These experts snipe at each other in academic journals and papers and are engaged in a statistical arms race of their own, sampling, adding and modifying databases to try to shed light on questions that have implications beyond the academic world. The disagreements begin with how fatalities are counted and then diverge more widely.
Authors from the International Peace Research Institute in Oslo, Norway (PRIO) in collaboration with Sweden's Uppsala University, say that since the collapse of the Soviet Union and end of East-West rivalry in 1989, civil wars, "though often intractable and devastating, have produced fewer battle deaths than their Cold War counterparts."
They relied on estimates by demographers, historians, and epidemiologists, supplemented by figures from the media, governments and non-governmental groups. Omitting "one-sided violence increases" such as the Rwanda genocide along with deaths from disease, hunger and "criminal and unorganized violence," PRIO arrived at a total of some 10 million battlefield deaths from 1946 to 2002 in conflicts where at least one warring party was a government.
Way too low, countered Harvard Medical School researcher Dr. Ziad Obermeyer, working with colleagues at the University of Washington and at the Gates Foundation-funded Institute for Health Metrics and Evaluation.
His group reconsidered the PRIO numbers in the light of World Health Survey sample interviews in 13 conflict-ridden countries where the U.N. asked families how many relatives they lost to war from 1955 to 2002. It found the true body count over a half-century was at least three times higher than PRIO's tally and concluded, "There is no evidence to support a recent decline in war deaths."
Their method, relying on memories of family members, allows for retrospective study of countries where census and death records have been disrupted by war and researchers cannot safely go into the countryside, Obermeyer told the AP.
"Rather than uncritically taking data as they are reported from the field, we are actually using random samples of populations to look backward at how many relatives of respondents were killed, and then using standard statistical methods to extrapolate and get a number for the whole population," Obermeyer said.
But the death toll also can be counted in other ways that drive it astronomically higher.
Going beyond direct battlefield casualties and counting the victims of genocide, deliberate famines, death camps and other warlike actions raises the tally to 41 million people slain since the end of World War II, says Milton Leitenberg, a senior researcher at the University of Maryland's School for International and Security Studies.
Leitenberg extends his count of victims of war and conflict back to the dawn of the 20th century, producing an appalling total of 231 million.
He says his figures for mass casualties are the most widely accepted totals based on statistics by U.N. agencies and humanitarian and human rights groups. But he concedes such figures can be colored by politics and says the true tally might be 20 percent higher or lower than his total.
What Leitenberg rejects is using only government-reported or one-sided battle fatalities - which would omit more than 70 million deaths from executions, repression and starvation in Stalin's Soviet reign of terror and the politically engineered Ukraine famine, Hitler's genocidal campaign against the Jews, and Mao Zedong's "Great Leap Forward" from 1959 to '61.
"It's basically a flawed construct, which gives you perhaps 10 percent of the real total," Leitenberg told the AP. "Social scientists have a nasty habit of setting arbitrary criteria."
Obermeyer also complained the PRIO methodology takes a limited view, excluding "deaths that are violent but occur when both sides don't have guns, that sort of one-sided violence."
"I don't think it's reasonable to not include those kinds of deaths in counts of war deaths," he added.
Andrew Mack, a former U.N. planning expert and now director of the Human Security Report Project at Canada's Simon Fraser University, believes battle deaths and indirect civilian deaths are both on the decline, backing PRIO's thesis.
His project judges the total number of battle deaths from 1946 to 2007 to be 10,095,152 - to be precise - using PRIO figures from the period 1946-2001 and data commissioned from Uppsala University for 2002-2007.
Mack supports Leitenberg's idea of trying to count all deaths, but told the AP that "census data are often unreliable in poor countries and often very out of date. As for mortality surveys, there are extraordinarily few of these - and Milton (Leitenberg) doesn't cite them. So his totals are sometimes just battle-deaths and sometimes battle-deaths plus guesstimates of 'indirect' deaths."
How the deaths are counted can lead to wild differences in totals. The Bangladesh/East Pakistan war of 1971 provides a glimpse of how these four systems come up with disparate results:
Mack and other backers of the PRIO theory of decreasing war deaths criticize the Harvard study for opening its timeline with 1955. That skews the trend, they say.
Mack also contends that the 13 conflicts cited by Obermeyer and Harvard were "a small minority of war-affected countries over this period. It is not possible to draw any conclusions from global trends with such a small sample."
Nils Peter Gleditsch of PRIO, one of the authors of the "declining mortality" study, told the AP that Harvard's "time frame excludes the Korean War, not to speak of World Wars I and II. Including these wars makes the decline in battle deaths even clearer." It also cuts out the huge death toll in China's civil war between Communists and the Nationalists.
Obermeyer responded that "I wouldn't base any trends on a cataclysmic event, like 1945 and World War II." He said his study began with 1955 because there are too few survivors of earlier conflicts to build a statistically significant sample. His analysis ends with 2002 because that's when WHO conducted its surveys.
Mack says both battle deaths and indirect civilian casualties are shrinking because "their major drivers, the number and deadliness of wars, have both shrunk. Plus - and this is critical - humanitarian assistance per displaced person has increased fivefold since the end of the Cold War, and become more cost-effective."
Gleditsch's co-author, Stanford University doctoral candidate Bethany Ann Lacina, noted that "the overall decline in battle deaths is driven by the decreasing incidence of internationalized wars involving rich countries with super-modern forces - those are the wars that cause the huge spikes in the data," such as the Korean, Vietnam and Afghanistan-Soviet wars.
American involvement in the 1991 Gulf War, after 2001 in Afghanistan, and the current Iraq war "all were definitely more deadly than most other conflicts in the same period. But prior to 1989 there were even more internationalized civil wars going on, with even larger troop deployments, and even more years of fighting," she told AP.
In the end, each of these studies by international research institutes contains major caveats about potential built-in methodological biases and leaves wiggle room for elaborating on the figures in later studies.
"There's always going to be uncertainty," Obermeyer said.