New York Review of Books
Volume 51, Number 18 · November 18, 2004


Review
Dishonesty in Science
By Richard C. Lewontin

Scientific Integrity in Policymaking: An Investigation into the Bush
Administration's Misuse of Science
a report by the Union of Concerned Scientists
February 2004, 42 pp.

The Great Betrayal: Fraud in Science
by Horace Freeland Judson
Harcourt, 463 pp., $28.00

1.
The founders of the American state understood that the proper
functioning of a democracy required an educated electorate. It is
this understanding that justifies a system of public education and
that led slaveholders to resist the spread of literacy among thier
chattels. But the meaning of "educated" has changed beyond
recognition in two hundred years. Reading, writing, and arithmetic
are no longer sufficient to decide on public policy. Now we need
quantum mechanics and molecular biology. The knowledge required for
political rationality, once available to the masses, is now in the
possession of a specially educated elite, a situation that creates a
series of tensions and contradictions in the operation of
representative democracy.

The problem of the role of elite knowledge in a democracy is an old
one. A version of a story in the Babylonian Talmud tells of four
rabbis walking in a field, engaged in a dispute over whether an oven
of a particular design can be purified. Three hold one opinion, while
the fourth has the opposite view. The lone holdout appeals to God,
asking that He send first thunder, then lightning, and then that the
lightning strike a lone tree in the field. Although each request is
granted, the others are not convinced. After all, thunder and
lightning are usual natural phenomena and in a lightning storm what
is more natural than that a tree standing in the middle of a field
should be struck? In desperation the dissenter calls on God to speak
directly to them. Sure enough, a voice from above is heard
proclaiming "IT IS AS HE SAYS." "So," asks the dissenter, "what do
you three have to say now?" "All right," they answer, "that makes it
three to two."

Science has replaced Jehovah as the source of privileged knowledge,
but the problems remain. How is the knowledge in the possession of
the scientific elites to be factored into a process of decision in
which considerations of economy, ideology, and political power also
enter? Is elite knowledge to be given absolute priority? Why should
we trust scientists, who, after all, have their own political and
economic agendas? On the other hand how can we decide by vote when
the voters and their representatives have no understanding of the
facts of nature?


The American government, like others, has attempted to solve the
problem by co-opting scientists into the apparatus of the state in
three ways. Most directly it has built an executive apparatus
including the president's science adviser, the Office of Science and
Technology, and regulatory bodies like the Environmental Protection
Agency. Second, it has created quasi-governmental bodies made up of
senior scientists, like the National Academy of Sciences and the
National Research Council, that are obliged to provide expert
scientific advice and evaluation on request from any government
agency. Finally, after the Second World War, the state became the
chief patron of science, currently committing about $35 billion
annually directly to basic and applied research.

Because of fears that federal support of research would result in
political interference with the research process, the pattern
established by Congress for the funding of research gave the
representatives of the scientific community itself the day-to-day
power to decide what research is to be done. Even in funding from the
Department of Defense and the Department of Energy (formerly the
Atomic Energy Commission), decisions about grants are made by a peer
review system in which the reviewers and agency administrators are
drawn from the research community and share in its general culture.

As a consequence, instead of producing a mass of grateful recipients
of state patronage, public research support has created a large and
prospering community of independent investigators-most of them
affiliated with universities-with immense public prestige and with
effective control over the distribution of funds for research. It is
no surprise that attempts by various administrations to make science
serve political and economic policy have been met by public
opposition from prestigious scientists speaking in the name of
disinterested objectivity. The most recent example is the report
issued in February 2004 by the Union of Concerned Scientists, whose
signers include twenty Nobel Prize winners and, ironically, nineteen
recipients of the American government's version of a knighthood, the
National Medal of Science.


Scientific Integrity in Policymaking accuses the Bush administration,
first, of deliberately suppressing scientific findings in the
interest of its own ideological and political ends and, second, of
packing various regulatory and review boards with unqualified members
who can be counted on to favor industrial profits or conservative
ideologies over public health and safety. Manipulation, distortion,
and suppression of scientific findings in the interest of industries,
the report shows, have affected research results on climate change,
on mercury emissions and other pollutants, on airborne bacteria, on
endangered species and forest management. The government's evidence
about Iraq's famous aluminum tubes is said to have been
misrepresented in the interests of building a case for war.

Three examples are given of the way in which education and
information about scientific findings have been manipulated to
support a conservative religious ideology. In order to demonstrate
that abstinence-only programs were effective, the Bush administration
instructed the Centers for Disease Control not to follow the actual
birth rate for participants in an abstinence-only test program, but
only their attendance and attitudes toward the program. In order to
hide the effectiveness of condom use in preventing HIV infection, the
CDC was directed to emphasize condom failure rates in its educational
material. Finally, the National Cancer Institute was directed to post
a claim on its Web site that abortion promotes breast cancer although
a large study had shown no connection between them.

The report also discusses a number of cases in which government
regulatory and review panels were packed with members favorable to
the administration. Moreover, it is reported that many potential
nominees for federal scientific advisory posts were questioned about
their political views and even whether they had voted for Bush. The
most transparent manipulation occurred in 2002 when the Center for
Disease Control Advisory Committee on Childhood Lead Poisoning was to
consider narrowing the criterion of lead poisoning, so that sources
of poisoning that were formerly banned became permissible. A panel of
new nominees for the Advisory Committee was proposed by the CDC and,
for the first time in the history of the committee, nominees were
rejected by the direct intervention of the secretary of health and
human services, Tommy Thompson, who replaced them with five persons
who were previously known to oppose tightening the standard. Two of
the five had financial ties with the lead industry.

In April the President's science adviser, John Marburger, issued a
reply to the Union of Concerned Scientists, providing an explanation
for each of the claimed abuses, including the defense that reports
were not suppressed but held up pending more complete studies.[1] In
the case of the abstinence-only program, Marburger says that it was
never designed as a scientific study, but as preliminary to long-
range evaluation of sexual abstinence. The accusation that seems
least easily dismissed is that judgments about political commitment
rather than about scientific competence were applied when people were
appointed to advisory panels. Obviously each claimed abuse can be
explained away, but whether the explanations are convincing or
whether a pattern of politicizing scientific policy makes itself
manifest will depend on what the reader is inclined to believe in the
first place.

In July yet another case of political prejudice was reported.[2] When
Torsten Wiesel, a Nobel laureate in physiology and medicine, was
rejected by Tommy Thompson's office as a candidate for the advisory
board of the Fogarty Center at the NIH, the director of the center
was told by an official from the Department of Health and Human
Services that Wiesel had "signed too many full-page letters in The
New York Times critical of President Bush." Indeed, the government
makes no apology for the use of criteria other than scientific
competence in its appointment policy. According to the report in
Nature, a spokesman for the DHHS has asserted that, in addition to
competence, a diversity of gender, race, geography, and political
opinion is a valid goal of appointments to scientific advisory boards.


This assertion brings us back to the original problem of the relation
of elite knowledge to the political process. Studies of climate
change, endangered species, acceptable pollution levels, or the
effect of sexual practices have not been called forth by pure
scientific curiosity. Like all processes that are of direct relevance
to human physical and psychic welfare, the costs and benefits of
decisions will fall differently on different people. Any amount of
lead is bad for your health. So what should be the minimum acceptable
level of lead in the bloodstream? Whose bloodstream? Acceptable to
whom? The worker in a lead refinery who lives in a badly polluted
neighborhood near the plant whose family will bear the cost to their
health and longevity of too much lead? The owner of shares in the
refinery who winters in Sedona and summers on Cape Cod, whose health
is not at issue but who will bear an economic cost of pollution
control? A popular bumper sticker in Vermont reads "Another Vermonter
for Global Warming." (That, of course, may be a scientific mistake,
since general global warming may make Vermont colder.)

My friends who are lawyers insist that the only general rule for
deciding legal disputes is "It depends on the jurisdiction," and that
rule applies equally to decisions about scientific questions of
public import. It is disingenuous to claim that scientists come to
their scientific work without prior ethical, economic, and social
values and motivations. Everyone I know who studies endangered
species cares about saving them. One never hears that the malarial
parasite is "endangered." To do science is to be political if only
because it is a political decision to spend some amount of limited
human energy and social resources on a particular question. Most
scientists are, at a minimum, liberals, although it is by no means
obvious why this should be so. Despite the fact that all of the
molecular biologists of my acquaintance are shareholders in or
advisers to biotechnology firms, the chief political controversy in
the scientific community seems to be whether it is wise to vote for
Ralph Nader this time. We might expect, then, that the actions of an
administration strongly protective of the interests of the owners of
capital and identifying itself culturally with religious
fundamentalism should be the cause of protest.

If knowledge about the natural world is to rationally influence the
decisions of an informed electorate, then people must believe that
scientists tell the truth about nature insofar as they know it. While
we might agree that prior political commitment could lead us to ask
one question rather than another, or to put more weight on the result
of a study that conforms to our prejudice rather than one that
refutes it, every scientist must agree that outright fraud is beyond
the pale. Putting aside the issue of morality, scientific
investigation would be destroyed as a useful human endeavor and
scientists would lose any claim on social resources if deliberate
falsifications were not exposed. So scientists must be on the alert,
ready to detect lies arising from within their institution. But this
leads to a contradiction. To survive, science must expose dishonesty,
but every such public exposure produces cynicism about the purity and
disinterestedness of the institution and provides fuel for
ideological anti-rationalism. The revelation that the paradoxical
Piltdown Man fossil skull was, in fact, a hoax was a great relief to
perplexed paleontologists but a cause of great exultation in Texas
tabernacles.


2.
Horace Freeland Judson, a science journalist who had previously
written a narrative of the development of molecular biology, has now
produced a nuanced and sophisticated yet accessible view of
scientific fraud.[3] The Great Betrayal is not simply a narrative of
scandals, but places var-ious instances of scientific bad behavior in
the context of general social pressures and their manifestation in
the scientific community. He reminds us that the drive for economic
suc-cess, personal power, and the gratification of one's ego has led
over and over to dishonesty, fraud, and wickedness in business, the
church, and the state. Why do we think that the devotees of Newton's
laws will be more saintly than those ruled by Cardinal Law?

Judson discusses fraud of three sorts: fabrication, falsification,
and plagiarism. Fabrication is the creation of claimed observations
and facts out of whole cloth. These are just plain lies.
Falsification is the trimming and adjustment of the results of
genuine experiments so that they come to be in agreement with a
desired conclusion. Numbers may be "adjusted"; there may be a
conscious dishonesty or, more subtly, observations that are not in
sufficient agreement with the theory may be discounted, often on the
basis of ad hoc criteria which the investigator comes to believe are
perfectly valid after the fact. There are, after all, many
experimental observations that are flawed for one reason or another
and ought to be discounted. The problem of how to cull observations
honestly is a constant preoccupation of statisticians and
methodologists. Judson includes in the category of plagiarism not
simply the copying of others' written texts without attribution, but
the appropriation of experimental design or data or experimental
material or credit for work that belongs to others. An acquaintance
of mine once was refused a strain of virus possessed by a prominent
investigator, so he cleverly soaked the letter of refusal in an
appropriate liquid and recovered enough virus to start his own
cultures.

Judson begins his excursion through the history of frauds with those
committed by both acknowledged heroes and villains of science: Isaac
Newton, Gregor Mendel, Louis Pasteur, Robert Millikan, Sigmund Freud,
and Cyril Burt. Some, like Burt and Freud, simply made up
observations out of their heads to justify their theories. Burt's
fabricated results on the heredity of IQ were so transparent (he even
invented fictitious collaborators) as to suggest real pathology.
Millikan's measurements of the electrostatic charge on the electron
were a classic case of discounting as aberrant the observations that
did not fit well with his theory.

Mendel's is a more interesting case, although we have only
speculation about his behavior. His reported numbers of different
types of offspring from various experimental crosses were too close
to his expected genetic ratios of 3:1 and 1:1 to be the outcome of
counting a relatively small number of seeds or plants in a real
sample. The toss of a real coin one hundred times is very unlikely to
give exactly 50:50, 51:49, or 49:51 heads to tails every time we do
it. Mendel was probably not dishonest but an unconscious innocent
victim of "optional stopping." If you are counting objects of
different types without a determination to count a total of exactly,
say, five hundred, then eventually you get tired of the whole thing
and decide that you have done enough. But if you have a prior theory
about how the results should look and are keeping a run-ning tally of
the counts there is a tendency, which must be consciously resisted,
to say "Enough!" when the results look good. That is not fraud, just
bad experimental practice, easy enough to fall into at a time when
neither statistics nor psychology was well developed.


A large part of The Great Betrayal is taken up with famous modern
cases of fraud, among them the so-called "Baltimore Affair." An
assistant professor at MIT, Thereza Imanishi-Kari, was engaged in a
collaboration with the laboratory of one of the most prominent
scientists of the day, David Baltimore, a Nobel Prize winner,
director of a large research institute, and soon to be president of
Rockefeller University. The actual work of the collaboration involved
some postdoctoral fellows and research associates and one result was
a paper, coauthored by several of these dependent investigators
together with Imanishi-Kari and Baltimore, which made an interesting
general claim about the nature of immune systems. A postdoctoral
fellow in Imanishi-Kari's laboratory, Margot O'Toole, discovered one
day while looking at some notebooks that Imanishi-Kari had made some
observations that contradicted the claims of the paper. O'Toole
herself had made similar observations but these had been angrily
dismissed by Imanishi-Kari.

O'Toole took her findings to authorities at MIT and at Tufts
University, where Imanishi-Kari had just been appointed, and this
resulted in a meeting with her, Baltimore, Imanishi-Kari, and
administrators of the universities. O'Toole's request that the pub-
lished paper be retracted or changed was refused and that seemed the
unsatisfactory end of the affair. Then the story reached the public
press and the affair escalated with an investigation by NIH
committees, a congressional hearing, and more attention in the
scientific and popular press. During the investigations and appeals
Imanishi-Kari introduced new documentary evidence of research records
intended to support her claims. A Secret Service examination,
undertaken at the request of a congressional committee, declared
these documents to be fraudulent constructions, thus adding the
accusation of deliberate fabrication to the original suspicion of
data suppression. However, a highly qualified forensic expert engaged
by Imanishi-Kari's attorney examined the documents and concluded that
the Secret Service's analysis was "erroneous" and did not support the
accusation of fabrication.

As a consequence of the uproar, O'Toole was not reappointed to her
postdoctoral position, Imanishi-Kari was suspended from her job at
Tufts, and Baltimore was forced to resign as president of Rockefeller
University. Finally, after ten years of dispute, the affair ended
with the Solomonic judgment of an NIH appeals committee that the
charges against Imanishi-Kari had not been proven "by a preponderance
of the evidence." In the end, O'Toole got a research position at a
research institute, Imanishi-Kari was reinstated at Tufts, and David
Baltimore was made president of California Institute of Technology.
[4]

The Baltimore Affair and a few others are notorious examples of more
numerous cases of scientific dishonesty, most of which do not reach
the attention of the public. We do not know how often scientific
fraud of various degrees of conscious dishonesty occurs, nor can we
ever know. Some scientific work is of sufficient general relevance
that false claims will eventually be contradicted by other
observations and in the end, after a certain amount of stumbling
around, the truth about nature will emerge. Moreover, some
falsification is in support of what turn out to be true theories, as
in the cases of Pasteur and Millikan.

Most of science, however, is immune to future verification or
refutation because the link between the reported findings and other
active branches of investigation are too weak to allow for
contradiction or because the reported results are in support of an
already verified phenomenon, or because the subject is so esoteric
and narrow that no one else cares to work on it. Judson gives the
data from a few sociological surveys in which respondents were asked
if they knew of cases of falsification or fabrication, but the
results cannot be used to estimate the frequency of such events among
all published scientific reports. The claim by a former editor of
Science, the journal of the American Association for the Advancement
of Science, that "we must recognize that 99.9999 percent of reports
are accurate and truthful" is either a fabrication or a
falsification, depending on whether he invented it on the spur of the
moment or was misrepresenting some actual data by throwing in some
extra 9's.[5]

Despite the sophistication of Judson's analysis he has missed a
pervasive dishonesty in the practice of science that makes a certain
level of intellectual corruption characteristic of the institution.
The dishonesty consists in the way credit for scientific research is
falsely ascribed to some of the authors of jointly signed scientific
papers. He brushes by this practice by referring to "gift
authorship," but, far from a willing gift, it is an exaction that the
powerful impose on the weak. Science is carried out for the most part
in a collection of cottage industries, work groups
called "laboratories," but that is a synecdoche. The group is headed
by a senior scientist, sometimes accompanied by a more junior but
established colleague, and includes postdoctoral fellows, research
associates, graduate students, visiting scientists, and technical
assistants all working in offices and laboratory rooms clustered
around the laboratory director's own space.

It is almost always the case that the laboratory director performs no
actual experimental work. There is considerable variation from
laboratory to laboratory and from project to project within the
laboratory in the degree to which the senior scientist participates
in the conception, planning, supervision, and eventual writing-up of
the work. In many cases the entire project from conception to
publication is without any significant input from the director. Much
of what is done, however, is supported by funds from various grants
and contracts obtained by the director as the euphemistically
named "principal investigator."

Regardless of the actual involvement of the laboratory director in
the intellectual and physical work of a research project, he or she
has unchallenged intellectual property rights in the project, much as
a lord had unchallenged property rights in the product of serfs or
peasants occupying dependent lands. The chief product of a laboratory
is in the form of published papers and the chief manifestation of the
director's intellectual property rights is that he or she will be
coauthor on every publication from the laboratory, sometimes
including even general review papers and book chapters written by
subordinate group members.

Such property rights explain how, for example, Professor Eugene
Braunwald of the Harvard Medical School came to be an author, at the
age of fifty, of over six hundred publications.[6] Unfortunately for
Braunwald, one of his protégés and coauthors, John Darsee, turned out
to be a detected fabricator. One wonders how many sleepless nights
Braunwald spent worrying about those other publications. But if
laboratory directors as a matter of course claim authorship of work
to which they have made no intellectual contribution or only a
trivial one then they are, year in and year out, committing an
intellectual fraud from which they reap immense rewards of ego,
prestige, income, and social power. Moreover, by an unconscious
affirmation on the part of the scientific community as a whole, these
rewards grow autocatalytic. Robert Merton, the founder of modern
social studies of science, called attention to a phenomenon he named
the "Matthew Effect" after Matthew 25:29:

For unto every one that hath shall be given, and he shall have
abundance; but from him that hath not shall be taken away even that
which he hath.

Irrespective of the order of authors on a paper, it is referred to
informally and sometimes formally by the name of the best-known
author. In laboratory libraries papers are filed under the name of
the "senior" author and remembered and discussed under his or her
name. I was an indignant witness to an extreme case of the Matthew
Effect. A graduate student in my laboratory had published a seminal
paper, without my name on it, on an enzyme called alcohol
dehydrogenase that everyone agrees has revolutionized the
experimental study of population genetics. Shortly afterward I gave a
lecture on a different subject, at the end of which a colleague came
up from the audience and said, "That was very interesting but what I
really admire is your paper on alcohol dehydrogenase." There is some
justice in the world, however, and the misappropriation of
intellectual property occasionally means that one may try to pass a
bad check. The Matthew Effect then does its work. The fraud
attributed to Imanishi-Kari becomes known as the "Baltimore Affair."
To them that hath it shall be given.

Scientists in training are conscious of the appropriation of credit
for their work by senior scientists and they resent it but feel that
they cannot protest. It is not that they place no value on the
details of authorship. They will fight bitterly with colleagues of
their own rank about who should be first author on jointly authored
publications. Yet when they too become seniors they will engage in
the same fabrications of intellectual credit. The fabrications and
falsifications of scientific results that we condemn as fraud are
carried out from the desire for fame, status, and economic reward.
But the misappropriation of credit by senior scientists arises from
the same motives. How can we expect scientists to hold literal truth
about nature as an inviolable standard, when they participate, en
masse, in a conscious everyday falsification about the production of
that truth? That is an aspect of what Judson calls "the culture of
fraud" that is far more relevant to scientific honesty than the
behavior of the executives of Enron on whom most scientists claim to
look with disdain.

Notes
[1] A report of Marburger's reply is given in Nature, April 8, 2004,
p. 589.

[2] Nature, July 15, 2004, p. 281.

[3] The book on molecular biology is The Eighth Day of Creation
(Simon and Schuster, 1979).

[4] For a much more detailed history of the case, see David Hull's
essay in The New York Review, December 3, 1998, and the book by
Daniel Kevles that he was reviewing, The Baltimore Case (Norton,
1998).

[5] Daniel E. Koshland Jr., editorial, Science, January 9, 1987.

[6] See Judson's account on p. 113.