S U P P L E M E N T
TO
L. M. Guenin, "Expressing a Consensus on Candour,"
Nature 402, 577-578 (1999).
It is suggested in the above-captioned commentary ("C") that self-promoting deception constitutes the common thread running though a consensus universe of research misconduct. In each case an investigator purports to have originated something that the investigator has not originated as described. Such deception may harm other scientists, funding agencies, and the public insofar as it may prevent them from seeing the investigator's work for what it is.
For convenience of reference, set forth in §1 hereof are the full texts of
[1] the Public Health Service ("PHS")'s and National Science
Foundation ("NSF")'s definitions of misconduct,
[2] the Office of Science and Technology Policy
("OSTP")'s proposed replacement definition ("the White House
proposal" or "WHP"), and
[3] the counterproposal introduced and defended in C,
to which are joined two procedural rules ("CP").
The discussion in the remainder of this supplement presupposes and does not purport to recount the arguments advanced in C. The comments offered in §2 identify further difficulties with WHP, difficulties that CP avoids. Additional remarks on the rationale for some components of CP appear in §3. An argument is given in §4 in support of OSTP's declination to proscribe certain other offences that do not constitute self-promoting deception. Apart from the logomachy about "misconduct," with respect to procedure it is argued in §5 that an accused scientist should be assured an agency hearing prior to any agency finding of guilt.
Erratum. In the print version of C, punctuation added by the editors inadvertently altered the sense of the last sentence, second paragraph, p. 577. The sentence appears correctly in a later PDF, and in the online text only version of C at http://www.nature.com: "On the premise that science deserves the best of scientific, moral, and legal thinking combined, I propose here a different definition--one that not only avoids counter-examples to which OSTP's well-intentioned proposal succumbs, but that sounds in the most fundamental notions of right and wrong pertinent to the experimental setting."
1. DEFINITIONS OF RESEARCH MISCONDUCT
1.1 Present Regulations
1.1[a] Public Health
Service
Misconduct or Misconduct in Science means fabrication, falsification, plagiarism, or other practices that seriously deviate from those that are commonly accepted within the scientific community for proposing, conducting, or reporting research. It does not include honest error or honest differences in interpretations or judgments of data (42 C.F.R. §50.102).
1.1[b] National
Science Foundation
Misconduct means
(1) Fabrication, falsification, plagiarism, or other serious
deviation from accepted practices in proposing, carrying out, or reporting
results from activities funded by NSF; or
(2) Retaliation of any kind against a person who reported or
provided information about suspected or alleged misconduct and who has not
acted in bad faith (45 C.F.R. §689.1[a]).
1.2 OSTP's Proposal (WHP)
[The following is excerpted from Office of Science and Technology Policy,
Proposed Federal Policy on Research Misconduct to Protect the Integrity of the
Research Record, 64 Fed. Reg.
55723 (Oct. 14, 1999), available at http://www.access.gpo.gov/su_docs/aces/aces140.html
(the "Notice"). The Notice goes on to set forth a
sketch, not reprinted here, of recommended investigatory and adjudicatory
principles. Those principles are general and do not appear to introduce any
innovations in the process heretofore administered by the PHS's Office of
Research Integrity ("ORI").]
I. Research Misconduct Defined
Research1 misconduct is defined as fabrication, falsification,
or plagiarism in proposing, performing, or reviewing research, or in reporting
research results.
§ Fabrication is making up results and recording or reporting them.
§ Falsification is manipulating research materials, equipment, or processes, or changing or omitting data or results such that the research is not accurately represented in the research record.2
§ Plagiarism is the appropriation of another person's ideas, processes, results, or words without giving appropriate credit, including those obtained through confidential review of others' research proposals and manuscripts.
§ Research misconduct does not include honest error or honest differences of opinion.
II. Findings of Research Misconduct
A finding of research misconduct requires
that:
§ There be a significant departure from accepted practices of the scientific community for maintaining the integrity of the research record;
§ The misconduct be committed intentionally, or knowingly, or in reckless disregard of accepted practices; and
§ The allegation be proven by a preponderance of evidence.
1Research,
as defined herein, includes all basic, applied, and demonstration research in
all fields of science, engineering, and mathematics.
2The research record is defined as the record of data
or results that embody the facts resulting from scientific inquiry, and
includes, for example, laboratory records, both physical and electronic,
research proposals, progress reports, abstracts, theses, oral presentations,
internal reports, and journal articles.
1.3 The Counterproposal (CP)
Research Misconduct Defined. It shall constitute research misconduct for an investigator in the conduct of research to commit misrepresentation, plagiarism, or misuse of another's work.
§ The conduct of research consists in (i) any inquiry funded by federal funds, or (ii) the preparation or communication related to any such inquiry of (a) a document published or provided to another, including any government agency, institution of higher education, research institution, or professional organization, or (b) a formal oral presentation.
§ An investigator is a person who participates in the conduct of research.
§ A fact is material if it or its contrary affects the plausibility, replicability, execution, or effect on safety of a result, hypothesis, or procedure, or is such that it would be rational for others to alter their course of conduct in reliance upon it.
§ Misrepresentation occurs if and only if an investigator, intentionally or recklessly indicating belief in the investigator's utterance, (i) delivers a false utterance concerning a material fact while believing the utterance false, or (ii) while believing the effect misleading, omits a material fact without which the utterance is misleading.
§ Plagiarism is the intentional presentation of the words of another as the presenter's own.
§ Misuse of another's work is the intentional presentation as the presenter's own, without attribution appropriate for the medium, of the ideas or work of another.
Procedure. No agency finding of research misconduct shall be made except after hearing. The agency shall bear the burden of proof to establish an alleged offence by a preponderance of the evidence.
2. FURTHER COMMENTS ON WHP
2.1 WHP Establishes Two Levels of Misconduct by Miscasting Elements of Offences as Conditions of a Finding
According to the first and
second clauses of WHP, II., an agency
may not enter a finding of misconduct unless "the misconduct"--said
to be defined in WHP,
Apart from the merits of
these requirements (already criticized in C), their status as
conditions on a finding serves to define two levels of wrong, (1) misconduct simpliciter
and (2) misconduct declared by an agency. This distinction wreaks havoc.
Misconduct as defined in WHP,
Only the third clause of WHP, II. is properly a condition of a finding. By this condition, WHP wisely preserves the status quo on burden of proof. The last sentence of CP states this rule more explicitly.
2.2 Blanket Standard of Intentionality or Recklessness Not Specific Enough
Even if inserted in the definition of a misconduct offence, the rule that "the misconduct be committed intentionally, or knowingly, or in reckless disregard of accepted practices" paints with too broad a brush. It produces such anomalies as "reckless fabrication" (how does one create something of whole cloth without specifically intending to do so?) and "reckless plagiarism."
Worse, this blanket phrase fails to state the specific intent necessary for any garden variety deception. With respect to misrepresentation, consider the following case. In the course of preparing a paper, with painstaking care Cloudsley converts a plethora of measured values from one system of measurement to another. He composes the paper in longhand, transcribes it with his word processor, submits the manuscript after careful study, and approves the proof. Some of the converted values are incorrect, but Cloudsley believes everything to be accurate. We would not say that Cloudsley has misrepresented anything. Yet in preparing the paper, at very many stages Cloudsley acted "intentionally" and "knowingly." According to the traditional view, we acquit Cloudsley because we infer that he formed no "intent to deceive." (According to WHP, Cloudsley's conversions were instances of "changing . . . data . . . such that the research is not accurately represented in the research record" but Cloudsley may plead "honest error." As we shall see in §2.3[i], "honest error" translates to action without intent to deceive.) Similarly might sanctions for deception attend "reckless disregard for the truth," but not recklessness simpliciter. In no case do "intentional" or "reckless" alone describe the mental condition required for misrepresentation.
As for "reckless disregard for accepted practices," this standard says nothing about the pertinent concern, the investigator's awareness that an utterance is false or misleading. Enforcing the former would require an agency to establish, and invite an accused to play sceptic against, an account of what practices are accepted. Where should they look? How much scientific testimony will be needed to resolve what is accepted in a given context? The debate may be internecine. This predicament stands among the objections to the ORI-NSF "deviation" clauses (§1.1).
The foregoing does not entail that one should adopt intent to deceive as the defining mental state for misrepresentation. A yet more precise result follows by adopting the signal and belief conditions that are said in C to define misrepresentation, of which more in §3.1[b].
With respect to plagiarism and misuse of another's work, in CP, "intentionally" couples with the notion of presenting something as one's own to form a necessary and sufficient mental condition for both offences.
2.3 Conjoining Blanket Standard With Exceptions for "Honest Error" and "Honest Differences of Opinion" is Convoluted, Imprecise, and Issues in Contradiction
2.3[a] "Honest Error"
The PHS and NSF definitions of research misconduct say nothing about intent. WHP mentions no specific intent. WHP's reference to "honest error" and "honest differences of opinion" rests in the following posture.
On its face, the PHS and NSF texts seem to condemn entirely inadvertent conduct. The only escape from this overbreadth in NSF's case must lie either in the claim that some specific intent is implicit in the respective offences or in the claim that a mere mistake does not constitute a "serious deviation from accepted practices." This escape is not open to WHP, for its definition of falsification and blanket mental state clearly condemns conduct such as Cloudsley's. The defense of Cloudsley under WHP must appeal to a sleight of hand in which WHP imitates the PHS text. "Honest error" and "honest differences of opinion," WHP declares, are not misconduct. Thus the right hand hath condemned but the left hand hath absolved.
Neither WHP nor the PHS text offer any definition of "honest." It is therefore unclear whether and how the foregoing legerdemain is expected to work. For as derived from honestus or "honorable," "honest" entails a complex of dispositions--a feature of persons, not utterances. Some time ago the Department of Health and Human Services Departmental Appeals Board was forced to sort this out. It cogently interpreted the exclusion of "honest error" to mean this: intent to deceive is a necessary element of deception (In re Popovic, Department of Health and Human Services Departmental Appeals Board, Research Integrity Adjudications Panel, Dec. No. 1446 [1993]). Having achieved this clarification, it is a step backward to revert to the two-handed legerdemain. To move forward, we should instead state a single condition. We may begin with the intent to deceive condition and improve upon it (§3.1[b]).
Moreover, to say that recklessness (or reckless disregard for the truth) is a sufficient mental state for misconduct and that honest error is not misconduct is self-contradictory. For one may err recklessly and yet "honestly" (here assuming for the latter either absence of intent to deceive or nonfulfillment of the signal and belief conditions). Suppose that (1) Smith recklessly publishes false data as a result of keeping sloppy records but (2) Smith does not believe the presentation to contain a false utterance or to be rendered misleading by omission of any material fact. Under WHP, Smith is innocent of misconduct by virtue of (2) and the exclusion of "honest error" (under any reasonable construal thereof), guilty by virtue of (1) and the sufficiency of recklessness. If recklessness is to remain as a sufficient mental state for culpable deception, this contradiction may be avoided only by dropping the exclusion of "honest error." Under CP, which needs no such exclusion, the matter is straightforward. Smith is innocent because Smith does not believe any utterance false or misleading.
2.3[b] "Honest Differences of Opinion"
When asked to apply the rule that "honest differences of opinion" are innocent, apart from asking for a definition of "honest," we must ask how an investigator would be accountable for "differences" in the first place. A "difference" presupposes the views of two or more persons. Whereas we test whether an investigator commits deception by reference to what the investigator alone says.
The important notion that an expression of opinion differs from a statement of fact ought to be clear by definition of deception, not ventured as an exception. CP secures a fact-opinion distinction long ago recognized in the interpretive history of misrepresentation. A person becomes culpable for misrepresentation only by an utterance concerning or an omission of a material "fact."
2.4 Asserting Jurisdiction Over "Reviewing" Research Can Achieve Only Overbreadth
According to WHP,
the scope of activity within which misconduct may occur consists in
"proposing, performing, or reviewing research" and "reporting
research results" (WHP,
As inserted in the definition of wrong, OSTP's well-intentioned alert backfires. We may observe its effects on three sets of persons: [a] mentors and collaborators, [b] referees and study section members, and [c] editors.
2.4[a] Mentors and Collaborators Become Culpable, If at All, By Preparing and Communicating Documents and Oral Presentations, Not By Reviewing Them
In respect of a mentor or collaborator, "reviewing" would appear to add nothing to "proposing," "performing," or "reporting" in WHP, I. To see this, we observe that the mere reviewing of a deceptive or plagiarized communication is insufficient for culpability. Misrepresentation, plagiarism, and misuse of another's work (and so too for the offences sketched in WHP) require more than mere reviewing. One need only consider the details of those offences to realize that it would be difficult to commit any of them without proposing, performing, or reporting research.
If "reviewing" is
not for this reason superfluous, all that the word could appear to add in WHP,
2.4[b] Referees and Study Section Members Do Not Commit Self-Promoting Deception in Their Capacities as Such
Suppose that referee or study section member Jeeves reads a submission, spots an intriguing passage, and immediately resolves to plagiarize it. Though we deplore Jeeves's mischievous intent, we cannot say that Jeeves's actions thus far--reading, learning, and intending to copy--constitute plagiarism or misuse. Plagiarism and misuse require that Jeeves present the submitter's work as Jeeves's own. Plagiarism or misuse must be a later event. If Jeeves plagiarizes, he does not do so qua reviewer. He does so qua author. Consequently in respect of referees and study section members, nothing is accomplished by a reference to "reviewing"--save for the detriments described in subsections [a] and [c].
2.4[c] No Tenable Reason for Trying to Reach the Work of Editors
WHP, II. invites charges against "reviewing" editors as accomplices in publishing those papers found to contain deceptive passages and unattributed borrowings. Thereby WHP either exceeds federal jurisdiction or extends it inappropriately. To prohibit conduct, the government needs some nexus between the conduct and something over which the government has authority. WHP offers none. A jurisdictional nexus is set forth in CP, which by its definition of "conduct of research," appropriately limits the concept of misconduct to an "inquiry funded by federal funds" and the preparation and communication of documents and formal oral presentations related to such an inquiry. Suppose an editor Biggles. Either Biggles is federally-supported (directly or indirectly) or Biggles is not. In the latter case, typically when Biggles works for a publisher, Biggles is immune from the principal penalty for misconduct, ineligibility for federal grants. In the former case, typically when he works in academic research, Biggles when acting as editor does not act on his own work. Should federal jurisdiction reach an editor's work on a manuscript concerning someone else's research unrelated to the publicly-funded research that the editor performs qua investigator? So long as an act of "reviewing" can constitute misconduct, this question requires an answer.
We can dispatch this matter as follows. There appears to be little incentive for an editor to conspire with a mendacious author. Rarely will an editor get the chance. A prevaricating author is unlikely to advise an editor of anything amiss in a manuscript. Hence there appears no serious risk that editors will become accomplices in deception. Therefore there is no reason to refer to "reviewing" as to them.
For all the foregoing reasons, no reference to reviewing appears in the definition of "conduct of research" or elsewhere in CP. This does not produce underbreadth. By greater precision, it avoids overbreadth. Within CP are encompassed all the appropriate acts of self-promoting deception by persons engaged in the conduct of research.
3. SUPPLEMENTARY REMARKS ON THE RATIONALE FOR CP
3.1 Misrepresentation
3.1[a] Misrepresentation in the Research Setting
Though misrepresentation has a long pedigree as a norm of conduct, the offence by that name in CP is one specially tailored with experimental research in view. The offence does not involve, for instance, reliance, causation, or damages. Nor, as earlier observed, is it predicated upon intent to deceive.
3.1[b] Signal and Belief Conditions More Specific and Precisely Targeted Than Intent to Deceive Condition
Courts have applied the notion of intent to deceive for centuries, but the notion succumbs to a counterexample. Suppose that Nelson, believing his utterance false, indicates his belief in the falsehood "The test result was positive." Charged with misrepresentation, Nelson rejoins, "My intent was not to accomplish deceit; I take no pleasure in deceiving. My intent was only to communicate my belief in 'The test result was positive.'" If misrepresentation requires intent to deceive, Nelson has cleverly refuted the charge. If the signalling and belief conditions in CP's definition of misrepresentation govern, Nelson is guilty. The latter accords with our common understanding. For this reason, in such definition CP employs the signal and belief conditions rather than the intent to deceive condition. The signal condition is met by "intentionally or recklessly indicating belief in the investigator's utterance"; the belief condition is met by "believing the utterance false" or, in the case of an omission, by "believing the effect misleading." Though it appears that application of the signal and belief conditions will often produce the same result as would insistence on intent to deceive, the advantages gained by employing them are that the signal and belief conditions
[i] identify the ingredients of intent to deceive,
[ii] state specific factual conditions into which a
factfinder is directed to inquire, and
[iii] avoid the Nelson counterexample.
3.1[c] Materiality
The notion of a "significant departure" from "accepted practices" (WHP, II., first clause), read together with the notion of "misconduct . . . committed" in disregard of accepted practices (WHP, II., second clause), describes an investigator's actions. Whereas for a claim of misconduct, the results of an investigators actions--typically the contents of a presentation--determines whether a wrong is serious enough to warrant sanctions. The point is made in C by analogy to judging pole vaulters. Results determine whether and to what extent an investigator has inflicted or threatened what makes deception wrong, namely, harm to others. (For public policy purposes, we neglect self-deception, long recognized as an insidious cost of untruthfulness.) The general name for a result threshold is "materiality." ("Significance" might also be used, but that risks confusion with the statistical and linguistic senses of that term.) Within CP, "material" works to take account of an utterance's effect with respect to the project to which it pertains and with respect to those whose learning or endeavors accompany or follow it.
3.1[d] Misleading Effect, the Standard for Omissions
Misrepresentation originally embraced only actions. Omissions gained recognition after it became clear that one could deceive by uttering only truths. As the history of interpreting the disclosure requirements of the federal securities laws has taught, to decide whether to condemn an omission requires one to decide whether what the utterer said is misleading. We reach the same conclusion if, following WHP, we ask whether the effect of an omission is that "the research is not accurately represented." We will not find an "inaccuracy" in the sense of a false utterance; if some falsehood had been uttered, that itself would have been a misrepresentation, and we would not be talking about an omission. To apply "accurately" here comes down to asking whether the utterer's presentation was misleading. This standard we hear in the everyday expression "His remarks would lead one to believe that . . . ." In establishing misleading effect and the investigator's belief in misleading effect (the latter being the pertinent belief condition) as elements of misrepresentation, CP presents two specific conditions. These capture what it is for an omission to be deceptive.
3.2 Rationale for Recognizing Misuse of Another's Work
The definitions of plagiarism and misuse of another's work follow the distinction between the two concepts developed by the American Historical Association, whose committees have unhappily had occasion to confront some clever instances of unattributed borrowing. In the AHA's words,
The misuse of the writings
of another author, even when one does not borrow the exact wording, can be
as unfair, as unethical, and as unprofessional as plagiarism . . . .
. . . . The clearest abuse is the use of another's language without quotation
marks and citation. More subtle abuses include the appropriation of concepts,
data, or notes all disguised in newly crafted sentences, or reference to a
borrowed work in an early note and then extensive further use without
attribution. All such tactics reflect an unworthy disregard for the
contributions of others ("Statement on Standards of Professional Conduct,"
pp. 13-14 [1993]).
The effect of the medium on the extent of attribution has been noted by both the AHA (ibid.) and the American Association of University Professors ("Statement on Plagiarism" [1989]).
4. OTHER OFFENCES WISELY NOT ENCOMPASSED
Some may recommend that misconduct include misappropriation of intellectual property. That offence consists in the intentional unauthorized appropriation and use of another's intellectual property. OSTP's choice not to proscribe that conduct in WHP is wise. To include misappropriation of intellectual property would be to sweep within research misconduct the following: patent infringement, misappropriation of trade secrets, and (perhaps) violation of nondisclosure agreements. (Copyright infringement is not listed here only because, from any conduct giving rise to a copyright infringement claim, a plagiarism claim is likely.) An intellectual property dispute is often vigorous and complex, sufficiently so that it may impede a university or agency investigation. Infringements also do not typically constitute, at least directly, deception in quest of distinction. They more resemble theft. Their usual motive is financial gain. Except for pursuing those unattributed borrowings defined in CP, to pursue infringement does not comport with the rationale for pursuing self-promoting deception. Governments pursue the latter in order to avert future waste of public funds.
On one occasion, ORI pursued a vandalism case against a scientist who allegedly destroyed another's work. Suit was brought under state law because the accused had violated no federal regulation about research misconduct (U. S. v. Arora, 860 F. Supp. 1091 [D. Md. 1994]). It has been reported (J. Kaiser, "A Misconduct Definition That Finally Sticks?", Science 286: 391 [1999]) that OSTP may view its definition of falsification as encompassing vandalism by one who disturbs the research record of another so that "the research is not accurately represented." Such a reading of WHP would be strained. Nothing in WHP's definitions mentions interaction with another or another's property. In any case, why would an agency sanction vandalism that disturbs records to the point of inaccuracy without sanctioning wholesale destruction that leaves no record to be inaccurate? If vandalism were to be proscribed, one could straightforwardly define it as intentional damage to tangible property used by another in research. But despite the possibility of a case in which A damages B's property so that A's work shows better in comparison with B's, vandalism does not usually accomplish self-promoting deception. In view of the apparently low incidence of this sort of conduct, it may be unnecessary at present to proscribe it.
5. PROCEDURE
As OSTP recognizes, the Notice's sketch of procedural goals speaks at a high level of generality. PHS may be said already to have in place the mechanisms that OSTP recommends. That is not to say that no changes in PHS procedures need occur. In the first instance, PHS has need of a single regulation codifying the procedures spread across PHS's present regulation and Federal Register procedural notices (56 Fed. Reg. 27384 [June 13, 1991] and 57 Fed. Reg. 53125 [November 6, 1992]). In the second place, it may be argued that the following matters should be addressed. The problem described in §5.1 obtains presently for PHS, but would seem likely to present itself to any agency that must fashion investigatory and adjudicatory procedures.
5.1 A Hearing Before a Finding
Current PHS procedure allows a hearing, and then only if the accused requests it, only after the agency has entered a finding of misconduct.
As noted, WHP allows a finding of research misconduct only if "the allegation be proved by a preponderance of the evidence." Perhaps OSTP did not intend by this to introduce a new procedural rule, but this provision appears to allow no misconduct finding until after a hearing. For if no hearing occurs, how is an allegation to "be proven"? If a law provides that a trucker may be convicted of carrying an excessive load only if the truck weighs more than x pounds, Jones cannot be convicted if the police never weigh Jones's truck. Whatever OSTP intended, a case may be made that an agency hearing should be a prerequisite for an agency finding of guilt. It may be contended that for a charge carrying such devastating consequences for a career as misconduct, no accused should be pronounced guilty by executive action without a hearing. (This point is further discussed in the work cited in C, n. 3, pp. 598-599, which offers multiple arguments against the suggestion that if a university conducts a hearing of some imagined quality, an agency might refuse a hearing.) Recent exonerations make obvious that government hearings are not superfluous. Thus the penultimate sentence of CP. The future agenda should address not only the legal but the financial availability of hearings. If a hearing were scheduled early in a case, not only would due process be enhanced, but costs to all parties might diminish merely because the case would not be protracted.
The unavailability of hearings prior to findings does not appear remedied by the recent reorganization effecting a shift of investigative responsibility in misconduct cases from ORI to the Office of Inspector General, Department of Health and Human Services. The notion has been broached that this move will solve another problem, the practice of allowing one entity to serve as both prosecutor and judge. The latter problem can scarcely be eliminated by a scheme that still leaves all functions within one department, but because it is not feasible to export adjudication from HHS, putting prosecution in one office and adjudication in another is at least an improvement. In respect of the availability of hearings, the new scheme offers no improvement. It may even be worse than the status quo ante. For it may be that for logistical reasons, the Assistant Secretary of Health will be disinclined to convene hearings prior to acting on the Inspector General's recommendations.
One cogent argument may be mustered against assuring a hearing in all cases. This is the argument that denial of grants is not analogous to punishment because no one is entitled to a grant, and hence that the extent of due process appropriate for a decree of grant ineligibility is less than that which one may fairly demand for other adverse agency actions. This argument gains support in the practice of other agencies. But the issue may be debated. Suffice it to say that consideration should be given to the interplay between nature of benefit and fairness in denying eligibility for it.
5.2 Privacy Act
While the Notice refers to the Freedom of Information Act, it is silent on more pressing issues about the status of investigatory records and case files under the Privacy Act. These issues cannot be discussed in detail here, but their pertinence may be indicated by posing a few questions. Should an agency declare its investigative office an "agency" for purposes of the Privacy Act, which would bar disclosure of records to other units of the agency? Or should an agency by administrative action exempt itself from requirements of the Act when records "compiled in reasonable anticipation of a civil action or proceeding" are exempted by the Act itself? Is it reasonable for an agency to withhold an entire file if only part of it would reveal the identity of an informant?
--L.M.G.