It was not the typical forum for an in-depth discussion of epidemiological methodology.

The complicated science of studying how and why people die is most often left to biostatisticians and medical journals.

But yesterday, as part of an ongoing effort to fuel the 20-year controversy over dioxin, the environmental group Greenpeace carried epidemiology to the general public, holding a news conference to charge that the design of four 10-year-old studies by the chemical giant Monsanto Co. of workers exposed to dixoin had been "cooked" and "manipulated" in order to convince federal officials that the chemical was not as dangerous as it really was.

In an exhaustive and densely written analysis, which has been submitted to the Environmental Protection Agency with a call to reassess the U.S. position on the suspected carcinogen, Greenpeace waded into a complex discussion of "surrogates" and "statistical power" and "exposed" and "unexposed" groups, footnoting heavily and referencing dozens of the most recent and sophisticated scientific publications on the safety of dioxin.

"These are studies of extraordinary influence" in shaping U.S. policy toward dioxin, said the report's author, Greenpeace analyst Joe Thornton. "There is a clear pattern of manipulation and scientific misconduct . . . to yield the predetermined result that dioxin was not harmful to human beings."

But Thornton acknowledged that he is not an epidemiologist or a scientist with graduate training and that the report was "not an epidemiological study of its own."

Monsanto officials said that report's charges came not from epidemiologists but attorneys engaged in litigation against them.

EPA officials appeared to be nonplused. Regardless of whether the Monsanto studies were right, they said, the government doesn't really use those types of epidemiology studies in forming U.S. policy. It uses toxicology studies.

"In general, the overall {epidemiological} data base on human carcinogen effects has been found to be inadequate," said William Farland, director of the EPA's office of health and environmental assessment. "Our risk assessment is based primarily on animal information."

The controversy over dioxin dates back to a series of mouse studies involving the chemical -- a byproduct of many industrial processes -- in the early 1970s that showed it to be highly carcinogenic at even the smallest doses.

Subsequently, dioxin became an environmental cause celebre, resulting in evacuation of the dioxin-contaminated communities of Times Beach, Mo., and Love Canal in Niagara Falls, N.Y., and sparking lengthy political and legal battles over the fate of Vietnam veterans exposed to the dioxin-containing defoliant Agent Orange.

In the past few years, however, many scientists and several Western countries have substantially relaxed estimates of the dangers of dioxin.

Part of the reason is that the animal studies used to extrapolate dioxin's human safety have been discredited and contrary data has emerged. Another reason is that repeated epidemiological studies of people exposed to higher than normal levels of dioxin -- as a result of, for example, working in a chemical plant -- have shown no serious ill-effects.

Greenpeace, in its petition to the EPA, takes aim at four of the most prominent of those epidemiological studies, done by Monsanto on workers at its Nitro, W.Va., plant.

The company, Greenpeace said, was able to show no ill-effects from dioxin among those exposed only through a series of "criminal and quasi-criminal" distortions of their studies.

The EPA said yesterday it is looking into Greenpeace's charges.

Among the allegations:That rather than, as is traditionally the case in an epidemiological study, comparing the health of those exposed to the chemical in question with the health of those who weren't exposed, Monsanto mixed up the two groups. Exposed workers were thus compared to exposed workers, making it impossible to tell whether dioxin had an independent deleterious effect. As proof, the group said that four employees who were listed as exposed in one study were switched to the unexposed group in another. That nine workers who developed cancer and 11 who developed heart disease were deliberately left out of the group whose health was studied, making the overall health of the workers look better than it actually was.

Monsanto officials say that none of these charges are evidence of fraud. Workers moved from exposed to unexposed groups, they said, because the definition of who was exposed varied from study to study.

One study, for example, looked only at those who were directly affected by a plant explosion in 1949. Another focused on those who received low, steady doses of dioxin over many years.

Some workers were left out, they said, because they died after the cut-off point for the study or did not meet's the study's requirements. Maintenance men, for example, even if they died prematurely of cancer, were not studied because there was no way to know whether they had been exposed to dioxin. In other cases, they said, simple mistakes were made in analyzing the volumes of 25-year-old employment information on Monsanto workers.

Jim Collins, head of Monsanto's epidemiology department, said the company has since recalculated its data, correcting earlier mistakes and adding people who died after the cut-off point. There is no difference in the end, he said. "No matter how you cut the data, you get the same results," he said.

Added Monsanto spokesman Dan Bishop, "Greenpeace has been using this trumped-up ploy about our health studies in its direct-mail fund-raising activities for several weeks, and is now going public to garner widespread publicity for their fund drive."