¹ War and the Minds of Men (New York 1950), 1.Google Scholar

² Guetzkow, Harold S.Brody, Richard A.Driver, Michael, and Beach, P., An Experiment on the N-County Problem through Simulation, Social Science Institute, Washington University (St. Louis 1960).Google Scholar

³ Brody, , “Some Systemic Effects of the Spread of Nuclear Weapons Technology: A Study through Simulation of a Multi-Nuclear Future,” Journal of Conflict Resolution, VII (December 1963), 663–753.CrossRefGoogle Scholar

⁴ Zinnes, , “The Expression and Perception of Hostility in Pre-War Crisis: 1914,” in Singer, J. David, ed., International Yearbook of Political Behavior Research (New York, forthcoming 1966).Google Scholar

⁵ The collections from which data was acquired were these: Germany: Montegelas, Max and Schuecking, Walter, eds., Outbreak of the War: German Documents Collected by Karl Kautsky, trans. Carnegie Endowment for International Peace (New York 1924)Google Scholar; Austro-Hungarian Monarchy: Ministerium des K. und K. Hauses und des Aeusseren, Oesterreich-Ungarns Aussenpolitik von der bosnischen Krise 1908 bis zum Kriegsausbruch 1914: Diplomatische Aktenstuecke des oesterreich-ungarischen Ministeriums des Aeusseren, ed. Bittner, Ludwig and Pribram, Alfred (Vienna and Leipzig 1930)Google Scholar; Russia: Komissiia po izdaniiu dokumentov epokhi imperializma [Commission for the Publication of Documents of the Era of Imperialism], Mezhdunarodnye otnosheniia v epokhu imperializma [International Relations in the Era of Imperialism], dokumenty iz arkhivov tsarskogo i vremennogo pravital'stv 1878–1915 gg., seria 3, IV, v (Moscow and Leningrad 1934)Google Scholar; France: Commission for the Publication of Documents Relative to the War of 1914, Documents diplomatique français (1871–1914), 3rd Series, X, XI (Paris 1936)Google Scholar; Great Britain: Foreign Office, British Documents on the Origins of the War, 1898–1914, XI, ed. Gooch, G. P. and Temperley, Harold (London 1926).Google Scholar

⁶ Hermann, Charles F., “A Validity Problem: Crises in Simulated Foreign Policy Organizations,” unpubl. paper delivered at Annual Meeting, American Political Science Association, September 1964.Google Scholar

⁷ Hypotheses 1–5, 10, and n were originally postulated and tested by Brody (pp. 713–46). The terms “alliance” and “bloc” will be used interchangeably to mean mutual ties of friendship and support between members.

⁸ Brody, 732.

⁹ For a complete description of the simulate data see Brody. A complete description of the 1914 data can be found in Zinnes.

¹⁰ Leary, Timothy F., Interpersonal Diagnosis of Personality (New York 1957).Google Scholar

¹¹ In contrast to most historical periods, there exists for the 1914 crisis a reasonably complete set of documents. This was due in part to the circumstances surrounding the conclusion of the First World War in which three of the governments of major-state participants (Germany, Austria-Hungary, and Russia) were overthrown. The new governments, in their rush to discredit the old, published all documents related to the war, particularly those that might incriminate the previous governments. The existence of these documents, and the absence of comparable documents for England and France, seemed to suggest that these latter two countries had something to hide. Revisionist historians subsequently charged England and France with direct responsibility for the outbreak of the war. Before long, England and France also felt compelled to publish most of their documents for the six-week crisis period. Consequently, we have a reasonably complete set of documents for this period. Only two omissions are significant: (1) Because of the British law requiring a fifty-year waiting period before the release of some official documents, not all British documents were available for research when the study began, and (2) we have almost no Serbian documents. Attempts were made to compensate for the latter, more serious, difficulty, but it did remain a problem throughout the study.

¹² A more detailed account of the content-analysis procedures utilized in this study can be found in Zinnes; and in North, Robert C.Holsti, Ole R.Zaninovich, M. George, and Zinnes, Dina A., Content Analysis (Evanston 1963).Google Scholar

¹³ For a complete description of the scaling procedure used see Zinnes; and North and others, 79–90.

¹⁴ Several additional tests were made in an attempt to solve this general question of comparability between the simulate and 1914 measure of xEy. Unfortunately, none of the tests was conclusive. Ideally, we would have liked simply to compute the Leary hostility index for the 1914 data, but the 1914 documents had not been coded for all the necessary themes. This deficiency led to a second approach which consisted of determining the relationship in the simulate data between the Leary hostility index and the three hostility themes on which it was based. If high positive correlations had been obtained it could have been argued that the Leary index was equivalent to the simple frequency counts of the hostility themes, and it would then have been reasonable to use the frequency of expressions of hostility in the 1914 data. The results, shown in the Appendix, did not support this contention: Surprisingly, the correlations between the frequency of each hostility theme separately and the sum of the frequencies for all three hostility themes against the Leary hostility index were all small and negative. This paradoxical result could not be explained and bears further study.

An additional attempt was made to solve the comparability problem. This consisted of recomputing those analyses made by Brody in which the xEy variable appeared, using the original raw data from the simulation, i.e., using the frequencies for each of the three themes and their sum. With only two exceptions, these results were also perplexing. In almost every case, when the Leary hostility index provided data that supported the hypotheses, the 1914 data also supported the hypotheses. In contrast, for these same tests, the use of any of the three hostility themes or their sum failed to produce adequate support for the hypotheses. These results also demand further investigation. For comparison purposes we have reported, in addition to the results obtained for the Leary hostility index, the results for the sum of the three hostility themes, for each relevant hypothesis tested.

¹⁵ Siegel, Sidney, Nonparametric Statistics (New York 1956).Google Scholar

¹⁶ The .05 level of significance is used throughout the study as the criterion for rejecting or accepting a hypothesis. U's are reported for the 1914 data since the n's in this case did not exceed 20 and U tables were applicable (n ₁ = 15, n ₂ = 16); the simulation data, containing seven rather than six nations, produced n's greater than 20 (n ₁ = 18, n ₂ = 24) which made U tables inapplicable and necessitated the use of z scores.

¹⁷ Since Hypotheses 5 through 8 involve the xEy variable, results have been given for both the Leary hostility index and the sum of the three themes. Note the generally poor performance of the sum of the three themes, and the perplexing negative trend of these correlations.

¹⁸ The use of a nonparametric statistic rather than the Pearson product moment correlation was dictated by the desire to use comparable statistics in both sets of data; the hypotheses tested by Brody used the Spearman formula.

¹⁹ It might be emphasized that these conclusions are based on the use of significance tests. That is to say, when both sets of data produce significant correlations, regardless of the actual magnitude of the correlations, this fact is here interpreted to mean that the relationship postulated in the hypothesis exists in both sets of data. Hence it is concluded that the two sets of data are comparable with respect to the hypothesized relationships. Other tests, of course, could have been used. One that has been suggested is to determine the difference between correlations obtained in simulate and historical data. This test was also made and a discussion and presentation of the results can be found in the Appendix.

²⁰ The same considerations discussed in n.19 are also relevant here. See Appendix.

²¹ In correspondence with the author, Dr. Dean Pruitt has suggested that Hypotheses 4–8 are not independent of Hypotheses 1–3, and that Hypotheses 11–13 are implied by Hypotheses 1–3 and 10. He argues, for example, that Hypotheses 2 and 3 would seem to imply Hypothesis 5. If there is a significant difference between interbloc perceptions of threat and intrabloc perceptions of threat (Hypothesis 2), it implies that there are a number of pairs of nations that are ranked high on perceptions of threat (the interbloc pairs), and a number of pairs that are ranked low on perceptions of threat (the intrabloc pairs). This is similarly the case for Hypothesis 3. Furthermore, the same pairs will be high (interbloc) or low (intrabloc) for both Hypotheses 2 and 3, thus implying that Hypothesis 5, which postulates a relationship between expression and perception of hostility, would have to be supported. However, while it can be argued that Hypotheses 2 and 3 imply Hypothesis 5, the reverse is not true, i.e., Hypothesis 5 does not imply Hypotheses 2 and 3. Hence these hypotheses are not equivalent and therefore tests of each hypothesis are still in order. In other words, it is possible that Hypotheses 2 and 3 may not be supported by the data, while Hypothesis 5 is supported; perhaps there exists a different alliance system than that being used here.

However, it is possible, as Pruitt has suggested, to perform tests of the hypotheses that would render them independent. These analyses are discussed in the Appendix.

²² The same considerations discussed in n.19 are relevant here. See Appendix.

²³ The 1914 data, for example, represent only one historical case, and possibly an atypical case since it was a crisis that ended in war. Perhaps decision-makers behave differently in noncrisis periods or in crises resolved short of war; or perhaps decisionmakers operating in a totally different international epoch behave according to different rules. For these reasons, then, the results here are only suggestive and must await further investigation before they can be considered conclusive. The simulation data, though acquired from a series of “runs” (i.e., plays of the game), also require further replication.

²⁴ T. M. Newcomb, “Autistic Hostility and Social Reality,” Human Relations, I, 69–86.

²⁵ International Relations Program, Northwestern University. Inter-Nation Simulation Participant's Manual, 4.

²⁶ Crow, Edwin L.Davis, Frances A.Maxfield, Margaret W., Statistics Manual (New York 1960), 161.Google Scholar