|
Two cases of scientific misconduct have recently appeared in the media (1,
2). One involved J. Hendrik Schön, a 32-year-old physicist
at Bell Labs, whose “discoveries” in the area of superconductivity
and molecular electronics, had they been true, might have merited a Nobel prize.
The other was the withdrawal of the claim that element 118 had been discovered
at Lawrence Berkeley National Laboratory. The original claim was based on Victor
Ninov’s computer-aided processing of raw data from cyclotron experiments.
Looking on the bright side, scientific misconduct continues to be newsworthy because
it continues to be rare. During the past decade, fewer than 200 cases have been
reported by the NSF and the NIH combined, out of hundreds of thousands of projects
those agencies supported. Both the media and the public believe that most scientists
are honest and trustworthy and that scientists themselves will root out cases
of misconduct, insuring that the overall enterprise remains unsullied. The two
recent highly visible cases ultimately reinforce that trust, but they also raise
significant questions about how science detects and deals with misconduct.
Each case involved a collaborative research project with many scientists. One
might think that collaboration would reduce the likelihood of misconduct, since
collaborators could be expected to detect forged or misinterpreted data, but that
did not happen. Prior to publication neither the collaborators nor the referees
and editors of distinguished journals detected the problems that subsequently
came to light. An advantage of collaboration is that everyone need not repeat
every experiment, and different people bring different skills to a good collaboration,
so it is impossible for everyone to check everything. Trusting colleagues to get
things right is essential. Nevertheless, these cases of scientific misconduct
illustrate that considerable vigilance is required among collaborators to make
certain that spurious data or incorrect interpretations of data are not propagated.
What implications do these cases of scientific misconduct have for us as teachers
of chemistry? We definitely need to address issues of scientific ethics. The NIH
requires that ethics be taught to graduate students in departments that receive
NIH grants, and this is a very good thing. In science we depend too often on things
that go without saying. It is much better to bring ethical decisions and behavior
out into the open and discuss situations in which ethics comes into play. Often
those situations are less clear cut than we might expect.
In retrospect, some of the graphs published by Hendrik Schön and colleagues
look far too good to be true. I would have expected my TAs in a first-year lab
course to question baseline date with no discernible noise and a straight line
from which none of the points deviated, but of course I would not fault a TA who
accepted without question really good data from a really good student. We are
prone to assume that an expert will get things right, and that appears to be what
Schön’s collaborators did. Also, Schön’s results confirmed
an idea initially proposed by a colleague. It is easy to be uncritical when one
of our ideas is validated and we can tell others about it.
In the Lawrence Berkeley case, only Ninov knew how to run the computer program
that was used to analyze the data, so none of his colleagues dealt directly with
the raw data until other labs were unable to replicate the results—long
after the results were published. Following standard practice, the journal that
published the results did not include the raw data either. It was more than a
year before a colleague learned to use the software, analyzed the original data,
and found that the results reported earlier were not there. Analysis of a computer
log file gave evidence that data had been cut and pasted and numeric values had
been changed. Ninov claims innocence and points out that many others had access
to the computer containing the data. In any case several research groups spent
considerable time trying to reproduce results that were not supported by the raw
data.
These cases provide a good basis for discussions of scientific ethics, particularly
with respect to the responsibilities of colleagues in collaborative projects.
With increasing numbers of students working in cooperative or collaborative groups,
there may be opportunities for more than just discussion—similar issues
of responsibility apply to the members of such groups. Further, this is an area
where, “no clear, widely accepted standards of behavior exist” (1).
Thus there is an opportunity to point out to students that scientific ethics,
like science itself, is incomplete and needs constant attention to issues that
result from new paradigms such as collaborative research. Finally, each of us
can resolve to pay more attention to the contributions we and our colleagues make
to collaborative projects, applying to our own work no less critical an eye than
we would cast on the work of those we don’t know at all.
Literature Cited
- Chang, Kenneth. On Scientific Fakery and the Systems to Catch
It. New York Times, Oct 15, 2002, p D1; see also this Bell Labs press
release and report
(both accessed Oct 2002).
- Johnson, George. At Lawrence Berkeley, Physicists Say a Colleague
Took Them for a Ride. New York Times, Oct 15, 2002, p D1.
|
Citation
