International Studies Review (2004) 6, 107–133

Pacifism and Fightaholism in

International Politics: A Structural History of
National and Dyadic Conflict, 1816–1992
ZEEV MAOZ
University of California, Davis

Can we put labels on states due to their history of conflict involvement? Popular
folklore as well as the rhetoric of politicians suggests that we can. Germany up to the
end of World War II and Japan in the same period were labeled ‘‘revisionist’’ or
‘‘aggressive’’ states. President Reagan called the Soviet Union ‘‘the Evil Empire,’’ due
to its seemingly expansionist ideology, but also due to its presumably aggressive
behavior. Israel is often depicted by many of its neighbors and other countries in and
outside the Middle East as ‘‘inherently expansionist.’’ These examples suggest a no-
tion that states can somehow be structurally characterized, independently of specific
policies, leaders, political parties or regimes in power, economic and social conditions.
If we can label states in structural terms, we can also label pairs or groups of
states. For example, President George W. Bush branded North Korea, Iran, and
Iraq as the ‘‘Axis of Evil,’’ due to these countries’ pursuit of weapons of mass
destruction. The Clinton administration identified Syria, Iran, and Iraq as a de-
stabilizing axis in the Middle East, confronting the latter two through a policy of
dual containment. The scholarly literature on international politics has identified
structural patterns of warring or conflicting dyads through such concepts as pro-
tracted conflict, intractable conflicts, andFmore analytically definedFthe concept
of enduring rivalries (Diehl and Goertz 2000; Maoz and Mor 2002).
How scientifically sound are such labels? More importantly, are such labels help-
ful in understanding the causes, courses, and consequences of international con-
flicts? In other areas of human and social inquiry, structural characterization of
units is of immense importance. Genetic research clearly indicates that certain
people are far more prone to some diseases than others. Research on addiction
attempts to identify structural propensities of drug or alcohol abuse. Research on
recidivism in criminology is intent on identifying structural propensities for crime.
Research on poverty systematically identifies structural characteristics including
individual, family, and even national correlates of poverty.
This study is motivated by the following empirical observation about international
conflict: the distribution of national and dyadic conflict involvement during the last two
centuries reveals extreme inequalities. A substantial number of states have engaged in
little or no conflict with other states, while a small group of states has participated in a
disproportionately high fraction of all conflicts. Likewise, a substantial number of po-
litically relevant dyadsFdyads that are expected by virtue of their geographic prox-
imity or span of strategic interests to be highly conflict proneFturn out to have little or
no conflict experience over their joint history. On the other hand, a handful of dyads
are responsible for most of the conflict activity in the international system.
This observation runs contrary to both explicit and implicit notions about inter-
national conflict in the literature. Studies influenced by realist conceptions assert that
conflict is an endemic feature of international anarchy. Hence, it follows that every
stateFif it survives long enoughFis bound to get involved in militarized interstate
r 2004 International Studies Review.
Published by Blackwell Publishing, 350 Main Street, Malden, MA 02148, USA, and 9600 Garsington Road, Oxford OX4 2DQ, UK.
108 Pacifism and Fightaholism in International Politics

disputes and wars. Likewise, if two states have the opportunity (Most and Starr 1989)
by virtue of their geographical or strategic contact and if both share a sufficiently long
stretch of history, they are bound to fight each other at some point (Waltz 1979:113).
The assumption that there is basic equality (or ‘‘normalcy’’) in conflict-involve-
ment patterns is implicit in most of the quantitative analyses of international con-
flict, which assumes a normal, log-normal, or Poisson distribution of the dependent
variable. Nathaniel Beck, Gary King, and Langche Zeng (2000) point out that these
assumptions are tenuous and that significant bias in the findings may result from
relying upon them.
The fundamental inequality in national and dyadic patterns of conflict involve-
ment may require us to redefine our approach to the study of conflict. Instead of
examining why states fight in general, we need to explore what makes some states
fundamentally pacifist in their international relations while others tend to be sub-
stantially conflict prone. Likewise, we need to investigate why some dyads fight
repeatedly while others do not fight at all.
Perhaps one of the most important implications of an ‘‘unnatural’’ distribution of
conflicts over states and dyads is that we must deal with structural characteristics of
specific groups of states rather than treat all states alike. For example, if there are
identifiable groups of states or dyads that are pacifist, it would be improper to ex-
amine them in the same terms that we study conflict-prone states or dyads. However,
identifying some states (or some dyads) as pacifist and other states (or dyads) as
conflict prone does not mean much unless we can systematically differentiate them in
terms of other properties that are related to their conflict proneness or pacifism.
Accordingly, the underlying aims of the present study are fourfold:
1. To explore long-term patterns of national and dyadic conflict involvement.
2. To classify states and dyads into distinct ‘‘risk’’ groups in terms of their
dispute and war-involvement patterns.
3. To identify the principal characteristics of states and dyads making up each
of the risk groups in terms of their conflict involvement.
4. To identify some preliminary correlates of the structural conflict propen-
sity of states and dyads.

This study focuses on the conceptualization and identification of structural pat-

terns of conflict, not on a theory thereof. In subsequent studies I will explore the
factors that account for the location of states or dyads in each of these risk groups,
or their move across the various groups over time.
The present study is designed as follows. Section two describes the distribution of
militarized interstate disputes and wars over the entire history of nations and dyads
and discusses the implications of these patterns. Section three defines the concepts of
pacifism, conflict proneness, and conflict-related addiction (or fightaholism) in the
context of other types of compulsive-obsessive behaviors, such as substance abuse,
recidivism, and behavioral addictions. Section four offers an empirical description of
these structural patterns of conflict and some of the typical correlates of these risk
groups. Section five discusses the implications of these issues for theories of inter-
national politics and conflict theory. (The research design is given in the appendix.)

The Inequality of Conflict: National and Dyadic Patterns

Some Observations about Units of Analysis in Conflict Research
Many studies of conflict employ the unit-year as their principal unit of observation.
The typical unit of observation is the nation-year (e.g., Maoz 1993, 2001),1 or the

1
Other investigators (e.g., Mansfield and Snyder 1995, 1996, 1997; Henderson 1998) hsve used the nation-
decade or the nation-half-decade as their unit of observation.
 ZEEV MAOZ 109

dyad-year (e.g., Bremer 1992, 1993; Maoz and Russett 1993; Thompson and
Tucker 1997; Beck, Katz, and Tucker 1998; Russett and Oneal 2001). The unit-year
scheme is designed to deal with the questions of ‘‘who gets involved in conflict
when’’ or ‘‘who fights whom when.’’ This approach has yielded important insights
(e.g., Bremer 2000). However, this scheme prevents detection of structural patterns
of conflict that can be conceptualized only within a temporal framework covering
the entire history of the unit. Testing for structural patterns of national and dyadic
conflict involvement requires long-term observation.
Studies on conflict-related structural properties of states or dyads are quite rare.
Quincy Wright (1944) briefly discussed the war-proneness of states. J. David Singer
and Melvin Small (1972; see also Small and Singer 1982), Charles Gochman and
Zeev Maoz (1984), and Maoz (1993) provide descriptive statistics of national war
and dispute proneness. More recently, the enduring rivalry literature (Goertz and
Diehl 1992, 1993; Diehl and Goertz 2000; Maoz and Mor 2002) has provided
structural information on the more dispute-prone dyads. Other studies have at-
tempted to identify potentially addictive patterns (e.g., Bremer 1980; Most and
Starr 1980). However, we still lack a good understanding of these tendencies as well
as of some of their causes and correlates. Thus, a study of these issues appears
timely.

National Conflict Involvement Patterns

The extent of inequality in the conflict involvement of states and dyads is analogous
to income inequality in societies. If all states were equally conflict prone, then the
least conflict-prone states would account for a similar proportion of systemic conflict
as the most conflict-prone states. If the distribution of conflicts over states were
normal, the cumulative distribution would provide us some approximation to the
line of perfect equality. Figure 1 presents empirical patterns of national conflict
involvement over the 1816–1992 period.
This figure suggests a fundamental inequality in the structure of national conflict
patterns: a small number of states accounts for a disproportionate amount of con-
flict in the system while a large number of states evidence very little conflict.2 Nearly
11 percent of all states with twenty or more years of independence were not in-
volved in any Militarized Interstate Dispute (MID) at all. Nearly 57 percent of all
states were not involved in any war. At the other end of the scale, the top 10 percent
of the states accounted for 49.7 percent of all MIDs, and for 56.8 percent of all wars
in the system.3

Dyadic Conflict Involvement Patterns

Inferences from the analyses conducted at the national level may be highly mis-
leading. States that are peace-prone are benign with respect to all other states; if
one avoids fights, one avoids fights with everybody equally. However, at the dyadic
level, states may not distribute their conflict activity uniformly over partners. If each
of two states is conflict prone, it does not follow that the dyad made up of these two
states would also be conflict prone. If state A fights with states C, D, E, F . . . and state
B fights states Q, R, S, T . . . , then both states may be individually conflict prone
while the dyad itself is pacific. Thus, the distribution of conflicts over dyads may

2
Normalized figures for the distribution control for the length of a state’s national history. The unnormalized
distributions of conflicts yield huge inequalities with Gini coefficients of 0.74 for MIDs and 0.85 for Wars. Similar
analyses were conducted for shorter stretches of time such as half-decades with nearly identical results. See ap-
pendix for details.
3
Matching the distributions of national MID and war involvement patterns with a Poisson distribution yields
extremely high chi-square values (for MIDs, the chi-square is over 63,000 with 9 DF and for war the chi-square is
889), suggesting strong evidence for addiction.
110 Pacifism and Fightaholism in International Politics

100%
Cumulative Normal (Poisson)
Distribution
80%
Cumulative Pct. Conflict

60%
Involvement

Gini(MIDs) = 0.59
40%
Line of Perfect Equality
20%
Gini(wars) = 0.81
0%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
−20%
Cumulative Pct. of States

MID/Year War/Year Equality Normal Dist.

FIG. 1. Lorenz Curve of National Conflict Involvement, 1816–1992 (normalized by length of national
history).

be significantly different from the distribution of conflicts over individual states.

Figure 2 shows the distribution of MIDs and wars over politically relevant dyads.
This figure suggests that states tend to be extremely ‘‘selective’’ in their choices of
enemies. A majority of all politically relevant dyads (57 percent) never have ex-
ercised threats, displays, or uses of force. Almost 87 percent of all politically relevant
dyads never fought a war. On the other hand, the upper centile of the dyads
accounted for 34 percent of all MIDs and for over 95 percent of all interstate wars for
this population.

Implications
So what if the national and dyadic distributions of conflict are not normally or
Poisson distributed? Many social phenomena are not normally distributed. The use
of international flights, the amount of money invested in stock exchanges by people
or firms, the ownership of golf clubs, and many other trivial patterns of behavior
display highly similar distributions to the ones discussed above. The fact that the
distribution of national and dyadic patterns of conflict involvement is highly un-
equal may be neither surprising nor of significant import.
100%
Cumulative Normal (Poisson)
90%
Distribution
80%
Cumulative Pct. of Conflict

70%
60% Gini(Wars) = 0.934
50%
40%
Gini(MIDs) = 0.819
30% Line of Perfect Equality
20%
10%
0%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
−10%
−20%
Cumulative Pct. of Dyads

Dyadic MID/Yr Dyadic War/Yr Perfect Equality Normal Dist.

FIG. 2. Lorenz Curve of Dyadic Conflict Involvement, 1816–1992 (normalized by year of joint dyadic
history).
 ZEEV MAOZ 111

While it is true that a great deal of trivial social and natural phenomena exhibit
substantial inequalities among units, it is also true of other, possibly less trivial,
patterns of deviant social behavior, such as the use of drugs and alcohol, the in-
volvement in criminal behavior, overeating, and obsessive overuse of certain habits
during leisure time (e.g., computers, television). One of the potential implications
of such skewed distributions of deviant behavior is that they are harmful not only to
the person engaged in them, but also to his/her environment.
The significance of the structural distributions of conflict involvement lies prin-
cipally in the following argument: our treatment of all states or dyads as if they were
a priori equally conflict prone is tenuous and making this assumptionFimplicitly
or explicitlyFin our studies may yield biased inferences. More generally, if we view
conflict involvement as self- and environmentally harmful, then the fundamental
inequality in the distribution of conflict carries important implications for the study
and prevention of conflict. Since conflict is a deviant behavior, the fact that a sub-
stantial proportion of all states and dyads systematically abstains from conflict sug-
gests a substantial lacuna in our knowledge. Specifically, we have a substantial
phenomenon of pacifism in world politics, but we do not know what the characteristics of
pacifist states and pacifist dyads are. The classification of states and dyads into low-
and high-conflict risk-groups may enrich our understanding both of the causes and
the consequences of conflict behavior.

Pacifism, Conflict Proneness, and Fightaholism: A Conceptual Exploration

I now turn to a discussion of three basic concepts that guide this study. These
concepts are interrelated, but each possesses distinctive features. The concepts of
pacifism and conflict proneness are self-explanatory. The concept of conflict-related
addiction or fightaholism is used as a metaphor for a structural pattern of behavior
not tapped by the other concepts. I discuss these concepts in this section.
Pacifism is an absolute concept. It refers to the absolute lack of conflict involvement
given sufficient opportunity. At the national level, an opportunity for conflict exists if a
state has a minimal number of immediate neighbors and if it has existed as an
independent system member for a sufficient amount of time. At the dyadic level,
opportunity is defined by the notion of political relevance (Maoz and Russett 1993;
Maoz 1996) and by a minimum amount of time of common history, that is, by a
sufficient number of years of joint national existence.4 If a state was involved in a
single dispute or war over its entire history, it can no longer be branded as pacifist.
A national commitment to avoid conflict is a necessary but not a sufficient condition
for pacifism. Other states must also avoid conflict with the focal state. Switzerland,
Holland, and Belgium wanted to avoid war in World War I and World War II, thus
declaring neutrality. Yet Germany respected the neutrality of the former, but not of the
latter two states. Dyadic pacifism is perhaps more meaningful. A pacifist dyad is one
wherein both states systematically avoid militarized conflicts with one another. I dis-
tinguish between dispute pacifism and war pacifism. A dispute pacifist is a state/dyad
that has experienced no militarized disputes during its entire history. A war pacifist is a
state/dyad that has never engaged in an interstate war during its entire history.
Conflict proneness refers to the number of conflict involvements per year of na-
tional or dyadic existence. This concept distinguishes between states that have been
involved in a large number of conflicts and those that have been involved in few
conflicts. Dividing by the number of years of existence normalizes for the temporal
opportunity for conflict. By definition, the conflict involvement score of dispute or
war pacifists is zero.

4
The notion of ‘‘opportunity’’ in terms of geographical proximity may be problematic. Nevertheless, it is fre-
quently used in the literature (e.g., Most and Starr 1989; Holsti 1991; Vasquez 1993).
112 Pacifism and Fightaholism in International Politics

The concept conflict-related addiction (fightaholism) reflects the evolution of conflict

involvement over time. Two states may have identical conflict-proneness scores but
their patterns of involvement may take on dramatically different forms over time.
For example, only about 18 percent of US militarized interstate disputes and 10.7
percent of its wars were fought in the nineteenth century. On the other hand, 59
percent of Mexico’s MIDs and all of its wars were in the nineteenth century.
Moreover, all except one of the 25 MIDs between the US and Mexico took place
before 1920. More generally, one state may have distributed its conflicts uniformly
over time, whereas another may have experienced a short and extremely intense
period of conflict while the rest of its history may have been relatively pacific.
To capture conflict proneness over distinct phases of states’ or dyads’ histories, I
have developed the notion of conflict-related addiction, or fightaholism. As noted, this
concept is based on a metaphor derived from other types of deviant behaviors.
Clearly, the metaphoric representation of any concept is necessarily imperfect. I do
not mean to suggest that certain states or dyads are addicted to conflict in the exact
same manner that some people are addicted to drugs, alcohol, or tobacco. Nor do I
mean to imply that conflict creates chemical reactions and feelings of craving and
psychological dependence as do certain substances. Rather, the behavioral impli-
cations of addiction or other kinds of obsessive-compulsive disorders are similar or
analogous to the behavioral patterns of conflict involvement of some states and
dyads over time.
In order to explore this metaphor, we must define different types of obsessive-
compulsive disorders and examine how they are diagnosed. These disorders
include various types of substance (drug, alcohol, tobacco) abuse that involve
chemical as well as psychological dependence. But they also include food addiction,
sexual addiction, compulsory gambling, workaholism, and addiction to television or
computers, as well as obsessive criminal behavior. The latter denote behaviors that
are based primarily on psychological dependence. Clearly, these types of addiction
are extremely diverse. However, they share a number of traits that are included in
the general definition of obsessive-compulsive behavior that places most of them
under the label of addiction.

Addiction is a dependence, on a behavior or substance, that a person is powerless

to stop . . . Addiction has been extended, however, to include mood-altering be-
haviors or activities. Some researchers speak of two types of addictions: substance
addictions (for example, alcoholism, drug abuse, and smoking); and process ad-
dictions (for example, gambling, spending, shopping, eating, and sexual activity).
There is a growing recognition that many addicts, such as polydrug abusers, are
addicted to more than one substance or process. (Fray 2001:50)

It is commonly noted that ‘‘the main characteristic of addictive behavior is the

compelling need to engage in a particular activity or to use the addictive substance
without deriving pleasure of it, and despite awareness of the damage caused by this activity to
oneself and to his or her family, friends, or co-workers’’ (Larson 1996:1129, emphasis added).
The elements of this general definition are visible in definitions of specific types
of addiction. Substance addiction is defined as excessive use of a certain substance
over a relatively long period of time. What excessive is depends on the nature of the
substance. With regard to some substances, for example, alcohol, nicotine, or cer-
tain ‘‘softer’’ drugs, addiction severity is a function of three factors: social norms,
the extent of deviation from the norm, and the period of use.5

5
The measurement of addiction in relation to a certain ‘‘norm’’ is very problematic. First, the ‘‘norm’’ itself might
vary across socieities. For example, the ‘‘norm’’ of alcohol use in Russia, France, and Italy is different from Iran and
Saudi Arabia. (See the figures on the cross-national differences in alcohol use in Clarke and Weisburd 1990:11.)
Second, the norm may change over time. This, for example, was the case with nicotine and drug abuse in most
Western societiess over the last three decades (Gottfredson and Hirschi 1990:40).
 ZEEV MAOZ 113

These definitions of addictive behavior identify five common traits that can be
compared to the notion of fightaholism discussed below:

1. Excessive and repetitive engagement in a given activity.

2. The activity may cause some sort of immediate gratification, but it pos-
sesses some elements that are harmful to oneself as well as to one’s en-
vironment.
3. Awareness that this behavior is self-damaging does not inhibit continuous
engagement in it.
4. Addiction is a self-reinforcing behavior. It may be induced by
chemical craving, by psychological dispositions, or even by rational rea-
soning.6
5. Efforts to discontinue this activity or to reduce the level of engagement in it
are brief and unsuccessful.

Addictive behavior is typically deviant. The number of people diagnosed as drug,

alcohol, nicotine, or food addicts or as compulsive gamblers (and even recidivists) is
generally small in the entire population, but their rates of consumption of sub-
stances or practice of the activities is disproportionately high. Thus, the study of
addiction focuses on a small number of individuals rather than on general char-
acteristics of a population due to the unequal distribution of deviance in the general
population.
Analogously, fightaholism is defined as excessive and repetitive engagement in milita-
rized disputes and wars over a long period of time with occasionally harmful consequences. At
the dyadic level, we define dyadic fightaholism as repeated and excessive engagement in
militarized disputes and war with the same partner over a long period of time, again with
harmful consequences that accrue to both members of the dyad. This behavior is accompanied
by policies that facilitate the conduct of conflict, such as militarization or alliance formation.
The definitions of these concepts allow us to classify states or dyads into four risk
groups regarding their conflict-related propensity:

1. Pacifists. Those states or dyads that did not engage in any conflict
throughout their history.
2. Normal. States or dyads that have engaged in relatively little sporadic
conflict during their history and whose rate of conflict involvement over
time has been sporadic.7
3. Conflict prone. States or dyads that have engaged in relatively high
amounts of conflict during their history but whose conflict history is spo-
radic rather than sustained.
4. Fightaholics. States or dyads that have engaged in excessively high
amounts of conflict in a sustained manner over their history.

The first three groups are mutually exclusive. The final group largelyFbut not
completelyFoverlaps with the conflict-prone group. All fightaholics are conflict
prone, but not all conflict-prone states or dyads are fightaholics. In the next section,
we discuss the empirical properties of these four groups.

6
Some theories of gambling and recedivism facilitate the application of rational decision or game theoretic
concepts to such processes. For example, sunk cost paradoxes in the Dollar Auction game and cost benefit cal-
culations in criminal behavior (where the probability of capture is discounted in the planning of later crimes than in
early ones due to experience and learning) suggest that even people who perform rational-like calculations can
become addicted to self-damaging behavior (see Maltz 1984 on recidivism, Brockner and Rubin 1985 on social
entrapment in a wide array of situations, and Maoz 1990 on sunk-cost paradoxes in war).
7
The notion of ‘‘normal dyads’’ is equivalent to what Diehl and Goertz (2000) and Maoz and Mor (2002) call
‘‘proto rivalries.’’
114 Pacifism and Fightaholism in International Politics

Identifying Pacifism, Conflict Proneness, and Fightaholism

General Empirical Properties of Structural Conflict Behavior
Pacifism. As noted above, the empirical identification of pacifism is relatively
straightforward. However, the probability of national and dyadic pacifism is re-
lated to the length of a state’s/dyad’s history; the longer the existence of states/
dyads as independent system members, the less likely they are to stay pacifist.
Figure 3 provides some sense of the distribution of national and dyadic pacifism
patterns by years of independence.
Figure 3 suggests that about 9 percent of all states with at least 20 to 50 years of
system membership were MID pacifist. However, none of the states with 50 or
more years of system membership absolutely avoided MID involvement. If states
live long enough, they end up getting involved in MIDs of one form or another.
War-related pacifism, however, is observed at some level even for states with
varying lengths of national histories. More than 55 percent of the states with
histories extending for at least 40 years completely abstained from war. For
‘‘older’’ states, this proportion levels off at an average of 15 percent regardless of
the length of national history.
About 58 percent of the politically relevant dyads with 20 or more years of joint
history are pacifists. This proportion drops to a rate of 32 percent MID-pacifism
for dyads with 50 to 80 years of joint history. It levels off at an average rate of 20
percent for dyads with 80 or more years of joint history. More significantly,
regardless of the length of joint history, 71 percent of all politically relevant dyads never
fought each other in an interstate war.
Thus, in contrast to the realist notion that conflict is a pervasive phenomenon
in world politics, pacifism turns out to be a significant phenomenon in an an-
archic world, and it calls for explanation.
Conflict Proneness. On the national level, there is no discernable increase in
conflict proneness over time. While a regression of MID and war proneness on
length of history yielded statistically significant effects, the coefficients were fairly
low for both measures of conflict (b ¼ .004, R2 ¼ 0.056 for MID proneness, and
b ¼ .0008, R2 ¼ 0.026 for warsFN ¼ 180). This suggests only a small effect of
duration in the system on the increase in rate of conflict proneness for individual
states.

1
0.9
Proportion of Pacifist States

0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170
No. of Years in System

National MID Pacifism Dyadic MID Pacifism

National War Pacifism Dyadic War Pacifism
FIG. 3. National and Dyadic Pacifism: Proportion of all states with zero conflict involvement by
number of years of national independence.
 ZEEV MAOZ 115

TABLE 1. Cross-Time Conflict Levels

1.1 A Hypothetical Addiction Table

Current Period Level of Conflict Involvement

Level of Conflict involvement Low Medium High

Past Period Low Nonadicted Nonadicted Possibly Addicted

Medium Rehabilitated Possibly addicted Addicted
High Rehabilitated Possibly addicted Addicted

Note: In cells denoted as possibly addicted and addicted, observed frequencies must be higher than expected fre-
quencies. If that is the case in a given contingency table, the cell is designated as consistent. If the reverse is true, the
cell is designated as inconsistent. In cells denoted as rehabilitated, observed frequencies must be smaller than expected
frequencies to be designated as consistent.

1.2. Examples: National Level Cross TIME RELATIONS

Level of MID Involvement

in Present Decade
Low Medium High Row Total
þ
Level of MID involvement in previous decade Low 470 n
116 70 n
656
(356.95) (139.41) (159.64)
Medium 109 þ 72 þ 69 þ 250
(136.03) (53.13) (60.84)
High 56 þ 60 þ 145 þ 261
(142.02) (55.47) (63.52)
Column Total 635 248 284 1,167
w2(4) ¼ 260.244; po0.001; mb ¼ 0.425

Level of MID Involvement

in Present Decade
Low Medium High Row Total
þ
Level of MID Involvement Two Decades Ago Low 369 100
n
83 n
552
(294.48) (117.57) (139.96)
Medium 105 þ 50 þ 58 þ 213
(113.63) (43.37) (54.01)
High 56 þ 60 þ 109 þ 221
(117.90) (47.07) (56.03)
Column Total 526 210 250 986
w2(4) ¼ 136.539; po0.001; mb ¼ 0.544
Note: entries in parentheses are expected frequencies.
n
Inconsistent cell.
þ
Consistent cell.

A similar finding can be reported for the effect of the length of dyadic his-
tory on the conflict proneness of the dyad. In the case of MID-proneness, the
effect of the length of dyadic history is statistically significant, but the slope is
very low (b ¼ .0002, R2 ¼ 0.003). For wars this effect is not even statistically
significant.
Fightaholism. How do we know that a state or a dyad is fightaholic? In the
appendix, I develop a number of measures of conflict-related addiction that build
on those bodies of literature. Here, I explore the empirical properties of these
measures. A first-cut identification and analysis of conflict-related addiction
requires us to examine time-dependence at various levels of conflict involvement.
116 Pacifism and Fightaholism in International Politics

Such time dependence could be prespecified so that any relationship between

conflict-involvement levels at one period and conflict involvement at another
period would correspond to the characteristics of addictive behavior discussed
above. Table 1 shows how an interpretation of addictive behavior would appear
in a contingency table that relates these two periods.
In the top part of this table, we designate the cells that are consistent with the
addiction specifications. The presence of addiction in the actual population of
states and dyads is based on the extent to which the difference between the
observed and expected frequencies in the cells that are designated as possibly
addictive or addictive contributes to the overall chi-square of the frequency ta-
ble.8 The mb statistic denotes the proportion of the chi-square accounted for by
the addiction hypothesis (Maoz 1996: 130–132). In the example given in part 1.2
of the table, we see that in most of the cells suggesting patterns of addiction,
observed frequencies are significantly higher than expected frequencies and in
most cells suggesting rehabilitation, observed frequencies are substantially lower
than the expected ones, consistent with the hypothesis of addiction-like behavior.
Table 2 provides the cross-time correlations for both national and dyadic ad-
diction levels over a variety of measures.
The results of Table 2 suggest evidence of addiction-like patterns of warfare
over time. States and dyads that exhibit disproportionately high levels of conflict
involvement during one period tend to be high in conflict in subsequent periods.
This trend extends over relatively long time-spans, implying that a state that
started its national history as a highly conflict-prone state will maintain its status
of addition-like conflict involvement over time.9
The measures of fightaholism offered herein apply the five characteristics of
addictive behavior to the extent of a state’s involvement in MIDs and wars over
time. A state/dyad is designated as fightaholic if its involvement in MIDs or wars
was far above the national/dyadic average for a given period and if this excessive
involvement lasted over a number of periods consistently. Based on this defi-
nition we can discuss the relationship between conflict-proneness, fightaholism,
and length of tenure in the club of nations.
Similar to conflict proneness, national fightaholism suggests an increasing
trend as a function of the length of the state’s national history. Less than 30
percent of all states with 40 or less years of history can be labeled fightaholics.
However, the 50-year history mark indicates a steep jump in fightaholism.
This rate increases gradually to the point where about 58 percent of the
states with histories of over 170 years were dispute-related fightaholics, and 50
percent of such states were war-related fightaholics. The rate of dyadic fighta-
holism increases gradually up to a peak of 35 percent for all dyads with 130 years
or less of national independence and declines thereafter. The ‘‘old’’ dyads
(with 160 years or more of joint history) exhibit a fightaholism rate of around
30 percent.

Placing States and Dyads in Risk Groups

Having described patterns of pacifism, conflict proneness, and fightaholism, we can
now identify the states and dyads that belong to each of these groups. Table 3 shows
the major pacifist and the major fightaholic states.

8
A similar procedure was developed by Most and Starr (1980). The measures below are a more accurate
representation of the relationship between conflict measures over time based on an a priori designation of specific
cells in the frequency table.
9
A similar analysis was performed for a five-year period split of national and dyadic history. The associations
between rates of conflict involvement over time for that analysis were even higher than those reported here.
 ZEEV MAOZ 117

TABLE 2. Addiction Levels in MID and War Involvement Rates of States and Dyads, 1816–1992

1-Decade 2-Decades 3-Decades First

/
Variable MID/War Lag Lag Lag Decade

1. National MID/War Involvement per Decade

>No. of Conflicts per Decade MID 0.425nn 0.544nn 0.442nn F
War 0.657nn 0.636nn 0.605nn
N ¼ 1,167 þ N ¼ 986 N ¼ 810
1-Decade Lag of No. Conflicts/Decade MID 0.411nn 0.482nn F
War 0.434nn 0.673nn
2-Decade Lag of No. Conflicts/Decade MID 0.485nn
War 0.668nn F
No. Conflicts/Year Next-to-Last Decade MID 0.459nn
War 0.662nn
N ¼ 161

Pearson Correlations, Next-to-Last MID / 0.905nn

Decade, Relative Conflict Rates War / 0.647nn
(see appendix) N ¼ 161

1-Decade 2-Decades 3-Decades First

Variable MID/War Lag Lag Lag Decade /

1. Dyadic MID/War Involvement per Decade

No. of Conflicts per decade MID 0.619nn 0.604nn 0.588nn F
War 0.617nn 0.589nn 0.656nn
N ¼ 5,443 þ N ¼ 4,436 N ¼ 3,429
1-Decade Lag of No. conflicts/Decade MID 0.651nn 0.627nn F
War 0.645nn 0.673nn
2-Decade Lag of No. conflicts/Decade MID 0.485nn F
War 0.620nn
No. Conflicts/Year Next-to-Last Decade MID 0.640nn
War 0.657nn
N ¼ 1,246
Pearson correlations, Next-to-Last MID / 0.273nn
Decade, relative conflict rates War / 0.088n
(see appendix) N ¼ 546
þ
Entries in each cell are mb statistics.
/
Pearson product moment correlation.
n
po.01; nnpo.001.

From the list in Table 3, it can be said, though with some caution, that most war
pacifists are smaller, less-developed states, while MID and War fightaholics seem to
be composed of two distinct groups. The first is a group of relatively ‘‘young’’ states
(less than 70 years in the system) in the Middle East and East Asia; the second group
is made up of major powers. We turn now to identification of pacifist and fight-
aholic dyads.
The list of pacifist dyads is large and cannot be displayed here. The dyads given
here are the ones with the longest history of dyadic pacifism that are also directly
contiguous. The geographic content of this list of pacifist dyads is interesting: it
consists of Western European and Latin American dyads only. We could not find
significant pacifist dyads in Africa, the Middle East, or Asia.
Turning to the list of conflict-prone and fightaholic dyads, we observe a
general resemblance in the members of the two lists, but they are not in
the same order. We will see below that the correlation between conflict
proneness and fightaholism is not as high as could be expected. Here again, the
members of the two lists are composed of the same groups of states. It is
118 Pacifism and Fightaholism in International Politics

TABLE 3. ‘‘Oldest’’ Pacifist States and Most Fightaholic States

Pacifists1 MID Fightaholism War Fightaholism

No. No. Proportion No. Proportion
Years in Years in of Decades Years in of Decades
State System State System Fightaholic State System Fightaholic

Sweden 177 Israel 45 1.000 Israel 45 0.800

Switzerland 177 Pakistan 46 1.000 Australia 72 0.625
Venezuela 152 India 46 1.000 Syria 47 0.600
Haiti 116 Jordan 47 1.000 Jordan 47 0.600
Uruguay 111 Syria 47 1.000 South Korea 44 0.600
Tunisia 100 India 46 1.000 Turkey 177 0.444
Dom. Repub. 92 United States 177 0.833 Iraq 61 0.429
Albania 75 (North) 39 0.800 China 133 0.429
Vietnam
Liberia 73 South Korea 45 0.800 Ethiopia 91 0.400
Afghanistan2 73 North Korea 46 0.786 India 46 0.400
Nepal 73 China 133 0.778 Pakistan 46 0.400
Costa Rica 73 Russia/Sov. Un. 177 0.722 Cambodia 40 0.400
Panama 73 France 174 0.722 (North) 39 0.400
Vietnam
Ireland 71 Japan 129 0.714 France 174 0.389
Luxemburg 70 Iraq 61 0.714 United States 177 0.333
Iceland 49 Utd. Kingdom 177 0.667 Bulgaria 85 0.333
Burma (Myn) 45 Zaire 43 0.600 Egypt 84 0.333
Sri Lanka 45 Cambodia 40 0.600 Japan 129 0.286
Indonesia 44 Saudi Arabia 66 0.571 Utd. 177 0.278
Kingdom
Taiwan 44 Turkey 177 0.500 Russia/SU 177 0.278
1
Only war pacifism. Recall that no state with a history of 50 years or more was a MID Pacifist.
2
Not a pacifist after 2001.

instructive to note, however, that the conflict-proneness and fightaholism

lists do not consist of all the typical dyads that are placed in the ‘‘enduring rival-
ries’’ lists, such as France–Germany, US–USSR, Greece–Turkey, and so forth
(e.g., Diehl and Goertz 2000; Thompson 2001; Maoz and Mor 2002). Fightaholism
and conflict proneness at the dyadic level are not one and the same as enduring
rivalries.
Table 5 provides intercorrelations between these structural patterns of conflict
involvement at the national and dyadic level of analysis.
The correlations between the measures of pacifism, conflict proneness, and
fightaholism are all statistically significant and generally high on both the national
and the dyadic levels of analysis. The correlations between different structural
patterns of conflict involvement are not as high as one might expect, given the
seemingly similar conceptions of conflict proneness and fightaholism. This suggests
that fightaholismFas a structural propensity for excessive involvement in con-
flictFoffers a distinctive characterization of states and dyads. This is notable in
addition to the lack of one-to-one empirical correspondence between such concepts
as enduring rivalries and dyadic fightaholics.

Correlates of Pacifism, Conflict Proneness, and Fightaholism

Despite the wide array of theories and empirical studies on the correlates
of addiction, the literature on substance and behavioral addiction and on chronic
 TABLE 4. Selected Pacifist, War Prone, and Fightaholic Dyads

Dispute War Fightaholic

Pacifist Dyads Prone Dyads1 Dyads2
No. Fightaholism
Dyad No. Years Dyad No. Years MDs/Year Dyad Years Score

Switzerland-Italy 177 Syria-Israel 45 1.011 Egypt-Israel 45 0.800

Sweden-Denmark 175 Egypt-Israel 45 0.813 Jordan-Israel 45 0.700
France-Switzerland 174 North-South Vietnam 22 0.568 Syria-Israel 45 0.600
UK-Belgium 166 India-Pakistan 46 0.541 North-South Korea 44 0.600
Belgium-France 162 Jordan-Israel 45 0.479 China-India 46 0.600
Venezuela-Brazil 155 North-South Korea 44 0.457 India-Pakistan 46 0.600
ZEEV MAOZ

Brazil-Uruguay 114 Cambodia-Vietnam 21 0.444 Iran-Iraq 61 0.583

Portugal-Morocco 105 US-North Korea 45 0.396 Russia-Japan 131 0.538
US-Argentina 97 Thailand-Cambodia 40 0.372 North-South Vietnam 22 0.500
Sweden-Norway 86 Somalia-Ethiopia 33 0.347 US-North Korea 45 0.500
Finland-Sweden 79 Iran-Iraq 61 0.344 China-(North) Vietnam 39 0.500
Ecuador-Brazil 68 China-India 46 0.327 Iraq-Kuwait 32 0.500
Costa Rica-Colombia 68 Russia-Japan 129 0.315 Iraq-Israel 45 0.500
1
Average of MID and war proneness (average no. of MIDs and wars for year of joint history).
2
Average of MID and war Figthaholism (average proportion of decades of joint history with high MID and War fightaholism scores).
119
 120

TABLE 5. Intercorrelations between Structural Attributes of Conflict Behavior

MIDs/Year of Wars/Year of Relative MID Prop. Decades Relative Prop. Decades

Variable Independence Independence Fightaholic MID Fightaholic War Fightaholic War Fightaholic

National Patterns (N ¼ 180)

Prop. Peace Years in History
0.541
0.351
0.200
0.605
0.182
0.466
No. MIDs/Year of Independence F 0.759 0.657 0.866 0.648 0.736
No. Wars/Year of Independence F 0.516 0.578 0.743 0.662
Relative MID Fightaholism F 0.506 0.884 0.438
Prop. Decades MID Fightaholic F 0.475 0.508
Relative War Fightaholic 0.697
Dyadic Patterns (N ¼ 1,172)
Prop. Peace Years
0.959
0.544
0.420
0.882
0.382
0.591
MIDs/Year of Independence F 0.518 0.882 0.810 0.541 0.585
Wars/Year of Independence F 0.374 0.392 0.844 0.877
Relative MID Fightaholic F 0.716 0.427 0.441
Prop. Decades MID Fightaholic F 0.359 0.452
Relative War Fightholic 0.748
Pacifism and Fightaholism in International Politics

Note: All correlations are significant at the po.0001 level.

 ZEEV MAOZ 121

criminal behavior has identified two basic clusters of correlates of these behaviors:
(1) Intrinsic correlates are factors associated with the individual’s physiology, psy-
chology, or history (both personal and family) and (2) extrinsic correlates refer to
factors in the individual’s social and physical environment that affect the individ-
ual’s behavior. (See Fray 2001 on the general correlates of a wide variety of drug
and behavioral addictions, Teichman 2001 on alcoholism, and Maltz 1984 on re-
cidivism.)
Typical ‘‘correlates’’ of international conflict fit well into this classification. Intrinsic
correlates of conflict at the national level include national characteristics such as
level of power, the (major-regional-minor power) status of the state, its economic
development, regime type, political stability, lateral pressure, and so forth (Rosenau
1966; Choucri and North 1975; Bremer 1982; Maoz 1993, 1996, 1997, 2001).
Intrinsic correlates of dyadic conflict refer to the characteristics of the dyad: its
power ratio, its power-related status, its alliance ties, contiguity, economic devel-
opment, regime type, level of trade, and so forth (Bremer 1982, 1993; Maoz and
Russett 1993; Oneal et al. 1996; Russett and Oneal 2001).
Extrinsic determinants of conflict focus on the politically relevant international
environment (PRIE) in which the state resides: the number of states in its PRIE,
their relative power, number of alliances, level of in the state’s environment, the
level of stability of the state’s environment, and so forth (Maoz 1996, 2001; Ward
and Gleditsch 2000). Studies of dyadic conflict have not generally explored ex-
trinsic correlates of conflict. The few exceptions that discuss such correlates focus
on factors that are similar to those identified at the monadic level (Bennett 1998;
Maoz 2000, 2001).
As noted, this is a preliminary investigation of the correlates of structural location
of states and dyads in different risk groups. A more theoretically informed discus-
sion will be presented and tested in a subsequent study. The purpose of the fol-
lowing analysis is to provide some basic indication of potential ‘‘causes’’ of national
or dyadic conflict patterns. These causes are derived from the quantitative liter-
ature on international conflict that repeatedly uses several basic variables in em-
pirical analysis.10
Table 6 includes a selected list of potential correlates that are frequently men-
tioned in the conflict literature. Since there are some cross-level differences in the
measurement of the variables, we can expect cross-level differences in the results.
In general, however, the relationship between each of the independent variables
and the risk group is expected to be the same across levels of analysis except for
the alliance variable. For this variable two different measures are employed at the
dyadic level. The first is the number of allies of the state with the fewest allies in the
dyad (Maoz 2000; 2001). The second measure denotes whether or not the dyad was
alliedFas used in most dyadic analyses of conflict.
Because this is an analysis of long-term structural traits of states/dyads, I average
most independent variables over time for any given state or dyad. The problem
with this procedure is that it may generate biased inferences, because the temporal
relationship between the dependent and independent variable is not clear. There
are several reasons for justifying this kind of measurement process, however. First,
pacifism as an absolute characteristic of a state can only be observed on a long-term
basis. Thus, uncovering the correlates of pacifism requires averaging potential fac-
tors over long stretches of time. Second, because certain measures (e.g., regime
scores, number of states in PRIE) are measured both at the point of the state’s/
dyad’s origin and over its entire history, it is possible to compare the results ob-
tained from the averaged measures to the same variables which are unaveraged. If

10
Discussions can be found in the literature on national and dyadic conflict involvement (e.g., Bremer 1992,
1993; Maoz 1993, 1996, 1997; Maoz and Russett 1993,and chapters in the handbook edited by Midlarsky 2000).
 122

TABLE 6. Potential Correlates of Pacifism, Conflict Proneness, and Fightaholism

Expected Correlation

Variable Type of Factor Definition Pacifism Proneness Fightaholism Sources

Military Capability Intrinsic National: Relative mil. Cap. State/PRIE.
þ þ Maoz (1996), Geller and Singer (1998),
Dyadic: Capability Ratio (strong/weak) Brecher and Wilkenfeld (1997)
Power Status Intrinsic National: Major/Minor Power status.
þ þ Singer and Geller (1998: 27);
Dyadic: Both majors, mixed, both minors. Bremer (1992)
Economic Development Intrinsic National: Level of economic wealth
þ þ Rosenau (1966)
Dyadic: Weak link level of wealth.1
Regime Type at Origin Intrinsic National. Democracy score. þ

Maoz and Abdolali (1989); Rousseau
Dyadic: Weak link democ. score. et al. (1996); Ray (1995); Maoz (2001).
Type of Acquisition of Intrinsic National. Revolutionary/
þ þ Maoz (1989; 1996), Walt (1996).
National Independence Evolutionary emergence as state.
Dyadic: Type of dyadic acquisition of inde.
Political Instability Intrinsic National. No. of violent regime changes.
þ þ Levy (1988); Maoz (1996).
Dyadic: Weak link regime changes
No. of States in PRIE Extrinsic National. Number of contiguous states.
þ þ Maoz (1996; 2001); Geller and
Dyadic. Weak link no. states Singer (1998: 27).
No. of Alliances Extrinsic National. Number of allies. þ
Dyadic: weak link no. of allies; 

Bremer (1992), Maoz (2000);
presence/absence of alliance in dyad. Geller and Singer (1998)
Pacifism and Fightaholism in International Politics

Conflict in PRIE Extrinsic National: No. of conflicts in PRIE.
þ þ Maoz (1996; 2000; 2001); Geller
Dyadic: Weak link No. of conflicts in PRIEs and Singer (1998: 27).
Regime in PRIE Extrinsic National: Average Regime score of PRIE þ

Maoz (1996; 2001)
Dyadic: Weak link avg. regime of PRIEs
Dyadic Affinity Extrinsic Dyadic: Tau-b (S) score of alliance affinity þ

Bueno de Mesquita (1981);
Signorino and Ritter (1998)
 ZEEV MAOZ 123

TABLE 7. Correlates of Risk Groups (National Level): Pacifism, Conflict Proneness, Fightaholism

MIDs War
þ þþ
Independent Variable Pacifism Proneness Fightaholism Pacifism Proneness Fightaholism

Mil. Capability
0.280nn 0.572nn 0.888nn
0.463nn 0.453nn 0.762nn

Power Status
0.254nn 0.527nn 0.871nn
0.442nn 0.418nn 0.801nn
Economic Wealth
0.198nn 0.422nn 0.775nn
0.343nn 0.312nn 0.641nn
Regime Score 0.193nn
0.090
0.024 0.043
0.060
0.028
Reg. Score at Origin 0.150n
0.251nn
0.286nn 0.298nn
0.234nn
0.296nn
Type of Independence 0.455nn
0.321nn
0.245nn 0.347nn
0.216nn
0.114
Political Instability
0.314nn
0.003 0.298nn 0.023
0.164n 0.214nn
No. States in PRIE
0.406nn 0.572nn 0.854nn
0.508nn 0.417nn 0.739nn
No. of Alliances
0.248nn
0.005
0.062 0.075
0.141n
0.108
Conflict in PRIE
0.114
0.017
0.314nn 0.065
0.014
0.275nn
Regime in PRIE
0.005
0.126n
0.170n 0.117
0.122
0.183nn
N 180 180 180 176 180 180
þ
Results reported here are based on measurement of pacifism as the average number of years without any dispute
over the state’s history.
þþ
Results reported here are based on relative fightaholism (see appendix).
n
po.05; nnpo.01.

results are similar, then we can infer that averaging does not lead to substantial
inferential bias.
It is important for the reader to remember that this empirical analysis
of ‘‘correlates’’ of risk groups is exploratory and is not intended to uncover
‘‘causes’’ of either pacifism or fightaholism. This type of analysis is equivalent to
similar analyses conducted in the study of drug and alcohol abuse as well
as in studies of behavioral addictions (Teichman 1989). With these caveats in mind,
Table 7 presents the results of the analysis linking potential correlates to risk
groups.
National Correlates of Pacifism. The results in Table 7 make several inter-
esting points. First, because there exist no states with more than 46 years of
independence that fit the definition of MID pacifism (that is, avoided MID par-
ticipation over their entire history), I focus the discussion on war-pacifism.
Out of 182 states with 20 or more years of independence, 103 states (over 56.5
percent) did not participate in a war. All of these states were minor powers; none
was a regional power or a major power. Thus, it appears that ‘‘minor power-
hood’’ is a necessary but not a sufficient condition for war-related pacifism. Being
a regional or major power, even for a short period, is a sufficient condition for
nonpacifism.
The most potent intrinsic correlates of war pacifism appear to be military
capability, power-related status, economic wealth, type of acquisition of inde-
pendence, and regime type of origin. Thus, it appears thatFbeyond the capa-
bility/status attributes of statesFthe basic circumstances in which the state
entered the system appear to determine to a significant degree whether or not
the state will be a pacifist. States that emerge into the system as a result of an
evolutionary process (Maoz 1989, 1996) are far more likely to end up as pacifists
than states that enter into the system through a revolutionary political process.
On a more general level, pacifist states tend to be militarily weak, less-developed
economically, have emerged into the system through an evolutionary process,
and were democratic at their origin.
Turning to the extrinsic correlates of pacifism, we note that the politically rel-
evant environment of pacifist states is composed of relatively few other states. Other
extrinsic factors do not appear to be correlated with their pacifist tendency.
124 Pacifism and Fightaholism in International Politics

National Correlates of Conflict Proneness and Fightaholism. Generally speaking, the

correlates of conflict proneness and fightaholism are the same as those of pac-
ifism, but in the opposite direction. The strength of associations, however, differs
somewhat across measures. Military capability, power status, and economic
wealth are significantly correlated with both conflict proneness and fightaholism
at medium levels. Here, too, the political circumstances of the state’s origin de-
termine its future course. States that came about through evolutionary processes
and whose regime at origin was increasingly democratic are less likely to be
conflict prone and fightaholics than states that have emerged through revolu-
tionary processes and were nondemocratic. Again, the only meaningful extrinsic
correlate of conflict proneness and of fightaholism is the number of states in the
focal state’s politically relevant environment.
These results highlight the predicament of the international ‘‘Gullivers.’’ The
rich and powerful find it difficult to escape war involvement. It is the poor, weak,
and relatively isolated states that inherit the peace.
Dyadic Correlates of Pacifism. The analysis of dyadic pacifism suggests that nearly
57 percent of the politically relevant dyads with at least 20 years of joint history
did not engage in an MID. War-related pacifism amounts to 86.5 percent. Table 8
examines dyadic MID- and war-related pacifism.
The results of this table suggest that pacifist dyads display high levels of ca-
pability disparity, they tend to consist of at least one minor power, they tend to be
jointly democratic, and they tend to emerge through evolutionary rather than
revolutionary political processes. The correlates of dyadic pacifism suggest sev-
eral things. First, with two notable exceptions, all pacifist dyads are composed of
no more than one major power. The only major power dyads that avoided war
were US–UK and US–Russia/Soviet Union.11 All MID pacifist dyads were com-
posed of strictly minor powers; all dyads involving at least one major power
experienced at least one MID in their lifetime.
A similar observation applies to dyads composed of regional powers. All pacifist
dyads have no more than one regional power. With one exception (the Israel-
Iran dyad in the Middle East over the 1975–1992 period), all dyads composed of
regional powers experienced both MIDs and wars. It appears that major and
regional power dyads are a sufficient condition for nonpacifism.
And in line with the paradoxical results on the alliance status of dyad members
found in previous studies (e.g., Bueno de Mesquita 1981; Bremer 1992), we find
that the higher the alliance-related affinity of dyad members, the more likely they
are to remain pacifists and the lower their dispute proneness score, but the
higher their likelihood of becoming fightaholics. This suggests that common
interests do not necessarily override conflicting interests. This paradoxical find-
ing seems to contradict realist notions that strategic interests dampen the prob-
ability of conflict (e.g., Farber and Gowa 1995; Gowa 1999). However, this
relationship between alliances, or strategic affinity, and nonpacifism may be al-
tered in a multivariate context as was the case in other studies.
The length of the jointly democratic history of the dyad seems to be correlated
with the likelihood of MID and war pacifism. However, this correlation is an
understatement of the relationship between democracy and pacifism. It turns out
that no two politically relevant states that had been jointly democratic for 70 percent or more
of their common history engaged in war. It also turns out that only 5 out of 101 dyads
that had been jointly democratic for more than half of their common history (US–France,
UK–France, US–Italy, Japan–Australia, and Japan–New Zealand) engaged in interstate
wars. Of course, all these wars occurred when one state was nondemocratic.
This is extremely powerful evidence in support of the democratic peace

11
The start of the temporal domain at 1816 ‘‘misses’’ the Anglo-American war of 1812.
 ZEEV MAOZ 125

TABLE 8. Correlates of Risk Groups (Dyadic Level): Pacifism, Conflict Proneness, Fightaholism

MIDs War
þ þþ
Independent Variable Pacifism Proneness Fightaholism Pacifism Proneness Fightaholism

Capability Ratio
0.119nn
0.066n
0.088nn
0.145nn
0.041
0.052n

Power Status 0.225nn
0.104nn
0.143nn 0.222nn 0.027 0.041
Economic Wealth 0.106nn
0.026
0.027 0.115nn 0.043 0.050
Regime Score 0.210nn
0.108nn
0.123nn 0.174nn
0.050
0.075n
Reg. Score at Origin
0.054
0.058n
0.091nn 0.090n
0.108nn
0.106nn
Type of Independence 0.202nn
0.209nn
0.227nn 0.143nn
0.115nn
0.114nn
Alliance Affinity 0.116nn
0.321nn 0.183nn 0.050n 0.031 0.056n
Regime Persistence 0.078n
0.003
0.077n 0.041 0.031
0.021
No. States in PRIE
0.071n 0.572nn 0.133nn 0.068n 0.083n 0.128nn
No. of Alliances
0.125nn
0.005
0.165nn 0.083nn
0.083n
0.091nn
Conflict in PRIE
0.123nn 0.233nn 0.218nn 0.044 0.220nn 0.271nn
Regime in PRIE 0.034
0.126n 0.100nn 0.079 0.026 0.021
N 1,172 1,172 1,172 1,172 1,172 1,172
þ
Results reported here are based on measurement of pacifism as the average number of years without any dispute
over the state’s history
þþ
Results reported here are based on measurement of fightaholism as the proportion of the state’ s decades in
which it was coded as fightaholic.
n
po.05; nnpo.01.

proposition. Long periods of joint democracy tend to spill over into pacifism even when at
least one member of the dyad is not a democracy. In addition, dyads in which both
members emerge through evolutionary processes are more likely to end up as
pacifist and less likely to end up as conflict prone or fightaholics. This corrob-
orates Maoz’s (1996) findings about the relationship between state-making proc-
esses and conflict propensity.
Among the extrinsic factors, the most potent correlate of the structural
characteristic of the dyad is the number of states in the dyad members’ res-
pective PRIEs. In addition, the larger the number of allies each member of the
dyad has, the less likely it is to be a MID pacifist but the more likely it is to be war
pacifist.
The level of conflict in dyad members’ PRIEs is inversely related to their
pacifist tendencies and positively related to their level of fightaholism. Other
variables do not exhibit significant relationships with the dependent variables.
The structural position of a state in terms of the various risk groups discussed
here appears, therefore, to have some rather distinct correlates. These correlates
are both intrinsicFbasic national or dyadic attributesFand extrinsic, referring
to the characteristics of states’ or dyads’ environments.

Conclusion
This survey of national and dyadic conflict-involvement patterns over long histor-
ical periods suggests the following observations:

1. National and dyadic patterns of conflict involvement are characterized by extreme

levels of inequality. A considerable number of states engaged in relatively few
MIDs and wars, while a small number of states accounted for a high pro-
portion of all MIDs and wars over the last two centuries. Likewise, a small
proportion of the politically relevant dyads accounted for most dyadic
conflicts.
126 Pacifism and Fightaholism in International Politics

2. There is consistent evidence of national and dyadic conflict-related addiction,

or fightaholism. Specifically, some, but hardly all, of the most conflict-
prone states engaged consistently in a pattern of excessive involvement in
conflict over a large portion of their national histories. These states tend to
form dyads that are engaged in repeated conflict, thus accounting for a
disproportionately high fraction of all dyadic MIDs and wars over the
1816–1992 era. The evidence for national and dyadic fightaholism is fairly
robust.
3. Empirical evidence establishes substantial rates of pacifism at the national and
dyadic level. Side-by-side the evidence of fightaholism, there is substantial
abstinence from conflict involvement at the national and dyadic levels.
Most states and most dyads are neither extreme users of conflict nor are
they complete pacifists. Nevertheless, just as a significant minority of fight-
aholic states and dyads exist, there are a substantial number of states and
dyads that do not fight at all. Just as we spend considerable effort trying to
account for conflict proneness and conflict addiction, it is time we started
studying pacifism systematically.

The preliminary analysis of the correlates of a state’s position within a given risk
group suggests the following conclusions:
4. Pacifism has distinct characteristics. Being a minor power is a neces-
saryFthough not a sufficientFcondition of pacifism. In addition, pacifist
states tend to be militarily and economically weak and tend to have rel-
atively few immediate neighbors.
5. By and large, the correlates of conflict proneness and conflict addiction are the flip
side of the correlates of pacifism; however, there are important exceptions. In gen-
eral, factors that correlate positively with conflict proneness tend to cor-
relate positively with conflict addiction and negatively with absolute or
relative pacifism, and vice versa.
6. It is possible and even important to understand national and dyadic patterns over
long periods of time. This is the key conclusion of this study. The analysis of
national and dyadic patterns over long historical stretches produces some
important insights into issues of conflict and peace. First, evidence about
pacifism, conflict proneness, and addiction may be a useful source for
theorizing about structural patterns of relations at various levels of analysis.
Second, comparing empirical findings from studies focusing on the tra-
ditional unit-by-year observation with studies focusing on the long-term
tendencies of states and dyads may corroborate existing evidence on the
correlates of conflict and warFas many of the findings of this study do.
Alternatively, this comparison may reveal important gaps in our knowledge
as we address different units of analysis. For years, cross-level paradoxes
have perplexed students of war and peaceFthe level-of-analysis puzzle in
alliance and democratic peace studies being only two notable examples
(Maoz 2000, 2001).

This study did not attempt to develop a theory of conflict or peace but rather to
document some basic patterns of conflict and peace across states and dyads. Gen-
eralizing from these patterns into coherent explanations of conflict is left for future
studies.

Methodological Appendix
This appendix discusses all the issues related to the research design of this study.
 ZEEV MAOZ 127

Data Sources
The principal sources of data for this study include the following:

Variable/s Dataset Source

MIDs/Wars Dyadic MID Dataset Maoz (1999)

Regime/Polity Formation Polity II–IV Jaggers and Gurr (1995)
Mil. Capability COW National Capabilities http://cow2.la.psu.edu/
Alliances COW Alliance Dataset Bennett and Stam (2000)

Spatial-Temporal Domain and Units of Analysis

The temporal domain covers the 1816–1992 period. The spatial domain consists of
all states with twenty or more years of national independence over this period
(N ¼ 182). The twenty-year threshold is due to definitions of units of analysis below.
Two units of analysis are employed. The national-history observation tracks each
state over its entire history, from the point it acquired independence to the end of the
period (or to the point the state lost its independence for those states that have not
persisted to the present). States that were independent prior to 1816 were treated as
if they acquired independence in 1815. With regard to measures of MID and war
addiction, state-histories required division of the temporal domain into decades (or
half-decades for validity tests of these measures), starting the counters with the first
year of independence and onwards. Thus, a state that acquired independence in
1838 ended its first decade in 1847, its second decade in 1857, and so forth.
The second unit is the politically relevant dyad history. A politically relevant dyad
is made up of contiguous states, a dyad where one of its members is a major power
with global reach capacity, or a dyad with a regional power that has regional reach
capacity (Maoz and Russett 1993; Maoz 1996, 2001). Each such dyad makes for a
single observation, measured over the entire joint history of the dyad. This history
is defined from the point when the ‘‘youngest’’ member of the dyad acquired
independence to the point where one member ceased to be an independent state,
or through 1992. Dyadic addiction measures entailed dividing dyadic histories into
decades (or half-decades). Dyads with less than twenty years of joint history were
deleted. The grand total for the dyadic analyses is N ¼ 1,172.12

Measurement of Variables
Dependent Variables: Measures of Conflict and Peace:

General Definitions of Conflict: All of the dependent variables are based on the
standard definition of MID involvement (Gochman and Maoz 1984:596) and of
war involvement (Small and Singer 1982).

Pacifism: I use two definitions of pacifismFabsolute and relative pacifism. Ab-

solute Pacifism is defined as the absence of conflict (MID/War) given sufficient
opportunity. Thus, a state/dyad is considered MID pacifist if it did not participate
in any MID (either as initiator, target, or joiner) during its entire (joint)
history. A state/dyad is considered a war pacifist if it did not participate in an
interstate war during its entire history. Relative Pacifism is measured as the average
number of years of peace (absent any MID/war involvement) of a state/dyad.

12
Analyses were performed also for the population of states/dyads with a minimum of 30 years of history. Results
were largely robust over these breakdowns.
128 Pacifism and Fightaholism in International Politics

Conflict Proneness: Conflict proneness is the number of dyadic MIDs/war in-

volvements of a state/dyad per year. Note that each dyadic MID underway in a
given year gets a score of one. Thus, if a state is involved in five MIDs (or in one
MID involving five actual opponents) during a given year, it gets a score of 5 for
that year. Likewise, if a state started a MID during one year and this MID ex-
tended over a period of four subsequent years, the state gets a score of one for
each year this MID was in progress.13

Measures of MID and War Fightaholism: Following the definitions of fightaholism

in section three of this study, the procedure for developing measures of addiction
was as follows. First, there are no strict benchmarks for defining ‘‘normal’’ in-
volvement in conflict and no strict benchmarks for defining ‘‘periods’’ pertaining
to intervals during which exposure to conflict is measured. Hence, I use two
alternative intervals of exposure to conflict. Also, due to the complexity of this
concept, I develop two alternative sets of measures. Both are used in the analyses
presented herein.
To start with, I take the date of a state’s entry into the system as the baseline for
measuring its conflict involvement. From this point of system entry, I break the
state’s history into two sets of alternative intervals, half-decades and decades.
Then I assess its relative addiction severity and cumulative addiction.
Relative Addiction Severity (RELADCT) is developed in several steps. First, each
nation’s history is divided into decades starting with its year of acquisition of in-
dependence and forward. Thus, for a state that became independent in 1951,
the first decade covers the years 1951 to 1960, the second decade covers the years
1961 to 1970, and so forth. Second, for each decade interval, and considering
all independent states during that interval, the average number of disputes or wars
per state is computed. This establishes the ‘‘normal level of conflict involvement.’’
Thus, over the decade 1951 to 1960, the ‘‘normal level of MID involvement’’ is
given by:
P P
1960 nt
DYDMIDit
i¼1
NORMID1951
1960 ¼ t¼1951 ; ð1Þ
P
1960
NOSTATESt
t¼1951

Where DYDMIDit is the number of dyadic MID involvements of state i in year t,

and NOSTATESt is the number of states in the system in year t. Next, for each
state, the relative level of conflict involvement is computed in the following
manner.
8 1960 9
> P >
>
> DYDMID >
>
>
>
it
>
>
< NORMID
t¼1951
if NORMID > 0 =
1951
1960
MIDADCTi;1951
1960 ¼ ; ð2Þ
>
> 0:1 if NORMID ¼ 0 >>
>
> >
>
>
: >
;
or MIDADCT ¼ 0
A score of 0.1 is assigned to MIDADCT when it assumes the value of zero in
order to avoid multiplication by zero in the subsequent stage. Thus, a state that
experiences above average levels of MID involvement over a decade gets a score
larger than one, while a state that experiences below-average involvement over a
decade receives a score between 0.1 and 1. This establishes the state’s rate of

13
See Maoz (1998) for a justification for focusing on disputes underway rather than dispute outbreak in studies
of international conflict.
 ZEEV MAOZ 129

conflict involvement over a decade relative to the ‘‘normal’’ or average level of

involvement over that decade.14
Recall that by addiction to conflict, or fightaholism, we refer to excessive con-
flict involvement over long periods of time. Thus, fightaholism increases with the
length of time (number of decades) a state is excessively involved in conflict.
Accordingly, I measure the relative fightaholism of a state over its entire history
using the formula:

1=k
k
RELADCTi ¼ P MIDADCTid ; ð3Þ
d¼1

In the formula, d indexes the nation-decade starting with the first decade of
independence and going to the last (or last decade prior to 1992). This last decade
is indexed by k. Thus, RELADCT is the geometric mean of the decade-rate of
addiction scores at the end of the period of observation. The same logic was applied
to time units of five years and to dyadic addiction. In the latter case, the i index in
Equations (1)–(3) refers to politically relevant dyads rather than to individual states.
Cumulative Addiction reflects the pro portion of a state’s duration in the system
during which it was characterized as addicted. First, I generate a relative posi-
tioning of the state in terms of its MID/war involvement during a given decade by:
 
1 if MIDit > 75 pctile MIDt
TFMIDit ¼ ð4Þ
0 if MIDit  75 pctile MIDt
In this formula, TFMIDit is the relative positioning score of MID involvement.
Thus, a state whose conflict involvement rate during a given decade was higher
than the conflict involvement of 75 percent of all states existing during this decade
gets a high addiction severity score for the decade. States whose MID involvement
rates were equal to or lower than the 75 percent most dispute-prone states re-
ceived got a score of zero.
Next, I cumulated across all consecutive ten-year periods where a state received
a TFMID score of 1, such that the first period gets a score of 1, the second 2, and so
forth. This variable is labeled CUMIDAD. The counter is set to zero whenever a
decade exists that a state gets a score of zero (that is, it drops out of the 25th upper
percentile of MID-prone states). The relative cumulative MID addiction score is
then computed by:
MaxðCUMIDADÞi
RELCUMIDi ¼ ð5Þ
MAXDECADi
MAXDECADi here is the number of decades during which the state was an
independent system member.15
Risk Groups: As is customary in research on obsessive behavior in the behavioral
or psychiatric sciences, subjects are placed into risk groups in order to uncover
correlates of addiction or other chronic obsessive behaviors. Risk groups are de-
fined as follows:
8
< 0 if state=dyad is paifist
Risk Group ¼ 2 if state=dyad fightaholic for more than half its history ð6Þ
:
1 otherwise

14
I use this measure rather than normal scores (Z-scores), because of the highly skewed distributions of conflict
and war involvement.
15
The same operations with appropriate adjustments are conducted for war-related fightaholism and for MID
and war fightaholism at the dyadic level.
130 Pacifism and Fightaholism in International Politics

Independent Variables:

National Attributes
(1) Military Capability: The average fraction of a state’s military expendi-
tures and military personnel of the system’s total.
(2) Economic Capability: Measured as the average fraction of a state’s iron
and steel production and its energy consumption.
(3) Power Status: The proportion of the state’s history during which it qual-
ified as a major or regional power (Maoz 1996:139; 2001).
(4) Regime Score: From Maoz and Russett (1993), the regime score is de-
fined as: REGIME ¼ (DEMOC–AUTOC)  CONCEN, where DEMOC is a
state’s democracy score, AUTOC is its autocracy score and CONCEN is its
power concentration score.
(5) Type of National Origin: Following Maoz (1989, 1996), this concept is
measured as 0 if the emergence of the state into the system was done in
an evolutionary fashion, and 1 if it entered the system through a rev-
olutionary/violent process.
(6) Regime Score at Origin: The regime score of the state at the year it
entered the system.
(7) Regime Stability: Number of regime changes divided by the length of the
state’s history.
(8) Number of States in PRIE: The average number of states in a given
nation’s PRIE.
(9) Number of Allies: Average number of states having alliance ties with the
focal state (Maoz 1996:169–170; 2000:140).
(10) Number of MIDs/Wars in PRIE: Number of dyadic MIDs/Wars in the
PRIE of the focal state, excluding MIDs/Wars involving the focal state.
(11) Average Regime Score in PRIE: Average regime score in the focal state’s
PRIE. (Also measured in some analyses as the proportion of states in
one’s PRIE that are democratic; see Maoz 1996: 171; 2001).
(12) Instability in PRIE: Average number of regime changes in the state’s
PRIE

Dyadic Attributes
(1) Capability Ratio: Average ratio of military capabilities of strongest to
weakest state in dyad. (Bremer 1992; Maoz and Russett 1993).
(2) Power Status (PS) of Dyad: Average (arithmetic mean) score of dyad on
the following scale:
8 9
>
> 0 Minor
Minor >
>
> 1 Minor
Re gional >
> >
>
> >
>
< =
2 Minor
Major
PS ¼
>
> 3 Re gional
Re gional >
>
>
> >
>
>
> 4 Re gional
Major >
>
: ;
5 Major
Major

(3) Alliance Status: Proportion of years during which dyad members were
allied.
(4) Similarity of Interests: Bueno de Mesquita’s (1981) tau-b measure of
similarity of interests based on similarity of alliance portfolios of mem-
bers of dyads. I also use Signorino and Ritter (1999) measures, labeled as
S and the weighted version of S labeled as Ws.
 ZEEV MAOZ 131

(5) Regime Stability: Average number of (violent or nonviolent) regime

changes as defined by Maoz (1996:128, 219–220).
(6) Type of Dyad Origin: This variable is coded as two if both dyad members
entered the system through a revolutionary process, one if one entered
the system in a revolutionary fashion and one entered through an ev-
olutionary process, and zero if neither member of the dyad entered the
system through revolutionary processes.
(7) Minimum Regime Score of Dyad: Smallest of average regime scores of
members of dyad.
(8) Minimum Regime Score at Dyad’s origin: Smallest of the regime scores
of dyad members at the year the ‘‘youngest’’ member entered the sys-
tem.
(9) Number of MIDs/Wars in Dyad’s PRIEs: Minimum average number of
MIDs/wars in the PRIEs of dyad members.
(10) Average Regime Score in Dyad’s PRIEs: Smallest of average regime
score in states’ PRIE.

References
BECK, NATHANIEL, J. N. KATZ, AND RICHARD M. TUCKER. (1998) Taking Time Seriously: Time Series,
Cross-Section Analysis with a Binary Dependent Variable. American Journal of Political Science
42(4):1260–1288.
BECK, NATHANIEL J., GARY KING, AND LANGCHE ZENG. (2000) Improving Quantitative Studies of
International Conflict. American Political Science Review 94(1):21–33.
BENNETT, SCOTT D. (1998) Integrating and Testing Models of Rivalry. American Journal of Political
Science 42(4):1200–1232.
BENNETT, SCOTT D., AND ALAN STAM. (2000) EUGENE: A Data Resource. International Interactions
26(2):179–204.
BREMER, STUART. (1980) The Trials of Nations: An Improbable Application of Probability Theory. In
The Correlates of War, Vol. 2, edited by J. David Singer. New York: Free Press.
BREMER, STUART. (1982) The Contagiousness of Coercion: The Spread of Serious International Dis-
putes, 1900–1976. International Interactions 9(1):29–55.
BREMER, STUART. (1992) Dangerous Dyads: Conditions Affecting the Likelihood of Inter-state War,
1816–1965. Journal of Conflict Resolution 36(2):309–341.
BREMER, STUART. (1993) Democracy and Militarized Interstate Conflict, 1816–1965. International In-
teractions 18(3):231–249.
BREMER, STUART. (2000) Who Fights Whom, Where, and Why? In What Do We Know about War? edited
by John A. Vasquez. Lanham, MA: Rowman and Littlefield.
BROCKNER, JOEL, AND JEFFREY Z. RUBIN. (1985) Entrapment in Escalation Social Conflicts: A Social Psy-
chological Analysis. New York: Springer Verlag.
BUENO DE MESQUITA, BRUCE. (1981) The War Trap. New Haven: Yale University Press.
CHOUCRI, NAZLI, AND ROBERT C. NORTH. (1975) Nations in Conflict. San Francisco: Freeman.
CLARKE, RONALD V., AND DAVID L. WEISBURD. (1990) On the Distribution of Deviance. In Policy and
Theory in Criminal Justice, edited by Don M. Gottfredson and Ronald V. Clarke. Aldershot, UK:
Avebury Press.
DIEHL, PAUL F., AND GARY GOERTZ. (2000) War and Peace in International Rivalry. Ann Arbor: University
of Michigan Press.
FARBER, HENRY S., AND JOANNE GOWA. (1995) Polities and Peace. International Security 20(2):123–146.
FRAY, REBECCA J. (2001) Addiction. In Gale Encyclopedia of Medicine. New York: Macmillan.
GOCHMAN, CHARLES S., AND ZEEV MAOZ. (1984) Militarized Interstate Disputes, 1816–1976: Proce-
dures, Patterns, and Insights. Journal of Conflict Resolution 28:585–615.
GOERTZ, GARY, AND PAUL F. DIEHL. (1992) The Empirical Importance of Enduring Rivalries. Inter-
national Interactions 18:151–163.
GOERTZ, GARY, AND PAUL F. DIEHL. (1993) Enduring Rivalries: Theoretical Constructs and Empirical
Patterns. International Studies Quarterly 37:147–171.
GOTTFREDSON, MICHAEL R., AND TRAVIS HIRSCHI. (1990) A General Theory of Crime. Stanford: Stanford
University Press.
GOWA, JOANNE. (1999) Ballots and Bullets: The Elusive Democratic Peace. Princeton: Princeton University
Press.
132 Pacifism and Fightaholism in International Politics

HENDERSON, ERROL A. (1998) The Democratic Peace through the Lens of Culture. International Studies
Quarterly 42:461–484.
HOLSTI, KALEVI J. (1991) Peace and War: Armed Conflict and International Order. Cambridge: Cambridge
University Press.
JAGGERS, KEITH, AND TED ROBERT GURR. (1995) Tracking Democracy’s Third Wave with The POL-
ITY III Data. Journal of Peace Research, 32(4):453–468.
LARSON, DAVID E., ED. (1996) Mayo Clinic Family Health Book. 2nd ed. New York: William Morrow.
MALTZ, MICHAEL D. (1984) Recidivism. Orlando: Academic Press.
MANSFIELD, EDWARD D., AND JACK SNYDER. (1995) Democratization and the Danger of War. Interna-
tional Security, 20(1):5–38.
MANSFIELD, EDWARD D., AND JACK SNYDER. (1996) Effects of Democratization on War. International
Security, 20(4):196–207.
MANSFIELD, EDWARD D., AND JACK SNYDER. (1997) A Reply to Thompson and Tucker. Journal of Conflict
Resolution, 41:457–461.
MAOZ, ZEEV. (1989) Joining the Club of Nations: Political Development and International Conflict,
1816–1976. International Studies Quarterly 33:199–231.
MAOZ, ZEEV. (1990) Paradoxes of War: On the Art of National Self-Entrapment. Boston: Linwin Hymann.
MAOZ, ZEEV. (1993) The Onset and Initiation of Militarized Interstate Disputes in the Modern Era.
International Interactions 19(2–3):117–141.
MAOZ, ZEEV. (1996) Domestic Sources of Global Change. Ann Arbor: University of Michigan Press.
MAOZ, ZEEV. (1997) The Strategic Behavior of Nations. Paper presented at the annual meeting of the
Peace Science Society (International), Houston, TX.
MAOZ, ZEEV. (1998) Realist and Cultural Critiques of the Democratic Peace: A Theoretical and Em-
pirical Reassessment. International Interactions 24(1):3–89.
MAOZ, ZEEV. (1999) The Dyadic MID Dataset, 1.0. Available at http://psfaculty.ucdavis.edu/zmaoz/
dyadmid.html.
MAOZ, ZEEV. (2000) The Street-Gangs of World Politics: The Origins, Management, and Consequences
of International Alliances, 1816–1986. In What Do We Know about War? edited by John A.
Vasquez. Lanham, MD: Rowman and Littlefield.
MAOZ, ZEEV. (2001) Democratic Networks: Connecting National, Dyadic, and Systemic Levels-of-
Analysis in the Study of Democracy and War. In War in a Changing World, edited by Zeev Maoz.
Azar Gat. Ann Arbor: University of Michigan Press.
MAOZ, ZEEV, AND BEN D. MOR. (2002) Bound By Struggle: The Strategic Evolution of Enduring Rivalries.
Ann Arbor: University of Michigan Press.
MAOZ, ZEEV, AND BRUCE RUSSETT. (1993) Normative and Structural Causes of Democratic Peace,
1946–1986. American Political Science Review 87:624–638.
MIDLARSKY, MANUS, ED. (2000) Handbook of War Studies II. Ann Arbor: University of Michigan
Press.
MOST, BENJAMIN, AND HARVEY STARR. (1980) Diffusion, Reinforcement, Geopolitics, and the Spread of
War. American Political Science Review 74:932–946.
MOST, BENJAMIN, AND HARVEY STARR. (1989) Inquiry, Logic, and International Politics. Columbia: Uni-
versity of South Carolina Press.
ONEAL, JOHN, FRANCES ONEAL, MAOZ ZEEV, AND BRUCE RUSSETT. (1996) The Liberal Peace: Economic
Interdependence, Democracy, and International Conflict, 1950–85. Journal of Peace Research
33(1):11–28.
ROSENAU, JAMES N. (1966) Pre-Theories and Theories of Foreign Policy. In Approache to Comparative
and International Politics, edited by R. Barry Farrell. Evanston, IL: Northwestern University Press.
RUSSETT, BRUCE, AND JOHN ONEAL. (2001) Triangulating Peace: Democracy, Interdependence, and Inter-
national Organization. New York: Norton.
SINGER, J. DAVID, AND MELVIN SMALL. (1972) The Wages of War. New York: Wiley.
SIGNORINO, CURTIS S., AND JEFFREY M. RITTER. (1999) Tau-b or not Tau-b: Measuring the Similarity
of Foreign Policy Positions. International Studies Quarterly. 43:115–144.
SMALL, MELVIN, AND J. DAVID SINGER. (1982) Resort to Arms: International and Civil Wars, 1816–1980.
Beverly Hills, CA: Sage.
TEICHMAN, MEIR. (1989) Life in a Different World: Alcohol, Drugs, and Human Behavior (Hebrew). Tel-Aviv,
Ramot: Tel-Aviv University Press.
TEICHMAN, MEIR. (2001) Alcohol and Alcoholism: Causes, Prevention, and Treatment (Hebrew). Tel-Aviv,
Ramot: Tel-Aviv University Press.
THOMPSON, WILLIAM R. (2001) Identifying Rivalries in World Politics. International Studies Quarterly
45:557–586.
 ZEEV MAOZ 133

THOMPSON, WILLIAM R., AND RICHARD M. TUCKER. (1997) A Tale of Two Democratic Peace Critiques.
Journal of Conflict Resolution 41:428–454.
VASQUEZ, JOHN A. (1993) The War Puzzle. Cambridge: Cambridge University Press.
WALTZ, KENNETH N. (1979) Theory of International Politics. Menlo Park, CA: Addison Wesley.
WARD, MICHAEL D., AND CHRISTIAN GLEDITSCH. (2000) War and Peace in Space and Time: The Role
of Democratization. International Studies Quarterly 44:1–29.
WRIGHT, QUINCY. (1944) A Study of War. Chicago: University of Chicago Press.
134