Who Is Anti-American in the European Union?

The term anti-Americanism has become common in public and academic debate in the last decade. Yet we have only limited knowledge of those who hold such views. From 2003, 2005, and 2006 Eurobarometer data, almost 20% of European Union (EU) respondents disapproved of U.S. policy in all five dimensions the surveys examined. Following the literature, this consistent opposition is defined as anti-American. Anti-Americans exhibit systematic differences in age, education, geographical location, policy preferences, and nationality. In addition, although anti-Americanism is associated with a preference for greater European independence, perhaps surprisingly, it is also linked to a desire for a less federal and hence less powerful Europe. In both sets of attitudes, to the United States and to the EU, there are substantial variations within countries between different types of locality, which reinforces the view that it is too simplistic to describe a country as being anti-American or being pro-European integration.


Introduction
After the election of President G. W. Bush and especially after 9/11, it seemed to some that the United States adopted radically changed attitudes to its allies and to its enemies. Allies were to be led, not accommodated, and alliances were to be temporary and purpose-specific. Enemies might be attacked preemptively, and their regimes changed, and if possible democratized in an American approved style. The United States exhibited a new isolationism of perspective along with a wider, more vigorous projection of its power (Iwama, 2004).
Unsurprisingly, this apparent major change created a European crisis of trust, which is now discussed openly by politicians, commentators, and scholars. We want to examine some solid data that could help us determine the strength of European citizens' views of the United States at this apparent point of change, and to allow an analysis of their different dimensions, and the characteristics of those who held them. In particular, we want to explore whether opposition to American policies was accompanied by a desire for a more integrated and powerful European Union (EU), one that might become a countervailing economic and political power to the United States. This leads us to try to link-at least empirically-two hitherto largely separate literatures: that on anti-Americanism with that on attitudes toward EU integration.
The data are drawn from three Eurobarometer surveys for the European Commission: those of March-April 2003, October-November 2005, and September-October 2006. The 2006 survey is the most recent available that includes the 2003 survey's questions on EU respondents' attitudes to American policy. For the 2003 survey, during whose collection period, the Iraq War began, some 23,000 citizens in the then 15-member union were questioned, using face-to-face home interviews in their national language. In the 2005 and 2006 surveys, almost 27,000 citizens in the now 27-country EU were surveyed. 1 The sample frame was those aged 15 and above, but we want to study the views of those who could influence policy, that is voters, so we analyze respondents aged 18 and above. The surveys' multi-stage random probability design ensures a representative sample with respect to metropolitan, urban, and rural areas. For each 584163S GOXXX10.1177/2158244015584163SAGE OpenLawson and Hudson research-article2015 1 University of Bath, UK 2 Matej Bel University, Banská Bystrica, Slovakia interviewee, the data included age, sex, educational level, income, and nationality, plus some attitudinal variables.
We focus on the EU electors' answers to two sets of questions: 1. Whether the respondent felt the United States played a positive, negative, or neither a positive nor negative role in (a) the search for world peace, (b) the fight against terrorism, (c) the fight against world poverty, (d) the protection of the world environment, and (e) world economic growth. We are less interested in the responses to any specific question than in the totality of such responses. In particular, building on Berman (2004), we define an anti-American mind-set as one that consistently disapproves of U.S. policies across a range of different policy dimensions. 2. Whether the respondent felt that independent EU defence, security and foreign policies were desirable. We look for possible links between respondents` answers to these questions and their views on U.S. foreign policy.
We aim to answer three basic questions: Research Question 1: Is there clear evidence of the existence of anti-Americanism as we define it? Research Question 2: Is anti-Americanism random, or is it linked to certain personal characteristics, for example, nationality, level of education, or residence in a particular part of a country? Research Question 3: Do those with anti-American views also share an identifiably similar set of opinions on either the foreign or security policies of the EU or its future political development?
The "Literature Review" section of this article discusses the definition of anti-Americanism and surveys the literature. It also reviews the literature on voters' attitudes to European Integration. The "Attitude Determination" section provides the justification for the model of attitude determination that underlies the empirics. The "Data and Analysis" section gives the results, and the "Conclusion" summarizes our conclusions.

Literature Review
Here, we define anti-Americanism and review the existing literature on its extent and causes.

Defining Anti-Americanism
As O'Connor (2012) observes, this term has been widely and sometimes loosely used. What is needed is a definition that corresponds to observable phenomena. For example, Spiro (1988) defines anti-Americanism as "a persistent pattern of gross criticism of the main values of the US Constitution" (p. 120). This is both too vague, for the values need to be defined, and too specific. For example, why is it confined to the Constitution?
Alternatively, Krastev (2004) defines anti-Americanism as "opposing any policy simply because it is endorsed by the US government" (pp. [6][7]. This is an appealing definition, because it incorporates a significant theoretical distinction between opposition based on the effects of American policy and opposition to the policy because it is American. However, Eurobarometer data do not allow us to establish why voters have specific views. Other definitions include "any hostile act or expression that becomes part and parcel of an undifferentiated attack on the foreign policy, society, culture and values of the United States" (Rubinstein & Smith, 1988, p.36) and "a psychological tendency to hold negative views of the United States and of American society in general" (Keohane and Katzenstein, 2007b, p.273).
Berman (2004) provides a more tractable definition that focuses on opposition to specific American policies, or groups of policies. This allows a search for clusters of opposition, across policies, countries, or respondents, without prejudging their cause. Such a definition is value neutral and operational. Blaydes and Linzer (2012) empiricalize this by averaging the normalized responses to seven questions relating to whether people had favorable views of the United States and Americans, attitudes to the spread of American customs, American ideas about democracy, ways of doing business culture, and scientific advances. Chiozza (2009a) also defines anti-Americanism as an attitude and a political belief that can be measured through the answers individuals give to survey questions.
We think it is worth extending the range of opinions that are used to assess the pattern of anti-Americanism, by including policy effects as well as policies. We also narrow our focus onto those who hold consistently negative views of U.S. policies or of their effects. Our measure of anti-Americanism is similar to that of Blaydes and Linzer (2012) in being based on several different responses. But unlike theirs, it is not a continuous variable but a binomial one based on those who consistently have a negative opinion of the United States. We extend the range to both policies and policy effects, because it is not clear that respondents distinguish between them, and Eurobarometer questions involve both. We confine our description of anti-American to those who have consistently negative views in all five of the examined U.S. policy dimensions because we want to explore links with their views on the EU. Opposition to an individual policy could arise for a myriad of reasons, and have little or no linkage to common concepts of anti-Americanism. In addition, if there is nothing unusual in our narrowly defined anti-American group's pattern of views on the EU, then in the survey data, there are likely to be few interesting links between the wider group of European voters' views of the United States and their views on the future of the EU.

The Literature on Attitudes Toward the United States and Its Policies
Opinion poll evidence for European anti-American views in the early years of this century, summarized by Linn (2004), revealed a growing negative view of the United States, though with marked differences across European states, and more positive views in recent EU members. Kohut (2003) quotes a 2001 State Department survey: Where 78% of Germans said they had a favorable view of the U.S. That fell to 61% in our (Pew Research Center) 2002 polland to 45% in the survey conducted (in spring 2003). Opinion of the U.S. in France has followed a similar track: 62% positive in 1999-2000, 63% (in 2002) . . . and 43% (in 2003). (pp. 1-2) However, according to both Linn and Kohut, although EU public opinion has changed, the U.S.-EU "value divide" is not widening. Linn (2004) defines "basic values" as "views regarding fundamental, life and death issues that are embedded in traditional culture and do not vary quickly or significantly over time" (p. 13). Such values might be the roles of the individual and the state, belief in God, or the necessary conditions for a just war. These he contrasts with "'public opinion' [which although it] refers to views and perceptions mainly on matters of current politics or events . . . " can and generally does "also cover and reflect basic values" (Linn, 2004, p.13). These authors believe that the recent growth in anti-Americanism reflects changing European public opinion rather than a basic value change.
However, even if there has been no profound change in European values, some American commentators have taken a very combative line. For example, Ceaser (2003) claims that anti-Americanism in Europe influenced Islamic terrorists. He traces anti-Americanism back to the 18th-century biological degeneracy thesis that all species, including human beings, thrived less well in the New World. By the late-18th century, this then dominant view also appeared in a political variant that asserted the inferiority of the American political system because it was not grounded in tradition. For Ceaser, anti-Americanism . . . is a construct of European thought . . . anti-Americanism makes rational discussion impossible, it threatens the idea of a community of interests between Europe and America. Indeed it threatens the idea of the West itself. (p. 16) One does not have to accept Ceaser's extreme account to agree that the United States's power, and hence its pervasive influence, has created the preconditions for distrust of its motives. The U.S. response to 9/11, and especially the invasion of Iraq, produced widespread European opposition. The more radical criticism went beyond policy disagreement and purported to examine alleged flaws in the American psyche that led to the violent and intimidating behavior that, some argued, has marked America since its inception (Sardar & Davies, 2002. A common view in the literature argues that attitudes to the United States are strongly influenced by views of American leaders and what they do, but also by views of the American people (Chiozza, 2009b, p. 283). A related view claims that non-Americans can and do distinguish between "what America is" and "what America does" (Keohane & Katzenstein, 2007a, p. 2). The latter argue that because America encompasses a very wide range of values and lifestyles, it provides foreigners with many reasons for attraction and repulsion. This helps explain the phenomenon of those who may oppose its foreign policy but still wish to work there.
The degree of distrust of America varies across European countries, most plausibly because of past contacts. However, opinions are constantly evolving. Germany and France provide two examples. Although relatively pro-American, Germans have recently become more distrustful. Berman (2004) claims, . . . there is a specific German continuity from pre-1989 Communist anti-Americanism to post-Communist anti-Americanism, which has been particularly relevant, given the role of the former Communist Party-the Party of Democratic Socialism (PDS)-and its ability to influence the larger German political landscape. (p.29) Other influences he cites are the . . . association of Americans with "capitalism in a negative sense"; a " . . . pre-democratic anti-Americanism (that) finds expression in contempt for aspects of American mass culture," and "a post-democratic anti-Americanism . . . driven by resentment that the United States has been unwilling to cede sovereignty to the structures of international governance." (p.30) However, French opinion is arguably the most anti-American. Meunier (2005) argues that " . . . France is the country with the deepest, most sedimented reservoir of Anti-American arguments" (p. 127). Toinet (1988), a French commentator sympathetic to the United States, argued that after its critical help to the United States in the War of Independence, France looked in vain for a "special relationship" with the new state. Instead, relations were punctuated by intermittent tensions. For example, France interpreted the 1898 American attack on the Spanish empire in the Caribbean and the Philippines as a threat to all European empires. The United States did not forgive France its First World War debts. Roosevelt initially favored Vichy over the Free French and wanted a retributive postwar military government for France. But even such an unpromising legacy cannot explain the severity of some criticisms. Toinet's explanation is that French commentators see a clash between the ideals of French civilization and American advocacy of its view of democracy. For the commentators, American democracy has led to and entails an unacceptably rapacious and uncaring form of capitalism (Grunberg, 2005;Todd, 2003).
It is possible to construct a systematic and wide-ranging explanation of anti-Americanism by focusing on its origins in domestic European politics. For example, Fabbrini (2002) notes that " . . . in (continental) Europe, anti-Americanism seems to be one of the few public philosophies that can unite large sections of the left, the right and the Catholic Church" (p. 3). He argues that there are clear, though different, reasons why these diverse movements should exhibit this philosophy. On the left is the legacy of the Cold War, former communist parties explicitly opposed to the United States, and socialist parties opposed to the perceived excesses of American capitalism. On the right are parties that are distrustful of American liberalism, individualism, and antipaternalism. The Catholic Church's skepticism he argues originates in the key role of protestant sects in American religious life, the separation of Church and State in the American constitution, and the individualism that lies at the heart of American identity. Despite this, the evidence presented in, for example, Keohane and Katzenstein (2007b) shows little evidence of any consequences of anti-Americanism on the United States itself.
In terms of specific empirical econometric analysis, Chiozza's (2009a) analysis of people across the world is of particular importance. He shows that variation in anti-American sentiment is explained more at the individual level rather than at the level of the country. He also finds that no single variable is of primary importance in explaining anti-Americanism and that the impact of socio-economic characteristics varies in different parts of the world. Finally, he shows that anti-Americanism, except perhaps in some Islamic countries, is not an entrenched philosophy around the world. Blaydes and Linzer (2012) in their study of anti-Americanism in the Islamic world link anti-Americanism to both individual socioeconomic characteristics and country-level variables. Haas (1961) argues that European co-operation would lead, through a "spillover" mechanism, to unintended consequences that would eventually lead to firms, governments, political parties, and citizens to orient their expectations and political behavior toward the EU. Moreover, the citizens of European countries would gradually come to view "Europe" as the natural focus for their political activities and loyalty (Haas, 1968). Thus, integration would gradually produce a convergence of beliefs, values, and aspirations that would unite the peoples of the European community. Haas (1976) did later become more skeptical of this view, as it seemed as if progress in the EU to further integration had slowed down, although the subsequent development of the single market, the Schengen Agreement, EU enlargement toward the east, and the birth of the euro expanded co-operation over both functions and territory (Fligstein, Polyakova, & Sandholtz, 2012). However, as Fligstein et al. (2012) also note, this increased co-operation does not seem to have been matched by a growth in public support for the EU or a shared European identity. In part, this may be because issues such as immigration, the "war on terror," and slow economic growth and the financial crisis have caused citizens to focus more on their national governments. But also following the 1992 Maastricht Treaty, the balance in the EU has shifted from economic toward political and cultural integration (Fligstein et al., 2012). For some Europeans, such developments and the expansion in the EU institution's role in governance threaten national identity while formalizing a European supranational identity (Fligstein et al., 2012). This may have resulted in the EU being looked on less favorably by some citizens (Shore, 2000). But equally perhaps, it may have led others to identify more with the EU; that is, it may have polarized views.

The Literature on Citizens' Attitudes to European Integration
These are important developments because there is evidence that less supportive citizens are more Eurosceptic (Usherwood & Startin, 2013). In addition, governments are unlikely to support further political, social, or economic integration if their citizens are not supportive of the EU (Down & Wilson, 2013). Hence, an understanding of the determinants of public support for the EU and for European integration is also important if we are to understand the direction the EU is likely to move in over the coming decades. Verhaegen, Hooghe, and Quintelier (2014) argue that supporting greater EU integration involved a cost-benefit calculation, a view empirically supported by Gabel (1998) and Sanchez-Cuenca (2000). This implies that the winners from EU integration will tend to be in favor of integration, and support the EU itself, whereas the losers will tend to hold opposite views. Consistent with this perspective, Fligstein et al. (2012) argue that European economic and cultural integration have tended to involve those from the higher end of the socio-economic spectrum. However, blue-collar, lower white-collar, service, and older workers have benefited less and will tend to favor their own governments over the EU. In terms of empirical work, Verhaegen et al. examine perceptions that EU membership is good for the country and perceptions of a European identity, rather than attitudes to further integration per se. They find personal characteristics such as age and gender to be significant in determining support for the EU, as well as occupational variables, the individual's financial position, and country variables linked to the economic situation, the date of joining the EU, and Eurozone membership. The pattern of significance tends to suggest, as indicated above, that the winners, in terms of those who are doing well, tend to support the EU more than others. Kuhn and Stoeckel (2014) analyze support for European economic governance, again using survey data. In practice, their dependent variables are averaged perceptions of whether a greater EU role in economic governance in several dimensions would or would not be effective, rather than whether the individual would support such a development. The most significant variable is, unsurprisingly, trust in EU institutions.
Other socio-economic variables, for example, age, gender, education, and occupation, are not that significant. Kritzinger (2003) presents an alternative view to the economic self-interest based approach and emphasizes the political, economic, cultural, and historical contexts in which attitudes developed. The key role of national concerns in European parliamentary elections supports this hypothesis (Franklin, Van der Eijk, & Marsh, 1995). Lacking specific EU knowledge, voters use their evaluations of domestic politics (Anderson, 1998). Much of their general EU information is also filtered through national politicians. Garry and Tilley (2009) also argue that apart from a cost-benefit calculation, attitudes are influenced by national identity. But there is little agreement on the nature of this influence. Carey (2002) links a strong sense of national identity with anti-integration views. Haesly (2001) finds the opposite. Garry and Tilley reconcile these conclusions by arguing that the impact depends on the country's situation, for example, with respect to income per capita. In passing, we note that their control variables included age, sex, employment status, and education level, finding support for the EU to be associated with being relatively highly educated, middle class, male, and satisfied with the country's economic performance.
Ordinary citizens may link their views of national and EU politics in one of two opposed ways. The "equal assessments" hypothesis suggests that people interpret both from similar viewpoints, perhaps because national politicians also participate in inter-governmental meetings (Anderson, 1998). A counter perspective sees national and EU politics as alternatives, even rivals: Those disillusioned with one may see the other as a more attractive way forward. Using 1994 Eurobarometer data, Kritzinger (2003) concludes that "the EU is used as a proxy for symbolic protest against the present national political and economic situation" (p. 237).
The impact of anti-Americanism on EU citizens' views on defense policy has been analyzed by Ray and Johnston (2007). They used Eurobarometer data for 2003 and 2004, for the EU-15 members, and concluded that anti-Americanism was positively associated with support for EU, rather than NATO-led defense policy. They also noted that " . . . anti-Americanism does push some Europeans away from multilateral defense policymaking altogether" (Ray & Johnston, 2007, p. 88). This latter group tended to favor a national approach to defense policy.

Respondents' Attitudes Toward U.S. Policies
Because anti-Americanism, the key variable that we analyze, is binary, we use probit regression. The literature review showed that there has been some empirical research into anti-Americanism, much of the early work focusing on differences between countries rather than on individuals within countries. However, more recent work has focused on individual attitudes across a range of countries (Blaydes & Linzer, 2012;Chiozza, 2009a;Chiozza, 2009b ). The novelties in this article are that it uses a different variable to measure anti-Americanism, one based on consistent opposition to American policies, rather than a continuous variable. It also analyzes the evolution of attitudes across several years. Finally, it includes the first empirical analysis that explores a possible link between EU citizens' opinions on U.S. policies and their opinions on EU integration. There has been speculation on whether anti-Americanism has had adverse effects on the United States, for example, on its political influence and economic interests. But there has been much less analysis of the effects on other countries.
A theoretical determination of citizens' views of the United States's policies is not set out formally here. Our purpose rather is to explore whether there are any interesting empirical regularities in the data. We do not presuppose that respondents' views are in the main irrational or prejudiced. Neither do we presuppose the contrary. A predisposition to take a positive or negative view of U.S. policy may be perfectly consistent with a set of goals that respondents could have articulated, but we cannot deduce from their recorded answers. We neither assume nor can deduce their answers reflect any predisposition.
The primary independent variables of interest are those relating to the individual's characteristics, the country fixed effects, and the individual's policy preference variables. We focus on four policy issues: terror, the economy, defense, and the environment, using a simple measure that reflects whether the individual thought they were important. These are included because, for example, individuals' attitudes to U.S. policies may in part depend on the relative importance they place on the environment as compared with terrorism as policy issues. Country-specific attitudes will differ, and again memory will be relevant, although this time, it is likely to be the collective memory of a nation that, as noted above, determines attitudes to the United States.
Following common practice in analyzing political attitudes, and also consistent with some of the work we have referenced, our control variables include the socio-economic variables of age, education, sex, occupation, a measure of the respondents' household income relative to their country's average household income, and a variable that reflects respondents' trust in big business. Underlying their inclusion is the belief that individual attitudes are partially determined by individual circumstances and knowledge. Not all these variables are available for all three surveys.
The variables are defined in the appendix. A brief rationale for their inclusion is as follows.
The literature survey indicated that views of U.S. policy or its effects may be influenced by respondents' views of the U.S. government, people, culture, power, and social and economic systems. Attitudes toward these factors have changed over time and location. Attitudes may be partly influenced by knowledge, including personal experience. To some extent, they may be related to social norms and hence reference groups, conventional wisdom, and even irrational bias. In turn, these influences can sometimes be linked to the independent variables used here.
For example, educated people may be more aware of issues such as the violation of human rights in Abu Ghraib, or climate change, which in turn could fuel a blanket opposition to U.S. policies. Variables such as age, sex, and education may be related to direct experience of American culture, including music, films, or fast food outlets that may be factors influencing respondents' attitudes to America. Age may also be relevant in providing a direct memory of periods such as the Cold War and the Second World War 2 when American power was often viewed more favorably than in the Bush era. Location can also influence exposure to American culture: For example, exposure to fast food outlets or U.S. multinationals is more likely in large cities than rural areas.

Respondents' Attitudes to European Integration
The focus is on policies with international spillovers, defense related foreign policy, and also the EU constitution. We follow Verhaegen et al. (2014) and others, and assume those who favor greater European integration feel that they will benefit from it-individually or altruistically, or both. Their views are uninfluenced by the appropriate service delivery level (Oates, 1999), because this is a given. International influence and defense capability increase with economic size. Hence, a united EU voice has more impact than any single member country, and possibly more than the sum of all members. But integration also entails a loss of sovereignty, and a country may become committed to support and finance policies it opposes.
Individual differences toward EU integration will thus arguably largely depend on how the individual perceives the national interest and the extent to which they are willing to trade national sovereignty for a greater global role. Such attitudes are likely to be driven by national pride and identity, and therefore, we include the socio-economic factors that are standard in this type of analysis as well as country fixed effects.
In addition, we also include attitudes to the United States. There are two possibilities. The first and most obvious is that anti-Americanism will drive European integration. This is consistent with an analysis of the institutional choices of EU constitutional delegates in the area of EU foreign policy (Jensen, Slapin, & Koenig, 2007). They found that delegates from states that had not deployed troops in support of the U.S.-led Iraq invasion were more likely to support EU integration. Their explanation was that such states would be more interested in creating a European foreign policy capable of counterbalancing the United States because they are more opposed to U.S. policy. But with individual citizens' attitudes, there is also the reverse possibility: that people are opposing a superpower and will not want to see a second superpower grow from the nations of Europe, although, or perhaps especially as, they will be part of that superpower.

Data and Analysis
We have already described the Eurobarometer sample frame and interview procedures. Our probit regressions use the survey's attitudinal and characteristics data to explore their inter-linkages. The variables are defined in the appendix.
The key variable is, of course, the one relating to anti-Americanism, and it is measured by sustained opposition to American policies across five dimensions of policies: peace, terrorism, poverty, the environment, and world growth.
Attitudes to the EU are measured in relation to voter support for a greater co-ordination of defense policy, foreign policy, an EU military rapid reaction force, a common crisis response position, and greater independence of foreign policy from the United States. In the two more recent surveys, data on the last three were not available, but there was information on support for an EU constitution. The constitution apart, these are predominantly foreign policy and defense issues that could be addressed by inter-state agreements. Hence, greater EU co-ordination implies a reduction in national sovereignty. Of course, to an extent, this is also the case for the EU constitution. In our analysis of the 2003 survey, the individuals' incomes were approximated by the midpoint of the income range in which the survey placed them. Their relative income was calculated as the ratio of this approximation to the country average. In the later surveys, data on income were not reported, but data on life satisfaction and, in the 2005 survey, on individual personal circumstances in the recent past were available and were used. Table 1 reports attitudinal variables. The first column in the top half of Table 1 records the proportion classified as anti-American in 2003, that is, disapproving of U.S. policy with respect to all five policy areas of peace, terror, poverty, environment, and growth. The alternatives include positive views and respondents with neither positive nor negative feelings. The proportions are quite high and much higher than would be the case if views were randomly distributed between alternatives across the different issues. Thus, in France, in the 2003 sample, randomly distributed attitudes would lead to 9.6% 3 disapproving of U.S. policy on all five dimensions, not the 34.7% we see, and in the United Kingdom, we would have seen less than 0.1% defined as anti-American not 9.2%.

Voters' Views on U.S. Policies
There are also considerable differences between countries for the three surveys (2003; 2005; 2006). At one extreme, we have France (34.7%; 28.1%; 27.4%), Spain (33.2%; 36.3%; 30.0%), and in particular Greece (68.0%; 64.1%; 62.9%) with a very high proportion of people classified as consistently disapproving of American policies. At the other extreme, Denmark (7.5%; 8.0%; 9.7%) and Sweden (10.2%; 8.2%; 10.4%) have only small minorities classified as anti-American. But across the 3 years, an average of seven EU-15  Table 2 reports the 2006 distribution of the population disagreeing with U.S. policies in terms of the number of issues. Ordering the distribution in this way reveals the wide range of approval or disapproval patterns across countries. Greece and Cyprus again stand out as having high levels of disagreement with the United States on all issues and the East European countries and Malta as having very low levels of disagreement. Ireland and Italy have high levels of zero disagreements, but a growing number of multiple disagreements. France, Germany, Finland, Belgium, and Luxembourg have few who are strongly supportive, but large numbers who are unsupportive on multiple issues. Austria, the United Kingdom, Spain, Portugal, and Slovenia show varying degrees of polarized disagreement patterns, with zero and five disagreements as the two modes. Spain and Portugal are strikingly polarized. Sweden exhibits the opposite pattern, with two as the mode; and the Netherlands's views are almost equally distributed across the categories. Table 3 gives the regression results for attitudes to U.S. policies. The use of squares of ages allows different age cohorts to have different reactions, perhaps reflecting the history of U.S. involvement in Europe, perhaps reflecting evolving self-interest. Not all of the empirical papers we have discussed included age squared as opposed to just age. But it does allow for attitudes to flatten out as people age and indeed even reverse. Note that because of the variation in some questions across the three surveys, replication of results cannot always be tested.  were dissatisfied with their life and whose personal situation had recently deteriorated tended to be more hostile to American policies. 4. People who lived in villages and rural areas tended to be less hostile to U.S. policies than those who lived in small towns, who again were less hostile than those who lived in cities.
Both the income and age impacts might have been expected, with richer voters and those with memories from Note. Values in parentheses denote t statistics. Equations estimated by probit in STATA. Country fixed effects included. χ 2 denotes the likelihood ratio statistic. The policy issues relate to respondents' concerns with these topics as issues, and so, differ from their perceptions of the United States on similar policy areas (see the appendix). The marginal effects, shown in columns 3.3, 3.6, and 3.9, are the average marginal effects. **and * denote significance at the 1% and 5% levels of significance.  Table 3. They have been normalized to a lower bound of 0 for Lithuania.
the height of the Cold War more supportive toward the United States. The United States's association with capitalism may attract the wealthy. Attitudinal variables are significant along several dimensions. Concern with terrorism tends to increase broad support for U.S. policies. The environment is also a key issue of increasing salience, with those who felt the environment to be important disapproving of American policies. But even more than the environment, attitude to big business is a defining issue. In the 2 years for which we have data, mistrust of big business is highly significant in increasing disapproval for American policies. Columns 3.3,3.6,and 3.9 show the marginal effects, that is, the impact on the probability of being anti-American of a change in the variable. Hence, we can see that for someone who lives in a village, the probability of being anti-American declines by .0212 in the 2006 regression, compared with someone who lives in a large city. Hence, if a city dweller's characteristics are such that the probability of being anti-American is .2012, for someone with the same characteristics who lives in a village or rural area, the probability would be just .180. The figure for 2005 is very similar. These are average responses and give a misleading impression of impact for age. Thus, for the 2006 regression, if we evaluate the impact at age 25 and 65, then the marginal effects are 0.0013 and −0.00079, respectively, both significant at the 1% level. These indicate that as someone moves from 25 to 26, the probability of being anti-American increases by approximately .0013 and as someone aged 65 moves to 66, then the probability declines by .00079. These are relatively small changes, but if we look at what happens when someone ages a decade, then the change from age 25 to 35 is to increase the probability from .168 to .178. Similarly, when someone ages from 65 to 75, the relative probabilities are .179 and .169, respectively.
The country coefficients reflect differences in attitudes compared with a base country-here Finland. The choice of Finland has no substantive impact on the results. Choosing a different country would have altered the size of the country dummy coefficients, but not their relative ordering. These country fixed effects are not shown in the tables, but Figure  1 reflects their average marginal effects values from the regressions 3.2 and 3.5 of Table 3. 4 Hence, given all other variables, it reflects an attitude to the United States, which characterizes the country as a whole. Of the EU-15, the least favorable to the United States are Greece, Spain, Luxembourg, and France. Sweden and Denmark are more favorable. In general, the new entrants are more favorable to the United States with the exception of Slovenia and Cyprus. But it is clear on this that "new Europe" is less disapproving of the United States than "old Europe." The difference between France and the United Kingdom has narrowed throughout the 3 years, with both countries showing some convergence. Nonetheless, based on the 2006 regression, 3.1 for a "representative individual," 5 the probability of being opposed to U.S. policies on all 5 issues is .23 in the United Kingdom and substantially greater at .31 in France.
We also show at the foot of the table the different log likelihoods from (a) excluding country fixed effects and (b) including just country fixed effects. The log likelihoods of the former are consistently much closer to zero, indicating higher explanatory power than the latter. This implies that most of the variability in responses is between countries rather than across socio-economic characteristics. Table 1 also shows basic data on respondents' attitudes toward EU policies. Specifically, it shows the proportion who approved of greater EU integration or co-operation out of all those surveyed, including "don't knows." In 2003, a large majority of respondents in all EU countries favored foreign policy independence from the United States. Other foreign and defense issues follow a similar pattern. The only exception was U.K. respondents' views on a more coordinated foreign policy, with slightly less than half of those questioned in favor. This general pattern of support was largely repeated in 2005 and 2006 even among the new EU entrants.

Voters' Views of EU Policies
In Table 4, we focus on the relationship between respondents' attitudes to the United States and their support for European integration. The regressions have been estimated by ordered probit. In the 2003 sample regressions, the binary variable, "anti-American," reflecting negative attitudes along the five dimensions of support for the United States, is significant. That is, those classified as anti-American are more likely to support an EU foreign policy more independent of the United States's. They are, however, less likely to support EU defense policy co-ordination and also foreign Note. Values in parentheses denote t statistics. Equations estimated by ordered probit in STATA. Country fixed effects were included. χ 2 denotes the likelihood ratio statistic. The policy issues relate to people's concerns with these topics as issues, and so, differ from their perceptions of the United States on similar policy areas (see the appendix). **and * denote significance at the 1% and 5% levels.
policy co-ordination, the creation of a rapid reaction force, and holding common positions in the face of crises. 6 Turning to the other variables, support for greater EU independence from the United States increased with the level of education and was higher for men. Men were also more favorable to an expanded EU role in foreign and defense issues, including a rapid reaction force. Country dummies (not shown) suggest that Denmark and Finland were the most hostile to the force, with Greece, Belgium, and France being the most favorable. Responses reflect both attitudes to integration and willingness to project military force. On the individual policy preferences, concern with the economy tended to be associated with support for a stronger EU. Finally, mistrust of national governments and, to an extent, big business was linked to reduced support for a stronger EU.
In the 2005 and 2006 samples, there were questions on attitudes to defense and foreign policy but not on the other issues. However, there was an additional question on the EU constitution. The results were largely consistent with those for 2003 with respect to the impact of anti-Americanism. They were also similar with respect to education and attitudes to the economy, although the latter was less significant in 2006. Concern for the environment and terrorism was, however, significant in affecting attitudes in both 2005 and 2006. In addition, those whose personal situation had deteriorated tended to be more hostile to a strengthening EU. Distrust of big business and government also tended to reduce support for EU integration.
We again show at the foot of the table the different log likelihoods from excluding country fixed effects and including just country fixed effects. In this case, the two are much closer together than for the anti-Americanism regressions. This implies that socio-economic characteristics are relatively much more important compared with country characteristics in determining attitudes to the EU than in explaining anti-Americanism. This result may well reflect the fact that the EU has a much greater direct impact on European citizens than American policies.

Conclusion
We defined anti-Americanism as consistent opposition to U.S. policy and focused our study on those EU voters who took a negative view of that policy in all of the five dimensions covered in the Eurobarometer surveys. Our three questions were as follows: Research Question 1: Is there clear evidence of the existence of anti-Americanism in the EU electorate? Research Question 2: Is anti-Americanism random, or is it linked to certain personal characteristics? Research Question 3: Do those who consistently disapprove of U.S. policies also share an identifiably similar set of opinions on the foreign policy or future political development of the EU?
The first of these questions may seem like something of a non-issue given the amount of work that has been done on this. But our results are based on a different measure of anti-Americanism than has been used before. In addition, our results show evidence for both anti-Americanism and pro-Americanism sentiments within the countries of the EU, although we focus almost solely on the former. But we do not find, as does Chiozza (2009a), that individual characteristics are more important than country ones, although, of course, this relative importance will depend on the heterogeneity within and between countries in the sample analyzed. Nonetheless, our results show strong evidence of anti-Americanism linked to age, education, policy preferences, national origin, and in particular, mistrust of big business. The latter strongly suggests that a root cause of anti-American attitudes is anti-capitalism. If so, the financial crisis and economic stagnation of 2008-2014 are unlikely to increase support for the United States, notwithstanding the considerably greater popularity of President Obama outside America, compared with that of his predecessor.
In "old Europe," anti-Americanism is less common in the United Kingdom, 7 Italy, Ireland, and Sweden. Greece, France, and Spain stand out in having consistently more unfavorable attitudes. As we noted, this pattern is consistent with these countries' historical experiences. However, attitudes to the United States are in general considerably more positive in "new" than in "old Europe." EU expansion does appear to have made its members, at least on average, less hostile to U.S. policy.
Excepting Ireland, this pattern of country support also matched explicit initial support for the Iraq war. It may also have adversely affected American multinationals and U.S. exports (Anholt & Hildreth, 2004). The lesson is that a "hard power" victory can come at the expense of a "soft-power" defeat (Nye, 2004).
European hostility to America also has implications for the EU. We showed that EU voter anti-Americanism leads to a demand for less dependence on the US, though not for an EU to rival America. Indeed, the reverse is the case-disapproval of American policies was not linked to a desire for greater European defense and foreign policy integration, nor for an EU constitution. Hence, we have identified an impact of anti-Americanism, not on the United States itself or its policies, but on the EU. In the main, EU voters who feel American policies have been bad for the world do not want a more powerful EU. This may be because the EU is perceived as closely aligned with the United States, or because the United States represents an example of what the EU could become. It may, however, be that in the future, the EU does develop in a way that brings it more into conflict with the United States both politically (McNamara, 2011) and economically. In this case, it is possible that anti-Americanism will be associated with support for a more integrated EU.
Finally, it is important to stress that there are important differences across the countries between urban and rural areas. Therefore, it can be an oversimplification to label a country as being pro-EU integration or anti-American policy. Our analysis has also shown that within country differences have a much more similar impact on attitudes to between country differences, than was the case for anti-Americanism. Hence, if the EU is to increase support for integration, then it has to focus on both levels, that is, on those within countries who are skeptical and on those countries that are most skeptical. In comparing our results with others, the diversity of empirical definitions that have been used to proxy both anti-Americanism and EU integration should be borne in mind.
The conclusions we have reached provide an extensive and nuanced answer to the question in our title "Who Is anti-American in the European Union." We relied on large scale and repeated surveys of public opinion. But it is worth recalling an important limitation of this type of analysis. The data do not allow us to distinguish opinion from distrust or indeed bias and prejudice. That limitation can be important if the views we have recorded influence EU or EU member states' policy. The obvious extension to our research would be to explore the impact of European anti-Americanism on European policy and relations with the United States.