Fraud Charge Shakes Faith in Ground Rules of Science

The dispute over doctored data raged for 10 long years, a spectacle more appropriate for Hollywood than the ivy-covered science labs of Cambridge, Mass.

There was the arrogant Nobel laureate with the above-it-all air; the dedicated young post-doc who dared to accuse her superior of scientific fraud; the congressman on a crusade to make a point about public accountability; a pair of self-appointed truth police who grabbed onto the case like a dog grabs onto a bone.

The story even had a bard--a Caltech professor who spun the tale in a New Yorker magazine page-turner that became the talk of the town (he since has received several nibbles from Hollywood studios).

In the end, the stage was strewn with shattered lives and ruined careers, even though the scientific data in dispute was ultimately a minor player--and no fraud was ever proven.

Instead, the saga took on out-sized proportions because the real issue became the state of science itself: its keepers, its credibility, its responsibility to the public. The issue was no less than scientific truth, who gets to act as its guardians, and whether those guardians had become complacent and corrupt.

Indeed, that a squabble over relatively minor data could escalate into open war over the substance of scientific truth says much about the ways some basic ground rules in science are changing due to the explosive success of biomedicine in recent years.

“Scientists had always been a form of priesthood,” said Caltech science historian Daniel J. Kevles, who wrote the New Yorker piece. “And now that priesthood can manipulate life. We fear them, like the shamans.”

What is perhaps most remarkable is how far apart the sides remain 10 years after the dispute surfaced--despite last month’s decision by a review board from the U.S. Department of Health and Human Services that fraud charges were without substance.

“I don’t think it’s ever going to go away,” said biologist Margot O’Toole, the whistle-blower who brought the fraud charges in the first place.

The facts, as far as they can be determined, are these:

In 1986, Nobel Prize-winning molecular biologist David Baltimore, then at the Massachusetts Institute of Technology, published a paper in the journal Cell describing research into immune responses in mice--research with possible implications for the human immune system. The paper included results of experiments by his colleague Thereza Imanishi-Kari, a promising immunologist from Brazil.

*

The paper didn’t make a major splash. But O’Toole, who worked in Imanishi-Kari’s lab, said she grew increasingly uncomfortable with inconsistencies in her superior’s work and concluded that it drew unsupportable conclusions. She reported her objections but they were brushed off by higher-ups at MIT and nearby Tufts University, where Imanishi-Kari took an assistant professorship. All the while, Baltimore stood firmly by his colleague.

The case languished until a pair of self-styled fraud-busters at the National Institutes of Health got wind of it. Walter Stewart and Ned Feder took on the O’Toole case as part of a personal crusade against scientific misbehavior.

Unable to get the science establishment to pay attention, Feder and Stewart enlisted the help of Rep. John D. Dingell, a Michigan Democrat who chaired the House Energy and Commerce Committee and who was only too happy to topple a pair of scientists on the receiving end of government grants who felt they were above public scrutiny.

Baltimore circled his wagons, essentially telling Dingell to keep his hands off matters he couldn’t possibly understand. The congressman and the scientist clashed repeatedly in public hearings and the media in a contest that escalated into what UCLA’s vice chancellor for research called “a national circus.”

Secret Service investigations, ordered by Dingell, suggested that Imanishi-Kari’s notebooks indeed had been doctored to make sloppy work appear more solid. Baltimore was scolded harshly in the press and deserted by many of his colleagues. A prominent group at Harvard even suggested he should give back his Nobel Prize.

Eventually, Baltimore was pressured to resign his prestigious position as president of New York’s Rockefeller University and is doing research again at MIT. In 1994, the newly established federal Office of Research Integrity made it official, concluding that Imanishi-Kari was guilty of fabricating data and trying to cover it up. She appealed.

Last month, in an abrupt about-face, a unanimous three-member review panel not only cleared Imanishi-Kari of all charges, but went on to chastise the Office of Research Integrity for basing its case on evidence that was inconsistent, irrelevant and unreliable. Last week, Imanishi-Kari took her first step toward getting her life back: reapplying for the research post at Tufts that dissolved 10 years ago.

So what gives? Trying to piece together an objective picture of the events is like trying to find compromise in a bitter divorce. Neither party can understand how any objective person could look at the facts and fail to come down on their side.

The only thing both sides agree on is that the scientific community was caught with its proverbial pants down, unprepared to deal fairly with allegations of research fraud. “It was not my fault and it’s not Thereza’s fault,” O’Toole said. “There was no consensus. There was total confusion. No one had any idea.”

Caltech Vice Provost David Goodstein, who wrote regulations designed to prevent such a nasty episode from erupting at his institution, agreed: “A junior person came along and accused a senior person of wrongdoing, and they didn’t know what to do.” Essentially, he said, the scientists O’Toole approached with her complaints told her “go away, don’t bother me. And that just escalated the whole thing.”

In the wake of what has become known as the Baltimore scandal, Caltech and every other research university was ordered to come up with a formal set of rules for dealing with scientific misconduct as a condition for applying for government grants. But a new set of rules will not solve everything.

*

Deep differences of opinion in the scientific community muddy the issue of fraud and what should be done about it. “It’s extremely difficult to define scientific misconduct because we don’t know what the scientific method really is,” Goodstein said.

To show how murky things can get, each year Goodstein tells his students the strange case of Caltech founding father and patron saint, the late Michael Millikan. Then he asks them to judge the Nobel laureate according to Caltech’s current ethics criteria. The students can vote not guilty, guilty of fraud or guilty of misconduct.

“And every year, the vote splits,” he said. Almost no one votes for fraud but half the class finds Millikan guilty of misconduct.

Millikan’s experiment involved taking measurements of tiny oil droplets to determine their electric charge. At one point in a paper, he states that his data comes from “all of the drops experimented upon during 60 consecutive days.”

But a look at Millikan’s notebooks makes it clear he didn’t use all of the drops. Data for some drops is marked “Beauty--publish.” Others are marked “Very low--something wrong.”

“That’s normal,” Goodstein said. “Everyone agrees you have to use judgment.” But Millikan stated baldly that he’d used every drop. “In other words,” Goodstein said, “Millikan lied.”

There are mitigating complications. The placement of the statement that all drops were used suggested Millikan may have meant it to apply only to a small subset of the experiment. “We’re dealing with the real world,” Goodstein said. “It’s not simple.”

What Goodstein finds remarkable is that here is a case where everything is known. And still, scientists don’t agree on whether fraud was committed.

Certainly, selecting which data to use is critical to all science. Not all measurements are worth keeping. A power failure could cause a spike in current that would skew an experiment. A researcher could contaminate a sample, or drop it on the floor.

“All readings are not data,” said Harvard science historian Gerald Holton. “Sometimes you have to have the feeling in the tips of your fingers to understand what the difference is.” Einstein called this fingertip feeling Fingerspitzengefuehl.

Knowing which measurements actually describe nature is especially difficult at the forefront of science, said UCLA molecular biologist David Eisenberg, because data are frequently spotty and confusing.

“The facts never speak for themselves,” he said. “They’re always interpreted. Creativity in science consists in being able to see a pattern in those new data.”

That’s why, he said, scientists responded with “general relief” at Imanishi-Kari’s exoneration last month. “The fear was there would be an attempt at regulating what goes on in the lab in a way that would reduce creativity.”

On the other hand, scientists concede, a researcher in a hurry to publish a result might be tempted to fudge the data--especially if a competitor is hot on the trail. “Once you fall into that trap, an individual might take shortcuts,” said UCLA Vice Chancellor C. Kumar N. Patel. “It’s a slippery slope.”

Millikan himself may have been sliding down just such a path. At the time he published his paper, a competitor was working on the same problem.

Imanishi-Kari was not Millikan, however. Both her supporters and detractors say her case was much more clear-cut: Supporters because, they say, there is no case for fraud, even of the Millikan variety; detractors because, they say, the fraud was obvious.

“She [O’Toole] never had a case,” said Tufts professor Henry Wortis, O’Toole’s thesis advisor and one of the first people to hear of her charges against Imanishi-Kari. “From the beginning, the allegations could not be backed up by independent observations.”

As for O’Toole, she has the opposite view. “Every single scientist who examined the evidence said that she [Imanishi-Kari] had committed fraud and that I was telling the truth.” (O’Toole was the only person on the “losing” side who would speak on the record for this story.)

Even a clear-cut case of fraud doesn’t necessarily lead scientists to the same conclusions. Fraud isn’t a major threat to science, some say, because all important experiments are repeated. Concocting data, while unethical, does no serious damage to collective scientific truth. “But it wastes time and money and that’s why it’s unacceptable,” Patel said.

Meanwhile, biomedicine has exploded in recent years, in terms of both money and researchers. “It became a big bucks science very quickly,” Kevles said.

That means that even if the percentage of scientists committing fraud remains tiny, the absolute numbers can get big enough to worry about. And because there are still many more researchers needing grants than there is money in the till, the competition for dollars is fierce.

When public money gets big enough, says Kevles, people get understandably suspicious about who’s minding the store.

Still, money alone can’t explain the explosive reaction sparked by O’Toole’s charges over Imanishi-Kari’s notebooks. The disputed experiment cost under $100,000, while legal costs to pursue the fraud charges ran many millions.

Historians Kevles and Holton both see a pervasive anti-science bias at work--especially in Washington.

To some extent, they say, scientists brought this backlash on themselves by getting too caught up in the Myth of the Nobel Scientist, as Caltech’s Goodstein calls it. And to be sure, in the past, some scientists have tended to brush aside all criticism coming from outside their narrow circle. That Baltimore responded to the attacks on his colleague’s methods with self-righteous indignation made his critics all the more eager to bring the big-name scientist down.

“They’re supposed to be more moral than the rest of us,” Kevles said. “When they aren’t, it’s like finding a pederast among the priests.”

Currently, the generally accepted criteria for scientific fraud are “fabrication, falsification and plagiarism.” But a Health and Human Services commission is currently pushing for new rules, including a clause that would cover “other practices that seriously deviate from those that are commonly accepted within the scientific community. . . .”

That last provision has many scientists up in arms. How can they make forefront discoveries, they argue, without deviating from the norm? “That could mean any kind of unconventional thinking,” Goodstein said.

Which brings things back to Thereza Imanishi-Kari’s controversial notebooks. Even her supporters admit they were messy and disorganized, her research paper “rife with errors.”

But was she just sloppy and eccentric? Or did she commit fraud?

To a lot of people involved in the case, guilt or innocence doesn’t matter nearly as much as process. Due to the lack of an established procedure, Imanishi-Kari never had the opportunity, until her last appeal, to study the evidence against her and face her accusers. “There was no due process,” Patel said. “What Dingell carried out was a public lynching.”

The widely perceived unfairness is the reason Kevles decided to chronicle the story for the New Yorker. “I really didn’t know if she was guilty,” he said. “I was offended by the lack of fairness.”

O’Toole and company say due process isn’t an appropriate standard for scientists. A defendant in a trial has the right to remain silent, to see evidence, to confront accusers. “But you do not have a constitutional right to a government grant,” O’Toole said. “When you’re asked questions, you do not have the right to remain silent.”

Science, in other words, requires a different kind of candor than legal models of due process would suggest. Eventually, O’Toole says, scientists will learn the hard way that “handing their profession over to the lawyers did not serve them well.”

In the end, says Kevles, the story had all the markings of a classic Greek tragedy. “Everyone was behaving in accordance with their own belief system. Everyone believed what they were saying; or if they were lying, they were lying to themselves.”

It’s difficult for people to fathom how such a mangled series of events could happen in science. “Science is a process, and people don’t understand that,” he says. “It’s not a definitive statement. It’s, ‘This is how it looks to me now,’ not ‘This is how it is.’ ”

Or as Wortis summed it up: “We have a big social problem. Our lives are dominated by science and people don’t understand how it works.”