We use cookies to distinguish you from other users and to provide you with a better experience on our websites. Close this message to accept cookies or find out how to manage your cookie settings.
Published online by Cambridge University Press: 18 July 2011
The quotations from Charles A. McClelland and Graham T. Allison represent two distinct approaches to the study of international relations: (i) international system analysis; and (2) foreign policy analysis. Essentially, international system analysts seek to explain interactions between nations by phenomena such as their prior interactions and the structure of the system. Foreign policy analysts, on the other hand, seek to explain foreign policy behavior as the output of subnational organizations following standard operating procedures or engaging in a problem-solving search. Given the international system and foreign policy approaches as contrasting points of departure, the goals of the present study are:
1 McClelland, Charles A., Theory and the International System (New York 1966), 20, 104.Google Scholar
2 Allison, Graham T., Essence of Decision: Explaining the Cuban Missile Crisis (Boston 1971), 87.Google Scholar
3 Singer, J. David, “The Level-of-Analysis Problem in International Relations,” in Knorr, Klaus and Verba, Sidney, eds., The International System (Princeton 1961), 80.Google Scholar It should be noted that the interaction approach is often distinguished from the international system approach. The latter orientation is based on the assumption that international politics is more than the sum of converging interactions and transactions; properties of the system as a whole are assumed to influence the behavior of individual nations.
4 The term strategic interaction in game theory often refers to the outcome of competing strategies. Here, interaction means the process where each actor pays attention to and responds to the prior patterns of his opponent.
5 See Simon, Herbert, “Some Strategic Considerations in the Construction of Social Science Models,” in Lazarsfeld, Paul, ed., Mathematical Thinking in the Social Sciences (Glencoe 1954), 388–415.Google Scholar Also see Simon, Herbert, Models of Man: Social and Rational; Mathematical Essays on Rational Human Behavior in a Social Setting (New York 1957), 241–60.Google Scholar
6 See Rapoport, Anatol, Two-Person Game Theory (Ann Arbor 1966).Google Scholar
7 Richardson, Lewis F., Arms and Insecurity: A Mathematical Study of the Causes and Origins of War (Pittsburgh 1960).Google Scholar
8 Ibid., 12.
9 Rapoport, Anatol, Fights, Games and Debates (Ann Arbor 1960), 15–107Google Scholar; and “Lewis F. Richardson's Mathematical Theory of War,” Journal of Conflict Resolution, 1 (September 1957), 249–99.CrossRefGoogle Scholar See also Boulding, Kenneth E., Conflict and Defense: A General Theory (New York 1962)Google Scholar; Smoker, Paul, “Fear in the Arms Race: A Mathematical Study,” in Rosenau, J. N., ed., International Politics and Foreign Policy (2nd ed., New York 1969), 573–82.Google Scholar
10 McGuire, Martin C., Secrecy and the Arms Race (Cambridge, Mass. 1965).Google Scholar
11 North, Robert C., “Research Pluralism and the International Elephant,” in Knorr, Klaus and Rosenau, James, eds., Contending Approaches to International Politics (Princeton 1969), 218–42Google Scholar; North, Robert C., “The Behavior of Nation-States: Problems of Conflict and Integration,” in Kaplan, Morton, ed., New Approaches to International Relations (New York 1968), 203–356Google Scholar; Charles A. McClelland and Gary D. Hoggard, “Conflict Patterns in the Interactions Among Nations,” in Rosenau (fn. 9), 711–24.
12 McClelland, Charles A., “The Acute International Crisis,” in Knorr and Verba (fn. 3), 182–204; “Access to Berlin: The Quantity and Variety of Events, 1948–1963,” in Singer, J. David, ed., Quantitative International Politics (New York 1968), 159–86.Google Scholar Event/interactions are international actions such as threats and promises (words) or uses of force and offers of proposals (deeds). Event/interactions are different from transactions such as trade and mail flows between nations. The present study deals only with connective event/interactions since there were too few cooperative interactions during the Berlin conflict of 1961 to perform statistical analysis.
13 See below, however, for a discussion of how recurring event sequences may be subsumed under learning models and how such models explain limited rational search behavior.
14 The mediated S-R model draws on internal attributes (perceptions) more than the other models. Similarly, game theory models applied to world politics focus on the rational intentions of decision-makers, which tap internal attributes of nations. A major criticism of game theory models, however, is their treatment of an actor as a black-box, ignoring psychological and behavioral attributes. See Harsanyi, John C., “Rational-Choice Models of Political Behavior vs. Functionalist and Conformist Theories,” World Politics, XXI (July 1969), 513–38CrossRefGoogle Scholar; Shapiro, Michael, “Rational Political Man: A Synthesis of Economic and Social-Psychological Perspectives,” American Potitical Science Review, LXIII (December 1969), 1106–19.CrossRefGoogle Scholar Simon modified game theory by incorporating attributes of the actor and then inferring a new decision-rule—“satisficing” (Simon, fn. 5, 241–60). Experimental gaming explicitly treats properties of the actors such as competitiveness, risk, and temptation, as well as rewards and punishment. Melvin Guyer, “A Review of the Literature on Zero-Sum and Non-Zero-Sum Games in the Social Sciences,” Mental Health Research Institute, University of Michigan, Mimeo, n.d.
15 Snyder, Richard C. and others, eds., Foreign Policy Decision-Making (New York 1962)Google Scholar; Robinson, James A. and Snyder, Richard C., “Decision-Making in International Politics,” in Kelman, Herbert C., ed., International Behavior (New York 1965), 433–63Google Scholar; Paige, Glenn, The Korean Decision (New York 1968)Google Scholar; Hermann, Charles F., Crises in Foreign Policy: A Simulation Analysis (Indianapolis 1969)Google Scholar; Robinson, J. A. and others, “Search Under Crisis in Political Gaming and Simulation,” in Pruitt, D. G. and Snyder, R. C., eds., Theory and Research on the Causes of War (Englewood Cliffs, N.J. 1969), 80–94.Google Scholar
16 Snyder, Richard C. and Paige, Glenn D., “The United States Decision to Resist Aggression in Korea: The Application of an Analytical Scheme,” in Rosenau, J. N., ed., International Politics and Foreign Policy (New York 1961), 196.Google Scholar
17 Simon (fn. 5); March, James G. and Simon, Herbert A., Organizations (New York 1958)Google Scholar; Cyert, Richard M. and March, James G., A Behavioral Theory of the Firm (Englewood Cliffs, N.J. 1963).Google Scholar
18 Allison (fn. 2).
19 Simon (fn. 5).
20 Allison (fn. 2), 87. Allison's “explanation” of present behavior as determined by prior behavior is not an explanation in the sense of specifying why the present behavior occurs. A learning model may be able to explain why organizations repeat or deviate from prior patterns.
21 Sorensen, Theodore C., Kennedy (New York 1965), 587.Google Scholar
22 This interpretation of the organizational model seems to imply that Kennedy increased the U.S. military budget because of his dissatisfaction with the State Department. External factors such as the WTO threat clearly should be considered to explain the increase in the military budget in this case.
23 See the essay by Nazli Choucri and R. C. North in this volume.
23a The distinction between event/interaction and organizational processes is for the sake of convenience of presentation. In a sense, there is only one model that contains interaction and organization parameters. Interaction parameters may be relatively more important at times, while organizational factors may be more significant at other times. See Tanter, 1972, for a more complete synthesis of interaction and organizational parameters than given here.
24 McClelland, in Knorr and Verba (fn. 3).
25 McClelland (fn. 1), chapter 4.
26 McClelland, in Singer (fn. 12) 159–86.
27 McClelland, in Knorr and Verba (fn. 3), 200–201. Note that one can explain event/interaction processes with an organizational model, a partial synthesis of the approaches of McClelland and Allison. Also, McClelland actually uses the term crisis where the interpretation in the text above refers to conflicts. The word crisis refers to the most intense phase of a conflict in the present study.
28 Young, Robert A., “Prediction and Forecasting in International Relations: An Exploratory Analysis,” unpub. Ph.D. diss., University of Southern California (June 1970).Google Scholar
29 Kaplan, Abraham, Conduct of Inquiry (San Francisco 1964), 332.Google Scholar
30 Hempel, Carl G., “Deductive-Nomological vs. Statistical Explanation,” in Feigl, H. and Maxwell, G., eds., Minnesota Studies in the Philosophy of Science (Minneapolis 1962), 98–169.Google Scholar
31 Langton, Kenneth P., Political Socialization (New York 1969), 3.Google Scholar
32 Simon (fn. 5), 274.
33 Learning models, unlike game theory, use a more bounded concept of rationality. Goals may not be ranked, and search for an alternative which satisfies a goal replaces choice of an optimal alternative.
34 McClelland, in Singer (fn. 12); McClelland and Hoggard, in Rosenau (fn. 9), 711–24.
35 Acknowledgments to Judith Tanter for assistance with the behavioral modification analogy. Subsequently, McClelland and his associates have begun to use learning models in their World Event/Interaction Survey. Thanks to Gary Hoggard and John Sigler for bringing these learning models to the author's attention. See McClelland's, “Verbal and Physical Conflict in the Contemporary International System,” Mimeo, August 1970, especially 4–8.Google Scholar
36 Young, Oran R., The Politics of Force: Bargaining During International Crises (Princeton 1968), 19, 28Google Scholar; Schelling, Thomas C., The Strategy of Conflict (Cambridge, Mass, 1960), 15–16.Google Scholar Note also that evidence suggests that perceptions become more important the more intense the conflictive interactions. See Ole Holsti and others, “Perception and Action in the 1914 Crisis,” in Singer (fn. 12), 123–58.
37 The temporal domain of the present study differs from the Choucri-North study. They base their study on observations covering the period 1870–1914, while the present study concerns the eight-month period immediately prior, during, and after the intense conflict over Berlin in 1961. While the important events in the Choucri-North study unfold over a period of years or even decades, the theoretically meaningful unit of time in the present study is a period of days.
38 Allison (fn. 2), explicitly acknowledges other models of foreign policy decisionmaking, e.g., Allison's rational actor model explicitly includes interaction.
39 Cyert and March (fn. 17), 101 and 113.
40 Allison (fn. a), 85.
41 Feldman, Julian and Kanter, Herschel, “Organizational Decision-Making,” in March, James G., ed., Handbook of Organizations (Chicago 1965), 662.Google Scholar
42 Donald W. Taylor, “Decision-Making and Problem Solving,” in March, ibid., 662.
43 Cyert and March (fn. 17).
44 Wildavsky, Aaron B., The Politics of the Budgetary Process (Boston 1964)Google Scholar, II ff.; also cf. Lindblom, Charles E., “The Science of Muddling Through,” Public Administration Review, XXXVI (Spring 1959), 79–88CrossRefGoogle Scholar; Braybrooke, David and Lindblom, Charles E., A Strategy of Decision: Policy Evaluation as a Social Process (New York 1963).Google Scholar
45 Crecine, John P., Governmental Problem-Solving: A Computer Simulation of Municipal Budgeting (Chicago 1969), 219Google Scholar; “Defense Budgeting: Organizational Adaptation to External Constraints,” RAND Corporation (March 1970).Google Scholar
46 Hunter, Floyd, Community Power Structure: A Study of Decision Makers (Chapel Hill 1953).Google Scholar
47 Dahl, Robert A., Who Governs? Democracy and Power in an American City (New Haven 1961).Google Scholar
48 SOP's in bureaucracies imply long-term stability of behavior, while the present analysis treats continuity of action over periods of several days. Nonetheless, the organizational literature may provide useful analogies for the study of short-term conflict.
49 Halperin, Morton H., Bureaucratic Politics and Foreign Policy, The Brookings Institute (March 1970).Google Scholar
50 Ibid.; Allison (fn. 2).
51 Blalock, Hubert M. Jr., “The Measurement Problem: A Gap between the Language of Theory and Research,” in Blalock, Hubert M. Jr. and Blalock, Ann B., eds., Methodology in Social Research (New York 1968), 5–27.Google Scholar
52 See the article by Oran R. Young in this volume regarding strategies that stress logical closure and those that emphasize the search for empirical regularities.
53 The term action intensity includes both word and deed intensities.
54 Corson, Walter H., “Conflict and Cooperation in East-West Relations: Measurement and Explanation,” paper delivered at the 66th Annual Meeting of the American Political Science Association, Los Angeles, September, 1970.Google Scholar Also, see Whiting, Allen S., “United States-Chinese Political Relations,” The University of Michigan, Mimeo, 1970, 17.Google Scholar
55 Corson constructed the scale in two phases. He administered questionnaires to 53 citizens of 13 non-Western and Western countries. In the first phase, there were 54 conflictive actions arranged in irregular order. With each action printed on a separate card, respondents arranged the actions in rank-order of increasing intensity. The responses from these questionnaires constituted information to compute a mean rankorder for each action, resulting in a 54-item rank-order conflict intensity scale. In the second phase, respondents had 14 conflictive actions selected from the original group of 54; these actions covered the full range of intensity. They were printed on separate cards and presented to respondents in irregular order. Respondents assigned a number to each action proportional to its intensity as they perceived it. Using the responses from these questionnaires, the geometric mean for each event reflected its intensity across respondents. From these data, he developed a 14-item conflict intensity scale and assigned intensity values by interpolation to the remaining 40 conflictive actions. Details of the scaling project are given in Corson, Walter H., “Conflict and Cooperation in East-West Crises: Dynamics of Crisis Interaction,” unpublished Ph.D. thesis, Harvard University, December, 1970.Google Scholar
56 The conflict phases outlined in this paper are based on empirical data from a specific conflict and describe only that conflict. Work is under way by the author and his colleagues on the development of a process model of conflict which will draw on this analysis but not be limited to it.
57 Corson originally identified five conflict phases: pre-crisis, intensification, peak, reduction, and post-crisis. For the present analysis, crisis includes intensification, peak, and reduction. Corson (fn. 55).
58 Holsti and others, in Singer (fn. 12); Hermann (fn. 15); Corson (fn. 55). Thanks to Paul Smoker for his thoughts on the study of time.
59 Corson (fn. 55), 186. Corson speculated that time period should be an aggregation of days rather than a single day. The criteria he employed are three: (1) if total conflictive intensity for NATO and WTO on a given day was less than 30 on the Corson scale, the intensity of conflictive actions on that day and the preceding six days predicted the intensity of conflictive actions for the next three days; (2) if total given intensity on a given day was between 30 and 150, action intensity on that day and the preceding four days predicted action intensity for the next three days; and (3) if total intensity on a given day was greater than 150, action intensity on that day and the preceding two days predicted deed intensity for the next two days. There are at least three difficulties with this method. First, it is difficult to implement this aggregation scheme without overlapping time periods. Second, the method is discontinuous when it should most properly be of the form
where t is the aggregation period, I, the conflictive intensity, and, a, the proportionality constant. As the intensity gets large the aggregation period gets small. Third, the method only represents more intuition than empirical finding.
60 Multi-lagged models were run under two hypotheses: (1) the connective intensity on any given day would be some linear combination of the conflictive intensities of the previous six days; (2) conflictive intensity would have a decreasing effect as time from the present increased. Neither of these two hypotheses were supported by the models or the data. The only different model which arose out of this analysis is found in footnote 73.
61 Aggregating to the alliance level as in the present study may result in a lack of fit between an organizational model and the alliance. The study assumes, however, that organizational models are equally valid irrespective of the level of analysis.
62 In the Berlin conflict of 1961, 28% of WTO actions recorded involved other WTO members acting with or without the U.S.S.R.; 48% of all NATO actions recorded involved other NATO members acting with or without the United States. See ibid.
63 Here is a summary of the methodology: The independent variables are prior WTO and/or prior NATO action intensities. Both word and deed intensity comprise the action category. The author standardized action intensity within the three conflict phases for each alliance, e.g., action intensity had a mean of zero and a standard deviation of unity, pre-conditions for path analysis. Standardized NATO and WTO action intensities were each regressed on standardized prior NATO and WTO action intensities, resulting in the path coefficients.
64 For a historical overview of the Berlin crisis, see: Bailey, George, “The Gentle Erosion of Berlin,” The Reporter (April 26, 1962)Google Scholar; Horelick, Arnold L. and Rush, Myron, “The Political Offensive Against Berlin,” Strategic Power and Soviet Foreign Policy (Chicago 1965)Google Scholar, chap. 10; Keller, John W., Germany, the Wall and Berlin: Internal Policies During an International Crisis (New York 1964)Google Scholar; Edward Smith, Jean, “Berlin: The Erosion of a Principle,” The Reporter (November 21, 1963)Google Scholar; Edward Smith, Jean, The Defense of Berlin (Baltimore 1963)Google Scholar; Speier, Hans, Divided Berlin (New York 1961)Google Scholar; Schick, Jack M., “The Berlin Crisis of 1961 and U.S. Military Strategy,” Orbis, VIII (Winter 1965)Google Scholar; Young, Oran R., The Politics of Force (Princeton 1966)Google Scholar; Charles McClelland, “Access to Berlin: The Quantity and Variety of Events, 1948–1963,” in Singer (fn. 12), 159–86.
65 For the purposes of this study, 1 May 1961 is the beginning of the Berlin conflict. This establishes a base line period several weeks prior to the WTO ultimatum in early June.
66 Corson (fn. 55).
67 Corson, ibid.
68 The meetings between Soviet Premier Khrushchev and Belgian Foreign Minister Spaak on 18–19 September mark the transition to the post-crisis phase. The analysis ends on 31 December 1961 because the frequency and intensity of actions began to approach the pre-crisis level of June.
69 Note that when variables are lagged, you lose one degree of freedom. Thus, N is always smaller than the number of conflict days in each phase.
70 Besides the organizational processes model, there are several other foreign policy type models that might explain incremental outputs during the pre- and post-crisis phases (cf. Allison and Halperin in this volume).
71 The path coefficients for the entire period can be seen in the following diagram:
72 See Donald E. Stokes, “Compound Paths in Political Analysis,” The University of Michigan, Mimeo, n.d.
73 This study does not use statistical inference procedures in evaluating the models. Here is an alternate model for NATO in the crisis phase, the only instance where the author felt the multiple lags contributed new information:
Here is an indication of the strength of the interaction relationships. NATO reacts strongest to WTO actions lagged by four days.,
74 Results of an earlier analysis using the aggregation periods defined in footnote 59 supported the hypotheses that organizational processes were more important in the preand post-crisis phases, and that organizational processes may have been important in the crisis phase as well. Because of the reasons stated in the text, most especially because of the ecological fallacy, these results are not presented here. Although the ecological fallacy demonstrates little effect on the functional relationship between variables, the regression coefficient, it has profound effects on the strength of that association, the beta weight. In path analysis only the beta weight is presented, which would be inflated because of the aggregation of days. Thus, it may be misleading to draw inferences, based as they might be on an artifact of aggregation. For a reference, see Blalock, Hubert M. Jr., Causal Inferences in Nonexperimental Research (Chapel Hill 1964), 97–114.Google Scholar
73 Given the alternative model in footnote 73, it appears that the event/interaction model has more explanatory power for NATO during the crisis phase.
76 Tanter, Raymond, “Dimensions of Conflict Behavior Within and Between Nations, 1958–1960,” Journal of Conflict Resolution, X (March 1966), 41–64.CrossRefGoogle Scholar
77 Tanter, Raymond, The Berlin Crises: Modelling and Managing International Conflicts (forthcoming, 1972).Google Scholar
78 This study, however, does not compare intensities for the three Berlin conflicts; rather, it only has data on the Berlin conflict of 1961. Thus, there are no hard data presented here on the routinization of conflict decision-making.
79 Process modelling is a research strategy designed to disaggregate a complex set of interrelated events and behaviors into stages representing discrete actions or distinct choice points. Process models serve several useful purposes. First, they direct our attention to processes such as learning, forgetting, or precedent search which underlie highly complex patterns of behavior. Thus, process models reduce complex situations to their basic elements, permitting an economy of description and explanation. Finally, process modelling could explain the breakpoints in a conflict—those points where the internal dynamics give way to external factors.
80 Halperin (fn. 49), 50.
81 Ibid., 9.
82 As stated previously, however, one must be careful to avoid selecting historical events in order “to prove” one's hypothesis. Thanks to Alexander George for the critique of the incrementalist thesis regarding the quality of programs.
83 Hayward R. Alker, Jr. and Cheryl Christensen, “From Causal Modelling to Artificial Intelligence: The Evolution of a U.N. Peace-Making Simulation,” Massachusetts Institute of Technology, Mimeo, n.d.
84 lbid, 21.
85 Bloomfield, Lincoln and Beattie, Robert, “Computers and Policy-Making: The CASCON Experiment,” Journal of Conflict Resolution, XI (March 1971)Google Scholar; Beattie, Robert, and Bloomfield, Lincoln, CASCON: Computer-Aided System for Handling Information on Local Conflicts (Cambridge, Mass. 1969)Google Scholar; also cf. Howe, Fisher, The Computer and Foreign Affairs (Washington 1967).Google Scholar
86 Bloomfield, Lincoln and Leiss, Amelia, Controlling Small Wars: A Strategy for the igjo's (New York 1969).Google Scholar
87 Paige (fn. 15).
88 Morgan, G. A., “Planning in Foreign Affairs: The State of the Art,” Foreign Affairs, XXXIX (January 1961), 278.Google Scholar The thrust of Morgan's argument is for selective planning. However, some authors advocate more planning—Ausland, J. C. and Richardson, J. F., “Crisis Management: Berlin, Cyprus, Laos,” Foreign Affairs, XLIV (January 1966), 291–303.CrossRefGoogle Scholar
89 Knorr, Klaus and Morgenstern, Oskar, Political Conjecture in Military Planning, Princeton University, Center of International Studies, Policy Memorandum No. 35 (1968), 10–15.Google Scholar
90 A conflict intensity scale produces more general classes than raw event data. That is, the scales allow an analyst to aggregate across a variety of events to calculate a general intensity score for the actor.
91 March, James G., “Some Recent Substantive and Methodological Developments in the Theory of Organizational Decision-Making,” in Ranney, Austin, ed., Essays on the Behavioral Study of Politics (Urbana 1962), 191–208.Google Scholar
92 Milburn, Thomas W., “The Management of Crisis,” Mimeo, 1970.Google Scholar
93 George, Alexander and others, The Limits of Coercive Diplomacy (Boston 1971), 8–15.Google Scholar
94 Cf. Azar, Edward, “Analysis of International Events,” Peace Research Reviews, IV (November 1970), 83.Google Scholar Azar asserts that, “We code events and measure their violence content with the 13 point interval scale. Although we realize that participants to a conflict situation do not use such an objective instrument, we maintain that they employ an implicit (or possibly explicit) scale which ranks signals by their violence content.” Also see Garrison, William A. and Modigliani, Andre, Untangling the Cold War: A Strategy for Testing Rival Theories (Boston 1971)Google Scholar, for an attempt to quantify and scale East-West interactions.
95 Also, see Sidney Verba, “Assumptions of Rationality and Non-Rationality in Models of the International System,” in Knorr and Verba (fn. 3), 93–117. Acknowledgments to Dennis Doolin for calling attention to the danger of freezing options on the basis of historical precedents with a system such as CACIS. There is a great need for what Doolin calls “… creative politics—which is really the essence and true genius of politics—and there seems to be a danger in an approach that could view routinization as a rule of action.” Letter from Dennis Doolin, 28 June 1971. CACIS attempts to address itself to Doolin's perceptive critique and to facilitate “creative politics.”
