Content uploaded by Bethany Lacina
Author content
All content in this area was uploaded by Bethany Lacina on Oct 22, 2015
Content may be subject to copyright.
Monitoring Trends in Global Combat: A New
Dataset of Battle Deaths
z
BETHANY LACINA
1,2,
* and NILS PETTER GLEDITSCH
2,3
1
Stanford University, Palo Alto, CA, U.S.A.;
2
Centre for the Study of Civil War (CSCW),
International Peace Research Institute Oslo (PRIO), Hausmanns gate 7, NO-0186 Oslo,
Norway;
3
Norwegian University of Science and Technology, Trondheim, Norway
(*Author for correspondence, e-mail: blacina@stanford.edu)
Received 26 February 2004; accepted in final form 22 June 2004
Lacina, B. and N. P. Gleditsch, 2005, Monitoring Trends in Global Combat: A New Dataset
of Battle Deaths, European Journal of Population, 21: 145–166.
Abstract. Both academic publications and public media often make inappropriate use of
incommensurate conflict statistics, creating misleading impressions about patterns in global
warfare. This article clarifies the distinction between combatant deaths, battle deaths, and
war deaths. A new dataset of battle deaths in armed conflict is presented for the period
1946–2002. Global battle deaths have been decreasing over most of this period, mainly due
to a decline in interstate and internationalised civil armed conflict. It is far more difficult to
accurately assess the number of war deaths in conflicts both past and present. But there are
compelling reasons to believe that there is a need for increased attention to non-battle causes
of mortality, especially displacement and disease in conflict studies. Therefore, it is
demographers, public health specialists, and epidemiologists who can best describe the true
human cost of many recent armed conflicts and assess the actions necessary to reduce that
toll.
Key words: battle deaths, casualties, combat mortality, conflict, war deaths
Lacina, B. et N. P. Gleditsch, 2005, Suivre les conse
´
quences humaines des conflits dans le
monde : nouvelles donne
´
es sur les de
´
ce
`
s lie
´
s aux combats, Revue Europe
´
enne de De
´
mographie,
21: 145–166.
Re
´
sume
´
. Que ce soit dans les publications de recherche ou les me
´
dias, l’usage de statistiques
disproportionne
´
es sur les victimes de conflits donne souvent une image de
´
forme
´
e des
conse
´
quences des ope
´
rations de guerre. Cet article distingue les morts de combattants, des
victimes des combats et des victimes de guerre. Il pre
´
sente un nouvel ensemble de donne
´
es
sur les de
´
ce
`
s dus aux combats sur la pe
´
riode 1946 a
`
2002. Le total des de
´
ce
`
s dus aux combats
a diminue
´
sur presque toute la pe
´
riode du fait d’une re
´
duction des conflits internationaux et
entre e
´
tats. Il est beaucoup plus difficile d’estimer le nombre total de de
´
ce
`
s dus a
`
la guerre
dans les conflits passe
´
s ou pre
´
sents. Mais il y a de fortes raisons de croire qu’il est ne
´
cessaire,
dans les e
´
tudes sur les conflits, de porter une plus grande attention aux causes de de
´
ce
`
s non
directement dues aux combats, notamment celles lie
´
es au de
´
placement des populations et a
`
la
diffusion de maladies. C’est pourquoi ce sont les de
´
mographes, les spe
´
cialistes de sante
´
European Journal of Population (2005) 21: 145–166 Springer 2005
DOI 10.1007/s10680-005-6851-6
publique et les e
´
pide
´
miologistes qui sont le plus a
`
meˆ me d’estimer le ve
´
ritable cou
ˆ
t humain
de beaucoup de conflits re
´
cents et d’identifier les actions ne
´
cessaires a
`
la re
´
duction de ce
fardeau.
Mots cle
´
s: de
´
ce
`
s au combat, de
´
ce
`
s de guerre, conflit, mortalite
´
au combat, pertes militaires
1. Introduction
Estimating how many deaths a war has caused is an exercise of obvious
importance but surprising complexity
1
. Of course, relevant information is
frequently concealed by parties to the conflict, destroyed in the course of
the war, or never recorded at all. But an additional layer of confusion
arises due to the complex and contradictory schemes that are used to
account for war losses. Few who go in search of such statistics pay close
attention to the maze of categories that militaries use to classify combat
losses, and those formal schemes are often difficult to apply or even
irrelevant in the context of civil wars or wars against informally organised
insurgents. The result is that inaccurate or misleading fatality figures are
frequently circulated widely, gaining credibility through mere repetition.
One example is the oft-repeated observation that 90% of the casualties in
today’s wars are civilians (Sivard, 1996, p. 17), as against only 5% in
World War I (Chesterman, 2001, p. 2). Yet, the 5% figure for World War
I is far lower than the range cited by most historians (Clodfelter, 2002,
p. 479), while the source of the estimate of 90% civilian casualties in
modern wars has long vanished (Mack, 2005).
In order to understand trends in warfare across time or space, we need
data that measure deaths due to armed conflicts in a consistent manner.
This article begins by distinguishing three principal ways of counting war
fatalities: combatant deaths, battle (or combat) deaths, and war deaths.
These three measures are appropriate for answering different research
questions, and we suggest some of the possibilities and limitations of each.
We also clarify the distinction between battle deaths and one-sided vio-
lence. We then present a new dataset of battle deaths in state-based armed
conflicts for the period after World War II (1946–2002). The new data
have been gathered for conflicts recorded in the Uppsala/PRIO dataset of
armed conflicts (Gleditsch et al., 2002; Harbom and Wallensteen, 2005).
Versions of the data are also available for use with the Correlates of War
(COW) data on wars 1900–1997 (Sarkees, 2000) and data on civil wars
1945–1999 compiled by Fearon and Laitin (2001). After presenting the
data, we display our estimate of the trends in global and regional battle
deaths 1946–2002. Battle violence has declined over the past 50 years due
to a decline in major interstate conflict and large internationalised civil
conflicts. However, in our final section, we point out that many conflicts
BETHANY LACINA AND NILS PETTER GLEDITSCH
146
are characterised by numbers of non-violent deaths due to humanitarian
crisis that far surpass the lives lost in combat. For that reason, demog-
raphers, epidemiologists and others who specialise in the scientific study of
population will play an important role in investigating the relationship
between conflict and humanitarian crisis and making recommendations on
appropriate international responses.
2. Distinguishing among fatality statistics
War fatality statistics may be compiled with very different research needs in
mind and there are a number of classification schemes for doing so. In this
article, however, we seek to distinguish three of the most common: counts of
combatant deaths, battle deaths (which we use synonymously with combat
deaths), and war deaths. The theoretical uses of each of these measures can
be quite different. We explain each term as we proceed, but Appendix A
contains the formal definition of battle deaths used to code our dataset.
Among political scientists, military experts, and legal scholars one of the
most commonly sought war fatality figures is that of combatant deaths. For
example, the Correlates of War (COW) dataset on interstate wars, widely
used by political scientists in quantitative study of conflict, records for each
state involved ‘‘the number of battle-connected fatalities among military
personnel’’ (Sarkees 2000, p. 128). Other compilations expand this definition
somewhat and account for soldiers who are not formally attached to any
state’s military.
Figures on combatant losses can be used to answer questions of strategy.
Comparing the parties’ battle losses may reveal their military capacity and
effectiveness, and be useful for evaluating their preparation for and execution
of a war and their capacity to continue fighting it. Counting the number of
non-violent deaths among combatants can be important to a study of cam-
paign conditions, or to comparing militaries’ culture or organisational
sophistication.
Comparisons of combatant to non-combatant deaths are also often used
to illuminate normative questions. Such an accounting of war losses implies a
distinction between legitimate targets in a war (combatants) and all other
persons (such as captured soldiers and civilians). The laws of war contained
in international treaties such as the Fourth Geneva Convention offer a degree
of protection for civilians that is not extended to combatants and hold mil-
itary forces responsible for the safety of civilians in the areas they control.
Thus, investigation of war crimes can turn on accounts of war deaths that
distinguish between combatants and civilians (e.g. Brunborg et al., 2003).
Just war theory sets the benchmark of proportionality in order to evaluate
the conduct of war by weighing the military objectives achieved against
‘collateral damage’ to non-combatants (Walzer, 1977). This kind of a
MONITORING TRENDS IN GLOBAL COMBAT
147
balancing analysis can be used to criticize certain modes of warfare as par-
ticularly indiscriminate, inhumane, or unjustifiably devastating to civilians.
By contrast, the measure we refer to as battle deaths includes all people,
soldiers and civilians, killed in combat. Measuring battle deaths answers the
question of how many people were killed in military operations during a war
and, therefore, it is the best measure of the scale, scope, and nature of the
military engagement that has taken place. It reflects the degree of military
parity between the sides, how heavily armed they are, and how frequently
and widely they engage each other.
Data on battle deaths are empirical measures of the size of combat
operations. They lack the normative concerns of an accounting of combatant
deaths that seeks to distinguish between legitimate and illegitimate targets
and participants. For this reason, the concept of battle deaths is readily
applied across a variety of types of conflicts. In today’s dominant forms of
conflict – civil wars, wars of insurgency, and asymmetric conflicts – the dis-
tinction between combatants and non-combatants may be very unclear or
even entirely fluid, in sharp contrast to an idealised model of a conflict fought
between formally organised state militaries. Even in wars fought by state
militaries there is an increasing reliance on private military firms, whose
personnel are not traditionally defined as combatants (Keefe, 2004). Thus a
focus on combatant deaths rather than battle deaths could seriously under-
estimate the scope of military combat in many, if not most, of today’s wars.
The number of battle deaths provides an exhaustive measure of how many
have died in combat operations. But it does not provide a remotely adequate
account of the true human costs of conflict. War kills people in less direct
(but highly predictable) ways, especially when it causes the collapse of a
society’s economy, infrastructure of health and human services, and public
safety systems. As Figure 1 lays out, the toll of a war is comprised of not only
battle deaths but deaths due to upsurges in one-sided violence (e.g. the
execution of prisoners of war or a genocidal campaign such as the Holocaust
or the Armenian Genocide); increases in criminal violence (e.g. an upsurge in
crime following the collapse of local policing, as in post-Baathist Iraq);
increases in unorganised violence (e.g. deadly food riots); and increases in
non-violent causes of mortality such as disease and starvation. A complete
accounting of the true human costs of conflict would include – in addition to
fatalities – non-fatal injuries, disability, reduced life expectancy, sexual vio-
lence, psychological trauma, displacement, loss of property and livelihood,
damage to social capital and infrastructure, environmental damage,
destruction of cultural treasures. Tallying the cost of a war quickly defies
straightforward accounting.
2
An account of war deaths must record all people killed in battle as well as
all those whose deaths were the result of the changed social conditions caused
by the war. Thus, measuring war related deaths involves comparing the
BETHANY LACINA AND NILS PETTER GLEDITSCH
148
number of deaths that occurred due to a conflict against the counterfactual
scenario of peace (Li and Wen, 2005).
3
It is necessary to judge whether
certain events – such as a famine or riot – would not have happened at all if
peace had prevailed, and to measure the degree of elevation (or depression) in
peacetime risks of mortality from factors like crime or malnutrition. Making
such estimates becomes quite difficult when there is no meaningful peacetime
benchmark to compare measured mortality rates against – as in Burma,
where civil war has been more or less continuous since independence – or
when a complex sequence of events that includes armed conflict lies behind
certain events or social changes. For example, conflict may have abetted the
spread of HIV/AIDS in Africa (Elbe, 2002), but the epidemic has other
causes as well. Separatist conflict in the Caucasus has fed into black market
economies and increased violent crime, but so has the difficult transition to a
market economy. Finally, when measuring war deaths it is difficult to
determine the relevant time frame. Indirect causes of higher mortality will
continue after the battles have stopped. How many years of elevated mor-
tality due to, for instance, depressed economic performance, environmental
degradation, or the spread of sexually transmitted diseases should be
attached to the terminated war and can those impacts be measured in a
reliable way?
Despite the definitional ambiguities of trying to account for war deaths,
study of the linkages between human insecurity and conflict and between
humanitarian crisis and conflict is vital. Throughout history wars have been
associated with humanitarian crisis. Even the most devastating instances of
battle violence in human history – World Wars I and II – are estimated to
have led to nearly as many non-battle deaths as combat fatalities, and
Battle
Deaths
WAR DEATHS
Non-Battle
Deaths
Soldiers and Civilians are
Killed in Combat
One-Sided Violence
Increases
Criminal & Unorganized
Violence Increases
Non-Violent Mortality
Increases
Figure 1. Sources of war deaths.
MONITORING TRENDS IN GLOBAL COMBAT
149
perhaps more (Clodfelter, 2002, p. 479 and 581). But although the elevated
mortality caused by war is predictable, attempts to prevent or ameliorate
non-battle impacts of war are too often thwarted by the political and security
dynamics of the conflict itself and by international indifference or clumsiness.
Also, the linkages between warfare, humanitarian crisis, and human inse-
curity have not been as widely studied as the political and military factors
that lie behind battle and battle deaths.
Some recent work has discussed or estimated the long-term public health
consequences of war (Krug et al., 2002; Murray et al., 2002; Black et al., 2003;
Ghobarah et al., 2003). There is also an increasing amount of original research
on specific populations in conflict by scholars in a variety of disciplines,
including recent studies of Afghanistan (Sliwinski, 1989; Benini and Moulton,
2004), Bosnia–Herzegovina (Brunborg et al., 2003), the Democratic Republic
of Congo (Roberts, 2000; Roberts et al., 2001, 2003), Guatemala (Ball et al.,
1999), and Rwanda (Verwimp, 2003). Relevant health and demographic data
are also continually being gathered by many humanitarian agencies, such as the
International Committee of the Red Cross (ICRC) or Epicentre, which works
with Me
´
decins Sans Frontie
`
res, in order to assess the humanitarian needs of
war-affected populations. The World Health Organization’s (WHO) Collab-
orating Centre for Research on Epidemiology and Disasters (CEDAT) is
currently building a repository for scientific studies of conflict-affected popu-
lations, and promises to become an important resource for those interested in
this topic. The second volume of the Human Security Report published by the
Centre for Human Security at the Liu Institute for Global Issues, University of
British Columbia will focus on ‘the war/disease nexus’ (Mack, 2005).
4
3. Distinguishing battle deaths from one-sided violence
Our definition of battle deaths includes a distinction between battle deaths and
one-sided violence. It may be necessary to explain in more detail why we do not
define one-sided violence to be battle-related, even though it may be of a
political character and intimately related to the issues at the heart of an ongoing
conflict. Examples of such one-sided violence include security forces firing on
unarmed protestors, summary executions of prisoners, and genocide.
Following most studies of armed conflict, we have conceived battle as a
two-sided phenomenon. Combat is political violence against any target,
military or civilian, in which the perpetrator faces the immediate threat of
lethal force being used by the opposing forces against him/her and/or allied
fighters. Again, our definition of battle deaths is empirical rather than nor-
mative. Rather than judging the legitimacy of certain targets or tactics for
collective violence as in humanitarian law (Schabas, 2001), we focus on the
degree of meaningful armed resistance. Thus, for example, terrorist attacks
against civilians were included as battle acts in our dataset because the
BETHANY LACINA AND NILS PETTER GLEDITSCH
150
perpetrators must take measures to avoid opposing security forces. The
execution of kidnapped civilians was excluded because these acts are carried
out in an environment of impunity. Words such as ‘massacre’ are sometimes
used to describe very lop-sided battle outcomes or attacks against soft tar-
gets. In general, we consider such events to be battle violence. We judge fatal
incidents to be one-sided violence – and thus exclude them from our count
of battle deaths – only when there is evidence of sustained destruction of
non-combatants taking place outside of the context of any reciprocal threat of
lethal force. The Cambodian and Rwandan genocides fall into this category.
We have attempted to make this distinction between battle violence and one-
sided violence because we believe that battle deaths are the best measure of
combat intensity. Battle deaths are not a representation of the human costs of
war, which should obviously include all violent death caused by the conflict, nor
the basis for a normative evaluation of war. Data on one-sided violence is of
critical importance for scholarship on genocide and politicide; for investiga-
tions of war crimes and crimes against humanity by courts and truth com-
missions; and for evaluation of the trends in one-sided violence and the effects
of changing international norms and institutions for evaluating and prose-
cuting war crimes.
5
Battle deaths data, however, provide a more accurate measurement of the
scope and scale of contested military engagement. On the one hand, we do
not recommend using a very narrow definition of combat deaths based on the
legitimacy of targets and tactics in order to measure the intensity of battle
violence. For example, although terrorist attacks against civilians are not a
sanctioned mode of warfare they are important to understanding the degree
of armed contest taking place in many conflicts.
On the other hand, by defining one-sided violence by the absence of armed
resistance it is possible to gain both a clear idea of the amount of military
engagement taking place (the number of battle deaths) and to study the
relationship of combat to one-sided violence. Measurements of the two are
not simply proxies for each other. For example, it has been argued that the
1994 genocide of Rwandan Tutsis could have been halted within a week by
even a small but credible United Nations force (Feil, 1998), suggesting that
the perpetrators lacked the capacity to continue their program of terror in the
face of even limited effective resistance. Demographic study has shown that
the intensity of the Rwandan genocide was greatest in those areas removed
from battles between the Hutu and Tutsi armies where Hutu Interhamwe
could act with impunity (Verwimp, 2003). Because one-sided violence often
depends on minimal threat of resistance or retaliation, battle violence and
one-sided violence have often varied inversely rather than directly, suggesting
that the relationship between them deserves careful study. Kalyvas (2004)
finds that selective execution of civilians during the Greek Civil War was
suppressed in areas of the highest combat intensity. The Khmer Rouge began
MONITORING TRENDS IN GLOBAL COMBAT
151
its ethnic and political terror, while it was still a rebel group, but the genocide
began in earnest when it won full military control of the country.
4. A new dataset of battle deaths
Several datasets have tracked fatalities in war. The best known is the list
of wars produced by the COW Project (Sarkees, 2000), which includes
interstate, intrastate, and extra-systemic (more commonly called colonial or
imperial) wars in which more than 1000 combatant deaths occurred (1000
deaths per year in the case of interstate conflict). The COW data estimate
combatant deaths by state participant, but do not disaggregate their data into
annual estimates. A related project, the Militarized International Dispute
Dataset, estimates the death toll among state militaries in smaller scale cla-
shes of a purely interstate nature. In the past, the COW and the MID projects
have drawn criticism for their exclusive focus on deaths among states’ armed
forces (Henderson, 2002). However, Lacina et al. (2005) demonstrate that the
COW data suffers from a more serious limitation: supposedly comparable
figures actually vary indiscriminately between recording combatant, battle,
and war deaths. For interstate wars, the COW project has tended to record
the number of military personnel killed in battle or the number of military
deaths from all causes, while for many extra-systemic and intrastate wars the
COW dataset estimates all war deaths, including those due to disease and
starvation. The result is that the COW data cannot be accurately compared
between types of war, decades, or regions.
Similar problems of comparability occur in fatality compilations that
attempt to record war deaths (Eckhardt, 1996; Rummel, 1997; Leitenberg,
2003). In these cases, the most serious problem is that reliable information on
war deaths in many long-terminated conflicts is simply not available and that
experts in conflict studies are often ill-equipped to estimate non-battle levels
of mortality. Reliable accounts of the earliest conflicts that appear in such
compilations are sparse or non-existent, let alone scientific demographic and
mortality data. For example, although the number of Herero and Nama
destroyed by the German military in south-western Africa in 1904–1905 was
of genocidal proportions (Hull, 2003), scholars can only speculate on the
number of people who were killed outright or forced into the desert to starve
because no pre-war census of the population was ever taken (Pakenham,
1992, pp. 614–615). There is necessarily great uncertainty in any compilation
of war deaths that attempts even modest backdating.
We present here a dataset of estimates for battle deaths in armed conflicts
from 1946 to 2002. It is, we believe, the first fatality compilation for use in the
study of armed conflict that focuses on battle deaths rather than combatant
or war deaths. As we argued above, battle deaths are the most appropriate
measure of the scale of military combat taking place in a conflict, especially
BETHANY LACINA AND NILS PETTER GLEDITSCH
152
given the fluidity of the lines between combatants and civilians in so many
armed conflicts. Tracking trends in the number of battle deaths is of great
interest to those who study patterns of combat activity world-wide and in the
context of certain eras, regions, and cases.
The new dataset was collected for the dataset produced by the Uppsala/
PRIO Conflict Data Project (Gleditsch et al., 2002; Harbom and Wallen-
steen, 2005), which records state-based armed conflicts that claim at least 25
battle deaths per year. Conflicts may be extra-systemic, interstate, internal
(i.e. civil), or internationalised internal struggles (for additional information
see the codebook: Strand et al., 2003). The new battle deaths data have also
been adapted for use with the Correlates of War data on extra-systemic,
interstate, and intrastate wars from 1900–1997 in which at least 1000 com-
batants died (Sarkees, 2000) and with the Fearon and Laitin (2001) dataset of
civil wars 1945–1999 that killed at least 1000 persons.
Our dataset draws on leading compendia of casualty statistics (e.g. Harff
and Gurr, 1988; Laffin, 1994; Bercovitch and Jackson, 1997; Rummel, 1997;
Brogan, 1998; Clodfelter, 2002; Ghosn and Palmer, 2003; State Failure Task
Force, 2003); on conflict monitoring projects (e.g. International Institute for
Strategic Studies, 2003; Project Ploughshares, 2003); on the annual tables of
major armed conflicts in the SIPRI Yearbooks (see, most recently, Wiharta
and Anthony, 2003); as well as consultations with regional experts. These
sources were augmented with studies of individual cases (e.g. Ball et al., 1999;
Sutton, 2001); archival materials from government sources (e.g. Anus
ˇ
aukas,
2000); media sources and published studies based on compiled media data
(e.g. Mueller, 1995; Dunlop, 2000); and original demographic and epidemi-
ological work where it was available.
6
5. Global trends in battle deaths: good and bad news
There has been a great deal of alarmist writing about the bloody post-Cold
War era, often painted in the media as a world of unprecedented internecine
conflict. Snow (1996, pp. 1–2, 105–113) holds, for instance, that armed
conflicts of the 1990s, which he calls ‘‘uncivil wars,’’ were less principled, less
focused on political goals, and therefore bloodier than many in the past.
Sarkees et al. (2003, p. 65) strike a similar note of pessimism, arguing that the
risk of suffering death in battle has trended neither up nor down since the
Napoleonic wars. Such remarks are surprising given that organisations that
monitor the global incidence of armed conflict have found that the late 1990s
and first years of the twenty-first century enjoyed a downward trend in
warfare between and within states (Gleditsch et al., 2002; State Failure Task
Force, 2003).
What do global trends in battle deaths tell us about the amount of armed
combat taking place in the world today? Figure 2 provides our estimate of
MONITORING TRENDS IN GLOBAL COMBAT
153
the global trend in battle deaths from 1946 to 2002 in state-based armed
conflicts, while Table 1 notes the five conflicts that inflicted the largest
numbers of battle deaths during that era. Figure 2 reveals that the long-term
trend in battle deaths has been sharply downward. The late 1940s and early
1950s were very grim; the combined impact of the Chinese Civil War and the
Korean War were augmented by the French Indochina War (estimated to
have killed 365,000 in battle) and the Greek Civil War (154,000 battle
deaths). The Vietnam War forms the next hump in the dataset, and the
combined impacts of the Iran–Iraq war and the Soviet intervention in
0
100,000
200,000
300,000
400,000
500,000
600,000
700,000
800,000
1946
1948
1950
1952
1954
1956
1958
1960
1962
1964
1966
1968
1970
1972
1974
1976
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
Year
Battle Deaths
Figure 2. Battle deaths worldwide, 1946–2002.
Table 1. Conflicts with the largest battle death totals
Conflict Years Best estimate of battle deaths*
Vietnam War 1955–1975 2,097,705*
Korean War 1950–1953 1,254,811*
Chinese Civil War 1946–1949 1,200,000
Iran–Iraq War 1980–1988 644,500
Afghan Civil War 1978–2002 562,995*
* The precision of these figures is due to exact accounting of Western losses; the accuracy
should be regarded as spurious.
BETHANY LACINA AND NILS PETTER GLEDITSCH
154
Afghanistan constitute the third peak. In the mid-1990s, the world seemed to
enter another trough in the number of battle deaths, until the war in the
Democratic Republic of Congo (DRC), with an estimated 145,000 battle
deaths in 1998–2001, and the interstate war between Ethiopia and Eritrea
that killed 50,000 from 1998 to 2000. Each peak is significantly lower than the
previous one and the trough in 2002 is the lowest for the entire period.
7
Thus, the good news about battle deaths since World War II is that
although there have been multiple major international security crises their
military scale has progressively diminished. This may reflect increasingly
pacific behavior among the great powers, who possess the resources and the
military technology (such as aerial power and heavy artillery) to inflict large
numbers of battle deaths in the wars they start, join, or provide with support.
Each of the five largest conflicts identified was a war of this type. The Korean
War and the Vietnam War were massive Cold War confrontations, and the
Chinese Civil War was also fed by superpower military assistance. The wars
between Iran and Iraq and the Soviet invasion of Afghanistan were also
driven in part by the logic of Cold War politics and the parties were armed by
the US and USSR. By contrast, declining tension between the superpowers
hastened the de-escalation of the Soviet war in Afghanistan and slowed rates
of battle deaths there in the late 1980s (Sliwinski, 1989, pp. 40–41). And while
the recent war in the DRC involved regional armies, it was not a proxy war
for major military powers. The most cataclysmic battles of the past
half-century were related to the now defunct ideological polarisation between
East and West.
The very large conflicts that these peaks represent almost overwhelm the
rest of the curve: together, the five conflicts in Table 1 constitute more than
half of the estimated toll of global battle deaths in the period 1946–2002,
accounting for about 5.76 million battle deaths out of a total of about
10 million. The five wars with the highest number of battle deaths, especially
the Vietnam and Korean Wars, are such outliers that they eclipse most of the
rest of the story of global warfare. Figure 3 shows the graph of global battle
deaths with those five largest conflicts excluded. What seemed like a down-
ward trend in battle violence vanishes and the world seems to fluctuate
between high and low war intensity years (although the final year is the
lowest for the entire period). Thus, the bad news about global battle deaths is
that, although the now-terminated Cold War played a devastating role in
driving major battle death events, there is a persistence of smaller scale, more
diverse conflicts, the trend in which is less obvious.
Even so, the data by no means support the sombre picture painted by
Sarkees et al. (2003) of a world of basically constant total rates of combat.
Table 2 shows that the mean and median numbers of battle deaths in a year
of conflict after the Cold War era were significantly lower than during the
Cold War. Nor do we find support for Snow’s hypothesis of increasingly
MONITORING TRENDS IN GLOBAL COMBAT
155
bloody civil conflicts, at least when measured according to the number of
battle deaths; the mean and median number of combat fatalities in a year of
internal or internationalised internal conflict have fallen dramatically since
the end of the Cold War.
Table 2. Battle deaths by time period and conflict type
Comparing Cold War vs. Post-Cold War conflict
All conflicts Internal conflicts*
Statistic 1946–1989 1990–2002 1946–1989 1990–2002
No. of conflict-years 1213 526 993 510
Mean 7430 2070 4250 1980
Median 500 300 439 308
Comparing types of conflict
Statistic Extrasystemic and
interstate
Internal conflicts*
No. of conflict-years 236 1503
Mean 20,620 3480
Median 945 373
* Includes internationalised internal wars.
0
20,000
40,000
60,000
80,000
100,000
120,000
140,000
160,000
1946
1948
1950
1952
1954
1956
1958
1960
1962
1964
1966
1968
1970
1972
1974
1976
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
Figure 3. Battle deaths worldwide, 1946–2002, removing the five largest conflicts.
BETHANY LACINA AND NILS PETTER GLEDITSCH
156
Civil war has been the dominant form of conflict for several decades, as
Table 3 demonstrates. For the whole period after World War II just over half
the battle deaths occurred in internal conflicts, but during the three decades
in the middle of the Cold War, there were more deaths in interstate and
extra-systemic conflicts.
8
Table 2 shows that the mean values for interstate
and extra-systemic conflict-years are much larger than for internal conflicts,
although the median value is less dramatically so. The fact that such conflicts
are increasingly rare no doubt accounts for some of the downward trend in
global battle deaths.
Figure 4 gives a regional picture of the trend in battle deaths (see Appendix
B for regional definitions). This is a stacked graph, in which each region is
represented by the area between those below and above it. The sum of the
stacked regional data provides the global estimate of battle deaths. The most
remarkable transformation in the security status of any region is that of East
and South East Asia. The greatest battle violence of the past 50 years took place
in China, Korea, and the Indochinese peninsula. By contrast, since the 1980s,
the region has been increasingly free of combat due to the de-escalation of Cold
War conflict in Indochina and the Western rapprochement with China.
In the most recent years, Sub-Saharan Africa and Central and South Asia
are the primary drivers of battle deaths. The most common conflict scenario
today is civil war and/or state failure in an impoverished society governed, if
at all, by a very weak post-colonial regime (Collier et al., 2003; Fearon and
Laitin, 2004; Mack, 2005). Most of these conflicts are neglected by major
powers, and the combatants often remain relatively ill-organised and poorly
equipped when compared to those who fought in the civil conflicts that
turned into proxy wars during the Cold War. The amount of actual military
engagement (rather than tactics of insurgency or banditry) in many modern
civil wars has been quite limited and sporadic, even desultory (Mueller,
Table 3. Share of battle deaths that occurred in internal conflicts (%)
Years Battle deaths in internal conflicts*
1946–1949 87
1950s 8
1960s 29
1970s 43
1980s 67
1990s 92
2000–02 93
1946–2002 52
*Includes internationalised internal conflicts.
MONITORING TRENDS IN GLOBAL COMBAT
157
2003). Thus these civil conflicts, though often intractable and devastating,
have produced fewer battle deaths than their Cold War counterparts.
6. Winning the battles and losing the wars?
Battle deaths do not tell the full story of the human cost of war. Although it
seems plausible to expect that the global downturn in battle violence over the
past half-century has been accompanied by some amelioration in numbers of
war deaths, we do not have the scientific data on war deaths available to test
this contention or provide a sufficient account of how war has affected
human populations over the past decades.
We suspect that war-related deaths were less severe in the late 1990s than
in previous decades primarily because the number of ongoing conflicts
declined. At the same time, it seems likely that the post-Cold War environ-
ment has been more effective in reducing the incidence of conflict and battle
deaths than in addressing the scourge of non-battle fatalities in those conflicts
that do occur. Excess mortality is probably increased because most armed
conflicts now take place in poor countries with a weak infrastructure and
limited medical facilities.
Table 4 lists nine African conflicts that caused very large numbers of non-
violent war deaths due to insecurity, displacement, deprivation, and disease.
0
100,000
200,000
300,000
400,000
500,000
600,000
700,000
800,000
1946
1948
1950
1952
1954
1956
1958
1960
1962
1964
1966
1968
1970
1972
1974
1976
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
Year
Battle Deaths
E & SE Asia & Oceania Middle East & N Africa Sub-Saharan Africa Central & S Asia Americas Europe
Figure 4. Battle deaths by region, 1946–2002.
BETHANY LACINA AND NILS PETTER GLEDITSCH
158
The estimates of battle deaths come from our dataset, the estimates of total
war related deaths are figures that are widely cited in media and conflict
literature, although, to our knowledge, only the figures for the Democratic
Republic of Congo (DRC) are based on a scientific study of the affected
population.
In the DRC, the International Rescue Committee conducted a series of
household surveys in order to estimate the impact of the internationalised
civil war fought there from 1998–2001 (Roberts, 2000; Roberts et al., 2001,
2003). The researchers estimate that roughly 350,000 persons died due to
violence in that time period, while the total toll of war-related deaths, pri-
marily driven by disease, is estimated at 2.5 million, a ratio of roughly one to
six. Combat deaths constitute an even smaller category than violent fatalities;
our best estimate of battle deaths in the DRC is 145,000 (see the DRC
conflict report within International Institute for Strategic Studies, 2003).
9
If
these figures are accurate, battle deaths constituted only about 6% of the
fatalities due to the war. Table 4 suggests that this disparity between battle
deaths and war deaths is not unique. Although there is great uncertainty
associated with these data, they strongly suggest that protracted conflicts in
poor countries claim the vast majority of their victims off the battlefield. This
is especially dramatic in cases where conflict causes famine, as has occurred in
Ethiopia and the Sudan.
Although poorly equipped and organised armies may have relatively little
capacity to cause large numbers of battle deaths or limited will to engage
other combatants, they may still be able to cause high numbers of war deaths.
In a very poor nation with weak state structures, it may not require great
Table 4. Deaths in selected conflicts in Africa
Country Years Estimates of total
war deaths
Battle
deaths
Percentage
battle dead
Sudan (Anya Nya rebellion) 1963–1973 250,000–750,000 20,000 3–8%
Nigeria (Biafra Rebellion) 1967–1970 500,000 to 2 million 75,000 4–15%
Angola 1975–2002 1.5 million 160,475 11%
Ethiopia (not inc.
Eritrean insurgency)
1976–1991 1–2 million 16,000 <2%
Mozambique 1976–1992 500,000 to 1 million 145,400 15–29%
Somalia 1981–1996 250,000 to 350,000
(to mid-1990s)
66,750 19–27%
Sudan 1983–2002 2 million 55,500 3%
Liberia 1989–1996 150,000–200,000 23,500 12–16%
Democratic Republic
of Congo
1998–2001 2.5 million 145,000 6%
MONITORING TRENDS IN GLOBAL COMBAT
159
military capacity to collapse the infrastructure of health and human security
and cause a full-blown humanitarian crisis. For example, a small force can
cut transportation links vital to food security, as demonstrated by the rela-
tively limited military intervention required to break the siege of Mogadishu
and relieve famine in Somalia in 1992–1993 (United Nations, 1996). With a
greater percentage of contemporary wars being internal conflicts in poor
states, it is likely that the number of global battle deaths has fallen far more
precipitously than the count of war deaths. All of the conflicts in Table 4
occurred in Africa, the region that now accounts for the greatest share in
global battle violence. In other regions, such as Central and South Asia and
Europe, it has also been weak and poor states that have fallen prey to
conflict.
Despite optimism in the early 1990s regarding the prospects for peace
enforcement by the newly unified community of major powers acting through
the United Nations, experiences with preventing the massive humanitarian
crises caused by war have been mixed. The UN deployment to Somalia ended
in humiliation, as did missions to the Balkans and Rwanda, dimming
enthusiasm for multilateral international intervention in civil conflicts. The
major powers have few strategic interests in many of the most conflict-prone
regions, and major humanitarian crises in West Africa and the DRC have
been largely ignored. In the summer of 2004 and early 2005, the grim scenario
of a displaced population at risk and a humanitarian relief process stymied
by insecurity and international inattention was playing out again in the
Darfur region of the Sudan. Today the nexus between conflict, one-sided
violence, and humanitarian crisis seems more important than ever.
7. Improving knowledge about deaths in war time
Media reports, government accounts, military data, and the analysis of his-
torians and political scientists all play a role in establishing statistics for
conflict deaths. However, the fields of demography and epidemiology are
better equipped to scientifically examine and describe the impacts of armed
conflict on human population, especially studying impacts beyond mortality
such as changes in fertility or migration. The tools of scientific studies of
population have already been used to provide data on a few recent conflicts
that is of a precision and clarity far beyond that offered by other sources.
Where possible, we have made use of such data to compile a dataset of
battle deaths over the past half-century. The results reveal interesting trends
in global battle violence, considered both in terms of the impacts of the end
of the Cold War and the changing fortunes of geographic regions. The
declining numbers of major interstate conflicts and internationalised civil
wars have led to a decline in global battle deaths. Presently, most warfare is
in the form of civil conflict and wars of state failure taking place outside of
BETHANY LACINA AND NILS PETTER GLEDITSCH
160
areas of the major powers’ strategic interest. We expect that many of these
conflicts will be characterised more by severe humanitarian crises than
combat of the intensity seen during the Cold War. Thus, we are encouraged
to see an increasing number of studies on the demography of conflict and
conflict and public health. Security and conflict analysts will increasingly
require such expertise in order to understand and address the true human
costs of war.
Notes
z
The work reported here has been carried out in collaboration with a number of colleagues
at the Centre for the Study of Civil War, the Uppsala Conflict Data Project, and the Centre for
Human Security at the Liu Institute for Global Issues, University of British Columbia. We are
grateful to our colleagues at all these three institutions as well as to Andy Mack, Daniel
Mun
˜
os-Rojas, Michael Spagat, and Juan Vargas for their comments, encouragement, and
constructive criticisms. The referees and editors of this journal also provided valuable
comments. Our work has been funded mainly by the Research Council of Norway, with
additional contributions from the Centre for Human Security. The dataset described here,
along with detailed documentation, can be downloaded from www.prio.no/cscw/cross/
battledeaths.
1
Throughout this article, we will use the terms ‘armed conflict,’ ‘conflict,’ and ‘war’ as
synonyms. We also use ‘battle’ and ‘combat’ synonymously to refer to the act of military
contest taking place within an armed conflict. For the more precise definition of state-based
armed conflict upon which our dataset of battle deaths draws see Appendix A.
2
For discussion of the long-term economic consequences of civil war, see the World Bank
report on civil war (Collier et al., 2003) as well as Murdoch and Sandler (2002). For a project
that undertakes comprehensive estimates of the impacts of war, see State Failure Task Force
(2003).
3
Of course, an estimate of battle deaths also involves comparison with a counterfactual. For
example, mortality rates among U.S. soldiers during the first Gulf War may actually have been
lower than those among similar cohorts living in the U.S. (Wolfson and Smith, 1993). In most
cases, however, the counterfactual scenario can probably be ignored when analyzing figures
for battle deaths.
4
A planning report for the second volume of the Human Security Report (Mack, 2005)
discusses the need to bring demography, public health, and epidemiology into conflict studies.
See Centre for Human Security (2004).
5
While studies of genocide and politicide (Harff, 2003) find relationships with some of the
same factors that predict to the onset of internal war (Fearon and Laitin, 2001; Hegre et al.,
2001), the explanatory models are not identical. See also Valentino et al. (2004) on why civil
wars involve differing amounts of targeting of civilians.
6
The dataset may contain much that is incomplete or inaccurate, but it has been compiled in
the hope of constant improvement based on users’ feedback. We have documented our source
materials and coding decisions for each armed conflict. In some cases, we provide low and high
estimates of battle deaths in a conflict, along with our best estimate.
7
The invasion and occupation of Iraq will undoubtedly form a new peak in the incidence of
battle deaths beginning in 2003. It is not clear whether the final toll in battle deaths will
surpass that in the DRC. On the basis of a sample survey, Roberts et al. (2004) estimate total
excess death among Iraqis in the 18 months following invasion at 100,000. This is seven times
the number of Iraqi citizens killed by coalition forces according to the leading monitoring
MONITORING TRENDS IN GLOBAL COMBAT
161
project based on press reports (www.iraqbodycount.net), but far less than the millions
estimated killed by all war-related causes in the DRC. However, the war in Iraq seems to have
claimed a far higher percentage of its victims through violence, especially aerial strikes.
8
This figure is strongly influenced by the coding of the Vietnam War. In the Uppsala/PRIO
data this conflict is an internal war from 1955–1964 and an interstate war from 1965–1975.
9
The authors of the IRC reports are quite explicit in emphasizing the criminal and/or
one-sided character of most deaths through violence they investigated.
10
A detailed list of references to sources used in compiling the dataset (but not cited in this
article) is found on www.prio.no/cscw/cross/battledeaths.
Appendix A: Definition of battle deaths
Our definition of battle deaths closely follows the definition of conflict used to create the
Uppsala/PRIO Armed Conflict Dataset (Gleditsch et al., 2002; Harbom and Wallensteen,
2005). According to codebook for the Uppsala/PRIO dataset (Strand et al., 2003, pp. 3–4):
• ‘‘An armed conflict is a contested incompatibility that concerns government and/or
territory where the use of armed force between two parties, of which at least one is the
government of a state, results in at least 25 battle-related deaths.’’
• The separate elements of the definition are operationalised as follows:
• Use of armed force : use of arms in order to promote the parties’ general position in the
conflict, resulting in deaths.
• Arms: any material means, e.g. manufactured weapons but also sticks, stones, fire, water,
etc.
• 25 deaths: a minimum of 25 battle-related deaths per year and per incompatibility.
• Party: a government of a state or any opposition organisation or alliance of opposition
organisations.
• Government: the party controlling the capital of the state.
• Opposition organisation: any non-governmental group of people having announced a name
for their group and using armed force.
• State: a state is an internationally recognised sovereign government controlling a specified
territory, or an internationally unrecognised government controlling a specified territory
whose sovereignty is not disputed by another internationally recognised sovereign
government previously controlling the same territory.
• Incompatibility concerning government and/or territory: the incompatibility, as stated by the
parties, must concern government and/or territory.
• Incompatibility: the stated generally incompatible positions.
• Incompatibility concerning government: incompatibility concerning type of political system,
the replacement of the central government, or the change of its composition.
• Incompatibility concerning territory: incompatibility concerning the status of a territory,
e.g. the change of the state in control of a certain territory (interstate conflict), secession, or
autonomy (internal conflict).
The Lacina and Gleditsch dataset defines the following terms:
• Battle deaths are deaths resulting directly from violence inflicted through the use of armed
force by a party to an armed conflict during contested combat.
• Contested combat is use of armed force by a party to an armed conflict against any person
or target during which the perpetrator faces the immediate threat of lethal force being used
by another party to the conflict against him/her and/or allied fighters. Contested combat
excludes the sustained destruction of soldiers or civilians outside of the context of any
reciprocal threat of lethal force (e.g. execution of prisoners of war).
BETHANY LACINA AND NILS PETTER GLEDITSCH
162
• Timeframe: we have collected annual battle deaths data which includes both deaths during
combat and deaths from wounds received in combat. Some of those considered dead of
wounds may have died in a year following that in which combat actually took place,
especially in the case of battles taking place late in the calendar year. These deaths were
included, however, if they were the direct and immediate result of injuries sustained during
combat violence. Long-term reduction in life expectancy because of wounds or disability
was not included.
Appendix B: Regional definitions
• Africa, Sub-Saharan: Angola, Benin, Botswana, Burkina Faso, Burundi, Cameroon, Cape
Verde, Central African Republic, Chad, Comoros, Congo (Brazzaville), Democratic
Republic of Congo (Zaire), Coˆ te d’Ivoire, Djibouti, Equatorial Guinea, Eritrea, Ethiopia,
Gabon, Gambia, Ghana, Guinea, Guinea-Bissau, Kenya, Lesotho, Liberia, Madagascar,
Malawi, Mali, Mauritania, Mauritius, Mozambique, Namibia, Niger, Nigeria, Rwanda,
Senegal, Sierra Leone, Somalia, South Africa, Swaziland, Tanzania, Togo, Uganda,
Zambia, Zanzibar, Zimbabwe.
• Americas: Argentina, Bahamas, Barbados, Belize, Bolivia, Brazil, Canada, Chile, Colom-
bia, Costa Rica, Cuba, Dominican Republic, Ecuador, El Salvador, Guatemala, Guyana,
Haiti, Honduras, Jamaica, Mexico, Nicaragua, Panama, Paraguay, Peru, Suriname,
Trinidad and Tobago, United States of America, Uruguay, Venezuela.
• Asia, Central and South: Afghanistan, Armenia, Azerbaijan, Bangladesh, Bhutan, Georgia,
India, Kazakhstan, Kyrgyzstan, Nepal, Pakistan, Sri Lanka, Tajikistan, Turkmenistan,
Uzbekistan.
• Asia, East and Southeast and Oceania: Australia, Brunei, Cambodia, China, East Timor,
Fiji, Indonesia, Japan, People’s Republic of Korea (North Korea), Republic of Korea
(South Korea), Laos, Malaysia, Maldives, Mongolia, Myanmar (Burma), New Zealand,
Papua New Guinea, Philippines, Singapore, Solomon Islands, Taiwan, Thailand,
Democratic Republic of Vietnam (North Vietnam), Republic of Vietnam (South Vietnam).
• Europe: Albania, Austria, Belarus, Belgium, Bosnia Herzegovina, Bulgaria, Croatia,
Cyprus, Czech Republic, Czechoslovakia, Denmark, Estonia, Finland, France, German
Democratic Republic (East Germany), German Federal Republic (West Germany),
Greece, Hungary, Iceland, Ireland, Italy, Latvia, Lithuania, Luxembourg, Macedonia,
Malta, Moldova, Netherlands, Norway, Poland, Portugal, Rumania, Russia (Soviet
Union), Slovakia, Slovenia, Spain, Sweden, Switzerland, Ukraine, United Kingdom,
Yugoslavia (Serbia and Montenegro).
• Middle East and North Africa: Algeria, Bahrain, Egypt, Iran, Iraq, Israel, Jordan, Kuwait,
Lebanon, Libya, Morocco, Oman, Qatar, Saudi Arabia, Sudan, Syria, Tunisia, Turkey,
United Arab Emirates, Yemen (Arab Republic), Yemen (People’s Republic).
References
10
Anus
ˇ
aukas, A., 2000. ‘A comparison of the armed struggles for independence in the Baltic States and
Western Ukraine’, in A. Anus
ˇ
aukas (ed.), The Anti-Soviet Resistance in the Baltic States. Du Ka:
Vilnius, 63–70.
Ball, P., Kobrak, P. and Spirer, H. F., 1999. State Violence in Guatemala, 1960–1996: A Quantitative
Reflection. Annapolis, MD: American Association for the Advancement of Science.
Benini, A. A. and Moulton, L. H., 2004. ‘The distribution of civilian victims in an asymmetrical conflict:
Operation Enduring Freedom, Afghanistan’, Journal of Peace Research 41(4): 403–422.
MONITORING TRENDS IN GLOBAL COMBAT
163
Bercovitch, J. and Jackson, R., 1997. ‘International Conflict: a Chronological Encyclopedia of Conflicts and
their Management 1945–1995’. Washington, DC: Congressional Quarterly.
Black, R., Morris, S. and Bryce, J., 2003. ‘Where and why are 10 million children dying every year?’,
Lancet 361: 2226–2234.
Brogan, P., 1998. World Conflicts. Lanham, MD: Scarecrow.
Brunborg, H., Lyngstad, T. H. and Urdal, H., 2003. ‘Accounting for genocide: how many were killed in
Srebrenica?’, European Journal of Population 19(3): 229–248.
Centre for Human Security, 2004. Deadly connections: the war/disease nexus workshop report, Liu
Institute for Global Issues, University of British Columbia, http://www.humansecuritycentre.org/
Deadly%20Connections%20Meeting%20Report.doc.
Chesterman, S., 2001. ‘Introduction: global norms, local contexts’, in S. Chesterman (ed.), Civilians in
War. Boulder, CO: Lynne Rienner.
Clodfelter, M., 2002. Warfare and Armed Conflicts: A Statistical Reference to Casualty and Other Figures,
1500–2000. Jefferson, NC.: McFarland.
Collier, P., Elliott, L., Hegre, H., Hoeffler, A., Reynal-Querol, M. and Sambanis, N., 2003. Breaking the
Conflict Trap: Civil War and Development Policy. Oxford: Oxford University Press.
Dunlop, J., 2000. ‘How many soldiers and civilians died during the Russo-Chechen war of 1994–1996?’,
Central Asian Survey 19(3/4): 329–339.
Eckhardt, W., 1996. ‘Wars and war-related deaths, 1900–1995’, in R. L. Sivard (ed.), World Military and
Social Expenditures 1996. Washington, DC: World Priorities, 17–19.
Elbe, S., 2002. ‘HIV/AIDS and the changing landscape of war in Africa’, International Security 27(2):
159–177.
Fearon, J. D. and Laitin, D. D., 2001. ‘Ethnicity, insurgency, and civil war’, American Political Science
Review 97(1): 75–90.
Fearon, J. D. and Laitin, D. D., 2004. ‘Neotrusteeship and the problem of weak states’, International
Security 28(4): 5–43.
Feil, Col. S. R., 1998. Preventing Genocide: How the Early Use of Force might have Succeeded in Rwanda.
Washington, DC: Carnegie Corporation of New York.
Ghobarah, H.A., Huth, P.K. and Russett, B., 2003. ‘Civil wars kill and maim people–long after the
shooting stops’, American Political Science Review 97(2): 189–202.
Ghosn, F. and Palmer, G., 2003. Militarized interstate dispute data, version 3.0, Correlates of War 2
Project, http://cow2.la.psu.edu.
Gleditsch, N. P., Wallensteen, P., Eriksson, M., Sollenberg, M. and Strand, H., 2002. ‘Armed conflict
1946–2001: a new dataset’, Journal of Peace Research 39(5): 615–637.
Harbom, L. and P. Wallensteen, 2005. ‘Armed conflict and its international dimensions, 1946–2004’,
Journal of Peace Research 42(5): 623–635.
Harff, B., 2003. ‘No lessons learned from the holocaust? Assessing risks of genocide and political mass
murder since 1955’, American Political Science Review 97(1): 57–73.
Harff, B. and Gurr, T. R., 1988. ‘Toward empirical theory of genocides and politicides: identification and
measurement of cases since 1945’, International Studies Quarterly 32(3): 359–371.
Hegre, H., Ellingsen, T., Gleditsch, N. P. and Gates, S., 2001. ‘Towards a democratic civil peace?
Democracy, political change and civil war, 1816–1992’, American Political Science Review 95(1): 33–
48.
Henderson, E., 2002. Democracy and War: The End of an Illusion. Boulder, CO: Lynne Rienner.
Hull, I. V., 2003. ‘Military culture and the production of final solutions in the colonies: the example of
Wilhelminian Germany’, in R. Gellately and B. Kiernan (eds), The Specter of Genocide: Mass Murder
in Historical Perspective. Cambridge: Cambridge University Press, 141–162.
International Institute for Strategic Studies, 2003. Armed conflict database, International Institute for
Strategic Studies, http://acd.iiss.org/armedconflict/.
Kalyvas, S., 2004. Techniques of violence in Greece and Vietnam. Unpublished paper presented at the
Workshop on Techniques of Violence in Civil War, Centre for the Study of Civil War, PRIO, Oslo,
20–21 August.
Keefe, P. R., 2004. ‘Iraq: America’s private armies’, The New York Review of Books LI(13): 48–50.
Krug, E. G., Dahlberg, L. L., Mercy, J. A., Zwi, A. B. and Lozano, R., 2002. World Report on Violence
and Health. Geneva: World Health Organization.
BETHANY LACINA AND NILS PETTER GLEDITSCH
164
Lacina, B., Gleditsch, N. P. and Russett, B., 2005. The declining risk of death in battle. Unpublished paper
presented at the 2005 Annual Convention of the International Studies Association, Honolulu, HI, 1–5
March, www.prio.no/cscw/cross/battledeaths.
Laffin, J., 1994. The World in Conflict: War Annual 6. London: Brassey’s.
Leitenberg, M., 2003. Death in Wars and Conflicts Between 1945 and 2000. Ithaca, NY: Cornell University.
Li, Quan and Ming Wen, (2005). ‘The immediate and lingering effects of armed conflict on adult mortality:
a time-series cross-national analysis’, Journal of Peace Research 42(4): 471–492.
Mack, A. (ed.), 2005. Human Security Report. New York: Oxford University Press, two volumes, in press.
Mueller, J., 1995. ‘The perfect enemy: assessing the Gulf War’, Security Studies 5(1): 77–117.
Mueller, J., 2003. ‘Policing the remnants of war’, Journal of Peace Research 40(5): 507–518.
Murdoch, J. C. and Sandler, T., 2002. ‘Economic growth, civil wars, and spatial spillovers’, Journal of
Conflict Resolution 46(1): 91–110.
Murray, Ch. J. L., King, G., Lopez, A. D., Tomijima, N. and Krug, E .G., 2002, ‘Armed conflict as a
public health problem’, BMJ 324(7333): 346–349.
Pakenham, T., 1992. The Scramble for Africa, 1876–1912. London: Weidenfeld and Nicolson.
Project Ploughshares, 2003. Armed conflicts report 2003, Project Ploughshares, http://www.plough-
shares.ca/content/ACR/ACR00/ACR00.html.
Roberts, L., 2000. Mortality in Eastern DRC: Results from Five Mortality Surveys. Bukavu/New York:
International Rescue Committee.
Roberts, L., Hale, Ch., Belyakdoumi, F., Cobey, L., Ondeko, R., Despines, M., IRC DRC Bukavu/
Kisangani and Keys, J., 2001. Mortality in Eastern Democratic Republic of Congo: Results from Eleven
Mortality Surveys. Bukavu/New York: International Rescue Committee.
Roberts, L., Ngoy, P., Mone, C., Lubula, Ch., Mwezse, L., Zantop, M. and Despines, M., 2003. Mortality
in the Democratic Republic of Congo: Results from a Nationwide Survey. Bukavu/New York:
International Rescue Committee.
Roberts, L., Lafta, R., Garfield, R., Khudhairi, J. and Burnham, G., 2004. ‘Mortality before and after the
invasion of Iraq in 2003’, The Lancet 364(9448): 1857–1864.
Rummel, R. J., 1997. Statistics of Democide: Genocide and Mass Murder since 1900. Rutgers, NJ:
Transaction Publishers.
Sarkees, M. R., 2000. ‘The Correlates of War data on war: an update to 1997’, Conflict Management and
Peace Science 18(1): 123–144.
Sarkees, M. R., Wayman, F. W. and Singer, J. D., 2003. ‘Inter-state, intra-state, and extra-state wars: a
comprehensive look at their distribution over time, 1816–1997’, International Studies Quarterly 47:
49–70.
Schabas, W. A., 2001. Introduction to the International Criminal Court. Cambridge: Cambridge University
Press.
Sivard, R. L., 1996. World Military and Social Expenditures 1996. Washington, DC: World Priorities.
Sliwinski, M., 1989. ‘Afghanistan: the decimation of a people’, Orbis 33(1): 39–56.
Snow, D. M., 1996. Uncivil Wars: International Security and the New Internal Conflicts. Boulder, CO:
Lynne Rienner.
State Failure Task Force, 2003. State failure problem set, 1955–2001. Center for International
Development and Conflict Management at the University of Maryland and Integrated Network for
Societal Conflict Research, http://www.cidcm.umd.edu/inscr/stfail/sfdata.htm.
Strand, H., Wilhelmsen, L. and Gleditsch, N. P., in cooperation with Wallensteen, P., Eriksson, M. and
Sollenberg, M., 2003. Armed conflict dataset codebook, version 2.0.: International Peace Research
Institute, http://www.prio.no/cwp/ArmedConflict/current/Codebook_v2_0.pdf.
Sutton, M., 2001. An index of deaths from the conflict in Northern Ireland. ARK: Northern Ireland Social
and Political Archive: Conflict Archive on the Internet (CAIN), http://cain.ulst.ac.uk/sutton/
index.html.
United Nations, 1996. The United Nations and Somalia, 1992–1996. United Nations, New York:
Department of Public Information.
Valentino, B., Huth, P. K. and Balch-Lindsay, D., 2004. ‘Draining the sea: mass killing and guerilla
warfare’, International Organization 58(2): 375–407.
Verwimp, P., 2003. ‘Testing the double-genocide thesis for Central and Southern Rwanda’, Journal of
Conflict Resolution 47(4): 423–442.
MONITORING TRENDS IN GLOBAL COMBAT
165
Walzer, M., 1977. ‘Just and Unjust Wars: A Moral Argument with Historical Illustrations’. New York:
Basic Books.
Wiharta, S. and Anthony, I., 2003. ‘Major armed conflicts’, in: SIPRI Yearbook 2003: World Armaments
and Disarmament. Oxford University Press, Oxford 87–125.
Wolfson, M. and Smith, R., 1993. ‘How not to pay for the war’, Defence Economics 4(4): 299–314.
BETHANY LACINA AND NILS PETTER GLEDITSCH
166
