Who Makes Foreign Policy Decisions and How: An Empirical Inquiry Author(s): Margaret G. Hermann and Charles F. Hermann Source: International Studies Quarterly, Vol. 33, No. 4 (Dec., 1989), pp. 361-387 Published by: Wiley on behalf of The International Studies Association Stable URL: https://www.jstor.org/stable/2600518 Accessed: 13-03-2020 21:00 UTC JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org. Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at https://about.jstor.org/terms The International Studies Association, Wiley are collaborating with JSTOR to digitize, preserve and extend access to International Studies Quarterly This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms International Studies Quarterly (1989) 33, 361-387 Who Makes Foreign Policy Decisions and How: An Empirical Inquiry MARGARET G. HERMANN AND CHARLES F. HERMANN Ohio State University At the apex of foreign policy making in all governments or ruling parties are actors with the ability to commit the resources of the government and the power to prevent other entities within the government from reversing their position-the ultimate decision unit. Although this decision unit may change with the nature of the policy problem and with time, its structure will shape a government's foreign policy. In this paper we propose three types of decision units: predominant leaders, single groups, and multiple autonomous actors. Each of these exists in one of several conditions that help to determine whether the decision unit affects foreign policy largely through the pre-existing knowledge, beliefs, and style of those participating in the unit (a self-contained unit) or whether factors outside the decision unit must be taken into consideration in understanding the results of the decision-making process (an externally influenceable unit). The hypotheses that self-contained units will engage in more extreme foreign policy behavior than externally influenceable units and that single group decision units will show more extreme foreign policy behavior than those comprised of multiple autonomous actors are examined using data from twenty-five nations during the decade from 1959 to 1968. Who makes foreign policy decisions? What is the effect of the decision unit on foreign policy? An examination of how governments and ruling parties around the world make foreign policy decisions suggests that authority is exercised by an extensive array of different entities. Among the decision units are prime ministers, presidents, politburos, juntas, cabinets, inter-agency groups, coalitions, and parliaments. Moreover, within any one government the pertinent decision units often change with time and issue. When cross-national comparisons of governmental decision-making bodies are contemplated, as in the comparative study of foreign policy, the number of possible kinds of decision units becomes formidable. This essay examines one way of classifying decision units, showing how it enhances our ability to differentiate and account for governments' behavior in the foreign Author's Note: This research was funded by a National Science Foundation grant (SES-8618438), the Mershon Center, and the Ohio State University Instructional and Research Computing Center. We would like to thank Greg Peacock for his help with the data analysis as well as Bahgat Korany, Roy Licklider, Jerel Rosati, and Harvey Starr for their thoughtful comments on earlier drafts of the paper. ( 1989 International Studies Association This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms 362 Who Makes Foreign Policy Decisions and How: An Empirical Inquiry policy arena. Although we recognize that numerous domestic and international factors can and do influence foreign policy behavior, these influences must be channeled through the political structure of a government that identifies, decides, and implements foreign policy. Within this structure is a set of authorities with the ability to commit the resources of the society and, with respect to a particular problem, the authority to make a decision that cannot be readily reversed. We call this set of authorities the "ultimate decision unit," even though in reality the unit may consist of multiple separate bodies rather than a single entity. It is our contention that the configuration and dynamics of such an ultimate decision unit help shape the substance of foreign policy behavior. Participants experienced in the foreign policy-making process as well as those involved in decision making in large, complex organizations often remind us of the elusive nature of decision. They point out that, in contrast to many decision theories, the actual process of choice may not be a clear, precise occurrence. Instead it may be a gradual, incremental process that transpires over an extended period without anyone being able to say "X" made the decision on a given date. They note that those who gather and analyze information supplied to policy makers shape and narrow subsequent options by determining what is passed along and how it is interpreted. Moreover, the implementors of someone else's decision may modify the original intent. It takes nothing away from these important insights about decision making, however, to observe that in the life of every organization actual points of decision do occur, although not always in a fashion visible to all who have participated in the process. Certainly key decisions and those who make them are constrained by available inputs, and subsequent implementation may lead to distortion; nonetheless choice points do occur with some regularity. Despite the need to recognize that decisions do not always get executed as intended, knowledge about how decisions are made remains a powerful source of insight into what complex entities, such as governments, do. These same participants familiar with governmental foreign policy making also feel uncomfortable with the usual requirement of decision theories that all decisions result from a similar process. In the reality of governmental foreign policy making-as in any organization dealing with an array of different kinds of complex issues-it is extremely unlikely that there will be one recurrent set of policy makers who will handle all problems in the same way. A contingency approach to modeling how governments make foreign policy decisions is needed that indicates under what conditions alternative decision units will engage in one or another different types of processes. In this essay we begin development of such a contingency approach, proposing a set of conditions that appear to affect how decision units make foreign policy decisions. In differentiating decision units, we build upon the growing research about foreign policy-making that focuses on competing bureaucratic organizations, on small groups, and on powerful individuals. Many analysts have employed notions from bureaucratic organizations to explain foreign policy (see Neustadt, 1970; Allison, 1971; Destler, 1972; Halperin and Kantor, 1973; Halperin, 1974; Steinbruner, 1974; Szanton, 1976; Brady, 1976; C. Hermann, 1983). Interest has also centered on the role that small groups play in shaping foreign policy (see Janis, 1972; C. Hermann, 1978, 1979; Tetlock, 1979; George, 1980; Semmel, 1982; Anderson, 1987). Still others have sought to explain foreign policy by examining qualities of single leaders (see Holsti, 1976; Walker, 1977; Etheredge, 1978; M. Hermann, 1978, 1980, 1984; George, 1979; Stuart and Starr, 1981-82; Jonsson, 1982; Rosati, 1985). Most of the work to date, however, has considered each of these configurations-bureaucratic agencies, small groups, or individuals-separately This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms MARGARET G. HERMANN AND CHARLES F. HERMANN 363 without asking when this unit, rather than another, comes into play and with what consequences for foreign policy behavior. In this essay we argue that all three types of decision units are relevant to the comparative study of foreign policy but under varying conditions. We propose a way of determining which type is the ultimate decision unit in a particular situation and we empirically test the effects that the different kinds of units have on governments' foreign policy behavior. The following are the conceptual underpinnings of our approach. Definition. Sooner or later, every issue that appears on a government's agenda is dealt with in one of two ways. Either it ends in limbo with no resolution (for example, because the opportunity for decision has been overtaken by external events or because certain policy makers do not want to deal with the matter) or one or more decisions are made (perhaps including the decision to do nothing). If there is a decision, it is made by an individual, group of individuals, or multiple actors who have both (a) the ability to commit or withhold the resources of the government in foreign affairs and (b) the power or authority to prevent other entities within the government from overtly reversing their position without significant costs (costs which these other entities are normally unwilling to pay). We refer to the decision unit that has these two characteristics for a given issue at a particular time as the "ultimate decision unit." This ultimate decision unit may vary with the nature of the problem. For issues of vital importance to a country, the highest political authorities probably will be part of the ultimate decision unit. With more routine problems, the ultimate decision unit may actually be at a much lower level in the government. 1 In a number of contemporary governments, where policy normally involves multiple bureaucratic organizations, the problem may be passed among many different groups-within one agency, across agencies, or among interagency groups. The issue also may move between parts of a government as, for example, between the executive and legislative branches. Moreover, individuals and entities outside the government may enter the decision process as full participants. The basic point, however, remains that eventually for most foreign policy problems someone makes a decision committing (or deciding to withhold) the resources of the nation that cannot readily be reversed; they constitute the ultimate decision unit for that issue at that point in time. Classification. It is possible to develop a comprehensive set of ultimate decision units such that one type is responsible for any given foreign policy case. If we postulate that we can, in principle, define the set of actors that comprise the ultimate decision unit with regard to a foreign policy issue, then the task becomes one of describing the relationship among the actors in that set. We believe that the research literature, previously noted, has isolated the major alternative types of ultimate decision units. They are: 1. Predominant Leader. A single individual has the power to make the choice and to stifle opposition. 2. Single Group. A set of individuals, all of whom are members of a single body, collectively select a course of action in face-to-face interaction and obtain compliance. ' If a unit lower in a government's hierarchy is the ultimate decision unit, there probably are higher units in the government that theoretically have the ability to reverse its decision. A lower unit becomes the ultimate decision unit through an act of delegation either for a specific issue or by being given carte blanche for a domain of problems. The more difficult question concerns problems where the decision or recommendation is thrashed out at one level and then ratified or legitimated at a higher level. If the approval is largely symbolic without thorough review or examination of alternative options, the lower unit is still the ultimate decision unit. This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms 364 Who Makes Foreign Policy Decisions and How: An Empirical Inquiry TABLE 1. Key control variables by type of decision unit. Unit Control variable End points Status Predominant Contextual Sensitivity (A) Insensitive Self-Contained Leader (B) Sensitive Externally Influenceable Single Prompt Consensus (A) Agreement Self-Contained Group (B) Disagreement Externally Influenceable Multiple Relationship Among (A) Zero-Sum Self-Contained Autonomous Actors (B) Non-Zero-Sum Externally Actors Influenceable 3. Multiple Autonomous Actors. The necessary actors are separate individuals, groups, or coalitions which, if some or all concur, can act for the government, but no one of which by itself has the ability to decide and force compliance on the others; moreover, no overarching authoritative body exists in which all the necessary parties are members.2 In cases of foreign policy decision making, the analyst should be able to classify the actors who can make authoritative decisions for the government into one of these three categories. In some countries, the same ultimate decision unit may prevail in nearly all foreign policy matters. It should be emphasized, however, that in many nations the unit may change with the issue under consideration or with the evolution of the regime. Thus, in the United States, for example, the President may make a spontaneous decision in response to an unexpected question at a press conference (/a predominant leader); for a military issue, the decision may be made by the Joint Chiefs of Staff (a single group); and for still another issue, such as arranging a treaty with a foreign government, the decision involves the President, his executive branch associates, and the Senate (the three constitute a multiple autonomous actor). Conceptualization of Control Variables. Each kind of ultimate decision unit exists in one of several states that determines not only the unit's direct effect on the final policy outcome but also the extent to which factors outside the decision unit must be considered in understanding what will happen in the foreign policy-making process. For each type of ultimate decision unit a key piece of information enables the analyst to know when to focus only on the decision unit itself to understand the making of a foreign policy decision and when to look outside the unit for factors that will influence the decision. The status of these "key control variables" determines how other elements enter into the decision calculus for that unit. The key control variables for each of the three types of decision units are presented in Table 1. The end points for each of the control variables labeled "A" in Table 1 indicate conditions for which the primary source of explanation for foreign policy resides in the nature of the decision unit itself-the internal dynamics of the unit shape the decision and the decision process is self-contained within the unit. By contrast, the end points for the control variables designated "B" indicate the circumstances in 2 In an earlier piece on decision units (Hermann, Hermann, and Hagan, 1987), we called this type of de unit multiple autonomous groups. Further consideration of the types of entities that comprise such units i that they could be individuals and organizations as well as groups, thus the change in the label from mult autonomous groups to multiple autonomous actors. This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms MARGARET G. HERMANN AND CHARLES F. HERMANN 365 which the unit is externally influenceable-that is, it is more susceptible to outside pressures on its decision-making process. Thus, we can identify ultimate decision units that are more likely to be self-contained and ultimate decision units that are more likely to be externally influenceable for each of the three kinds of units depending on the state of the key control variables. As we shall see in describing the three types of decision units and the key control variables, these alternative conditions lead to different decision-making processes and, in turn, affect the nature of the foreign policy behavior of a government. Three Types of Ultimate Decision Units Predominant Leader When the ultimate decision unit is a predominant leader, a single individual has the power to make the choice for the government. When such a leader's position is known, those with differing points of view generally stop voicing alternative positions out of respect for the leader or fear of political reprisals. Even if others are allowed to continue discussing alternatives, their points of view are no longer relevant to the political outcome. The predominant leader decision unit is illustrated by a statement attributed to Abraham Lincoln in a cabinet meeting: "Gentlemen, the vote is eleven to one and the one has it." Only Lincoln's vote mattered; in this case, he was a predominant leader. In this type of decision unit, it becomes important to learn about the personal characteristics of the predominant leader. The leader's traits shape his initial inclinations and determine whether and how he will regard advice from others, react to information from the external environment, and assess the political risks associated with various actions (see M. Hermann, 1978, 1980, 1984). Of particular relevance in explaining a predominant leader's reaction to a foreign policy problem is knowledge about the leader's orientation to foreign affairs-his or her composite set of views about how governments should act in the foreign policy arena. An orientation defines the leader's conception of his nation's role in the world and it presupposes a specific political style in dealing with foreign policy problems (see D'Amato, 1967; Etheredge, 1979; George, 1980; Walker, 1983; Rosati, 1985; M. Hermann, 1987). Most important for purposes of the present discussion, orientations indicate how sensitive the leader will be to advice and information from the environment when making a foreign policy decision. If a leader's orientation suggests that he has a strongly held view of the world and uses his view as a lens through which to select and interpret incoming information, the leader is likely to be looking only for cues that confirm his beliefs when making foreign policy decisions. As a result, he will be relatively insensitive to discrepant advice and data.3 These leaders have been variously called in the literature "crusaders" (Stoessinger, 1979), "ideologues" (Ziller et al., 1977), "autocratic leaders" (Bass, 1981), and "low self monitors" (Snyder, 1974, 1979, 1982); they are guided by their dispositions, taking an "inside looking outward" perspective on life (Gardner, 1983) and selectively using incoming information to support their predispositions. Such leaders tend to choose advisors who define problems as they do and who are generally enthusiastic about the leader's ideas. Moreover, they value congruence between "who they are" and "what they do." Libya's Quaddaffi and Cuba's Castro are examples of predominant leaders whose orientations appear to predispose them to 3 Of course, most leaders at some time become relatively insensitive to new and potentially disturbing information and analysis of their environ-ments-e.g., when they exper-ience stress. The type iden-tified here, however, routinely rejects or reinterprets such information. This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms 366 Who Makes Foreign Policy Decisions and How: An Empirical Inquiry be relatively insensitive to information that does not conform to what they want to do. By knowing the foreign policy orientations of these two leaders, we know the positions they are likely to press on their governments. If, however, the leader's orientation leads him to be sensitive and open to others' opinions and to incoming information, we will need to know something about the environment in which the predominant leader is operating to predict what the government is likely to do. Because such leaders are more "pragmatic" (Ziller et al., 1977; Stoessinger, 1979)-more guided by the situation and interpersonal influences (Snyder, 1974, 1979, 1982; Bass, 1981; Gardner, 1983), our analysis must take into account the context in which the leader finds himself. The sensitive leader will want to ascertain where others stand with regard to the problem, to consider how other governments are likely to act, and to examine conflicting information before making a decision. Such leaders perceive themselves as flexible and adaptible, shrewdly and pragmatically able to tailor their behavior to fit the demands of the' situation. Jordan's Hussein and Zambia's Kaunda are examples of this type of leader. Knowing the foreign policy orientations of leaders like these two will provide clues about what part of the environment will be most influential on the leader, but we must still understand the influences emerging from that part of the environment to understand what the leader will do. In sum, when the ultimate decision unit is a predominant leader, the key question is whether or not the leader's orientation to foreign affairs leads him to be relatively sensitive or insensitive to information from the political environment. If the leader is relatively insensitive, knowledge about the leader's personality will provide us with cues about what his government's foreign policy behavior is likely to be. The insensitive predominant leader is a self-contained decision unit. If the leader is more sensitive, we need to understand other aspects of the political system in order to suggest what the government will do in response to a foreign policy problem-personality data will not be sufficient. In effect, the sensitive predominant leader becomes an externally influenceable ultimate decision unit. Single Group When no one individual has the ability to routinely determine the position of the government on a class of foreign policy issues-or if such an individual declines to exercise that authority-then an alternative decision unit must operate. The single group represents one option. A single group acts as the ultimate decision unit if all the individuals necessary for allocation decisions participate in a common group and the group makes decisions through an interactive process among its members. Single group ultimate decision units are common in contemporary governments. The Politburo of the Communist Party in the Soviet Union, the Standing Committee of the Communist Party in China, the National Security Council in the United States, and the cabinet or subcabinet groups in various parliamentary governments illustrate single group decision units. To be an ultimate decision unit a single group does not have to be legally or formally established as an authoritative agent. Instead it must have, in practice, the de facto ability to commit or withhold resources without another unit engaging at will in the reversal or modification of its decision. 4To make our presentation less complicated we have been discussing leaders who are sensitive or insen actuality, however, there is a continuum of sensitivity along which leaders differ in their degree of openess to information from the environment. Moreover, sensitivity may not be a general phenomenon but may change with issues or with level of interest or expertise in the area of the foreign policy problem. Thus, even though we will continue to use the terms sensitive and insensitive, we recognize that this variable has some nuances we are not confronting directly here. This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms MARGARET G. HERMANN AND CHARLES F. HERMANN 367 Moreover, it is not necessary for all group members to concur on every decision of the unit or to have equal weight in the formation of group decisions. However, if some formal members of the group are never essential to establishing a group decision, then it would be more accurate to recognize the existence of a subgroup that excludes such persons. When the ultimate decision unit for a particular foreign policy problem is a single group, the analyst must determine if the group can achieve a prompt consensus about the disposition of the problem under consideration (see Janis, 1972; C. Hermann, 1978, 1979; George, 1980). If substantial agreement is achieved quickly among the members (that is, during one meeting or a few sessions held under conditions of tight secrecy over the course of only a few days), factors outside the group that are not already known by its members can have very little effect on the decision. With prompt consensus members of the group do not look elsewhere for either recommendations or support for their positions. As a result, elements outside the group at the time it considers the problem remain excluded from the process. The members reinforce each others' predispositions and feel secure in their collective decision. Should disagreement persist, however, other forces outside the group can become influential, as members of the group seek supporting information for their positions, reinterpretation of the problem, or ways to resolve the conflict. The likelihood of prompt group consensus is affected by a variety of factors. It is more likely if group members share a common ideology or regime orientation-in other words, a common set of values and beliefs with regard to the problem at hand. For example, the group may have a strongly shared set of beliefs about certain traditional enemies and how they should be treated. Prompt consensus is also more likely if the group has certain structure and process characteristics (see Collins and Guetzkow, 1964; Allison, 1971; C. Hermann, 1979; McGrath, 1984). Thus, consensus is more likely if the information the group receives is from a common source, is shared among group members, and is similarly interpreted by members. Moreover, it is more likely if the group is small, if members have their primary loyalty to the group, and if power is unequally distributed among group members (that is, there is a strong, dominant leader). When information comes from diverse sources and is not shared by all group members, when groups get larger, when group members are representatives of other groups or organizations with their primary loyalty to these outside groups, and when power is more equally distributed, consensus becomes more difficult and the possibility for disagreement among the members increases. Aspects of the situation may also make prompt consensus more difficult to achieve. When the problem is highly ambiguous or complex and controversial, the likelihood increases of members of the group disagreeing and introducing outside influences into the decision-making process. Therefore, a key to understanding the foreign policy behavior that will be advocated when a single group is the ultimate decision unit is information indicating whether the group can achieve prompt consensus. When consensus occurs quickly, we can focus exclusively on the group's internal dynamics in estimating how the group is likely to deal with the foreign policy problem-the single group becomes a self-contained decision unit. If, on the other hand, group members have difficulty reaching consensus, we need to ascertain what other forces outside the group are likely to affect the process in order to determine resulting foreign policy decisions. The single group is in this case an externally influenceable decision unit. Multiple Autonomous Actors It should be evident that another alternative exists when the ultimate authority in foreign policy making is neither a single individual nor a single group. In this case we This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms 368 Who Makes Foreign Policy Decisions and How: An Empirical Inquiry have multiple-two or more-separate entities with independent authority structures, none of which can commit the resources of the regime without the agreement of all or some of the others. To be one of the actors in the set classified as the ultimate decision unit, a group, organization, or individual must be capable of giving or withholding support that, when combined with the support (or lack thereof) from the other actors, is sufficient to determine whether regime resources will be allocated. One actor can block another's initiatives by (1) using a formal, sometimes constitutionally defined, veto power; (2) threatening to terminate a ruling coalition by withdrawing from it or overthrowing it with force; (3) withholding part of the resources necessary for action or the approval needed for their use; or (4) initiating countermeasures that can seriously harm the other actors or their objectives. For a set of multiple autonomous actors to be the ultimate decision unit, the decision process cannot involve any superior group or individual that can independently resolve differences existing among the actors or that can reverse any decision these entities reach collectively. Representatives of multiple autonomous actors can interact, so long as any resulting decision is not official unless approved by each constituent party. Unlike participants in single group decision units, representatives of multiple autonomous actors have no authority except as agents of their respective entities. A classic example of an ultimate decision unit composed of multiple autonomous actors is the coalition government in a parliamentary system such as those recurrent in the Fourth Republic of France, in Italy during the past two decades, and in Israel under the Labour-Likud coalition. In these governments, cabinets are composed of members from several parties, none of which has a majority of seats in the parliament. The members of the coalition depend on each other to retain control of the government. This situation gives each party the ability to block potential policies advocated by the other parties in the cabinet with the threat of bringing down the government by withdrawing from the coalition. Ultimate decision units composed of multiple autonomous actors are not limited to parliamentary regimes. In presidential democracies with their independent executive, decision making can involve multiple autonomous actors on those issues where the president must receive the approval of the legislature. Multiple autonomous actors as ultimate decision units can also exist in authoritarian regimes. Following Perlmutter (1981), we note that authoritarian regimes typically consist of three components: the state or governmental apparatus; the single, official party; and a variety of "parallel" or "auxiliary" structures which support the regime (such as militant gangs, the secret police, and the military). Generally, a stable authoritarian regime like that in the Soviet Union is characterized by the dominance of one group-in this case a strong, cohesive Communist party. During certain periods, however, relations among these three components can become unstable, with none of the competing groups or factions having dominance. This situation is particularly likely if there are no accepted rules or procedures for allocating resources or transferring political power. The government takes on the form of an unstable coalition. Such unstable coalitions are commonplace among Third World regimes, many of which are internally fragmented and continuously threatened by military intervention. When multiple autonomous actors form the ultimate decision unit, a foreign policy behavior must result from an agreement forged among the set of entities involved. When multiple autonomous actors are unable to reach agreement on any meaningful course of action, they deadlock. Deadlocks result because (by definition) no entity has the capacity to act alone on behalf of the regime. One or more actors are always in a position to block the initiatives of the others. Each actor can take only very limited action on its own (typically in the form of verbal pronouncements). No This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms MARGARET G. HERMANN AND CHARLES F. HERMANN 369 coordinated regime foreign policy activity involving the use of resources is possible; meaningful actions and commitments are postponed. We do not mean to indicate, however, that deadlock is automatic. The relationship among the multiple autonomous actors determines when deadlock is likely to occur. Actors that accept each other's right to exercise power often develop some formal or informal "rules of the game" or procedures for engaging in political competition that may include rules for establishing agreement (see Leiserson, 1970; Dodd, 1976; Druckman, 1977; Salmore and Salmore, 1978; Austin and Worchel, 1979; Hinckley, 1981; Wilder, 1986). Such entities, in effect, have a non-zero-sum relationship. Because they do not regard negotiation and compromise as defeats or as concessions to an illegal entity, they have a better chance of making a decision than those actors who deny each other's legitimacy and recognize no limits on practices that keep the others from participating politically. In other words, in cases where the actors grant each other political legitimacy, they are likely to have established procedures for negotiation and reaching agreement that reduce the chances of deadlock. When the various actors can anticipate the effects of applying established rules for agreement, they can estimate the chances of their preferences prevailing and make a decision whether to remain resolute or to make concessions. In this fashion the rules of the game increase the chance of some agreement because the actors whom the rules do not favor in a given instance have added incentive to bargain with the others. In sum, when multiple autonomous actors are the ultimate decision unit, the key to ascertaining the nature of any foreign policy behavior is whether the political relationship among the multiple autonomous actors is zero-sum or non-zero-sum with respect to recognizing the legitimacy of each entity to seek and share power. When multiple autonomous actors have a zero-sum relationship, they typically try to deny one another power by all possible means and see each party as benefitting at the other's expense. Usually, interdependent actors locked in such a conflict can do nothing or almost nothing in the foreign policy arena. They become a self-contained decision unit fighting among themselves for power and authority, and they are open to little outside influence. When multiple autonomous actors have a non-zero-sum relationship, however, there is basis for agreement. In order to ascertain the nature of the agreement, we have to examine the bargaining process among the parties, the formal or informal rules of the game governing such a process, and the external forces that may influence the process. Multiple autonomous actors with a non-zerosum relationship are an externally influenceable decision unit: they seek as a guide to action information on each other and the environment as well as on how any rules of the game are likely to affect them. Determining the Nature of the Ultimate Decision Unit Having defined three types of ultimate decision units, we must develop a way to determine which of the three types occurs in a particular government dealing with a specific foreign policy problem. In approaching this task we have made several assumptions. First, the ultimate decision unit can vary from one problem to another, although in some countries there may not be much variability. Second, decision units with the power to make a decision may not always elect to do so. Third, certain kinds of problems increase the likelihood of particular units exercising ultimate authority. For example, a head of state is likely to deal with his counterpart on high-level protocol issues; in crises or other critical situations, decisions tend to be made by a high-level, small group (see C. Hermann, 1972). Fourth, the identification of the ultimate decision unit for any particular problem is almost always an inferential task in which the analyst must exclude possibilities as well as find evidence with which to estimate the probable unit. This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms 370 Who Makes Foreign Policy Decisions and How: An Empirical Inquiry 1. Begin by identifying an immediate substantive problem that the regime has No recognized. 2. Is there in the regime's leadership a 3. Has the leader manifested in the single individual with the power/authority Yes past a general, active interest in Yes to commit (or withhold) the regime's as well as involvement with resources in the affected issue area foreign and defense issues? regardless of opposition by others? I No 4. Is the immediate foreign policy problem one perceived by the regime leadership to be critical Yes for the well-being of the regime or society or does it entail highlevel diplomacy and protocol? | No 5. Does the immediate problem concern No Yes a matter known to be of personal interest to the leader? 6. Does the leader, after setting the general policy direction for No coping with the problem, evidence regular, active participation in the decision process? l Yes 7. Does the leader include selected others as part of the decision Yes process throughout the examination of the problem and give them a ",veto" over any decision? | No DECISION UNIT IS PREDOMINANT LEADER FIG. 1. Decision tree for determining ultimate decision unit. These assumptions provide the foundation for the questions in Figure 1 which, i turn, we have used to infer the nature of the ultimate decision unit for a country fo a given problem. The decision tree in Figure 1 indicates, for a particular foreign policy issue, which individuals and groups are likely to have the authority to commit or withhold the resources of the nation and the ability to prevent other entities in the This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms MARGARET G. HERMANN AND CHARLES F. HERMANN 371 8. Does the problem as defined fall within the domain of an issue area for which Yes there is only one known dominant policy group? I No 9. When a foreign or defense problem is perceived to be critical for the well-being of the regime or society, is 10. Is the there one group that can Ye s current Yes exercise final authority to problem commit or withhold the re- perceived gime's resources in coping to be with the problem? critical? | No 11. Are the policy groups likely to be involved with the No immediate problem arrayed in a hierarchical manner in relation to one another such that somewhere in the chain of Yes_ command one group can make an authoritative decision to commit or withhold the regime's resources? 12. Is everyone within the regime whose support is essential to No commit or withhold the regime's resources for coping with the immediate problem a member of the No same policy group, such that a group decision cannot readily be altered by outside opposition? I Yes 14. Are there two or more separate 13. Is the issue area of which the actors (groups, organizations) immediate problem is a part one in who do not combine into a which the regime is dependent upon single decision unit and none the approval of an external of the actors can by itself (foreign) entity? (i.e., without the concurrence of one or more others) commit or withhold the regime's No resources in coping with the immediate problem? 4 Yes / DECISION UNIT IS SINGLE GROUP DECISION UNIT IS MULTIPLE AUTONOMOUS ACTORS government from reversing their decision. This figure can be compared to a sorting machine that separates apples of different sizes into alternative bushel baskets. The smaller apples are siphoned off first as they pass through the sorter, followed by the medium-sized apples, with the large apples left at the end of the chute. The initial questions seek to determine if the ultimate decision unit is a predominant leader; if This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms 372 Who Makes Foreign Policy Decisions and How: An Empirical Inquiry this is not the case, questions are posed to ascertain if the ultimate decision unit is a single group; if not, the remaining units must be multiple autonomous actors. As indicated in Figure 1, the first step-question (Q) 1-is to identify the immediate foreign policy problem the government is facing. Next, the concern is whether any regime leader qualifies as a predominant leader (Q 2). Even if the regime has a single, powerful leader who meets the criteria for predominant leader, that person may not exercise authority over and deal with the current problem. The left side of Figure 1 provides a series of checks to ascertain whether a predominant leader exercises his authority. At issue are whether the leader has a general interest in foreign and defense issues (Q 3), whether the problem is critical or involves high-level protocol (Q 4), whether the problem is of special interest or concern to the leader (Q 5), whether the leader continues to participate beyond specifying broad policy directions (Q 6), and whether the leader opts to share decision-making responsibility with selected others (Q 7). Depending on the status of these conditions, the predominant leader may or may not actually serve as the ultimate decision unit for the problem at hand. If the predominant leader does not exercise his authority, attention shifts to the right side of Figure 1 and the possible existence of a single group as ultimate decision unit. A single group may have ultimate authority for making the decision regarding the current problem because it has been assigned responsibility for all problems like the present one (such as a task force) (Q 8), because the problem is a crisis or other critical problem (Qs 9 and 10), or because the decision units in the government are hierarchically arranged so that at each level one group has authority to resolve particular kinds of problems (Q 11). An important requirement is that no individual or group whose participation is essential for making a decision be excluded from the group (Q 12). This requirement also extends to foreign governments or other international actors that may exercise a veto over decisions in certain areas (Q 13). If all these various inquiries have failed to establish the ultimate decision unit as either a predominant leader or a single group, then the ultimate authority likely rests with multiple autonomous actors (Q 14). To apply this classification system to the actual foreign policy activities of some selected governments, we solicited the help of a group of area and country specialists. Rather than ask these experts directly the often unanswerable question of who made the decision in dealing with a given problem, we sought answers to questions that informed area experts might reasonably know. We asked them about the nature of the regime, the types of issues that faced the regime during a given time, and the probable relationships among known policy groups and individuals. Although the questions often required considerable judgment, most could be answered with a degree of confidence by the respondents. Problems that the experts had in answering any questions were resolved through discussion with the researchers. (This research is reported in detail in C. Hermann, 1981.) The information from the area specialists was combined with a set of assumptions about decision making (for example, problems critical to a regime are more likely to be handled at the top) to enable us to use Figure 1 to make plausible and replicable estimates of the likely types of ultimate decision units that dealt with various events in a given regime. Data were collected on the decision units in twenty-five nations between 1959 and 1968 and have been used to estimate the probable type of ultimate decision unit for a number of foreign policy problems that confronted these countries during this decade.5 The particular nations and the estimated ultimate 5 The twenty-five nations examined here represent that portion of the thirty-eight nations in the Comparative Research on the Events of Nations (CREON) Project for which we could both identify and gain the cooperation of an expert in the foreign policy of the country for the decade 1959-68. We chose specialists based on their This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms MARGARET G. HERMANN AND CHARLES F. HERMANN 373 decision units are presented in Table 2. Where specific issues are not indicated beside a decision unit in Table 2, the data from the area experts suggested that the particular decision unit exercised ultimate authority across a broad array of problems for the regime during the time period noted (see footnote b to Table 2). How Ultimate Decision Units Shape Foreign Policy Behavior To be useful our classification of decision units and associated control variables must lead to insights about foreign policy behavior. In the discussion to follow we will describe a preliminary empirical inquiry into the effects that decision units can have on a government's foreign policy actions. The analysis will focus on the distinction noted earlier between self-contained decision units (the insensitive predominant leader, the single group able to reach prompt consensus, and multiple autonomous actors with a zero-sum relationship) and those that are externally influenceable (the sensitive predominant leader, the single group in continuing disagreement, and multiple autonomous actors with a non-zero-sum relationship). We hypothesize that the self-contained decision units are less constrained by the complexities and nuances of the specific setting in which they find themselves than the externally influenceable units and, as a result, on balance are more likely to engage in extreme foreign policy behavior. By extreme behavior we mean both very minimal activity as well as assertive, highly committed actions. In general, it is our contention that the self-contained decision units are less affected by the distinctive aspects of the immediate problem than their externally influenceable counterparts. Self-contained units are likely to have strong predetermined beliefs about how to handle almost all international situations or to be so absorbed in dealing with the internal dynamics and politics within the decision unit that they ignore the particular features of any given problem or concerned outside interests. This is not to say that self-contained units are always oblivious to the current political demands of the situation or that their responses are invariably insensitive to the existing situation, but, by ignoring situational complications and the special requirements of the problem, self-contained decision units can be expected more often to pursue a simpler, more unqualified course of action. Thus, the self-contained units are more likely to push their positions and are less likely to compromise or to take small, incremental steps toward their goals than are the externally influenceable units. They believe they know what should be done in response to the problems they face and are ready to do it. For the insensitive predominant leader and the single group able to reach prompt consensus, in particular, we hypothesize that this belief will lead them to engage in more conflictual behavior than their counterparts in the externally influenceable units. Their strong convictions also enable them to either commit their resources extensively or withhold their resources entirely and to use economic and military instruments of statecraft (instruments which generally require some commitment of resources in their use) in addition to or in place of diplomatic channels. In the case of the multiple autonomous actors with a zero-sum relationship, because each separate party believes in its own infallibility and the sinister nature of the other parties to the decision, each perceives that its position must prevail. As a result, none is likely to yield and deadlock occurs. In contrast to the other self-contained units, activities of zero-sum multiple autonomous actors tend to occur reputations, seeking individuals "who had demonstrated published scholarship on the country's foreign policypreferably including the decade under examination" (C. Hermann, 1981:219). We asked one expert for each country to complete the questionnaire regarding the internal decision processes involved in the making of foreign policy for that country. This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms 374 Who Makes Foreign Policy Decisions and How: An Empirical Inquiry TABLE 2. Ultimate decision units for twenty-five nations during the decade 1959-68. Nation Decision unitr Years and issuesb Canada Cabinet (SG) 59-4/63 Cabinet (SG) 4/63-4/68 Chile Ministry of Foreign Affairs (SG) 59-11/64 Frei (PL) 11/64-68 China (PRC) Standing Committee of the Polit- 59-8/66 buro (SG) Cultural Revolutionary Group 8/66-68 (SG) Cuba Informal Advisory Group to Cas- 59-60 tro (SG) Castro (PL) 61-68 Czechoslovakia Politburo (SG) 59-67 Politburo (SG) 68 Egypt Nasser (PL) 59-68 France DeGaulle (PL) 59-68 Ghana Nkrumah (PL) 59-2/66 National Liberation Council (SG) 2/66-68 India Nehru (PL) 59-5/64 Inner Cabinet (SG) 5/64-65 Inner Cabinet (SG) 66-68 Israel Ben-Gurion (PL) 59-6/63 Cabinet (SG) 6/63-68 Kenya Inner Circle (SG) 12/63-68 (for economic issues and issues dealing with relations with Western countries) Ministry of Foreign Affairs (SG) 12/63-68 (for intra-African issues) Mexico Mateos (PL) 59-11/64 Ordaz (PL) 11/64-68 New Zealand External Minister and Staff (SG) 12/60-64 External Minister and Staff (SG) 65-68 Norway Cabinet and Storting Foreign Af- 59-68 fairs Committee (MAA) Philippines Macapagal (PL) 61-65 (for issues dealing with Southeast Asian relations) Marcos (PL) 66-68 (for issues dealing with Southeast Asian relations) Macapagal and the Americans 61-65 (MAA) (for issues dealing with the economy and East-West alignments) Marcos and the Americans (MAA) 66-68 (for issues dealing with the economy and East-West alignments) Soviet Union Khrushchev (PL) 59-10/64 Politburo (SG) 10/64-68 Spain Franco (PL) 59-68 Switzerland Federal Council (SG) 59-68 (foreign labor) Foreign Minister and Advisers 59-68 (SG) (neutrality, aid, relations with Socialist countries) This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms MARGARET G. HERMANN AND CHARLES F. HERMANN 375 TABLE 2. (continued) Nation Decision uni't Years and issuesb Defense Ministry (SG) 59-68 (for defense issues) Transport Ministry (SG) 59-68 (for issues dealing with international transport) Finance Ministry and National 59-68 Bank (MAA) (for issues dealing with international finance) Foreign Affairs Ministry, Foreign 59-68 Trade Section of Department of (for issues dealing with EEC and Economy, and Standing Delega- European integration) tion for Foreign Trade Negotiations (MAA) Thailand Sarit (PL) 59-11/63 Cabinet and Military High Com- 11/63-68 mand (MAA) Tunisia Bourguiba (PL) 59-68 Turkey Gursel (PL) 5/60-10/61 Inonu (PL) 11/61-2/65 Demirel (PL) 10/65-68 Uganda Obote (PL) 10/63-68 Venezuela Betancourt (PL) 59-64 Foreign Minister and Advisers 65-68 (SG) Yugoslavia Tito (PL) 59-68 Zambia Kaunda (PL) 10/64-68 a The abbreviation-s in par-en-theses stan-d f Leader (PL), Single Group (SG), and Multiple Autonomous Actors (MAA). b If no issues are listed under a date, the area experts did not perceive that the ultimate decision unit differed across the issue areas we identified for the country for that time period. Both we and the area experts realize that a more detailed case study analysis of specific problems will lead to a more finely tuned identification of ultimate decision units for many of these countries that would differ by issue. Thus, these data are seen as providing only an initial cut at specifying ultimate decision units for these nations and time periods. only at one end of the spectrum of extreme behavior-that entailing minimal physical action. Deadlocks generally do not lead to the commitment of resources or resource-using instruments such as economic and military action. Thus, little resource commitment and heavy concentration on diplomacy are hypothesized for this type of ultimate decision unit. Although physical action may be minimal, expressions of feelings toward others need not be. Given that each of the multiple autonomous actors in the zero-sum condition may unilaterally use diplomatic channels to try to enhance its own position in the decision process, the government's behavior may entail high negative verbal outbursts. Being more responsive to and constrained by what is happening outside the decision unit-for example, by the positions taken by the opposition, by the need to gather more information, by the changing nature and special features of the current situation, by the need to build a consensus or reach a compromise on the issue-the externally influenceable units display more caution and deliberation. These units, more often than their self-contained counterparts, detect the complexity in the circumstances they face and the absence of unambiguous alternatives. They are This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms 376 Who Makes Foreign Policy Decisions and How: An Empirical Inquiry more aware of the uncertainties associated with any action they take and, as a result, engage in more moderate, provisional behavior. Thus, we expect the sensitive predominant leader and the single group in continuing disagreement to display foreign policy behavior that focuses primarily on diplomacy as the instrument of statecraft, that is cooperative to neutral in nature, and that involves committing only moderate amounts of resources to dealing with a problem at any one time-the incremental approach to action. As we noted earlier, the multiple autonomous actors with a non-zero-sum relationship, unlike their counterparts in the self-contained units, often can come to some agreement following a set of rules or political norms that permit bargaining and coalition. Accordingly, we hypothesize that such decision units will choose behavior that is more cooperative, involves more commitment of resources, and uses a more diverse array of instruments of statecraft than the multiple autonomous actors with a zero-sum relationship. The fact that the zero-sum actors usually deadlock makes the behavior of the non-zero-sum actors seem somewhat more extreme, even though their actions normally result from compromise and con- sensus.6 Up to this point we have considered the possible effect of different conditions of the ultimate decision units on governments' foreign policy behavior. In effect, we have suggested that regardless of which of the three decision units prevails, those in the externally influenceable mode will tend to engage in less extreme foreign policy behavior than those which are self-contained. But the reader may reasonably ask whether there are differences among the three types of decision units themselves with respect to their tendency toward more or less extreme action. To date there is little theoretical or empirical work that compares all three types of decision units (or ones roughly analogous to those we have proposed). The most instructive insights come from work in social psychology comparing individual and group tendencies to select more risky choices. Initially referred to as the "risky-shift" phenomenon, early studies (Stoner, 1961; Wallach, Kogan, and Bem, 1962) found that following a group discussion, members of the group chose more risky responses than these same individuals had done privately before the discussion. Later research has sustained this shift in individual position following group discussion, but has shown that the shift can be toward either a more risky or a more conservative position. The common finding is that collective decisions and the positions of individual members after group discussion move in the direction of the individual positions most frequently held before the discussion. This more generalized position-now referred to as the "polarization effect"-has been explained in various ways. Some have made a cultural argument; that is, individuals change positions to support the more culturally valued position revealed by discussion. Others have drawn on group phenomena proposing that individuals tend to alter their own positions in response to group pressure for conformity and, as a result, diffuse responsibility throughout the group for any negative consequences of the action. Still others have posited that the explanation resides in the fact that members in the minority are persuaded by the merits of the stronger and more diverse arguments made for the position held by the majority of the members. (For reviews of this literature, see Lamm and Myers, 1978, and McGrath, 1984.) The effect in each of these explanations is reinforcement of extreme positions in the group. This research suggests that single group ultimate decision units will tend toward more extreme behavior (in either the conservative or risky direction) than the predominant leader ultimate decision units. 6 The rationale for the specific hypotheses for each of the types of decision units presented here is develope more detail in Hermann, Hermann, and Hagan (1987). This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms MARGARET G. HERMANN AND CHARLES F. HERMANN 377 It is difficult to compare the behavior patterns of multiple autonomous actors with those of the other two types of decision units. The simple idea that extreme behavior varies inversely with the number of persons whose views must be accommodated, which might have provided a basis for ordering the three types of units, is inconsistent with the polarization effect we have just discussed. The relative autonomy of the participants in multiple autonomous actors units suggests that decision making among them may be more complex and that agreement may involve some compromise or bargaining. A process of mutual concessions (even if not always equal) implies that the collective decision will fall between the initial extreme positions of the individual actors. Thus, the decisions of multiple autonomous actors units are likely to be less extreme than the preferences of their constituent members. There are several difficulties with this perspective, however. First, the argument does not indicate how the decisions of multiple autonomous actors will compare to decisions by single groups or predominant leaders. Second, if the multiple autonomous actors cannot reach agreement-which we noted earlier may be the case under certain conditions-the resulting deadlock represents an extremely conservative position (that is, no action). Despite these caveats, we speculate that because of the greater complexity involved in coalition formation and negotiation among members of a multiple autonomous actors unit, its behavior will be less extreme than that of either the single group or predominant leader units. In effect, we hypothesize that single groups will show the most extreme foreign policy behavior, followed by predominant leaders and, then, by multiple autonomous actors. Assessing the Control Variables To examine our hypotheses empirically, it was necessary not only to identify the ultimate decision units for a series of foreign policy problems but to (a) construct measures of extreme and moderate foreign policy behavior and (b) assess the three key control variables (contextual sensitivity, difficulty in reaching consensus, and relationship among the actors). We will describe our measures of the key control variables first. Contextual Sensitivity. Contextual sensitivity is the control variable that we have proposed differentiates self-contained from externally influenceable decision units when the ultimate decision unit is a predominant leader. The contextual sensitivity of a predominant leader was determined by assessing his conceptual complexity revealed by analyzing press interviews with him; the procedure is described by M. Hermann (1980, 1984).7 This measure of conceptual complexity focuses on the degree of differentiation that an individual shows in characterizing other people, places, policies, ideas, or things. The more conceptually complex person can see varying reasons for a particular position, is willing to entertain the possibility that there is ambiguity in the environment, and is flexible in reacting to objects or ideas. More conceptually complex people are more responsive to cues from their environment and more likely to monitor their environment for information (see Nydegger, 1975; Driver, 1977; Ziller et al., 1977; Streufert and Streufert, 1978). Scores on conceptual complexity were divided at the median to determine which predominant leaders were relatively sensitive to the context and which were relatively insensitive. Those with scores below the median were considered relatively insensitive to their context and, thus, self-contained decision units; those with scores 7 A codebook (M. Hermann, 1983) describing how to assess conceptual complexity through content analysis of press interviews is available from the authors. This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms 378 Who Makes Foreign Policy Decisions and How: An Empirical Inquiry above the median were considered relatively sensitive to the context and, thus, externally influenceable decision units. The median for this sample of predominant leaders was a standard score of 50; scores ranged from 34 to 67. Difficulty in Reaching Agreement. When a single group is the ultimate decision unit, the control variable that differentiates a self-contained from an externally influenceable unit is the ability to reach prompt consensus in response to a problem. In our previous discussion of factors that affect how fast a single group can reach consensus, we noted both structure and process variables. Prompt consensus is more likely if the group is small, members have their primary loyalty to the group, and power is unequally distributed among members. Prompt consensus also is more likely if the information the group receives is from a common source, if the core political beliefs of the group are shared and homogeneous, if affective relations among the group members are harmonious, and if there is little substantive conflict in general over the foreign policy issues facing the nation. The potential for disagreement grows as more of these process and structure conditions are not satisfied because members have a greater diversity of positions, values, information, and perceptions of influence that must be reconciled before consensus is possible (see C. Hermann, 1978, 1979, 1981). To determine how much difficulty a particular single group acting as the ultimate decision unit would have in reaching consensus, we asked area experts to answer a series of questions about the foreign policy-making units in the countries they studied. The questions requested information about group structures and process characteristics like those mentioned in the previous paragraph. As there were eight questions and three degrees of possible disagreement for each question, scores could range from 8 to 24-8 indicating the possibility for reaching prompt consensus, 24 suggesting the likelihood of continuing disagreement. An example of one of the questions may illustrate this scoring process: "Were core political beliefs about foreign policy for this foreign policy-making unit: Very homogeneous, partially shared by members, or very heterogeneous?" A response of "very homogeneous" suggested the potential for reaching prompt consensus and was scored a one; a response of "partially shared by members" indicated some potential for disagreement and was scored a two; and a response of "very heterogeneous" posed the potential for much disagreement among the members and was scored a three. The questions the area experts answered are described in detail in C. Hermann (1981).8 To ascertain which of the single group decision units were self-contained and which externally influenceable, the scores were divided at the median for the sample. The median score was 13, with scores ranging from 9 to 18. Nature of Relationship Among Actors. When the ultimate decision unit consists of multiple autonomous actors, the control variable that differentiates self-contained from externally influenceable units is the nature of the relationship among the parties. Is it zero-sum or non-zero-sum? To assess the relationship among the entities 8 In addition to homogeneity of political beliefs among group members, the other seven questions used to construct this index dealt with the size of the group, the distribution of power among members in the group, the loyalty of group members to the group, the commonality of information sources among group members, substantive conflict among members of the group, affective relations among members of the group, and procedures used to deal with conflict in the group. The reader will note that our measure of difficulty in reaching agreement is not based on the actual performance of the group but on an inference from structural and process characteristics of the group. Since cross-national performance data on single groups is not readily available, we have used a surrogate process to estimate this key control variable by asking area experts to describe aspects of groups that are more easily discernible. A copy of the qulestioninaire the area experts wer-e asked to complete and the instructions they received before responding to the questionnaire are available from the authors. This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms MARGARET G. HERMANN AND CHARLES F. HERMANN 379 in decision units comprised of multiple autonomous actors, we asked area experts to indicate on a four-point scale the degree of controversy surrounding each of the issues the units confronted. Was the issue extremely controversial, so that the parties were deeply divided on what should be done; did the issue generate moderate controversy among the actors; or was there limited or no controversy over the issue? Decision units composed of multiple autonomous actors were considered self-contained (to have a zero-sum relationship) if all the issues they confronted were rated as extremely or moderately controversial by the area experts. Decision units were considered externally influenceable (to have a non-zero-sum relationship) when the issues they faced were rated as involving limited or no controversy. We have made the assumption here that the more beset by controversial issues a cluster of actors is, the more likely they are to have a zero-sum or antagonistic and competitive relationship; the less controversial the issues such decision units face, the greater chance their relationship has of being non-zero-sum or non-adversarial.9 A description of the rating process that the area experts followed in judging the controversial nature of the issue areas is described in C. Hermann (1981). Measuring Foreign Policy Behavzor The measures of foreign policy behavior used in this research come from the Comparative Research on the Events of Nations (CREON) event data set (C. Hermann et al., 1973) and are conceptualized and operationalized in Callahan, Brady, and Hermann (1982). Specifically, the measures determine the type of affect, degree of commitment, and instrument of statecraft involved in a nation's foreign policy behavior. The measures are event-based; that is, measures are assigned to each event in the CREON data set. By estimating the type of unit likely to have been involved in making the decision in each event, we can examine the relationship between the nature of the decision unit and the resulting foreign policy behavior. We matched decision units to events by issue areas. Thus, for example, if a decision unit for a particular nation had ultimate authority for foreign economic issues, it was matched to all events for that nation that dealt with foreign economic problems. Affect. Affect refers to policy makers' hostility or friendship toward other governments as expressed by what they are saying and doing at the moment (see M. Hermann et al., 1982). The affect measure used in this analysis is a three-point scale: (1) positive (friendly, supportive), (2) neutral, and (3) negative (hostile, confrontational). In the present research we are examining the type of affect expressed toward the source of the problem in each of the CREON events. Thus, each event has a positive, neutral, or negative affect score attached to it. Commitment. As defined by Callahan (1982), commitment involves the degree to which a government's current actions limit its future options either through the allocation of resources or the generation of expectations in others. The commitment measure used in the present research is based on the four-point CREON commitment scale which ranges from minimal commitment (symbolic verbal behaviors) to high commitment (the irreversible allocation of resources and the signing of international agreements). In between these two extremes are low and moderate 9 It would clearly be preferable to have a direct indicator of the zero- or non-zero-sum status of the relationship among multiple autonomous actors. But since such a question was not put to the area experts in our study, we are obliged here to use this approximation based on issue conflict. Although the source of the severity of issue conflict was not established, it is possible that it is a function, in whole or part, of the relationship among the actors. Such is the logic for the measuremen-t used here. This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms 380 Who Makes Foreign Policy Decisions and How: An Empirical Inquiry commitment. Each event in the CREON data set was classified at one of these points based on the degree of commitment evidenced in the government's behavior in response to the problem it confronted. For purposes of the present analysis this four-point scale was recoded with the extreme scale points (minimal and high commitment) combined to indicate extreme commitment behavior (coded 2) and the moderate scale points (low and moderate commitment) combined to indicate moderate commitment behavior (coded 1). Instruments of Statecraft. Instruments of statecraft are the skills and resources used in the formation and implementation of foreign policy (see C. Hermann, 1982). In this research each event was classified as involving one of the following three types of instruments: (1) diplomatic instruments; (2) diplomatic in combination with nondiplomatic instruments such as economic, military, propaganda, or science/technology skills and resources; or (3) non-diplomatic instruments. We considered this categorization to be equivalent to a three-point scale. Results and Discussion To examine the hypothesized effects of the key control variables and decision units on foreign policy behavior, we performed a series of two-way analyses of variance. The data were the 5,185 events identified in the CREON data set for the twenty-five countries listed in Table 2 during the decade 1959-68. Each event represented one decision. The results of the analyses of variance are presented in Tables 3 through 5, one table for each of the foreign policy behaviors we examined. Since the number of events differed by type of decision unit, we used an unweighted-means analysis of variance (see Winer, 1962:222-24, 241_44).10 The analysis of variance for expressions of affect is reported in Table 3. The table shows that there are significant main (or independent) effects of both type of decision unit and the key control variables on the kind of affect expressed in foreign policy behavior. As hypothesized, the single group decision units engaged in the most extreme behavior of the three types of decision units, evidencing the most conflictual behavior. Multiple autonomous actors were the least conflictual, with predominant leaders in between. The differences between each type of decision unit and the other two are significant using t-tests (tpL&SG = 10, p < .01; tPL&MAA = 2.8, p < .01; tSG&MAA = 6.71, p < .01). Also as hypothesized, the self-contained decision units were significantly more conflictual-that is, more extreme in their behavior-than the externally influenceable units. The insensitive predominant leader, the single group that can reach consensus quickly, and multiple autonomous actors with a zero-sum relationship were more conflictual than their opposites. Table 4 presents the results of the analysis of variance for the degree of commitment in a government's foreign policy behavior. An examination of these data indicates a significant main effect for type of decision unit and a significant interaction effect for type of decision unit by key control variable. The significant interaction effect suggests that the mean for a particular type of decision unit and condition of the key control variable is not predictable from knowledge of either the type of decision unit or the key control variable by itself. One has to examine the 10 We have used an unweighted-means analysis of variance here instead of a least squares solution because the differences in the number of events by type of decision unit are primarily a function of sources of variation irrelevant to the hypotheses being examined and, thus, should not be allowed to influence the estimation of the population means (see Winer, 1962:224). As reflected in footnote b of Table 2, we believe more detailed case studies of the occasions for decision under study would reveal more ultimate decision units involving multiple autonomous actors and single groups and fewer involving predominant leaders. This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms MARGARET G. HERMANN AND CHARLES F. HERMANN 381 TABLE 3. Analysis of variance for affect by type of ultimate decision unit. Means Across decision Self-contained Externally Decision unit units unit influenceable unit Predominant Leader 1.84 1.85 1.82 (N = 3256) (N = 2750) (N = 506) Single Group 2.04 2.11 1.98 (N = 1680) (N = 801) (N = 879) Multiple Autonomous 1.70 1.77 1.58 Actors (N = 249) (N = 156) (N = 93) Mean 1.91 1.90 (N = 3707) (N = 1478) Summary of analysis of variance Source SS df MS F p Decision Unit 39.25 2 19.62 29.28 <.001 Key Control 5.61 1 5.61 8.37 <.01 Variable Decision Unit by 1.12 2 .56 .84 n.s. Key Control Variable Error 3520.43 5179 .67 TABLE 4. Analysis of variance for commitment by type of ultimate decision unit. Means Across decision Self-Contained Externally Decision unit units unit influenceable unit Predominant Leader 1.59 1.60 1.55 (N = 3256) (N = 2750) (N = 506) Single Group 1.60 1.63 1.58 (N = 1680) (N = 801) (N = 879) Multiple Autonomous 1.61 1.60 1.62 Actors (N = 249) (N = 156) (N = 93) Mean 1.61 1.58 (N = 3707) (N = 1478) Summary of analysis of variance Source SS df MS F p Decision Unit 5.047 2 2.52 10.5 <.01 Key Control .28 1 .28 1.17 n.s Variable Decision Unit by 7.01 2 3.50 14.58 <.01 Key Control Variable Error 1241.80 5179 .24 This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms 382 Who Makes Foreign Policy Decisions and How: An Empirical Inquiry interaction between the variables to understand what is happening (see Winer, 1962:148-49). Focusing on the interaction effect, we note that the means for the predominant leader and single group decision units that are self-contained and externally influenceable support our hypothesis. In other words, the insensitive predominant leader and single group that can reach consensus quickly manifested more extreme commitment behavior than their counterparts. The reverse of our hypothesized relationship, however, holds for the multiple autonomous actors decision units: the self-contained units with a zero-sum relationship among the actors displayed more moderate commitment behavior than the externally influenceable units with a non-zero-sum relationship. A plausible explanation for this difference between the multiple autonomous actors decision units and the other two types of units was posited in our earlier discussion of the hypotheses. Since deadlock is the most probable outcome for multiple autonomous actors with a zero-sum relationship, in contrast to the other self-contained units, their activities are likely to be at one extreme of the spectrum of behavior, that entailing minimal physical action. Using the four-point CREON commitment scale with minimal commitment having a score of one and high commitment a score of four, we note a significant difference between the multiple autonomous actors decision units with a zero-sum and non-zero-sum relationship (t = 1.97, p < .05). Those with a zero-sum relationship engage in behavior that tends more toward minimal commitment (mean = 2.09) than those with a non-zero-sum relationship (mean = 2.40). The interaction effect helps us understand why multiple autonomous actors rather than single groups evince the most extreme commitment behavior and why predominant leaders show the least extreme commitment behavior. For both single group and predominant leader decision units, the condition of the key control variable influences the resulting foreign policy behavior. In the cases of the single group and predominant leader decision units, there is a significant difference between self-contained and externally influenceable units (tSG = 2.08, p < .05; tPL = 2.27, p < .05). While single group decision units that are self-contained show the most extreme commitment behavior of any of the six types of decision units, this behavior moderates when the single groups are externally influenceable-indeed, in this condition single groups rank fifth in degree of commitment among the six types of units. Similarly, the commitment behavior of the predominant leader decision units is more extreme when the units are self-contained than when they are externally influenceable. In fact, predominant leader decision units that are externally influenceable manifest the least extreme commitment behavior of any of the six types of decision units. The data suggest that for degree of commitment of resources, what occurs in single group and predominant leader units that are self-contained is not the same as what occurs in these units when they are externally influenceable-the decision process affects what the government does. The analysis of variance for instruments of statecraft reported in Table 5 shows significant differences among the three types of decision units and among the two values of the key control variables, as well as a significant interaction between the decision units and key control variables. Examining the interaction places the main effects into perspective. As hypothesized, the predominant leaders in self-contained units (the insensitive leaders) use more economic and military instruments of statecraft than those in the externally influenceable units (the sensitive leaders). And the multiple autonomous actors in self-contained units (with a zero-sum relationship) emphasize diplomatic instruments more than their counterparts in externally influenceable units (with a non-zero-sum relationship). Both these differences are significant (tpL = 2.5, p < .05; tMAA = 5.55, p < .01). This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms MARGARET G. HERMANN AND CHARLES F. HERMANN 383 TABLE 5. Analysis of variance for instruments by type of ultimate decision unit. Means Across decision Self-contained Externally Decision unit units unit influenceable unit Predominant Leader 1.55 1.57 1.47 (N = 3256) (N = 2750) (N = 506) Single Group 1.50 1.46 1.54 (N = 1680) (N = 801) (N = 879) Multiple Autonomous 1.78 1.55 2.16 Actors (N = 249) (N = 156) (N = 93) Mean 1.54 1.56 (N = 3707) (N = 1478) Summary of analysis of variance Source SS df MS F p Decision Unit 44.86 2 22.43 32.98 <.001 Key Control 16.82 1 16.82 24.74 <.001 Variable Decision Unit by 36.45 2 18.22 26.80 <.00 1 Key Control Variable Error 3535.49 5179 .68 The difference between single groups in self-contained units and those in externally influenceable units is also significant (tSG = 2.22, p < .05) but in the reverse direction from that hypothesized. Single groups in the self-contained units (those able to reach consensus quickly) focused more on diplomatic instruments of statecraft than those in externally influenceable units (those in continuing disagreement over what should be done). Ability to reach prompt consensus on a response to a problem may be easier when all members can agree to attempt a diplomatic solution-at least at the outset of a problem-avoiding more costly and, perhaps, more controversial instruments. When some members argue that economic or military instruments should be used, more debate and a delay in decision may result. In fact, the significant main effect for the key control variables suggests that one of the results of opening up the decision-making process to forces outside the decision unit may be an increased push for a non-diplomatic solution to the problem. The externally influenceable units used more non-diplomatic instruments than the self-contained units. With the opportunity to exert influence on a decision, a wider variety of interests may become active in pushing for the solution that furthers their mission and status in the government. It may also be the case that when a decision unit contemplates the use of more costly instruments of statecraft, it opens up the process in an effort to gain support for its anticipated action. Given the data, such arguments seem particularly applicable to single groups in continuing disagreement and multiple autonomous actors with a non-zero-sum relationship. Sensitive predominant leaders, on the other hand, may become overwhelmed by the pressure of these outside elements and seek to paper over the differences among the special interests by resorting to a strictly diplomatic initiative. This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms 384 Who Makes Foreign Policy Decisions and How: An Empirical Inquiry The significant main effect for decision unit indicates a difference in the use of diplomatic and non-diplomatic instruments among the three types of decision units. Single group decision units, whether self-contained or externally influenceable, focus more on diplomatic instruments than the other two types of decision units (tSG&PL = 2.02, p < .05; tSG&MAA = 5, p < .01). Although the multiple autonomous actors decision units are the most likely to use non-diplomatic instruments, this result stems primarily from the use of such instruments by the multiple autonomous actors in the externally influenceable units-those with a non-zero-sum relationship. Multiple autonomous actors are likely to include advocates for the various non-diplomatic instruments who probably can succeed in a non-zero-sum environment in seeing that their organizations' skills and resources are involved in a decision even if the decision represents a compromise. Looking across the results in Tables 3 through 5, the data suggest that the nature of the decision unit can have an effect on what a government does in the foreign policy arena. Specifically, the analysis shows that there are differences in behavior among predominant leaders, single groups, and multiple autonomous actors decision units that can be accentuated depending on whether the particular unit is open (externally influenceable) or closed (self-contained) to forces outside itself in the decision-making process. In many respects this research opens up a domain for empirical inquiry in the comparative study of foreign policy. An important next step is to specify what the processes are within each type of decision unit and key control condition that lead to the differences we observed here in foreign policy behavior. A provisional elaboration of these processes is proposed in Hermann, Hermann, and Hagan (1987). The results of the present study add incentive to take this next, more complicated step. Conclusions In this essay we have argued that at the apex of foreign policy making in all governments or ruling parties there are actors with the ability to commit the resources of the government and the power to prevent other entities within the government from reversing their position-the ultimate decision unit. Although this decision unit may change with the nature of the foreign policy problem and with time, its structure will shape a government's foreign policy. Our proposal is that the decision process in each type of decision unit channels the impact of the wider domestic and international environment on foreign policy behavior. In effect, internal and external pressures may predispose a government to act in a particular manner, but the precise character of its actions will be modified by properties of the ultimate decision unit. We have postulated that there are three types of ultimate decision units: predominant leaders, single groups, and multiple autonomous actors. Each of these types of units exists in one of several states or conditions that help to determine whether the decision unit affects foreign policy largely through the pre-existing knowledge, beliefs, and style of those participating in the unit or whether factors outside the decision unit must be taken into consideration in understanding the decision-making process. By ascertaining which of the three types is the ultimate decision unit for a particular foreign policy problem and its state or condition, we can say something about the nature of the foreign policy behavior the decision unit is likely to choose. The framework we have outlined in this paper provides some basis for making cross-national comparisons among governmental decision-making bodies. It defines concepts that can be applied to a variety of different political systems. And it enables us to put into perspective the extensive array of different entities within a This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms MARGARET G. HERMANN AND CHARLES F. HERMANN 385 government that contribute to foreign policy. Furthermore, the framework gives us a means for comparing different types of decision units. In effect, it makes the decision unit a more accessible unit of analysis for the student of comparative foreign policy. References ALLISON, G. T. (1971) Essence of Decision: Explaining the Cuban Missile Crisis. Boston: Little, Brown. ALLISON, G. T. AND P. SZANTON. (1976) Remaking Foreign Policy: The Organizational Connection. New York: Basic Books. ANDERSON, P. A. (1987) Prospects of Viewing Decision Making as a Social Process for the Comparative Study of Foreign Policy. In New Directions in the Study of Foreign Policy, edited by C. F. Hermann, C. W. Kegley, Jr., and J. N. Rosenau, pp. 285-308. London: Allen & Unwin. AUSTIN, W. G. AND S. WORCHEL. EDS. (1979) The Social Psychology of Intergroup Relations. Monterey, Calif.: Brooks/Cole. BASS, B. M. (1981) Stogdill's Handbook of Leadership, rev. ed. New York: Free Press. BRADY, L. P. (1976) Bureaucratic Politics and Situational Constraints in Foreign Policy. Sage International Yearbook of Foreign Policy Studies, vol. 4. CALLAHAN, P. (1982) Commitment. In Describing Foreign Policy Behavior, edited by P. Callahan, L. P. Brady and M. G. Hermann, pp. 177-206. Beverly Hills, Calif.: Sage Publications. CALLAHAN, P., L. P. BRADY AND M. G. HERMANN, EDS. (1982) Describing Foreign Policy Behavior. Beverly Hills, Calif.: Sage Publications. COLLINS, B. AND H. GUETZKOW (1964) A Social Psychology of Group Processes for Decision Making. New York: Wiley. D'AMATO, A. (1967) Psychological Constructs in Foreign Policy Behavior.Journal of Conflict Resolutio 11:294-311. DESTLER, I. M. (1972) Presidents, Bureaucrats, and Foreign Policy. Princeton, N.J.: Princeton University Press. DOOD, L. C. (1 976) Coalitions in Parliamentary Government. Princeton, N.J.: Princeton University Press. DRIVER, M. J. (1977) Individual Differences as Determinants of Aggression in the Inter-Nation Simulation. In A Psychological Examination of Political Leaders, edited by M. G. Hermann, pp. 337-53. New York: Free Press. DRUCKMAN, D. (1977) Negotiations: Social-Psychological Perspectives. Beverly Hills, Calif.: Sage Publica- tions. ETHEREDGE, L. S. (1978) A World of Men: The Private Sources of American Foreign Policy. Cambridge, Mass.: MIT Press. ETHEREDGE, L. S. (1979) Hard-Ball Politics. Political Psychology, 1:1-26. GARDNER, H. (1983) Frames of the Mind: The Theory of Multiple Intelligences. New York: Basic Books. GEORGE, A. L. (1979) The Causal Nexus Between Cognitive Beliefs and Decision-Making Behavior: The "Operational Code" Belief System. In Psychological Models of International Politics, edited by L. S. Falkowski, pp. 95-124. Boulder, Colo.: Westview Press. GEORGE, A. L. (1980) Presidential Decision Making in Foreign Policy. Boulder, Colo.: Westview Press. HALPERIN, M. H. (1974) Bureaucratic Politics and Foreign Policy. Washington, D.C.: Brookings Institution. HALPERIN, M. H . AND A. KANTOR. (1973) The Bureaucratic Perspective: A Preliminary Framework. In Readings in American Foreign Policy, edited by M. H. Halperin and A. Kantor, pp. 1-42. Boston: Little, Brown. HERMANN, C. F. (1972) International Crises: Insights from Behavioral Research. New York: Free Press. HERMANN, C. F. (1978) Decision Structure and Process Influences on Foreign Policy. In Why Nations Act, edited by M. A. East, S. A. Salmore and C. F. Hermann, pp. 69-102. Beverly Hills, Calif.: Sage Publications. HERMANN, C. F. (1979) The Effects of Decision Structures and Processes on Foreign Policy Behavior. Paper presented at the annual meeting of the International Society of Political Psychology, Washington, D.C., May 24-26. HERMANN, C. F. (1981) Judgments of Country Specialists about Foreign Policy Formation: The Operation of Internal Decision Processes. In Expert Generated Data: Applications in International This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms 386 Who Makes Foreign Policy Decisions and How: An Empirical Inquiry Affairs, eclited by G. W. Hopple and J. A. KLihlman, pp. 209-39. BouLlder, Colo.: Westview Press. HERMANN, C. F. (1982) Instruments of Foreign Policy. In Describing Foreign Policy Behavior, edited by P. Callahan, L. P. Brady and M. G. Hermann, pp. 153-74. Beverly Hills, Calif.: Sage Publications. HERMANN, C. F. (1983) Bureaucratic Constraints on Innovation in American Foreign Policy. In Perspectives on American Foreign Policy, edited by C. W. Kegley, Jr. and E. R. Wittkopft, pp. 390-409. New York: St. Martin's Press. HERMANN, C. F., ET AL. (1973) CREON: A Foreign Events Data Set. Sage Professional Papers in International Studies, 2:02-024. HERMANN, M. G. (1978) The Effects of Personal Characteristics of Political Leaders on Foreign Policy. In Why Nations Act, edited by M. A. East, S. A. Salmore and C. F. Hermann, pp. 49-68. Beverly Hills, Calif.: Sage Publications. HERMANN, M. G. (1980) Explaining Foreign Policy Behavior Using Personal Characteristics of Political Leaders. International Studies Quarterly 24:7-46. HERMANN, M. G. (1983) Handbook for Assessing Personal Characteristics and Foreign Policy Orientations of Political Leaders. Mershon Occasional Papers, April. HERMANN, M. G. (1984) Personality and Foreign Policy Decision Making: A Study of Fifty-three Heads of Government. In Foreign Policy Decision Making, edited by D. Sylvan and S. Chan, pp. 53-80. New York: Praeger. HERMANN, M. G. (1987) Leaders' Foreign Policy Orientations and the Quality of Foreign Policy Decisions. In Role Theory and Foreign Policy Analysis, edited by S. G. Walker, pp. 123-40. Durham, N.C.: Duke University Press. HERMANN, M. G., C. F. HERMANN ANDJ. D. HAGAN. (1987) How Decision Units Shape Foreign Policy Behavior. In New Directions in the Study of Foreign Policy, edited by C. F. Hermann, C. W. Kegley, Jr. and J. N. Rosenau, pp. 309-36. London: Allen & Unwin. HERMANN, M. G., C. F. HERMANN AND G. L. HUTCHINS. (1982) Affect. In Describing Foreign Policy Behavior, edited by P. Callahan, L. P. Brady and M. G. Hermann, pp. 207-22. Beverly Hills, Calif.: Sage Publications. HINCKLEY, B. (1981) Coalitions and Politics. New York: Harcourt Brace Jovanovich. HOLSTI, 0. R. (1976) Foreign Policy Formation Viewed Cognitively. In Structure of Decision, edited by R. Axelrod, pp. 18-54. Princeton, NJ.: Princeton University Press. JANIS, I. L. (1972) Victims of Groupthink. Boston: Houghton Mifflin. JONSSON, E., ED. (1982) Cognitive Dynamics and International Politics. New York: St. Martin's Press. LAMM, H. AND D. G. MYERS. (1978) Group Induced Polarization of Attitudes and Behavior. In Advances in Experimental Social Psychology, Vol. 11, editecl by L. Berkowitz., pp. 145-95. New York: Academic Press. LEISERSON, M. (1970) Power and Ideology in Coalition Behavior: An Experimental Study. In The Study of Coalition Behavior, editecl by S. Groennings, E. W. Kelley and M. Leiserson, pp. 323-35. New York: Holt. McGRATH, J. E. (1984) Groups: Interaction and Performance. Englewood Cliffs, NJ.: Prentice-Hall. NEUSTADT, R. E. (1970) Alliance Politics. New York: Columbia University Press. NYDEGGER, R. V. (1975) Information Processing Complexity and Leadership Status. Journal of Experimental Social Psychology 11:317-26. PERLMUTTER, A. (1981) Modern Authoritarianism: A Comparative Institutional Analysis. New Haven, Conn.: Yale University Press. ROSATI, J. A. (1985) Developing a Theory of Foreign Policy: The Evolution of American Foreign Policy. Paper presented at the Conference on New Directions in the Comparative Study of Foreign Policy, Ohio State University, May 9-11. SALMORE, B. G. AND S. A. SALMORE. (1978) Political Regimes and Foreign Policy. In Whly Nations Act, edited by M. A. East, S. A. Salmore and C. F. Hermann, pp. 103-22. Beverly Hills, Calif.: Sage Publications. SEMMEL, A. K. (1982) Small Group Dynamics in Foreign Policy Making. In Biopolitics, Political Psychology, and International Politics, edited by G. W. Hopple, pp. 94-113. New York: St. Martin's Press. SNYDER, M. (1974) Self-Monitoring of Expressive Behavior. Journal of Personality and Social Psychology 30:526-37. SNYDER, M. (1979) Self-Monitoring Processes. In Advances in Experimental Social Psychology, Vol. 12, edited by L. Berkowitz, pp. 86-128. New York: Academic Press. This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms MARGARET G. HERMANN AND CHARLES F. HERMANN 387 SNYDER, M. (1982) When Believing Means Doing: Creating Links Between Attitudes and Behavior. In Consistency in Social Behavior: The Ontario Symposium, Vol. 2, edited by M. Zanna, E. Higgins and C. Herman, pp. 1 Hillsdale, N.J.: Lawrence Erlbaum. STEINBRUNER, J. D. (1974) The Cybernetic Theory of Decision. Princeton, N.J.: Princeton University Press. STOESSINGER, J. G. (1979) Crusaders and Pragmatists: Movers of American Foreign Policy. New York: W. W. Norton. STONER, J. A. F. (1961) A Comparison of Individual and Group Decisions Involving Risk. Unpublished Masters Thesis, Massachusetts Institute of Technology. STREUFERT, S. AND S. STREUFERT. (1978) Behavior in the Complex Environment. Washington, D.C.: V. H. Winston. STUART, D. AND H. STARR. (1981-82) The "Inherent Bad Faith Model" Reconsidered: Dulles, Kennedy, and Kissinger. Political Psychology 3:1-33. TETLOCK, P. E. (1979) Identifying Victims of Groupthink from Public Statements of Decision Makers. Journal of Personality and Social Psychology 37:1314-24. WALKER, S. G. (1977) The Interface Between Beliefs and Behaviors: Henry Kissinger's Operational Code and the Vietnam War. Journal of Conflict Resolution 21:129-68. WALKER, S. G. (1983). The Motivational Foundations of Political Belief Systems: A Re-Analysis of the Operational Code Construct. International Studies Quarterly 27:179-201. WALLACH, M. A., N. KOGAN AND D. BEM. (1962) Group Influence on Individual Risk Taking.Journal of Abnormal and Social Psychology 65:75-86. WILDER, D. A. (1986) Social Categorization: Implications for Creation and Reduction of Intergroup Bias. In Advances in Experimental Social Psychology, Vol. 19, edited by L. Berkowitz, pp. 293-355. New York: Academic Press. WINER, B. J. (1962) Statistical Principles in Experimental Design. New York: McGraw-Hill. ZILLER, R. C., ET AL. (1977) Self-Other Orientations and Political Behavior. In A Psychological Examination of Political Leaders, edited by M. G. Hermann, pp. 176-203. New York: Free Press. This content downloaded from 147.231.63.14 on Fri, 13 Mar 2020 21:00:43 UTC All use subject to https://about.jstor.org/terms