Attributes, Behaviors, and Conflict:

A Statistical Analysis of the Use of Force

By

James Cerven

Matthew Hedberg

Joshua Lilly

Brooke Loesby

Jack E. Vincent

In this paper we will explore the relationship between national attributes and behaviors, and the international use of force. First, we will review the literature on this subject. Second, we will construct a predictive model about the use of force, drawing on the literature and based on our own assumptions. Third, we will perform various multivariate statistical analyses, relevant to our model, using national attributes (UN based) and behavioral (WEIS based) data obtained from the Martin Institute Archives (housed at UI). Fourth, will then interpret the results and discuss the possible policy implications.

Within the literature, there are several nation-state characteristics that are identified as causally related to the use of force. For example, the increase in a nation's military capabilities, either through arms imports or domestic arms developments, has been linked to the use of force (Sample 1998, Sample 1997, Kinsella 1998, Schampel 1993). It appears that nations will develop military capabilities when they expect conflict to occur in the future. In order to acquire these vast amounts of military resources, however, the state must first have the economic capacity to purchase them. For example, it has been found that states with large GNP's will be more likely to use force and/or engage in conflict (Vincent 197, De Soysa, O'Neal & Park 1997, Rummel 1994). We believe the reasons for these linkages are twofold. First, nations that have a large GNP/GDP will be interested in preserving their economic status from external threats and may use force to do so. Thus, such nations may be forcefully active in the international system in order to prevent damage to their economies, e.g. the Persian Gulf War. Second, those nations effectively arming for conflict will require a large GNP/GDP in order to maintain their social and economic infrastructure. In this connection, it has also been found that population size and population growth may have a significant impact on the likelihood of armed conflict (Tir &Diehl 1998, Rummel 1994, Gizewski & Homer-Dixon 1995). Also, nations with comparatively large populations may be more likely to have the economic resources and manpower resources necessary to develop significant military capabilities. It has also been found that democratic governments, when compared to non-democracies, are less likely to use force (Rummel 1983, Rummel, 1988). In addition "power" is one of the most often cited causes of conflict; that is, states with the most power are the most likely to become involved in armed conflict (Van Evera 1998, Lemke & Werner 1996, Kim and Morrow 1992, Houweling &Siccama 1991, Vincent 1996). Power is usually operationalized as a mix of military capabilities, economic wealth, and national resources. Although analysts debate about the specific relationship between power and force, many agree that the states that use the most force are likely to be the most powerful nations. Finally, we predict that the indicators of conflict and cooperation will tend to co-vary since states that frequently engaged in the use of force may also frequently engage negotiation etc., in an effort to reduce the level of violence as the US and North Vietnam did during the war in Vietnam. In this connection, Jerris (1993: 239) has argued "cooperation and conflict are so closely linked that we can hardly analyze one without paying attention to the other…"

. Based on the above review, with our own added assumptions about the relationship between national attributes and the use of force, we offer a "capabilities" interpretation regarding the use of force, i.e., those nations that maintain the greatest capabilities (large population, large GNP, and large military resources etc., will be the most likely to use force. Similarly, we expect that indicators of wealth (high scores on number of physicians, life expectancy, degree of urbanization, and low scores on fertility, death rate, and population growth) may also correlate with high force use. Contrary to the literature, however, we do not believe that regime type (whether a nation is democratic or non-democratic) will be as significantly related to the use of force. According to our theory, there is no necessary strong connection between regime type and the development of capabilities. As a matter of history, we know that non-democratic nations, such as the Soviet Union, China, Iran, and Iraq, have developed significant military capabilities along with democratic states, such as the US, France and Great Britain. This is not to refute the argument now well established that democracies tend not to fight other democracies (Rummel, 1988). Since we are making a capabilities argument, however, highly capable may be just as force prone as highly capable non-democratic states. In fact, it is hard to identify a non-democratic state in the cold war era, continuing to the present, that approaches the United States in the use of force. For example, between 1970 and 1989, WEIS records indicate that the US used 57 times as much force as the USSR.

Using this predictive model, we can expect a significant relationship (consistent with the model) between national attributes and national behavior and the degree of forceful conflict.

The test of our model will be facilitated by a factor analysis of the independent variables, given the large number involved. Factor analysis also guarantees that we will encounter no danger of linear dependence between the predictive indicators since the factor scores generated by the methods used (principle components, rotated by Kaiser's Varimax solution) are, by mathematical necessity, unrelated to each other. If the factors that heavily load variables indicating military capacity, economic development, population size, and international involvement are significantly correlated with the use of force our model will be supported empirically.

While we believe our methods to be very useful in determining force, we concede that there are several limitations to our research. First, because we chose to use a number of cross-time variables, a few countries such as Vietnam were not included due to the fact that it was not in existence throughout the whole period of analysis. However, the vast majority of the countries and most of the important force participatory states are included. Second, we recognize that variables included are probably not the only reason for states tend to use force in the international system. For example, the territorial position of a state (Vasquez, 1995) could be major factors in whether or not states will display forceful behavior. In addition, the study in no way addresses the possibility that the personality of specific actors may be strongly linked to the use of force, i.e., the personality of Adolph Hitler is assumed by many to have triggered the large scale use of force during WW II. However, such historic figures tend to be transient in the system, compared to national attributes. Therefore, although there are other theoretical variables (many of which may be virtually impossible to quantify, such as Hitler's personality type) that one can postulate as being linked to the use of force, we believe that correlating national attributes and measurable behaviors with force is an reasonable approach to evaluating whether a portion (and possibly a significant portion) of the force that is used in the system is, in fact, related to such variables.

The data used in this paper was taken from the Martin Institute Archives. The Martin Institute routinely merges initially non-compatible sets, such as World Bank Economic Indicators, WEIS Behavioral Indicators, etc. Our study used two types of analysis attribute and behavioral. Attributes include variables such as gross national product, armed forces per capita, and political and civil rights. This includes attribute differences computed between 1970 and 1989. The WEIS collection records newsworthy actions taken by one state towards another, such as a "protest." The Archives aggregates and defines the intensity and nature of these events in terms of the basic WEIS categories, which will be listed shortly. Since Martin Archive files contain such measurements from 1969 to 1989, we decided to use that time frame. We did so with the assumption that a longer time period would maximize the applicability of the model. In addition, we chose to look at every state located on the variables in the international system. We recognize that a few highly conflictual actors, such as North Vietnam, disappear from the world of nations over our time frame, but we accept this in the interest of maximizing the time-length of our project.

By using the Varimax method of factor rotation, our analysis generated four important dimensions linked to the use of force. Spearman Rho correlation's were used to avoid the problem of using Pearson's r on likely non-normal distributions. Spearman's Rho, of course, is a non-parametric test that does not assume normality. The following table shows the linkages between force and the factors that were significantly linked. F1 indicates that Factor 1 is associated with force at the level of .478 Rho. That is, those high of the force scale also tend to score high on F1 and those that score high on F1 tend to have many passenger cars etc. Those scoring low on force, of course tend to have the opposite pattern (few passenger cars etc.). Multiple R was computed by squaring the significant Rho's, summing and taking the square root. This is permissible since the factors are highly orthogonal, i.e., the rank variance explained by each factor is basically unique and therefore can be squared and summed.

Clearly the capability argument is strongly supported by this analysis. If you have got it, you tend to use it. In a sense, power begets force usage. Our prediction of a linkage between GNP/per capita and force, however, is not supported. GNP and population appear far more related to force that wealth as measured by GNP/per capita. In other words, it is certain kinds of capabilities rather than capabilities in general that appear to explain force usage. As predicted, cooperation and conflict appear highly linked and both are related to force usage. That is, states that exhibit a high degree of cooperative behavior may also tend engage in conflictual behavior and tend to use force. Conflict and cooperation appear to be part and parcel with one another. It is instructive to view these two seemingly disparate characteristics as possible part of a circle of behavior. It is possible that as cooperation between states breaks down, conflictual behavior rises to the surface. During the periods of high conflict, a number of cooperative behaviors may occur, as was the case in the Vietnam War. Therefore, while one may logically suspect that conflict is a harbinger of force, cooperation attributes that result within conflictual relations also help to predict force. Finally, as predicted, democracy appears irrelevant to the use of force. The democratic dimensions simply did not link to force usage.

The results of this study are noteworthy because of the twenty-year time -frame. By incorporating the time dimension, we are more fully able to understand not only the role that attributes play in prediction force, but also whether or not shifts in those attributes over time are meaningful in predicting force. The results in this study show that, as countries become more militaristic over time, they become more forceful. This stands in stark contrast to those who argue that the road to peace is through military power, a belief so deeply ingrained in the US political system, that Presidential candidates must support it with the same enthusiasm as they do motherhood, or face immediate rejection. We find that militarism shifts over time provide a prime indicator of use of force.

J. David Singer once challenged the academic community, Aacademe needs to do a better job of helping governments and their leaders get a handle on the causes of war (Singer 1990 p. 52). We believe that this project has taken a small step in meeting Singer's challenge through identifying some of the key attributes and behaviors of those states that use the most force in the system. Similarly, we have identified some of the key characteristics of those states that do not use force in the system. Although we do not claim that these characteristics cause war, we do suggest that there is a relationship between capabilities and the use of force; large and better-armed nations perpetrate most of the conflict in the international system. An understanding of this relationship might help government officials predict the occurrence of conflict in the system, and devise better responses to such conflict.

In an effort to be even more helpful, we suggest the following policy changes that take into account the relationships we have stipulated. First, a two part international arms limitation agreement that (1) reduces or eliminates the international arms trade (reducing V16&19) and (2) reduces national armed forces, weapons development, and military expenditures (reducing V 17&18). Second, instituting population control measures in developing countries and, possibly, encouraging some countries to split into large smaller nations. The latter procedure would reduce the ability for states to tap vast pools of human resources for the purposes of war. Third, adopt wealth redistribution programs that would shift funds from military budgets to quality of life budgets. Similarly, if such large, wealthy nations were disassembled into smaller units, and with less money devoted to the military, then those smaller nations might devote more resources to increasing the GNP per capita (which does not enhance the probability of force usage). Fourth, steps might be considered to depopulate large metropolitan cities, since those nations with a high percentage of urbanization were high force users. Some researchers have stipulated that living in such an area lends itself to aggressive behavior. A nation with fewer large cities might become more peace prone.

These proposals, of course, are not likely to be politically popular. They presented because they consistent with our findings and may stimulate thinking and new approaches to the problems of violence in the international system.

Although our study has not attempted to clearly identify the "true"cause(s) of war, we do believe that we have accurately stipulated the relationship between capabilities, certain behaviors and force through a large portion of the cold war period. Those states that have the greatest capabilities, e.g. wealth, population, and military resources, tended to use the most force. Further, those states have also been the most active states in the international system. Conflict and cooperation may be the two sides of the same coin, since the use of force appears to be linked to a high level of participation as defined by the WEIS data bank.

Vincent, Jack E. 1987. 'Analyzing Shifts in International Conflict from Quality of Life Indicators', Journal of Peace Research, vol. 24, no. 4, 393-405.

Vincent Jack E. 1996 'The Relative Importance of Power, Economic Development And Political System During the Middle of the Cold War Cold War,' Martin Journal of Peace Research, No. 2.