3
Conflict ... is a theme that has occupied
the thinking of man more than any other,
save only God and love.
Anatol Rapoport,
Fights, Games, and Debates
THE INSTITUTIONS, persons, and policies of rulers have inspired
the violent wrath of their nominal subjects throughout the history
of organized political life. A survey of the histories of European
states and empires, spanning twenty-four centuries, shows that
they averaged only four peaceful years for each year of violent disturbances.1
Modern nations have no better record: between 1961
and 1968 some form of violent civil conflict reportedly occurred in
114 of the world's 121 larger nations and colonies.2
Most acts of
group violence have negligible effects on p olitical life; but some
have been enormously destructive of human life and corrosive of
political institutions. T e n of the world's thirteen most deadly conflicts
in the past 160 years have been civil wars and rebellions; 3
since 1945, violent attempts to overthrow governments have been
more common than national elections. The counterpoise to this
grim record is the fact that political violence has sometimes le d
to the creation of new and more satisfying political communities.
The consequences of the American, Turkish, Mexican, and Russian
revolutions testify in diffe rent ways to the occasional b e neficence
of violence.
In this study political violence refers to all collective attacks
1
Pitirim Sorokin, Social and Cultural Dynamics, Vol. III: Fluctuations of Social
Relationships, War and Revolutions (New York: American Book Co., 1937), 409-
475. Twelve countries and empires were studied over the period 500 B.C. to 1925
A.D., none of them for the entire span. Only "important" disturbances were recorded,
i.e., those mentioned in standard histories.
2
Based in part on data for 114 polities reported in Ted Robert Gurr, "A Comparative
Study of Civil Strife," in Hugh Davis Graham and Ted Robert Gurr, eds.,
Violence in America: Historical and Comparative Perspectives (Washington, D.C.:
National Commission on the Causes and Prevention of Violence, 1969).
3 According to data collected by Lewis F. Richardson, Statistics ofDeadly Quarrels
(Pittsburgh: The Boxwood Press, 1960), 32-43, to which events after 1948 are
added here. The conflicts most destructive of life were World Wars I and II. Of the
other eleven all but one were primarily internal wars and all caused more than
300,000 deaths each; the Tai-P'ing Rebellion, 1851- 64; the American Civil War;
the Great War in La Plata, 1865-70; the post-revolutionary Civil War in Russia,
1918-20; the first and second Chinese Civil Wars, 1927-36 and 1945-49; the
Spanish Civil War; the communal riots in India and Pakistan, 1946-48; the Vietnam
War, 1961-present; the private war between Indonesian Communists and their
opponents, 1964-66 (casualty figures are problematic); and the Nigerian Civil
War, 1967-.
1. Explanations of Political Violence
4  ✻  Chapter One
within a political community against the political regime, its actors
-including competing political groups as well as incumbentsor
its policies. The concept represents a set of events, a common
property of which is the actual or threatened use of violence, but
the explanation is not limited to that property. The concept subsumes
revolution, ordinarily defined as fundamental sociopolitical
change accomplished through violence. It also includes guerrilla
wars, coups d'etat, rebellions, and riots. Political violence is
in turn subsumed under "force," the use or threat of violence by
any party or institution to attain ends within or outside the political
order. The definition is not based on a prejudgment that political
violence is undesirable. Like the uses of violence qua force by the
state, specific acts of political violence can be good, bad, or neutral
according to the viewpoint of the observer. Participants in political
violence may value it as a means of expressing political demands
or opposing undesirable policies. Limited violence also can be
useful for rulers and for a political system generally, especially
as an expression of social malaise when other means for making
demands are inadequate. Ethical judgments are held in abeyance
in this study to avoid dictating its conclusions. But it does not require
an ethical judgment to observe that intense violence is destructive:
even if some political violence is valued by both citizens
and rulers, the greater its magnitude the less efficiently a political
system fulfills its other functions. Violence generally consumes
men and goods, it seldom enhances them.4
Despite the frequency and social impact of political violence, it
is not now a conventional category of social analysis. Yet some common
properties of political violence encourage attention to it rather
than more general or more specific concepts. Theoretically, all
such acts pose a threat to the political system in two senses: they
challenge the monopoly of force imputed to the state in political
theory; and, in functional terms, they are likely to interfere with
and, if severe, to destroy normal political processes. Empirical
justification for selecting politicalviolence as a universe for analysis
is provided by statistical evidence that political violence comprises
events distinct from other measured characteristics of nations,
and homogeneous enough to justify analysis of their common
characteristics and causes. For example, countries experiencing
4
It is possible that political violence can increase the sum total of satisfactions
of society's members. This can be true if violence and its immediate effects are
intrinsically valued more than the material and human resources it consumes, or
if violence serves a popularly approved regulatory function, as it did for the American
vigilante movements. A hypothetical relationship of this type is shown schematically
in fig. A. It is likely that high magnitudes of violence destroy more than
they create, at least in the short run. When the time dimension is taken into ac-
Explanations of Political Violence  ✻  5
extensive political violence of one kind-whether riots, terrorism,
coups d'etat, or guerrilla war- are rather likely to experience other
kinds of political violence, but are neither more or less likely to
be engaged in foreign conflict.5
The properties and processes that
distinguish a riot from a revolution are substantively and theoretically
interesting, and are examined at length in this study, but
at a general level of analysis they seem to be differences of degree,
not kind.6
The searchfor general causes and processes of political
violence is further encouraged by the convergence of recent case,
comparative, and theoretical studies. One striking feature of these
studies is the similarity of many of the causal factors and propositions
they identify, whether they deal with revolution, urban rioting,
or other forms of political violence. This similarity suggests
that some of their findings can be synthesized in a more efficient
set of testable generalizations.
However good the prospects seem for a general analysis of political
violence, research on it has been quite uneven, both in subcount,
however, intense political violence, though it destroys much in the short
run, may have the long-run payoffs either of stimulating rulers to increase outputs
or of restructuring society in such a way that total satisfactions are substantially
increased. This kind of relationship is sketched in fig. B.
Gain '
Overall
Value
Position
Magnitudes of
Loss violence
Figure A
high
Figure A. Hypothetical effects of violence
on satisfactions in a society in
which violence is valued, at time X.
Gain
Overall
Value
Position
Loss
Figure B
Figure B. Hypothetical effects of intense
violence on satisfactions in a
society in which violence leads to reforms,
over time.
5
Rudolph Rummel, "Dimensions of Conflict Behavior Within and Between
Nations," General Systems Yearbook, VIII (1963), 1-50; and Raymond Tanter,
"Dimensions of Conflict Behavior Within and Between Nations, 1958-1960,"
j ournal of Conflict Resolution, X (March 1966), 41-64. There are statistically significant
relations among some forms of internal and external conflict, but they are
relatively weak. For factor-analytic evidence that types of political violence are
independent of most measured characteristics of nations, see Rudolph Rummel,
"Dimensionality of Nations Project; Orthogonally Rotated Factor Tables for 236
Variables," Department of Political Science, Yale University (New Haven, July
1964), mimeographed.
6 Billington points out, for example, that the Russian Revolution comprised an
"insurgency" from below-actually extensive rioting and mutinies-followed by a
coup d'etat and a protracted civil war. James H. Billington, "Six Views of the Russian
Revolution," World Politics, XVIII (April1966), 452ff.
6  ✻  Chapter One
stance and in disciplinary approach. There is considerable European
historical scholarship on segments of the subject, notably the
peasant rebellions of the twelfth through nineteenth centuries and
the great revolutions of England, France, and Russia. American and
European scholars, most of them also historians, have in recent
years contributed a modest case-study literature. American policy
scientists have written a small flood of treatises on the causes and
prophylaxis of subversive warfare, most of which seem to have had
neither academic nor policy impact. The lapses of attention are
striking by comparison. Of all the riotous mobs that have clamoured
through the streets of history, only the revolutionary crowds of
eighteenth- and nineteenth-century Europe and the ghetto rioters
of twentieth-century America have attracted much scholarly attention.7
There are relatively few case studies of political violence in
the non-Western world, and fewer systematic comparative studies
or attempts at empirical theory. Experimental studies dealing
with social-psychological mechanisms of collective violence can
be counted on one hand.8 Among social scientists the historians
have been by far the most active; American political scientists have
until recently neglected the subject.9 Of 2,828 articles that appeared
in the American Political Science Review from its estab-
7 A useful survey of current knowledge of urban and rural turmoil in eighteenthand
nineteenth-century Europe is George Rude, The Crowd in History: A Study
of Popular Disturbances in France and England, 1730-1848 (New York: Wiley,
1964). A representative survey of current knowledge of urban violence in modern
societies is Louis H. Masotti and Don R. Bowen, eds., Riots and Rebellion: Civil
Violence in the Urban Community (Beverly Hills: Sage, 1968).
8 The studies, using a va1·iety of experimental techniques, include Norman
Polansky, Ronald Lippitt, and Fritz Redl, "An Investigation of Behavioral Contagion
in Groups," Human Relations, III (No. 3, 1950), 319-348; G. E. Swanson,
"A Preliminary Laboratory Study of the Acting Crowd," American Sociological
Review, XVIII (October 1953), 522-533; Norman C. Meier, G. H. Mennenga, and
H. J. Stoltz, "An Experimental Approach to the Study of Mob Behavior," Journal
of Abnormal and Social Psychology, XXXVI (October 1941), 506-524; Kurt Lewin,
Ronald Lippitt, and Ralph K. White, "Patterns of Aggressive Behavior in Experimentally
Created Social Climates," Journal of Social Psychology, X (May 1939),
271-299; and David Schwartz, "Political Alienation: A Preliminary Experiment
on the Psychology of Revolution's First Stage," paper read at the Annual Meeting
of the American Political Science Association, 1967. Many other experimental
studies are relevant to collective violence; these deal explicitly with social-psychological
factors facilitating violence.
9 The renaissance of systematic theoretical and empirical work on political violence
by political scientists can be dated from the publication of three articles that
appeared in 1962 and 1963: James C. Davies, "Toward a Theory of Revolution,"
American Sociological Review, XXVII (February 1962), 5-19; Harry Eckstein,
"Internal War: The Problem of Anticipation," in Ithiel de Sola Pool et al., Social
Research and National Security: A Report Prepared by the Research Group in Psychology
and the Social Sciences (Washington, D.C.: Smithsonian Institution,
March 5, 1963), published in revised form as "On the Etiology of Internal Wars,"
History and Theory, IV (No.2, 1965), 133-163; and Rummel, "Dimensions of Conflict
Behavior."
Explanations of Political Violence  ✻  7
lishment in 1906 through 1968, only twenty-nine appear from their
titles to be concerned with political disorder or violence. Moreover
twelve of the twenty-nine were concerned specifically with
revolution, and fifteen appeared after 1961.10
Political scientists might be expected to have a greater concern
with political violence than others. Authoritative coercion in the
service of the state is a crucial concept in political theory and an
issue of continuing dispute.11 Some have identified the distinctive
characteristic of the state as its monopoly of physical coer~ion.
Max Weber, for example, wrote that violence is a "means specific"
to the state and that "the right of physical violence is assigned to
all other associations or individuals only to the extent permitted
by the state; it is supposed to be the exclusive source of the 'right'
to use violence." 12 Thomas Hobbes, dismayed by the brutish anarchy
of men living outside the restraint of commonwealths, conceived
the sovereign's control of coercion to be the foundation of
the state and the social condition.13 Schattschneider sees conflict,
which subsumes violence, as the central concept of political
science.B Nieberg emphasizes the positive functions of nonauthoritative
violence and its threatened use as an instrument of
social change.15 From any of these perspectives the occurrence of
collective, nonauthoritative violence appears to pose two fundamental
questions for political science: From what sources and by
what processes does it arise, and how does it affect the political
and social order?
What Is to Be Explained?
This study proposes some general answers to three basic questions
about our occasional disposition to disrupt violently the
order we otherwise work so hard to maintain: What are the psy-
10 By contrast, 111 articles deal specifically with Great Britain, some 140 with
the U.S. Congress, and about 250 with constitutions or constitutional issues. These
counts were facilitated by the use of the Key-Word-in-Context index of article titles
in Kenneth Janda, ed., Cumulative Index to the American Political Science Review:
Volumes 1-57: 1906-1963 (Evanston: Northwestern University Press, 1964).
11 A useful analysis of alternative conceptions of authoritative violence is E. V.
Walter, "Power and Violence," American Political Science Review, LVIII (June
1964), 350-360. A survey of the origins, functions, and forms of legitimate and illegitimate
violence, within and among states, is Chapter 8, "Civil Conflict and War,"
Alfred de Grazia, Politics and Government, Vol. 1: Political Behavior, rev. edn.
(New York: Collier, 1952, 1962), 265-301.
12 Max Weber, "Politik als Beruf," Gesammelte Politische Schriften (Tubingen:
J. C. B. Mohr, 1958), 494, translated in E. V. Walter, 359.
13 Thomas Hobbes, Leviathan (Oxford: Basil Blackwell, 1946).
14 E. E. Schattschneider, "Intensity, Visibility, Direction and Scope," American
Political Science Review, LI (December 1957), 933-942.
15 H. L. Nieburg, "The Threat of Violence and Social Change," American Political
Science Review, LVI (December 1962), 865-873, and Political Violence: The
Behavioral Process (New York: St. Martin's Press, 1969).
8  ✻  Chapter One
chological and social sources of the potential for collective violence?
What determines the extent to which that potential is
focused on the political system? And what societal conditions affect
the magnitude and form, and hence the consequences, of violence?
The study has four primary objects of analysis. Two are
intervening variables: the potential for collective violence and the
potential for political violence. Propositionally, potential for collective
violence is a function of the extent and intensity of shared
discontents among members of a society; the potential for political
violence is a function of the degree to which such discontents are
blamed on the political system and its agents. The remaining objects
of analysis are dependent variables: the magnitude ofpolitical
violence and the forms of political violence, both of which are discussed
below.
Theories of revolution are usually concerned with specifying a
relationship between some set of preconditions and the occurrence
of revolution. Political violence, however, is a pervasive
phenomenon, as was suggested above: few contemporary or historical
societies have been free of it for long. It may be useful for
microanalysis to specify whether political violence is likely in a
given society at a particular point in time. For macroanalysis, however,
the more interesting questions are the determinants of the
extent of violence and of the forms in which it is manifested. If
one's interest is the effects of political violence on the political
system, the questions of its magnitude and kind are both highly
relevant. And if one is concerned in an ethical way with political
violence, then almost certainly one wants to assess its human
and material costs, and consequently the determinants of its magnitude.
Various measures of the relative extent of political violence
have been used in recent comparative studies. Sorokin combined
measures of the proportion of a nation affected (social area), proportion
of population actively involved, duration, intensity, and
severity of effects of violence in assessing the magnitude of internal
disturbances. Tilly and Rule make use of man-days of participation.
Rummel and Tanter have used counts of numbers of events. The
Feierabends have developed a scaling procedure that takes account
of both number of events and a priori judgments about the
severity of events of various types.16 Some researchers have used
16 Sorokin, op. cit.; Charles Tilly and James Rule, Measuring Political Upheaval
(Princeton: Center of International Studies, Princeton University, Research Mono- ·
graph No. 19, 1965); Rummel, "Dimensions of Conflict Behavior ..."; Raymond
Tanter, "Dimensions of Conflict Behavior Within Nations, 1955-60: Turmoil and
Internal War," Peace Research Societies Papers, III (1965), 159-164; and Ivo K. and
Rosalind L. Feierabend, "Aggressive Behaviors Within Polities, 1948-1962; A
Cross National Study," journal of Conflict Resolution, x (September 1966), 249-
271.
Explanations of Political Violence  ✻  9
the grisly calculus of number of deaths resulting from violence.17
The proposed relation between perceived deprivation and the
frustration concept in frustration-anger-aggression theory, to be discussed
in chapter 2, provides a rationale for a more general defini.
tion of magnitude of violence and a more precise specification of
what it comprises. The basic frustration-aggression proposition is
that the greater the frustration, the greater the quantity of aggression
against the source of frustration. This postulate provides the
motivational base for an initial proposition about political violence:
the greater the intensity of deprivation, the greater the magnitude
of violence. (Other perceptural and motivational factors are also
relevant to political violence, but many of them can be subsumed
by the deprivation concept, as is suggested in chapter 2.) Intense
frustration can motivate men either to intense, short-term attacks
or to more prolonged, less severe attacks on their frustrators. Which
tactic is chosen is probably a function of anticipated gain, opportunity,
and fear of retribution, which in political violence are
situationally determined. Hence the severity of deprivation affects
both the intensity of violence, i.e. in the extent of human and physical
damage incurred, and its duration. Moreover there are evidently
individual differences-presumably normally distributedin
the intensity of frustration needed to precipitate overt aggression.
Extension of this principle to the deprivation-violence relationship
suggests that the proportion of a population that participates
in violence ought to vary with the average intensity of perceived
deprivation. Mild deprivation will motivate few to violence,
moderate deprivation will push more across the threshold,
very intense deprivation is likely to galvanize large segments of a
political community into action.
This argument suggests that magnitude of political violence
has three component variables that ought to be taken into account
in systematic analysis: the extent of participation within the political
unit being studied (scope), the destructiveness of action (intensity),
and the length of time violence persists (duration).
Sorokin's empirical work takes all three aspects into account; so
does mine.18
The intensity and scope of relative deprivation and magnitude of
17 Among them Richardson, especially chap. 2 and 4; Rummel, "Dimensions of
Conflict Behavior"; Tanter; and Bruce M. Russett, "Inequality and Instability:
The Relation of Land Tenure to Politics," World Politics, XVI (April 1964), 442-
454.
18 Ted Gurr with Charles Ruttenberg, The Conditions of Civil Violence: First
Tests of a Causal Model (Princeton: Center of International Studies, Princeton
University, Research Monograph No. 28, 1967); Ted Gurr, "A Causal Model of
Civil Strife: A Comparative Analysis Using New Indices," American Political Science
Review, LXII (December 1968), 1104-1124.
10  ✻  Chapter One
violence are unidimensional variables. Theoretically, and empirically,
one can conceive of degrees or quantities of each in any
polity. The forms of violence, however, are attributes that do not
form a simple dimension. A society may experience riots but not
revolution, revolution but not coups d'etat, coups d'etat but not
riots. Hypotheses about forms of violence as dependent variables
thus are necessarily different from those about deprivation and
magnitude of violence. They are expressed in terms of probabilities
(the greater x, the more likely y) rather than strict concomitance.
The question is how many forms of political violence ought to be
accounted for in a general theory. The principle of parsimony,
which should apply to dependent as well as independent variables,
suggests using a typology with a small number of categories, events
in each of which are fairly numerous.
Conventional taxonomies, of which there are many, provide
little help. Some, like that of Lasswell and Kaplan, provide simple
typologies for revolutions but not for political violence generally.19
Eckstein proposes a composite typology comprising unorganized,
spontaneous violence (riots), intraelite conflicts (coups), two varieties
of revolution, and wars of independence.20 Perhaps the most
complex typology is Rummel's list of twenty-five types of domestic
conflict, the analysis of which provides an empirical solution to the
problem of a parsimonious typology. In Rummel's analysis, and in a
number of subsequent studies, data on the incidence and characteristics
of various types of political violence were collected and tabulated
by country and the "country scores" (number of riots, assassinations,
coups, mutinies, guerrilla wars, and so on, in a given time
period) were factor analyzed. Whatever the typology employed,
the period of reference, or the set of countries, essentially the same
results were reported. A strong turmoil dimension is characterized
by largely spontaneous strife such as riots and demonstrations. It
is quite distinct both statistically and substantively from what can
be called a revolutionary dimension, characterized by more organized
and intense strife. This revolutionary dimension has two
components that appear in some analyses as separate dimensions:
internal war, typically including civil war, guerrilla war, and
some coups; and conspiracy, typically including plots, mutinies,
and most coups.21
19 Harold Lasswell and Abraham Kaplan, Power and Society: A Framework for
Political Inquiry (New Haven: Yale University Press, 1950), 261-268. The types
are palace revolution, political revolution, and social revolution.
20 "On the Etiology oflnternal Wars," 135-136.
21 Rummel's typology appears in "Dimensions of Conflict Behavior," 25-26.
Two summary articles on the factor analyses are Rudolph J. Rummel, "A Field
Explanations of Political Violence  ✻  11
These types are not absolutely distinct. The analyses mentioned
on pp. 4-5 indicate that, at a more general level of analysis, political
violence is a relatively homogenous universe. Within that universe,
however, some kinds of violence tend to occur together, and the
occurrence of some types tends to preclude the occurrence of other
types. The principal distinction between turmoil and revolution
is the degree of organization and focus of violence, a distinction
also made by Eckstein in his composite typology. A major difference
between the internal war and conspiracy components of the
revolutionary dimension is one of scale. General definitions of
the three forms of political violence examined in this analysis are
as follows:
Turmoil: Relatively spontaneous, unorganized political violence
with substantial popular participation, including
violent political strikes, riots, political
clashes, and localized rebellions.
Conspiracy: Highly organized political violence with limited
participation, including organized political assassinations,
small-scale terrorism, small-scale guerrilla
wars, coups d'etat, and mutinies.
Internal war: Highly organized political violence with widespread
popular participation, designed to overthrow
the regime or dissolve the state and accompanied
by extensive violence, including large-scale
terrorism and guerrilla wars, civil wars, and revo-
lutions.22
In summary, this study is an attempt to analyze, and develop
testable general hypotheses about, three aspects of political violence:
its sources, magnitude, and forms. The processes by which
the potential for violence develops and the kinds of conditions
and events that channel its outcome are examined as part of this
Theory of Social Action With Application to Conflict Within Nations," Yearbook
of the Society for General Systems Research, x (1965), 183-204, and Tanter, "Dimensions
of Conflict Behavior."
22 Eckstein, in "On the Etiology of Internal Wars," reintroduced the term "internal
war," but defined it considerably more broadly as "any resort to violence
within a political order to change its constitution, rulers, or policies," 133. In ordinary
language, however, the phrase connotes a degree of participation and organization
of conflict that is not characteristic of the events I have separately categorized
under turmoil and conspiracy. The three categories differ from the categories
with the same labels in Gurr, "A Comparative Study of Civil Strife," and Gurr,
"A Causal Model of Civil Strife," only in their exclusion of nonviolent strife and of
violent conflict between apolitical groups.
12  ✻  Chapter One
analysis. Two topics often examined in theories of revolution are
examined here only in passing: the immediate precipitants of violence,
about which most generalizations appear trivial; and the
long-run outcomes of various kinds of political violence, about
which there is little empirical evidence or detailed theoretical
speculation.
Toward an Integrated Theory of Political Violence
The basic model of the conditions leading to political violence
used in this study incorporates both psychological and societal
variables. The initial stages of analysis are actor-oriented in the
sense that many of the hypotheses about the potential for collective
action are related to, and in some instances deduced from, information
about the dynamics of human motivation. The approach is
not wholly or primarily psychological, however, and it would be
a misinterpretation of the arguments and evidence presented here
to categorize it so. Most ofthe relationships and evidence examined
in subsequent stages of analysis are those that are proposed or
observed to hold between societal conditions and political violence.
The psychological materials are used to help provide causal
linkages between and among societal variables and the dependent
variables specified above: the potential for collective and political
violence; the magnitude of political violence; and the likelihood
that political violence will take the form of turmoil, conspiracy, or
internal war. Use of psychological evidence in this way makes
certain kinds of social uniformities more clearly apparent and
comprehensible, and contributes to the simplification of theory.
At the same time the analysis of societal relationships is crucial for
identifying the sources of some common psychological properties
of violence-prone men and for generalizing about the many facets
of political violence that have no parallels in psychological dynamics.
The goal of this analysis, at best only partly realized, was
proposed by Inkeles in the context of a discussion of social structure
and personality: "What is required ... is an integration or
coordination of two basic sets of data in a larger explanatory
scheme-not a reduction of either mode of analysis to the allegedly
more fundamental mode of the other." 23
The outlines of the theory can now be sketched briefly. The primary
causal sequence in political violence is first the development
of discontent, second the politicization of that discontent, and
23 Alex Inkeles, "Personality and Social Structure," Sociology Today (New York:
Basic Books, 1959), 272, quoted in Marvin E. Wolfgang and Franco Ferracuti,
The Subculture of Violence: Towards an Integrated Theory in Criminology (London:
Social Science Paperbacks, 1967), 8.
Explanations of Political Violence  ✻  13
finally its actualization in violent action against political objects
and actors. Discontent arising from the perception of relative deprivation
is the basic, instigating condition for participants in collective
violence. The linked concepts of discontent and deprivation
comprise most of the psychological states implicit or explicit in
such theoretical notions about the causes of violence as frustration,
alienation, drive and goal conflicts, exigency, and strain (discussed
in chapter 2).
Relative deprivation is defined as a perceived discrepancy between
men's value expectations and their value capabilities. Value
expectations are the goods and conditions of life to which people
believe they are rightfully entitled. Value capabilities are the
goods and conditions they think they are capable of attaining or
maintaining, given the social means available to them. Societal
conditions that increase the average level or intensity of expectations
without increasing capabilities increase the intensity of discontent.
Among the general conditions that have such effects are
the value gains of other groups and the promise of new opportunities
(chapter 4). Societal conditions that decrease men's average
value position without decreasing their value expectations similarly
increase deprivation, hence the intensity of discontent. The inflexibility
of value stocks in a society, short-term deterioration in a
group's conditions of life, and limitations of its structural opportunities
have such effects (chapter 5).
Deprivation-induced discontent is a general spur to action.
Psychological theory and group conflict theory both suggest that
the greater the intensity of discontent, the more likely is violence.
The specificity of this impulse to action is determined by men's
beliefs about the sources of deprivation, and about the normative
and utilitarian justifiability of violent action directed at the agents
responsible for it.
Societal variables that affect the focusing of discontent on political
objects include the extent of cultural and subcultural sanctions
for overt aggression, the extent and degree of success of past political
violence, the articulation and dissemination of symbolic appeals
justifying violence, the legitimacy of the political system, and the
kinds of responses it makes and has made to relative deprivation
(chapters 6 and 7). The belief that violence has utility in obtaining
scarce values can be an independent source of political violence,
but within political communities it is most likely to provide a
secondary, rationalizing, rather than primary, motivation. Widespread
discontent provides a general impetus to collective violence.
However, the great majority of acts of collective violence
in recent decades have had at least some political objects, and
14  ✻  Chapter One
the more intense those violent acts are, the more likely they are to
be focused primarily or exclusively on the political system. Intense
discontent is quite likely to be politicized; the primary
effect of normative and utilitarian attitudes toward violence is to
focus that potential.
The magnitude of political violence in a system, and the forms
it takes, are partly determined by the scope and intensity of
politicized discontent. Politicized discontent is a necessary condition
for the resort to violence in politics. But however intense
and focused the impetus to violence is, its actualization is strongly
influenced by the patterns of coercive control and institutional
support in the political community. Political violence is of greatest
magnitude, and most likely to take the form of internal war, if
regimes and those who oppose them exercise approximately equal
degrees of coercive control, and command similar and relatively
high degrees of institutional support in the society. The coercive
capacities of a regime and the uses to which they are put are crucial
variables, affecting the forms and extent of political violence in both
the short and long run. There is much evidence, some ofit summarized
in chapters 8 and 10, that some patterns ofregime coercive control
increase rather than decrease the intensity of discontent, and
can facilitate the transformation of turmoil into full-scale revolutionary
movements. Dissidents, by contrast, use whatever degree ofcoercive
capacities they acquire principally for group defense and for
assaults on the regime. The degree of institutional support for dissidents
and for regimes is a function of the relative proportions
of a nation's population their organizations mobilize, the complexity
and cohesiveness of those organizations, their resources,
and the extent to which they provide regularized procedures for
value attainment, conflict resolution, and channeling hostility
(chapter 9). The growth of dissident organization may in the short
run facilitate political violence, but it also is likely to provide the
discontented with many of the means to alleviate deprivation in
the long run, thus minimizing violence.
The preceding three paragraphs are an outline of the framework
in which the hypotheses and definitions of this study are developed,
and a summary of some of its generalizations. The hypotheses
and their interrelationships are summarized more fully and systematically
in chapter 10. The Appendix lists all hypotheses developed,
categorized according to their dependent variables, and
the chapters in which they are proposed.
The three stages in the process of political violence- those in
which discontent is generated, politicized, and actualized in political
violence- are each dependent on the preceding one, as the
Explanations of Political Violence  ✻  15
outline indicates. It is likely but not necessarily the case that
there is a temporal relationship among the three stages, whereby
a sharp increase in the intensity of discontent precedes the articulation
of doctrines that justify politically violent action, with shifts
in the balances of coercive control and institutional adherence occuring
subsequently. The conditions can be simultaneously operative,
however, as the outbreak of the Vendee counterrevolution in
1793 demonstrates: implementation of procedures for military
conscription intensified the discontent of workers and peasants
already sharply hostile to the bourgeoisie and the government it
ruled. Mass action against the bourgeoisie began in a matter of
days; the social context for dissident action was provided in part
by preexisting communal and political organization, action that was
facilitated by the concurrent weakness of government forces and
institutions in the region.24
The point is that many of the attitudes
and societal conditions that facilitate political violence may be
present and relatively unchanging in a society over a long period;
they become relevant to or operative in the genesis of violence
only when relative deprivation increases in scope and intensity.
Intense politicized discontent also can be widespread and persistent
over a long period without overt manifestation because a
regime monopolizes coercive control and institutional support.
A weakening of regime control or the development of dissident
organization in such situations is highly likely to lead to massive
violence, as it did in Hungary in 1956 and China in 1966-68, and
as is likely at some future date in South Africa.
The concepts, hypotheses, and models of causes and processes
developed in the following chapters are not intended as ends in
themselves. Intellectually pleasing filters through which to view
and categorize the phenomena of a disorderly world are not knowledge.
Systematic knowledge requires us to propose and test and
reformulate and retest statements about how and why things happen.
We know enough, and know it well enough, only when we can
say with some certitude not just why things happened yesterday,
but how our actions today will affect what happens tomorrow,
something we can always hope to know better, though never perfectly.
This analysis may demonstrate that too little is known about
the violence men do one another, and that it is known too weakly
and imprecisely. It is designed to facilitate the processes by which
that knowledge can be increased.
24
Charles Tilly, The Vendee (Cambridge: Harvard University Press, 1964),
passim.
16  ✻  Chapter One
The Nature of Social Theory
(In this section I wish to comment on some criteria for social
theory and their application to political violence. These criteria
are not essential for understanding the substance of the book, but
they help explain its approach to analysis andform ofpresentation.
Readers not concerned with these questions may turn directly to
chapter 2.)
Many descriptive generalizations and propositions can be made
about the origins, processes, and effects of political violence generally,
and of its several forms. Variables commonly examined include
the kinds of societal and political structures most susceptible
to violence, revolutionary ideologies, the developmental sequence
of revolutionary movements, the effects of force on the outcome of
violence, the nature of revolutionary organization, and much else.
Some less common variables for analysis are the nonideological
motivations of revolutionaries, the effects of institutional patterns
on the forms and objects of violence, the efficacy of different kinds
of governmental response, and the long-range consequences of
political violence.
Two paradigmatic approaches to "theory" about political violence
can be identified. One is to analyze a related phenomenon,
like revolutionary organization or ideology, in order to generalize
about its origins, characteristics, or effects. An interest in explaining
the occurrence or outcome of political violence is often implicit
or explicit in such an analysis, but typically the analysis does not
pretend to be complete; the relevance of other variables ordinarily
is recognized. The second is to choose a particular violent event or
class of events as an object of explanation and to specify some more
or less exhaustive set of conditions or variables that determines the
occurrence, extent, or outcome of the events. Theories of revolution,
like those of Edwards, Brinton, Pettee, and Timasheff, are
representative of this second approach.25 A variant of this approach
is to select a common, quantifiable property of a variety of events
for analysis and explanation (see pp. 8-9).
"Theory" is a term loosely used in the social sciences generally
and in the study of political violence specifically. The kind of
theory aimed at in this study is an interrelated set of general, falsi-
25 The three best-known American theories of revolution are Lyford P. Edwards,
The Natural History of Revolutions (Chicago: University of Chicago Press, 1927);
Crane Brinton, The Anatomy of Revolution (New York: Norton, 1938); and George
Pettee, The Process of Revolution (New York: Harper, 1938). A more recent theory
of this type is proposed by Nicholas S. Timasheff, War and Revolution (New York:
Sheed and Ward, 1965).
Explanations of Political Violence  ✻  17
fiable hypotheses that specify causal or concomitant relationships
between independent and dependent variables. The nature of
such relationships can be argued inductively from observations
about the relevant phenomena at various levels of analysis, or deduced
from other statements comprising the theory, or both.26
Such "explanatory" theory is distinct from what Eckstein calls
pretheoretical exercises, such as the formulation of classificatory
schemes and conceptual frameworks, which are sometimes called
theory.27 Nor are descriptive generalizations about relationships
among variables theory in the sense intended here, unless the
sources and nature of such relationships are stated explicitly and
the generalizations themselves interrelated.
Most hypotheses developed in this study are derived from the
juxtaposition and generalization of relationships observed in
studies of political violence and of the individual behaviors manifest
in it. The approach reflects an assumption that social science
theory ought to build on what is already known about the subject
being theorized about, and that it ought to be consistent with, or
at least not directly contradictory to, what is more generally known
about the nature and processes of individual and aggregate human
behavior. This is not to say that hypotheses must be consistent with
what Levy calls "going common sense," but rather that it is nonsensical
and inefficient to invest research effort in testing hypotheses
that contradict what is precisely known, unless there are compelling
logical or empirical grounds for questioning what is assumed
to be "precisely known."
The central scientific criterion for theory is that it be subject
to empirical assessment. Four attributes of theory that facilitate
its assessment are its falsifiability, definitional clarity, identification
of relevant variables at various levels of analysis, and applicability
to a large universe of events for analysis. The first two
are necessary conditions for assessment, the others desirable.
A fundamental limitation of most older theories and conceptions
26 For a critique of what passes for theory in the social sciences and a proposed
set of criteria for good theory, see Marion J. Levy, Jr., "'Does It Matter if He's
Naked?' Bawled the Child," in Klaus Knorr and James Rosenau, eds., Contending
Approaches to International Politics (Princeton: Princeton University Press,
1969). For a statement of criteria for theory in a field related to the subject of this
book, and a survey of extant "theory," see Wolfgang and Ferracuti, chap. 2.
- 27 Harry Eckstein, "Introduction: Toward the Theoretical Study of Internal
War," Internal War: Problems and Approaches (New York: The Free Press, 1964),
7-29, classifies pretheoretical operations as delimitation (statement of the boundaries
of the subject); classification and analysis (meaning by "analysis" the dissection
of a subject into its components); and problemation (the formulation of specific
problems for theory construction). Most of these "operations" with respect to
political violence are attempted in this chapter.
18  ✻  Chapter One
of revolution is the difficulty of deriving falsifiable hypotheses
from them.28 Few if any were formulated with reference to applicable
empirical methods. The fact that very few case or comparative
studies make use of them is further evidence of their limited
usefulness even for taxonomic or conceptual purposes. Two general
conventions regarding falsifiability are used in social science.
One is to state that an independent (causal) variable is a necessary
and/or sufficient condition for the dependent variable one wants
to explain. This usually means that both variables are defined in
dichotomous terms- "disequilibrium" does or does not occureven
if one or both are continuous variables, and typically leads
to propositions that are trivial or falsifiable by a single deviant case.
The second is to rely on tests of statistical significance of relationships
between variables, which has led to the proliferation of
weakly supported but not-quite-falsified hypotheses which have
yet to be integrated into a more general and parsimonious theory.
A third convention is used in this study: a minimum strength of
relationship is postulated for each hypothesis in correlational
terms; if the relationship found in an empirical test is weaker than
stipulated, and no serious sample or instrument error is likely,
the hypothesis is rejected. For example, one basic hypothesis
is that the greater the intensity and scope of relative deprivation
in a population, the greater the potential for collective violence.
The proposed relationship is a strong one: if interval-order measurement
of both variables for a substantial number of cases gives
a product-moment correlation coefficient less than .45 (less than
20 percent of the variance explained), the hypothesis is rejected.29
28 For example, Edwards, Brinton, and Pettee. A sampling of the older theoretical
speculation about revolution and political violence, which seldom finds its
way even into the footnotes of current writings, would include the work of Gustave
Le Bon, especially The Psychology of Revolution, trans. Bernard Miall (London:
Unwin, 1913); Brooks Adam, The Theory of Social Revolutions (New York:
Macmillan, 1913); Everett Dean Martin, The Behavior of Crowds: A Psychological
Study (New York: Harper, 1920), chap. 7, "The Psychology of Revolutionary
Crowds"; Charles A. Ellwood, The Psychology of Human Society: An Introduction
to Sociological Theory (New York: Appleton, 1925), chap. 8, "Changes Within
the Group: Abnormal"; Robert Hunter, Revolution: Why, How, When? (New York:
Harper, 1940); and Mark A. May, A Social Psychology of War and Peace (New
Haven: Yale University Press, 1943), chap. 7, "Aggressive Social Movements."
The flavor of the literature to 1925 can be sampled in Dale Yoder, "Current Definitions
of Revolution," American journal of Sociology, XXXII (November 1926),
433-441. Some writings of this period are curious indeed, for example, Sidney A.
Reeve, The Natural Laws of Social Convulsions (New York: Dutton, 1933).
29 The same strength of relationship is required of multiple regression coefficients
(R) for studies in which multiple measures of a particular independent variable are
used. I have reported evidence to the effect that the .45 threshold is by no means
too stringent for the hypotheses mentioned in Gurr, "A Causal Model of Civil
Strife." Three composite measures of deprivation related to magnitude of civil
strife for 114 polities give an R of .60, despite the presence of considerable measurement
error.
Explanations of Political Violence  ✻  19
Other hypotheses specify relationships of moderate strength,
requiring a minimum r of .30 if they are not to be discarded.
Hypotheses about independent variables that do not meet these
criteria are not necessarily false in the conventional sense, but they
denote relationships that are too weak to contribute to parsimonious
theory.30
Independent and dependent variables must be defined with
sufficient precision so that a researcher can determine what particular
conditions or set of events constitute "X" for purposes of
empirical assessment. The definitional inadequacy of many theories,
new and old, contributes to the difficulty of assessing them.
For example, a great many conditions have been said to "cause"
or to constitute a basic potential for collective violence, among
them general properties of sociopolitical systems labeled "cramp,"
"disequilibrium," and "strain"; motivational characteristics of
violent men such as "frustration" and "discontent"; and particular
institutional patterns such as repressive government and inadequate
circulation of, or divisions within, an elite.31 A difficulty of
explanations citing such general properties is that the conditions
as defined, if indeed they are defined, can usually be found in
most societies and among most men, whether or not revolutionary;
and usually they are not formulated precisely enough to permit
one to assess the effects of their various elements and degrees.
The empirical referents of some concepts are more easily isolated
than those of others. The concept "participation in clandestine
organization" is more readily made operational than "revolutionary
mobilization," and revolutionary mobilization is a more malleable
concept than "power deflation." This is not to argue that
more general concepts be avoided, but that the more general they
are the more necessary it is to define them fully and to catalog their
manifestations sothat they can be evaluated empirically.
It is desirable that theoretical statements be testable at various
levels of generalization. Feldman, for example, attributes revolutionary
potential to the increasing salience of goal conflicts between
the increasingly numerous subsystems that are said to result from
30 These two criteria imply an absolute maximum of five strong variables or
eleven moderately strong variables for any theoretical system. Hypotheses that
are not supported at the .30 level but that nonetheless meet statistical tests of
significance should lead to postulation of more general hypotheses that might
account for a set of such weak relationships.
31 See Pettee, passim; Chalmers Johnson, Revolutionary Change (Boston: Little,
Brown, 1966), especially chap. 4; Neil J. Smelser, Theory of Collective Behavior
(New York: The Free Press, 1963), passim; Davies, passim; and Ronald G. Ridker,
"Discontent and Economic Growth," Economic Development and Cultural Change,
XI (October 1962), 1-15. A categorization of specific and general causal factors cited
in the literature is given in Eckstein, "On the Etiology of Internal Wars," 143-
144.
20  ✻  Chapter One
structural differentiation in a society.32 Ideally it should be possible
to observe manifestations of such a process at both the macrolevel
of an entire social system, for example eighteenth-century
France, and at the level of the community and among small, faceto-face
groups.33 Small-group phenomena are not likely to resemble
simply a scaled-down model of macrophenomena, but it should
be possible to identify the small-group interactions and individual
behaviors that are comprised by macroevents. If this translatability
cannot be achieved directly, then at least the macrophenomena
postulated in theory should be compatible with what is
known about microphenomena. This is not a reductionist argument
that analysis of social systems or collective behavior can or
should be reduced to analysis of component individual behaviors.
The point is that analysis on one level can and should inform the
other, and that hypotheses whose relationships are manifested
at, and subject to examination at, different levels of analysis are
usually more interesting and fruitful than those which refer to
one level of analysis only.34
The fourth attribute of theory that facilitates its assessment concerns
the types and numbers of cases or settings in which the proposed
relationships can be examined. The preference is for theory
that can be subject to test both in case studies and in large-sample
comparative studies using statistical methods. Case studies are
useful for elucidating the fine structure of revolutionary events
and for providing a sense of understanding of how general variables
act and interact. They also can be used to test theoretical
statements expressed in dichotomous terms: one can define "accelerators
of dysfunction" and "revolution" independently and
in sufficiently rigorous fashion that a single case study may be sufficent
to falsify the proposition that accebrators of dysfunction are
a necessary precondition of revolution.35 But many variables of
interest to social theory can be dichotomized only at great loss of
information, and scholars undertaking case studies often find it
difficult to distinguish between general relationships and the
32 Arnold Feldman, "Violence and Volatility: The Likelihood of Revolution,"
in Internal War, lll-129.
33 Sidney Verba, Small Groups and Political Behavior: A Study of Leadership
(Princeton: Princeton University Press, 1961), demonstrates the usefulness of relating
concepts and findings at one level of analysis to those at another. Tilly's
case study of an internal war, The Vendee, shows how analysis can proceed from
macrolevei concepts such as urbanization and social change to the individual
level, and back again.
34 On the relative merits of using different levels of analysis in social inquiry
see David Singer, "The Level of Analysis Problem in International Relations,"
World Politics, XIV (October 1961), 77-92.
35 This proposition is adapted from Johnson, op. cit.
Explanations of Political Violence  ✻  21
unique historical and cultural circumstances of each case. Hypotheses
that specify systematic relations among continuous
variables, or between a continuous and a dichotomous variable,
are subject to scientific assessment only if substantial numbers of
cases are examined. This provides further justification for the
choice of political violence rather than revolution as the subject
of this study: the former is far more common than the latter.
In this study hypothetical relationships are formally stated
and given alphanumeric descriptors. The first term in each hypothesis
is its dependent variable. Definitions ordinarily are proposed
in the paragraphs immediately following the first appearance of
a theoretical term in a hypothesis. Underlining of a term in this
context indicates that the sentence in which it appears is a formal
definition. Synonyms are used for several of the frequently cited
theoretical concepts to avoid the mind-deadening effects of repetition.
"Relative deprivation," defined as a perceived discrepancy
between men's value expectations and value capabilities, is represented
by its initials, RD, and the synonyms deprivation, discrepancy,
and, loosely, frustration. "Discontent," the psychological
state said to be caused by RD, has as its synonyms anger, rage,
and dissatisfaction. Synonyms for "value expectations" are wants,
aspirations, and expectations. "Values" are the goods and conditions
of life which men seek; the phrase "goods and conditions of
life" is used synonymously with values. "Value opportunities,"
the courses of action people have available to them for attaining or
maintaining their desired values, are more simply referred to as
"means." These synonyms are used only when there is no ambiguity
about their referent; the precisely defined terms are used
in the development of the basic theoretical arguments.