THE NUMBERS GUY

By CARL BIALIK

Death Tolls Can Be Distracting

May 23, 2008; Page A11

When it comes to counting the dead from disasters such as the Myanmar cyclone and the China earthquake, your guess is about as good as anybody's.

Myanmar's military rulers first reported two weeks ago that 22,000 had been killed by Cyclone Nargis. Now, the official number has climbed to 78,000 and is expected by nongovernmental organizations to top 100,000. In China last week, early tallies by officials reported nearly 10,000 people killed by an earthquake in Sichuan province. The count has now climbed to more than 51,000.

THE NUMBERS GUY

[go to blog]

Blog: The Media's Favorite Disaster Numbers

In both countries, tens of thousands of people remain missing.

People killed by natural disasters don't need relief supplies, and no court can prosecute their amorphous killers. As a result, counting bodies is rarely a top priority, and accurate numbers may come months later -- or never. "I'm not really sure beyond press reporting what function these numbers serve," says Patrick Ball, a statistician who studies human-rights violations.

Yet, death tolls calibrate news coverage, which builds public awareness and, in turn, drives donations. The estimates also are recorded in history books and serve as rankings for disasters. They are also used to build software models for quickly assessing earthquakes' damage. (Storms are more variable and so harder to measure in a single number akin to magnitude for earthquakes.)

What's more, the high estimates generate more scrutiny of relief efforts. The reported tolls have brought attention to the Myanmar government's initial reluctance to accept aid and to China's challenges in disbursing aid to the remote earthquake zone.

"If just 1,000 died, the junta wouldn't be getting nearly the notoriety it is now" for blocking some international relief efforts, says Jeffery Wright, World Vision's senior program officer for emergency response. In China, people are protesting the shoddy construction of school buildings that collapsed on hundreds of schoolchildren.

Beyond practical considerations, some aid workers see value in an accurate picture, rather than ignoring the magnitude of the human suffering. "People should die in dignity," says Aung Myo Min, director of Human Rights Education Institute Burma, which has aided cyclone relief efforts. "All the dead persons should be reported correctly."

Still, some aid workers lament the media's preoccupation with figures. "When a small country like Madagascar or Mozambique has cyclones and floods that wipe out homes and decimate crops, media are not likely to cover it unless the death toll also is huge," says Lurma Rackley, spokeswoman for the poverty-fighting group CARE. "And people are not as likely to be moved toward helping if they don't see images of the disaster."

Some believe the task of enumerating the dead distracts from tending to the living. Barbara Butcher, chief of staff of New York City's medical examiner's office, says Myanmar officials must focus on survivors' needs. "Sadly, I would say they should not worry about identifying the dead, nor any kind of preparation for burial," says Ms. Butcher, who spearheaded the count of World Trade Center victims and advised similar efforts following Hurricane Katrina and the 2004 Asian tsunami. She notes that Americans expected a relatively precise count after Katrina -- roughly 1,700 -- but that figure took more than a year to calculate. Poorer countries can't afford to gather such a list of victims.

In general, early counts are likely based on rough techniques such as extrapolating the death rate from one village to all affected areas, or using aerial photography to estimate building damage.

To improve on that process, David Wald, a U.S. Geological Survey seismologist, is analyzing disaster-fatality rates for the development of software that will predict death tolls quickly after earthquakes. The aim is to guide aid agencies' early planning. "If the U.S. government were to put out a number that was pretty accurate within a few hours, that would put a little more pressure on parties to respond more quickly," Dr. Wald says.

The software would incorporate measurements of quake magnitude from USGS sensors and partners around the world, as well as population maps and historical earthquake data. Dr. Wald calculates, for example, that an earthquake in Pakistan would kill roughly 5,000 times the number of people as a similar-size quake exposing the same number of people in California, because of varying construction practices. China's fatality rate, he says, is closer to Pakistan's than to California's.

To project the death toll from an earthquake, the program would calculate how many people were exposed to shaking, and extrapolate a figure from a prior quake in the same country or, if earthquakes are rare there, from an analogous one.

One problem is that the software must rely on death tolls from historical earthquakes to project future ones. Also, the researchers use only the death toll -- not the number of people injured or the number of buildings damaged -- because it's the one figure most commonly reported after disasters.

The software, then, may not improve the accuracy of death-toll counts, though it may help get the estimates out faster.

Max Wyss, who has implemented fatality-projection software for the Swiss government, has made projections reliable enough to determine whether the Swiss government should mobilize aid. Yet, his initial projection for the China quake was 1,000 to 4,000 deaths -- far too low -- in part because of an initially low USGS estimate of the earthquake's strength. When the USGS revised its estimate hours later, Mr. Wyss increased his projection to 40,000 to 100,000 dead. Similarly, his initial estimate for the earthquake near Bam, Iran, in 2003 was for a maximum of 1,010 casualties, when more than 26,000 people died.

"We can do better," Mr. Wyss says.

Read Carl Bialik's daily commentary about numbers and join him in a discussion with readers at the free blog, WSJ.com/NumbersGuy. Email him at This e-mail address is being protected from spambots. You need JavaScript enabled to view it .