Taking Foreign Policy Personally: Personal Values and Foreign Policy Attitudes

Previous research shows that, when it comes to foreign policy, individuals have general orientations that inform their beliefs toward more speciﬁc issues in international relations. But such studies evade an even more important question: what gives rise to such foreign-policy orientations in the ﬁrst place? Combining an original survey on a nationally represen- tative sample of Americans with Schwartz’s theory of values from political psychology, we show that people take foreign policy personally: the same basic values that people use to guide choices in their daily lives also travel to the domain of foreign affairs. Conservation values are most strongly linked to “militant internationalism,” a general hawkishness in international relations. The value of universalism is the most important value for predicting “cooperative internationalism,” the foreign- policy orientation marked by a preference for multilateralism and cosmopolitanism in international affairs. This relatively parsimonious and elegant system of values and foreign-policy beliefs is consistent across both high- and low-knowledge respondents, offering one potential explanation for why those people who are otherwise uninformed about world politics nonetheless express coherent foreign-policy beliefs.

A large body of scholarship holds that values not only play a role in the mass public's attitudes toward domestic politics, but that they also help people make sense of attitudes toward international affairs. In particular, existing research finds that ordinary Americans have orientations that inform their attitudes toward more specific issues in international relations (Hurwitz and Peffley 1987;Herrmann et al. 1999). Scholars generally agree that these orientations originate in underlying values. Hurwitz andPeffley (1987, 1105) and Rathbun (2007, 379) write of "core values;" Herrmann et al. (1999, 553) assert a role for "core dispositional values;" and Chittick et al. (1995, 314) stress the importance of "value orientations." Against the venerable Almond-Lippmann view that sees foreignpolicy beliefs as shapeless and incoherent (e.g., Almond 1950;Lippmann 1955), we now know that foreign policy attitudes have structure and that values supply the mortar that holds them together.
Despite this progress, however, international-relations scholars lack a systematic understanding of, first, which values matter and, second, how they contribute to the architecture of foreign-policy views. In one strand of literature, scholars infer values indirectly from the patterns of association among specific foreign-policy attitudes (Chittick et al. 1995). However, the lack of serious attempt to measure values directly and to estimate their impact on foreign-policy postures limits the utility of this work. Another vein of scholarship avoids these pitfalls but assumes that the more abstract values that drive the formation of foreign-policy preferences on specific issues are unique to the foreign-policy domain. That is, it treats these values as structuring only views on international affairs and not other aspects of individuals' social interactions. This seems implausible (Hurwitz and Peffley 1987).
We hypothesize that foreign-policy dispositions derive from values that structure not only political life but social life in general. A number of social psychologists identify models of values based on two universal human needs. These needs often conflict with one another. On the one hand, societies need to provide physical safety to their members: to protect their group from internal and external threats. On the other, societies also need to foster consideration for others and reciprocal exchange so as to reap the gains of cooperation-even with others outside the group. Values serve both of these needs. We utilize the Schwartz (1992) value framework. Not only does it seem most capable of subsuming other similar models, but it is also the most prominent model found in the recent literature. According to it, the "conservation" values of conformity, tradition, and security work to create the solidarity and cohesion necessary to guard against internal and external threats. "Self-transcendence" values foster prosocial bonds between individuals-leading them not only to care about, but also cooperate with, others.
We find evidence of a relatively parsimonious and elegant system of values and foreign-policy beliefs in which different clusters of Schwartz values predict the core components of foreign policy orientations. Conservation values are most strongly linked to "militant internationalism," a general hawkishness in international relations. Universalism-a self-transcendence value that indicates an identification and concern for all human beings-is the most important value for predicting "cooperative internationalism," the foreign-policy orientation marked by cosmopolitanism and multilateralism. The values that matter for foreign-policy beliefs, in other words, are not just specific to the domain of international affairs. They are bigger than politics. Individuals take foreign policy personally.
Our integration of Schwartz into the study of foreignpolicy views makes two important contributions to existing scholarship. First, unlike much of the existing research in social psychology that focuses on one-off relationships between particular values and foreign policy preferences, we study the role of values as part of a broader system. As Kertzer et al. (2014, 828) argue, political scientists are "confronted with a cornucopia of values to choose from." This raises questions about why we should study one particular value but not another; the Schwartz value framework provides us with a unified and coherent framework. Second, we measure values directly, ex ante, rather than divining them ex post from the patterns of covariation found among foreign policy attitudes.
This article calls into serious question pessimistic readings of the mass public's ability to form political judgments without the help of elite cue-givers (e.g., Zaller 1992; Lupia and McCubbins 2000;Berinsky 2009). Previous research shows how foreign-policy predispositions allow individuals to derive attitudes on specific foreign-policy events or questions on which they lack good information (Rathbun 2007;Reifler et al. 2011;Kertzer and McGraw 2012). We demonstrate that those dispositions are themselves grounded in even more fundamental values. Converse (1964) was correct in arguing that many Americans are "innocent of ideology," but we show how these low-knowledge individuals prove just as able to connect their personal values to foreign policy preferences as their high-knowledge counterparts, despite the latter being more ideological than the former.
In the following sections, we first review the literature on core values in foreign policy. We highlight the two most important approaches: the cooperative/militant internationalism framework and the vertical hierarchy model. We then introduce the Schwartz value framework, derive hypotheses, and present results. Subsequently we discuss the implications of those results for prominent traditions in foreign-policy opinion studies.

Cooperative and Militant Internationalism
Two strands of research on the foreign policy attitudes of the mass public stress the importance of values. Perhaps the most prominent tradition in this research on foreignpolicy beliefs-offered by Wittkopf (1986Wittkopf ( , 1990Wittkopf ( , 1994a n d Holsti and Rosenau (1986, 1988-consistently finds that American foreign policy attitudes organize along two related, but distinct, dimensions: cooperative and militant internationalism. Cooperative internationalism describes the extent to which one embraces the world with open arms, militant internationalism captures beliefs about willingness to meet the world with a clenched fist. Cooperative internationalism is an orientation toward international affairs that stresses concern for others abroad, with whom one should work toward common goals. Chittick et al. (1995) discuss cooperative internationalism as a dimension capturing distinctions between self and other in international affairs. They write, "What all these [cooperative internationalism] questions seem to have in common is a concern for the wider community. We believe that those who emphasize the importance of these goals have a more inclusive identity than those who de-emphasize these same goals" (1995,318). Similarly, Nincic and Ramos (2010) write of "other-regarding" objectives, while Rathbun (2007, 338) defines this dimension as denoting a "sense of obligation to the broader international community." Global solidarity constitutes a key element of cooperative internationalism, but cosmopolitanism concerns more than just self-sacrifice and service to others: cooperative internationalists also believe that cooperation leads to mutual gains. Accordingly, previous work finds that support for international institutions, multilateralism, and international collaboration all load on the same cooperative internationalism dimension (Rathbun 2007;Wittkopf 1990;Rosenau 1988, 1990).
Militant internationalism, on the other hand, generally marks the familiar division between hawks and doves over the importance, effectiveness, and desirability of using force to reach foreign-policy objectives. Hurwitz and Peffley posit a "dimension of militarism ... anchored, on the one end, by a desire that the government assume an assertive, militant foreign-policy posture through military strength and on the other by a desire for a more flexible and accommodating stance through negotiations" (1987,1107). According to some, hawkish and dovish postures rest on different cognitive "models" about the effectiveness of force (Jervis 1976;Tetlock 1983). Hawks embrace the "deterrence model," in which strength and the demonstration of resolve best achieves peace. In this worldview, lack of credibility and signs of weakness invite challenges by aggressive foes in a dangerous environment.
Doves, in contrast, point out the often self-defeating nature of such displays, which risk inciting fear on the other side and escalating hostilities in such a way that leaves both sides worse off.
Scholars utilizing the cooperative internationalism/ militant internationalism framework sometimes also specify a third isolationist dimension in the structure of foreign-policy beliefs. This dimension involves a general support for the United States turning inwards and withdrawing from international affairs (Chittick et al. 1995;Rathbun 2007). An isolationist disapproves of both foreign military engagements and collaborative multilateral efforts to solve global problems. The cooperative/militant internationalism framework was originally developed to distinguish between two kinds of internationalists, each of which opposed isolationism. Early public-opinion research on foreign-policy attitudes tended to assume a single isolationist-internationalist dimension (Holsti 1979). However, isolationism stands apart from both cooperative and militant internationalism as a general belief that the United States should avoid political entanglements with other countries (Kertzer 2013). This difference receives support in the frequent detection of a third independent dimension in factor analysis. It involves more than just a low score on cooperative and militant internationalism, but instead represents a phenomenon in its own right (Rathbun 2008, 395). While historically-oriented scholars demonstrate that much of what observers construed as isolationism in 20 th -century American foreign policy actually amounted to nationalist unilateralism (Dueck 2006;Rathbun 2012)-which public-opinion scholars now call militant internationalism-isolationism still remains a discernible, albeit small, force in American foreign policy.
Analysts in this tradition generally assume that these foreign policy orientations are grounded in underlying values. But they never measure them directly. Instead, they surmise them from the clustering in factor analyses. Murray and Cowden (1999, 458), for instance, write of the "hidden organizing principles" that bring order and coherence to foreign policy belief systems. They are treated as latent values that explain the pattern of factor loadings. We argue that, instead, we should measure values directly in order to better understand their nature. While the general structure of foreign-policy beliefs remains wellestablished, existing studies remain ambiguous as to what underlies these postures.

Vertical Constraint Models
Another related strand of research places values at the top of a hierarchical structure of attitudes. They inform foreign-policy orientations-such as constructs similar to cooperative and militant internationalism-which in turn manifest themselves in specific foreign-policy preferences such as those concerning defense spending or cooperation with the United Nations. Hurwitz andPeffley (1987, 1105) identify two primary values at the top of the chain: "ethnocentrism" (essentially patriotism or nationalism) and beliefs about the morality of force.
Although these models represent an important contribution to our study of foreign-policy beliefs, the past several decades of research on value systems in social and political psychology suggest three related concerns. First, the reason why values are so analytically powerful is because they are trans-situational, working across domains (Jacoby 2006). In this sense, what we believe matters in our personal lives should influence our judgments in domains like foreign affairs as well. Yet classic hierarchical models argue instead for a silo-like understanding of political beliefs in which each domain has its own set of overarching principles from which everything else follows. They thus imply that our personal values are separate from our foreign-policy values (Hurwitz and Peffley 1987).
Second, and relatedly, because these models are domain-specific, they use values that are not very distant from the specific foreign-policy attitudes they are trying to explain. This choice leaves open the question of whether less proximate values may also shape opinions on international affairs. For instance, Hurwitz and Peffley (1987) argue that militarism, defined by preferring force over diplomacy, is a function of beliefs about the morality of warfare. Is the latter any more fundamental than the former? To truly examine how values affect foreign policy thinking, we need measures of values uncontaminated by the substance of foreign affairs. Grounding specific foreign-policy beliefs in more general personal values also makes it easier to distinguish cause from effect, a common complaint about the public opinion literature in foreign policy (e.g. Fordham and Kleinberg 2012).
Finally, the choice of core foreign-policy values in the vertical constraint literature comes across as somewhat ad hoc: beliefs about the righteousness of using force undoubtedly matter, but preferences about multilateral cooperation might be just as fundamental; how do we decide ex ante what counts as a value worth studying? Research that focuses on a handful of values in a given study necessarily elides broader questions about the universe of values, as well as where these stand in relation to one another. Insofar as value systems -as opposed to one or two values -matter for public opinion, this omission proves problematic. Feldman (2003, 480)

Schwartz's Theory of Values
We believe it is possible to improve on both strands of research on the structure of foreign-policy beliefs by using a more theoretically grounded model of values that explains decision-making across multiple policy domains. We integrate the insights of both these approaches while making improvements. Like the proponents of the vertical constraint model, we measure values directly and hypothesize how these more abstract principles might vertically structure more specific foreign policy postures: the orientations of cooperative internationalism, militant internationalism, and isolationism derived from other work. However, rather than taking an ad hoc approach to the study of values, we follow a more systematic approach made possible by Shalom Schwartz's universal model of values (1992Schwartz's universal model of values ( , 1994) that tests whether values unspecific to foreign policy nevertheless influence foreign policy postures.
Building on the work of Rokeach (1973), Schwartz (1994, 20) conceptualizes personal values as i) abstract beliefs ii) about desirable end states or behaviors that iii) transcend specific circumstances and contexts, iv) guide evaluation and behavior, and v) can be rank-ordered in terms of relative importance. Schwartz identifies 10 broad value domains: benevolence, universalism, self-direction, stimulation, hedonism, achievement, power, security, conformity, and tradition. The content and structure of personal values can be illustrated using the circumplex shown in Figure 1, which arrays the value types along a circular motivational continuum. Adjacent values share similar motivations (and should be positively correlated with one another); opposed values share conflicting motivations (and should be weakly or negatively correlated with one another). Values should have the highest positive correlations with those directly next to them in the circumplex, so that, for instance, achievement and power have a stronger positive association than achievement and security.
When the motivational bases underpinning the identified universe of personal values are considered en masse,f o urs uperordinate values emerge. As indicated in Figure 1,t h e capstone principles are (1) self-transcendence values, which emphasize the acceptance of and concern for other individuals (known and unknown), groups, society writ large, and even the global community; (2) self-enhancement values, which call for the pursuit of one's own self-interest, success, satisfaction and dominance, over others if necessary; (3) openness values, which privilege independent feeling, thought, creativity, and action; and (4) conservation values, which emphasize self-restraint, social stability, resistance to change, and deference to established traditions and cultural dictates.
To illustrate, tradition values call for deference to established familial, cultural, and religious norms and practices. They are compatible with adjacent conformity values that prioritize the goals of impulse control, self-restraint, and submission to social expectations. Together as conservation values they stand opposed to stimulation values, which prioritize excitement and novelty, and self-direction values, which emphasize thought and action free of socially imposed norms or internally imposed restraints. Hence, the realization of one value (e.g. self-direction) can obstruct attainment of another (e.g. tradition or conformity).
Over the past 25 years, Schwartz and others working in this vein have collected data from over 400 independent samples covering some 80 countries. Generally speaking, the model performs quite well in most applications.
Researchers find the empirical patterns in the data usually correspond to the hypothesized value structure (Schwartz 1992;Spini 2003;Davidov et al. 2008). Given the number of independent samples and the fact that they are drawn from diverse cultures, linguistic traditions, age groupings, probability and non-probability samples, and different points in time, it seems fair to conclude this model of value content and structures rests on a strong empirical foundation. As a result, the Schwartz value framework has become "the most widely employed model of values" in social psychology (Hitlin and Pinkston 2013, 322), and "the standard model in values research" (Gollan and Witte 2014, 452). At the time of this writing, the five most popular pieces by Schwartz developing this framework have been cited over 21,000 times. Thus, although the Schwartz value framework is by no means the only value framework within social psychology, it is among the most widely employed, and as Duckitt and Sibley (2009, 100) note, there is a striking similarity between Schwartz's factors and those utilized by others, which enables us to better develop hypotheses about the connection between core values and foreign policy orientations.

Schwartz Values and Politics
Psychologists see personal values as standards that guide perception, judgment, and behavior in all walks of life. Values operate at a higher level of evaluative abstraction than attitudes toward or beliefs about concrete objects; therefore, the former are well-positioned to shape the latter. To take some examples, values shape lifestyle choices, consumption decisions, altruistic behavior, social interactions, college majors, what people worry about, and many other judgments and behaviors (Bardi and Schwartz 2003;Roccas et al. 2002;Verplanken and Holland 2002). It is precisely because of values' centrality in political judgment that elites turn to value frames when they want to mobilize public opinion (Zaller 1992;Nelson et al. 1997;Brewer 2001).
Since people use values in a trans-situational manner to guide evaluation in all (or at least many) elements of their lives, the application of values to the formation of political preferences follows naturally. Indeed, Converse (1964, 211) anticipated this possibility when he noted "a few crowning postures-like premises about survival of the fittest in the spirit of social Darwinism-serve as a sort of glue to bind together many more specific attitudes and beliefs." Recent research has found that basic human values shape public opinion on domestic policy views and electoral choice in both the United States and in other countries (e.g. Caprara et al. 2006;Piurko et al. 2011;Schwartz et al. 2010). Conservation values of security, conformity, and tradition tend to predict support for the political right, and self-transcendence for the political left, for instance.
Application of the Schwartz framework to the foreign policy domain, however, has been much less systematic. No study to date applies the Schwartz value framework systematically to foreign policy, focusing instead on how individual values affect specific policy stances, such as the use of force in a particular instance, or at best how values affect one particular orientation, such as militarism (which might be considered synonymous with militant internationalism). We know for instance that self-transcendence values are important for understanding concern about human rights ( http://isq.oxfordjournals.org/ construct of cooperative internationalism, which encompasses concern for the human rights of those abroad, much less isolationism. In addition, most studies in this tradition do not rely on representative samples, although some do go beyond the undergraduate student body (Mayton et al. 1999;Schwartz et al. 2010;Bayram 2015).
Theoretical Expectations Schwartz (1992Schwartz ( , 1994 argues that values enable the smooth functioning and survival of groups and coordinated social interaction. Understanding values in this way helps us identify their likely effects on foreign policy dispositions. We also buttress those proposed links with findings from other value frameworks, most of which overlap significantly with that of Schwartz. Conservation values serve to create in-group solidarity to protect from internal and external threats. In order to provide for their safety, individuals must trade some degree of autonomy (Feldman 2003). Jost et al. (2003, 339) argue that an "existential" motivation underlies this set of values. Braithwaite (1997Braithwaite ( , 401, 1998 calls it the "security" dimension of social beliefs, attitudes, and values. As Schwartz notes, both conformity and tradition entail subordination of the self in favor of socially imposed expectations. They create social solidarity, making sure everyone is behaving properly and willing to sacrifice for the group. Tradition and security both stress preserving existing social arrangements that give certainty to life, and conformity and security emphasize production of order and harmony in social relations. Those who score high on these values also embrace particular moral foundations of in-group loyalty and deference to authority that serve the common goal of "binding" the group together (Graham et al. 2009(Graham et al. , 1029Graham et al. 2011, 366). Authority is necessary for keeping individuals in line and for leading a cohesive group against internal and external threats. Duckitt (2001Duckitt ( , 2006 and others (Duckitt et al. 2002;Van Leeuwen and Park 2009) demonstrate that those that embrace these values are driven by a conception of the world as a dangerous and threatening place. Feldman and Stenner (1997) similarly show that they go hand-in-hand with a sense of threat. Conservation values are marked by an "avoidance" motivation; they seek to prevent negative outcomes through protection and security rather than ensure positive ones through the provision of goods to others (even if avoiding negative happenings requires the provision of defense) (Janoff-Bulman et al. 2007, 1091Janoff-Bulman 2009, 120). Threats, of course, might be either external or internal.
We expect that these conservation values will have the greatest effect on support for militant internationalism, as the military and force are used to protect the in-group from threats. For those who hold conservation values, the use or threat of force would be a necessary element for controlling an unpredictable environment where there is no recourse to a higher authority. Thus, they will be relatively more hawkish. Conservation values mighta l s ol e a dt ol o w e rs c o r e so n cooperative internationalism. Conservation values bind individuals together in the presence of some external "other," which could inhibit a sense of global identity and cosmopolitanism. However, studies show that in-group identification and out-group derogation are related but not reducible to one another. For instance, greater patriotism does not necessarily lead to nationalistic feelings of group superiority (Kosterman and Feshbach 1989;Herrmann et al. 2009).
Of course, we recognize that conservation values also vary by the situational context. The motivational goals underlining right-wing authoritarianism (RWA) -concern for conformity, tradition, and security -rise in situations of acute threat (Duckitt 2001;Duckitt and Fisher 2003). Although it goes beyond the scope of our cross-sectional data, we know from foreign policy practice that there are often calls for greater conformity and tradition when international threats become more acute as a way of making society more cohesive and better able to face an external challenge.
Self-transcendence values indicate concern for and acceptance of others. They are the expression of one of the defining characteristics of human society, its high degree of altruism, which facilitates cooperation and allows for a higher degree of social organization (Cosmides 1989). Selftranscendence values are associated with different moral foundations than conservation values, those of fairness and caring for others (Graham et al. 2009;Graham et al. 2011). Braithwaite (1997Braithwaite ( , 1998) calls this the "harmony" dimension of values, oriented toward improving the welfare of other individuals. Janoff-Bulman et al. (2007,1091) identify these values as having an "approach" motivation. They drive individuals toward bringing about positive outcomes, "providing" for others rather than "protecting" them. Schwartz (1994, 22) distinguishes between benevolence, which is caring for those in one's local community, and universalism, marked by "understanding, appreciation, tolerance, and protection for the welfare of all people and for nature." Universalism should be associated with a belief in the interdependent nature of international relations and a commitment to the collective resolution of common problems, what we might call multilateralism. Bayram (2015) finds that self-transcendence values like universalism are predictive of cosmopolitanism. Universalism will also potentially lead to negative support for militant internationalism, as the imposition of force is harmful to its targets and universalists will demonstrate a greater concern for their welfare, even if they are not part of one's immediate ingroup. Nevertheless we expect that the primary effect of universalism will be on cooperative internationalism.
Benevolence on the other hand will likely lead in the opposite direction. Conservation values call for in-group solidarity; for this reason benevolence sits adjacent to the conservation set of values in the Schwartz circumplex that encourage deference to authority and conformity to common social standards. Those high on benevolence are therefore likely to score higher on militant internationalism and lower on cooperative internationalism for the same reasons as those who are high on conservation values.
To the extent that self-enhancement values -one's concern for one's own achievement and power -are directly opposed to self-transcendence, particularly universalism, we should expect that they predict support for militant internationalism but are antagonistic to cooperative internationalism. However, the relationship between what one values for oneself and how one relates to broader groupings is an open (and ultimately empirical) question. It might be, for instance, that one's desire for personal achievement does not come at the expense of one's commitment to others. Indeed one might desire power and achievement for oneself in order to help those in one's community and beyond. Since foreign policy is fundamentally about groups interacting with one another, it could be that the personal values that define individuals' relationships with broader groups and help coordinate behavior within them should matter most in determining foreign-policy beliefs. Values based purely on individual needs, such as achievement or power, would demonstrate less of an effect. In this view, values that serve the most basic (or even base) needs of individuals -such as individual status within a group or pleasure-seekingare the least likely to affect foreign-policy beliefs. Coveting status, positions of leadership, and respect will not necessarily translate into support for military superiority, for instance, since the latter concerns the actions of one's group rather than oneself.
Lying between self-transcendence and self-enhancement (and directly across from conservation) in Schwartz's circumplex are "openness" values such as self-direction and stimulation. To explain this positioning, consider that selfdirection is a value that justifies the pursuit of selfish interests. However, as a general principle applied to others it indicates a belief that all individuals should pursue their own conception of the good life. It is indicative of a prosocial, universalist approach to individuals that emerges from the Enlightenment, in which all humans are thought to have intrinsic worth.
Openness likely helps individuals expand in-group boundaries or at least venture beyond them. It allows individuals to look for mutually beneficial cooperation outside of one's immediate circle. Those who do so have been found to do better than those who confine their relations to those who are just like them (Yamagishi and Yamagishi 1994, 141;Messick and Kramer 2001;Rathbun 2012;Orbell and Dawes 1993;Orbell et al. 1984). We expect therefore that openness values will be predictive of cooperative internationalism. To the extent that openness values, which celebrate individuals, are directly opposed to conservation values, they should also lead to higher support for militant internationalism.
Scholars have had considerable difficulties identifying the individual-level attributes that consistently account for variation in isolationism. For instance, while liberalism is associated with higher levels of cooperative internationalism and lower levels of militant internationalism, political ideology explains little of the variance in this general desire for disengagement from the world. One might argue that we could disregard the dispositional causes of isolationism. In contemporary studies of the structure of foreign policy attitudes, it is always the weakest dimension, captured in statistical terms by a lower eigenvalue than the other two foreign policy orientations (e.g., Rathbun 2007, 100). Research also shows that it likely has strong situational determinants, whether changes in the White House (Urbatsch 2010), the economy (Fordham 2008), or the world stage (Kertzer 2013). However, we argue that even while isolationism might not be as much a dispositional phenomenon as other foreign policy postures, values might still help us understand why some are inclined toward removing their countries from the world.
We propose that isolationism is partially driven by a sense of threat and is a different strategy for creating security. Rather than preparing to fight so as to deter negative outcomes, isolationists might simply prefer flight to remove themselves from dangerous situations through selfsufficiency and non-involvement. Exemplified by what Mead (2002) calls the "Jeffersonian" tradition in US foreign policy, this strand of isolationism should therefore be associated with high levels of conservation and low levels of universalism. Indeed historically isolationism has been highly nationalistic, even chauvinistic in character and hostile to liberal schemes for multilateral order (Dueck 2006). At the same time, that desire for self-sufficiency likely makes isolationists feel less of a bond with their fellow countrymen, which would be expressed in higher scores on self-enhancement and lower scores on benevolence.

Hypotheses
Based on the previous discussion, we hypothesize a relatively simple and elegant structure of foreign-policy beliefs and personal values. Since it is generally accepted that foreign policy postures help people derive more specific policy attitudes (e.g., Reifler et al. 2011), our interest is higher up in the vertical chain: the effect of Schwartz's constructs on three foreign policy orientations. 1 Our most important hypotheses are H1 and H2, as we expect the largest impact for those two core sets of values that are the basis of multiple two-dimensional models: H1: Universalism will be positively related and highly predictive of cooperative internationalism.

H2: Conservation will be positively related and highly predictive of militant internationalism.
While we also expect universalism to be negatively associated with militant internationalism and conservation to be negatively associated with cooperative internationalism, we believe that effects of these values will be more attenuated given the primary purposes they serve.
As concerns self-enhancement values, we have two different hypotheses: H3: If self-enhancement values are directly opposed to universalist values, they will have the opposite effect of universalism, leading to lower cooperative internationalism and higher militant internationalism.

H4: If self-enhancement values are oblique or orthogonal to universalist values, conservation and self-transcendence values (particularly universalism), self-enhancement values will not have much effect on foreign policy attitudes.
As for openness, we predict,

H5: Openness will be associated with greater cooperative internationalism and lower militant internationalism.
We also have two different expectations for isolationism.

H6: Isolationism will be associated with higher levels of conservation and lower levels of universalism.
H7: Isolationism will be associated with high levels of self-enhancement and low levels of benevolence.

The CVP Survey, Personal Values, Rival Independent Variables, and Controls
In late January 2011, a nationally representative sample of 1,200 American adults was given the "Core Values Project" (CVP) survey, which we believe is the most comprehensive 1 One line of critique, which we do not explore here, concerns the restrictive top-down nature of hierarchical models. See . survey to date on personal values, foreign policy outlooks, and political attitudes. The survey, fielded by YouGov/ Polimetrix, was specifically designed to test the claims presented in this paper that values structure foreign policy thinking. Our empirical strategy is straightforward: we measure the underlying trans-situational values people hold, and use them to explain foreign policy outlooks while controlling for other variables.
We utilize the cooperative/militant internationalism framework, supplemented by isolationist measures, as a large and well-established body of research has repeatedly found that it accounts for the overall structure of the foreign-policy beliefs of Americans, even across epochs (Murray 1996). Moreover, the evidence of coherent attitudes does not seem to be unique to the American context: scholars have also found evidence of a similar structure of foreign-policy beliefs in at least Sweden, Canada, and Great Britain (Bjereld and Ekengren 1999;Reifler et al. 2011). We therefore measure our dependent variables of interest -participants' foreign policy orientations rather than specific issue attitudes -by asking participants about the extent to which they agree or disagree with a series of 10 statements, each of which taps into one of these three classic orientations from the foreign-policy beliefs literature (e.g. Rathbun 2007;Wittkopf 1990;Rosenau 1988, 1990;). Four of the items measure militant internationalism (sample item: "The United States must demonstrate its resolve so that others do not take advantage of it"), four capture cooperative internationalism (sample item: "The United States needs to cooperate more with the United Nations"), and two reflect support for isolationism (sample item: "The U.S. should mind its own business internationally and let other countries get along the best they can on their own.") As would be expected given the popularity of these three sets of measures, a variety of factor analytic techniques point to the presence of three distinct factors. Parallel analysis (Zwick and Velicer 1986) recommends a three factor solution. Exploratory factor analysis with principal axis factoring and varimax rotation shows that the three factors correspond to the militant internationalism, cooperative internationalism, and isolationism scales (see Appendix §4.1 for the pattern matrix), and produces a good fit according to standard model fit criteria (RMSEA: 0.042, RMSEA.LB: 0 . 0 3 ,T L I :0 . 9 6 7-s e ePreacher et al. 2013). In the main analysis below, we employ factor scores for these scales to obtain more precise estimates of our constructs of interest; in the appendix we replicate the analysis using simpler additive scores and find the results hold.
To measure our independent variables of interestparticipants' personal values -we employ a 20-item version of the Schwartz Portrait Value Questionnaire (PVQ) in which respondents are presented with verbal "portraits" of individuals (sample item for universalism: "She thinks it is important that every person in the world should be treated equally. She believes everyone should have equal opportunities in life.") gender-matched with each respondent, and asked to indicate "how much each person is or is not like you" (see Appendix §2 for the full instrumentation). We use a shorter 20-item version of the Schwartz value scale based on the ESS-PVQ-21 employed on the European Social Survey rather than the longer 40-item version, which is less practical to field on nationally representative samples because of cost considerations (for its measurement properties, see Davidov et al. 2008). As is standard in this type of survey research, we also include the usual demographic controls, measuring age, race, gender, and so on. Importantly, the Schwartz items are farther removed from the policy preferences they are supposed to explain than many of the standard items employed by political scientists in surveys such as the American National Election Studies (ANES) or General Social Surveys (GSS).

Personal Values Shape Foreign Policy Orientations
As an initial cut at the results, Table 1 presents a series of regression models where we estimate the effect of personal values on foreign policy orientations by aggregating upwards and employing additive scores for each of Schwartz's four superordinate value categories (conservation: a ¼ 0.77, openness to change: a ¼ 0.74, self-enhancement: a ¼ 0.84, self-transcendence: a ¼ 0.77). Models 1, 3, and 5 include survey weights, while models 2, 4, and 6 include a series of demographic control variables, although both sets of results tell a substantively similar story, suggesting four findings in particular. First, self-enhancement values, devoted to the meeting of personal needs, never significantly predict foreign policy orientations, while interpersonal values -particularly conservation and self-transcendence -play a larger role. Thus, as anticipated given the sociotropic nature of foreign policy preferences, interpersonal values more strongly predict foreign policy orientations than personal ones. Openness also adds little. Second, conservation values are statistically significant predictors of support for militant internationalism: individuals who embrace values that emphasize group survival like conformity, tradition, and security are the most supportive of the United States employing tough, coercive means of foreign policy abroad. Third, self-transcendence strongly predicts support for cooperative internationalism: respondents who seek to promote the welfare of others endorse foreign policies that do the same. Fourth, in comparison with militant and cooperative internationalism, isolationism appears to be less rooted in personal values. There is some support for H6: conservation's effect is statistically significant, but substantively weak, and even combined, the values explain relatively little of the variation in isolationist views, echoing other research highlighting the difficulties of rooting isolationism in personal values .
However, there are several reasons to be cautious about these results: although parsimonious, restricting our focus to the four superordinate value categories may mask potential heterogeneity, belying the possibility that values that occupy a similar structural position within Schwartz's framework have distinct effects on foreign policy postures. For example, as mentioned above, both universalism and benevolence are self-transcendence values, yet since universalism involves promoting the welfare of all while benevolence involves taking care of your friends, we would expect cooperative internationalism to be more strongly predicted by the former than the latter. Testing this hypothesis thus requires a lower level of aggregation. Moreover, the use of additive scales tacitly assumes that each indicator contributes equally to a participant's score, when certain questions may more closely proxy the underlying construct being measured than others.
To obtain a more precise measure of our quantities of interest and strike a balance between parsimony and empirical richness, we used principal axis factoring with varimax rotation to generate factor scores from a six factor solution, in which two of the factors correspond to superordinate value categories discussed above (conservation and openness to change), and the remaining factors refer to the two self-transcendence values (universalism and benevolence), and two self-enhancement values (achievement and power). We present the pattern matrix and more details about the model selection procedure in Appendix §4.2, but four points are worth noting here: the six factor solution has a good model fit (RMSEA: 0.043, RMSEA.LB: 0.037, TLI¼0.951), is theoretically interpretable, and avoids the multicollinearity concerns that would result from including additive scales for all 10 Schwartz values in a regression model simultaneously, all while obtaining more precise measures of our quantities of interest than would be possible from simple additive scales. Table 2 replicates the models from Table  1, this time using the factor score measures of the Schwartz values; the coefficient plot in the top panel of Figure 2 visually depicts the quantities of interest.
The top panel of Figure 2 presents a set of coefficient plots from a series of regression models regressing foreign policy orientations on the Schwartz values, while the bottom panel follows Ward et al. (2010) by presenting the same results another way, comparing each variable's statistical significance (jzj) and contribution to the model's insample predictive power (DR 2 ). Both sets of plots reconfirm that universalism is the driving value behind cooperative internationalism, that conservation (and to a lesser extent, universalism) underpin militant internationalism, and that the Schwartz values have little impact on isolationism.
Many of the patterns we saw in Table 1 also manifest themselves in Table 2: conservation, for example, is strongly positively associated with support for militant internationalism. At the same time, however, several dynamics change. The effect sizes are consistently larger than their predecessors in the previous set of models. Moreover, by disaggregating the self-transcendence values, we see that universalism and benevolence indeed have very different effects. As expected, it is the former rather than the latter that is strongly predictive of cooperative internationalism; although less statistically significant, benevolence appears to be negatively associated with cooperative internationalism, consistent with previous research that shows that in-group loyalty decreases support for humanitarianism abroad . Benevolence is also positively associated with support for militant internationalism, as we would expect if its effect is to facilitate in-group solidarity based on our theoretical review above, although displaying a substantively smaller effect than conservation.
As we expected in H3, if self-enhancement values were not directly opposed to self-transcendence values, they have no real effect, statistical or substantive, on foreign policy postures, with the exception of power's slight negative effect on cooperative internationalism. We find some support for H5: openness is associated with higher cooperative internationalism, although not with lower militant internationalism. We expected the latter if openness was not orthogonal to conservation, as indicated by the two value sets forming separate dimensions in the factor analysis.
Finally, we find some support for both H6 and H7. Isolationism is indeed associated with higher levels of conservation, consistent with the idea that conservation values are a way of meeting threats, and that isolationism is a particular way of protecting one's country. And yet, isolationists do not have the same communal solidarity as others who embrace conservation values; they score lower on benevolence. In this way, isolationists have a different value combination than either cooperative internationalists or militant internationalists. With the militant internationalists, they share conservation values, consistent with the idea that both fight and flight are the responses we expect in situations of threat, toward which conservation values are oriented. However, they do not have the same sense of communal solidarity expressed through benevolence values. Isolationism is about self-sufficiency and removing oneself from others.
Values are therefore helpful in resolving a longstanding puzzle in the literature on foreign policy attitudes. Wittkopf (1990) and Rosenau (1988, 1990) conceptualize isolationism as a combination of low internationalism, both militant and cooperative. Yet it often appears as a separate dimension in factor analyses of survey data, as it does here. That suggests that isolationism is more than just the absence of other foreign policy orientations. Our data offer a helpful step in understanding this: isolationism is a different construct because its adherents have a different constellation of values than internationalists of either type. Nevertheless, we do note that values are a weaker determinant of isolationism than other foreign policy postures, consistent with the idea that the question of whether to engage with the international system has stronger external determinants than the question of how.
Since, as Ward et al. (2010) show, statistical significance does not necessarily lead to predictive power, we illustrate these results another way in the bottom panel of Figure 2, which replicates the regression results presented in Table 2, but this time plotting the effects of each of the personal values as a function of its statistical significance (on the x axis, represented here as the absolute value of its z statistic), and the change in in-sample predictive power (on the y axis, represented here by the change in the R2 statistic) incurred by dropping the personal value from the model. These results reconfirm the intuitions from the previous sets of results: universalism is the driving value behind cooperative internationalism (explaining almost 20% of the variance), while conservation -and to a lesser extent, universalism -underpin militant internationalism, and the Schwartz values have relatively little impact on isolationism. The substantive effects of universalism on cooperative internationalism and conservation on militant internationalism are substantively large: holding all other variables at their means, a change from the 5th to the 95th percentile in universalism is associated with a change from the 17th to the 70th percentile in cooperative internationalism, while a change from the 5th to the 95th percentile in conservation is associated with a change from the 29th to the 67th percentile in militant internationalism. To summarize, we have shown that personal valuesespecially those serving interpersonal goals -shape foreign policy orientations: Americans who place a great deal of emphasis on promoting the welfare of others are more likely to be cooperative internationalists favoring working through the United Nations to resolve global problems, while respondents who are heavily concerned with the stability and security of the in-group are more likely to be militant internationalists favoring the use of force abroad. Americans, in other words, take foreign policy personally. Both the cooperative/militant internationalism and the vertical constraint models challenge the pessimism of early studies of American public opinion that found the mass public incapable of forming coherent beliefs about foreign policy, largely because they could not connect attitudes about international relations with their political ideology. Despite consistent findings that foreign policy attitudes are structured, this idea persists, most recently in the form of "elite cue" theory in which the mass public is viewed as passively receiving the wisdom of elites and the media. Goren (2013) challenges this pessimistic view of voter competence, showing that even while low-information voters who do not follow politics closely have a hard time identifying what a conservative or a liberal position is, they can nevertheless form coherent beliefs based on core values. He shows that values demonstrate a similar effect on policy attitudes of both low and high knowledge respondents but that in the case of the latter those views are more highly mediated by political ideology.
We therefore turn to nonparametric mediation analysis (Imai et al. 2011), estimating a set of nonparametric mediation models in which the effects of personal values are mediated by political ideology. Doing so not only serves as a means of mapping the ideological pathways through which personal values structure foreign-policy beliefs, but also offers a way to study the role of political ideology without inducing post-treatment bias (King and Zeng 2007): if personal values are truly pre-political, estimating their effects on foreign policy preferences while controlling for ideology would bias our effect estimates. We estimate separate mediation models for high-and lowknowledge participants, based on how they performed on a political knowledge test in our survey. 2 If the public is truly as helpless as pessimistic readings of elite cue theory would suggest, low-knowledge respondents should have difficulty connecting their values to their foreign policy orientations; they should need information from trusted elites in order to structure their foreign policy preferences for them, to provide the "constraint" from above that is unavailable from below. If, in contrast, even low-knowledge individuals are capable of grounding their foreign policy orientations in their core values, we should see the total effects of core values not significantly varying across both subgroups, even if the indirect effects through ideology differ. Low-knowledge individuals may indeed be as "innocent of ideology" as Converse (1964) claims, and will thus display weaker indirect effects, but the total effects will remain the same across both subgroups, as even unsophisticated respondents can anchor their foreign-policy beliefs on their core values, despite the absence of elite cues from above. Given the complexity of the modelwith six treatments and three dependent variables -in order to save space we relegate the full set of results to Appendix §6, and focus our attention here on the effects of conservation and universalism, the two Schwartz values that are not only of the most theoretical interest, but also played the most important role in the previous analyses. Figure 3 plots three quantities of interest for each personal value. First is the average causal mediation effect (ACME), which refers to each personal value's effect on foreign policy orientations transmitted through political ideology. Second is the direct effect, which refers to the personal value's effect on foreign policy postures channeled through all other potential mechanisms; in supplementary analyses in Appendix §6, we estimate a set of multiple mediation models to test whether ethnocentrism and the morality of force, the two foreign policy values espoused by Hurwitz and Peffley (1987), play a role here, although we focus on the role of ideology in our main analysis. Third is the total effect of the personal value on foreign policy orientation, regardless of the mechanism. Figure 3 presents selected results from a pair of nonparametric mediation models (Imai et al. 2011) calculated using N ¼ 1500 simulations and 95% quasi-Bayesian confidence intervals, in which personal values' effects on foreign policy orientations is mediated through political ideology. The results for low-knowledge respondents are shown in grey, and high-knowledge ones in black. Importantly, although the ACMEs differ -a greater proportion of the total effect is mediated by ideology for higher-knowledge participants -the total effects do not. We present just the results for the two most substantively important personal values (conservation and universalism) here; see Appendix §6 for the full set of results.
We median-split the sample on respondents' levels of political knowledge, and estimate two sets of models: the point estimates shown in black are from the high-knowledge subsample, and the point estimates in grey are from the lowknowledge one. Three points are important here. First, although ideology's mediation effects are significant for our core values of interest, personal values continue to matter independently of ideology for both groups of respondents: in the high-knowledge subsample, for example, only 36% of conservation's effect on militant internationalism is transmitted through political ideology. We know that political ideology matters in foreign policy (Gries 2014), but these results tell us that personal values predict foreign policy postures above and beyond the usual differences displayed by liberals and conservatives. 3 Second, the magnitude of the mediation effects is systematically larger for high-knowledge participants than for low-knowledge ones: the ACME for the effect of universalism on cooperative internationalism is 0.16 in the high-knowledge group (40% of the total effect), but only 0.04 in the low-knowledge group (9% of the total effect). Similarly, the ACME for the effect of conservation on militant internationalism is 0.11 in the high-knowledge group (36% of the total effect), and only 0.04 in the low-knowledge group (15% of the total effect). Consistent with much of the public opinion literature, then, high-knowledge respondents are more ideological than low-knowledge ones. Yet the sizes of the total effects do not significantly differ from one another: the total effect of universalism on cooperative internationalism is 0.43 in the low-knowledge group, and 0.40 in the highknowledge one; the total effect of conservation on militant internationalism is 0.27 in the low knowledge group, and 0.31 in the high-knowledge one.
In other words, we find, as did Goren in other domains, that values ground low-knowledge respondents' foreign policy dispositions just as they do for high-knowledge respondents. The latter simply demonstrate a tighter link between values, ideology and foreign policy postures. Where we differ from other work, however, is that the values we use are horizontal in character; they transcend the foreign policy domain.

Conclusion
Our findings suggest that personal values play a role in how ordinary people make sense of world politics. Other scholars of public opinion about foreign policy study the role of values, but typically do so by looking at one-off effects of individual values rather than exploring how values connect to a broader system. Although Schwartz's theory of universal values now constitutes a prominent and prolific research agenda in political and social psychology, we are, to our knowledge, the first to systematically investigate the relationship between these personal values and (a)( b)( c) Figure 3. Nonparametric mediation models: even low-knowledge respondents can anchor on core values. Low-knowledge respondents in grey, high-knowledge in black foreign-policy orientations via a nationally representative sample of Americans.
In general, our analysis shows that those values that define individuals' relationships with broader groups-as well as help coordinate behavior within them-play a larger role in foreign-policy beliefs than values based purely on individual needs. Conservation values provide strong predictors of militant internationalism. Universalism values are strongly associated with cooperative internationalism. In other words, the same fundamental values that shape our beliefs and behavior in our daily lives also predict our foreign-policy preferences; people take foreign policy personally. Values also help us better understand the phenomenon of isolationism. Isolationists embrace a unique combination of personal values: they are high on conservation but low on benevolence. Like Hurwitz and Peffley (1987), we claim that values and foreign policy orientations are related at least largely in a vertical sense, with the former driving the latter. However, personal values take the form of abstract beliefs about desirable end states or behaviors that transcend specific situations. They can therefore guide evaluation and behavior across the water's edge and into the domain of foreign affairs.
Our study carries with it a number of important implications. First, the Schwartz value scheme is universal. It encompasses multiple aspects of life, and works across cultures and contexts. Because other research demonstrates the significance of this same value framework for domestic political issues (e.g., Caprara et al. 2006;Schwartz et al. 2010), our results offer further evidence that foreign-policy beliefs reflect a universal value scheme. Our findings indicate that personal values matter for the mass public on those political issues and foreign-policy matters that they are the most removed from. This reveals the power of values.
Future research should explore similar linkages between personal values and foreign-policy preferences in other countries (e.g., Bayram 2015), and whether cross-national variation in foreign-policy preferences can be traced to cross-national variation in values as well. Typically in International Relations we assume that shared values across states promotes cooperation-as in, for example, liberal theories of the democratic peace (Russett 1993), and research on the role of ideological distance on conflict (Haas 2005) -but because conservation values serve the stability and well-being of the in-group, it may turn out that common values also facilitate tensions and conflicts.
Second, and relatedly, when international-relations scholars typically talk about values, they tend to employ the kind of constructivist or ideational bent commonly associated with concepts such as legitimacy, socialization, norms, and identity (Klotz 1995;Bukovansky 2007;Finnemore 2009). The Schwartz values, however, are biologically grounded. Each value cluster links to specific needs to survive as organisms and individuals. In this sense, there exists a "rump materialism" in the Schwartz value scheme, which may offer a potential bridge between public-opinion research and work examining the biological underpinnings of conflict and decision-making (e.g., Hatemi and McDermott 2012).
Third, because the mass public is relatively uninformed about events on the world stage, a large literature emphasizes the extent to which citizens use endorsements from trusted elites as heuristics to shape their judgments (Zaller 1992;Berinsky 2009). Yet personal values provide another way for generally uninformed individuals to establish coherent foreign-policy preferences-even in absence of elite cues. We show that those who score low on political knowledge connect their personal values to their foreign-policy orientations; they simply do so without going through the mediating factor of ideology-as highknowledge respondents do.
Fourth, our results suggest that while basic human values structure foreign-policy beliefs, some values carry more weight than others. This is where the distinction between inter-personal or society-centered values (universalism and conservation) and intra-personal or self-centered values (self-enhancement values) becomes important. In the foreign-policy domain, public discourse centers on how to ensure national security and advance the national interest abroad: Are military armaments or the tools of statecraft better suited to advance American interests? Should the US work through the UN or go it alone in international arena? Given the sociotropic focus of issues like these, values that prioritize socially-focused goals should find expression more readily through policy opinions than those egocentric values that prioritize personal advancement and selfgratification. That is precisely what we found. We demonstrated that socially orientated values of conservation and universalism exert more consistent and powerful effects on foreign-policy postures than the personally focused values of achievement, power, and self-enhancement. In sum, students of international relations and public opinion have long seen values as prime candidates for shaping citizens' views about foreign policy. We have affirmed this proposition, but with the critical qualification that some values are more consequential than others.