The Civilians to Combatants Killed Ratio in Gaza Is Probably At Least 3.4:1.

Debunking Eli Lake

Introduction

The journalist Eli Lake (pictured above) has written a piece for Bari Weiss’ The Free Press. Lake’s piece attempts to discredit the common accusations that the death toll in Gaza is extremely high, and that the overwhelming majority of those killed are civilians. It has made the rounds with the usual suspects. Andrew Sullivan, who has been getting in touch with his neoconservative side through his recent coverage of the Israel-Gaza war, celebrates Lake as having “largely debunk[ed] the civilian casualty stats coming out of Gaza.”

In his analysis, Lake makes two main claims.

First, Lake contends that “only” a slight majority of 54.3% of killed Gazans are women and children, in contrast to the figure of about 70% women and children cited by the Ministry of Health in Gaza (MoH). Second, as a result of the lower figure of 54.3% women and children among the killed Gazans, Lake contends that

the ratio between combatants and civilians killed in Gaza is closer to one to one, a figure much closer to counterinsurgency wars America and its allies have fought in the Middle East.

Let's assume for argument's sake claim 1 is true: that is, as Lake (citing analysis from the AP) suggests, the figures for unidentified deaths are falsified, and only the identified deaths are accurate. I reject this claim,¹ but I’ll assume it’s true for purposes of this analysis. Therefore, I join Lake in assuming the percentage of killed Gazans who are women and minors is not 70%, but that only a narrow majority of killed Gazans are women and children.

I will focus instead on debunking his second claim, that “the ratio between combatants and civilians killed in Gaza is closer to one to one, a figure much closer to counterinsurgency wars America and its allies have fought in the Middle East." This inference is bogus. Indeed, as I will show, Lake’s own data suggest that the civilians to combatants killed ratio (CCR) in Gaza will be at least 3.4:1, and most likely higher or much higher.

My estimate of a CCR of least 3.4:1 follows from two assumptions.

Two Basic Assumptions

First, I assume that demographic groups of killed Gazans literally 100% of whom—according to the B’Tselem database—were civilians in the previous two major Gaza wars are also all or virtually all civilians in the current war. These specific 100% civilian demographics are males 0-14, males 65 and over, and females of all ages.²

Second, I assume that least as many Gazan male civilians 15-64 have been killed as have female civilians 15-64.

I’ll reiterate these assumptions later, but it’s still important to keep them in mind for now. For now, Let’s get back to Lake’s analysis.

Lake’s Bungled Demographic Analysis

In inferring a ≈1:1 CCR, Lake seems to assume that "men" are basically all combatants and "women and children" are basically all civilians. Thus, he suggests that (1) the roughly 46% of the 24,685³ identified killed Gazans (through the end of April) who were men are combatants; and (2) the roughly 54% of 24,685 identified killed Gazans who were women and children are civilians, for a roughly 1:1 ratio.

Lake’s assumptions are specious. First, the “killed men” category includes elderly men, none or virtually none of whom are going to be combatants. Second, the “children” category includes older male minors—teens 15-18⁴—among whom there are very likely to be many combatants. Finally, it is completely erroneous to assume, as Lake (with his inference of an approximately 1:1 CCR) seems to do, that all or almost all combat-aged adult men kill will be combatants.

Better Demographic Categories

A better demographic breakdown than Lake’s would separate killed Gazans into the following two categories: “Males 15-64,” and “Everybody Else.” This demographic breakdown will help us sus out combatants vs civilians, because the former category will contain virtually all the combatants, while the latter will be virtually entirely composed of civilians.

The “Everybody Else” category comprises three-subcategories: Male Children 14 and Under, Elderly Males 65 or Over, and All Females regardless of age. As I previously noted, in the two major recent Gaza Wars, Cast Lead (2008-2009) and Protective Edge (2014), literally 100% of all killed Gazans in these three-subcategories were civilians, according to the B’Tselem database.

Conversely, the killed males 15-64 category will be mixed, comprised of many combatants and civilians. In Cast Lead and Protective Edge, there were plenty of combatants and plenty of civilians among killed Gazan males 15-64.

I have spent weeks looking through, counting, and recounting⁵ the data of identified killed Gazans through the end of April—these are the same data Lake is citing—and can break them down as follows: Males 15-64 equals 46.4% of the total killed Gazans (11,453/24,685), and Everyone Else equals 53.6% of the total killed Gazans (13,232/24,865). (“Everyone Else” includes 8,996 Females of All Ages, 3,463 Males 0-14, and 793 Males 65 and over, for a total of 13,232.)

Our goal here is to estimate the percentage of killed identified Gazans who were civilians. Based on the historical record, we can define 13,232 or 53.6% of killed Gazans—our “Everyone Else” category, namely all females, males 14 and under, and males 65 and up—as all or virtually all civilians. So, before we address killed Gazan males 15-64, we start with approximately 13,232 Gazans or 53.6% of the total killed in Lake’s data being civilians.

What of killed Gazans males 15-64? Among them, there will be plenty of civilians and also plenty of combatants. We can give a low-ball estimate of how many of them are civilians by assuming that at least as many civilian males 15-64 were killed as female civilians 15-64.

How Many Gazan Civilians 15-64 Have Been Killed?

So we will use our figure for killed female civilians 15-64 to estimate the number of male civilians 15-64 killed. But how can we determine the former figure? We can do so because literally 100% of killed females (15-64, and of all ages) have been civilians in the previous two major recent Gaza wars. Thus, we can assume female civilians 15-64 killed=approximately 100% of all females 15-64 killed, or approximately 5,835.

A core assumption I’m making here is that least as many male civilians 15-64 were killed as females. How can I know that?

In the first place it’s important to highlight and resist a tempting fallacy. Specifically, it is tempting to infer from the fact that the percentage of females killed who are civilians is much higher than the percentage of men killed who are civilians, that a higher raw number of female civilians are killed than male. But this would be as illogical as inferring that, because the percentage of one-year-olds killed in war who are civilians (100%) is much higher than the percentage of adult males killed who are civilians, that a larger raw number of civilians killed in war are babies than adult male civilians.

When we look at the actual historical record, we find that male civilians are much more likely to be killed in war than female. This is because men are going to be more likely to be out and about in war and running risks than women. (We see this all the time in clips of the streets in Gaza, where civilians roaming about are skewed male.) And civilian men of military age are also more likely to be wrongly suspected of militant activity and killed for this reason. 

This phenomenon of many more male civilians killed than female is borne out in statistics of wartime casualties in general, and in the previous Gaza Wars more specifically. For example, in Operation Protective Edge (2014), 878 male civilians were killed versus 491 female civilians, according to B’Tselem. So 79% more male civilians died than female. Among killed Gazans in our critical demographic (aged 15-64), 592 male civilians were killed versus 280 female civilians, a difference of 111%. These differences—and the over-representation of overall male civilians and male civilians 15-64 killed relative to females—are even more dramatic for Cast Lead.

Thus, it is extremely reasonable and conservative to assume that at least as many Gazan males civilians 15-64 have been killed as Gazan female civilians of the same age. This is the assumption we will make in pursuit of a lower-bound (“at least”) estimate. Therefore, of the 11,453 Gazan males 15-64 killed, we assume that at least 5,835 were civilians, equal to our figure of female civilians 15-64 killed.

Final Calculation

Let us finally calculate our estimate of the percentage of civilians among the identified killed Gazans in the data set cited by Eli Lake.

If we add up all killed Gazans from our “Everyone Else” demographics—males 14 and under, males 65 and over, and females of all ages—literally 100% of whom⁶ were civilians in the two major recent Gaza Wars, we get, as I noted above, 13,232 Gazans killed. Approximately all of them are civilians. If we add this figure of 13,232 to the at least 5,835 civilian males 15-64 killed, we get a total of at least 19,067 civilians killed, or 77.2% of the total (19,067/24,685).

Meanwhile, our estimate for combatant Gazan males aged 15-64 equals 11,453 (all Gazan males killed who were 15-64) – 5,835 (our minimum estimate for killed civilian males 15-64)=at most 5,618. Based on the historical data described above, all other categories of killed Gazans are projected to be 100% civilian. So at most a total of 22.8% of killed Gazans in Eli Lake’s data are combatants.

This equates to a civilians to combatants ratio of at least 3.4:1 (77.2/22.8). Rather different from Eli Lake’s apologetic claim of about a 1:1 ratio.

3.4:1 is an ultraconservative, lower-bound estimate. The ratio is probably higher or even much higher. After all, it is overwhelmingly likely that the number of male civilians 15-64 killed is not equal to the number of killed female civilians 15-64, but greater or much greater. Moreover, we are for argument’s sake following Eli Lake (and the AP) in discarding the data of unidentified killed Gazans; if we included these, the ratio would surely be much higher than 3.4:1.

The Implications of a 3.4:1 (or Worse) Ratio

It is important to contextualize what a civilians-to-combatants killed ratio of at least 3.4:1 means. According to a popular myth—sometimes repeated by ignorant bureaucrats at the EU and UN, and leveraged by propagandists for Israel in the current war—90% of persons killed in war are civilians. But this nonsense. As the eminent international-relations scholar Adam Roberts has shown in an essay debunking the myth, 9:1 is massively higher than the average, and since World War II only genocidal and arguably genocidal wars have had such a ratio. I would add that in World War II, even the genocidal urban battles of the Eastern Front, such as Stalingrad and Leningrad, had ratios well below 9:1.

Our estimate of at least 3.4:1 for Israel would be a hideous ratio. The siege of Aleppo, whose main protagonists were war criminals Bashar Al-Assad and Vladimir Putin, had a ratio of 3:1, i.e. considerably better than our lower-bound projection for Israel in Gaza. As to the American counter-terrorism campaigns Eli Lake scurrilously compares to the razing of Gaza, the solid majority of persons killed by the Coalition against ISIS in Mosul and Raqqa were combatants. This is despite the fact that the ISIS enemy engaged in human shielding and these battles were fought in densely populated areas (conditions we are told excuse the horrible civilian death toll inflicted by the IDF).

1. I believe that both categories of killed Gazans—unidentified deaths and identified deaths—suffer from sampling biases which lead to a relative under-representation of males in the former, and over-representation in the latter. Hence, I expect that the overall percentage of women and children among the dead is higher than the identified death data suggest, and lower than the unidentified death data suggest, but that neither figure is faked. Rather, these data reflect real deaths whose demographic balances are distorted by sampling biases. There are two sampling biases I believe likely exist.

First, regarding unidentified deaths (which are largely drawn from media reports), I believe that the press is more likely to report on airstrikes that disproportionately kill women and children. The existence of such a sampling bias would be consistent with the well-documented phenomenon of women and children victims of war getting more public sympathy and notice than adult male victims. This sampling bias would lead to an over-representation of women and children among non-identified deaths, but would not entail the reporting of faked deaths.

Second, regarding identified deaths, I believe that (for logistical reasons) killed Hamas fighters are more likely to be fully reported and identified to the MoH than are killed Gazan civilians. Since Hamas fighters are nearly all adult males, this would predictably lead to a relative over-representation of men in the reported deaths, at least since November, when the MoH began having to rely on media reports to document casualties. Assuming the MoH is not double counting, this sampling bias would also—along with the previous one I mentioned—help explain the radical under-representation of adult males among the unidentified Gazan deaths. After all, Hamas men have already been counted in the identified deaths, thereby (unless the MoH is not engaged in double counting) reducing the number of potential pool of killed men who could be covered in the media reports.

2. I include two female police officers killed in Cast Lead as civilians. There is zero evidence that these two women took up arms or joined militant groups; hence they are civilians under the Law of Armed Conflict.

3. I exclude one child listed by the MoH whose age is not identified. Hence my total figure for identified killed Gazan children—and identified killed Gazans—through 30 April is one fewer than the MoH’s 7,797.

4. The MoH, perhaps for the propagandistic reason of ensuring that a majority of killed Gazans remain women and children, counts 18-year-olds (but not 19-year-olds) as “children.”

5. I thank my friend Veronica Brouchard for her research assistance in this regard.

6. According to B’Tselem.