HomePersonalDemocratic PeaceDemocide20th C. DemocideMegamurderersLesser MurderersWhy DemocideDimensionsConflictMethodsTheoryPolicyLinks



Volume 4

Expanded Contents | Figures | Tables

Chapters

1. Perspective And Summary
2. International Relations
3. The International Actors
4. International Behavior Space-Time
5. International Expectations And Dispositions
6. International Actor And Situation
7. International Sociocultural Space-Time
8. Interests, Capabilities, And Wills
9. The Social Field Of International Relations
10. Latent International Conflict
11. International Conflict: Trigger, Will, And Preparations
12. The Balancing Of Power
13. Comparative Dynamics Of International Conflict
15. Empirical Dynamics Of International Conflict
16. Causes And Conditions Of International Conflict And War
17. Ending Conflict And War: The Balance Of Powers
18. The International Conflict Helix
19. Theoretical And Empirical Conclusions On Conflict And War 20. Principles Of Peace And Conflict

Appendices

15A. Phasing Propositions and Their Evidence on International Conflict
16A. On Causes of International Conflict
16B. Propositions and Their Evidence on the Causes and Conditions of International Conflict Behavior
16C. Evidence on the Causes and Conditions of International Conflict Behavior
17A. Propositions and Evidence on the Causes and Conditions of Ending International Conflict Behavior
18A. Descriptive Propositions on International Conflict
19A. Overall Evidence on 54 Social Field Propositions on International Conflict
19B. Primary Propositions on Social Conflict
II. Event Data: Bases of Empirical Conflict Analysis
III. Characteristics of Published Quantitative International Relations Studies


Other Volumes

Vol. 1: The Dynamic Psychological Field
Vol. 2: The Conflict Helix
Vol. 3: Conflict In Perspective
Vol. 5: The Just Peace


Other Related Work

Conflict And Violence page

Democratic Peace page



UNDERSTANDING CONFLICT
AND WAR: VOL. 4:
WAR, POWER, PEACE

Appendix I

Unpublished Research And Results
On International Relations And Conflict*


By R.J. Rummel



Over the years 1965 to 1977, my assistants and I completed a number of relevant research projects of which results are unpublished in any form, including research reports. These projects have been consolidated in the following list under general headings for easy reference. A few introductory definitions should be helpful.

References in tables (such as Table 16C.3.3) to these projects are made in terms of this Appendix, general listing, and specific list number. Thus, for example, (Appendix I: 2.3) refers to this Appendix, Heading I.2 titled "Dimensions of International Behavior," and the third listed Project No. 5 titled "Dyadic Foreign Conflict and Cooperation Dimensions." The specific list number and the project number (Project No. 5 in this case) will generally differ because projects have been reordered here to place similar analyses under the same general heading.


I.1. DIMENSIONS OF FOREIGN CONFLICT BEHAVIOR

1.1. (Project No. 2) "Dyadic Foreign Conflict Dimensions."

Samples: all directed dyads with reported conflict behavior = 289. Data Year: 1960.

Data Sources: New York Times Index using Rummel (1966) code sheet.

Variables: 21 kinds of foreign conflict behavior.

Analysis: (A) a component analysis of the 21 variables and orthogonal rotation of all factors with eigenvalues greater than 1.00; (B) a reanalysis of a reduced set of 15 variables.

Results: dimensions delineated in both the 21 and 15 variable analyses are (1) negative communications and negative sanctions, (2) military violence, (3) warning and defensive acts, and (4) boycotts and embargoes and aid to object's enemy.

Assistant: George Omen.


1.2. (Project No. 4) "Dyadic Foreign Conflict Dimensions"

Samples and Data Years: all directed dyads with reported foreign conflict behavior for 1950 (286 dyads, 19 variables), 1955 (341 dyads, 16 variables), 1960 (290 dyads, 21 variables), 1963 (276 dyads, 24 variables), 1965 (305 dyads, 24 variables).

Data Sources: daily The New York Times, New York Times Index, using Rummel (1966) code sheet.

Variables: as indicated (the variation in the number of variables was due to some being omitted because of an insufficient number of events).

Analysis: (A) a separate component analysis and orthogonal rotation of factors with eigenvalues over 1.00 for each year; (B) a super-P component analysis of all years together--same rotation.

Results: dimensions delineated for all years were (1) negative communications, (2) military violence; in addition, consistently appearing for a number of years were (3) incidence of violence, (4) negative sanctions, (5) warning and defensive acts, (6) antiforeign demonstrations, and (7) aid to rebels.

Assistant: Sang-Woo Rhee.


1.3. (Project No. 3) "Dyadic Foreign Conflict Dimensions."

Samples and Data Years: the samples for the years indicated in 1.2 (Project No. 4) were augmented by additional collection and corrected where errors were found and, in addition, an hypothetical peace dyad without any conflict was added to the sample for each year to give in total: 1950 (285 dyads, 16 variables), 1955 (336 dyads, 15 variables), 1960 (286 dyads, 15 variables), 1963 (275 dyads, 15 variables), 1965 (366 dyads, 16 variables).

Data Sources: same as in 1.2 (Project No. 4), above, augmented by Keesings Contemporary Archives, Facts on File, and other sources as necessary.

Variables: as indicated above.

Analysis: (A) a separate component analysis and orthogonal rotation of seven factors (the seventh had an eigenvalue around 1.00) for each year; (B) a super-P image analysis, with rotation same as above (results published in Omen, 1975).

Results: dimensions delineated in all analyses are: (1) negative communications, (2) military violence; in addition, consistently appearing in a number of analyses were (3) antiforeign demonstrations, (4) aid to rebels, (5) expelling or recalling diplomats, (6) boycotts and embargoes, and (7) war.

Assistant: George Omen.


1.4. (Project No. 27) "Dimensions of Dyadic Foreign Conflict"

Sample: 47 symmetrical dyads with conflict.

Data Years: 1955-1957.

Data Sources: diverse (see Rummel, 1963).

Variables: 12 foreign conflict variables measured as presence or absence.

Analysis: component analysis and orthogonal rotation of eigenvalues over 1.00.

Results: dimensions of mobilization and military violence, negative communications, troop movements and antiforeign demonstrations, and negative sanctions accounted for about 73% of the variance.

Assistant: Mary Hirschoff.


1.5. (No Separate Project No.) "Dyadic Foreign Conflict Dimensions."

Sample: all directed dyads with foreign conflict behavior.

Data Year: 1955.

Data Sources: Same as Project No. 4, above.

Variables: 16 (same as Project No. 4, above, for 1955).

Analysis: (A) common factor analysis, image factor analysis, canonical factor analysis, and alpha factor analysis--all with the same rotations and eigenvalue-one criteria for the number of factors; (B) a comparison of the above results to each other and to component analysis.

Results: (A) common and image analyses delineate military violence and negative communications as independent factors; (B) alpha, canonical, and component analysis essentially agree on military violence, negative communications, negative sanctions, antiforeign demonstrations, and diplomatic protests as independent dimensions.

Assistant: unrecorded.


1.6. (Project No. 15) "Assessing Political Risk."

Sample: All states with significant political conflict = 82).

Data Periods: June 1, 1976-August 31, 1977.

Data Sources: diverse public media.

Variables: 21 foreign conflict scales.

Analysis: super-P (time series) component analysis with orthogonal rotation of factors with eigenvalues over 1.0.

Results: (1) dimensions of negative communications and action, foreign intervention, warning and defensive acts, antiforeign demonstrations, clashes and discrete military action, and killed; (2) the first unrotated dimension accounts for 21 percent of the variance and is general to all the variables.

Assistant: Gary Murfin.


I.2 DIMENSIONS OF INTERNATIONAL BEHAVIOR

2.1. (Project No. 22) "Dimensions of International Behavior."

Sample: 348 directed dyads, 182 of which were nonrandomly selected.

Data Year: 1955.

Variables: 40 diverse types of international behavior.

Analysis: (A) component analysis of the raw data and orthogonal and oblique rotation of factors with eigenvalues over 1.00; (B) higher order image analysis.

Results: (A) 10 dimensions delineated, which are general transactions, alliances (UN voting and bloc membership), relative exports, embassy and legations, mail and emigration, students, relative mail, relative international organizations, negative sanctions, and military violence and negative communications: (B) conflict and cooperative behaviors are independent; (C) all higher order dimensions are weak, indicating a near complete independence among all behavioral dimensions

Assistant: M. Hart.


2.2. (Project No. 23) "Dimensions of International Behavior."

Sample: 182 selected, directed dyads.

Data Years: 1950, 1955, 1960, 1963, 1965.

Variables: 50 diverse types of international behavior.

Analysis: separate image and component analyses of the raw data for each year and orthogonal rotation of eigenvalues around 1.00.

Results: (A) consistent dimensions for each year involve exports/GNP, students, emigrants, UN voting, relative international organizations, embassy or legation, aid, tourists, military violence, negative actions and negative communications; (B) cooperative and conflict behaviors are consistently independent.

Assistant: Yong-Ok Park.


2.3. (Project No. 5) "Dyadic Foreign Conflict and Cooperation Dimensions."

Sample: 182 selected, directed dyads.

Data Year: 1967.

Data Source: daily The New York Times, New York Times Index.

Variables: (A) 23 foreign conflict variables aggregated from Rummel (1966) code sheets; (B) 22 WEIS (Table II.2 of Appendix II) event variables, which involved 10 cooperation and 12 conflict ones.

Analysis: (A) separate component analyses and orthogonal rotation of different numbers of factors of 23 foreign conflict variables, 22 WEIS event data variables, 12 WEIS conflict variables, and 10 cooperation variables; (B) canonical analysis of the similarity in factors from the 23 foreign conflict variables on the one hand and the WEIS factors on the other; (C) canonical analysis of the similarity of events collected for the 23 variables on the one hand and for the WEIS variables on the other (Phillips, 1972).

Results: (A) there is about a 60% overlap between conflict events collected for 23 Rummel variables and 12 WEIS conflict variables and similar dimensions can be delineated in each; (B) there are two dimensions of cooperation (positive communication, bargaining); (C) there is little relationship between the cooperation and conflict events; (D) the dimensions of the 23 Rummel foreign conflict variables are negative communications and sanctions, military action plus warning and defensive actions, aid to rebels, antiforeign demonstrations, and expel or recall diplomats plus sever diplomatic relations; (E) the dimensions of the 12 WEIS foreign conflict variables are negative communications and sanctions, military action, threat communication.

Assistants: Warren Phillips and Mike Zavatsky.


2.4. (Project No. 24) "International Behavioral Groups."

Sample: 182 selected, directed dyads.

Data Year: 1955.

Data Sources: the component analysis of international behavior results published in Rummel (1969a).

Variables: standardized distances between dyads on (A) unweighted 10 dimensions of international behavior for raw data, and (B) unweighted 11 dimensions of data transformed towards normality.

Analysis: separate direct component analyses of distance matrices on raw data and on transformed data dimensions, and orthogonal rotation in each case.

Results: (A) for each analysis, seven distinct clusters of dyads were defined; (B) in each analysis, one cluster's profile involved high negative communications, great UN voting distance, low tourists, low relative emigrants, low relative students, low exports/GNP, but medium to high relative international organization comemberships, and low to medium relative (to total for actor) embassies and legations.

Assistant: Betty Bockelman.


2.5. (Project No. 17) "Peace Systems of International Behavior."

Samples: (A) 139 selected, directed dyads with no conflict behavior; (B) 151 randomly determined, directed dyads with no conflict behavior.

Data Year: 1955.

Variables: 36 diverse types of nonconflict, international behaviors.

Analysis: two separate component analyses of each sample (and orthogonal and oblique rotations) of (A) raw data, (B) data transformed towards normality.

Results: most consistently delineated across the four analyses were the dimensions of treaties, bloc alliances, UN voting, relative exports, students, diplomatic (embassies and legations), emigrants, and international organizations.

Assistant: M. Hart.


2.6. (Project No. 18) "Conflict Systems of International Behavior."

Sample: 43 selected, directed dyads with conflict behavior.

Data Year: 1955.

Analysis: two separate component analyses (and orthogonal and oblique rotation) of (A) raw data, (B) data transformed toward normality.

Results: (A) conflict behavior is an independent dimension; (B) international organizations, embassies and legations, alliances (UN voting and bloc membership), are independent dimensions; (C) tourists, mail, and relative exports are independent dimensions.

Assistant: M. Hart.


2.7. (Project No. 19) "Peace versus Conflict Systems of International Behavior."

Sample: 182 selected, directed dyads.

Data Year: 1955. Data Source: results of 2.5 and 2.6 (Project No.17 and Project No.18) listed above.

Variables: component dimensions for (A) 36 raw data variables, (B) 36 transformed data variables.

Analysis: (A) raw data, peace system dimensions were transformed into the space of the conflict system (using the factor comparison method of Ahmnaavara (Rummel, 1970: Section 20.2.3) in order to compare and contrast the two systems; (B) same analysis for transformed data dimensions.

Results: (A) international organizations, embassies and legations, treaties, alliances, and UN voting distances form patterns that change little between the two systems; (B) in the conflict system, exports more directly link into treaties, mail, and tourists, relative mail and emigration form a dimension not existing in the peace system, and tourists and translations also form a dimension not found in the peace system; (C) the behaviors most changing their patterns from peace to conflict system are (in decreasing order of change) foreign mail/domestic mail, relative translations, emigration, relative tourists, and exports.

Assistant: M. Hart.


I.3 DIMENSIONS OF STATES

3.1. (Project No. 37) "Consistent Dimensions of States."

Samples and Data Years: 72 states (1950), 82 states (1955), 87 states (1960), 107 states (1963), 113 states (1965).

Analysis: for each year, component analysis and separate orthogonal rotation of dimensions with (A) eigenvalues over 1.00, and accounting for about 90% of total variance.

Results: (A) consistent dimensions across all five years are wealth, (liberal democracy versus) totalitarianism, power bases, diversity, authoritarianism, and density, (B) the largest dimensions are wealth, power, and politics (totalitarianism) accounting for about 29 percent of total variance.

Assistant: Sang-Woo Rhee.


3.2. (Project No. 42) "Principal Space-Time Components of States."

Sample and Data Year: same as 3.1 (Project No. 37), above.

Variables: 90 diverse attributes of states listed in Rummel (NAB) plus a time variable.

Analysis: systematic comparison of the unrotated results of (A) component analyses of raw and transformed data (rotated results reported in Rummel, 1979), (B) image analysis of transformed data (rotated results reported in Rummel, 1979).

Results: wealth, power, and politics are consistently the largest space-time (unrotated) principal axes (components).

Assistant: none.


3.3. (Project No. 38) "Stability of Dimensions of States: Cross-time."

Sample: 82 states.

Data Years: 1955, 1963.

Variables: 79 variables common to both the 1955 dimensions published in Rummel (1972) and the 1963 component analysis of 3.1 (Project No. 37), above.

Analysis: (A) component analysis or raw data for each year and orthogonal rotation; (B) factor comparison using Ahmnaavaara's method (Rummel, 1970: Section 20.2.3); (C) canonical analysis to determine predictability of 1963 dimensions from 1955.

Results: the same largest dimensions of wealth, power bases, and totalitarianism exist for both years, but over all the dimensions there was a considerable shift in correlation and some dimensions were unique to each year. The 1955 dimensions (loadings) predict about 47% of the variation in the 1963 dimensions.

Assistants: Paul McCarthy and Dennis Hall.


3.4. (Project No. 39) "Stability of Dimensions of States: Cross-models."

Sample: 113 nations. Data Year: 1965.

Variables: 90 diverse attributes used in 3.1 (Project No. 37), above.

Analysis: (A) image analysis and orthogonal rotation; (B) comparison to results of component analysis, described in 3.1 (Project No. 37) above.

Results: major and important minor dimensions are similar for each model.

Assistant: Sang-Woo Rhee.


3.5. (Project No. 40) "Stability of Dimensions of States: Cross-missing Data Estimation Techniques."

Sample, Data Year and Variables: same as in 3.1 (Project No. 37), above.

Analysis: parallel component analysis for each year and super-P component analysis on data with missing data (A) estimated through least squares; B) estimated through a randomized procedure ("pseudoestimates")--see Wall and Rummel (1969).

Results: essentially similar across missing data estimation techniques.

Assistants: Sang-Woo Rhee and George Omen.


3.6. (Project No. 41) "Higher Order Space-time Dimensions."

Sample, Data Year, and

Variables: same as 3.1 (Project No. 37), above.

Analysis: a higher order component analysis (and orthogonal rotation) of the super-P, image analysis of transformed data published in Rummel (1979: Chapter 5).

Results: Data too complex for a good higher order (oblique results do not achieve simple structure), implying that the unrotated results should be emphasized.

Assistant: George Omen.


3.7. (Project No. 46) "Taxonomy of States."

Sample and Data Year: same as 3.1 (Project No. 37), above.

Variables: 87 standardized, diverse attributes (listed in Rummel, 1979).

Analysis: (A) separate Q-component analysis and orthogonal rotation of (1) full data matrix and (2) transformed data matrix; (B) Q-component analysis and orthogonal rotation of 10 transformed indicators of the dimensions of states listed in Rummel (1979).

Results: 3 major empirical types of state-societies exist: coercive, authoritative, and exchange (Chapter 34 of Vol. 2: The Conflict Helix).

Assistant: Yong-Ok Park.


I.4 TESTING DIFFERENT PREDICTION
MODELS OF INTERNATIONAL CONFLICT

4.1. (Project No. 7) "Model I: Distance Vectors versus Squared Magnitudes."

Sample: 182 selected, directed dyads. Data Year: 1955.

Variables: (A) factor scores on four conflict dimensions of dyadic international behavior from Rummel (APSR); (B) basic indicators of 13 dimensions of 236 state-attributes from Rummel (1972).

Analysis: (A) regression analysis of the dependence of conflict dimensions on distances (j-i differences) and, separately squared distances; (B) canonical analysis of the dependence of conflict dimensions on distances (j-i differences) and, separately, squared distances.

Results: (A) differences account for almost no variation in dyadic conflict behavior (canonical traced squared = .03); (B) squared differences explain about 28% of the variation in conflict behavior; (C) disparity in power, similarity in development, and relative freedom most significantly explain dyadic conflict

Assistant: Dave McCormick.


4.2. (Project No. 8) "Model II and the Conflict Helix: Distance Magnitudes and Time."

Samples: 14 selected samples of directed dyads, each with the same actor and 13 different object states. Data Year: 1955.

Variables: (A) factor scores on the five dimensions of 16 foreign conflict variables for 340 dyads reported in Rummel (1967a, 1968); (B) eight theoretical, absolute distances variables, including time since last formation of the structure of expectations for a dyad.

Analysis: regression analysis.

Results: (A) average of 35% of the variation in dyadic conflict explained by the eight variables; (B) parity rather than disparity in power best explains conflict and is the most significant of all independent variables--next in importance is political dissimilarity.

Assistant: Tong-Whan Park.


4.3. (Project No. 9) "Model I and the Conflict Helix: Distance Magnitudes, Power, Time, and Time Interaction."

Sample: 182 selected, directed dyads. Data Year: 1955.

Variables: (A) factor scores on the dimensions of 16 foreign conflict variables for 340 dyads reported in Rummel (1967a, 1968); (B) 19 theoretical variables, including power parity and time since the formation of the last structure of expectations for a dyad, absolute distances, distances x time, and joint power.

Analysis: (A) common factor analysis of the 22 conflict and theoretical variables and orthogonal rotation of factors with eigenvalues over 1.00; (B) regression analysis with each conflict dimension as a dependent variable and the theoretical variables as independent.

Results: (A) conflict forms a dimension clearly independent of the theoretical variables; (B) about one-fourth of the conflict is explained by the independent variables.

Assistant: Tong-Whan Park.


4.4. (Project No. 10) "Models I and II and the Conflict Helix: Distance Magnitudes and Time."

Sample: 182 selected, directed dyads. Data Year: 1955.

Variables: same as 4.2 (Project No. 8), above.

Analysis: canonical analysis of (A) all dyads; (B) all dyads with Israel as actor; (C) all dyads not involving Israel as actor.

Results: (A) dyadic conflict has a very low fit to distance magnitudes and parity (trace correlation squared across actors = .02); (B) best fitting canonical variates for all actors have a correlation of .48, with military violence and negative communication most dependent on language and political dissimilarities; (C) Israel's conflict makes no difference in results overall, but analysis of the Israel dyads alone show Israel's conflict well predicted (trace correlation = .67), especially by language dissimilarity and time since the last structure of expectations was formed with her Arab antagonists.

Assistant: Tong-Whan Park.


4.5. (Project No. 11) "Model I and II and the Conflict Helix: Distance Magnitudes and Time."

Sample, Data Year, and Variables: same as 4.4 (Project No.10), above.

Analysis: (A) regression analysis with each conflict factor as dependent variable and distance magnitudes and time as independent; (B) a profile analysis of largest and smallest regression residuals; (C) a rerun of the regression for all dyads having Israel as actor; (D) a rerun of the regression for all dyads not having Israel as actor.

Results: (A) about 12% of the variance in conflict is accounted for by distance magnitudes and time; (B) power parity and political distance are the most consistent predictors of conflict; (C) power parity, Time, and economic distance account for the best predictions (smallest residuals) of military violence: (E) power disparity, little time since the formation of the structure of expectations, and political distance best account for the poor predictions (largest residuals) of military violence.

Assistant: Tong-Whan Park.


4.6. (Project No. 12) "Model I and the Conflict Helix: Curvilinear Test of Distance magnitudes and Time."

Sample, Data Year: same as 4.4 (Project No. 10), above.

Variables: (A) same as 4.4 (Project No. 10) above; (B) all independent variables transformed to natural logarithms.

Analysis: (A) regression analysis, with each conflict factor as dependent and log distance magnitudes and log time as independent; (B) canonical analysis of conflict factors against log distances magnitudes and log time.

Results: (A) there is little curvilinear relationship of conflict to distance magnitudes across actors; (B) the best relationship is that involving political distance and racial-language similarity which account for about 23% of the variation in military violence and negative communications; (C) power parity significantly predicts separately to military violence and negative communication.

Assistant: Tong-Whan Park.


4.7. (Project No. 13) "Models I and II and the Conflict Helix: Interaction of Distance Magnitudes and Time."

Sample, Data Year: same as 4.4 (Project No. 10), above.

Variables: (A) dependent same as 4.2 (Project No. 8), above; (B) eight theoretical absolute distance variables multiplied by the time since the last structure of expectations was formed, in addition to time by itself.

Results: (A) overall about 8% of conflict behavior is explained; (B) parity x time is a significant predictor of conflict behavior, especially military action.

Assistant: Tong-Whan Park..


4.8. (Project No. 14) "Model I and the Conflict Helix: Distance Magnitudes and Time."

Sample: 166 randomly selected, directed dyads. Data Year: 1955.

Variables: same as 4.2 (Project No. 8), above.

Analysis: regression of the different conflict factors onto distance magnitudes and the time since the last formation of the structure of expectations for a dyad.

Results: (A) little variance in conflict predicted (average R2 = .05); (B) power parity is the most significant predictor Q of the 4 significant regression coefficients).

Assistant: Tong-Whan Park.


4.9. (Project No. 28) "Model I: Theoretical Variables."

Sample: 91 selected, symmetrical dyads.

Data Years: 19S5-57.

Variables: (A) dependent are the factor scores on four rotated and first unrotated dimensions of dyadic foreign conflict delineated in 1.4 (Project No. 27), above; (B) independent are UN voting similarity, economic similarity, power parity, overall similarity, and joint power.

Analysis: regression analysis.

Results: (A) similarity in UN voting and joint power are the best predictors (especially for negative communication); (B) power parity is a poor predictor; (C) overall, about one-third of the variation in negative communications, and warning and defensive acts is dependent on the five predictors.

Assistant: Betty Bockelman.


4.10. (Project No. 29) "Models of the Conflict Helix."

Sample, Data Year: same as 4.9 (Project No. 28), above.

Variables: (A) dependent same as 4.9 (Project No. 28), above; (B) independent are models of the (logged and unlogged) latent conditions underlying the conflict situation and the conditions theoretically correlated with conflict behavior.

Analysis: bivariate regression analysis.

Results: log of conflict behavior most related to logged prediction model of conflict behavior (accounts for 31% of the variance), and the model of the conflict situation moderately predicts (27% of the variance) the log of conflict behavior.

Assistant: Betty Bockelman.


I.5 TESTING DIFFERENT PREDICTION
MODELS OF INTERNATIONAL BEHAVIOR

5.1. (Project No. 16) "Model I: Squared Distance Magnitudes."

Sample: 182 selected, directed dyads. Data Year: 1955.

Variables: (A) factor scores on 12 orthogonally rotated factors of a component analysis of dyadic behavior of 182 dyads on 44 variables (see Rummel, 1969a); (B) distance magnitudes on 14 indicators of the factors of a component analysis of 236 variables for 82 nations (see Rummel, 1972) plus two geographic distance variables.

Analysis: (A) curvilinear regression analysis of each dimension of dyadic behavior onto the squared 14 distance variables, plus geographic distance; (B) curvilinear regression analysis of each dimension of dyadic behavior onto the log of wealth and power distances, squared distances for the remaining dimensions, plus geographic distance.

Results: (A) about 24% of the variation in international behavior (20% in conflict behavior) is explained by distance magnitudes; (B) power parity, political similarity, and geographic distance significantly predict to negative sanctions; (C) power disparity and political dissimilarity significantly predict to military violence plus negative communications; (D) political disparity, rich-poor gap, and geographic distance significantly predict to UN voting distance; (E) wealth, power, and political distances are most significant for predicting international behavior.

Assistant: none.


5.2. (Project No. 30) "Model I: Attributes, Distance Magnitudes, and Joint Power."

Sample: (A) 182 selected, directed dyads; (B) 166 randomly determined, directed dyads; Data Year: 1955.

Variables: (A) 21 to 30 types of international behavior as dependent variables; (B) 23 independent variables, including actor's attributes, absolute distances, and joint power.

Analysis: canonical analysis.

Results: (A) attributes and distances account for about 26% of the variation in behavior; (B) an actor's wealth and power, plus the joint power and overall dissimilarity between actor and object, predict to about 90% of the variation in a pattern of mail, transactions, tourists, migrants, and negative communications.

Assistant: Paul McCarthy.


5.3. (Project No. 25) "Model I: Distance Vectors and Squared Magnitudes."

Sample: 182 selected, directed dyads. Data Year: 1955.

Variables: (A) dependent variables are factor scores on 12 dimensions of international behavior delineated through component analysis; (B) independent variables are indicators of 15 dimensions of cross-national attributes delineated through component analysis (see Rummel, 1977, Chapter 4 for a discussion of these independent and dependent variables); (C) geographic distance and power parity.

Analysis: (A) separate regression analyses onto distance vectors and squared distance magnitudes; (B) separate canonical analyses involving distance vectors and squared distances, plus geographic distance and power parity.

Results: (A) differences (distance vectors) in general account for little (13%) of international behavior across dyadic actors, the best being difference in wealth and power (see Rummel, 1977: 90-91, and Gleditsch, 1970, for some of these results); (B) squared distances account for about 30% of the variation in behavior, the best being squared distances in power and political freedom.

Assistant: Dave McCormick.


5.4. (Project No. 26) "Model I: Theoretical Predictors."

Sample: 42-105 randomly selected, directed dyads with no missing data. Data Year: 1955.

Variables: (A) dependent are factors scores or indicator variables of dimensions of foreign conflict and international behavior from Rummel (1969a), merged into scales of total cooperation, total behavior (cooperation plus conflict), and net cooperation (cooperation minus conflict); (B) independent are distance magnitudes plus geographic distance and joint power.

Analysis: regression analysis.

Results: (A) cooperation (or cooperation plus or minus conflict) highly dependent (86% of the variance); (B) best predictors are joint power (71 %), power parity (4%), political similarity (3%), and racial similarity (3%).

Assistant: Betty Bockelman.


5.5. (Project No. 31) "Model I: Testing Theoretical Linkages Between International Behavior and Attributes."

Samples: (A) 182 selected, directed dyads' (B) 166 randomly determined, directed dyads. Data Year: 1955.

Variables: separate raw and transformed 80 variable data sets of (A) 44 types of diverse international behaviors; (B) 36 characteristics of a dyad, involving (1) attributes of the actor, (2) absolute distances on the indicators of the dimensions of 236 national attributes (see Rummel, 1972), (3) rank distance, (4) distance on Russett's socioeconomic dimensions (1967), (5) joint power of actor and object, (6) measures of geographic distance, (7) racial, language, and religious similarities.

Analysis: target rotation: factors linking 44 behavior and 36 attributes rotated to a best fit to theoretical linkages involving wealth, joint power, geographic distances, and cultural similarity.

Results: Negative--actors are not similarly influenced by the indicated variables in the theoretical direction.

Assistants: L. Levine and M. Hart.


I.6 THE LINKAGE OF INTERNATIONAL BEHAVIOR
TO NATIONAL ATTRIBUTES AND DISTANCES

6.1. (Project No. 1) "Model I: Attributes and Absolute Distances."

Samples: (A) selected sample of 182, directed dyads; (B) randomly determined sample of 166, directed dyads. Data Year: 1955.

Variables: (A) for international behavior: factor scores of 12 dimensions of behavior for the selected sample (Rummel, 1969a), and II dimensions of behavior for the random sample; (B) for attributes and absolute distances: 26 variables measuring an actors attributes, absolute distances between actor and object, cultural similarity, and geographic distance.

Analysis: (A) a component analysis of 26 measures of attributes and absolute distances separately for each sample; (B) a canonical analysis of the dependence of behavioral factor scores on the factor scores for the attribute and absolute distance dimensions, done separately for each sample.

Results: (A) in general, national attributes and absolute differences across dyads explain about 18% of the variation in dyadic international behavior; (B) rank distance (absolute differences in power and wealth) and political distance are the best predictors of behavior across actors, accounting for about 50% of some components of behavior.

Assistant: Paul McCarthy.


6.2. (Project No. 21) "Model L Components of National Attributes and Behavior."

Sample: 348 directed dyads, 182 of which were nonrandomly selected.

DataYear, and Variables: same as 5.5 (Project No. 31), above.

Analysis: component analysis and orthogonal rotation of all dimensions with eigenvalues over 1.00.

Results: no strong linkages appear: an actor's wealth and power have a small link to his volume of transactions; power distance to diplomatic importance; political similarity to UN voting agreement and alliances; contiguity to military violence.

Assistant: M. Hart.


6.3. (Project No. 20) "Model I: Common Factors of National Attributes and Behavior."

Sample, Data Year, Data Source, and Variables: same as 5.5 (Project No. 31), above.

Analysis: (A) two separate common factor analyses (orthogonal and oblique rotations) for each sample of (1) raw data, (2) data transformed towards normality; (B) a factor comparison of all four sets of common factors, using the method of Ahmnaavara (Rummel, 1970: Section 20.2.3).

Results: (partially described in Rummel, 1972: 407-410): (A) the only strong linkage appearing is of UN voting similarity and alliances to socioeconomic similarity: (B) consistent weak relationships across the four analyses involve (1) transactions linked to wealth, power, political similarity, and geographic closeness, (2) conflict linked to the actor's power, rank distance, total distance across all attributes, and joint power of actor and object, and (3) UN voting distance linked to Catholic culture distance, and political distance.

Assistants: L. Levine and M. Hart.


6.4. (Project No. 32) "Model I: Common Components of National Attributes and Behavior."

Samples, Data Year, and Variables: same as 5.5 (Project No. 31), above.

Analysis: component analysis of the raw data of each sample and orthogonal rotation of eigenvalues over 1.00.

Results: (A) similar to those for 6.3 (Project No. 20), above; (B) unrotated components show (1) general transactions have significant dependence on power, wealth, and geographic distance; (2) independently, general conflict has a significant dependence on power, wealth, and political distance, (3) two secondary transaction components are dependent on geographic distances. These five components account for about 50% of the variation in the 70 dyadic behavioral and attribute variables.

Assistant: L. Levine.


6.5. (Project No. 33) "Model I: Most General (Higher Order) Factors of National Attributes and Behavior."

Sample, and Data Year: same as 5.5 (Project No. 31), above.

Variables: biquartimin factors and components from the common and component factor analyses of 6.2 and 6.3 (Project No. 21 and Project No. 20), above.

Analysis: second and third-order image factor analyses.

Results: similar to those for unrotated components listed in 6.4 (Project No. 32), above.

Assistant: L. Levine.


6.6. (Project No. 34) "Peace Versus Conflict Behavioral System Linkages to Attributes and Distances."

Sample: (A) 139 selected, directed dyads with no conflict behavior; (B) 43 selected directed dyads with conflict behavior. Data Year: 1955.

Variables: same as 5.5 (Project No. 31), above.

Analysis: (A) separate common factor analysis and orthogonal and oblique rotation of raw and transformed data for each sample; (B) peace system linkages compared to conflict linkages by (1) transforming peace results into conflict space (Ahmnaavara's factor comparison technique), and (2) separately transforming into the general space of the results listed in 6.3 (Project No. 20), above.

Results: (A) the attributes most changing in their linkages to dyadic behavior between conflict and peace systems are racial, religious, and language similarity; (B) shift in patterns between peace and conflict systems along which conflict occurs involve transactions, political distance, and geographic distances. Power, wealth, overall dissimilarity, or joint power linkages do not distinguish peace and conflict systems.

Assistant: L. Levine.


6.7. (Project No. 43) "Models I and II: Dependence of Space-Time, International Behavior on Distance Magnitudes."

Sample: 186 selected, directed dyads.

Data Years: 1950, 1955, 1960, 1963, 1965.

Variables: (A) absolute distances on the factor scores for 11 dimensions of national attribute space-time and dyadic factor scores on eight dimensions of behavioral space-time (from Rummel, 1979); (B) indicators for above dimensions used in place of factor scores for attributes and behavior.

Analysis: (A) separate canonical analyses of A and B sets of variables for all dyads for each year (Model I); (B) separate canonical analyses of A and B sets of variables across all years for each subsample of 13 dyads with the same actor (Model II).

Results: (A) for all dyads (Model I), behavior is most dependent on absolute political distances, especially bloc alignment and UN voting agreement; (B) absolute economic and power distances (power parity) have little relationship to behavior; (C) for actor specific analyses, absolute political distance still most important determinant of behavior, especially of trade, aid, UN voting, bloc alignment, and international organization comemberships; (D) distance magnitudes predict behavior almost as well as differences (distance vectors) for individual actors (Model II), and are much better predictors over all dyads (Model I).

Assistant: Yong-Ok Park.


6.8. (Project No. 36) "Model II: U.S. Foreign Relations and Absolute Distances."

Sample: 81 object nations of U.S. behavior. Data Year: 1955.

Variables: (A) absolute distances between United States and object nations on 14 indicators of dimensions of national attributes from Rummel (1972); (B) factor scores from Rummel (1972a) of 81 dyads with United States as actor on six dimensions of U.S. dyadic behavior.

Analysis: (A) canonical analysis of dependency of U.S. dyadic behavior on U.S.-object, absolute distances; (B) rerun of canonical analysis with distances on indicators of wealth and power log transformed; (C) rerun of canonical analysis with all distances log transformed.

Results: (A) overall, absolute distances accounted for 40% of U.S. behavior; (B) power parity and, secondarily, similarity in wealth are tightly linked to Western European cooperation or, secondarily, negative communications and military violence; (B) similarity in political system, dissimilarity in religion, and geographic closeness are linked to diplomatic cooperation.

Assistant: Richard Van Atta.


I.7 MISCELLANEOUS

7.1. (Project No. 6) "Correlates of Escalation."

Sample: major opponents in 39 historical conflicts.

Data Years: World War I to 196 5. Data Source: Wright (1965).

Variables: (A) one variable measuring four levels of escalation; (B) twenty variables involving four from Wright's escalation model, five descriptive of the conflicts, and 11 measuring absolute distances between antagonists.

Analysis: (A) component and image analysis and orthogonal and oblique rotation of factors with eigenvalues greater than 1.00, (B) higher order factor analyses and orthogonal rotation--same factor cutoff.

Results: (A) escalation is independent of absolute political and cultural distances across actors and the absolute differences in size; (B) escalation depends on the perception of the parties of their national interest that is involved and (to a lesser extent), their comparison of their relative vulnerability, power parity, and absolute difference in wealth; (C) escalation is more likely in recent years.

Assistant: L. Levine.


7.2 (Project No. 44) "Testing the Empirical Assumptions of Détente."

Sample: (A) 24 distinct canonical analyses of the dependence of dyadic behavior on attribute differences (distance vectors), some published in Rummel (1977, 1979), involving as actors China, USSR, United States; (B) 86 distinct canonical equations involving 14 national actors.

Data Years: 1950-1965.

Data Sources: diverse (see Rummel, 1979).

Variables: about 350 diverse, underlying behavioral and attribute variables.

Analysis: a systematic comparison of the canonical linkages found empirically against those linkages assumed by détente (as formulated by Henry Kissinger) for both samples A and B.

Results: (A) the assumption that power is related to conflict is correct, but not as power parity; rather, the greater the power of the object nation, the more conflict directed towards it; (B) the assumption that. transactions are inversely related to conflict is falsified (these results underlie the assertions of Rummel, 1976a).

Assistant: none.


7.3. (Project No. 45) "Synthesizing Empirical, International Behavior, Space-Time linkages."

Sample: 197 canonical equations for 14 selected nation-actors linking dyadic behavior and attribute differences (distance vectors).

Data Years: 1950-1965.

Variables: about 350 underlying behavioral and national attribute variables.

Analysis: a systematic tabulation and cross-tabulation of all the linkages regarding (A) coincidence of linkages (scope), (B) strength of coincidence (intensity).

Results: (A) for differences in power, politics and wealth, the first has the greatest influence on dyadic behavior, the third the least; (B) for dyadic contractual, familistic, and conflict behavior, the first is most dependent on attribute differences; the first and third are less and equally dependent; (C) the strongest linkages are between power and political differences and contractual behavior; the weakest is between wealth and political differences and conflict behavior; (D) considering the linkages by type of actor, the strongest linkages are between power difference and contractual behavior for poor, weak, or authoritarian actors; (E) the weakest linkages are for political differences and conflict behavior for the poor, weak, or totalitarian actors; and wealth difference and familistic behavior for totalitarian actors. See Appendix 9A for partial results.

Assistant: Lei Rummel.


7.4. (Project No. 47) "Test of the Conflict Helix for International Behavior."

Samples: results of 2.1, 2.5, 2.6, 6.2, 6.3, 6.4, and 6.6 (Projects No. 22, 17, 18, 21, 20, 32, and 34, respectively), above.

Data Years: 1955.

Variables: see projects listed under samples, above.

Analysis: a systematic comparison of unrotated dimensions for peace versus conflict systems to test (1) hypothesis that peace systems are more generally integrated by a dominant structure, while being much more pluralistic in behavior, (2) hypothesis that peace has a more general linkage to attributes while having overall more independent of attributes (the conflict system will have more diverse linkages and less unique behavior).

Results: hypotheses confirmed.

Assistant: none.


7.5 (Project No. 48) "Libertarian Systems and War."

Samples: 50 interstate wars.

Data Years: 1816-1965.

Data Sources: Singer and Small (1972).

Variables: political system (libertarian, authoritarian, totalitarian), number of wars, number killed.

Analysis: crosstabulation.

Results: There were no wars between libertarian (and ambiguously libertarian) systems, 14 wars between libertarian and authoritarian systems, and 36 between nonlibertarian systems.

Assistant: none.


NOTES

* Scanned from Appendix I in R.J. Rummel, War, Power, Peace, 1979. For full reference to the book and the list of its contents in hypertext, click book. Typographical errors have been corrected, clarifications added, and style updated.


For citations see the Vol. 4: War, Power, Peace REFERENCES




Go to top of document