One Disaster, Too Many Explanations
In the aftermath of modern U.S. disasters, science is tasked with coming up with unbiased data and irrefutable analysis. If only life were that simple, especially when it all goes to court.
One month after a retaining wall collapsed at a Tennessee Valley Authority coal plant just before Christmas, sending 5.4 million cubic yards of fly ash into the Emory River, a group of researchers at Duke University released their analysis of what exactly the gray sludge contained. Radium and arsenic particulates embedded in it, they concluded, could pose health problems for the people living near the spill site.
“When we published our data and had a press release of our report,” said Avner Vengosh, a professor of earth and ocean sciences who led the study, “we received some calls from people who said, ‘You’re being paid by TVA to shut you up,’ which we found kind of ironic.”
Ironic because of the reaction Vengosh received from TVA itself.
“They were not happy at all,” he said.
That Duke touched a nerve with both camps suggests its research was independent. But even more than that, the reaction underscores the particularly contentious nature of science in the wake of disasters — be they man-made, natural or an ambiguous mix of the two — when science holds the weighty task of not just reconstructing what happened and explaining what could happen next, but also assigning blame.
The pattern is also familiar after Hurricane Katrina, Exxon Valdez, Three Mile Island, and Love Canal. Multiple testing from multiple parties reaches multiple conclusions. The public, in all this multiplicity, is unsure which science to trust.
“There are certain themes you find in the wake of a lot of disasters,” said Gregory Button, an environmental anthropologist at the University of Tennessee, Knoxville. “There’s an inevitable period of time of scientific uncertainty to be expected to some degree. Part of the problem is often people in the public don’t expect it. They think science knows everything. And when (science) is not as forthcoming, they think that’s sort of suspicious.”
Studies in epidemiology, public health, environmental containment and engineering — the very areas disasters tend to funnel researchers into — seem particularly fraught with chose-your-own-science accusations of bias and agenda. As the scope of disasters increases — and around events where no single, respected investigator exists like the FAA for plane crashes — so, too, do the stakes of litigation.
“I’m sure it would be more sensitive than, say, research into bias in decisions in referees for the National Basketball Association,” said Steven Wing, a University of North Carolina epidemiologist. “Because people could sue, or industry could be told they have to clean up — things that would be very expensive.”
Vengosh’s research didn’t actually turn up substantially different data from TVA’s own sampling; he just interpreted it differently. Duke warned of the public health danger should particulates in the fly ash become airborne and then inhaled by local residents. TVA, on the other hand, stressed that such a situation did not, at the moment, exist.
“I think part of the issue is, from his perspective, he was looking at his data, and he was making some observations based upon his scientific knowledge,” said Joe Hoagland, TVA’s vice president for environmental science, technology and policy. “The media folks who took that data and published it for the general public were trying to find a smoking gun.”
TVA officials approached Vengosh with the possibility of cooperating on future testing. Outside researchers, either through universities or environmental groups working in the area, have so far not been allowed to test directly on TVA property. Other testing has been done at the site by the federal Environmental Protection Agency and the Tennessee Department of Environmental Conservation.
Vengosh would love to test there — continued sampling is necessary due to the airborne potential as well as the possibility that the process of cleaning the river could further unsettle the ash — but he’s cautious of any conditions that would be placed on the research by TVA.
“If it only means they give us access to the location and then we’re required to give them data, I have no problem with that,” Vengosh said. “I would not accept any conditions in terms of the ability to publish and transparency of the data.”
Hoagland said TVA is working on a plan for the Oak Ridge Associated Universities to administer a project that would invite university researchers to conduct testing with a uniform protocol that would allow all the data to be compared “apples to apples.”
“We will provide the funding, but then we’re out of it,” Hoagland said. TVA’s interest, he said, is in arranging for data collected by experts that can be cross-examined — which is why every environmental group and private citizen with a water bottle is not permitted to test on TVA land.
Perception within a skittish community is as important as the arrangement itself. The TVA, the largest public utility in the country and a major development project of the 1930s New Deal, brought both jobs and electricity to a region in poverty. But it also took over many peoples’ land in the process, breeding over the decades a love-hate relationship with the community, in Button’s words.
Duke originally undertook its testing in January at the request (but not the funding) of United Mountain Defense, an anti-strip-mining group that had been doing its own air and water sampling. UMD, out of concern that its own results could be attacked by TVA as biased, urged university researchers to provide another alternative.
Vengosh, in later releasing his findings, said Duke wanted to distance itself from UMD to underscore its independence in a standoff increasingly pitting TVA against environmental groups.
“There are a whole lot of data here, and this is probably true of any disaster,” said Hoagland, who holds a doctorate in physical chemistry, “but there are people who have their own personal agenda that has nothing to do with the science itself, so they’re going to take the data and make an interpretation that focuses on their agenda.”
His veiled criticism mirrors the barb environmental groups lobbed back at TVA.
UMD has continued collecting its own samples, videotaping the process to post on the Internet and inviting community members to participate in the collection.
“I definitely think that it is in TVA’s favor to paint us as radicals, or extreme activists that have an alternative agenda, when in reality our agenda is to offer disaster support for the community,” said Bonnie Swinford, a UMD board member. It’s also in TVA’s best interest, she said, for the data to show that the water and air around the spill site are safe, minimizing TVA’s exposure to legal charges and cleanup costs.
With its results, UMD has embraced a sort of open-source public data dump on its Web site. The philosophy is the opposite of the process behind most scientific publication: Let the public see the data for themselves rather than running it first through expert peer-review that might weed out inconsistencies or insignificant data.
TVA is now, perhaps out of necessity of public perception, posting its results on its Web site too, revealing that, as the site states, particulate matter and metals in the air “are less than any levels of concern.”
Critics, though, can’t get past the suspicion that the agency responsible for the spill should not also be in charge of measuring its consequences.
Three Mile Island
Wing, the UNC epidemiologist, cautions that the solution to a tense disaster cleanup is not necessarily independent university researchers.
“There’s nobody with no agenda,” he said. “Everybody has an agenda.”
University scientists are beholden to department chairs, who have to worry about research funding, which largely comes from government and industry sources. Wing believes admitting those social and political contexts, rather than trying to avoid them, is the path to objectivity.
And context, he says, comes not just from funding conflicts, but the assumptions upon which research is designed: where to test, where not to test, what to test for.
Wing analyzed cancer research in the Pennsylvania community around the Three Mile Island nuclear power plant that experienced a partial core meltdown — a mini-Chernobyl — in 1979.
Researchers at Columbia University were commissioned by the nuclear industry-financed TMI Public Health Fund to do a cancer study in the community after the accident. They concluded, based upon what was officially considered to be a small amount of radiation released in the accident, that increased cancer rates in the area were likely due to stress.
Several years later, Wing was approached by the plaintiffs in a lawsuit against TMI that alleged health problems due to radiation exposure. They asked him to reanalyze the Columbia data — a scientific request to which he admits he brought his own set of assumptions.
His “point of view,” he says, was that mass hysteria couldn’t explain all the symptoms people around Three Mile Island experienced — and certainly not the symptoms of people who reported nausea, hair loss or a metallic taste in their mouth before learning of the accident.
“I also had my own experiences seeing that researchers weren’t given all the information before in other cases around nuclear plants,” he said. “So all of that affected my assumption that there could have been more radiation released than we were told.”
Those assumptions led to decidedly different conclusions than the Columbia study, using the same data set. Wing concluded that there was evidence cancer rates increased in communities directly in the path that radiation plumes were estimated to have traveled from the plant, and not in the other directions. Columbia had concluded that the radiation couldn’t have caused an increase in cancer because the radiation doses were too low.
“But that’s not something that was testable,” Wing said. “If they assumed the doses were too low, they didn’t need to do a study of cancer in the first place.
“We tried to come up with alternative explanations, and what Columbia didn’t do was come up with the possibility that there was greater radiation released than the government said. I think that’s actually a pretty simple one.”
The lawsuit was dismissed in late 1999, in part after some of Wing’s analyses were thrown out on the defendants’ use of another assumption — that radiogenic lung cancer (as we understood it in the wake of atomic bombs dropped on Japan) has a minimum latency period of 10 years, a timeframe beyond the length of the original TMI study. Wing points to the two studies, though, as an illustration that data alone cannot give objective answers to questions of causation and consequence in the wake of disasters. The answers all depend on the assumptions we bring to the data.
Three Mile Island reveals another set of problems inherent in some post-disaster research: The interest — and funding — for long-term studies often expire before an event’s full consequences can be known, and, in the case of epidemiologic studies, the exposed often live in small communities that hardly constitute statistically significant sample sizes.
Hurricane Katrina this spring spawned a similar lawsuit against the Army Corps of Engineers that the judge has called the “first real trial” about Katrina — but one that is, more than anything else, about science. At issue is whether a navigation channel built by the Corps in the 1960s — identified locally as “Mr. Go” — served as a runway for Katrina’s destruction as it tore into the city.
“We know that Mr. Go probably destroyed 140 square miles of wetlands, but the scientists disagree about what the effect of the loss of the those wetlands are,” said Robert Verchick, a law professor at Loyola University, New Orleans, and a research scholar at the Center for Progressive Reform. “We know wetlands buffer storms and storm surges, but scientists disagree about what would have happened if we had those wetlands.”
The suit will turn on two questions: one scientific, one legal. Did the channel amplify the destruction of the hurricane, and did the Corps breach its duty to the public in planning and maintaining the channel, beyond making reasonable discretionary decisions?
Judge Stanwood Duval could have ruled on the latter legal point without taking the case to trial. Instead, he chose to hear out the first question.
“Essentially the trial is more about the science, because the judge can’t decide the science just by looking at papers filed by the court,” Verchick said. “What the defendants would love to do is win four-square on the discretionary function standard so they don’t have to go and talk about the science.”
Trials, Verchick said, are inevitably the time when such cases become about science.
Shortly before the trial began in late April, LSU decided that it would not renew the contract of Ivor van Heerden, the professor who led a study critical of the Corps and key to Katrina’s scientific debate. Some academics cried that LSU fired Van Heerden out of fear that his research would jeopardize government grants. Others argued van Heerden was speaking outside his area of professional expertise; he holds a doctorate in marine sciences, not engineering. The university, for its part, declined to offer an explanation.
Alongside the academic scandal, the scientific controversy around Mr. Go has risen to such a pitch that it seems unlikely one judge could mete out clarity.
“We don’t have a very good way of making people whole after terrible events like this, so we come to rely on these lawsuits, which are a very arbitrary way of solving the problem,” Verchick said. “You can win one if you get a good lawyer, if you’ve got the right scientists, or you can lose one. And if you lose one, you’re talking maybe hundreds of thousands people who are out of luck.”
He suggests establishing guidelines for Congress to create compensation funds similar to the one that aided the victims of Sept. 11. That fund — created by Congress in less than two weeks and ultimately used by 97 percent of the people eligible — required takers to swear off suing the airlines or the city of New York.
Such arrangements, which have been used by the government in only sporadic cases, would remove the need for post-disaster lawsuits that turn on one sworn-in scientist against another.
Replacing the threat of lawsuits might also tamp down the contentious climate in which disasters are studied. But one problem remains: the uncertainty itself. Who is willing to tell a rattled public that we don’t know why a disaster occurred, that we don’t know what its health effects will be? The alternative is to stake out some scientific ground, somewhere.
“Their mantra,” Button said of the TVA, “has been from Day One, ‘everything’s all right; there’s no problem,’ which is hard to believe because when it first occurred, how could we know?”
Back in East Tennessee, the EPA announced May 11 that it was taking over supervision of the cleanup, that it would resume its own testing for the first time since early January and audit the TVA’s continued sampling as well. The EPA’s order
also requires TVA to provide $50,000 for the community to hire an expert to help decipher technical documents. An EPA spokeswoman said the development wasn’t an indictment of TVA’s handling of the incident thus far, but members of the community, including Button, have greeted the news the same way they have every other announcement since the spill — with skepticism.
Are you on Facebook? Become our fan.
Follow us on Twitter.