Circling the Wagons: Soft News and Isolationism in American Public Opinion

This study investigates the differences in coverage of foreign policy by the soft and hard news media, and the implications of such differences for public attitudes regarding the appropriate U.S. role in the world. I find that, relative to traditional news outlets, the soft news media place greater emphasis on dramatic, human-interest themes and episodic frames and less emphasis on knowledgeable information sources or thematic frames, while also having a greater propensity to emphasize the potential for bad outcomes. I then develop a conceptual framework in order to determine the implications of these differences. I argue that the style of coverage of soft news outlets tends to induce suspicion and distrust of a proactive or internationalist approach to U.S. foreign policy, particularly among the least politically attentive segments of the public. I test this and several related hypotheses through multiple statistical investigations into the effects of soft news coverage on attitudes toward isolationism in general, and U.S. policy regarding the Bosnian Civil War in particular. I find that among the least politically attentive members of the public, but not their more-attentive counterparts, soft news exposure—but not exposure to traditional news sources—is indeed associated with greater isolationism in general, and opposition to a proactive U.S. policy toward Bosnia in particular.

On August 20, 1998, just three days after testifying about his relationship with Monica Lewinsky, President Bill Clinton ordered cruise missile strikes against suspected terrorist sites in Afghanistan and Sudan. The missile strikes began at 1:30 p.m., EST. By that evening, before newspapers could appear the next morning, almost three quarters of the public knew about them.¹

Over the next week the three major networks’ evening newscasts covered the story intensely, offering 69 stories on the subject.² Surprisingly, so did such entertainment-oriented “soft news” programs as Entertainment Tonight (“ET”), Access Hollywood, Extra, and The Tonight Show.

Many people learned about the missile strikes from network newscasts, which typically attract about 30 million viewers per day. Others probably learned about them from local TV news or all-news cable networks. Yet far more viewers regularly watch soft news programs than all of the all-news cable networks combined, while about as many, and possibly even more, watch soft news shows as watch the nightly network newscasts (Baum, 2003b). This suggests that a typical television viewer may have been about as likely to encounter the missile strike story on soft news shows as on the network news.³

Why should it matter where people learned about the missile strikes? The answer is that these different venues presented starkly different portrayals of events. Whereas network news coverage focused primarily on military tactics and the political ramifications of the attacks, the soft news media focused almost exclusively on the uncanny parallels between real-world events and a relatively obscure (until then) movie, called “Wag the Dog.” In the film, a fictional president hires a Hollywood producer to “produce” a phony war in order to distract the public from a sex scandal.

A review of transcripts from 12 soft news programs revealed that in the week following the attacks, 35 out of 46 soft news stories on the subject (or 76%) addressed the “Wag the Dog” theme, repeatedly raising the possibility that the president may have launched the missile strikes in order to distract the nation from the Lewinsky scandal. During the same period, the three network evening newscasts, combined, mentioned “Wag the Dog” or “Monica Lewinsky” in only 11 of 69 stories on the missile strikes (or 16%).⁴

In post-attack opinion polls, as many as 40% of respondents—including 25% of self-described Democrats—indicated they believed distracting the nation was one of the considerations motivating President Clinton.⁵ These suspicions were most widespread among less educated Americans—among the primary consumers of soft news (Davis and Owen, 1998; Baum, 2002, 2003b). In one post-attack survey (Star Tribune, 1998), respondents with less than a 12th grade education were nearly twice as likely as their college-educated counterparts (60% vs. 31%) to believe that the president's decision to order the missile strikes was influenced “a great deal” by his political problems stemming from the Lewinsky scandal.

Baum (2002) finds that despite their entertainment orientation, the soft news media provide substantial coverage of U.S foreign policy crises. As a consequence, segments of the American public that traditionally tuned out politics entirely are exposed to information about such issues as an incidental by-product of seeking entertainment. The Afghanistan-Sudan anecdote, in turn, suggests that the soft news media may cover foreign crises quite differently than traditional news outlets. This raises the possibility that exposure to information about foreign policy via the soft news media may influence viewers in very different ways than exposure via traditional news outlets.

Additional research (e.g., Key, 1961; Iyengar and Kinder, 1987) has shown that the nature of the information people consume can influence both the extent (i.e., breadth) and form (i.e., valence and diversity) of their opinions. Such effects may influence outcomes at the ballot box, as well as alter a president's calculus concerning the likelihood of sustaining public support for his policies. This study investigates whether the soft and hard news media do, in fact, differ systematically in the manner in which they cover foreign crises, as well as the implications of any such differences for public attitudes regarding foreign policy.

In the next section, I introduce a conceptual framework to explain why different types of individuals might respond differently to information about foreign policy presented in the soft news media. I then present the results of a content analysis of soft and hard news coverage of recent U.S. military interventions. Based on the findings from my content analysis, I derive several hypotheses concerning the relationship between soft news consumption, political engagement, and the propensity to support an internationalist approach to American foreign policy. Next, I present the results of a series of statistical tests of my hypotheses. I find that the least politically attentive members of the public who consume large amounts of soft news are less likely than their more-politically-attentive or non-soft-news-consuming counterparts to support an internationalist U.S. foreign policy. In the final section, I discuss my findings and offer conclusions.

Soft News and Public Attitudes: A Conceptual Framework

Borrowing from Patterson (2000), I define “soft news” as a set of story characteristics, including sensationalized presentation, human-interest themes, emphasis on dramatic subject matter and the absence of a public policy component. Elsewhere (Baum, 2002), in turn, I refer to those media outlets that focus primarily on such material—including entertainment and tabloid news shows, network newsmagazines, and daytime and late-night talk shows—as the soft news media. These outlets differ in many respects. Yet they all focus primarily on soft news themes, and they all tend to cover similar types of political and foreign policy issues in similar manners (Baum, 2002, 2003b). Their audiences also tend to have comparatively little education or interest in politics (Baum, 2003b; Hamilton, 2003).

To help explain how exposure to soft and hard news coverage of foreign policy is likely to influence the opinions of different types of consumers, I develop a theoretical framework based upon a two-step reception-acceptance model of public opinion change (McGuire, 1968; Zaller, 1992). Zaller (1992) argues that individuals at different levels of political awareness differ in their propensities to be exposed to political information (step one), as well as in their responses, given exposure (step two). The least aware individuals tend not to expose themselves to political information. Hence, they frequently have no basis upon which to alter their opinions. The opinions of highly politically aware individuals are also relatively stable, but for a different reason. Political sophisticates have well-formed belief systems that allow them to resist accepting dissonant messages, even if they receive them.

I refer to this latter propensity to reject information inconsistent with pre-existing attitudes, given exposure, as selective acceptance. The reception-acceptance model suggests that because they possess a broad range of political attitudes, highly politically aware individuals are better equipped to employ selective acceptance, given message reception, than their less aware counterparts, who possess fewer political attitudes upon which to draw. Moreover, because the soft news media are in the business of making information accessible to relatively apolitical audiences (Baum, 2002), low-awareness individuals would appear more likely to be exposed to a political story in the first instance, and hence to accept it, if it is covered by a soft news outlet.

According to the reception-acceptance model, politically inattentive individuals are unlikely to tune in to traditional political news, and so are unlikely to be exposed to hard news stories about foreign affairs. Even if exposed, they are less likely than political sophisticates to attend to or understand the relatively more complex information presented by such shows (Hamill and Lodge, 1986; Hamilton, 2003). It follows that exposure to coverage of foreign affairs via the soft news media is likely to have a stronger effect on the attitudes of inattentive individuals than exposure through traditional news.

Yet depending on the tenor of soft news coverage, and the pre-existing attitudes of low-awareness consumers, the net effect of such exposure could be to raise or lower support for an internationalist U.S. foreign policy. Hence, before predicting how exposure to soft or hard news will influence public attitudes, we must first consider whether and how the soft and hard news media are likely to differ in their coverage of foreign policy. Elsewhere (Baum, 2002), I argue that in order to appeal to their largely apolitical audiences, soft news outlets tend to cover only the most dramatic and accessible aspects of foreign crises. One implication is that they will tend to rely disproportionately on episodic frames (which focus on the experiences of specific individuals or small groups), rather than thematic frames (which place stories into a broader social, economic, or political context). This is because episodic frames tend to be more compelling and accessible to politically inattentive viewers (Iyengar, 1991).

Additionally, far more than traditional news media, soft news outlets tend to pass moral judgment on the subjects of their reporting, usually in the context of stories involving intense human drama (Media Monitor, 1997). It follows that soft news outlets are also more likely to focus on the circumstances that tend to give rise to human drama—most notably danger, violence, and tragedy—and to do so in stark, moralistic terms. Whether intended or not, a prevalence of episodic stories of individuals facing grave dangers while struggling against “evil” adversaries seems likely to make the world appear an inhospitable place. Indeed, one unintended consequence of soft news programming strategies geared toward entertaining mostly apolitical audiences is therefore likely to be a tendency to emphasize negative or pessimistic aspects of U.S. foreign policy.

Soft vs. Hard News Coverage of U.S. Military Interventions

To determine whether and how soft and hard news outlets differ in covering foreign crises, I, along with several research assistants, reviewed a randomly selected sample of 38 “soft” and 55 “hard” news reports pertaining to a series of U.S. military conflicts in the 1990s.⁶ The key results from my content analyses are summarized in Table 1.⁷ (See Appendix 2 for a summary of the data and the coding form.)

1

Content Analysis of Soft vs. Hard News Coverage of U.S. Military Operations, 1992–1999

	Soft News	Hard News	Soft News Difference
Use of Episodic and Thematic Frames
1.Probability report discussed origins of conflict.	.08	.44	−.36***
2.Probability origins of conflict described as beginning prior to U.S. involvement.	.00	.24	−.24**
3.Probability report mentions a country, faction or group as an adversary/antagonist in conflict.	.21	.55	−.34*
4.Probability report featured an individual, family or small group(s).	.95	.69	+.26***
5.Probability report featured the issue or conflict as a whole or the circumstances surrounding the conflict.	.16	.69	−.53***
Cues from Credible or Non-Credible Sources
6.On average, how many members of president's party sourced per-report?	.16	.60	−.44***
7.On average, how many members of opposition party sourced per-report?	.05	.09	−.04
8.On average, how many experts (including senior military officers) sourced per-report?	.45	.78	−.33^
9.On average, how many celebrities sourced per-report?	.77	.00	+.77**
Support for U.S. Decision-Makers
10.Probability report includes comment by celebrity critical of U.S. decision-makers?	.30	.00	+.30***
11.Probability report includes comment by celebrity supportive of U.S. decision-makers?	.18	.00	+.18
12.Probability report includes comment by member of president's party supportive of U.S. decision-makers?	.11	.24	−.13^
13.Probability report suggests conflict has good chance of being resolved through U.S. action.	.00	.11	−.11**
Comparisons to U.S. Foreign Policy Failures
14.Probability report mentioned a past U.S. foreign policy crisis/conflict.	.45	.25	+.20^^
15.On average, how many of past foreign policy crises/conflicts mentioned per-report represented clear U.S. foreign policy failures?	.45	.15	+.30^^
16.On average, how many comparisons per-report to Vietnam War?	.16	.00	+.16**

^

p<.10,

^^

p<.07,

*

p<.05,

**

p<.01,

***

***p<.001

The first set of questions address the prevalence of episodic and thematic frames in soft and hard news coverage of U.S. military interventions. The results indicate that, compared to traditional national newscasts (ABC, CBS, NBC, and CNN), soft news outlets are far less likely to employ thematic frames, such as describing the origins of the conflict (.08 vs. .44, p<.001), noting whether the conflict began prior to U.S. involvement (.00 vs. .24, p<.001), or identifying the countries and groups involved in the conflict (.21 vs. .55, p<.05).⁸ More generally, soft news reports are far less likely (.16 vs. .69, p<.001) to feature discussion of the conflict as a whole, or the circumstances surrounding it. The exception concerns drawing analogies to past U.S. foreign policy experiences, which is far more common in the soft news media (.45 vs. .25, p<.06). Conversely, while episodic frames appear in 69 percent of the hard news reports, they are virtually ubiquitous in soft news coverage; ninety-five percent of all soft news reports employed episodic frames. This difference is significant at the .001 level.

In referencing past U.S. military activities, the soft news reports were three times as likely as the hard news reports to draw comparisons between the current crisis and past U.S. foreign policy failures, such as terrorist attacks against U.S. interests or the Vietnam War (.45 vs. .15, p<.05).⁹ In fact, 16 percent of the soft news reports—representing nearly one-fifth of all soft news mentions of past U.S. conflicts—drew explicit comparisons to Vietnam. In contrast, Vietnam was never mentioned in the 55 traditional news broadcasts. Not surprisingly, this difference is significant at the .01 level.

Additionally, not once did a soft news outlet suggest or imply that the issue or dispute that prompted U.S. involvement was likely to be successfully resolved through U.S. military action. The corresponding probability in hard news reports was 11%—also quite low, but far from zero. Once again, this difference in “optimism” is statistically significant at the .01 level.

Finally, soft news reports were far less likely to feature comments by individuals likely to possess significant knowledge about U.S. foreign policy, including leaders of the president's party (.16 vs. .60, p<.001), opposition party leaders (.05 vs. .09, insig.) or expert commentators (.45 vs. .78, p<.07). Instead, the soft news media emphasized commentary by sources far less likely to possess important knowledge regarding U.S. foreign policy. In particular, the soft news media featured comments by celebrities more frequently than comments by all three of the other groups combined (.79 celebrities-per-report, compared to no celebrity comments in traditional news reports).

This latter difference, which is significant at the .01 level, is potentially important because research (e.g., Druckman, 2001) has shown that source credibility is an important prerequisite for persuasion. For many Americans, the opinions of celebrities presumably carry less weight than the opinions of policy experts and leaders of the party or administration in power. Yet this may be less true for politically inattentive soft news enthusiasts. Indeed, there are some celebrities, like Oprah Winfrey, whose opinions are considered highly credible, at least to their fans.¹⁰

To the extent that some soft news viewers do consider some celebrities to be credible, it is worth noting that in the one-third of soft news reports in which celebrities took any discernible position on a U.S. conflict, they were more likely to criticize U.S. decision-makers than support them. The probability of a critical celebrity comment in a soft news report was .30, while that of a supportive comment was .18. This difference is significant at .001. In contrast, hard news reports were substantially more likely (.24 vs. .11, p<.09) to include a supportive comment from a leader of the president's party. Taken together, these data identify stark differences in the coverage of U.S. foreign policy by the soft and hard news media—differences, in each instance, consistent with my earlier conjectures.

Implications for Public Attitudes Regarding Foreign Policy

Research in social psychology (Reeder and Spores, 1983; Skowronski and Carlston, 1989) has shown that negative information tends to outweigh positive information in people's evaluations of most objects, particularly those related to moral judgment. Because soft news enthusiasts tend to be less politically attentive than typical hard news consumers (Baum, 2002), they are less likely to possess sufficient contextual knowledge to successfully counter-argue any dissonant messages to which they are exposed. Moreover, as we have seen, soft news coverage is more likely to emphasize negative aspects of U.S. military engagements, such as analogies to past high profile—and hence highly accessible to most Americans—U.S. foreign policy failures.

Moreover, politically inattentive Americans are typically far more receptive to isolationist messages than their highly attentive counterparts (Holsti, 1996; Kull and Destler, 1999). For instance, nearly half of the least politically informed respondents in the 2000 American National Election Study (NES) agreed with the proposition that the United States “would be better off if we just stayed home and did not concern ourselves with problems in other parts of the world.”¹¹ Research in cognitive psychology, in turn, has shown that individuals find it easier to accept information consistent with their pre-existing attitudes (e.g., Sniderman, Brody, and Tetlock, 1991; Popkin, 1994). Hence, for many politically inattentive soft news consumers, there is little incentive to counter-argue any isolationist messages they might encounter. (For convenience, in my statistical analyses, I refer to respondents who agreed with the above proposition as “isolationists” and those who disagreed as “internationalists.”)

The logic of selective acceptance therefore implies that inattentive isolationists who are exposed to isolationist messages in highly accessible formats are likely to accept and be influenced (or at least reinforced) by such messages. Given their limited range of accessible political attitudes, in turn, inattentive internationalists are likely to be less effective at counter-arguing isolationist messages than their more attentive counterparts. Hence, despite their internationalist sensibilities, they may be influenced by isolationist messages, albeit presumably less so than inattentive isolationists.

The overwhelming majority of highly politically aware individuals, in contrast, are internationalists (Holsti, 1996; Kull and Destler, 1999). For instance, only about 10% of the most-highly-politically-informed respondents in the 2000 NES agreed with the aforementioned isolationist position. Given their internationalist sensibilities, most politically attentive individuals, unlike their less-attentive counterparts, possess both the motive and capability (i.e., contextual knowledge) to effectively counter-argue any isolationist messages they encounter.

The aforementioned differences between the soft and hard news media's dependence on episodic and thematic frames are also likely to produce differences in consumers’ reactions. Most news outlets make liberal use of episodic frames (Iyengar, 1991). Yet I found that episodic frames are virtually ubiquitous only in the soft news media. In contrast, unlike traditional news outlets, the soft news media rarely employ thematic frames. Research in social psychology, in turn, has found that episodic frames are more likely to capture peoples’ attention, particularly if they are relatively uninvolved, ex ante, with the object of the frame (Chaiken and Eagly, 1976; Fiske and Taylor, 1991). Additional research has shown that people are more likely to remember episodic frames, because they are more vivid (Lynn, Shavitt, and Ostrom, 1985), and that vivid information—because it is emotionally involving and concrete—disproportionately influences people's judgments and causal explanations (Nisbett and Ross, 1980; Larson, 1985).

Moreover, Iyengar (1991) found that thematic frames are more likely to cause people to turn to government for a solution to whatever problem is at issue. Thematic frames engender a sense of social responsibility, and hence a motivation to see society take action to address a given problem. In contrast, while episodic stories of individual hardship or disaster may tug at people's heartstrings and evoke sympathy, Iyengar found that they tended to cause people to attribute responsibility for the problem, and for the solution, to the individuals involved, rather than to society as a whole. Hence, while episodic frames make good television drama, they tend not to engender support for governmental or societal intervention.

Given the overwhelming soft news reliance on episodic frames, it is not surprising that politically inattentive individuals—among the primary consumers of soft news (Baum, 2003b) and the very individuals most likely to be influenced by framing (Stanovich and West, 1998; Druckman, 2001)—would tend to express skepticism of internationalism. After all, the almost exclusively episodic and disproportionately negative information about foreign crises they receive via soft news tends to reinforce a belief that whatever the problem, the United States should not be responsible for fixing it.

These patterns appear to indicate that consuming soft news is likely to be associated with reduced sympathy for a proactive or internationalist U.S. foreign policy, especially among the least politically attentive members of the public. A number of hypotheses follow, four of which I test below:

• H1: The Isolationism Hypothesis. Among politically inattentive individuals, increased soft news consumption will be associated with reduced support for U.S. military interventions and greater support for isolationist positions and themes.

• H2: The Reinforcement Hypothesis. The isolationism-enhancing effect of soft news will be stronger among politically inattentive individuals who are sympathetic, ex ante, to isolationism, than among inattentive individuals who are not sympathetic to isolationism.

• H3: The Ideologue Hypothesis. As individuals’ political attentiveness increases, the pattern described in H1 will weaken, and may reverse among highly attentive individuals.

• H4: The Hard News Hypothesis. Consuming hard news will influence politically inattentive individuals less than consuming soft news. The effects on attentive individuals will be relatively modest, but may or may not be weaker than the effects of soft news.

Of course, it is possible that politically inattentive isolationists might seek out soft news outlets because they anticipate finding reinforcement for their isolationist worldviews. Similarly, in seeking to appeal to an audience suspicious of internationalism, soft news outlets might self-consciously bias their coverage of foreign policy in an isolationist direction. If so, this would make any soft news effects endogenous to the pre-existing preferences of individual consumers. Yet, in the 2000 NES, respondents’ attitudes toward isolationism explain less than one half of one percent of the variance in their daytime talk show consumption. Recent research (Baum, 2002; Prior, 2003), in turn, suggests that, unlike consumers of traditional news, people consume soft news shows primarily for entertainment, and not due to their occasional coverage of foreign policy in times of crisis. This strongly suggests that if consumers are learning about the world through the soft news media, they are doing so primarily as an unintended consequence of seeking entertainment. Hence, it seems highly unlikely that politically inattentive individuals are self-selecting into soft news audiences in order to reinforce their isolationist predispositions. Rather, any exposure to the types of themes that stoke isolationism seems likely to be largely incidental. It therefore also seems improbable that soft news outlets would bother tailoring the content of foreign affairs coverage to suit the isolationist sensibilities of a portion of their audiences who are not, after all, tuning in to learn about politics.

Finally, it is also possible that viewers tune in to soft news programs due to an overall alienation from Washington politics, and, hence, an aversion to political news. Such viewers, in turn, may be systematically more prone to isolationism. If so, any relationship between soft news exposure and isolationism might be spurious. Yet, in the 2000 NES, trust in government and the media are each positively related to daytime TV talk show consumption and negatively related to isolationism, while political efficacy, interest in government, and respondents’ attitudes on a variety of non-foreign policy political issues (e.g., affirmative action and gun control) are unrelated to talk show consumption. This strongly suggests that soft news consumption is not merely a proxy for political disaffection.

Statistical Investigations

Soft News and Support for U.S. Military Involvement in Bosnia

My first test of the Isolationism, Ideologue, and Hard News hypotheses focuses on the Bosnian Civil War of the early 1990s. I employ two surveys: a May 1994 North Carolina statewide telephone poll, conducted by the News and Observer newspaper, and a May 1993 Harris telephone poll (#931103). (Henceforth, I refer to the two surveys as the Carolina and Harris polls, respectively.)

The mid-1990s were characterized by a gradual escalation of U.S. and NATO involvement in Bosnia. In mid-1993, the Clinton administration sought—with little success—to convince its allies to support exempting anti-Serbian forces from a UN-imposed embargo on weapons sales to Bosnia. This proposal was aimed primarily at helping Bosnia's Muslims, who were widely perceived to be suffering disproportionately from the war due to the Serbian ethnic cleansing campaign. At the same time, the United States was engaged in diplomatic negotiations and public hand-wringing regarding the U.S. role in ending the civil war. In May 1993, the UN Security Council endorsed a U.S.-initiated NATO plan to enforce a no-fly zone over Bosnia, intended to protect its Muslim population.

By mid-1994, in response to Serbian bombing of a civilian marketplace in Sarajevo, the United States and its NATO allies were in the midst of an air strike campaign against Serbian targets, and President Clinton was promising to intensify the air strikes if the Serbs did not agree to talk peace. Indeed, the U.S. military engagement in Bosnia was gradually escalating, causing some in Washington to fear that Bosnia might become a quagmire, like Vietnam. Hence, both surveys took place at times when Bosnia was a regular topic in the news and, hence, a significant concern for many Americans.

For my dependent variable in the Carolina poll, I focus on a question asking respondents whether they supported or opposed “U.S. military action in Bosnia.” Responses were coded as follows: 1=“strongly oppose”, 2=“somewhat oppose”, 3=“somewhat favor”, and 4=“strongly favor.” My key causal variables are derived from a series of questions asking respondents how frequently they watch a variety of soft and hard news TV programs, which I combined into two scales measuring soft and hard news consumption, respectively. The soft news scale is based on respondents’ viewing of: (1) “Talk shows like Oprah, Donahue, or Geraldo,” (2) MTV, (3) “Shows about celebrities, like Entertainment Tonight,” and (4) “Real-life drama shows, like A Current Affair, Hard Copy, or Rescue-911.”¹² The hard news scale, in turn, is based on respondents’ viewing of: (1) “National and International News on NBC, ABC, CBS, or CNN,” (2) “Discussions of News and Business on Public Television,” and (3) “Local evening news at 6:00 or 11:00.”¹³ Responses to each question ranged from 1 (“never”) to 4 (“often”).¹⁴

Turning to the Harris poll, my dependent variables consist of two questions asking respondents (1) whether they supported or opposed “allowing the Bosnians to buy weapons to defend themselves against the Serbs,” and (2) how they would “rate President Clinton's handling of events in the former Yugoslavia, including Serbia and Bosnia.”¹⁵ I coded the responses 1 if a respondent supported lifting the arms embargo or rated President Clinton's handling of events in the former Yugoslavia as “excellent” or “pretty good,” and 0 if she opposed lifting the embargo or rated Clinton's handling of Bosnia as “only fair” or “poor.”

My key causal variable for the Harris poll (TV Talk Shows) is based on a series of questions asking respondents whether or not they watch the TV talk shows hosted by Phil Donahue, Arsenio Hall, Sally Jessy Raphael, Jay Leno, David Letterman, Regis Philbin and Kathy Lee Gifford, Joan Rivers, Geraldo Rivera, Montel Williams, and Oprah Winfrey. Each question was coded 1 if a respondent indicated that he or she watches a given talk show, and 0 otherwise. I combined these questions into a 10-item scale measuring talk show viewing habits.¹⁶ Unfortunately, the Harris poll did not include any traditional news consumption questions. Hence, in this instance, I cannot test my Hard News hypothesis.

The control variables in my analyses of these two polls, which vary according to availability, fall into four primary categories: (1) demographics (age, education, gender, family income, marital status, and race); (2) political partisanship (party identification, presidential approval, and whether respondent voted in 1992); (3) salience of U.S. policy toward Bosnia (measured as whether the respondent mentioned Bosnia as the nation's “most important problem” in an open-ended question); and (4) attitudes toward the UN and U.S. allies with ground troops in Bosnia (Britain and France).¹⁷

In order to test my hypotheses, it is necessary to estimate respondents’ political awareness. Unfortunately, neither survey contains any questions explicitly measuring this construct. An alternative indicator is education. Though certainly far from a perfect indicator, numerous studies have found education to be closely related to political awareness and have employed it as an indicator of such in statistical analyses (e.g., Converse, 1964; Bennett, 1995; Krause, 1997).

In fact, education is reasonably closely correlated with political knowledge. For instance, in the 1996 and 2000 NES surveys, education and political information correlate at about .50 and .44, respectively. In the latter survey, education also correlates at .48 with a scale derived from 10 factual political knowledge questions. Also, according to numerous surveys (e.g., Pew Center, 1998 and 2002a; NES, various years), better-educated individuals are more likely to be interested in international affairs and more likely to consume traditional news media than their less educated counterparts. This suggests highly educated individuals are likely to be less dependent upon the soft news media for information about politics (Baum, 2003b).¹⁸ Hence, I employ education as a crude, yet best available indicator of political awareness. In order to capture the effects of media consumption on respondents at different levels of political awareness, I interact the media consumption indexes with respondents’ education level. The results from the Carolina and Harris poll investigations are reported in Models 1 and 2–3, respectively, in Table 2.¹⁹

2

Logit and Ordered Logit Analyses of Effects of Watching Soft and Hard News on Attitudes Toward U.S. Involvement in Bosnia and Isolationism, as Political Information or Education Vary

	Carolina	Harris				NES 2000
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8 Talk Shows	Model 9 TV News
	U.S. in Bosnia	Bosnia Gun Ban	Clinton on Bosnia	Isolationism	Clinton For. Policy	Isolationism	Clinton For. Policy	Clinton For. Policy	Clinton For. Policy
	Media Consumption
Soft News Index	−.194	—	—	—	—	—	—	—	—
	(.094)*
Hard News Index	.064	—	—	—	—	—	—	—	—
	(.172)
TV Talk Shows	—	−.182	.141	—	—	—	—	—	—
		(.092)*	(.082)^
Daytime Talk TV	—	—	—	.604	−.410	.918	−.758	−.927	−.009
				(.332)^	(.228)^	(.447)*	(.295)**	(.553)^	(.180)
TV News (National or Local)	—	—	—	.059	.032	−.100	.019	.044	.191
				(.041)	(.028)	(.049)*	(.038)	(.020)*	(.072)**
Newspaper	—	—	—	—	—	−.067	−.010	−.037	−.040
						(.039)^	(.029)	(.047)	(.045)
Political Talk Radio	—	—	—	—	—	.109	−.029	−.117	−.075
						(.235)	(.177)	(.262)	(.264)
Internet Access	—	—	—	—	—	−.279	−.094	.075	.017
						(.251)	(.191)	(.279)	(.281)
	Demographics
Education	.058	−.223	.148	—	—	−.123	.021	.031	.038
	(.562)	(.129)^	(.121)			(.050)**	(.038)	(.063)	(.065)
Age	−.059	−.022	−.054	—	—	−.004	−.020	−.027	−.031
	(.035)^	(.032)	(.030)^			(.034)	(.027)	(.042)	(.041)
Age2	.001	.0003	.0004	—	—	.00002	.0002	.0001	.0002
	(.000)*	(.0003)	(.0003)			(.00033)	(.0003)	(.0004)	(.0004)
Family Income	−.099	.103	.117	—	—	−.067	.019	−.002	−.014
	(.064)	(.053)*	(.052)*			(.038)^	(.025)	(.033)	(.031)
Male	−.041	.325	−.202	—	—	−.133	−.130	.227	.238
	(.183)	(.164)*	(.171)			(.233)	(.169)	(.263)	(.264)
Married	.320	—	—	—	—	.365	−.092	−.145	−.053
	(.208)					(.236)	(.179)	(.266)	(.257)
Black	1.320	.017	.403	—	—	−.325	−.220	−.472	−.583
	(.427)**	(.528)	(.531)			(.693)	(.420)	(.688)	(.683)
White	.645	.468	.540	—	—	−.708	.369	.789	.853
	(.384)^	(.461)	(.471)			(.396)^	(.373)	(.469)^	(.449)^
Hispanic	.083	−.407	−1.184	—	—	−.770	.434	1.080	1.107
	(.577)	(.380)	(.423)**			(.724)	(.474)	(.712)	(.707)
		Political Characteristics
Political Information	—	—	—	−.624	−.100	−.431	.081	−.080	.217
				(.140)***	(.089)	(.187)*	(.133)	(.156)	(.200)
Approve of President	—	.124	.765	—	—	.015	2.108	2.129	2.192
		(.066)^	(.070)***			(.249)	(.177)***	(.267)***	(.270)***
Party ID	−.042	.088	−.007	—	—	.016	−.249	−.234	−.250
	(.065)	(.105)	(.106)			(.061)	(.043)***	(.061)***	(.062)***
Attention to Public Affairs	—	—	—	—	—	−.005	−.143	−.193	−.215
						(.138)	(.108)	(.154)	(.154)
Voted in 1992	−.195	—	—	—	—	—	—	—	—
	(.216)
Regular Voter	—	—	—	—	—	−.184	.040	.172	.167
						(.154)	(.130)	(.196)	(.189)
Isolationism	—	—	—	—	—	—	—	−.490	.112
								(.300)^	(.426)
Bosnia as Important National Problem	—	.510	−.153	—	—	—	—	—	—
		(.420)	(.383)
Affect on UK	—	−.271	−.034	—	—	—	—	—	—
		(.120)*	(.123)
Affect on France	—	.187	.068	—	—	—	—	—	—
		(.082)*	(.082)
Support United Nations	—	.120	−.377	—	—	—	—	—	—
		(.091)	(.100)***
Political Trust and Efficacy
External Efficacy	—	—	—	—	—	−.695	.411	−.006	−.051
						(.297)*	(.213)*	(.299)	(.298)
Internal Efficacy	—	—	—	—	—	.108	−.308	.093	.135
						(.243)	(.185)^	(.266)	(.264)
Trust Government	—	—	—	—	—	−.300	.368	.393	.383
						(.135)*	(.098)***	(.152)**	(.149)**
Trust Media	—	—	—	—	—	−.237	.323	.479	.486
						(.164)	(.116)**	(.179)**	(.174)**
Interaction Terms
Soft News Index×Education	.038	—	—	—	—	—	—	—	—
	(.031)
Hard News Index×Education	−.037	—	—	—	—	—	—	—	—
	(.060)
TV Talk Shows×Education	—	.077	−.053	—	—	—	—	—	—
		(.030)**	(.027)*
Political Information×Daytime Talk TV	—	—	—	−.137	.204	−.221	.232	.313	—
				(.100)	(.072)**	(.134)^	(.089)**	(.147)*
Political Information×TV News	—	—	—	.019	−.006	.030	−.004	—	−.033
				(.012)	(.008)	(.014)*	(.011)		(.019)^
Isolationism ×Daytime Talk TV	—	—	—	—	—	—	—	−1.125	—
								(1.044)
Isolationism×Political Information×Daytime Talk TV	—	—	—	—	—	—	—	.285	—
	—	—	—	—	—	—	—	(.340)
Isolationism ×TV News	—	—	—	—	—	—	—	—	−.256
									(.083)**
Isolationism×Political Information× TV News	—	—	—	—	—	—	—	—	.049
									(.021)*
Intercept 1	−3.884	−.185	−1.093	.839	.863	5.554	−2.101	−2.624	−3.734
	(1.733)	(.986)	(.951)	(.443)^	(.308)**	(1.244)***	(.988)*	(1.493)^	(1.550)*
Intercept 2	−2.163	—	—	—	—	—	—	—	—
	(1.718)
Intercept 3	−.745	—	—	—	—	—	—	—	—
	(1.716)
χ2	37.19***	40.15***	193.13***	45.69***	20.50***	104.40***	333.17***	177.24***	180.84***
Pseudo R2	.03	.06	.20	.06	.01	.15	.30	.33	.33
(N)	(N=464)	(N=964)	(N=990)	(N=743)	(N=1423)	(N=626)	(N=1222)	(N=600)	(N=600)

^

p<.10,

*

p<.05,

**

p<.01,

***

***p<.001

Note: All models employ heteroscedasticity-consistent (“robust”) standard errors. Carolina and Harris poll models employ probability weighting. (Standard errors in parentheses)

In these data, soft news appears to have a substantial influence on respondents’ views of the U.S. intervention in Bosnia, particularly among those with relatively little formal education. Figure 1 presents three graphics, in which I have transformed the logit and ordered logit coefficients into probabilities of supporting U.S. policy in Bosnia, as education and soft news consumption vary. The graphic in the top right-hand quadrant presents the probabilities of strongly supporting a U.S. military role in Bosnia, from the Carolina poll; the graphics in the bottom left- and right-hand quadrants, respectively, from the Harris poll, track respondents’ probabilities of support for lifting the arms embargo and for President Clinton's handling of U.S. policy toward Bosnia.²⁰

1

Probability of Supporting U.S. Policy Toward Bosnia, as Soft News Consumption and Education Vary

Beginning with the Carolina poll results (shown in Model 1 of Table 2), consistent with my Isolationism hypothesis, among all respondents with less than a post-graduate education, as soft news exposure increases, support for a U.S. military role in Bosnia declines. The declines are statistically significant (p<.05 or p<.10) for the 55 percent of respondents who did not attend college. Among respondents who did not complete high school, as soft news exposure increases from its minimum to maximum values, the probability of strongly supporting a U.S. military role in Bosnia declines by 37 percentage points (p<.05). The corresponding effect on the probabilities of strongly or somewhat opposing a U.S. military role are increases of 18 and 24 percentage points, respectively (p<.05). As anticipated by my Ideologue hypothesis, these patterns weaken in a stepwise fashion as respondents move up the education ladder. Finally, consistent with my Hard News hypothesis, consuming traditional news does not influence respondents’ attitudes regarding the U.S. role in Bosnia.

Turning to the Harris poll results (Models 2 and 3 in Table 2), the two graphics in the bottom half of Figure 1 indicate that among the 40 percent of respondents who did not attend college, increased talk show exposure is associated with reduced support for U.S. policy toward Bosnia.²¹ Among respondents who did not complete high school, as talk show consumption increases from its minimum to maximum values, the probabilities of supporting lifting the arms embargo and approving of President Clinton's handling of Bosnia decline by 24 and 20 percentage points, respectively. As before, the relationships weaken in a stepwise fashion as respondents’ education level increases. These results offer additional support for my Isolationism and Ideologue hypotheses.

In both surveys, consuming soft news had the opposite effect on college-educated respondents. For instance, among those with a post-graduate degree, as talk show consumption increases from its lowest to highest values, the probabilities of favoring lifting the embargo and approving of Clinton's Bosnia policies increase by 36 and 30 percentage points, respectively. These results are consistent with my Ideologue hypothesis. Among political sophisticates, increased exposure to information about Bosnia via soft news appears to be associated with support for a more proactive U.S. role in Bosnia, such as that advocated by President Clinton.

The debate surrounding the arms embargo illustrates why this might be the case. Politically inattentive individuals, lacking a broader context for understanding U.S. policy toward Bosnia, may see lifting the arms embargo as, in effect, adding fuel to the fire. Soft news reports detailing the brutality of the civil war may induce such individuals to oppose any policy that seems likely to exacerbate the violence, even in the short term. In contrast, highly attentive individuals may be more likely to focus on longer-term effects. They might, for instance, conclude that, even given a short-term escalation of the war, enhancing the Bosnian Muslims’ capacity for self-defense would reduce the likelihood of a larger-scale U.S. intervention at a later date. Hence, a political sophisticate might see the same soft news report as a less attentive individual yet draw the opposite conclusion. Regardless, these data suggest that among respondents lacking a college education, but not among their better-educated counterparts, consuming soft news is associated with greater suspicion of a proactive U.S. policy toward Bosnia.

Daytime Talk Shows, Isolationism, and the Clinton Policy Agenda

In 2000, the NES, for the first time, included a question asking respondents how frequently they watch daytime talk shows. This makes it possible to explore the effects of watching these programs on respondents’ attitudes regarding foreign policy. Specifically, I focus on the following two questions: (1) “Do you agree or disagree with this statement: This country would be better off if we just stayed home and did not concern ourselves with problems in other parts of the world,” and (2) “Do you approve or disapprove of the way Bill Clinton is handling our relations with foreign countries?” (Responses were coded 0=“disagree” or “disapprove,” 1=“agree” or “approve.”)

The key causal variable (Daytime Talk TV), in turn, is based on the following question: “How many times in the last week have you watched daytime television talk shows such as ‘Oprah Winfrey,’‘Rosie O'Donnell,’ or ‘Jerry Springer’?” I collapsed the responses into four categories: 0=never, 1=1–2 times, 2=3–5 times, 3=6+ times or every day. To capture the effects of consuming hard news, I created a scale based on three questions measuring consumption of local and national TV news.²² The control variables fall into four categories: (1) demographics (age, education, family income, gender, race, marital status), (2) political interest and knowledge (political information, party identification, attention to public affairs, regular voter, presidential approval), (3) political trust and efficacy (trust in government and media, internal efficacy, external efficacy), and (4) other media consumption habits (newspapers, political talk radio, Internet).²³ (See Appendix 1 for variable coding and definitions.)

For this analysis, a more direct indicator of political awareness is available: the NES interviewers’ assessment of respondents’ overall level of political information (a 5-point scale with 5 representing maximum information).²⁴ Once again, I include an interaction term to determine whether consuming talk shows has a stronger effect on inattentive respondents, as well as to determine whether, consistent with the Reinforcement hypothesis, any such effects vary with respondents’ ex ante support for isolationism. Models 4–7 in Table 2 present a series of logit analyses employing both dependent variables. Due to the broad range of control variables, I begin (in Models 4–5) by excluding all controls, in order to insure that my results are in no way artifacts of model specification. Models 6–7 then replicate the analyses with all controls included. These models test my Isolationism, Ideologue, and Hard News hypotheses. The preliminary models are broadly consistent with the full multivariate models, suggesting that the relationships are quite robust. Hence, I focus on the fully specified models.

Beginning with the Isolationism and Ideologue hypotheses, the results offer additional support in each instance. To assess the substantive effects of consuming talk shows, I translate the logit coefficients into probabilities of supporting isolationism or approving of President Clinton's foreign policy management, as daytime talk show consumption and political information vary from their lowest to highest values. Consistent with the Isolationism hypothesis, among the least informed respondents, a maximum increase in talk show viewing is associated with a 45 percentage point increase in isolationist sentiment (from .27 to .71, p<.05) and a 36 percentage point decrease in support for Clinton's handling of foreign policy (from .72 to .36, p<.01). Consistent with my Ideologue hypothesis, these relationships weaken in a stepwise fashion as respondents’ political information increases and reverse at the highest information level. Among the most-highly informed respondents, the corresponding increase in talk show consumption is associated with a statistically insignificant 5 percentage point decline in isolationism (from .17 to .12) and a 15 percentage point increase (from .75 to .90, p<.05) in support for Clinton's management of foreign policy.

These latter results are consistent with my previous investigation, and with my conceptual framework. After all, a soft news report about foreign policy presumably enhances a political sophisticate's overall knowledge about foreign affairs by only a very small degree, if at all. Moreover, for the reasons discussed earlier, it seems more likely to reinforce their relatively strong internationalist predispositions rather than challenge them.

The results from Models 3 and 4 also support the Hard News hypothesis. Among the least informed respondents, a maximum increase in TV news viewing is associated with a 29 percentage point decrease in isolationism (from .48 to .19, p<.05) and a statistically insignificant 6 percentage point increase in support for President Clinton's handling of foreign policy (from .64 to .70). For both dependent variables, as predicted, the effects on the least informed respondents are far more modest than the corresponding effects of talk show consumption. Moreover, in both cases, they move in the opposite direction—toward reduced isolationism and enhanced support for Clinton's foreign policy. Once again, this suggests that the different content of hard and soft news shows produces substantially different effects on politically inattentive Americans. Moreover, consistent with my Ideologue hypothesis, these inverse relationships weaken as respondents’ political information increases. Among the most highly informed respondents, the relationships are substantively small and statistically insignificant for both dependent variables.

I turn next to my tests of the Reinforcement hypothesis. For this investigation, I employ the isolationism question as an explanatory variable, interacting it with talk show consumption and political information. As my dependent variable, I employ the aforementioned question asking respondents whether they approve of President Clinton's management of foreign policy. Given Clinton's avowed support for a proactive, multilateral approach to U.S. foreign policy, it seems highly probable (indeed virtually certain) that isolationists would be less likely than internationalists to approve of his handling of foreign affairs. Indeed, in the 2000 NES, 71 percent of internationalists approved of President Clinton's foreign policy, compared to just 57 percent of isolationists.

Models 8–9 in Table 2 present the results of my tests of this hypothesis.²⁵ Beginning with the least politically informed isolationists, as daytime talk show consumption increases from its lowest to highest levels, the probability of approving of President Clinton's handling of foreign policy declines by 60 percentage points (from .69 to .09, p<.05). Among the least informed internationalists, the corresponding effect is a nearly-significant (p<.15), yet substantially smaller, 39-percentage point decline (from .78 to .39). Finally, among highly politically informed internationalists, a maximum increase in talk show consumption is associated with a 21-percentage point increase in support for President Clinton's foreign policy management (from .72 to .93, p<.05).²⁶

Given the nature of soft news coverage of foreign crises, these results are entirely consistent with selective acceptance. The attitudes of politically inattentive isolationists are strongly reinforced by exposure to daytime talk shows. Their more-internationally-oriented yet also inattentive counterparts appear to have resisted some of the isolationism-enhancing effects of daytime talk shows; such effects among this group are smaller and are not quite statistically significant. They do not, however, appear to be able to resist such messages entirely. After all, the direction of the effect is the same as for low-information isolationists, while the magnitude of the isolationism-enhancing effect of consuming talk shows remains substantial. At least in these data, consuming talk shows is clearly associated with increased isolationism among low-information respondents, regardless of their views concerning the appropriate U.S. role in the world, albeit more strongly so for those predisposed to support isolationism. Additionally, as expected, highly informed internationalists are able to counter-argue any isolationist messages in daytime talk shows, while finding reinforcement for their internationalist predispositions. Taken together, these results strongly support the Reinforcement hypothesis.

Turning, finally, to the effects of consuming hard news on support for President Clinton's foreign policy management, as one might anticipate given my content analyses, exposure to TV news increases support for Clinton's foreign policy among all but the least informed isolationists. Among these respondents, a maximum increase in TV news watching is associated with a statistically insignificant 22-percentage point decrease in the probability of supporting Clinton's foreign policy (from .53 to .31). In contrast, among the least informed internationalists, a maximum increase in TV news watching is associated with a 44 percentage point increase (from .51 to .95, p<.01) in the probability of supporting Clinton's management of foreign policy. The corresponding increase among highly informed internationalists is a statistically insignificant 9 percentage points (from .70 to .79).

In this instance, given the relatively more internationalist tenor of hard news, any reinforcement effect should apply primarily to internationalists. Hence, it is not surprising that the strongest effects emerge among these respondents, whose attitudes are reinforced by exposure to TV news shows, regardless of their level of political information. In order to evaluate these results against the Hard News hypothesis, the appropriate comparison therefore involves the reinforcement effect of talk show consumption on low-information isolationists, relative to that of hard news consumption on low-information internationalists. In fact, consistent with my hypothesis, and with selective acceptance, the reinforcement effect of daytime talk shows among low-information isolationists is more than one-third larger than that among low-information internationalists associated with consuming TV news shows. Similarly, though statistically insignificant in both instances, the effects of consuming TV news on low-information isolationists are far weaker than the effects of talk show consumption on low-information internationalists.

Conclusion

Each of my statistical investigations indicated that consuming soft news enhances skepticism of an internationalist American foreign policy, with the strongest effects emerging among the least politically attentive segments of the population. Among highly attentive individuals, however, exposure to soft news appears to essentially reinforce internationalism. These patterns are consistent with selective acceptance, and, for politically inattentive respondents, they are precisely what one would anticipate arising from exposure to disproportionately negative and episodic framing of foreign policy issues and events, combined with an emphasis on violence, tragedy, and other forms of highly accessible human drama, such as prevails in the soft news media.

Hard news had far less impact than soft news on inattentive individuals. Moreover, in this instance, consistent with the reception-acceptance model, highly attentive respondents mirror their less attentive counterparts; they also appear relatively immune to being influenced by hard news, presumably due to their capacity to counter-argue dissonant information. The exception concerns the Reinforcement hypothesis; in these data, internationalists at all levels of political information found some measure of reinforcement for their attitudes through consuming hard news. Nevertheless, consistent with the Hard News hypothesis, among inattentive respondents this effect was smaller than the reinforcement effect of soft news on isolationists.

The soft news media self-consciously package information about foreign crises in ways designed to appeal to apolitical consumers (Baum, 2002). One presumably unintended consequence is that the same episodic framing of human drama in a violent context that makes a story compelling may also produce a repellent effect, particularly among viewers who lack sufficient knowledge about an issue to place such information into a broader context. After all, while millions of people enjoy watching Arnold Schwarzenegger and Bruce Willis vanquish terrorists and other “evil-doers” on screen from the safety of their movie theater seats, how many would actually choose to confront real-world dangers comparable to those of their celluloid heroes?

A large volume of research has shown that public opinion does influence U.S. foreign policy, even with respect to the use of military force (e.g., Key, 1961; Ostrom and Job, 1986; Sobel, 2001). This gives policymakers an incentive to seek to influence all public sources of news about foreign affairs, including the soft news media. The findings reported in this study suggest that failure to do so may result in reduced, or at least narrowed, public support for future U.S. foreign policy initiatives.

In fact, U.S. policymakers appear to recognize the importance of soft news coverage of war. In the run-up to the March 2003 U.S. invasion of Iraq, such decidedly apolitical media outlets as MTV, Rolling Stone, and People magazine were granted coveted reporting slots “embedded” within U.S. combat units. Explaining the Pentagon's rationale for granting these scarce slots to soft news reporters, Bryan G. Whitman, Deputy Assistant Secretary of Defense for Media Operations, explained: “It is a recognition that not everyone gets their news from The Washington Post and The Wall Street Journal…. We consciously looked at those news organizations that have reach and impact and provided them with the greatest possible opportunities…. Each of them [embedded reporters] reach a different audience. Our goal was to dominate the information market…” (Carr, 2003).

It is, of course, possible that the differences between soft and hard news coverage described herein may be limited to the period I investigated. At minimum, there are doubtless exceptions to these patterns, such as, perhaps, foreign crises that enjoy broad, bipartisan elite support. For instance, politically inattentive and less educated Americans were just as likely to support the post-9/11 U.S. attack against the Taliban regime in Afghanistan, and the War on Terrorism in general, at least initially, as their more attentive and better-educated counterparts.²⁷

In an environment of overwhelming bipartisan popular and elite support, it is certainly possible that coverage of the War on Terrorism in the soft news media, while emphasizing different aspects of the story than traditional news outlets, would nonetheless reinforce, rather than undermine, the public consensus, at least in the short term. Zaller (1994) refers to this as a “mainstreaming effect,” which arises when a president's policy enjoys broad elite support, behind which the news media typically fall into line in lock step. Absent media coverage of elite dissent, traditional partisan divisions in the public recede, and the president tends to gain widespread public support.

The events of 9/11, and the subsequent War on Terrorism, produced a dramatic mainstreaming effect in the traditional news media (Rosensteil et al., 2002). Yet, my content analysis indicated that the soft news media depend far less than their hard news counterparts on official government sources. Hence, it is unclear whether the mainstreaming effect, as originally conceived, even applies to the soft news media. To the extent that it does, this is most likely attributable to support for the president's policies among celebrities, rather than political elites. After all, soft news outlets feature far more comments by the former group. In this instance, even Hollywood celebrities who usually oppose U.S. military engagements were largely supportive of the War on Terrorism, at least initially.

Another factor most likely contributing to the near universal popular support for the War on Terrorism is the fact that it followed a direct attack on the United States. Research (Jentleson, 1992) has shown that the American people are more willing to support a use of force when U.S. goals are perceived as defensive, rather than offensive. The Bush Administration successfully framed the post-9/11 conflicts in Afghanistan and Iraq (initially, at least) as defensive uses of force. Moreover, whereas many politically inattentive individuals are generally suspicious of a proactive U.S. foreign policy, it does not necessarily follow that they will oppose a forceful U.S. response to a direct attack against American soil. Indeed, due to the morally unambiguous and near-universally accessible “good vs. evil” frame associated with the War on Terrorism, such individuals would seem likely to support military action with at least as much enthusiasm as political sophisticates.

Zaller (1994) further argues that elite discord regarding a U.S. military conflict can produce a “polarization effect” in which elite debate is reflected in more critical media coverage, resulting, in turn, in heightened partisan divisions among the public. In the 1990s, virtually all post-Desert Storm U.S. military engagements provoked substantial elite discord. Hence, all of the events addressed in my content analyses were prime candidates for polarization effects. Yet, as we have seen, the tenor of soft news coverage cannot be attributed to variations in elite support for a president's policies. The disproportionate soft news reliance on celebrities – many of whom frequently express anti-war sentiments – suggests that the threshold for a mainstreaming effect in the soft news media is most likely quite high. This, in turn, suggests that, absent further attacks on American soil, the War on Terrorism may prove to be an exception rather than a new post post-Cold War rule. If so, soft news coverage of foreign policy will most likely continue to be associated with enhanced isolationism among politically inattentive Americans. This represents a potential challenge to America's post-World War II internationalist foreign policy consensus. In fact, the embedding of soft news reporters with U.S. combat troops in Iraq suggests the current administration is already attempting to respond directly to this challenge. Yet, at least as of this writing, the long-term effectiveness of such efforts remains unclear.

Appendices

Appendix 1. NES Variable Definitions and Coding

Internal Efficacy:“Please tell me how much you agree or disagree with these statements…. Sometimes politics and government seem so complicated that a person like me can't really understand what's going on.” Coding: 1=“agree,” .5=“don't know,” 0=“disagree.”

External Efficacy: Scale constructed from 2 questions (1) “Please tell me how much you agree or disagree with this statement…. People like me don't have any say about what the government does.”; and (2) “Please tell me how much you agree or disagree with this statement…. I don't think public officials care much what people like me think.” Coding: 1=“agree,” .5=“don't know,” 0=“disagree.”

Regular Voter: 3-point scale, running from 0 to 2, constructed from 2 questions asking respondents whether they voted in the 1996 and 2000 elections. Respondents were given 1 point for each election in which they claimed to have voted.

Trust Government: 6-point scale constructed from 2 questions: (1) “How much of the time do you think you can trust the government in Washington to do what is right?” Coding 4=“just about always,” 3=“most of the time,” 2=“only some of the time,” 1=“never”; (2) “Do you think that people in the government waste a lot of the money we pay in taxes, waste some of it, or don't waste very much of it?” Coding 1=“don't waste very much,” 2=“waste some,” 3=“waste a lot.”

Trust Media: Question: “How much can you trust the media?” Coding: 1=“almost never,” 2=“only some of the time,” 3=“most of the time,” 4=“just about always.”

Attention to Public Affairs: Question: “Some people seem to follow what's going on in government and public affairs most of the time, whether there's an election going on or not. Others aren't that interested. Would you say you follow what's going on in government and public affairs most of the time, some of the time, only now and then, or hardly at all?” Coding: 1=“hardly at all,” 2=“only now and then,” 3=“some of the time,” 4=“most of the time.”

Political Information: Interviewer's estimate of respondent's “general level of information about politics and public affairs.” Coding: 1=“very low,” 2=“fairly low,” 3=“average,” 4=“fairly high,” 5=“very high.”

Watch TV News: Scale running from 0–21, based on three questions asking respondents how many days in the past week (0–7) they watched the national network TV news, as well as early- and late-evening local TV news.

Read Daily Newspaper=Number of days in past week respondent read a daily newspaper.

Listen to Political Talk Radio: “There are a number of programs on radio in which people call in to voice their opinions about politics. Do you ever listen to political talk radio programs of this type?” Coding: 0=“no,” 1=“yes”. “Don't know” responses were coded as missing.

Internet Access=variable coded 1 if respondent has access to the Internet and 0 otherwise.

Appendix 2. Content Analysis Coding Form and Data

Coding Form

1. Did the report discuss the origins of the conflict?

2. If you answer yes to question 1, was the crisis described as having origins long before the current actual or contemplated U.S. involvement?

3. Did the report identify a country, faction, or group as an adversary/antagonist in the conflict?

4. Was the crisis presented as one that had a good chance of being resolved through US action?

5. How many members of the US President's political party were sourced in the report?

6. How many members of the opposition political party were sourced in the report?

7. How many experts (includes senior military officers) were sourced in the report?

8. How many celebrities were sourced in the report?

9. How would you characterize the assessment of the US decision-makers offered by those [presidential party members/opposition party members/experts/celebrities]? (1) supportive; (2) critical; (3) mixed; (4) no assessment offered.

10. Were any past foreign policy crises mentioned in the report? If yes, list them below, along with a brief description of the context in which they were mentioned.

11. For each past foreign policy crisis mentioned in the report, note if the crisis was a reasonably clear-cut success or failure for U.S. foreign policy (or is widely perceived as such by most elites and members of the public), or if it was not clear-cut either way.

12. Did the report focus on, or address the circumstances or involvement of, an individual, family or other small group?

13. Did the report address the general or broader issues involved in the conflict or crisis that was the subject of the report (as opposed to the effects on, or involvement of, specific individuals, families or small groups)?

A1

Soft and Hard News Programs Included in Content Analysis Investigation

Date	Duration	Program	Topic
Traditional News Programs
1. 12/25/92	2:06	CNN Prime News	Somalia
2. 12/31/92	4:56	CNN Prime News	Somalia
3. 1/3/93	2:27	CBS Evening News	Somalia
4. 1/12/93	1:14	CBS Evening News	Somalia
5. 1/24/93	2:15	CNN Prime News	Somalia
6. 3/10/93	2:27	CNN Prime News	Somalia
7. 3/25/93	2:05	CBS Evening News	Somalia
8. 5/1/93	3:42	CBS Evening News	Bosnia
9. 6/18/93	3:33	ABC World News Tonight	Somalia
10. 7/1/93	2:03	CNN Prime News	Somalia
11. 10/2/93	2:03	CNN Prime News	Somalia
12. 10/4/93	2:02	CBS Evening News	Somalia
13. 10/5/93	3:16	ABC World News Tonight	Somalia
14. 10/7/93	2:15	CBS Evening News	Somalia
15. 10/20/93	2:02	CBS Evening News	Somalia
16. 9/26/94	3:53	ABC World News Tonight	Haiti
17. 9/26/94	3:51	ABC World News Tonight	Haiti
18. 9/30/94	4:50	CNN Prime News	Haiti
19. 9/30/94	2:14	ABC World News Tonight	Haiti
20. 10/1/94	5:36	CNN Prime News	Haiti
21. 10/1/94	3:21	CBS Evening News	Haiti
22. 6/10/95	5:14	ABC World News Tonight	Bosnia
23. 6/11/95	2:25	CNN Prime News	Bosnia
24. 6/12/95	2:04	CNN Prime News	Bosnia
25. 6/12/95	3:33	ABC World News Tonight	Bosnia
26. 6/12/95	3:48	CBS Evening News	Bosnia
27. 6/12/95	7:24	CNN Prime News	Bosnia
28. 12/1/95	2:59	ABC World News Tonight	Bosnia
29. 12/1/95	2:47	ABC World News Tonight	Bosnia
30. 12/31/95	2:16	CBS Evening News	Bosnia
31. 1/22/96	3:54	ABC World News Tonight	Bosnia
32. 1/24/96	2:49	CNN Prime News	Bosnia
33. 1/25/96	2:05	CNN Prime News	Bosnia
34. 12/19/97	5:03	ABC World News Tonight	Iraq
35. 1/30/98	2:27	CNN Prime News	Iraq
36. 2/28/98	2:02	NBC Nightly News	Iraq
37. 8/20/98	2:20	NBC Nightly News	Cruise Missile Attacks on Afghanistan & Sudan
38. 8/20/98	2:03	CBS Evening News	Cruise Missile Attacks on Afghanistan & Sudan
39. 8/21/98	2:57	CNN Prime News	Cruise Missile Attacks on Afghanistan & Sudan
40. 8/22/98	2:09	CBS Evening News	Cruise Missile Attacks on Afghanistan & Sudan
41. 8/24/98	2:20	ABC World News Tonight	Cruise Missile Attacks on Afghanistan & Sudan
42. 12/16/98	2:21	ABC World News Tonight	Iraq
43. 12/17/98	2:00	CBS Evening News	Iraq
44. 12/18/98	19:43	CNN Prime News	Iraq
45. 12/18/98	2:15	NBC Nightly News	Iraq
46. 2/15/99	2:58	CBS Evening News	Kosovo
47. 3/6/99	3:42	CNN Prime News	Kosovo
48. 3/15/99	3:11	ABC World News Tonight	Kosovo
49. 3/27/99	3:21	NBC Nightly News	Kosovo
50. 4/8/99	2:52	NBC Nightly News	Kosovo
51. 4/10/99	2:18	CBS Evening News	Kosovo
52. 5/25/99	2:09	ABC World News Tonight	Kosovo
53. 6/29/99	2:04	CBS Evening News	Kosovo
54. 1/29/99	3:40	ABC World News Tonight	Kosovo
55. 4/29/199	2:29	CNN Prime News	Kosovo
Soft News Programs
1. 7/24/94	2:39	Hard Copy	Bosnia
2. 9/27/94	5:26	Inside Edition	Iraq
3. 9/27/94	5:32	Inside Edition	Haiti
4. 6/13/95	7:45	Inside Edition	Iraq
5. 04/01/96	2:16	Inside Edition	Bosnia
6. 07/02/97	3:13	Inside Edition	Bosnia
7. 2/27/98	2:00	Entertainment Tonight	Iraq
8. 3/4/98	2:44	Access Hollywood	Iraq
9. 8/18/98	3:23	Access Hollywood	Cruise Missile Attacks on Afghanistan & Sudan
10. 8/20/98	3:03	Extra	Cruise Missile Attacks on Afghanistan & Sudan
11. 8/20/98	2:20	Access Hollywood	Cruise Missile Attacks on Afghanistan & Sudan
12. 8/21/98	2:39	Access Hollywood	Cruise Missile Attacks on Afghanistan & Sudan
13. 8/21/98	2:34	Extra	Iraq
14. 8/21/98	3:26	Extra	Osama Bin Laden
15. 8/21/98	2:29	Extra	Iraq
16. 8/22/98	1:44	Daily Show	Cruise Missile Attacks on Afghanistan & Sudan
17. 8/23/98	2:15	Entertainment Tonight	Cruise Missile Attacks on Afghanistan & Sudan
18. 8/25/98	2:27	Access Hollywood	Interview with George H. W. Bush on Gulf War
19. 8/26/98	2:39	Extra	South Africa Nightclub Bombing
20. 8/27/98	2:09	Extra	South Africa Nightclub Bombing
21. 8/27/98	0:51	Access Hollywood	South Africa Nightclub Bombing
22. 8/28/98	1:12	Access Hollywood	South Africa Nightclub Bombing
23. 9/7/98	3:05	Extra	Terrorism
24. 9/10/98	7:18	Extra	Osama Bin Laden
25. 12/17/98	1:35	Extra	Iraq
26. 12/17/98	1:33	Extra	High Tech Weapons
27. 12/18/98	1:42	Entertainment Tonight	Iraq
28. 12/18/98	1:55	Extra	Iraq
29. 12/18/98	1:02	Access Hollywood	Dangers to Journalists of Covering Wars
30. 3/24/99	1:33	Inside Edition	Afghanistan
31. 3/24/99*	2:35	Extra	Kosovo
32. 3/29/99	2:15	Extra	Kosovo
33. 3/29/99	2:02	Extra	Iraq
34. 3/29/99	0:21	Access Hollywood	Kosovo
35. 4/5/99	2:26	Extra	Kosovo
36. 4/5/99	2:16	Extra	Kosovo
37. 4/5/99	2:37	Extra	Kosovo
38. 4/5/99	1:32	Extra	Kosovo

*

estimated air date (precise date was unavailable).

Footnotes

Author's note: An earlier draft of this article was presented at the 2002 meetings of the American Political Science Association and the Peace Science Society (International). I wish to thank the following individuals for their valuable comments and suggestions: James Druckman, Tim Groeling, Karen Hult, David Lake, Stephen Wayne, John Zaller, and Amy Zegart, as well as several anonymous reviewers and the ISQ editors. I also wish to thank Scott Hoaby, Angie Jamison, Liz Stein and Eric Zusman for research assistance, and Phil Gussin for coordinating the content analysis project. Finally, I am grateful to Paul Rosenthal for assistance in obtaining the soft and hard news transcripts and videotapes employed in my content analyses.

1

Source: CBS/New York Times poll, 8/20/98.

2

Source: Lexis-Nexis, August 20-27, 1998.

3

Baum (2003b) reports that in October 2002, the total average daily audience for four popular entertainment news programs (ET, Inside Edition, Extra, and Access Hollywood) was over 22 million viewers. The corresponding totals for three daytime (The Oprah Winfrey Show, Dr. Phil, and Live with Regis and Kelly) and late-night (Jay Leno, David Letterman, and Conan O'Brien) talk shows were 22 and 16 million viewers per day, respectively. In contrast, the top dozen cable news shows, combined, attracted an average of 8.4 million viewers per day during the first quarter of 2002 (Baum, 2003a).

4

The soft news programs I reviewed included: ET, Access Hollywood, Extra, The Daily Show, E! News Daily, The Tonight Show with Jay Leno, Late Night with Conan O'Brien, Politically Incorrect, The Howard Stern Show, 60 Minutes, 20/20 and Dateline. If I could not determine from Lexis-Nexis abstracts whether the “Wag the Dog” theme was addressed in a given story, I counted it as non–“Wag the Dog”-related coverage of the missile strikes.

5

Source: Newsweek poll, 8/20/98-8/21/98.

6

The soft news programs included: Extra, Access Hollywood, Inside Edition, ET, Hard Copy, and The Daily Show. The hard news programs included: ABC World News Tonight, CBS Evening News, NBC Nightly News, and CNN's prime time news. The average duration of soft news reports was 2:38. The hard news reports averaged 591 words in length. Because many of these did not include information on duration, I calculated the number of words-per-second where both word counts and duration were available. Applying the words-per-second average to the other reports yielded an estimated average duration for hard news reports of 2:56.

7

The coders agreed on 85% of all coding decisions, for both soft and hard news programs and for both written transcripts and videotaped story segments. Calculating Cohen's Kappa score (Cook, 1998) indicates that this degree of inter-coder agreement is highly unlikely to be attributable to chance.

8

All significance levels reported in this section are based on 2-tailed t-tests.

9

Only unambiguous failures are coded as such. Uses of force involving ambiguous outcomes are not included in these totals. The coders agreed on about 90% of all coding decisions on this question.

10

For instance, in one survey (Pew Center, 2002b), Oprah Winfrey's score on a “believability” scale was nearly identical to those of Alan Greenspan and George W. Bush.

11

This figure is based on the NES interviewer's placement of the respondent on a 5-point scale based on their level of political information.

12

The question wording is as follows: “How often do you watch _____? Would you say often, sometimes, rarely, or never?”

13

According to a study of 49 stations in 15 cities (Rosensteil et al., 2000), “politics and government” is second only to “crime and law” as the most prevalent topic on local TV newscasts. A factor analysis, employing the Carolina poll, indicated that local, network, and public TV news shows load on a common underlying factor, at .57, .50 and .41, respectively, while TV talk shows, MTV, entertainment news shows, and real-life/tabloid news shows load on a second common factor, at .53, .29, .47 and .54, respectively. This suggests that local TV news is more accurately characterized as hard, rather than soft, news.

14

Cronbach's alpha reliability scores for the hard and soft news scales are .57 and .54, respectively. Price and Zaller (1993) find that survey respondents tend to overstate their media consumption, particularly for questions addressing relatively broad media categories. The questions I employ are fairly specific – the NES, for instance, asks respondents to recall their consumption of specific programs during the prior week– thereby mitigating this problem. Moreover, there is no theoretical reason to believe that differences in the predicted probabilities of holding a given attitude, as news consumption varies, should be affected by systematic over-reporting of the level of news consumption.

15

In 1993, advocating an end to the embargo was widely perceived as an expansion of U.S. engagement in Bosnia and a reversal of the Bush Administration's policy of preserving the status quo. In his presidential campaign, Bill Clinton regularly criticized President Bush for failing to intervene more proactively. Upon assuming office, Clinton escalated U.S. diplomatic and military involvement in Bosnia. He was therefore clearly associated with advocating an expanded U.S. role in Bosnia.

16

Chronbach's alpha reliability score for the talk show index is .73.

17

The latter questions asked respondents to rate each country on a 4 point scale, coded 1=“close ally,” 2=“friendly [but] not ally,” 3=“not friendly,” and 4=“enemy.” The UN question, in turn, is an additive scale derived from the following three questions: (1) “Overall, how would you rate the job the United Nations is doing in working for peace in the world – excellent, pretty good, only fair, or poor?” (Coding: 1=“excellent” or “pretty good,” 0=“only fair” or “poor”); (2) “Do you feel that the UN is an effective organization for peace or not?” (Coding: 1=“is effective,” 0=“is not effective”); and (3) “All things considered, do you think the UN is worthwhile or not worthwhile?” (Coding: 1=“UN worthwhile,” 0=“not worthwhile”). Responses of “not sure” were coded as missing.

18

According to the 2000 NES, the least politically informed respondents and those lacking a high school diploma consume 87 and 40 percent more daytime TV talk shows, respectively, than their highly informed or college educated counterparts. According to the Carolina and Harris polls, non-high school graduates consume about 1/3 more entertainment news programs and TV talk shows, respectively, than their college-educated counterparts. In prior research (Baum 2003a), in turn, I report that Nielsen ratings in October 2002 for a series of entertainment news shows and daytime TV talk shows were, on average, 26 and 33 percent higher, respectively, for households lacking a college-educated adult, compared to college-educated households. This suggests that the self-declared viewing habits in the surveys I employ are reasonably accurate estimates of actual soft and hard news viewing patterns.

19

In each instance, I first tested models including only the key causal variables. The results (not shown) were largely consistent with the fully specified models.

20

To derive predicted probabilities and significance levels, I employ a simulation technique developed by King et al. (2000). This procedure, however, does not allow probability weighting, which is arguably an important element in analyzing public opinion surveys – particularly those that, unlike the NES, do not attempt to insure that the respondents are representative of the national population. Hence, wherever probability weighting materially affects the results, I instead employ the standard logit transformation. Unfortunately, the latter procedure does not allow an assessment of the statistical significance of variations in predicted probabilities.

21

In this instance, probability weighting produced fairly substantial effects. Hence, I am unable to employ the King et al. (2000) procedure. Nonetheless, since the base category for talk shows is significant at the .01 level, we can be confident that, at minimum, the relationships shown in Figure 1 are statistically significant among the least educated respondents.

22

Cronbach's alpha reliability score for the TV news index is .69.

23

A factor analysis indicated that newspaper and political talk radio consumption do not load well on a common underlying factor with local and national TV news. The TV news variables (national network, early evening local, and late evening local), however, do load strongly on a common underlying factor, at .65, .58 and .50, respectively.

24

Zaller (1992) found that this variable performs as well as or better than any other NES-based indicator of political awareness.

25

Due to the limited number of observations and large number of interaction terms in the latter models, I separate my analyses of the effects of soft and hard news consumption. The results, however, differ only modestly if all interaction terms (6 in total) are included in a single model.

26

Unfortunately, because only 10% of highly politically informed respondents adopted the isolationist position, it is difficult to draw meaningful inferences concerning the effects of talk show consumption on isolationism among political sophisticates. Moreover, the isolationism question was asked of only half of the respondents. As a result, there are an inadequate number of highly informed isolationists (nearly 50% fewer than any of the other categories) to allow a meaningful estimate of the substantive effects of talk show consumption on this sub-group. Hence, I do not report probabilities for this group.

27

In two Pew Center surveys (2001 and 2002a), non-high school graduates differed only modestly from college graduates in their support for U.S. military action against Iraq – with (.59 vs. .60) or without (.76 vs. .70) the assumption that the conflict would produce thousands of U.S. casualties – as well as in their support for using military force to fight terrorism (post 9/11) in Somalia (.74 vs. .76) and Sudan (.80 vs. .74) and for keeping U.S. troops in Afghanistan to maintain civil order (.67 vs. .70).