****. (.)****. (.) Gallup presidential approval. (.). (.)****. (.). (.) Monthly coalition casualties in Afghanistan. (.)**. (.)***. (.)**. (.)*. (.). (.). (.). (.) Monthly total Afghanistan-related stories, NYT front page President s party in Senate minority. (.). (.). (.). (.) Monthly Afghan mentions, White House. (.). (.). (.). (.) Bulletin Monthly US casualties in Iraq. (.)****. (.). (.). (.)**** Constant. (.). (.)****. (.). (.) Notes * p.. ** p.. *** p.. **** p.; Strategic ch.indd // ::

19 T. Groeling and M.A. Baum that we should only look at evaluations of the administration s handling (positive or negative) and try to predict which of those will be positive. Looking at the simple models first, we see that for both Model (which looks at all evaluations) and model (which just looks at handling evaluations), there appears to be no significant difference in evaluations across networks for Bush. For Obama, however, the simple models suggest MSNBC is significantly more likely to air an evaluation praising the administration s handling of Afghanistan. Adding in a full set of controls in models and makes this conclusion somewhat less clear, however. With the added controls, Fox s decreased likelihood to air praise of the administration (which had been statistically insignificant for Bush in models and ) is larger and more significant. In contrast, MSNBC s predicted likelihood of airing administration praise (while still positively signed) has dropped to insignificance in Model s full population model, and p <. significance for Model s handling evaluations. Table. repeats this type of analysis for the evaluations for how the situation is going in Afghanistan (with the dependent variable = when the senator answers favorably). Arguably, this evaluation (which isn t asking for an evaluation of a partisan figure like handling or a preference in a contentious policy like the troop levels question) should more closely reflect the real situation on the ground in Afghanistan, versus the preferences of the speaker or news outlet. While suggestive, the marginal significance of the news outlets in the equation suggests that there isn t a clear linkage between the outlet s presumed preferences and their guests assessments of the Afghan conflict. In the final step of this part of the analysis, Table. looks for differences in the probability of calling for maintaining or increasing troop levels in Afghanistan (compared to making any other observation (models and ) or just making an evaluation about troop levels (models and )). Models and show no significant differences across outlets for the Bush administration, but then large and significant differences between the networks under Obama. These differences, in which Fox is significantly more likely to support maintaining or increasing troops, while MSNBC is significantly less likely to do so, continue or even strengthen in some cases once we add controls in models and. Analysis of competing strategic narratives We begin our analysis of competing strategic narratives with Figure., which shows the aggregate- level prevalence of oppositional (dark bars) and supportive (light bars) narrative frames in the Afghanistan coverage of each of our networks (CBS, MSNBC, and Fox News, respectively). Each bar represents the proportion of total Afghanistan stories for each network. The black line (with its own scale on the right side of the chart) Strategic ch.indd // ::

20 Table. Logit analysis of probability of airing a senator praising the way things are going in Afghanistan Model (population = all evaluations) Model (population = any going evaluation) Bush Obama Bush Obama n = Pseudo r =. n = Pseudo r =. n = Pseudo r =. n = Pseudo r = Fox. (.). (.). (.). (.) MSNBC. (.). (.). (.)^. (.) Constant. (.)***. (.)***. (.). (.) Model (population = all evaluations) Model (population = any going evaluation) Bush Obama Bush Obama n = Pseudo r =. n = Pseudo r =. n = Pseudo r =. n = Pseudo r =. Fox. (.). (.)****. (.). (.) MSNBC. (.)****. (.). (.). (.) IEDs. (.)**** (.). (.)****. (.)* Consumer sentiment. (.). (.). (.). (.) US troops deployed to Afghanistan. (.)****. (.). (.)*. (.) Gallup presidential approval. (.). (.). (.)****. (.) Monthly coalition casualties in Afghanistan. (.)*. (.). (.)*. (.). (.). (.). (.). (.) Monthly total Afghanistan-related stories, NYT front page President s party in Senate minority. (.)****. (.). (.). (.). (.). (.). (.). (.) Monthly Afghan mentions, White House Bulletin Monthly US casualties in Iraq. (.). (.). (.). (.) Constant. (.). (.). (.)****. (.) Notes * p.. ** p.. *** p.. **** p.. Strategic ch.indd // ::

21 Table. Logit analysis of probability of airing a senator calling for troop levels in Afghanistan to be maintained or increased Model (population = all evaluations) Model (population = troop level evaluations) Bush Obama Bush Obama n = Pseudo r =. n = Pseudo r =. n = Pseudo r =. n = Pseudo r = Fox. (.). (.)**. (.). (.)**** MSNBC. (.). (.)**. (.). (.)** Constant. (.). (.)*. (.)***. (.)**** Model (population = all evaluations) Model (population = troop level evaluations) Bush Obama Bush Obama n = Pseudo r =. n = Pseudo r =. n = Pseudo r =. n = Pseudo r =. Fox. (.). (.)*. (.). (.)** MSNBC. (.). (.)**. (.). (.)* IEDs. (.)*. (.)****. (.). (.) Consumer sentiment. (.)****. (.)****. (.)****. (.) US troops deployed to Afghanistan. (.). ()*. (.)****. (.) Gallup presidential approval. (.). (.)**. (.)^. (.) Monthly coalition casualties in Afghanistan. (.). (.). (.). (.) Monthly total Afghanistan-related stories,. (.). (.). (.)****. (.)**** NYT front page President s party in Senate minority. (.). (.) Dropped. (.)****. (.). (.). (.)****. (.) Monthly Afghan mentions, White House Bulletin Monthly US casualties in Iraq. (.)*. (.)****. (.)****. (.)**** Constant. (.). (.)****. (.)****. (.) Notes Non-presidential Senate majority predicted success perfectly in observations and was dropped. * p.. ** p.. *** p.. **** p.. Strategic ch.indd // ::

22 Percentage of stories with supportive or oppositional frame CBS,,,,,, n =, n =, n = n = n = n = n = n = n = n = n = n = Oppositional.%.%.%.%.%.%.%.%.%.%.%.% Supportive.%.%.%.%.%.%.%.%.%.%.%.% Total stories Percentage of stories with supportive or oppositional frame n = n = n = n = n = n = n = n = n = n = n = n = Oppositional.%.% Supportive % % Total stories Percentage of stories with supportive or oppositional frame.% %.% %.%.% MSNBC.% %.% %.% %.% %.%.%.%.%.%.% n = n = n = n = n = n = n = n = n = n = n = n = Oppositional.%.% Supportive.%.% Total stories.%.%.%.%.%.% Fox News.%.%.%.%.%.%.%.%.%.%.%.%.%.%,,,,,,,,,,,, Figure. Total Afghanistan stories and relative proportion of supportive and oppositional strategic narrative frames, by network. Total number of stories Total number of stories Total number of stories Strategic ch.indd // ::

23 T. Groeling and M.A. Baum shows the total number of Afghanistan stories on the outlet. Note that, as discussed earlier, total Afghanistan coverage varies substantially across networks and years. Beginning with CBS, we initially see high volumes of coverage with a major emphasis on narratives supportive of the Bush administration s preferred strategic narrative regarding the war. While CBS volume of coverage drops sharply after, President Bush s first term is largely characterized by high frequency of supportive frames and very low frequency of opposition frames. The narrative frames on CBS become somewhat less favorable in Bush s second term with more opposing frames and fewer supportive frames but the coverage still remains generally supportive. Interestingly, Barack Obama s first term in office received the lowest level of coverage to date on CBS, but a sharp increase in both supportive and oppositional frames in coverage of Afghanistan. MSNBC s coverage differs starkly from CBS, with relatively few stories aired on Afghanistan (particularly in Bush s second term), but continually high levels of supportive and oppositional frames within the network s relatively modest level of coverage. This result seems consistent with prior studies of MSNBC that highlight its disproportionate emphasis on opinionated (versus straight news) coverage of events (Jurkowitz et al. ). Finally, for Fox News, both the level of coverage and use of narrative frames falls in between MSNBC and CBS, particularly during the Bush administration. In contrast, they offered their highest volume of Afghanistan coverage during the first Obama term. During this same period, they featured more oppositional narrative frames and fewer supportive frames. Finally, in Table. we disaggregate the terms that were most often associated with supportive and oppositional coverage of Afghanistan across CBS, Fox News, and MSNBC. As one might expect, the most common terms used in discussing Afghanistan following the / attacks in Bush s first term are all directly linked to terrorism or the Taliban regime. Remarkably, over percent of all MSNBC stories about Afghanistan during this four- year period invoke terror or terrorism, and about percent also invoke al- Qaeda or Osama bin Laden. Freedom and liberty were invoked far less frequently, particularly on CBS, and allied support and WMD- related issues were invoked even less often. Oppositional rhetoric in the first Bush term focused on the comparatively difficult time the Soviet Union faced in the aftermath of their invasion, protests, and lack of domestic and international support for the invasion, Afghan and tribal politics, and concerns about how to withdraw or get out of the conflict. Especially on CBS and Fox, this rhetoric is relatively uncommon when discussing Afghanistan during President Bush s first term. During Bush s second term, overall levels of coverage drop sharply, but the linking of Afghanistan to terror continues in support of the administration s strategic narrative. Fox and MSNBC also begin to place Strategic ch.indd // ::

24 The U.S. s narrative on Afghanistan comparatively higher emphasis on freedom, democracy, and safety in Afghanistan during this time. In terms of oppositional frames, the links to the Soviet experience in Afghanistan were less prominent, with greater emphasis on calls for withdrawal and getting out of Afghanistan particularly on MSNBC (consistent with our prior elite rhetoric results). Finally, during President Obama s first term in office, coverage continues to emphasize the Taliban and al- Qaeda. However, almost half of all stories on MSNBC and CBS now raise the issue of ensuring safety in Afghanistan. In the meantime, getting out/withdrawal continues to be a focus of oppositional narratives, with questions about Afghan and tribal concerns also becoming more prominent. Conclusion During the presidency of George W. Bush, it was typical to find commentators bemoaning the unprecedented partisanship of national politics and the news. Setting aside the question of whether such partisan rancor was truly historically aberrant, if the analysis above is correct it seems clear that the Bush years were a metaphorical walk in the park compared to the partisan divisions exposed during the Obama presidency. In fact, we confess to being surprised at the similarity in elite rhetoric across disparate outlets in the Bush years (especially early in the war), given the large differences in coverage we have previously observed in contemporaneous coverage of the war in Iraq. Regardless of the cause, it appears that any Era of Good Feeling in the Afghanistan conflict is now behind us as partisan (and news) polarization steadily increases under President Obama. These results suggest that partisan differences in news coverage do, in fact, appear to depend on the partisan political environment. When partisan elites were mostly in agreement in support of the war in Afghanistan or at least in their general disinterest in it news coverage reflected this relative elite harmony. This characterized much of the Bush administration, particularly following the start of the Iraq War in March. However, once President Obama assumed office, partisan discord among elites in Washington began to increase. This discord was clearly evident in news coverage, with conservative- leaning Fox News significantly less likely than (relatively centrist) CBS, and liberal- leaning MSNBC significantly more likely than CBS, to air reports supportive of the Obama administration s policies in Afghanistan. The opposite differences emerged during the Obama administration with respect to the merits of maintaining or escalating troop levels in Afghanistan, with Fox more and MSNBC less supportive than CBS. Finally, during the Obama administration, but not during the Bush administration, Fox was significantly less likely than MSNBC to feature Democratic senators evaluations of the war and significantly more likely to feature Republican war- related evaluations. In each instance, these results are Strategic ch.indd // ::

25 Table. Analysis of top ten competing (supportive vs. oppositional) strategic narratives in Afghan coverage of CBS, MSNBC, and Fox News (%) First term, George W. Bush CBS (n =,) Fox News (n =,) MSNBC (n = ) Supportive Oppositional Supportive Oppositional Supportive Oppositional. Taliban. Afghan. terror. Soviet. terror. get out. bin Laden. Soviet. Qaeda. get out. Qaeda. Soviet. Qaeda. tribal. bin Laden. Afghan. bin Laden. occur. terror. protest. Taliban. tribal. threat. Afghan. threat. get out. threat. protest. free. protest. safe. civilian casualties. safe. occur. Taliban. tribal. free. occur. free. withdraw. safe. withdraw. allies. safe haven. allies. civilian casualties. allies. international law. camps. withdraw. WMD. safe haven. WMD. safe haven. WMD. opium. camps. int l law. democrac. civilian casualties Second term, George W. Bush CBS (n =,) Fox (n = ) MSNBC (n = ) Supportive Oppositional Supportive Oppositional Supportive Oppositional. terror. Afghan. terror. get out. terror. get out. Qaeda. protest. Qaeda. protest. free. withdraw. Taliban. tribal. threat. withdraw. safe. protest. bin Laden. get out. free. Afghan. Qaeda. Soviet. NATO. withdraw. Taliban. tribal. threat. occur Strategic ch.indd // ::

26 . safe. safe haven. bin Laden. Soviet. bin Laden. tribal. threat. Soviet. safe. safe haven. democrac. neoconservative. free. opium. democrac. occur. Taliban. Afghan. democrac. poppy. NATO. alienate. allies. safe haven. allies. occur. allies. international law. defeat. alienate First term, Barack Obama CBS (n = ) Fox (n =,) MSNBC (n = ) Supportive Oppositional Supportive Oppositional Supportive Oppositional. Taliban. protest. Taliban. get out. Qaeda. get out. safe. get out. Qaeda. protest. Taliban. Afghan. free. Afghan. terror. Afghan. threat. protest. threat. withdraw. free. withdraw. safe. Soviet. Qaeda. safe haven. threat. Soviet. free. withdraw. terror. occur. safe. safe haven. terror. tribal. NATO. tribal. NATO. tribal. bin Laden. safe haven. allies. Soviet. allies. occur. NATO. occur. bin Laden. civilian casualties. bin Laden. civilian casualties. democrac. nation building. defeat. opium. defeat. alienate. defeat. alienate Strategic ch.indd // ::

27 T. Groeling and M.A. Baum consistent with the media indexing. They suggest that to some extent the media s partisan gloves only come off upon receiving elite cues that a given issue falls within the boundaries of acceptable partisan warfare. While it has become common to refer to Afghanistan as the forgotten war, we were surprised to find that in one- third of the months we examined during the Bush and Obama administrations, none of the news outlets we investigated offered a single senatorial evaluation of the war. Indeed, it is arguably appropriate to characterize public opinion on the Afghanistan War throughout most of its duration as a mile wide and an inch deep. That is, many people expressed support or opposition when asked in opinion polls which were themselves relatively few and far between from mid- through yet few were paying much attention to Afghanistan, as all eyes were on Iraq for the duration of that latter conflict. Consequently, it is perhaps a stretch to think of strategic narratives as playing an important role in influencing public attitudes toward Afghanistan, at least in the United States. To the extent the Bush administration deployed such narratives whether subsequently contested or not the pervasive lack of public attention placed a stark ceiling on the ultimate efficacy of such efforts, especially during the Bush administration s second term. Notes Search terms for the Afghanistan search were ((pg. AND SECTION(A)) and (#GC#)) [the Nexis country code for Afghanistan]. The Iraq search terms were ((pg. and section (A)) and (#GC#)) [the Nexis country code for Iraq]. So, for example, the Iraq search became ((iraq w/ (casualty or casualties))) and ((#GC#)). Moreover, it is commonly perceived that MSNBC s shift toward a more liberal perspective and brand occurred during the period of our study. While the exact timing of the shift is hard to define, by late the New York Times discussed it in a story entitled Cable channel nods to ratings and leans left (Steinberg ). Our coders evaluated over, records. We assigned stories to coders in random- order packets to ensure no single coder was solely responsible for a particular range of stories. To be included in our analysis, the senator s statement either had to be an actual soundbite delivered by the senator, or another verbatim source, such as a letter or other form of direct attributed quotation. We are grateful for the hard work provided by Noah Grand and our outstanding team of coders who continued hand- coding these evaluations throughout the summer. Does the Senator make a statement that indicates s/he thinks the situation in Afghanistan is going well/better or poorly/worse? Note that the options are not mutually- exclusive (except no opinion vs. the expressed opinions, obviously: Make sure you turn off no opinion below whenever one of the other items applies for that evaluation). Does the Senator praise the then- current president or administration s handling of any aspect of the Afghanistan conflict? [Note: exclude praise or criticism specifically directed at soldiers, troops, etc., rather than their elected leadership]. Strategic ch.indd // ::

28 The U.S. s narrative on Afghanistan Does the Senator reach any explicit conclusions about whether the U.S. should reduce troop strength or withdraw from Afghanistan, versus staying the same or increasing our troops there? Our Lexis- Nexis search for rhetoric consistent with this strategic narrative was as follows: (((((free* or democrac* or defuse or defeat or pre- empt or preempt or wmd or (weapons of mass destruction) or NATO or taliban or qaeda or (bin laden) or camps or terror* or abandon or threat or evildoers or (evil doer*) or stabil* or safe* or allies or (defend the peace) or liberty or responsib* or extremist or islamist or (islamic fundamentalism)) and ((#GC#))) and [date range]). Note that #GC# is the Lexis- Nexis geographical code for Afghanistan. Our Lexis- Nexis search for the presence of these critical narratives was (((((((protest* or unilateral or alienate* or (international law) or blowback or (war of aggression) or neoconservative or realist or (safe haven) or opium or poppy or withdraw* or (eyes off the ball) or occur* or (nation building) or tribal or hatr* or (civilian casualt*) or soviet or (get out) or (long history))) and ((#GC#)) and Date. From through November ; remainder from Figure. in www. brookings.edu/~/media/files/programs/fp/afghanistan%index/index.pdf. Data adapted from figure. from Programs/FP/afghanistan%index/index.pdf and figure. in pdf. Available at: Available at: names.xls. Note that data on Afghan opium production, school enrollment, civilian casualties, and other suitable data were either unavailable in a continuous series or in density sufficient for our analysis. Somewhat surprisingly, only of our evaluations come from John Edwards, John Kerry, Barack Obama, Joe Biden, or John McCain in their respective presidential election years. Afghanistan data courtesy Gary Jacobson. Note that we are unable to tell whether these differences reflect real changes or possible variations in the quality of the Lexis- Nexis holdings for each network over time. As a result, we will spend most of this section discussing the relative prevalence of different narratives within the search results available for each network. While senators are clearly among the most consequential partisan figures in national politics outside the White House, it is possible that our missing months problem might be exacerbated by excluding other potential evaluators. Also, because we lack a measure of the true unmediated population of potential soundbites available to CBS, Fox, and MSNBC, we are unable to resolve the socalled unobserved population problem (Groeling and Kernell ; Groeling ). That is, establishing whether a news organization is presenting a biased selection of news requires some measure of the full population of news from which those stories were selected. Despite our best efforts to gather such an indicator, we were unable to do so to our satisfaction. References Almond, Gabriel A. (). The American people and foreign policy. New York: Praeger. Baker, William D. and John R. Oneal. (). Patriotism or opinion leadership? Strategic ch.indd // ::