Professional Documents
Culture Documents
American Scientist
the magazine of Sigma Xi, The Scientific Research Society
This reprint is provided for personal and noncommercial use. For any other use, please send a request to Permissions,
American Scientist, P.O. Box 13975, Research Triangle Park, NC, 27709, U.S.A., or by electronic mail to perms@amsci.org.
©Sigma Xi, The Scientific Research Society and other rightsholders
Macroscope
M
ly published by others.
them from their We shall avoid speculating about the
uch has been written on the motives for plagiarism here. Generally,
ethics of plagiarism. One context represents an however, the ethical dilemma seems to
aspect that has received less notice be analogous to the person who robs
is plagiarism’s role in corrupting our act of plagiarism a store to feed his or her family, or the
ability to learn from data: We propose politician who lies to achieve a larger
that plagiarism is a statistical crime. It political goal. In all of these cases, the
involves the hiding of important infor- behavior in question is generally recog-
mation regarding the source and con- A statistical perspective on plagia- nized to be unethical, so if the broader
text of the copied work in its original rism might seem relevant only to cases context in which the action takes place
form. Such information can dramatical- in which raw data are unceremoniously is deemed ethical, it can only be thus be-
ly alter the statistical inferences made and secretively transferred from one urn cause the unethical action serves some
about the work. to another. But statistical consequences larger, more important goal. In Weg-
In statistics, throwing away data is also result from plagiarism of a very man’s case, no such argument about a
a no-no. From a classical perspective, different kind of material: stories. To un- larger context has been made (perhaps
inferences are determined by the sam- derestimate the importance of contex- because that would require admitting
pling process: point estimates, confi- tual information, even when it does not the ethical violation in the first place).
dence intervals and hypothesis tests concern numbers, is dangerous. The Wegman case came to public
all require knowledge of (or assump- Perhaps the most prominent statisti- notice after the Canadian blog Deep Cli-
tions about) the probability distribu- cian to have repeatedly published mate- mate found the first few pages of mate-
tion of the observed data. In a Bayesian rial written by others without attribu- rial in the report to be plagiarized from
analysis, it is necessary to include in tion is Edward Wegman, formerly of the a book by Ray Bradley, one of the au-
the model all variables that are relevant Office of Naval Research and currently thors whose work was attacked in that
to the data-collection process. In either a professor at George Mason Univer- document. The blog post stirred oth-
case, we are generally led to faulty in- sity. The case is especially interesting ers to study this and other documents
ferences if we are given data from urn because Wegman has a distinguished written by Wegman and his students,
A and told they came from urn B. record of public service and scholar- at which point additional incidents of
ship (he received the Founders Award copying without attribution turned up.
in 2002 from the American Statistical In 2011, a related article by Wegman
Andrew Gelman is a professor in the departments of Association) and because one of the pla- and a collaborator in the journal Com-
statistics and political science at Columbia Univer- giarized documents was part of a report putational Science and Data Analysis was
sity, New York, and the author of Red State, Blue on climate change delivered to the U.S. formally retracted by the publisher on
State, Rich State, Poor State: Why Americans
Congress. The ethical dimensions of this grounds of plagiarism.
Vote the Way They Do (Princeton University Press,
2008). Thomas Basbøll is an independent writing
copying seem clear enough: By taking Despite the human and political dra-
coach and an external lecturer in the department of others’ work without giving credit— ma of the Wegman case, it may not ap-
management, politics and philosophy at the Copenha- even copying from Wikipedia at one pear immediately interesting from the
gen Business School. Address for Gelman: Columbia point (see the appendix to this essay at standpoint of statistics. Perhaps counter-
University, 1016 Social Work Building, New York, American Scientist’s website)—Wegman intuitively, a purely qualitative example
NY 10027. E-mail: gelman@stat.columbia.edu and his research team were implicitly reveals why this appearance is wrong.
168 American Scientist, Volume 101 © 2013 Brian Hayes. Reproduction with permission only.
Contact bhayes@amsci.org.
Stephanie Freese
A poem by Miroslav Holub that appeared in the Times Literary Supplement in 1977 tells the story of a Hungarian reconaissance unit caught in
a snowstorm in the Alps. Holub recounts how the lieutenant who sent the unit out feared for their lives—but that the unit returned after three
days, saying that one of their number had a map. The map, however, turns out to be of the Pyrénées, not the Alps. The story has been widely
retold in the field of organization studies. Whether and how its source is cited, the authors argue, is a matter of statistical concern.
Snowstorm, Map, Conundrum nied by the slogan, “When you are as Kahneman noted in his talk, some
An anecdote that has been widely circu- lost, any old map will do.” It was even irregularities in Weick’s referencing (or
lated in the organization studies litera- retold by noted psychologist Daniel lack thereof) have emerged.
ture goes something like this: A group of Kahneman at the 2009 Digital Live In 2006, one of us (Basbøll), and a
soldiers are sent out by their leader and Design conference as part of an ac- Ph.D. student in his department, Hen-
get lost in a snowstorm in the Alps. After count of the importance of confidence. rik Graham, published a paper show-
discovering that one of their number has Kahneman attributed the story to the ing that Weick had simply transcribed
a map, they regain their confidence, wait “famous organizational psychologist the story from a poem by Miroslav Hol-
out the storm and return to camp. Only Karl Weick.” Weick, like Wegman, is an ub that had been published in 1977 in
afterward do they realize that the map award-winning and highly regarded the Times Literary Supplement. The text
was not of the Alps but of the Pyrénées. scholar in his field, and he is the com- has minor changes but is nearly identi-
This story has made the rounds in monly cited source for the anecdote in cal to Holub’s—without the line breaks,
management circles, often accompa- the organization studies literature. But, of course. (See the online appendix to