ON WAR (Part 3)

by Johan M.G. van der Dennen

2.10 Dissimilarity, heterogeneity and war

Dissimilarity of national attributes between two States seems to have had little influence on the likelihood of their going to war with each other. Singer (1972) finds that most of the wars in the "Correlates of War" project were between nations in close geographical proximity and similar on most attribute dimensions. Studying national dyads for 1955-1957, Rummel (1966) found only slightly more conflict as attributes of the two nations diverged. Divergence in the ethos of the elites of different nations was found by Rosecrance (1963) to be neither a necessary nor a sufficient condition for instability in international systems since 1740. Likewise Richardson (1960) found similarity or difference of language of little influence on the incidence of war.
Haas (1972) has considered the "heterogeneity" of a State (in ethnic, linguistic, religious, racial and national terms) and found it to correlate consistently with frequency of wars, military actions, and foreign conflict casualties. Voting support in the United Nations, for either the East or West Cold War Bloc, is not associated with foreign conflict, though the extent of participation in foreign affairs (in terms of bloc prominence, giving or getting aid, and the number of representatives at the United Nations) seems to be correlated in a positive way (Haas, 1972).
The DON project results include the general observation that: "The more similar two nations are in economic development, political orientation, Catholic culture, and density, the more aligned their voting in the UN and the less conflictful their interaction will be". Further: "the more dissimilar two nations are in economic development and size and the greater their joint technological capability to span geographic distance is, the more overt conflict they have with each other". And finally that: "Racial distance is the most important characteristic distinguishing between peace and conflict in international systems" (Rummel, 1969).
Richardson (1960), on the other hand, found little relationship between economic inequality and a propensity for war, though the distance between States on the dimensions of language and religion did correlate in particular unambiguous ways.
Finally, Russett (1967), after a detailed study of inter-State integration, concludes that : "At best, cultural similarity and voting behaviour (in the United Nations) make essentially no difference in the probability of conflict... But countries belonging to the same groupings by organizational membership, proximity or trade are more than twice as likely to fight than are nations which belong to different groups, or to none".
Richardson (1960) and Luard (1968) both found economic causes of minor importance in the wars they surveyed. Richardson also found relative wealth of little value in predicting war, but Rummel (1966), in his study of attribute similarities within dyads, found a high correlation between conflict and "high economic distance". Generally, the clustering of States into regions does have significance in terms of the likelihood of war, and conflict and integration are related, though in a three-step way: when the States involved are mutually irrelevant, war is not likely to occur; however, it is likely to be common when "capabilities and salience are moderately and narrowly focused". When capabilitie s are "numerous and varied", though, war is unlikely again (Russett, 1967).
Rummel (1968) has attempted to correlate the amount of national cooperation (in terms of membership of international organizations, treaties, aid and diplomatic representation) and the level of a State's international communications or transactions (in terms of mail, economic aid and trade measures) with foreign conflict behaviour, but has found very little connection. "Alliances" and foreign conflict have also been examined for covariant effect. Richardson (1960) concluded that previous alliance did reduce the likelihood of war between those who had been allied, though with many exceptions. Singer and Small (1968) have discovered a strong association between the number of alliance bonds which a nation has and the amount of war it subsequently experiences in the following years.

2.11 War and international organizations

Singer and Wallace (1970) as well as Rittberger (1971) carried out longitudinal studies of international organization-building and its effects on the incidence of international war for the periods 1815-1964 and 1865-1965, respectively. Singer and Wallace measured their independent variable, international organization-building, by counting and adding all intergovernmental organizations (IGOs) which existed during any given five-year period since 1815; in addition, they constructed two derived measures weighting individual IGOs on the basis of membership size and ascribed international status of their members. Rittberger, on the other hand, proceeded more selectively and focused on the UN system and its predecessor organizations only.
In both studies the results of the statistical tests showed unequivocally that, contrary to what one would hope, no negative association exists between measures of international organization-building and measures of incidence of international war for the time periods covered by these investigations. Rather, correlation coefficients are around zero in the case of frequency of war and even weakly-to-moderately positive in the cases of magnitude and severity of international war. However, further analyses revealed that these positive correlations are largely spurious: both the measures of international organization-building and the measures of magnitude and severity of international war are themselves positively correlated with a third variable, the advance of industrial civilization. This correlation simply reflects the growing destructiveness of war as a result of technological progress and its diffusion which has been made possible by the advance and spread of industrial civilization (Rittberger. 1973).

2.12 The domestic/foreign conflict linkage

Russell and Russell (1968) draw attention to a striking alternating sequence of domestic and foreign conflict which is provided by the story of the Plantagenet kings of England:

Henry III   Civil war
Edward I   Foreign war
Edward II   Civil war
Edward IIl   Foreign war
Richard II, Henry IV Civil war
Henry V   Foreign war
Henry VI   Civil war

They comment:

This particular sequence, and its significance, were not lost on William Shakespeare, who made a very careful study of the English chronicles. He presented his findings in the longest and most impressive series of historical plays ever written. He saw clearly that the alternation was explained by mass direction, promoted by the more energetic rulers, and this time he smuggled the message past the censors. In the last act of Henry IV, Part II, he expressed in one and a half lines the underlying mechanism of these seven plays. "Be it thy course", says the dying king to his heir Prince Hal,

    Be it thy course to busy giddy minds
    With foreign quarrels...

The sequel, in the next play, is the battle of Agincourt - for Henry V took his father's advice.

Considerable work has been done on the characteristics of a national political system and its likelihood to engage in domestic or international violence. Some of these studies are based on a combination of theoretical analysis and intuitive selected historical evidence. Societies in deep crisis, it has been plausibly argued many times (e.g. Lasswell, 1930; Haas, 1968), are more likely to elevate psychopathological personalities to key decision-making roles, with consequences for their international behaviour. Societies which fail to provide security of income and status for large parts of their populations are likely to give rise to extremist political movements, and their rulers may then have to choose between letting the increased aggressive attitudes of these groups explode in domestic strife, or assuaging them by rapid and far-reaching reforms, or channeling them into international conflict - with theory and history suggesting a preference for the third of these responses (Lasswell, 1935).
If this theory of alternative responses to domestic tensions were correct, then foreign conflicts would serve as substitutes for domestic ones. According to the "substitution theory", the frequency of the two types of conflict ought to be inversely correlated. Unfortunately, another version of the theory predicts the opposite. High degrees of tension ought to create a higher demand both for domestic and for international violence. According to this "joint demand theory" or "spill-over theory", the frequency of the two types ought to vary together. Finally, a mixed theory would suggest that the effects of substitution and joint demand ought just about to balance each other, and that there should be no observable correlation of any kind between the frequency of domestic and foreign conflict (Deutsch and Senghaas, 1973).
Several studies have dimensionalized domestic conflict behaviour by factor analysis (Rummel, 1963; 1966; Tanter, 1966; Feierabend and Feierabend, 1966; Bwy, 1968; Banks, 1972; Wilkenfeld, 1969; Gurr, 1968 et seq.). The results distinguish either two or three major dimensions of domestic conflict. In all orthogonally rotated factor analyses, a distinct turmoil dimension appears which is centred around strikes, riots and demonstrations. Terrorism and assassinations load on this factor but also load on the other two factors. The turmoil factor, therefore, involves disruptions with rather limited objectives and usually rather limited violence, though the violence can easily get out of hand as in the case of some riots. But widespread violence is not initially the explicit intent of the perpetrators (Finsterbusch, 1974).
The remaining domestic conflict variables produce either one or two factors. Rummel (1966) obtained a "revolution" factor involving coups, mutinies and plots, and a "subversion" factor involving civil war and extended violence. The latter is related to large-scale terrorism and riots. In Rummel's (1963) and Banks' (1972) factor analyses, the highest loading on the "revolution" factor is revolutions and on the "subversion" factor is guerrilla warfare. The Tanter, Wilkenfeld and Bwy factor analyses combine revolutions and guerilla warfare into a single "organized violence" factor and produced no significant third factor. These three analyses contained only 7 anti-establishment conflict variables as compared with 12 such variables in Rummel's 1966 analysis and 14 variables in the Feierabends' analysis. The second and third (or the two combined) factors, which we call, following Gurr, "conspiracy" and "internal war", differ from the turmoil factor in terms of the purpose of the perpetrators. Turmoil events usually involve public demands by lower groups for relatively specific actions on the part of the persons or groups with authority or are relatively goalless spontaneous reactions to an accumulation of frustrations. The conspiracy and internal war factors are attempts to take over the centres of power. When conspiracy and internal war are differentiated, the basis is mainly the scale of operation (Finsterbusch, 1974).
Lee (1931) found that internal violence in China clustered before, during and after external wars. Sorokin (1937) carried out a massive, longitudinal analysis of the ancient empires of Greece, Rome, Byzantium and a number of European nations over 14 centuries (525-1927). Sorokin reported that although there seemed to be a slight indication that internal disturbances tend to occur more frequently during and around years of war, upon closer examination the two processes tend to be independent of one another. Sorokin's analysis was based on time intervals of quarter centuries and centuries and thus does not yield very discriminant results, but rather explores gross trends for major outbreaks of internal and external violence. He states: "So far as century periods are concerned, each process has led a course independent of the other without positive or negative association".
Cattell (1949 et seq.), in a number of attempts aimed at the discovery of culture pattern profiles, performed factor analyses on a number of variables representing national characteristics from 1837 to 1937. The first two studies (Cattell, 1949; 1950) used 69 nations and yielded 12 orthogonally rotated factors. The first two factors contain the variables of interest to the present study. The internal and external violence dimensions appear to be independent of one another. In a third study (Cattell, 1951), 29 nations, whose data coverage had been considered poor, were dropped from the analysis. The 40 nations that were left were chiefly comprised of the "modern industrial nations". This analysis yielded quite different findings. By limiting the sample population, Cattell found that the two processes were not as independent as the previous studies had shown, but rather that both tended to load highly on the same dimension.
A series of studies, which also utilized factor analysis and, in addition, multiple regression techniques, were initiated by Rummel in the Dimensionality of Nations (DON) Project. All subsequent research in the linkage between internal and external behaviour of nations has relied heavily on the DON project for either methodology and/or data base. The relationship between domestic and foreign conflict behaviour in these studies was based on an analysis of twenty-two variables for all nations with populations over 800,000 in the years studied. For the Rummel data (1955-1957) the sample was 77 nations, while the Tanter data (1958-1960), as a result of population growth and the inclusion of newly independent nations, included a universe of 83 (Stohl, 1976).
The first Rummel (1963) and Tanter (1964) studies found a slightly positive association between domestic conflict behaviour and the more belligerent forms of foreign conflict behaviour, although for the most part, the independence of the two were reconfirmed. Subsequently, the 1955-1957 data were annually reanalysed by Chadwick (1963) and, with one-year time lag, this independence was further supported. Tanter (1966) lagged the 1958-1960 data on the 1955-1957 data and found that there was a slight increase in the relationship. He suggested that: "There may be no simple relationship between domestic and foreign conflict behaviour, but there may be a causal relationship which is being obscured by other phenomena". The studies also identified separate independent dimensions within both the internal and external dimensions themselves. Rummel (1963) states: "The domestic conflict behaviour of nations varies along three uncorrelated dimensions: turmoil, revolutionary and subversive. The foreign conflict behaviour of nations varies along three uncorrelated dimensions: war, diplomatic and belligerent".
Tanter (1964) found that the revolutionary and subversive dimensions were subsumed under one internal war dimension, while the turmoil dimension and the foreign conflict behaviour dimensions compared favorably with the Rummel study. The early DON project data base did not include the majority of African nations, since they were not independent in 1955. Collins (1969), employing the same general research strategy put forth by Rummel and his successors, investigated the relationship between foreign conflict behaviour and domestic disorder in Africa in the period 1963-1965. In this study, the first to investigate the relationship within one "region", Collins reports that there were differences between his results and the preceding work: "Foreign violence is related to conditions of domestic disorder more so in African States than elsewhere, although the size of the correlations indicates that foreign violence is a product of other factors as well, which have not been tapped in the present research".
Using Richardson's (1960) data over the 1820-1949 period, Denton (1966) described a relationship between a civil conflict dimension and inter-nation conflict with peaks in the former preceding peaks in the latter, and Denton and Phillips (1968) used Q. Wright's (1942; 1965) data for 1480-1900 to arrive at a correlation of .39 between percentage of civil wars and an index of international violence (Dowty and Kochan, 1974).
Haas (1965; 1968) confirmed the low correlations between foreign conflict and both legitimate domestic conflict and illegitimate domestic conflict (revolution, guerrilla), and in a smaller sample of 10 nations from 1900 to 1960 found little support for linking "use of military means" to "internal stresses", but he also uncovered a very high connection between foreign conflict and anomic domestic conflict (riots).
Feierabend and Feierabend (1966) and North and Choucri (1969) reported strong positive relationships between internal tensions or frustrations and external aggressiveness.
Also Weede (1975) found domestic and international conflicts closely related in the fifties and sixties of the twentieth century. However, he concludes that it is false that domestically unstable nations tend to be aggressive against other nations. Instead, they become victims of superpower intervention.
Haas (1965) finds more foreign conflict in "non-constitutiona1" governments, a claim supported very indirectly, according to Dowty and Kochan (1974), by Kaplan's (1968) observation that in three historic international systems studied, the threat to stability always came from the "sub-system dominant, directive" actors, and by Naroll's (1969) conclusion from his broad historical survey that States with greater centralization tend to be involved in more wars.

There seems to be at least equal evidence, however, against a connection between warlikeness and type of regime. Richardson (1960) argues that no States in his survey can be properly characterized as inherently belligerent or pacific, and Luard's (1968) survey of 62 wars in the 1865-1965 period also offers "little evidence" of such a connection. On a broader level, Haas' (1972) study of the linkage between warlike behaviour and some 200 societal characteristics concludes that there is little support for any of the classic theories on such linkage.
Most important of these contra-indications is Wilkenfeld's (1968 et seq.) study of the hypothetical link between the nature of a State's political system and its foreign-conflict behaviour, including war. Rummel and Tanter did not differentiate between the States they used, and may have concealed thereby positive connections that run against their results. Working with 74 states, Wilkenfeld divided them into three factored groups-politically "personalist" (Latin American dictatorships for example), "centrist" (socialist and Middle Eastern regimes) and "polyarchic" (economically developed, Western and westernized) - and correlations were then carried out using these groups for all the possible pairs of foreign and domestic conflict dimensions that Rummel identified. Correlational tests were made using zero time lags and lags of one year and two. In the "personalist" group internal conflict was accompanied by foreign conflict of the "diplomatic" sort (diplomatic expulsions and troop movements). Using a two-year time lag a significant association was discernible between the "turmoil" dimension (demonstrations, riots, government crises) and external "belligerency". Most interesting of all, however, was the occurrence of "war" and the apparent outbreak two years later of the subversive domestic activities of assassination and guerrilla revolt. This link Rummel's (1964) study, in its generalized, perhaps overgeneralized way, explicitly denied. For the "centrist" group domestic "revolutionary" activity (purges, general strikes, revolutions and numbers killed within the State) was followed in one or two years by all the 13 types of foreign-conflict behaviour. Wilkenfeld's results here might support the traditional argument that foreign conflict is used by State elites to distract from domestic disorder. Finally, "polyarchic" States experiencing "turmoil" (anti-government demonstrations, riots and major government crises) engage in all types of foreign-conflict behaviour, and vice versa. The relationship is a mutually reinforcing one, which likewise supports the traditional view.
Wilkenfeld's earlier results are further supported by those of Zinnes and Wilkenfeld (1971): the governmental structure of a State is important in predicting the relationships between foreign and domestic conflict behaviour.
Another finding, also generally consistent with Wilkenfeld's earlier study, is that some relationship exists between domestic conflict behaviour and foreign conflict behaviour. The relationship, however, is very specific. Internal warfare affects the transitions in the level of belligerency only for polyarchic States, while turmoil affects the transitions in the levels of belligerency only for centrist States. Again, there appears to be no relationship between domestic conflict behaviour and foreign conflict behaviour for personalist States.
A final result that emerged from the analyses was the effect of very extensive domestic conflict on international conflict behaviour. It was generally found that high levels of domestic conflict tended to be associated with a subsequent reduction in the level of international conflict commitments (Zinnes and Wilkenfeld, 1971).
In a third study Wilkenfeld and Zinnes (1973) employed Markov analysis to determine if foreign conflict behaviour affects the changes or transitions over time between levels of domestic conflict behaviour. Once again the Rummel and Tanter data were utilized and the factor scores were the data for the Markov analysis. When all nations were analysed "the foreign conflict behaviour war" was measured by the variables most highly loading on this factor (military actions, wars, mobilizations and foreign killed) affects transitions in domestic conflict behaviour as captured by the turmoil factor (strikes, riots, demonstrations) primarily when domestic conflict behaviour is at a very high level" (Wilkenfeld and Zinnes, 1973). When the nations were analysed by nation type it was again found that foreign conflict behaviour primarily affects transitions in domestic conflict for the personalist and polyarchic States.
Stohl (1971) reanalysed the Rummel (1963) data after dividing the nations into groups by political type corresponding once again to the Banks and Gregg study. Supporting Wilkenfeld's conclusion, political type of nation was found to play an important role in determining the conflict patterns within and between nations. The factor analysis of the measures of conflict behaviour found that there were different patterns of domestic and foreign conflict behaviour for each of the political types. Further exploration of the relationship, through multiple regression analysis with a one-year time lag, revealed moderately strong relationships in polyarchies, between diplomatic exchanges and general internal strife (r = .54), and between war and internal crises (r = .45). In the personalist nations increases in foreign conflict behaviour were associated with small increases in domestic conflict behaviour (multiple R's - .25, .45, .33, .40); while in the centrist cluster, there was no relationship discovered. It was hypothesized that the two major factors differentiating the three groups, level of social control and decision latitude, could account for the differences. As degree of social control and decision latitude decreases, elites are faced with greater pressure to justify their foreign conflict behaviour to their populations. It was suggested that higher information levels and inadequate elite justifications of behaviour may help to explain the higher associations between foreign and domestic conflict behaviour found in polyarchies and personalist nations (Stohl, 1976).
Stohl (1973; 1974; 1975) attacks the linkage problem via another route. Using a quasi-experimental design, he has looked for the effects of American war involvement on domestic levels of political violence. His results are mixed; there is no uniform pattern that holds both for each of the five wars (from the Spanish-American War to the Viet Nam War) and for each of the different dimensions of violence.
Two more recently reported studies of the relationship have attempted a more sophisticated analysis through the use of canonical analysis. Canonical analysis attempts to maximize the linear correlation between sets of variables. Phillips (1970), in a study of the impact of the conflict environment of nations via a regression analysis of the residuals in the canonical analysis of the DON data for 1963, found evidence for a relationship between internal and external violence. He states:

The conclusion here is that nations displaying domestic violence, having a low percentage of population in agriculture, who have tended to experience unlawful changes of offices in the recent past, and have a high cost-of-living index, tend to send more military violence to the environment than would be expected, given normal exchange with the environment. In other words, modernizing nations experiencing inflation and internal violence possibly associated with unlawful change of leadership, are likely to respond militarily to their environment.

It should be noted that these types of states more closely resemble the personalist type differentiated by Wilkenfeld and Stohl and the African States studied by Collins, lending support to those findings (Stohl, 1976).
The most ambitious use of the Rummel and Tanter data to date has been the attempt by Hazlewood (1973) to adapt this data for use within a general systems model. Hazlewood's main hypothesis concerning this chapter's interest was that "to the extent that internal variety is more extensive than internal constraint, the system stresses are likely to be manifested in foreign conflict behaviour at a later time period".
Employing the Tanter and Rummel foreign conflict behaviour factors and subsequent canonical and path analysis, Hazlewood found that: "Existing internal variety (societal diversity and turmoil), even without extreme economic expansion to activate it, is strongly associated with external conflict behaviour. Economic stability, societal heterogeneity and internal turmoil predict best to war". However, the path analysis revealed that, "The strongest path in the model relates turmoil to war for 1958-1960 through prior warfare". This would indicate that past foreign conflict behaviour is more robust than domestic conflict in predicting war.
Kegley, Richardson and Richter (1978) isolated for further study the size and importance of the military as a coercive resource of a State. They thus focus upon Hazlewood's finding that relatively highly militarized States experiencing domestic turmoil are disinclined to engage in foreign conflict. The correlation of their militarization measure with internal conflict is .15 and with foreign conflict .24. These results indicate that the militarization variable is at most only modestly correlated with either sphere of conflict. Thus, the coefficients fall somewhere between the contradictory results of, first, Haas (1974) and Rummel (1968), the latter concluding that "the military capabilities of a nation have little relationship to its foreign conflict behaviour", and, more recently, Choucri and North (1975) who discovered instead a strong correlation between the same two factors.
Overall, the impression one gets from the evidence is that military armament levels tend slightly to co-vary with the incidence of foreign conflict, but that military preparedness is not a very reliable predictor of external hostility. Suffice it to suggest here that militarization appears to predict potently to neither foreign nor domestic conflict in a direct fashion.
The empirical question Kegley, Richardson and Richter (1978) finally raise is, "What happens to the domestic-foreign conflict linkage when we take into consideration government types as measured by differences in level of military spending?". Here the 73 countries of their sample are divided into three subsets according to the levels of their military expenditures: low, medium and high; and the bivariate correlation between domestic and foreign conflict is computed separately for each of the three groups of countries. These coefficients are revealing.
The 39 polities comprising the "low" militarization category show a very small positive relationship between their domestic and foreign conflict behaviour, whereas those in the "medium" militarization group manifest virtually no relationship whatsoever between their domestic and foreign conflict propensities. To recapitulate, the bulk of the nations comprising the international system show moderate or low levels of military spending. And, for this large majority, internal strife does not appear to serve as a stimulus to or catalyst for foreign conflict. The picture quickly changes, however, when we turn to the highly militarized societies. For these 10 countries (14 percent of the population under investigation), the relationship between domestic and foreign conflict behaviour is inverse (r = -.49), and this association is significant at the .07 level, corroborating Hazlewood's determination. This evidence tells us that in militarized countries, the higher the level of civil strife, the lower the level of external conflict and, conversely, that militarized nations experiencing low levels of domestic turmoil tend to be more conflictual in the behaviour they direct toward foreign targets. Moreover, the finding informs us more broadly that only in highly militarized societies does a patterned relationship between civil strife and foreign conflict exist.
In sum: when countries are grouped according to their relative efforts to militarize, civil strife is negatively related to (later) foreign conflict in the most militarized societies, consistent with Hazlewood's inquiry into governmental responses to domestic stress (Kegley, Richardson and Richter, 1978).
Sloan (1978) studied dyadic linkage politics in Lebanon, and concluded:

There is a strong relationship between domestic conflict and conflict directed towards external supporters for the dissident elements in a national conflict, especially if the possibility of national `disintegration', though not necessarily the disappearance of the nation as a political entity, exists. It is the direct threat to the political fabric of the State, and not a physical threat to the continued existence of the nation, that results in a strong correlational linkage between domestic conflict levels and dyadic external conflict.

The nature or state of a country's economy has also been tested for the difference in the way it and foreign-conflict behaviour concur. Rummel (1968) finds no correlation between the two, measuring the economic development dimension in terms of telephones per capita, G.N.P. per capita, energy consumption per capita and percentage of the population engaged in agriculture. He cites other studies that corroborate this conclusion.
Zinnes (1972), however, in reviewing Rummel's empirical results, notes a set of positive correlations within them that would seem to indicate that the more developed countries (measured by their G.N.P., rate of population increase, number of calories consumed in relation to the number nutritionally required, steel production and electricity generated) engage in more foreign-conflict behaviour, particularly of the protest, expulsion of lesser diplomatic personnel and troop movement sorts. Haas (1965), too, has found that "the most, and, to a lesser extent, the least developed countries, which one might expect to have a high degree of economic stability, exhibit more significant foreign conflict than do the underdeveloped and intermediate types".
Retesting this result, using U.N. data to estimate national wealth, Haas (1965) has confirmed that rich States have more foreign conflict than most developing countries. In a later study (Haas, 1968), however, he draws the opposite conclusion: "rural international systems", he says, "are more peaceful than transitional industrial international systems... [but as] industrialization proceeds, war is less necessary for solving internal problems". Here, however, indices of social "stress" and "strain" have intervened.
Haas (1965) in particular has developed a set of domestic measures in these terms, and looked for foreign-conflict correlations with them. In terms of "stress", he tested a limited sample of ten countries for the years from 1900 to 1960 and found a positive connection between "unemployment" and the frequency of war. Rural countries differed markedly from States with large urban populations in this regard, though the sample itself was loaded with industrialized nations.

Lacking an industrial base for prolonged or total war, rural states are much more aggressive in entering war as an immediate escape from sudden stress than are urban countries. Nevertheless, rural nations are so often isolationist... that they respond to few of their economic crises in a violent external manner. Urban countries are less immediately aggressive, but many of them find it convenient to eliminate the unemployment problem. And the effect of militarization by several countries has been to feed fears and suspicions of other States, thus triggering fateful arms races. (Haas, 1965)

That stress and strain in a societal system might spill over into State behaviour on the international level is consistent with the evidence presented by Haas (1968) based on the analysis of 10 countries for the 1900-1960 period. He presents a theory of coping mechanisms linked to a continuum from rural to urban societies. A completely rural society is called Type I; a totally urban society is Type V; these are ideal constructs. Intermediate positions are Types II, III and IV. Forms of deviance, using Parsons' (1959) scheme, can be associated with each point on the continuum. Because revolutions and homicides are associated with rural settings, and suicides and alcoholism most prevalent in urban milieus, active orientations are located towards the rural end, passivity towards the urban end. In Type V, it is postulated that stress is eliminated before it is strainful, so conformity is its model orientation. This pattern is presented in table 7.

TABLE 7: Distance and the rural-urban continuum

Types of Societies
Deviant Orientations (Parsons) Main Deviant Acts Main Types of Wars


I. Rural- agricultural Active-Person Focused Homicides Outlet Nomadic tribes
II. Rural and Semi-Industrial Active-Norm Focused Homicides Suicides Outlet
Finland 1900-1960
Japan 1900-1945
Norway 1900-1918
Spain 1900-1960
III. Mixed Rural-Urban Passive-Norm Focused Alocholism
Arms Race Australia 1900-1960
France 1900-1960
Germany 1900-1918
Japan 1946-1960
Norway 1919-1960
Switzerland 1900-1945
U.S.A. 1900-1918
IV. Mostly Urban-Industrial Passive-Person Focused Suicides Arms Race
Germany 1919-1960
Great Britain 1900-'60
Switzerland 1946-1960
U.S.A. 1919-1960
V. Fully Urban- Industrial None (Conformity) None (Conformity) Deterrence Orwell's 1984

The next problem is to fit coping mechanisms into the picture. "Outlet" wars are related to lack of mechanisms in rural and semi-industrial countries for handling stress directly; "arms race" wars are associated with the use of militarization as a means of dealing with economic crisis in the mixed societies; "deterrence" involves mostly urban industrial States. Accordingly, it may be argued that there are at least 5 distinct causal chains linking stress conditions with the use of military means of international conflict resolution, one corresponding to each type. An extended examination of the process of coping with stress is also presented by Haas (1968).
If societies and economies become very complex, a recent study has demonstrated according to Haas (1968), suicide, homicide and compulsive alcoholism are less socially available, but deaths due to ulcers and hypertension increase. Rudin (1963) found that the "need for power" is related to violent-demonstrative death rates (rho = .41) and the "need for achievement" correlates (rho = .66) with deaths due to inhibition and repression.

2.13 Population and war

In the Cattell (1949) study, population density has a +.32 score on the 'Cultural Pressure' factor, and urbanization has a +.78 loading; war participation had a +.62 position. In other words, war, population density and urbanism are highly intercorrelated with the dimension of societal stress. Logically war could not of itself breed high population density or urbanization, so one might infer that urban population density is one antecedent to wars.
Haas (1965) used the "Cross-Polity Survey" (Banks and Textor, 1963) to test this hypothesis. He concludes that: "the stress within overcrowded cities bears more relation to international conflict than the stresses associated with life in densely settled farmlands" (Haas, 1965).
In support here we might include Rummel's (1964) findings that demographic variables (population density, birth rate, infant mortality and rate of population increases) were the most important predictors of conflict within and between States. Instead of the expected correlation between the degree of foreign conflict and population density, Haas (1965) finds "a slight inverse relation" in his study of societal variables contributing to foreign conflict behaviour of nation-States. Finding no relationship between the rapid increase of world population from 1820 and 1949 and any proportionate increase in the frequency of, and losses from, war, Richardson (1960) concluded: "There is a suggestion, but not a conclusive proof, that mankind has become less warlike since A.D. 1820".
Singer (1972) too finds "no significant association between a nation's growth rate in population or density and its war-proneness" in his study of 93 international wars during the period 1816-1965.
There is no definite evidence that overpopulation or density per se leads to war. Empirical correlations are generally spurious and do not hold cross-culturally. Many densely populated areas in the world have been highly stable, and many States in Latin America and Africa with low population densities have been highly unstable (Weiner, 1971). However, the search for raw materials and markets has often been a prominent cause of international conflict. The interactive effects of the population growth, technological development and resource constraints have been traditionally associated with the extension of national activities outside of legal boundaries and with the ensuing competition for resource and territorial control. For example, nineteenth-century colonial expansion was accompanied by considerable population growth in Europe in combination with increases in economic productivity and technological capabilities (Weiner, 1971).
In a simulation model of international conflict Choucri et al. (1971) have shown that population growth, economic and technological development, resource utilization, national expansion and international behaviour are intensely interdependent and linked with requirements for basic resources.
Choucri and Bennett (1972) and Choucri and North (1972) use population figures as part of a more general equation that leads in a general way to foreign conflict and war. In this equation, however, there are at least two important intervening variables - the contemporary demand for resources, and the level of technology. An increase in population must bring about increased demands for resources and a greater level of technological development before "lateral pressures", competition and crisis are likely to lead to violent conflict.
In her monograph "Population Dynamics and International Violence", the most extensive quantitative analysis available to date, Choucri (1974) presents the following conclusions:

The qualitative evidence from Part I and the more systematic evidence from Part II both strongly suggest that population factors indeed have a pronounced effect upon the development of conflict situations, and can often be critical determinants of violence and warfare. But the linkages between population and violence are rarely direct: complex intervening networks are at work. Major wars, as well as local conflicts, often emerge by way of a two-step process: First, in terms of internally generated pressures and demands occasioned by growing needs associated with added population: and then in terms of reciprocal comparisons, rivalries and conflict for control over resources, territory, valued goods or spheres of influence. Each step is closely related to the other, and each can be traced to the interaction among the population, resource and technological attributes of a society. In those terms, population factors amount to critical determinants of violent conflict.


There is no evidence to suggest that density by itself emerges as an important determinant of conflict and violence. Far more critical appears to be the location and distribution of population in relation to resources (as distinct from pressures upon resources) and location in relation to national borders. In the developing world, at least, these two factors emerge as critical in conflict situations. The location of population and its distribution in relation to resource may often exacerbate tensions between the resource-rich and resource-poor areas, making the task of developing a viable political community considerably more difficult.

2.14 In summary

Vasquez (1976) presented a synoptic inventory of null, promising and possible statistical findings. Among these the following are pertinent:

Vasquez comments: "the notion that the national attributes of a nation or internal stress within a nation may be related to international conflict is not supported. The notion that alliances or the balance of power are related to conflict or violence is not supported. The belief that transactions lead to peace is not supported".

Update of the Macroquantitative Literature until 1998