This depiction of Israeli military tactics, in particular the equivalence drawn between Israeli and Palestinian tactics, in turn provides cover for the embrace of revisionist interpretations of current Israeli security policy. Thus, insofar as Israel's current goal is allegedly "to dominate, expel, starve, and humiliate an entire people,"4 current Israeli policy does one of two things: it systematically targets noncombatants in an effort to terrorize the Palestinians into submission; or, in a scarcely less egregious version, it makes an insufficient effort to avoid the systematic and indiscriminate targeting of noncombatants. In either case, the results, it is claimed, are the same: the killing of "huge numbers of children, women, and elderly."5
Is this true? What does an analysis of mortality data show? It shows that the allegedly huge numbers of noncombatant casualties are in fact fictional. Israeli military strikes, it turns out, have been remarkably discriminate, in marked contrast to those of Palestinians.
The Case against Israel
Two types of evidence have been used to indict Israel. The first refers to specific deaths and instances of human rights abuse. This is the case-study approach to analysis. It draws its strength from highly charged descriptive or visual anecdotes, which are both easy grist for journalistic mills and also highly effective public relations tools. The televised death of 11-year-old Muhammad ad-Dura, caught with his father in the crossfire of a sudden gun battle in the early stages of the violence, is a prime example of the power of this approach. Highly emotive, it was a catalyst for demonstrations throughout the Arab world. More general visual stimuli of this sort include footage of Israeli tanks arrayed against Palestinian stone-throwers, lines of Palestinian cars at Israeli checkpoints, etc. These purport to give an accurate flavor of life under occupation. Finally, this category also includes reports by human rights groups that profess to have identified several cases in which the Israeli Defense Forces (IDF) have used excessive force.6
The second piece of evidence is the total number of deaths in the conflict. The recitation of simple aggregates - x Palestinians and y Israelis have died - is now a formulaic part of media coverage of this conflict. Thus far, roughly three times as many Palestinians have died as Israelis. This implicitly provides a basis for extrapolating from the case studies noted above, suggesting that more Palestinian deaths are directly equivalent to more Palestinian noncombatant deaths.
While rhetorically potent, these arguments are not in the least sufficient to establish that Israel systematically and indiscriminately targets noncombatants. Making that leap on the basis of these arguments disregards the most basic inferential rules in social science.
The key problem with the case-study approach is that it provides no reliable basis for generalization, which in turn means that it is unclear how accurately either selected poster children or alleged instances of human rights abuse summarize the parameters of the conflict. On what basis can or should one generalize from the (contested) case of Muhammad ad-Dura? (Especially since some evidence suggests he was killed by Palestinian fire.) Or from tanks parked outside a Palestinian town? Or from the alleged human rights abuses, whose easy investigation (and concentrated journalistic coverage) is a direct product of Israeli democratic freedoms? A single taste, to carry the metaphor a little further, may not represent the overall flavor.
But it does drive the battle over image. In typical Palestinian accounts, the "Israeli war machine" has perpetrated "all forms of atrocities" against the Palestinian people, including the indiscriminate shelling of civilian areas, the shooting of civilians, the assassination of legitimate political or resistance leaders, and the mining of routes followed by school children. These are the tactics allegedly responsible for the supposedly huge numbers of children, women, and elderly casualties.
Typical Israeli accounts are, of course, very different. Its attacks on Palestinian Authority (PA) targets, Israel asserts, have been measured, proportionate, solely in self-defense and have made every reasonable effort to limit casualties among noncombatants.
In fact, the only way to verify any of these claims is to explore the distribution of all deaths across age groups and sex (i.e., what demographers refer to as a disaggregation of mortality). It is a simple analytic approach that, in this case, allows for the identification of sectors of the population that are indisputably noncombatant.
The Mortality Data
We have compiled lists of all Israeli and Palestinian deaths caused directly by violence in the 16 months between September 29, 2000, and January 31, 2002, from several sources. The variation in total number killed between the various lists is primarily caused by different editorial strategies with respect to who should be included in the list. Associated Press's list, for example, includes suicide bombers and alleged Palestinian collaborators with Israel who were killed by Palestinians, while Reuters figures do not.7 With the exception of the Israeli government, however, the only source that contains anything other than a simple aggregate (that is, any descriptive data about each casualty) is B'Tselem, the Israeli Information Center for Human Rights in the Occupied Territories.
As its subtitle suggests, B'Tselem is primarily concerned with violations of rights in the West Bank and Gaza (henceforth referred to as the territories). Since the beginning of the current round of violence in September 2000, it has noted deaths of all Palestinians and Israelis killed both in the territories and inside the Green Line, that is, within Israel's pre-1967 international borders. (It also has data on all Palestinian deaths going back to 1993.) The database is updated monthly.
Database information about each of the deceased includes name, age, place of death/event, and cause of death (e.g., killed by gunshot, explosion, missile, knife, etc.). Gender is not explicitly given in the English version of the site but is easily inferred from the parallel Hebrew site since each entry is descriptive rather than numeric, and Hebrew differentiates between gender, thus allowing us to identify gender through linguistic form, cross-checking it with the apparent gender of the name (Hebrew and Arabic names, other than nicknames, also differentiate clearly between gender).
B'Tselem's data on Israeli deaths correspond to data made available by the Israeli foreign ministry, so we assume that they are fully accurate.8 The Palestinian data may be slightly less so, but we assume that any inaccuracies are too marginal to affect our estimates.9This assumption is based on three things. First, the wide media coverage of events within the Green Line and in the territories means that it is unlikely that any deaths go unreported, making it relatively easy for B'Tselem to collect these data. Second, B'Tselem claims to expend considerable effort in verifying each of those deaths through the careful cross-checking of its own agents' fieldwork with relevant documents, official government sources, and information from other sources, among them Israeli, Palestinian, and other human rights organizations.10
Finally, B'Tselem's self-referential claims are essentially trustworthy because B'Tselem cares about its international reputation. It is a past winner of the Carter-Menil Award for Human Rights and has received financial support from numerous foreign governments and international non-governmental organizations (NGOs). Its reports have frequently been cited by other leading international human rights organizations, such as Human Rights Watch, Amnesty International, and Derechos. Similarly, both its publications and data are also referred to in official Palestinian Authority reports and in reports by Palestinian NGOs such as the Palestinian Society for the Protection of Human Rights and the Environment (PSPHRE) and the Palestinian Human Rights Monitoring Group (PHRMG).11 In short, if there are any errors in the B'Tselem data on Palestinian mortality, they are more likely to overestimate than underestimate the number of deaths.
Late Spring 2002, marked by waves of Palestinian suicide attacks and a subsequent Israeli offensive, saw heightened mortality on both sides. Although B'Tselem data are not yet available for the whole of this period, currently available data through the end of March show almost identical age- and sex-specific patterns as presented below in relation to the first 16 months. These data are available from the authors.
The coming months will reveal whether or not these patterns were maintained throughout the Israeli offensive. If the Jenin experience is indicative - Human Rights Watch declared that no massacre or systematic killing of Palestinian civilians by Israeli forces occurred during heavy house-to-house fighting, notwithstanding Palestinian claims and heavy international criticism of Israel - then it can reasonably be assumed that the pattern is more or less fixed.
We use 12 of the 15 categories used by B'Tselem, all of which refer to deaths of Israeli and Palestinian civilians, soldiers, and other security or paramilitary forces in the 16-month period ending January 31, 2002. All three of the 15 B'Tselem categories that are not used refer to deaths of foreigners, that is, 12 individuals not normally resident in Israel or the PA, four of whom were killed by Israeli security forces, three by Palestinian civilians in the territories, and the remaining four by Palestinian civilians within the Green Line. The data also do not include:
- 47 suicide bombers who died in the 16-month period, since these people cannot reasonably be considered victims of the violence;12
- 29 Palestinians suspected of collaborating with Israel, all of whom are male;
- 13 Palestinian citizens of Israel, all of whom are male;13
- 22 Palestinians (11 of whom are female) who, it has been claimed, have died from medical complications after too long a delay at Israeli checkpoints (though we include these in some tangential estimations in order to show that their exclusion does not substantively change any of the estimates);
- An unspecified number of Israelis killed in the increased number of road-traffic accidents. The number of fatal road accidents in Israel increased 14 percent between 1999 and 2001, leading to 61 more deaths.14This increase is considered to be a direct product of the violence, since repeated ambushes of Israeli civilian drivers have both increased the speed at which people drive and led to a reduced police enforcement of speed restrictions in certain areas.
As implied above, the key analytic aim is to explore mortality among groups who, we can reasonably be assured, are noncombatants. Ideally, we would adopt B'Tselem's own categories since their lists claim to differentiate between deaths of civilians and security forces. Unfortunately, B'Tselem's definitions of each of these two are dangerously misleading.
The key problem is that they largely disregard the nature of the conflict and the context of given attacks. The sole determinant of a person's civilian status in the B'Tselem lists is whether they belong to a uniformed military or paramilitary group that officially represents either Israel or the PA. However reasonable this may sound in the abstract (presumably it is intended to establish unambiguous criteria), it ignores a fundamental difference between Israeli and Palestinian civilian categories. In fact, there are thousands of non-uniformed Palestinian combatants who are not members of the official PA security forces but act in the name of Hamas, Islamic Jihad, Tanzim, and so on. The largest category in B'Tselem's lists, for example, "Civilian Palestinian Casualties in the Territories," is therefore an absurd amalgam of women, children, the elderly, and young men killed, for example, while activating an explosive device against a passing bus (Muhammad ‘Imad and Mazin Badawi, killed January 31, 2002, in an attack for which Hamas subsequently claimed responsibility).15 Indeed, Palestinian attackers have been listed as civilians even when they have been killed while wearing Israeli uniforms (e.g., attacks of October 2 and 26, 2001). Israeli security forces are, of course, all members of a conventional armed force and so, by definition, are classed as combatants - even if they were not killed in combat.
Because B'Tselem makes no distinction between combatant and noncombatant civilians, there is a fundamental mismatch between the structure of the B'Tselem data and the structural characteristics of the conflict. The lists engage in a statistical deceit of the highest order since they structure the data in a way that prevents true comparison of Israeli and Palestinian civilian deaths. This, in turn, instantly inflates the number of Palestinian civilians killed, directly implicating Israeli policy.
Rather than adopt these specious categories, we therefore substitute a more inductive approach that first identifies sectors of the population that are highly unlikely to include combatants, and then explores mortality in those groups.
PA chairman Yasir Arafat himself, in a speech in Doha, identified the three subgroups whose deaths are emotive: children, women, and the elderly. This provides us with our starting point. Women and the elderly, we assume, can safely be considered noncombatants during the period under survey, since there were no reports that members of these groups had been involved in fighting. (Only on January 29, 2002 - two days before the end of the period under survey - did the first female suicide bomber detonate herself.)
Children are more problematic, however, since there is considerable evidence that children of various ages have been heavily involved in the violence. It is crucial to review that evidence.
A standard Israeli position since September 2000 has been that (a) young Palestinian teens and children actively participate in violent activities, especially in the throwing of Molotov cocktails and stones; and (b) Palestinian gunmen have frequently fired at Israeli positions from among these young demonstrators, inviting Israeli return fire.16 Journalists, non-partisan international observers, and human rights monitors have tended to agree with these claims. In August 2000, before the outbreak of the violence, The New York Times reported the existence of summer camps in which 27,000 Palestinian children had participated the past summer, learning guerrilla tactics, how to operate firearms, practicing kidnapping, etc.17 In October 2000, the United Nation's Children's Fund (UNICEF) urged the PA to take energetic measures to discourage those under 18 years of age from participating in any violent action because such action places them at risk.18 In an early B'Tselem report, for example, a foreign journalist reports filming confrontations between Palestinians and Israeli security forces in El-Bireh in October 2000:
Suddenly a blue commercial vehicle appeared and stopped around 20 meters away from us, some 30 meters from young Palestinians who were at the front of the demonstration. Three Palestinians, 20 to 30-years-old, were inside. They called to the children, gave orders, and distributed Molotov cocktails. I asked my photographer to film it. One of the children noticed, shouted out a warning, and within 15 seconds we were surrounded. The vehicle drove ahead 20 meters and stopped. The three men inside ran to the back and snatched the camera from the photographer. One of them shouted, "Kill, kill."19Indeed, wide media coverage of the PA's use of children even generated some debate in Arabic-language newspapers about the ethics and religious legality of either allowing or encouraging children to attack Israelis and prompted subsequent demands that the PA limit the participation of children and close down the special training camps.20
In response, Yasir ‘Abd Rabbu, the PA's information minister, asserted that, as of October 2000, Palestinian political parties and forces decided to prevent children and youngsters under 16 from participating and agreed to establish field committees (which will be present at the locations) to implement the decision.21 Follow-up fieldwork by B'Tselem, however, drawing both on testimonies of children participating in demonstrations and direct observation of those demonstrations, found no evidence to indicate that these official guidelines were being followed. PA officials at these locations, the authors asserted, made no serious effort to prevent children from reaching the site of demonstrations or from participating in them. And they made no attempt to move children away so that they would not be injured in Israeli return fire.22 This situation appears to have continued through 2001,23prompting a reiterated call in November 2001 by UNICEF's special representative to the West Bank, Gaza, and Jerusalem for the PA to expand measures to discourage those under 18 from participating in any violent action.24
In short, Palestinian children have participated in this round of violence from its earliest stages. While it is impossible to generalize about the frequency of their participation (these, too, are case-studies, and so share the flaws listed above), children as a group cannot be considered noncombatants. Or at least this is true for older children. Only younger children, we can reasonably assume, have not been actors in the violence.
In summary, a conservative definition of groups that we can safely assume to be noncombatant would include, women, the elderly, meaning those past fighting age, arbitrarily set at 55 and over, and all children aged less than 10. Together, these groups account for 69.9 percent of the Palestinian population, and 68.1 percent of the Israel population.25
Net of these exclusions, the B'Tselem lists include 1,004 deaths between September 29, 2000, and January 31, 2002. Of these, 755 (75.2 percent) were Palestinian and 249 were Israeli.
Mortality data for both the Palestinian and Israeli populations are presented in Table 1. Out of the 755 Palestinian deaths, only 19 were females aged 10-54, 10 were elderly (males or female older than 55), and eight were young children aged less than 10. The distribution is, therefore, completely dominated by deaths among Palestinian prime-age and adolescent males. They account for 718 deaths, or 95.1 percent of all Palestinian deaths. Non-combatant groups of females, elderly, and young children together account for only 4.9 percent of the total number of deaths, even though they represent more than two-thirds of the population. These mortality patterns, therefore, clearly do not support the Palestinian claim that there have been huge numbers of children, female, and elderly casualties.
Table 1. Mortality among noncombatant groups as percentage of total of national mortality, by demographic group and nationality.
More random mortality patterns with respect to sex and age can be seen among Israelis. Of the 249 Israeli deaths, for example, there were 67 deaths to females aged 10-54, 33 to the elderly, and five to young children. Deaths of Israeli males aged 10-54 therefore accounted for only a small majority (57.8 percent) of all deaths, leaving 105 deaths (42.2 percent) in the noncombatant categories.
These data also suggest that there is a complete inversion of the combatant versus non-combatant distribution. This is shown explicitly in Figure 1, which presents the distribution of deaths among adult males, females, and the elderly. It shows that while 83.3 percent of all deaths among males aged 10-54 have been Palestinian, 77.9 and 76.7 percent, respectively, of all female and elderly deaths in the 16-month period have been Israeli. This is a highly significant crossover in the distribution.
Figure 1: Distribution of mortality among men, women, and the elderly, by nationality.
Nor do the data support another frequent argument: that Palestinian noncombatant deaths inevitably result from Israeli missile attacks on, or shelling of military targets in, civilian areas. Of the 37 Palestinian noncombatant deaths, only five (two women, one older person, and two children) were caused by a shell, missile, or some other type of explosion. All others were caused by gunshot.
Palestinian claims, journalistic summaries, Kofi Annan's comments, and instances of excessive force to the contrary, the mortality data show no sign of systematic targeting of Palestinian civilians by Israeli forces. Nor do they show any signs that the Israeli forces are systematically failing to avoid targeting Palestinian civilians. On the contrary, the fact that less than 5 percent of Palestinian casualties are either women, elderly, or young children, in comparison to more than 40 percent of Israeli casualties, supports the key Israeli claim: that the higher number of Palestinian deaths reflects the high number of Palestinian attacks on Israeli targets, not the reverse. This is the reason for the huge bulge in male mortality from late childhood through middle age among the Palestinians.
The data also suggest that Israeli and Palestinian military tactics in this conflict are completely dissimilar. Israel appears to have systematically targeted Palestinian combatants, while the latter have been more concerned with targeting any Israeli, combatant or civilian. If Kofi Annan, the editors of The New York Times, Time, and others are interested in retailing more than Palestinian rhetoric, they should acknowledge this point and cease drawing invalid parallels between Israeli and Palestinian tactics. They also need to rethink the simplistic assumption that firing upon combatants in civilian areas is necessarily indiscriminate and causes high numbers of noncombatant casualties. Even in built-up areas, it is possible to wage discriminating warfare.
More generally, the differences in Israeli and Palestinian mortality highlighted in this paper suggest that opinion-makers, leaders, and commentators in particular, need to extrapolate more warily from various instances of excessive or indiscriminate use of force and also cast a more critical eye over single aggregates. Put another way, responsible commentators must use data responsibly. At a minimum, this means asking about the legitimacy of given types of generalization and highlighting heterogeneity in statistical distributions. In this case, we think, the heterogeneous mortality patterns signal a fundamental difference between Israeli and Palestinian military cultures. One side appears to draw a careful distinction between combatants and others. The other does not.
Alexander A. Weinreb is National Institutes of Child Health and Human Development (NICHD) Postdoctoral Fellow at the Population Research Center & NORC, University of Chicago. Avi Weinreb is financial journalist at Globes,Israel's business newspaper.1 Kofi Annan, United Nations Security Council, Mar. 11, 2002, SG/SM8159, SC/7325.
2 The New York Times, Mar. 12, 2002.
3 Time, Mar. 12, 2002.
4 "Courage to Refuse, Combatant's Letter," at http://www.seruv.org.il.
5 Yasir Arafat, 2001, speech to the meeting of the ministers of foreign affairs of the Organization of Islamic Conference, Doha, Qatar, May 26, 2001, at http://www.palestine-pmc.com/statments/26_5_2001_arafat_speech.html.
6 B'Tselem press release, "29.7.01: In Broad Daylight: Abuse of Palestinians by IDF Soldiers on July 23, 2001," at http://www.btselem.org/English/Press_Releases/2001/010729.asp; and Excessive Force: Human Rights Violations during IDF Actions in Area A (Jerusalem: B'Tselem, 2001), at http://www.btselem.org/Download/Excessive_Force_Eng.doc.
7 Associate Press, personal communication, Mar. 5 2002.
8 Data at the Israeli foreign ministry at http://www.israel-mfa.gov.il/mfa/go.asp?MFAH0ia50.
9 It is normal for data sets to include some error, but it has to be significant in scale or biased in its distribution to affect results in the type of simple bivariate analysis that we conduct here. These issues are discussed extensively by Richard A. Zeller and Edward G. Carmines, Measurement in the Social Sciences: The Link between Theory and Data (New York: Cambridge University Press, 1980).
10 "About B'Tselem," at http://www.btselem.org/English/About_BTselem/index.asp.
11 See PA reports, at http://www.pna.net, and Palestinian NGO reports, at http://www.lawsociety.org/Intifada2000/articles/btselem.htm and http://www.phrmg.org/monitor2001/feb2001.htm.
12 Of these 47, there was one woman and one bomber aged more than 40, so their exclusion has no effect on the overall age-and sex-specific mortality patterns described below.
13 The totals in these last three categories are from "Summary of Palestinian Fatalities, 29/09/2000 to 02/03/2002," Palestinian Human Rights Monitoring Group (PHRMG), at http://www.phrmg.org/#Summary%20of%20Palestinian%20fatalities, as accessed on Mar. 5, 2002.
14 "Road Accidents with Casualties, and Casualties by Severity," Israel Central Bureau of Statistics, 2002, at http://www.cbs.gov.il/hodaot2002/21_02_17t1.htm.
15 Hamas communiqué, at http://22.214.171.124/hamas/communiques/comm_text/2002/31jan02.htm.
16 "A Year of Violence: Overview of the Violence in the Territories, September 2000-September 2001," Israel Defense Forces, 2001, at http://www.idf.il/geut/english/main.html.
17The New York Times, Aug. 3, 2000.
18 Statement by Carol Bellamy, UNICEF executive director, to the Special Session of the Commission on Human Rights, at http://www.unicef.org/newsline/00prmideast.htm.
19 Illusions of Restraint: Human Rights Violations during the Events in the Occupied Territories, 29 September-2 December 2000 (Jerusalem: B'Tselem, 2000), at http://www.btselem.org/Download/Illusions_of_Restraint_Eng.zip, p. 42.
20 Ash-Sharq al-Awsat, Oct. 27, 2000, at http://memri.org/bin/articles.cgi?Page=archives&Area=sd&ID=SP14700;Al-Hayat,Apr. 16, 2001, at http://memri.org/bin/articles.cgi?Page=archives&Area=sd&ID=SP20601.
21 Illusions of Restraint, p. 19.
22 Ibid.,p. 14.
23 Dougherty, Jon E. and David Kupelian, "Children of the Jihad: Palestinian Kids Raised for War Taught to Hate, Kill Jews through ‘Sesame Street'-Type TV Show," WorldNet Daily, Nov. 3, 2000, at http://www.worldnetdaily.com/news/article.asp?ARTICLE_ID=17707.
24 UNICEF press release, Nov. 15, 2001, at http://www.unicef.org/newsline/01pr87.htm.
25 These proportions are based on data from the Palestinian Bureau of Statistics, covering population in the territories (i.e., combined West Bank and Gaza) in the 1997 Palestinian census, and from the Israeli Census Bureau, covering Israel's population in 2000.