QUANTIFYING GENOCIDE: Darfur Mortality Update, August 6, 2010

What we learn from the new report by “Darfurian Voices” (July 14, 2010)

Eric Reeves
August 6, 2010

(this update is available, on request, as a Word document attachment, preserving
all quantitative formatting, word and number highlighting, and links)

In the late summer of 2004, during the most violently destructive phase of the
Darfur genocide, the US State Department commissioned the International
Coalition for Justice (CIJ) to oversee a systematic interviewing of Darfuri refugees
who had fled to eastern Chad. It was on the basis of the report that emerged from
these interviews (“Documenting Atrocities in Darfur” at
http://www.reliefweb.int/rw/rwb.nsf/db900sid/KHII-64S2UH?OpenDocument )
that Secretary of State Colin Powell would make his September 2004
determination that genocide was being committed in Darfur. The personnel
conducting the research included human rights experts, law enforcement officials,
genocide scholars, forensic experts, and those with significant experience in the
tribunals for the former Yugoslavia and Rwanda. They were provided with ample
resources, including a full complement of translators.

On the basis of 1,136 carefully randomized interviews, conducted among the
Darfuri refugee population in Chad at 19 camp locations along the border, the
Coalition for International Justice (CIJ) found that “sixty-one percent [of those
interviewed] reported witnessing the killing of a family member.” The
implications of this extraordinary figure were registered by only a very few among
those who purported to speak about levels of mortality in Darfur. My own efforts
to draw implications from these data (first in fall 2004,
http://www.sudanreeves.org/Sections-article226-p1.html ) were followed by those
of Professors John Hagan and Patricia Parker (April 2005, commissioned by the
Coalition for International Justice) and by Dr. Jan Coebergh, an independent
researcher (http://h-net.msu.edu/cgi-bin/logbrowse.pl?trx=vx&list=h-
w= ). They yielded comparable results using different statistical methodologies.

But because “Documenting Atrocities in Darfur” was not specifically designed as a
mortality study, its consequential findings were dismissed by professional
epidemiologists, in nearly all cases with painfully little contextual understanding of
the Darfur conflict—and ultimately by the US Government Accounting Office,
which similarly dismissed the data from “Documenting Atrocities in Darfur” and
the reports that utilized them. This ham-fisted dismissal has been smugly accepted
as definitive in some quarters. My own finding was that by the beginning of 2005,
215,000 people had been killed (this based primarily on CIJ data) and that another
200,000 had died as a result of disease and malnutrition directly consequent upon
ethnically-targeted violence (my primary sources for this latter figure were UN
World Health Organization mortality studies, particularly of August/September
2004; see http://www.sudanreeves.org/Sections-article482-p1.html ). Total
mortality, I argued, was over 400,000 (400,000 was the figure that Hagan and
Parker would arrive at several months later, though making assumptions that I
believe worked to understate violent morality and overstate mortality from disease
and malnutrition).

There have been no subsequent mortality studies that have used the CIJ mortality

But those data have recently (July 14, 2010) found remarkable corroboration and
amplification in a report entitled “Darfurian Voices”
(http://www.darfurianvoices.org/ ). The report indicates that 72% of Darfuri
refugees (in a population of approximately 250,000) had one or more (up to ten)
“immediate family members (parents, siblings, spouses, children) [who] were
killed in attacks related to the current conflict” (page 14).

What are the statistical implications of these figures (pages 14 – 15 of the report)?

Let us assume for clarity in treating percentage data 100 respondents (the actual
number of respondents was approximately 1,250; see citation of “Footnote 2” in
Appendix 1b); let us assume as well a rather large average immediate family size
of 10 (see Appendix 2), so that the 100 respondents represent a total statistical
cohort of 100 families or 1,000 refugees; and finally, let us assume no duplication
in family representation in the survey, an assumption warranted by the statistical
methodology of the survey.

[specific percentage values here are from the bar graph on page 14, rounded to
nearest percentage integer]:

Percentage of 100 families
who experienced “x deaths”
per immediate family:

0 deaths 28% = ( 0 deaths within the statistical cohort of 1000)
1 death 20% = (20 deaths within the statistical cohort) (1 x 20)
2 deaths 16% = (32 deaths within the statistical cohort) (2 x 16)
3 deaths 12% = (36 deaths within the statistical cohort) (3 x 12)
4 deaths 8% = (32 deaths within the statistical cohort) (4 x 8)
5 deaths 5% = (25 deaths within the statistical cohort) (5 x 5)
6 deaths 3% = (18 deaths within the statistical cohort) (6 x 3)
7 deaths 3% = (21 deaths within the statistical cohort) (7 x 3)
8 deaths 2% = (16 deaths within the statistical cohort) (8 x 2)
9 deaths 1% = (9 deaths within the statistical cohort) (9 x 1)
10 deaths 2% = (20 deaths within the statistical cohort) (10 x 2)

100% of 229 total deaths within the statistical cohort
100 families (1,000 refugees)

The statistics represented here in simplified form are derived from approximately
1,250 interviews of persons 18 years of age and older residing in eight of the
twelve official Darfur refugee camps (see Appendix 1c). The report uses a
“stratified random sampling approach” (though see Appendix 1 concerning
selection of leadership figures). The survey was able to field a full complement of
translators (speaking Arabic as well as Fur, Massalit, Zaghawa, and other African

The report assumes a total Darfuri refugee population of approximately
250,000 (249,744), based on a May 2009 calculation by the UN High Commission
for Refugees. (The “Darfurian Voices” data were collected between April and July

Assuming no family duplication in representation (again, such duplication is highly
unlikely given the sampling method) and an average immediate family size of 10:

For 100 respondents—each representing 10 family members, for a total
represented population of 1,000 refugees—there is confirmation of 229
“immediate family members (parents, siblings, spouses, children) [who] were
killed in attacks related to the current conflict” (in other words, 22.9% of the
1,000 represented refugee population).

If we assume that this figure is fully statistically representative for Darfuri refugees
in the camps of eastern Chad (as it was designed to be), the total figure for violent
mortality experienced by this population is approximately 57,250 (229 x 250 [250
= 250,000 ÷ 1,000 persons] as of July 2009. Alternatively, we may calculate
22.9% of 250,000—again 57,250. If we assume a smaller average size for families
in the camp, the total number of deaths increases proportionately. (Such an
assumption does not increase the number of family members killed in the sense of
numbers per family, but rather increases our estimate of the number of families
represented in the camp, and thus the total number of deaths; see Appendix 2.)


Since the total displaced population in eastern Chad and Darfur has grown over
more than seven years, we require an average figure on which to base any
extrapolation to the internally displaced populations inside Darfur. Using UN
figures, such averaging moves from a starting number of 0 (zero) for early 2003 to
approximately 1.66 million in December 2004 to approximately 2 million in
October 2006 to the present figure of 2.7 million, first reported by the UN in
October 2008 and relevant through the gathering of “Darfurian Voices” data in
April – July 2009. Finding a precise average number of displaced persons,
including refugees, with proper weighting of particular time-frames, is not feasible.
More troublingly—from a statistical perspective—displacement is less and less
directly related to violent attacks—and thus violent mortality—after 2004,
although it is still typically related to concerns about insecurity, and significant
violent mortality continues. But if precision is impossible, reasonable estimates
can be made from available figures for displacement.

If we assume an average internally displaced population of 830,000 (830,000 = (0
+ 1.66 million) ÷ 2) for the period of February 2003 to December 2004, and if we
assume that the Chadian refugee mortality figure is fully representative of this
population, then we may estimate that the total number of family members killed is
229 x 830 (830 = 825,000 ÷ 1,000 persons) or almost 190,000 (190,070)
“immediate family members (parents, siblings, spouses, children) [who] were
killed in attacks related to the current conflict.” Alternatively, we may calculate
22.9% of 825,000—again 190,000 (190,070).

The average total of Internally Displaced Persons (IDPs) for January 2005 to
October 2006 may be estimated as 1.83 million (1.83 million = (1.65 million + 2
million) ÷ 2). Thus the average increase in the internal displacement population
for the period of January 2005 to October 2006 is 1.01 million (that is, this figure
represents the increase from the average population of internally displaced persons
for February 2003 through December 2004: 1.83 million minus 825,000). If we
assume that by this point in the conflict the Chadian refugee violent mortality rate
overstates by a factor of 5 violent mortality among internally displaced Darfuris,
then we may estimate that the total number of family members killed is (229 ÷ 5) x
1,010 (1,010 = 1.01 million ÷ 1,000 persons) or approximately 46,000 (46,258)
“immediate family members (parents, siblings, spouses, children) [who] were
killed in attacks related to the current conflict.” Alternatively, we may calculate
(22.9% ÷ 5) of 1.01 million—again 46,000 (46,258).

The average total of IDPs for November 2006 to July 2009 may be estimated as
2.35 million (2.35 = (2 million + 2 .7 million) ÷ 2). Thus the average increase in
the internal displacement population for the period of November 2006 to July 2009
is 520,000 (that is, this figure represents the increase from the average population
of internally displaced persons for January 2005 to October 2006: 2.35 million
minus 1.83 million). If we assume that by this point in the conflict the Chadian
refugee violent mortality rate overstates by a factor of 10 violent mortality among
internally displaced Darfuris, then we may estimate that the total number of family
members killed is (229 ÷ 10) x 520 (520 = 520,000 ÷ 1,000 persons) or
approximately 12,000 (11,900) “immediate family members (parents, siblings,
spouses, children) [who] were killed in attacks related to the current conflict.”
Alternatively, we may calculate (22.9% ÷ 10) x 520,000—again 12,000 (11,900).

Violent mortality through July 2009 is thus estimated to be:

57,250: violent mortality among Darfuri refugees in Chad
190,000: violent mortality among Darfuri IDPs, February 2003 – December 2004
46,000: violent mortality among Darfuri IDPs, January 2005 – October 2006
12,000: violent mortality among Darfuri IDPs, November 2006 – July 2009

305,250: total violent mortality among displaced Darfuris through July 2009

Obviously these calculations all depend upon how we answer a basic question:
how representative are Darfuri refugees in eastern Chad of the displaced
population within Darfur itself? And how does the answer to this question change
over the course of the conflict? “Darfurian Voices” attempts no extrapolation of
the sort offered here, confining the relevance of all their data on all issues to
eastern Chad alone. But the question of larger relevance is imperative, given the
paucity of data on violent mortality in Darfur, especially in the first year of the
conflict. Moreover, we are not completely without suggestive, if more localized,

One important consideration in answering this central question is the reason for
civilian flight. Nearly all the refugees in eastern Chad fled from violence. For an
analysis of the relationship between violence and human displacement within
Darfur itself, especially valuable data come from an early study published in The
Lancet (October, 1, 2004, “Violence and mortality in West Darfur, 2003-2004” at
http://www.ncbi.nlm.nih.gov/pubmed/15474133 ). This study suggests an
extremely high correlation between violent displacement and displacement per se,
precisely what we find among refugees in eastern Chad. (In fact, this study
presents data indicating that violence within Darfur during the first two years of
conflict is associated even more closely with displacement than among the Darfuri
refugees as reported in “Darfurian Voices.”) Thus it supports in particular the
calculations here for violent mortality for the period of February 2003 to December

Certainly we may safely assume that the 1,250 interviews conducted by “Darfurian
Voices” are highly statistically relevant, especially given the ethnic and
geographical range of those interviewed (virtually all were from non-Arab or
African tribal groups, with 82% from the Massalit, Zaghawa, and Fur tribes; eight
geographically distributed camps were represented). At the same time, there is no
purely statistical justification for these deflator factors of 5 and 10; they are
estimates based on close textual analysis of many hundreds of reports of violence
during the periods in question.


Although the “Darfurian Voices” study is deeply impressive in its research, it must
also be noted that, as was the case with the CIJ study, there are several limiting
factors in any statistical assessment of the responses that were recorded, factors
that might well drive total violent mortality within displaced Darfuri populations
either higher or lower (see Appendix 2):

[1] Neither the CIJ study nor the “Darfurian Voices” study can take into account
families in which all members were killed, and who thus had no reporting presence
in the camps where interviews took place. There is circumstantial evidence that
this number may be significant (e.g., the survival of only one member of a large
family would suggest that this family came perilously close to being

[2] Further, families with small numbers of survivors are under-represented in the
statistical sampling vis-à-vis families that survived largely intact; to the extent that
these small families of survivors have been diminished by killings, their under-
representation works to understate total violent mortality.

[3] The CIJ study reports that 28% of those interviewed “directly witnessed”
persons dying from the consequences of displacement before reaching Chad. These
deaths must be considered the direct consequence of violence, if not violent deaths
per se, and would significantly increase violent mortality totals suggested by the
CIJ data. There is no comparable set of disaggregated figures in the “Darfurian
Voices” report; such deaths, the direct consequence of antecedent violence, may or
may not be included in the phrase “were killed in attacks related to the current
conflict.” If they are not included, then the total violent mortality figure would rise

[4] The possibility for double-counting of families in using the “Darfurian Voices”
survey data cannot be overlooked. If there are approximately 25,000 refugee
families in eastern Chad, we cannot be sure to what degree they are separated
geographically. Thus, if statistical method ensures that there was no overlap
between the families within the “Darfurian Voices” sample, the same cannot be
said of families represented in the sample but that are also part of the population
total not included in the sample (by virtue of separation and geographic dispersal).
Here there may be overlap and in effect “double counting.”

The only situation that would eliminate the problem is one in which each refugee
family, in its entirety, lived in the same location. The UN High Commission for
Refugees has done a great deal of work over the past six years in re-uniting refugee
families within the twelve camps in eastern Chad, but to the degree they have not
been fully successful, a kind of double-counting through overlap may have
occurred in my statistical extrapolation. The degree of such overlap is unclear, but
I believe it is relatively small.

One other possible source of double-counting is the “double role” some individuals
may play in different “immediate families,” i.e., someone might be both a father in
one family and brother to a father in another family. If both brothers were selected
for their responses in the survey, their reports would double one another. Again,
given the randomizing strategy of the survey, this seems quite unlikely to produce
a significant reduction in estimated total mortality.

[5] The UN High Commission for Refugees reports than more than 500,000 have
been newly displaced since January 1, 2008 (the point at which UNAMID formally
took over protection activities in Darfur). Within this population are people who
have been displaced two, three, even four times (although these multiple
displacements do not add to the total figure for internal displacement:
approximately 2.7 million). Violent mortality is certainly associated with these
additional displacement experiences, often the direct result of renewed violence
against civilians within Darfur; however, we have no data by which to quantify this
mortality. It may well be significant, as suggested by the large displacement
among previously displaced populations in the Muhajeria area of South Darfur in
January/February 2009
eria )

[6] Given the relative security of the Chadian refugee camps vis-à-vis the much
larger displaced populations inside Darfur, violent morality may actually have been
higher within Darfur at times during the years after 2004, when the refugee exodus
was largely over (90% of the Darfuri refugee population had moved into eastern
Chad by the end of 2004). Nonetheless, the assumptions here in calculating violent
mortality among internally displaced Darfuris are that the violent mortality rate for
the Chadian refugee population overstates by a factor of five for the period January
2005 – October 2006 and by a factor of ten for the period November 2006 – July
2009. (Deadly cross-border raids by Janjaweed militia and Khartoum’s regular
Sudan Armed Forces beginning in 2005 may or may not be included in the
“Darfurian Voices” report; if they are not, this would also suggest a higher violent
mortality total for Darfuri refugees.)

[7] There was significant ethnically-targeted violent mortality in Darfur in the
years immediately prior to the outbreak of war in February 2003. We have no data
that permits quantification of a total, but there are a number of important reports
indicating the scale of the violence. For example, Julie Flint and Alex de Waal in
Darfur: A Short History of a Long War (2005) note that in “October 2002,
government-supported Janjawiid from the camps in South Darfur launched a major
offensive, the first of its kind against Fur civilians” (page 64). Many civilians were
killed and “hundreds of villages burned.” They note also that in 1998 Janjaweed
forces had attacked Dar Massalit, the Massalit homeland in West Darfur, killing
more than a thousand people and burning some 30 villages in a three-month period
(pages 60 – 61). There are many other examples.


Almost all studies of mortality issues find that the majority of “excess deaths” in
Darfur after the initial phase of the counterinsurgency (2003 – 2004) are a function
not of direct violent acts but the health consequences of violence and the
displacement it has produced. This has been the consistent conclusion of my own
work after 2004. If in fact 304,250 Darfuris have died violently, and more than
240,000 have died from causes other than violent killing (see mortality study from
The Lancet, January 2010, below), then total mortality in Darfur and eastern
Chad now exceeds 500,000.

Here it is critical that we keep in mind the fact that mortality from malnutrition and
disease (primarily diarrhea) is a function of antecedent violence that destroyed the
livelihoods and security of millions of Darfuris. In other words, these “excess
deaths” would not have occurred except for the genocidal violence that exploded in
spring 2003. People fled to camps—many dying along the way, often
unwitnessed—because their villages had been systematically and comprehensively
destroyed on an ethnic basis. High mortality rates even within the camps were
reported early in the conflict. The previously cited study in The Lancet, October 9,
2004 (“Violence and mortality in West Darfur, 2003-2004”) reports that in West
Darfur, once displaced persons arrived in camps their Crude Mortality Rates
(CMR—deaths per day per 10,000 of population) dropped back from
catastrophically high levels (5.9 to 9.5), “but remained above the emergency
benchmark (a CMR of 1.0), with a peak in 5.6 in el-Geneina.” The same study

“Before arrival [of displaced persons] at displacement sites, mortality rates
(expressed as deaths per 10,000 per day), were 5.9 (95% CI 2.2 – 14.9) in Zalingei,
9.5 (6.4 – 14.0) in Murnei, and 7.3 (3.2 – 15.7) in Niertiti. Violence caused 68 –
93% of these deaths.”

These are staggering mortality rates for large populations of civilians in or near
major towns of West Darfur—and violence “caused 68 – 93% of these deaths” (see
sample extrapolation from these data below). Moreover, shortly after The Lancet
study, Doctors Without Borders/Médecins Sans Frontières (MSF) reported similar
results in North Darfur:

“Mortality studies carried out by MSF show that during the early phases of the
Darfur conflict the pattern of repeated violence and consequent displacement was
the cause of very high mortality [emphasis added].” (MSF press release,
Brussels/Khartoum, December 1, 2004, at
http://www.doctorswithoutborders.org/press/release_print.cfm?id=364 )

Mortality rates have stabilized in most of the camps, at least for the present; but
there are extremely ominous humanitarian indicators that suggest CMRs may again
rise sharply, and that the fate of humanitarian operations in Darfur hangs in the
balance (see my recent two-part analysis of current humanitarian conditions in
Darfur and eastern Chad, at http://www.sudanreeves.org/Article264.html ).


The earliest study to bring serious attention to the scale of mortality in Darfur from
disease and malnutrition was released by the UN World Health Organization
(WHO) in August 2004, for the period 15 June and 15 August 2004
(http://www.who.int/mediacentre/news/releases/2004/pr63/en/index.html ); it
focused on establishing excess Crude Mortality Rates in Darfur (excess deaths per
day per 10,000 people). On the basis of this study, WHO announced in September
that 50,000 people in camps for the displaced had died from disease and
malnutrition in the five-month period from April 2004 to August 2004 (in October,
WHO estimated mortality for the six-month period from March 2004 to October
2004 at 70,000). The figure of 50,000 deaths was erroneously reported by many
news organizations as a new figure for total mortality in Darfur, gravely distorting
the scale of the Darfur catastrophe. The error was explicitly corrected in an email
to me from Dr. David Nabarro, head of WHO emergency response efforts in

“I fear that remarks I made at a Press Briefing on September 13th 2004 were
misquoted. I said that we estimate that at least 50,000 Internally Displaced Persons
have died from disease (in some cases exacerbated by malnutrition) since April
2004.” (email from Dr. David Nabarro to this writer, September 16, 2004;
emphasis added)

It is important here to consider what was left out, as shocking as the figure of “at
least 50,000” deaths is in itself. It excluded all mortality—from violence, disease,
and malnutrition—prior to April 2004. As we have seen, this total was already
well in excess of 100,000. The total excluded all mortality in Chad. It excluded
most violent mortality, which occurred largely away from the camps surveyed. It
excluded mortality from disease and malnutrition among populations beyond the
reach of camp humanitarian workers (this population was estimated to be 500,000
by the UN Humanitarian Coordinator for Sudan in September 2004). If this
inaccessible population—described in the UN report as “conflict affected and in
need of assistance”—experienced mortality rates comparable to those in the camps,
then an additional 17,000 people died of malnutrition and disease. In short, it was a
highly circumscribed study, and yet still yielded a terrifyingly large number of
“excess deaths.” (Again, in October WHO increased the mortality figure to 70,000
excess deaths between March 2004 and October 2004.)

Further, the report was the basis for extrapolating a future monthly mortality rate
of 6,000 – 10,000 from the same causes within the same population WHO had
surveyed (this is reflected in the October 2004 WHO estimate). Dr. Nabarro again
clarified this monthly figure, telling me explicitly that the WHO figure for monthly
mortality should be closer to 10,000 in the “6,000 to 10,000 deaths per month”
range indicated in the report.

Predictably, the WHO report so angered the Khartoum regime that they made the
one subsequent global mortality study virtually impossible to complete, and
ensured through threats against WHO workers that there would be no further such
studies in Darfur; insecurity continued to escalate and has made mortality
assessments increasingly difficult. As a consequence, we have only localized and
piecemeal mortality reports on “excess deaths” from disease and malnutrition from
2005 to the present (as opposed to truly global mortality surveys, with well-
established Crude Mortality Rates for all areas of Darfur). At present we also lack
global data and analysis of morbidity and malnutrition, largely because of
Khartoum’s intimidation of UN agencies and international humanitarian
organizations that in the past have generated and promulgated such data.


Very usefully, many of these smaller-scale mortality reports have been extracted,
collected, and analyzed by Olivier Degomme and Debarati Guha-Sapir in “Patterns
of mortality rates in Darfur conflict,” The Lancet, January 23, 2010 (pages 294-
300, http://www.thelancet.com/journals/lancet/article/PIIS0140-
6736%2809%2961967-X/abstract ). Their account draws in particular on the
“Complex Emergency Database” (http://www.cedat.be/guidelines ) of the Center
for Research on the Epidemiology of Disasters (CRED). While excluding from its
estimates of violent mortality any consideration of the CIJ data (“Darfurian
Voices” data were of course unavailable), their study offers important conclusions
about deaths from disease and malnutrition that are likely to be approximately
right, though excluding mortality in eastern Chad as well as mortality from the
period February 2003 to “early 2004” (see below for discussion of this critical

This highly technical paper, statistically of enormous potential richness, offers
some very clear conclusions: the authors “estimate the excess number of deaths in
Darfur to be 298,271 (95% Confidence Interval, 178,258 – 461,520)” in the time
period “from early 2004 to the end of 2008” (N.B. the terminus a quo and the
terminus ad quem). Of these approximately 300,000 excess deaths, they argue that
“more than 80% of the excess deaths were not as a result of the violence” but from
“diseases such as diarrhoea,” at least on the basis of the numerous studies archived
at CRED’s “Complex Emergency Database.” This yields a figure of roughly
240,000 deaths from disease (which presumably includes the effects of
malnutrition, which has in various times and places in Darfur been extremely
high), and a corresponding figure of roughly 60,000 deaths from violence in the
period “from early 2004 to the end of 2008”

Unfortunately, Degomme and Guha-Sapir seem incapable of recognizing the direct
connection between deaths from disease and malnutrition and the antecedent
violence that was responsible for these deaths—this is so even as they speak
abstractly of “excess deaths.” But of course these deaths are “excessive” because
of the genocidal violence that created the conditions in which people died of
malnutrition and disease. And there is nothing abstract about this brutal violence.
For the moment, however, we may ignore this peculiar myopia.

In the course of their study, the authors explain their limitations. They have not
included mortality in Chad (again, 52,750 people killed according to data from
“Darfurian Voices,” in addition to victims of disease and malnutrition). They have
not included mortality from December 2008 to the present. But even more
tellingly they confess that, “Another constraint was that we could not identify any
survey that included the first few months of the conflict before the deterioration in
September, 2003.”

In fact, there are no mortality studies at all for Darfur for the year 2003 in CRED’s
“Complex Emergency Database”: entering this year and any of the three Darfur
states into the site’s search engine yields only the message, “No entries found,
please try broaden your search parameters.” This is why Degomme and Guha-
Sapir’s narrative indicates that the period actually represented by their conclusions
is January or February 2004 to December 2008: “298,271 (95% Confidence
Interval, 178,258 – 461,520)” excess deaths in the time period “from early 2004
to the end of 2008.” (Tellingly, the authors do not specify precisely what is meant
by “early 2004.”)

This delimitation of time period is highly significant. In excluding the period from
February 2003 to “early 2004”—an approximately yearlong period of
extraordinary violent mortality, as well as highly significant mortality from other
causes—the authors leave out an essential part of the global mortality picture with
almost no acknowledgement. Of the half-year period from February 2003 to
August 2003 (“phase 1” in their Panel 1) they say only: “not included in any
retrospective survey, and mortality data should therefore be estimated by other

The truth is that they have not identified any Darfur mortality study for any months
in 2003. Even so, Degomme and Guha-Sapir push on to offer what is a
transparently untenable figure from the US State Department for this critically
omitted period of time (“February [2003] to August 2003”): “between 1,000 –
4,500 deaths” (from all causes).

In fact, this figure is an erroneous citation by the authors: the State Department
“fact sheet” (“Sudan: Death Toll in Darfur,” March 25, 2005)—as originally
promulgated on the State Department website—found that “4,100 – 8,800 excess
deaths are estimated to have occurred primarily in North and West Darfur [during
the period March – September 2003].” Notably, though unsurprisingly, Degomme
and Guha-Sapir do not cite the URL for this State Department estimate; rather they
cite the US Government Accounting Office report on mortality studies that cites
this four-page “fact sheet.” The inability to cite the URL for the document in
question derives from the State Department’s removal of this statistical travesty
from its website (www.state.gov/s/inr/rls/fs/2005/45105.htm is now a dead link).
But at the time of its initial distribution I analyzed in detail
(http://www.sudantribune.com/spip.php?article9364 ) the “fact sheet’s” numerous
methodological problems, its highly consequential factual errors, the lack of
citation or statistical analysis, and a consistent disingenuousness (document
accessed by this writer from the State Department website on April 23, 2005 at the
now defunct URL).

Critically for present purposes, Degomme and Guha-Sapir leave ambiguous
whether there are relevant mortality studies or considerations for the period
between September 2003 and January 2004, this despite their subsequent
parceling out of various “phases of the Darfur conflict,” including “September
2003 to March 2004” (“phase 2” of Panel 1—again, N.B. the terminus ad quem
for this “phase”). In their statistical analysis for this period they indicate “excess
deaths” of “45,137 (95% Confidence Interval, 27,320 to 73,380).” But it remains
unclear what data were used for September 2003 through the end of December
2003: again, there is nothing in CRED’s “Complex Emergency Database.” Nor is
it clear what data was used to calculate violent mortality for the period January
2004 to their “March 2004” endpoint for “phase 2.”

The authors conspicuously refuse to confront the implications of the CIJ study
(which looked retrospectively at Darfur mortality data from February 2003 to
August 2004) in speaking about “phase 1”; but even more tellingly—and all too
characteristically for professionals with no real knowledge of the historical context
in Darfur—they ignore evidence and reports from the ground that make clear that
tens of thousands of people were dying from a range of causes during the period
“February [2003] to August 2003,” causes that included violence, flight from
violence, and life in makeshift camps that were systematically denied humanitarian
assistance, often woefully inadequate in any event. The same may be said of their
“Phase 2” (“September 2003 to March 2004”), which plays such an ambiguous
role in their analysis.

Let us be clear about the significance of the figure “1,000 – 4,500” deaths—from
all causes—in the period February 2003 through August 2003 (by which time
Khartoum had substantially deployed its Janjaweed militias, as well as its regular
military forces, in genocidal destruction). Degomme and Guha-Sapir conclude that
total excess mortality in Darfur is 298,271 “from early 2004 to the end of 2008.”
But by a tawdry statistical sleight-of-hand, this time-frame is disingenuously
expanded by almost a year to become “about 300,00” “between March 2003 and
December 2008.” A year that includes some of the greatest violent mortality in
the Darfur genocide (February 2003 to “early 2004”) is brought within their final
estimate and time-frame on the basis of a completely vitiated State Department
“fact sheet,” which purportedly supports an estimate of “between 1,000 – 4,500
deaths” in this time period.

This is simply untenable, and here in the form of an historical compendium is some
of what cannot be accommodated to Degomme and Guha-Sapir’s account:


[1] The UN High Commission for Refugees (UNHCR) found in late summer

“…that about 75,000 people are currently scattered on the edge of the Darfur
conflict zone, on the Chadian side of the border. Most of them, spread out over 600
km, are women and children. With temperatures of up to 40 degrees Celsius during
the day and minus 15 at night, as well as sand storms and rain, respiratory
infections among the refugees are on the rise. They have little shelter, relying on
torn rags and bits of plastic sheeting.” (UN Integrated Regional Information
Services, October 8, 2003)

Doctors Without Borders/Médecins Sans Frontières (MSF) found that within this
refugee population, “…there is no clean drinking water, forcing people to dig in
sandy riverbeds with their bare hands to find ‘a dirty brown liquid.’ With scarce
food supplies, malnutrition is also on the increase.” [ ]

“On top of the hardship faced in Chad, the refugees have already had to flee their
homes to escape aerial bombardments and militia attacks in Darfur, MSF pointed
out. ‘Without exception, the refugees are traumatised by the violence they have
been subjected to. Many no longer have the emotional strength to do anything
except lie in the sand day and night, letting events unfold around them.’ [ ] ‘The
military response by the government, which escalated throughout the summer
forcing more and more people to become displaced or to flee to Chad, is being
backed up by systematic attacks by Arab militias who seek to crush the rebellion
and terrorise villagers,’ MSF said.” (UN IRIN, October 8, 2003)

Emphasis should be placed heavily here on MSF’s declaration that, “The military
response by the government, which escalated throughout the summer forcing
more and more people to become displaced or to flee to Chad, is being backed up
by systematic attacks by Arab militias who seek to crush the rebellion and terrorise
villagers [emphasis added].” There are many similar contemporaneous accounts of
civilians who fled large-scale, organized violence but remained within Darfur.

This is summer 2003—precisely the period Degomme and Guha-Sapir have
statistically minimized (“1,000 – 4,500 deaths” from all causes).

[2] As previously indicated, an early study of violent mortality in West Darfur
appeared in The Lancet (October 9, 2004 [9442]); it reported that in West Darfur,
once displaced persons arrived in camps their Crude Mortality Rates (CMR—
deaths per day per 10,000 of population) dropped back from catastrophically high
levels (5.9 to 9.5), “but remained above the emergency benchmark (a CMR of 1.0),
with a peak in 5.6 in el-Geneina [capital town of West Darfur].” The same study

“Before arrival [of displaced persons] at displacement sites, mortality rates
(expressed as deaths per 10,000 per day) were 5.9 (95% CI 2.2 – 14.9) in Zalingei,
9.5 (6.4 – 14.0) in Murnei, and 7.3 (3.2 – 15.7) in Niertiti. Violence caused 68 –
93% of these deaths.” [Zalingei, Murnei, and Niertiti are all major towns in West

Again, these are staggeringly high mortality rates among very sizeable populations
in or near major towns in West Darfur. For example, the number of violent deaths
over the course of 60 days for a population of 200,000 would have been
approximately 9,000, depending on the distribution of populations between the
three general locations, with their differing CMR’s.

Shortly afterwards, MSF would issue its own assessment: “Mortality studies
carried out by MSF show that during the early phases of the Darfur conflict the
pattern of repeated violence and consequent displacement was the cause of very
high mortality.” (MSF, Brussels/Khartoum, December 1, 2004)

“…during the early phases of the Darfur conflict the pattern of repeated violence
and consequent displacement was the cause of very high mortality.”

MSF was the first major international humanitarian organization with a substantial
presence in Darfur. My own communications with humanitarian organizations
operating in Darfur in 2003 confirm MSF’s conclusions about violent mortality.

[3] The scale of the catastrophe in the period excluded by Degomme and Guha-
Sapir is also captured in a November 10, 2003 statement by the UN Humanitarian
Coordinator for Sudan—very shortly after the period excluded in the State
Department estimate (but preceding Degomme and Guha-Sapir’s “early 2004”
starting point for their calculation of total mortality):

“The number of displaced people continues to increase thanks to the escalation of
armed conflict in the region since February 2003. The estimated 500,000 to
600,000 newly displaced people live in North, South and West Darfur. 70,000
people have sought refuge in Chad, and one million others have been affected by
the war.” (Statement from UN Humanitarian Coordinator for Sudan, issued
Nairobi/Khartoum, November 10, 2003)

Degomme and Guha-Sapir estimate that at the end of August 2003 the total civilian
population displaced or conflict-affected was 200,000 (graph from Figure 1). This
is in shocking contrast with the assessment of the UN Humanitarian Coordinator
for Sudan at the time, who estimated that approximately 1.6 million civilians were
displaced or conflict-affected. This enormous disparity is all too revealing.

[4] The Khartoum regime was in summer 2003 already hampering humanitarian
relief efforts, certainly costing many lives. The issue came to a head publicly in
October/November 2003:

“New regulations [from Khartoum] on travel permits that entered into force on 1
October 2003 have not been followed consistently. As a result, travel procedures
remain slow and cumbersome and, in some cases, permission to visit affected areas
is withheld.” (Statement from UN Humanitarian Coordinator for Sudan, issued
Nairobi/Khartoum, November 10, 2003)

“Sudan's government is hampering an adequate response to an escalating
humanitarian crisis in the war-ravaged Darfur region by reneging on a pledge to
process aid workers' travel permits speedily, the UN accused on Monday. ‘Some
aid operations haven't been able to start. Aid workers who are ready to go (to
Darfur) are getting stuck,’ because their permit applications have not been turned
around within a promised 24-hour period, Ben Parker, the Nairobi-based
spokesperson for the UN's Humanitarian Coordinator in Sudan, Mukesh Kapila
said.” (Agence France-Presse, November 10, 2003)

[5] In December 2003, Ambassador Tom Vraalsen, special UN envoy to Sudan for
humanitarian affairs, gave a clear retrospective, noting that while the Khartoum
regime characterized “the security situation [in Darfur] as steadily improving,” this
account “sharply contrasted with first-hand reports that I received from tribal
leaders and humanitarian actors on the ground. They reported that [Khartoum-
backed Arab] militias were launching systematic raids against civilian populations.
These attacks included burning and looting of villages, large-scale killings,
abductions, and other severe violations of human rights. Humanitarian workers
have also been targeted, with staff being abducted and relief trucks looted”
(emphasis added). (Tom Vraalsen, Note to the Emergency Relief Coordinator;
“Sudan: Humanitarian Crisis in Darfur,” December 8, 2003)

“Systematic attacks against civilians populations,” “large-scale killings,” “targeting
of humanitarian workers”—Vraalsen is speaking out in early December 2003, but
his characterization is relevant for much of late summer 2003, during which time
Degomme and Guha-Sapir would have us believe only “1,000 – 4,500” people
died from all causes. And of course Vraalsen’s characterization is fully relevant
for the entire period prior to Degomme and Guha-Sapir’s vague terminus a quo,
“early 2004.”

[6] Humanitarian conditions were much worse by the end of summer 2003 than
Degomme and Guha-Sapir seem aware of. Invoking the terrible famine in Bahr el-
Ghazal province in 1998—in which perhaps 100,000 people died—Save the
Children (UK) reported that “current overall malnutrition rates are reported to be
alarmingly high, with Global Acute Malnutrition rates reaching 25% in some of the
affected areas of Darfur, which are accessible to relief workers” (“Sudan
Emergency Statement,” December 10, 2003). Such an extraordinarily high GAM
rate (the emergency threshold is 15%) was months in the making—again taking us
well back into the February 2003 – September 2003 period Degomme and Guha-
Sapir essentially exclude.

Under-Secretary Jan Egeland, UN Emergency Humanitarian Coordinator at this
time, declared: "‘The humanitarian situation in Darfur has quickly become one of
the worst in the world. Access to people in need is blocked by the parties in
conflict and now, as the need for aid grows, stocks of relief materials are
dwindling…’" (UN News Center, December 5, 2003). Again, though speaking in
early December 2003, Egeland’s assessment reflects conditions extending well
back into summer 2003.

[7] Evidence of sustained “ethnic cleansing” and genocide were clearly in
evidence by October/November 2003, and the December 11, 2003 report from the
International Crisis Group (“Sudan: Towards an Incomplete Peace,”
http://repository.forcedmigration.org/show_metadata.jsp?pid=fmo:3714 ) was soon
confirmed by many other human rights investigations:

“[G]overnment-supported militias deliberately target civilians from the Fur,
Zaghawa, and Massalit groups, who are viewed as ‘African’ in Darfur and that the
latest attacks by these Khartoum-backed Arab militias occurred deep inside the Fur
tribal domain, against unprotected villages with no apparent link to the rebels other
than their ethnic profile.” (page 19)

“The killings and resulting forced displacement prompted Mohamed Barka,
member of parliament for the district of Kebkabiya, to warn in early November
[2003] against “genocide” in Darfur. In a statement to Alwan newspaper
[November 6, 2003], he indicated that in one month of raids, Janjaweed had burnt
150 villages and forcibly displaced 280,000 inhabitants in his district. Other hard
hit areas include Kass, Zaleinge, and Jabal Marra districts, all within the Fur
heartland.” (page 19)

Doctors Without Borders/Médecins Sans Frontières (MSF)—again the first
humanitarian organization with a large response presence in Darfur—characterized
the initial months of the war as, “Mass violence against civilians in Darfur [that]
began with a wave of attacks against villages in February 2003 and escalated in
summer 2003.” (MSF “Field News,” November 1, 2004)

By June 2003, Musa Hilal—the most brutal and notorious of the Janjaweed
leaders—had returned to Darfur from Khartoum, and immediately set out to recruit
both for the Janjaweed and the paramilitary Popular Defense Forces. He was
enormously successful, and military equipment delivered from Khartoum
continued to become more lethal, making the Janjaweed an ever more effective
fighting force (military preparations had been ongoing throughout 2002). The shift
to deliberate civilian destruction accelerated in September/October 2003, but we
have ample evidence of deliberate civilian targeting throughout the summer of
2003 (i.e., through the end of September). Julie Flint and Alex de Waal report in
Darfur: A Short History of a Long War (2005):

“The [Sudanese Armed Forces] air force continued in its leading role. Large
numbers of displaced had congregated in Habila by August 2003. It seems that
their presence was the reason why the town was heavily bombed, killing thirty
people on a single day that month. The UN Commission of Inquiry found ‘no
evidence that there was any rebel activity or structures in the vicinity that could
have been the target of this attack.” (page 106)

Darfuris insisted that this bombing was not a “mistake” by Khartoum’s forces, an
explanation the credulous UN Commission appears to have accepted. They
pointed out that, “‘Before the bombing of [Habila], Tunfunka, Tulus, Andranga,
Hajjar and Bayda had all been burned and everyone had run to Habila’” (cited by
Flint and de Waal, page 107).

The signature cruelty and genocidal intent of the Darfur conflict was also in
evidence early in 2003:

“‘These rapes [by the Janjaweed] are…orchestrated to create a dynamic where the
African tribal groups are destroyed,’ an aid worker said. ‘It’s hard to believe that
they tell them [African women and girls] that they want to make Arab babies, but
it’s true. It’s systematic.’ Nor was sexual violence limited to rape. Early in 2003,
a young woman called Mariam Ahmad was stopped at a roadblock and force to
watch while Janjawiid cut the penis off her 21-day-old son, Ahmad. The child
died soon after in her arms. In Bargai, a village near Zalingei, a young mother who
had just given birth to twins was killed with her legs tied to her neck, exposing her
genitals. Her babies were thrown into a container of boiling water that had been
brought for the birth.” (Flint and de Waal, pages 108 – 109)

Diplomats in Khartoum were speaking privately of a continuing campaign of
“ethnic cleansing”:

"Diplomats have described the fighting in Darfur as 'ethnic cleansing' with Arab
militias, possibly backed by the government, destroying entire villages inhabited
by dark-skinned people who speak African languages." (BBC, December 11, 2003)

Of course direct support of the Janjaweed militia by the Khartoum regime was
soon definitively established, most compellingly by Human Rights Watch in
“Entrenching Impunity: Government Responsibility for International Crimes in
Darfur” (December 2005, http://www.hrw.org/node/11496 ).

The grim assessment of genocidal ambition in Darfur was of course also coming
directly from non-Arab/African Darfuris:

“'I believe this is an elimination of the black race,' one tribal leader told IRIN (UN
IRIN.” [dateline: el-Geneina] December 11, 2003)

To suggest in this context that “1,000 – 4,500” people died as a result of the Darfur
conflict from February 2003 to September 2003—or indeed from February 2003 to
“early 2004”—is in fact more than myopia on the part of Degomme and Guha-
Sapir. It is a deliberate, finally disingenuous effort to preserve the supposed
integrity of their figure for total mortality. For by using this estimate, they can
then give a more inclusive time range for the mortality total they profess to have
determined statistically: “the overall number of excess deaths in Darfur
between March, 2003 and December, 2008 was about 300,000.” Again, this
slippery figure of “about 300,000” is obviously the result of their adding their
precise calculation of “298,271 (95% Confidence Interval, 178,258 – 461,520)”
(“early 2004 – December 2008”) and the completely untenable US State
Department estimate of “1,000 – 4,500” (February 2003 – September 2003)—
and evidently ignoring the period from September 2003 to “early 2004” (again,
there are no data in CRED’s “Complex Emergency Database” for the year 2003,
for any of the three Darfur states).

Given the high degree of politicization that has defined mortality estimates for the
Darfur conflict, it is deeply dismaying to see such disingenuous exclusion and
tendentiousness on the part of experienced epidemiologists. They contribute to
what has already been a terrible moral discounting of Darfuri lives.


Why does any of this matter? Here it’s useful to recall that in February 2004—one
year into the most violent and destructive phase of the Darfur genocide—the
official UN estimate for total human mortality was 3,000. In retrospect this is of
course an absurdly low number, although there was no apparent UN effort to
deceive. But this figure gave a deeply misleading picture of the scale of the
conflict, and that misapprehension continues, mutatis mutandis, particularly among
those intent on denying that the human destruction in Darfur is genocidal. We
catch a glimpse of this political disposition in the refusal of Degomme and Guha-
Sapir to connect explicitly the ethnically-targeted violence in Darfur with what
they themselves describe as “excess deaths” from disease and malnutrition.

In the end the challenge to this early distorting UN figure came not from
professional epidemiologists like Degomme and Guha-Sapir, but from activists;
from Asma Jahangir, UN Special Rapporteur on extrajudicial, summary or
arbitrary executions, who reported in late June 2004 that the “number of black
Africans killed by Arab militias in the Darfur region of Sudan is ‘bound to be
(http://www.un.org/apps/news/story.asp?NewsID=11191&Cr=sudan&Cr1= );
from Sudan Focal Point/South Africa, which gathered critical mortality data from
eastern Chad in early 2004; from the large team of forensic specialists, genocide
scholars, legal experts, and law enforcement officials of the Coalition for
International Justice (who, I was told by a member of the team in September 2004,
unanimously regarded what they had uncovered as evidence of genocide); from the
UN World Health Organization under David Nabarro; and from independent
researchers such as Dr. Jan Coebergh. Most recently, “Darfurian Voices” takes an
essential place in this history of efforts to convey the real scale of human
destruction in Darfur. One can only hope that it will be read with honesty and
moral integrity.


[I am very grateful to those with training in statistics, epidemiology, medicine, as
well as significant experience in Sudan and Darfur, for their critical reviews of this
assessment—Eric Reeves, August 6, 2010]

Appendix 1: In section III of the “Results” section of the report from “Darfurian
Voices,” there are several important statistical results and two important
qualifications that should be borne in mind. The footnote to this section is
reproduced here in its entirety, as is the explanation by the authors of “Darfurian
Voices” of the selection process for refugee leaders.


“Nearly all respondents were exposed to serious violence while in Darfur. 74% of
interviewees [81% of men; 67% of women] left Darfur because of immediate
violence that they personally experienced and 21% [15% of men; 26% of women]
left because of fear of future violence. 91% had all of their property and
possessions stolen or destroyed. 32% [37% of men; 28% of women] had
personally suffered physical violence in an attack related to the conflict. 72% of all
respondents reported that at least one member of their immediate family was killed
in the conflict; the median number of family members killed was 4.2. 90%
personally witnessed attacks on family members.”

[b] Footnote 2 (page 15):

“While collecting data from the fourth [of 12] camps, we discovered an ambiguity
in the Arabic wording of the question regarding the number of family members
killed in the conflict. In English, the question read: ‘How many of your immediate
family members (parents, siblings, spouses, children) were killed in attacks related
to the current conflict?’ In Arabic, the parenthetical statement was accidentally
omitted. To correct for this, we inserted the parenthetical statement and re-trained
all interviewers. An ad hoc comparison showed that the mean number of family
members killed did appear to differ significantly between the first four and last
eight camps [p<2*10-6]. Therefore, all figures presented in this report regarding
the number of family members killed in conflict does not include data from the
first four camps: Gaga, Farchana, Bredjing, and Treguine.”

[c] Selection of sample population:

“The civilian sample consisted of persons aged eighteen or over who were
randomly selected in each of the 12 refugee camps in eastern Chad, using stratified
random sampling approach. As such, it was representative of the adult refugee
population in the camps. The civilian sample size was 1,872. Unless otherwise
indicated, statistics attributed to respondents, interviewees, or participants
refer to the civilian sample. The leader sample consisted of tribal, civil society,
and rebel leaders, who were given a more in-depth survey than the one
administered to civilians. The leader survey included additional open-ended
questions as well as all the same questions as the civilian survey. Participants in the
leader survey were not randomly selected. Leaders were drawn from a combination
of a sample of convenience and from an effort to interview the sheikh of every
block in each refugee camp. Consequently, the leader sample cannot be said to be
representative of the leadership population as a whole. The leader sample size was
280, including 250 refugee leaders who were interviewed in the camps as well as
30 rebels and rebel leaders who were interviewed outside the camps. The data
collected from the 30 leaders and rebels interviewed outside the camps were used
for qualitative purposes, but are excluded from the leader statistics computed and
presented in this report. Statistics drawn from the leader sample will always be
explicitly attributed to leaders.” (emphasis added)

Appendix 2:

A number of the variables in this analysis are not purely statistically derived, nor
could they be. What I have done whenever introducing a statistical variable that is
not itself statistical derived is to be at once reasonable and conservative.
Reasonable in the sense of asking, “does this make sense, given all we know about
circumstances, context, and qualitatively relevant reports bearing on Darfuri
refugee and displaced populations, as well as numbers of reported attacks, satellite
imagery of destroyed villages, ecological study of Darfur’s terrain during the
period of conflict, and eyewitness accounts as reported directly and by human
rights and humanitarian organizations?” “Conservative” in the sense that I
deliberately sought figures that were reasonable, but which if anything were more
likely to understate than overstate mortality. This was the basis for the choice of
deflators in calculating violent mortality rates among displaced populations at
different stages in the genocide.

The difference such assumed variables can make is perhaps best illustrated by
looking at family size, where the assumption in this assessment is of an
“immediate family” whose members include “parents, siblings, spouses, children.”
I have used a figure of 10, which I believe to be quite conservative; for the smaller
the family size in extrapolating from “Darfurian Voices” data, the larger the total
for violent mortality. An average family size of 9, for example, increases the
mortality total among Darfuri refugees in eastern Chad from approximately 57,250
to approximately 63,500.

But there are certainly problems in arriving at any firm figure for family size. The
fourth of my “caveats” indicates one way in which family size might be uncertain
in a situation where someone is both a father in one family and brother to the father
in another family. The most fundamental problem, however, is simply the range of
estimates for the “average” Darfur family size.

CHF International based their work in Darfur’s IDP camps on the assumption of
“5.5 family members per household.” (There is an obvious potential ambiguity in
the language here.)

USAID reports in May 2004 that at in Dubai, South Darfur, it provided “temporary
shelter for 6,400 families,” and then went on to note that, “the average family size
in Darfur is 5.”

But a month later, USAID would report that, “Based on data collected during
Médecins sans Frontières’ (MSF) recent nutritional survey, the average family size
among the conflict affected population in Darfur is 7 persons.”

CARE, in an April 2004 emergency report on Darfur, found the average family
size to be 5.5.

Last June (2009) a UNAMID assessment at Masri village (North Darfur) reported:
“Residents revealed that about 2,300 households with an average family size of 8
individuals had returned to the area.”

In an August 2005 article in the journal Criminology, John Hagan, Wenona
Rymond-Richmond, and Patricia Parker note:

“The patriarchal and polygamous practices of these [Darfuri] African tribes is
reflected in the larger average family size reported by males (7.1, compared to
females, 6.2), who may have more than one wife.

Notably, all of these figures are considerably under the assumed figure of 10
members for the “immediate family” of Darfuri refugees. And again, the smaller
the family size—using the data from “Darfurian Voices”—the greater the total of
violent mortality.

A future mortality update will provide more specific ranges for results from the
variables utilized throughout this analysis of “Darfurian Voices” data.