Foreign Policy Worldviews and US Standing in the World

Transcription

1 Foreign Policy Worldviews and US Standing in the World By Matthew A. Baum (contact author) Harvard University John F. Kennedy School of Government 79 JFK Street Cambridge, MA Phone: and Henry R. Nau George Washington University Abstract: What do Americans think about the US role in world affairs and why do they think as they do? Existing scholarship identifies some general attitudes Americans hold toward world affairs, rejecting isolationism and favoring multilateralism, but few studies explore more specific attitudes such as assessments of US standing in the world (defined as foreign views of America s capability, credibility and esteem abroad). American National Election Study data from provide one such data point, which shows a strong correlation between party identification and attitudes toward US standing defined as weakness. When Democrats occupy the White House, Republicans generally see US standing falling. The reverse holds true when Republicans hold the White House. Past studies conclude that this correlation is primarily a matter of partisanship and domestic political ideology (conservative vs. liberal). In this article we investigate a deeper and more novel explanation rooted in the independent influence of individuals foreign policy worldviews. Respondents assess US standing based on nationalist, realist, conservative and liberal internationalist views of the world. Across multiple statistical investigations, we find that while party ID remains a powerful heuristic for defining attitudes toward standing, foreign policy worldviews also exert a distinct influence on such attitudes, especially for more politically sophisticated respondents. Prepared for the panel on State Development, Representation, and Grand Strategy in the Great Powers, at the annual convention of the American Political Science Association, New Orleans, Louisiana, August 30, 2012.

2 What do Americans think about the US role in world affairs and what causes them to think that way? Americans typically do not think about foreign policy most of the time, and as a consequence know relatively little about it (Almond 1950; Lippmann 1955; Erskine 1963; Edwards 1983; Sobel 1993; Holsti 2004; Page and Bouton 2006, Berinsky 2007). While foreign policy issues can become salient when major international events arise (like 9/11 and the Iraq War) or when political candidates focus on foreign policy (Aldrich, et al. 1989), ceteris paribus, Americans know and care more about domestic politics (Delli-Carpini and Keeter 1996; Holsti 1994 & 2004). Consequently, typical Americans are broadly aware of foreign policy, and have some available attitudes about it (Page and Bouton 2006; Aldrich et al. 1989). However, except in the face of political priming by elites or exogenous shocks, such attitudes may not be broadly accessible when making political decisions, like voting. Scholars have made substantial progress understanding the general attitudes the American public holds toward foreign affairs. At least since World War II most Americans have consistently rejected isolationism (Kull 2001; Holsti 2004) in favor of robust U.S. engagement with the world. Moreover, Americans tend to prefer multilateral over unilateral approaches to foreign policy (Page and Bouton 2006; Holsti 2004; Todorov and Mandisodza 2004). Although these attitudes appear firm, they disguise substantive subtleties such as the fact that most Americans, while preferring to act multilaterally, assume that their fellow citizens prefer to act unilaterally (Todorov and Mandisodza 2004) and that multilateral policies generally produce results that conform to what the U.S. would prefer if it acted unilaterally (Stewart and Bennett 1991). Scholars know much less about American attitudes toward more specific aspects of world affairs, such as US standing in the world. This was the subject of a APSA Presidential Task Force (APSA 2009). The Task Force s final report defines US standing as an attribute 1

3 assigned to the United States by other actors such as foreign leaders and peoples, international organizations, transnational groups, and of course, assessed by American voters (3). Standing in this sense has many aspects but the report emphasizes two in particular: credibility and esteem. Credibility refers to the U.S. government s ability to do what it says it is going to do. This dimension captures the reputation, or standing up, concerns that have long dominated studies of deterrence as well as U.S. leadership more broadly. Esteem refers to America s stature, or standing for, for other countries and the American image in international politics (APSA 2009). In other words, the Task Force defines standing as the assessment of America s role in the world by other countries based on what America does (that is, credibility of its commitments) and what America is (that is the esteem in which it is held). Notably, this definition omits the possibility of defining standing primarily in terms of respect for American power (what America has), whether America is weaker or stronger as in the ANES data. How do Americans assess or think about this concept of US standing? Do they define it as the Task Force did, or do they see it through multiple lenses some defining it primarily in terms of diplomacy/credibility, others in terms of values/esteem, and still others in terms of security/power? And what factors shape these assessments? One of the most significant statistical finding over time (going back well before the Iraq years), illustrated in Figure 1, is the strong link between party ID and assessments of US standing, defined as relative weakness or power. Republican respondents consistently see US position in the world as weaker than Democrats during Democratic administrations, and Democratic respondents consistently see US position in the world as weaker than Republicans during Republican administrations. [Figure 1 here] 2

4 This pattern raises the question of whether this association is a simple matter of party reflexivity, or whether party ID serves as a heuristic for other factors influencing American views of US standing. It also raises the question of whether Democrats and Republicans think differently about America s standing in the world because they belong to different parties, because they hold different domestic political philosophies (liberal/conservative), or because they espouse different foreign policy worldviews (e.g., nationalism, realism, conservative or liberal internationalism). This study explores the independent and interactive influence of these three variables party ID, domestic political ideology and foreign policy worldviews in affecting assessments of US standing by US citizens. By drilling deeper than existing studies, it finds that even after taking into account party ID and domestic ideology, foreign policy worldviews directly and significantly influence attitudes toward the standing of the United States abroad, especially for more politically sophisticated or knowledgeable respondents. Party ID and Foreign Policy Attitudes The evidence concerning the influence of party ID on foreign policy attitudes in general that is, not on specific attitudes such as US standing -- is weak. Page and Bouton (2006), for instance, report that party identification significantly mediated general attitudes for only three of twenty possible U.S. foreign policy goals they investigated. Party does tend to be highly correlated with domestic political ideology (consistently in the neighborhood of about.40 across 10 Pew Center surveys we sampled conducted between 2001 and 2006). However, as Page and Bouton (2006) report, once ideology is accounted for, party typically drops out as an influential factor mediating Americans attitudes regarding U.S. foreign policy goals. Domestic ideology thus 3

5 appears to do the heavy lifting of causality. Klinker (2005) reaches a similar conclusion (see also Rauch 2007), arguing that partisan differences on foreign policy issues, while sometimes statistically significant, are in most cases not particularly large. He finds similar patterns with respect to the goals of U.S. foreign policy, the means of achieving those goals, and Americans values, such as patriotism and national pride. He reports that the exception to these patterns is ratings of President Bush s foreign policy, where a large partisan gap is apparent. This is consistent with Jacobson (2006, 2007), who finds that the partisan gap associated with the Iraq War was far larger than for any prior U.S. military conflict. This suggests that when a debate regarding American foreign policy becomes highly polarized along partisan dimensions, party ID may become more consequential as a predictor of individual attitudes on that (and related) foreign policy issue(s). Busby and Monten (2008) offer some evidence in support of this latter conjecture, finding that while Americans have remained predominantly internationalist throughout the post-wwii era: parties have become more ideologically homogenous, more regionally concentrated, and more extreme in their voting patterns on foreign policy (465). In Congressional roll call studies, Kupchan and Trubowitz (2010) confirm this conclusion, but Chaudoin, Milner and Tingley (2010) dispute it. Our first hypothesis tests the evidence that partisanship may be a primary factor causing attitudes toward specific foreign policy issues such as US standing: H1: Typical individuals will view US standing as relatively higher when their own party controls the White House than when the other party does so. Domestic Political Ideology and Foreign Policy Attitudes As already noted above, the literature offers stronger evidence that domestic political ideology (liberal/conservative) influences individuals attitudes toward specific US foreign policy 4

6 issues. The most widely employed model that disaggregates American attitudes toward foreign policy in terms of domestic political ideology is the MI/CI index, where MI and CI represent militant and cooperative internationalism, respectively (e.g., Maggiotto and Wittkopf 1981; Holsti 2004; Holsti and Rosenau 1999). This work defines domestic political ideology, the independent variable, along two independent dimensions: economic issues (economic regulation, redistribution of wealth, etc) and social issues (promoting minority interests, opposition to death penalty, etc.). This classification yields four types of respondents in terms of domestic political ideology: liberals (liberal on both dimensions), conservatives (conservative on both dimensions), populists (conservative on social issues, liberal on economic issues), and libertarians (liberal on social issues and conservative on economic issues). This literature then defines specific foreign policy attitudes, the dependent variable, in terms of support for militant internationalist policies (use of force, defeating adversaries, believing in the domino theory, etc.) versus cooperative internationalist policies (disarmament, support for United Nations and foreign aid, etc.). Four types of respondents, in terms of foreign policy attitudes, emerge: isolationists (opposed on both dimensions), internationalists (support both dimensions), hardliners (oppose cooperative and support militant internationalism), and accommodationists (support cooperative and oppose militant internationalism) (Wittkopf and Maggiotto 1981). Correlating these measures of domestic political ideology and foreign policy attitudes across four surveys conducted between 1984 and 1996, Holsti (2004) finds that an average of 78% of liberals are accommodationists, who support CI and oppose MI, 34% of conservatives are hardliners who support MI and oppose CI, and another 41% of conservatives are internationalists who support both MI and CI. In other words, 75% of conservatives, but less than 25% of liberals, support MI. The MI-CI index has proven impressively reliable at predicting support or opposition to 5

7 U.S. approaches toward foreign policy in general. It may also apply to specific policy initiatives. For instance, Holsti (2004, 143) reports that accommodationists were about half as likely as hardliners to view the U.S. victory in the first Persian Gulf War as a great victory for the United States (45 vs. 83 percent) and over five times as likely to believe that the U.S. will be too ready to use military force and go to war again (56 vs. 10 percent). This suggests that there is, in fact, a strong domestic ideological component to respondent attitudes toward foreign policy issues, both general (use of force) and specific (Persian Gulf War). Summarizing the most recent research on this topic, Nincic and Ramos (2010) conclude that where international affairs are concerned, conservatives are more likely to favor self-regarding ends [nationalist] and punitive means [militant internationalism], with liberals more apt to endorse other-regarding objectives [internationalist] and policy means based on positive incentives [cooperative internationalism]. This suggests a second hypothesis: H2: Typical individuals whose domestic political ideology is conservative will view US standing as relatively higher when a Republican controls the White House than when a Democrat is in power, while liberal individuals will view US standing as relatively higher when a Democrat controls the White House than when a Republican does. Foreign Policy Worldviews and US Standing However, studies to date of the influence of domestic ideology on attitudes toward foreign policy are based on incomplete definitions of both domestic political ideologies and foreign policy attitudes. For example, Holsti s two components of domestic ideology economic and social views exclude a third potentially crucial political component political views toward the relative 6

8 importance of freedom vs. equality and limited vs. activist government. Conservatives generally prioritize freedom and limited government, liberals equality and activist government. Similarly, the foreign policy dimensions measured in the MI-CI index emphasize the means (military versus diplomatic) but not the goals of foreign policy. As indicated in the quote above, Nincic and Ramos (2010) find that foreign policy views also encompass foreign policy goals self-regarding or nationalist vs. other-regarding or internationalist as well as attitudes towards the use of force militant internationalism vs. cooperative internationalism. Could a more complete delineation of domestic ideology and foreign policy beliefs better account for the full range of Americans opinions regarding foreign policy, and especially US standing? Due to limitations in data availability, with one exception below, we leave for later research a refinement of the domestic ideology variable (liberal vs. conservative) to include attitudes toward government as well as economic and social issues. In this article, we elaborate and refine the independent variable of foreign policy worldviews to capture more systematically respondents different views about the goals and means of foreign policy. The concept of foreign policy worldviews suggests that Americans assess foreign policy outcomes such as US standing through different foreign policy perspectives or schools of thought. Some Americans, drawing from realist perspectives, assess US standing largely in terms of security threats and power (capabilities or what America has); others, drawing from institutional perspectives, assess US standing primarily in terms of diplomacy and multilateralism (credibility or what America does); still others, drawing from ideological perspectives, assess it more in terms of US democratic values, support for human rights and the 7

9 like (esteem or what American is). 1 The worldview from which one assesses standing determines in large measure whether one sees it rising or falling, and whether one considers that rise or decline as important. For example, American relative power increased in the early 1980s, the onset of the new Cold War under Ronald Reagan. But American diplomacy was widely criticized around the world. Domestic groups that defined US standing largely in terms of American power might have been inclined to see US standing as rising. Those that defined it more in terms of approbation of American diplomacy might have seen it as declining. Similarly, under George W. Bush, observers widely reported that American diplomacy was in ill repute. Yet American military power, as least as measured by the overthrow of the Taliban and Saddam Hussein regimes in 2001 and 2003, respectively, was arguably never greater. It is therefore possible, as we observe in Figure 1, that Democrats, who assess standing more in terms of diplomacy than power, were more troubled about America s standing in the world than Republicans because Democrats perceived a decline of American diplomacy, while Republicans, who assess standing more in terms of power than diplomacy, were less troubled than Democrats because Republicans perceived a rise in American power. Thus the foreign policy worldviews of individuals may be a broad mediating factor between domestic political ideology and US standing. When a Democrat is in the White House, Republicans disapprove of that administration not just for domestic ideological reasons but also because they distrust the administration s judgment in foreign affairs, expecting it to depend too much on diplomacy and too little on force, such that threats build up in the world and America s 1 Here we include a definition of standing in terms of US power which the APSA Task Force excluded or included only in relationship to US credibility (behavior) and esteem (values). 8

10 reputation suffers. Similarly, when a Republican is in power, Democrats oppose the administration for domestic ideological reasons but also because they distrust that administration s judgment in foreign affairs, expecting it to use too much force and too little diplomacy, increasing terrorism, unilateralism and consequently causing America s standing to fall. A third hypothesis follows: H3: Typical individuals worldviews (net of party and domestic ideology) will influence their opinions concerning US standing. Political Sophistication and Foreign Policy Worldviews Thus far, we have argued that many Americans possess worldviews of the sort identified above and at least sometimes bring them to bear in assessing the merits of U.S. foreign policy actions. But exactly which Americans bring such worldviews to bear, when they are likely to do so, and why? Worldview is a more cognitively demanding consideration than domestic ideology or party ID. Hence, to answer these questions we might need to differentiate among respondents in terms of which ones are more or less capable of employing it. One key factor that seems likely to mediate individuals capacities to employ worldviews in assessing foreign policy is their overall level of political sophistication. A vast literature in cognitive and political psychology shows that typical individuals rely on information shortcuts, or heuristic cues (Sniderman, Brody, and 1991; Popkin 1994), including the opinions of trusted political elites (Iyengar and Kinder 1987; Zaller 1992; Rahn 1993) and party ID (Rahn 1993; Popkin 1994; Nelson and Garst 2005). Individuals interpretations of heuristic cues depend in significant measure on their preexisting belief systems (Hurwitz and Peffley 1987; 9

11 Hermann et al. 1997), for which party ID is typically an important (Rahn 1993; Popkin 1994; Lupia and McCubbins 1998; Nelson and Garst 2005) if incomplete (Hermann et al. 1999; Holsti 2004) element. The party affiliations of information sources (e.g., elites) and receivers (e.g., citizens) in interaction thus serve as a cognitive filter, mediating the selection and implications of the information shortcuts typical individuals rely on in making political judgments. 2 Party ID is an important heuristic because it is highly accessible. Nearly every adult citizen possesses it, and nearly everyone understands its implications, at least in general terms. Party ID and its implications are continually reinforced by political elites seeking to maintain the value of the party brand as a signal representing a set of policies a given candidate will likely tend to support or oppose. A worldview can also be thought of as an information shortcut, allowing an individual to assess the likely merits of a policy without necessarily delving into all of its details. However, relative to party ID, worldviews are less universally recognized than party ID, less frequently primed or reinforced by elites, and require a great deal more information to comprehend or apply to particular circumstances. This makes them more costly to employ as a heuristic for assessing a foreign policy activity. It further raises the questions of who is likely to employ such a demanding heuristic cue, given the ready availability of cheaper ones, and under what circumstances are they likely to do so? Consistent with prior research (e.g., Holsti 2004), we argue that political sophisticates individuals who pay a lot of attention to and understand politics are more likely than political novices who neither attend to nor understand politics to employ worldview as a heuristic. 2 For an investigation into the effects of partisan cues on public opinion regarding U.S. foreign policy, see Baum and Groeling (2010). 10

12 Sophisticates are more likely to possess the means and motivation to employ the more complex and hence demanding heuristic of foreign policy worldview. After all, if the goal is to assess the relative merits of a foreign policy activity, then one is likely to do a better job by employing a heuristic that carries with it a great deal of topically pertinent information. The more general, catch-all brand of party ID is simply less apt for this purpose. A fourth hypothesis follows. H4: Politically sophisticated individuals will rely more than political novices on worldview in assessing U.S. standing. However, if human beings are cognitive misers, expending the minimum necessary effort to reach the appropriate decision (Zaller 1992), then it seems unlikely that even political sophisticates would always elect to employ a cognitively demanding heuristic. Rather, they seem likely to prefer to match the complexity and precision of the heuristic to the task at hand. If there is significant doubt about the merits of a policy, it is more likely to be worth the effort to employ a complex heuristic. Conversely, if there is relatively little doubt about a policy, then a simpler heuristic, like party ID, may suffice. How can we anticipate, ex ante, when individuals are likely to favor a more high demand/high precision heuristic (worldview) over a more low demand/low precision one (party ID)? One obvious answer concerns the ex ante probability that a given individual will be inclined to support or oppose a given foreign policy activity. As noted previously, partisans are highly likely ex ante to assume, absent information to the contrary, that their fellow partisan presidents will pursue policies, including foreign policies, consistent with their own preferences. 3 Hence, party ID that of the citizen respondent relative to that of the president -- is a sufficient heuristic 3 For instance, from the Eisenhower to the George W. Bush administrations, an average of over 80% of presidents fellow partisans have approved of their job performance, compared to only 49% of opposition partisans. 11

13 when an individual shares the partisanship of the president. However, when the opposition party holds the presidency, then a typical individual is unlikely to be willing to assume that the president s policies are consonant with his or her own interests. In addition, partisanship is likely to offer a less clear signal, all else equal, in foreign relative to domestic policy, as Americans tend to know and care less about the former (see previous citations). Thus, on foreign policy, individuals may seek to ground their objections to the opposition party incumbent president in the logic of foreign policy worldviews rather than pure partisanship. They may feel obliged to present an opposing or alternative foreign policy view to that of the incumbent administration. Or they may simply find that their party affiliation or domestic ideology does not provide a clear roadmap for assessing the opposition party president s foreign policy activities. In this case, such individuals might conclude that it is worth the cost to employ the more demanding cognitive heuristic of worldview in order to more accurately determine whether they ought to support or oppose the president s foreign policy initiatives. A final hypothesis follows: H5: Members of the non-presidential party will base their assessment of U.S. standing relatively more than members of the presidential party on their personal worldviews. Delineating Four Distinct Foreign Policy Worldviews As noted above, the MI/CI matrix captures foreign policy attitudes toward the means of foreign policy whether respondents favor primarily militant internationalism (MI) or cooperative internationalism (CI). It does not include attitudes toward the goals of foreign policy that is, whether respondents believe that foreign policy should aim primarily for balance of power goals such as security or value goals such as the spread of democracy. If we include both 12

14 goals and means, a two-by-two matrix emerges. Figure 2 distinguishes among four distinct foreign policy worldviews: nationalist, realist, conservative internationalist, and liberal internationalist. Nationalists tend to emphasize power or security goals and military means to achieve those goals. Realists also emphasize power or security goals but now favor more activist diplomatic means to achieve these goals. Conservative Internationalists favor more ambitious goals of spreading democracy and more assertive military means to achieve them. And, liberal internationalists pursue the goal of spreading democracy but prefer to reduce the role of military means and strengthen diplomatic institutions to achieve this end. These distinctions are matters of emphasis not exclusion. But relatively (as in the case of the distinction in domestic ideology between liberal and conservative), worldviews differ across the two dimensions of foreign policy goals and means. 4 [Figure 2 here] Each worldview assesses US standing differently. Nationalists assess US standing primarily in terms of respect abroad for American power, realists primarily in terms of respect abroad for American alliances (power) and diplomatic skills (credibility), conservative internationalists primarily in terms of progress abroad toward promoting democracy and defeating militants, and liberal internationalists primarily in terms of support abroad for international institutions, economic development and the rule of law. Table 1 summarizes the 4 These traditions are variously labeled but well established in the history and study of American foreign policy (Perkins 1952; Nordlinger 1995; Nau 2002; Jentleson 2007). The conservative internationalist tradition includes neoconservatives, who are more hawkish than liberal internationalists, but also more idealistic realists and nationalists who favor the promotion of human rights and democracy. 13

15 key features of each worldview. [Table 1 here] Statistical Analyses We investigate three distinct datasets to explore the effects of the aforementioned worldviews on U.S. standing, independently and in combination with one another. This allows us to test our hypotheses against several, independently derived operationalizations of our independent and dependent variables, and thereby, we hope, make it possible to support our argument more strongly than would be possible through any single empirical test. Unfortunately, available survey questions limit our ability to test consistently for all four sets of worldviews. In our first two analyses, based on 1997 and 2004 surveys by the Pew Center for the People and the Press, we investigate all four worldviews, but in the third, which focuses on the 1996 and 2004 American National Election Studies (ANES), out of necessity we collapse the four views into a linear scale with nationalists at one end, conservative internationalists and realists in the (relative) middle, and liberal internationalists at the other end. In addition, the several data sets implicitly define standing, our dependent variable, differently. The dependent variable in the ANES data assesses US standing in terms of whether America is weaker or not, a seeming reference to American power; the Pew data assess US standing in terms of whether America is more or less respected, more of a reference, arguably, to US diplomacy and values than American power. We take these differences into account in our interpretations of the statistical results. 14

16 Pew Center 2004 Survey In July 2004, the Pew Center conducted a broad survey on Americans attitudes toward foreign policy, a time during which the war in Iraq dominated public opinion regarding foreign policy. Our analysis of this study tests H1 through H3. In this survey, respondents party ID and political ideology correlated significantly with attitudes toward Iraq. Based on a six-question scale we constructed (see Appendix Table A3), party and ideology correlated with attitudes toward the Iraq war at -.68 and -.40 respectively, indicating, unsurprisingly, that liberals and Democrats are less supportive of the war than conservatives or Republicans. In other words, if we assume that party ID and domestic ideology precede attitudes toward Iraq (as seems likely for most individuals), then partisanship, and to a somewhat lesser extent ideology, are extremely strong predictors of attitudes regarding Iraq. 5 (We measured worldviews based on scales explicated in Tables A1 and A2 in the Appendix.) The correlation analysis reveals extremely strong relationship between party and support for Iraq, and a fairly strong relationship between ideology and support for Iraq. The relationships between these three variables on the one hand, and worldviews on the other, are more varied. For 5 We also considered the influence of party and ideology on worldviews, as well as that of worldviews on attitudes toward Iraq. The Iraq support scale correlations are as follows: Ideology (-.41); Party (-.68), Liberal Internationalists (.06), Nationalists (.18), Realists (.49), and Conservative Internationalists (.57). The Ideology (from conservative to liberal) correlations are as follows: Party (.45), Liberal Internationalists (.01), Nationalists (-.23), Realists (-.38), and Conservative Internationalists (-.39). Finally, the Party (from Republican to Democrat) correlations are as follows: Liberal Internationalists (-.01), Nationalists (-.13), Realists (-.36), and Conservative Internationalists (-.41). 15

17 the nationalist and liberal internationalist scales, the correlations are fairly weak (.18 and.06, respectively). The correlation is stronger for realists and conservative internationalists. The realist scale correlates with the Iraq support scale, ideology, and party at.49, -.36, and -.38, respectively. The corresponding correlations for the conservative internationalist scale are.55, -.42, and -.39, respectively. This indicates that conservative internationalists and realists are far more likely than liberal internationalists and nationalists to support the Iraq War, as well as substantially less likely to be ideologically liberal or affiliated with the Democratic Party. For liberal internationalists and nationalists, however, party and political ideology account only modestly for worldview (with party and ideology measured on standard right-to-left continua). In short, worldview seems to be influenced by, but also vary independently from, party and ideology. To isolate the significance of worldviews from the partisan salience of Iraq, we next turn to a regression analysis, controlling for Iraq attitudes (as well as other variables such as party ID, domestic ideology, interest in politics and socio-economic characteristics). Our model thus included a series of questions and coding results (shown in Tables A1 and A2) from which we constructed indexes for each of the above-referenced worldviews 6 as well as a two-part question that we employed to measure attitudes toward U.S. standing. 7 6 For our worldview scales, we normalized each item in Table A1 to a 0-1 interval and then summed all items within each worldview category. This produced four scales, based on differing numbers of elements. We standardized the four scales, by normalizing each to a 0-1 interval. 7 We constructed a 5 category scale, coded -3 if respondents believed U.S. respect in the world was declining and that this was a major problem, -2 if respondents believed U.S. respect was declining and it was a minor problem, -1 if they believed U.S. respect was declining and it was not a 16

18 To validate our coding of the 19 questions included in our scales, we surveyed experts in U.S. foreign policy. We contacted all 600 members of the APSA Foreign Policy Section listserve. We received 81 completed responses 8, which we summarize in Table A1. As Table A1 indicates, the experts overwhelmingly validated our coding (with but one exception out of 19 questions non-proliferation, where conservative internationalists not liberal internationalists were most aggressive). This increases our confidence in the validity of our scales. 9 problem, 0 if they believed the U.S. was as respected as in the past or if they responded don t know, and 1 if they believed the U.S. was more respected than in the past. The two questions were: (1) Compared with the past, would you say the U.S. is MORE respected by other countries these days, LESS respected by other countries, or AS respected as it has been in the past? (Coded: 1=More respected, 2=Less respected, 3=As respected as in the past, and 9=Don t know/refused); and (2) ASK IF LESS RESPECTED. Do you think less respect for America is a major problem a minor problem or not a problem at all? (Coded: 1=Major problem, 2=Minor problem, 3=Not a problem, and 9=Don t know/refused). Because testing revealed that the values on this scale are not perfectly ordinal and evenly distributed, we employ a multinomial logit estimator. 8 We believe the relatively high dropout rate which increased over the course of the survey -- was primarily due to its substantial length. 9 Alpha reliability scores and factor loadings for the four scales are as follows: liberal internationalist scale =.64 (factor loadings range from.72 to.87), conservative internationalist scale =.68 (factor loadings range from.37 to.53), realist scale =.58 (factor loadings range from.30 to.53), and nationalist scale =.62 (factor loadings range from.24 to.57). Each of these scales thus approaches or achieves the standard level of acceptable scale reliability, while the elements of the various scales load moderately well, and in the case of liberal internationalists 17

19 We expected, per H3, that in Summer 2004, liberal internationalists who assess US standing more in terms of support for international institutions would be most likely to see respect for America declining and that this constituted a major problem, while conservative internationalists who assess US standing more in terms of alliances (coalitions of the willing) and defeating militants would be least likely to hold this view. Realists and nationalists, we anticipated, would fall somewhere in between, with realists perhaps being somewhat more concerned than nationalists. Table 2 presents the results from an unordered multinomial logit analysis testing H1 to H3. 10 In the top half of Table 3, in turn, we employ Clarify (King et al 2000) to transform the coefficients from Table 2 into expected probabilities. The results generally, albeit imperfectly, support our expectations, thereby suggesting that respondents worldviews do indeed matter when they are asked to assess U.S. standing. [Tables 2 and 3 here] As the top half of Table 3 shows, moving from one standard deviation below to one standard deviation above the mean score on the liberal internationalist scale with scores on all other scales held constant at one standard deviation below the mean -- is associated with a 23 percentage point increase in the probability of believing that the U.S. is less respected (than before 2004) and that this is a major problem (p<.01), a.12 point decline in the probability of believing the U.S. is less respected and that this is a minor problem (p<.01), a 3 point decline in quite strongly, on common underlying factors. Additionally, among our foreign policy experts, the average standard deviations are equivalent to only about ¼ of the variability of the question scales. This reflects the relatively high level of agreement among our experts regarding the relationships between the four worldviews and responses to the 19 individual questions. 10 The models in Table 2 exclude several extreme residual outliers. 18

20 the probability of believing the U.S. is less respected, but this is not a problem (p<.10) and an 8 point decline in the probability of believing that U.S. respect in the world has not changed (p<.05). This represents the strongest set of effects across the four worldview groups. Also as anticipated, conservative internationalists represent almost the mirror opposite of liberal internationalists. The corresponding increase on the conservative internationalist worldview scale is associated with a 26 percentage point decline in the probability of believing that the U.S. is less respected (than before 2004) and that this is a major problem (p<.01) and a 27 point increase is believing the U.S. is less respected, but that this is only a minor problem (p<.10). For both effects, the sign is reversed relative to liberal internationalists. In this instance, realists resemble liberal internationalists. A two standard deviation increase on the realist scale is associated with a 21 percentage point increase in the probability of believing that the U.S. is less respected and that this is a major problem (p<.10) and an 18 point decline in the probability of believing the U.S. is less respected and that this is a minor problem (p<.05). The effects are weaker for nationalists. The only statistically significant effect among nationalists is a 4 point decline in the probability of believing that the U.S. is more respected than in the past (p<.10). These results support H3 and are largely consistent with our expectations concerning the likely views associated with the four worldview groups on U.S. standing circa Consistent with H1, relative to Independents, Republicans the incumbent party in are 7 percentage points less likely to believe the U.S. is less respected and this is a minor problem (p<.05) and 7 points more likely to believe the U.S. is about as respected as in the past, though this latter effect is statistically insignificant (p<.15). Conversely, and also relative to Independents, Democrats the opposition party in are 7.4 points more likely to believe the U.S. is less respected in the past (p<.10) and 6.2 points less likely to believe the U.S. is about as respected as 19

21 in the past (p<.10) Interestingly, and contrary to H2, net of worldviews and Iraq attitudes, domestic ideology appears in these data to wield no significant effect on views regarding the trend in U.S. respect. This appears consistent with our earlier conjecture that worldviews may underlie at least some of the apparent effects of ideology on attitudes toward US standing. However, support for the Iraq war in these models is, as we previously demonstrated, strongly collinear with ideology. Consequently, attitudes toward Iraq and worldviews may be absorbing much of the independent effect of ideology and presumably to a lesser extent, party -- on attitudes toward US standing. It is not difficult to imagine why liberal internationalists and realists, who were unhappy with the Bush Administration s handling of the Iraq conflict from the beginning and even more so by 2004, would view U.S. standing defined in terms of respect for and influence of U.S. diplomacy and values abroad -- as on the decline. But why would nationalists be less likely to believe that the United States was more respected in 2004 relative to the past? Recall that nationalists base their concerns over U.S. standing on respect for American power and selfdefense, rather than approval of US diplomacy and values. By 2004, sophisticated nationalists were unhappy with the perceived change of focus in the Iraq conflict from self-defense (eliminating weapons of mass destruction) to democracy promotion. This factor, combined with the apparent open-ended nature of the commitment, may have caused them to conclude that American power, and hence standing, was declining. Regardless, these results are consistent with the patterns identified in the initial analysis shown at the top of Table 3. Taken together, these initial results generally support our expectations and appear to justify including worldviews in our theoretical model as a distinct causal variable. For additional tests of our model, we now turn to our second Pew survey analyses. 20

22 Pew Center 1997 Survey Our second analysis focuses on the first year of the Clinton Administration s second term in office. For this analysis, the dependent variable is captured by the following question: Do you think the United States plays a more important and powerful role as a world leader today compared to ten years ago, a less important role, or about as important a role as a world leader as it did ten years ago? Standing is defined in terms of some combination of power and importance (credibility/esteem). With a Democrat in office, our first two hypotheses anticipate that Democrats and liberals should be more positive regarding the importance of the U.S. role than Republicans or conservatives. With respect to H3, which anticipates an independent effect for worldviews, we expect nationalists to believe that the U.S. role was less important than before because there was no longer a major threat to U.S. security. We further anticipate that realists will believe that the U.S. role was less important or about the same because, while threats are less, the United States is the only great power actor to secure global peace. Conservative internationalists, in turn, should believe that the U.S. role was more important because the United States had unprecedented power and opportunity to maintain American security and ideological preeminence. Finally, we expect liberal internationalists to believe that the U.S. role was more important because the United States was the leading supporter of international institutions and the rule of law. The response options to our dependent variable question form a scale, which we coded as follows: -1=less important, 0=as important or don t know, and 1=more important. Our four worldview scales are based on a similar, albeit somewhat different, set of questions as those employed in the 2004 worldviews scales. (In Table A4 in the Appendix we present the questions 21

23 and coding for the four scales employed in this analysis.) 11 As with the 2004 survey, we employ an unordered multinomial logit estimator, which unlike an ordered logit -- does not assume that the various categories in the scale forming the dependent variable are both ordinal and equally distributed. Table 4 presents the results from our analysis testing H1 to H3. 12 In the bottom half of Table 3, in turn, we employ Clarify (King et al 2000) to transform the coefficients from Table 4 into expected probabilities. The results once again largely support our expectations, further indicating that respondents worldviews influence assessments of U.S. standing independently from party and domestic ideology. As the bottom half of Table 3 shows, moving on the liberal internationalist scale from one standard deviation below to one standard deviation above the mean score with scores on all other scales once again held constant at one standard deviation below the mean is associated with an 11 percentage point increase in the probability of believing that the U.S. is playing a more important role than 10 years before 1997 (p<.10), and a 14 point decline in the probability of believing the U.S. is playing about as important a role as in the past (p<.10). In this 11 Alpha reliability scores and factor loadings for the four scales are as follows: liberal internationalist scale =.71 (factor loadings range from.62 to.80), conservative internationalist scale =.69 (factor loadings range from.28 to.57), realist scale =.68 (factor loadings range from.34 to.59), and nationalist scale =.64 (factor loadings range from.27 to.55). Each of these scales thus approaches or achieves the standard level of acceptable scale reliability. The factor loadings, in turn, indicate that for the most part the elements of the various scales load moderately well, and once again in the case of liberal internationalists quite strongly, on common underlying factors. 12 The models in Table 2 exclude several extreme residual outliers. 22

24 instance, realists occupy the opposite extreme. The corresponding increase on the realist scale is associated with a 41 percentage point increase in the probability of believing the U.S. is playing a less important role than in the past (p<.05) and a 16 point decrease in the probability of believing the U.S. is playing a more important role (though the latter effect is not statistically significant, at p<.15). Conservative internationalists are 9 points less likely to believe the U.S. is playing a less important role than in the past (p<.05), while nationalists are 11 points more likely to believe the U.S. is playing a less important role (p<.01) and 13 points less likely to believe the U.S. role is about as important as in the past (p<.05). These results generally accord with H3. Compared to ten years before, nationalists and realists are much more likely to believe that the United States is playing a lesser role in the world, while liberal and conservative internationalists are more likely to believe it is playing a more important role. Realists seem even more likely than nationalists to favor a lesser role but that may be due to the prominence in 1997 of the debate over NATO expansion, which many realists opposed given the absence of a Russian threat to Europe. The strongest proponents of NATO and a more important U.S. role in the world were in fact the conservative and liberal internationalists, as anticipated by our hypotheses. Unfortunately, the lack of a question on domestic ideology in this survey prevents us from testing H2. However, we can test H1, which focuses on the effects of partisanship. In this instance, the results offer clear support for the hypothesis. Republicans the opposition party in are 8 percentage points more likely than Independents to believe that the U.S. is playing a less important role than in the past (p<.01) that is, 10 years prior to and 5 points less likely to believe the U.S. is playing a more important role (p<.05). Conversely, Democrats the incumbent party in are six points less likely to believe the U.S. is playing a less important 23

25 role than in the past (p<.01) and five points more likely to believe that the U.S. is playing a more important role (p<.05). We turn next to our analysis of the 1996 and 2004 American National Election Studies vs American National Election Studies (ANES) Analysis While ANES includes consistent questions about party ID and U.S. standing from 1958 on, it includes relatively few other questions that could be coded to worldviews over this same period of time. Of eleven questions we identified that could be coded in terms of worldviews, only five were asked across Democratic and Republican administrations, allowing us to investigate the effects of variations in partisan control of the presidency (Democrat Bill Clinton in 1996 and Republican George W. Bush in 2004). Unfortunately, these few questions (listed in Table A-4 in the Appendix) were insufficient to delineate four distinct worldviews. We thus created a scale by adding together the five questions each normalized to a 0-1 interval, with liberal internationalist at the low end, realists and conservative internationalists in the (relative) middle and nationalists at the high end. On this collapsed scale, nationalists define the goals of US foreign policy in the most limited way national or at best hemispheric defense and favor the strongest reliance on military force to fend off attacks but not to establish world order or spread democracy. Conservative internationalists also favor an assertive use of military force but now seek more expansive goals to spread democracy by regime change. Realists are more cautious in the use of force and restrict US goals to pursuing world order, rather than regime change. And liberal internationalists favor spreading American democracy but primarily through multilateral institutions and mutual disarmament rather than through the assertive use of force (particularly unilaterally). Thus, as we move along this scale, 24

26 foreign policy preferences shift from global to national scope, from multilateral institutional to military means, and from democracy to security goals, with the exception that on the latter dimension conservative internationalists fall closer to the liberal internationalist end of the spectrum than realists. To measure U.S. standing, the best available question in the ANES surveys was the following During the past year, would you say that the United States' position in the world has grown weaker, stayed about the same, or has it grown stronger? (VCF9045). As we noted above, the reference here appears to relate more directly to America s power than to its diplomacy (credibility) or values (esteem). As with our Pew data, we validated our coding of the five questions employed in our ANES worldviews scale through a survey of members of the APSA Foreign Policy Section listserve. We received 228 responses to this much briefer survey, which we summarize in Table A5. As before, the expert results largely substantiated the implicit coding of worldviews underlying our scale (see Table A5), thereby increasing our confidence in its validity Alpha reliability testing indicates that, perhaps unsurprisingly given the limited range and diversity of available questions, the scale is not particularly reliable among low-sophistication respondents (excluding college educated respondents, α=.28), becomes somewhat more reliable among college educated respondents (α=.41) and even more so among college educated respondents rated by the interviewer as relatively high (a 4 or 5 out of 5) on the political information scale (α=.56). The scale reaches standard levels of reliability (α=.70) for college educated ideologues (that is, strong conservatives or strong liberals). (Highly educated ideologically oriented individuals are presumably the individuals best suited to employ relatively cognitively complex concepts like worldviews in assessing foreign policy.) 25