NATIONAL SCIENCE FOUNDATION OFFICE OF INSPECTOR GENERAL OFFICE OF INVESTIGATIONS CLOSEOUT MEMORANDUM 11 Case Number: A04050029 11 Page 1 o f 1 We received an allegation that the subject' had plagiarized text from an NSF award abstract2 posted on the NSF website into one of his NSF proposals.3 Our initial inquiry revealed three additional NSF proposals,4 which the subject submitted as the sole PI, containing 124 lines of apparently copied and unattributed text. The allegation was referred to the subject's university.5 The university's investigation concluded that although the subject had knowingly plagiarized 7 lines of text and recklessly plagiarized 117 lines of text, his actions constituted "low-level plagiarism" and was therefore not research misconduct. We disagreed and continued our investigation with a review of the subject's other NSF proposals. We identified two additional unfunded proposals6 and an NSF award7 into which the subject had copied text without proper attribution. We determined that in total the subject knowingly plagiarized 156 lines of text and recommended in the attached Investigation Report that NSF send the subject a letter of reprimand with a finding ofresearch misconduct; require the subject to certify completion of an ethics course; bar the subject from serving as an NSF peer reviewer for 3 years; and for the next 5 years require the subject to provide certifications and assurances with each proposal or report he submits to NSF stating that the proposal or report does not contain plagiarized, fabricated, or falsified material. NSF accepted our recommendations. This memo along with the report and the DD's letter to the subject constitute the close for this case. Accordingly, this case is closed. 11' NSF OIG Form 2 ( 1 1102) NATIONALSCIENCE FOUNDATION 4201 WILSON BOULEVARD ARLINGTON, VIRGINIA 22230 OFFICE OF THE DEPUTY DIRECTOR CERTIFIED MAIL --RETURN RECEIPT REQUESTED SEP 2 3 2006 Re: Nodce of Misconduct in Science Deternzination documented in the attached lnvestigative Report prepared by NSF's Office of Inspector General ("OIG"), these proposals contained text that was plagiarized. Scientific Misconduct and Proposed Sanctions Under NSF's regulations, "research misconduct" is defined as "fabrication, falsification, or plagiarism in proposing or performing research funded by NSF . . ." 45 CFR 8 689.1(a). A finding of research misconduct requires that: (1) There be a significant departure from accepted practices of the relevant research community; and (2) The research misconduct be committed intentionally, or knowingly, or recklessly; and (3) The allegation be proven by a preponderance of evidence. 45 CFR 5 689.2(c). Page 2 In your proposals, you misappropriated text from several source docGments without providing proper attribution for such material. By submitting proposals to NSF that copy the ideas or.words of another without adequate attribution, as described in the OIG Investigative Report, you misrepresented someone else's work as your own. Your conduct unquestionably constitutes plagiarism. I therefore conclude that your actions meet the definition of "research misconduct" set forth in NSF's regulations. Pursuant to NSF regulations, the Foundation must also determine whether to make a finding of misconduct based on a preponderance of the evidence. 45 CFR 689.2(c). After reviewing the Investigative Report, NSF has determined that, based on a preponderance of the evidence, your misconduct was committed knowingly and constituted a significant departure from accepted practices of the relevant research community. I am, therefore, issuing a finding of research misconduct against you. NSF's regulations establish three categories of actions (Group I, 11, and 111) that can be taken in response to a finding of misconduct. 45 CFR §689.3(a). Group I actions include issuing a letter of reprimand; conditioning awards on prior approval of particular activities from NSF; requiring that an institution or individual obtain special prior approval of particular activities from NSF; and requiring that an institutional representative certify as to the accuracy of reports or certifications of compliance with particular requirements. 45 CFR §689.3(a)(l). Group I1 actions include award suspension or restrictions on designated activities or expenditures; requiring special reviews of requests for funding; and requiring correction to the research record. 45 CFR §689.3(a)(2). Group IrI actions include suspension or termination of awards; prohibitions on participation as NSF reviewers, advisors or consultants; and debarment or suspension from participation in NSF programs. 45 CFR 689.3(a)(3). In determining the severity of the sanction to impose for research misconduct, I have considered the seriousness of the misconduct; our determination that it was knowing; and the fact that your conduct did not have an impact on the published research record. I have also considered other relevant circumstances: 45 CFR fj 689.3 (b). In light of the foregoing, I am bamng you from serving as a reviewer of NSF proposals until October l , 2009. Ln addition, until October 1,2011, for each proposal or report you submit to NSF, I am requiring that you: (1) certify that the proposal or report does not contain plagiarized, falsified or fabricated material; and (2) submit assurances by a responsible official of your employer that the proposal or report does not contain plagiarized material. Lastly, you must complete an ethics training course on research misconduct by April 1, 2007. You may not submit any additional proposals to NSF on which you are a principal investigator or co-principal investigator until you complete this course. You must certify in writing to the OIG that such training has been completed. Page 3 Procedures Governing Appeals Under NSF's regulations, you have 30 days after receipt of this letter to submit an appeal of this decision, in writing, to the Director of the Foundation. 45 CFR §689.10(a). Any appeal should be addressed to the Director at the National Science Foundation, 4201 Wilson Boulevard, Arlington, Virginia 22230. If we do not receive a response to this notice within the 30-day period, this decision will become final. For your information we are attaching a copy of the applicable regulations. If you have any questions about the foregoing, please call , Assistant General Counsel, at (703) 292-8060. Sincerely, Kathie L. Olsen Deputy Director Enclosures - Investigative Report - 45 C.F.R. 689 Summary The Office of Inspector General (OIG) has concluded that the subject' committed knowing plagiarism of 124 lines fiom nineteen sources in four NSF research proposals.2 The Subject's ~ n i v e r s i t ydespite ,~ finding that he had recklessly (1 17 lines) and knowingly (7 lines) plagiarized text as a "violation of the institutional standard of scholarly integrity,"4 did not make a finding of research misconduct under its policies and procedures.5 We disagreed with the University's final conclusion because our assessment of the evidence supports a finding of research misconduct. We continued with our investigation, supplementing the evidentiary material from the University's investigation. We identified another 29 lines of text from two additional sources that the Subject knowingly plagiarized in three additional research proposals,6 one of which NSF awarded, and an additional 3 lines in one of the original four proposals, thereby bringing the total number of lines knowingly plagiarized to 156. Therefore, we recommend that NSF make a finding of research misconduct and that NSF: send a letter of reprimand to the Subject informing him that NSF has made a finding of research misconduct; require the Subject to certify completion of an ethics course covering research misconduct before submitting any proposals to NSF on which he is a principal investigator (PI) or co-principal investigator (Co-PI); require the Subject to certify each time he submits a proposal or report to NSF that the proposal or report does not contain plagiarized, falsified, or fabricated material for 5 years commencing with the date of NSF's finding of research misconduct; require the Subject to submit assurances by a responsible official of his employer each time he submits a proposal or report to NSF that the proposal or report does not contain plagiarized, falsified, or fabricated material for 5 years commencing with the date of NSF's finding of research misconduct; and bar the Subject from serving as a reviewer of NSF proposals for 3 years commencing from the date of NSF's finding of research misconduct. OIG's Inquiry We reviewed an allegation that the Subject had copied text from an NSF CAREER award ,abstract7into one of his proposals (Proposal 2).' We analyzed Proposal 2 and determined that approximately 5 1 lines of text appeared to be copied verbatim andlor with de minimis alteration from other sources. We also determined that most of the remaining text was copied from one of the Subject's previous publications on the same topic.9 Although the Subject was free to use his previously published text without its constituting plagiarism, the large portion of the proposal attributed to this one source suggested a possible attempt to obtain funding for work already completed.'0 Therefore, we analyzed three of the Subject's other NSF proposals (Proposal 1," Proposal 3,12 and Proposal 4,13) to determine the apparent extent of the Subject's alleged conduct. Our analysis revealed copied text in Proposal 1 (40 lines), Proposal 3 (1 1 lines), and Proposal 4 (22 lines), all of which the Subject submitted to NSF as the sole PI. We wrote to the Subject to ask him about the apparently copied text in Proposals 1 through 4.14 We also asked about the substantial use of text from his previous publication in Proposal 2 to determine whether the proposal was an attempt to receive funding for work already completed. The Subject responded by letterI5 in which he grouped the questioned text into three categories: (i) similarities in my proposals to descriptions of methods and principals, [sic] such as definitions, (ii) similarities in my proposals to descriptions of instruments and commercial software packages, and (iii) cases of misunderstandings andlor honest errors in my proposals, which could have been avoided if more care would have been taken.[I6] through Proposal 4 based on the order in which the Subject subrnitte; the proposals to NSF. Tab 32.U. The ~ u b j e c t ' s ~ c i e n article. ce See also Proposal 2 at Tab 29 where the text from the Science article is underlined in green in those sections of the proposal related to the proposed work. We did not question the rn text from the article that comprised the majority of the text in the Background section of Proposal 2. l o In our review we noted that at least one reviewer commented that it was difficult to distinguish between the Subject's earlier work left off and the new work began. See Tab 36, page 1. " Tab 28, Proposal 1). '' Tab 30, (Proposal 3). " ~ a 31,b (Proposal 4). 14 Tab 4, OIG Inquiry Letter to the Subject. 1 5 Tabs 17-27. The Subject's responseeto OIG's Inquiry Letter. l6 Tab 17, page 2. The Subject offered excerpts of somewhat similar text from other source^'^ but did not explain the verbatim and near-verbatim text in the four proposals as compared to the sources we identified in our inquiry. The Subject also pointed out that Source 1'' had a publication date in February 2004 whereas he submitted Proposal 2, which contained the questioned text, "several months" before the publication date.19 We agreed that Proposal 2 predated the publication of Source I, but we could not rule out the possibility that the Subject had access to the unpublished manuscript. In response to our query regarding the possibility that he had already published the work he proposed to do in Proposal 2, the Subject stated that the work involved newly synthesized molecules20and that the reviewers had commented the proposal was "too ambitious" and not "repetitive and narr~w."~'However, we found that at least one reviewer noted: Reference 4 . . . presents preliminary results and is featured in the proposal. This publication [Source U] includes [the Subject's postdoctoral mentor] as a co-author . . . . It is not made clear in the proposal where [the mentor's] work ends and the PI'S begins. In general the proposal suffers from numerous grammatical errors that seemingly could have been avoided with careful proofreading.[221 Although this statement does not appear to raise an allegation directly, it suggested to us, in light of the amount of copied text from Source U, that there was sufficient -substanceto investigate further. Therefore, we concluded that further investigation was warranted and referred this issue to the University. Because the Subject's response did not resolve the allegations of plagiarism and an attempt to acquire funding for work already published, we referred an investigation to the University and deferred our investigation pending the University's final report.23 The University's ~ n v e s t i g a t i o n ~ ~ The University completed an investigation and forwarded its report to us. The ad hoe investigation committee (the ~omrnittee)~'considered the allegations of plagiarism in Proposals 1 through 4 (Sources A thorough T) and seeking funding in Proposal 2 for work I' Tab 18, the Subject's Appendix A. l8 Tab 32.1. l 9 ~ a 18, b page 17. 20 Tab 18, page 18. 21 Tab 18, page 18. 22 Tab 36. 23 Tab 3, OIG Referral of Investigation Letter. he other two Committee member' Electrical Engineering and Computer Science: already published in Source u . The ~ ~Committee reviewed the material we provided in our referral letter, its communications with the Subject, and a sizeable body of materials the University library staff produced at its request. As summarized by the University's Deciding Official, 27 the Committee evaluated the evidence and concluded: That plagiarism was found in 26 of the 37 instances[281alleged by NSF. Of 1 1 instances foundnot to be plagiarism, one finding was unanimous, and 10 were by a 2-1 majority. ... That the allegation that [the Subject] requested funding from NSF for work that he had already completed and published is not supported by the evidence. With regard to the 26 instances of plagiarism . . . : None was "careless" in the sense of being inadvertent. Twenty-five were "reckless." One was "knowing." There was a clear pattern of low-level copying in the "reckless category." But: That the instances of plagiarism were not a breach of community standards because "the threshold at which low-level plagiarism becomes a clear violation of ethical standards is neither well defined nor widely understood within the Physical Science and Engineering research communities."[291 The Committee provided a tally of its members' votes and rationale for each "instance" of text that we questioned.30 The Committee classified twenty-seven "instancesM3under the heading ' "Seriousness, impact"32as "violates institutional standard of scholarly integrity."33 The 26The Committee apparently interpreted our concern about the possibility that the Subject had obtained the Source I text through a journal's peer review process as an allegation of a breach of reviewer confidentiality. With respect to that allegation, jurisdiction would more properly rest with the publisher andlor the University, although it would have relevance to our consideration of a~broaderpattern of misconduct. Therefore, we relv on the committee's investigation report only for its evidentiary value in accessinp the allegation of plagiarism. 27 28 - . . as a separate The Committee investigation treated each discrete section of text that we identified in each woposal allegation rather than looking at all of the copied text in a proposal as evidence supporting an allegation of plagiarism. The numbering scheme, which used letters to designate source documents with numbers to identify sections of text from that source, was a means of correlating multiple sections of text from a single source with the text as it appeared in the proposals. The numbering scheme does not delineate separate allegations. 29 Tab 2, page 3. We note our review of the Committee's tabulated rationale for each questioned passage showed 27 "instances," not 26 "instances," of plagiarism (Tab 8). 30 Tab 8. 3 ' See footnote 29. 32 See Tab 8, page I.. Committee also utilized the library staff to identify other potential sources for the questioned text from which it concluded: . . . the Library search does not support the contention that copying for the purpose of providing context or to describe methodology is widespread in the literature of the field, or that apparent instances of copying are the result of using common technical vocabulary or modes of expression.[341 The Committee relied on the library staff to research plagiarism in general. As a result the Committee appeared to rely heavily on published articles about NSF 's pre-2002 misconduct regulation and the previous Department of Health and Human Services' (DHHS) misconduct regulation.35 The Committee concluded: "clearly enunciated standards for plagiarism are needed within the physical science and engineering cornmunitie~."~~ Despite no finding of misconduct, the Committee recommended an educational requirement for the Subject to complete in the area of research ethics. The Subiect's Response to the University Committee's Investigation Report The Subject accepted the Committee's report, writing: Moreover, I am in agreement with the conclusion of the report in that the evidence does not support NSF's allegations of research misconduct. '' The Committee defined "violates institutional standard of scholarly conduct" as: A fifth category [of seriousness, impact] involving violation of general institutional standards of copying in scholarly research as defined by the University . . ., has been added to the Levels of Seriousness defined by NSF. This category applies when copying does not result in a harm classified in categories 11-14, [Tab 8, page 5.1 Categories 11-14, as identified by the Committee, are defined as seriousness or impact on: 1) "the research record"; 2) "stealing others' results or methods"; 3) "Institution"; and 4) "public welfare." (Tab 8, page 1.) In response to our request for clarification of "violates institutional standard of scholarly conduct", the University explained: The University in its review of allegations of wrongdoing in the conduct or reporting of research has found many grades of seriousness. A finding of no academic misconduct does not imply that a concern does not involve a questionable research practice; only that it is not a "hanging offense," warranting a determination of research misconduct. [Tab 34, page 6.1 Thus, we conclude that the Committee's use of "violates institutional standard of scholarly conduct" indicates a departure from accepted practices of the community. The Committee's method of treating each "instance" of copying as a discrete allegation rather than considering the copied text in a proposal in the aggregate could reasonably lead to its conclusion that an individual "instance" was of minimal seriousness. In contrast, our general treatment of all of the copied text in the aggregate, in a proposal regardless of its proximate locat~onto other copied text, warrants interpretation of "violates institutional standard of scholarly conduct" as a significant departure of accepted practices. This is consistent with NSF's treatment of a proposal as the unit of submission. 34 Tab 7, page 7. 35 Tab 7, pages 8-9. 36 Tab 7, page 9. Although I agree with the Committee that the apparent similarities between my proposals and the documents identified by the NSF [OIG] do not establish a "breach of community standards," I strongly feel that more care should have been taken to avoid these instances. . . . Specifically, I feel the proposed educational component . . . could be [sic] valuable asset to my further career development."37 The Universitv's Action Against the Subiect Although the University did not make a finding of misconduct regarding the Subject's copying of text, it required that the Subject: work under the mentorship of a senior [departmental] faculty member to learn the unwritten best practices in technical writing plagiarism, to complete an assignment to explore best practices at other institutions, and to prepare a written document of those best practices for the benefit of the [department].[381 OIG's Investiaation We determined that the University assembled an adequate evidentiary record for its investigation but that we could not accept the University's conclusions. The Committee found no misconduct despite finding of numerous "instances" of knowing and reckless copying as part of a pattern of copying across the four questioned proposals. The Committee found the copying to be a violation of the "institutional standard of scholarly i n t e g r ~ t y . " ~ ~ ~ ~ ~ We wrote to the University for clarification on the apparent contradictions between its conclusions and the evidentiary support.41The University but its response did not resolve our concerns. Therefore, we were unable to accept the University's conclusions. Where a necessary, we will address particular points made by the Committee below as they pertain to our analysis and conclusions. 37 Tab 15, page 2. The Subject's response to the Committee's report. 38 Tab 2, page 4. 39 See footnote 33. 40 Tab 7, pages 6-9. The Committee requested an impressive search of what it perceived to be the relevant literature on the accepted practices of chemical, materials science, and engineering disciplines. However, we found that most of the sources discussed were either outdated or inapplicable. 4 ' Tab 33. 42 Tab 34. The Subject allegedly copied three pages (143 lines) of material into four NSF proposals from twenty source documents (Sources A through T ) . ~The ~ copying included verbatim copying and substantially similar text. During the course of the investigation, the evidence showed that the Subject legitimately used 19 lines of the questioned text from Source J , leaving ~ ~ 124 lines45of text unexplained. As noted above, the Subject subdivided the sections of questioned text into three subgroups when he responded to our inquiry letter.46 We have used the Subject's own subgroups to address the sections of text in summary, although we have reordered the groups for greater clarity. We have also summarized the contributions of each of the sources to the proposals in Table 1. 41 Tab 32.A-32.T. 44 Tab 25, the Subject's Alternate Source for OIG Source J (Tab 32.5). The Subject provided a proposal on which he, as a postdoctoral fellow, was named as an investigator along with the author of OIG's Source J. Although the Subject provided this document in his response to our inquiry, he did not discuss it in his letter. (Tab 17.) On its face, the document lacks sufficient information to identify the intended funding source although it appears to be a DARPA proposal. 45 This number reflects the total number of lines from 36 of the 37 passages identified in our initial inquiry. 46 Tab 17, page 2. Table 1. Summary of Number of Lines Copied by Proposal and Source Document The same three lines of text from Source A that appear in Proposal 1 also appear in Proposal 2. W e originally questioned 19 lines of text from Source J, for which the Subject later provided evidence that he used the text without citation appropriately. 'One line of text from Source K (specifically K1) appears in Proposal 2, Proposal 6, and Proposal 9. Of the remaining 4 lines from Source K found in Proposal 6 , 2 also appear in Proposal 2. '% The same 3 lines of text copied from Source W appear in both Proposal 2 and Proposal 6. Misunderstandings and/or honest errors (2 Proposals, 51 lines). The Subject grouped Sources A, B, C, L and M under this category. It was among these five sources that we found the more egregious copying as described below. Source was the abstract of an NSF CAREER award to another PI:* from which the Subject copied approximately ten lines of text related to broader impacts into Proposal 2 and three linesinto Proposal 1.49 The Subject explained the copied text: 47 Tab 32.A. 48 See footnote 7 I initially wrote this section in the abstract with the intention to revise it, once the actual activities were formulated. This occurred during an earlier period of the proposal writing process, when I was indeed reading several abstracts of successful NSF proposals (potentially including that abstract listed as [Source] A). During the further development, the proposal underwent several iterations and I finally decided not to include the activities that involve . At that time, I dealt with the abstract and proposal as two different entities, and when the proposal was submitted, a non-updated version of the abstract was included. I clearly should have avoided to include this passage (i) in the ,current form and (ii) without addressing in it in the text to ensure sufficient credit to relevant activities outside of the proposal. However this happened as a consequence of a technical mistake and without intention and is, as such, an honest error.[501 The Subject's assertions regarding Source A do little to dissuade us fiom the conclusion that he knowingly copied these texts verbatim into both Proposal 1 and Proposal 2. Source B involved approximately eleven lines of text,51copied verbatim into Proposal 152 along with the reference^^^ used by the Source B authors.54 We concluded that the references were integral to the copied text. The selection of references in Source B represents the intellectual effort of the original authors to support the content of the text the Subject copied. Thus, we considered the text and references fiom Source B as a whole rather than as discrete allegations.55 The references occurred in the same order in Source B as in Proposal 1 with the only difference being selected omissions from the list in the source document. The Source C material56was ap roximately three lines of text, most of which was contiguous and verbatim in Proposal 1.P7 The authors of the source text cited one reference; 49 Three of the lines in Proposal 2 were the same as the three lines in Proposal 1 . 50 Tab 17, page 3. 51 Tab 32.B, page 1 . 52 Tab 28, project description, page I . 53 For discussion purposes we are distinguishing between references as items in the reference section of the proposal and citations as the notation in the proposal text directing the reader to a specific reference. 54 The Committee report noted: The text is clearly copied. Two members of the Committee judged that plagiarism has not been demonstrated for B3-6 since [the Subject] has a right to cite references as he wishes, and it is not clear that the references were simply copied. The third member disagreed, considering the references to be an integral part of the text. [Tab 32.B, page 1 .I See Tab 8, pages 5-6. We agreed with the third member of the committee, who was most representative of the Subject's own discipline. 5 5 The Subject and the Committee frequently refer to the passages we identified in our inquiry as discrete, stand- alone allegations. 56 Tab 32.C, page 648. 57 Tab 28, project description, page 1 1 . however, the Subject provided a citation to another reference." The Subject explained that the text was taken from one of his 2001 publications (reference 41 in Proposal 1) and that when he took the text from his paper he mistakenly copied the citation number, "" from his paper into his proposal. He went on to explain that reference 25 in his 2001 publication was a citation to Source C. We confirmed that the copied text does appear in the Subject's 2001 publication and that the citation in his 2001 publication refers to a 1998 paper that predates Source C (published in 1999). The 1998 paper and Source C have authors in common. Thus, it appears that the authors of Source C and the 1998 paper were the original authors of the text that the Subject copied into both his 2001 paper and Proposal 1." The Subject included no citations to the 1998 paper or Source C in Proposal 1. Furthermore, the Subject did not cite reference 4 1 in the vicinity of the copied text in Proposal 1.60 The thirteen lines of text copied from both Source L and Source M into Proposal 2 described recent advancements in the particular area of science. With regard to Source L, the Committee found: [Source] L involves a paper . . ., which [the Subject] cites in the appropriate place of the reference list . . .. That citation, however, is unnumbered, evidently the result of an oversight. The intention to reference the science is clear and appropriate, although the copying of the text of the review article . . . is inappropriate.[611 The Subject did make an attempt to cite the source that the is cited in Source L as the original source. However, the Subject's citation appeared in the same place in the copied text as it did in the Source L text. We interpret this to be less an effort to provide the appropriate citation and more indicative of copying the Source L text. We conclude similarly for the Source M text, which was a catalog of applications cited to numerous references in Source M. The Subject copied both the text and the references from Source M . ~ ~ With regard to the text copied from Source A, C, L, and M and part of the copied material from.Source B, the Committee found that the copying violates the "institutional standard of scholarly integrity."63 Therefore with respect to the text in these passages, we concluded that the Subject significantly departed from accepted practices. 58 See footnote 53 for the distinction between citations and references. 59 This would tend to suggest that the Subject's 2001 publication contained a statement copied verbatim from either the 1998 paper or Source C. We note that the work reported in the 2001 publication does not appear to have been conducted with NSF funds. A detailed examination of the Subject's published works did not appear warranted given that a pattern of copying in NSF proposals, as described below. Therefore, we have not conducted a full examination of the Subject's 2001 publication and other publications. 60 The Committee found that the Subject "properly referenced" the source of the text in the proposal; however, that conclusion was unsupported. See Tab 8, page 6 note 6). 61 Tab 8, page 6 note (m). 62 The Committee's only comment was that the Source M text was "generic material to provide context and an introduction" (See Tab 8, page 5 note (d)). Although, the Committee did conclude that the text was copied without proper attribution (See Tab 8, page 4). 63 Tab 8, pages 2-4. See also footnote 33. Descriptions of methods and principles, such as definitions (4 Proposals, 39 lines). The Subject included under this category Sources D, E, F, G, H, I, N, 0 , and Q . The ~ ~Subject challenged Source I as being published after his submission of the proposal and Source Q as "a dubious French hornepage" that he was unable to access.65 While in the inquiry phase, we were unable to rule out the possibility that the Subject had pre-publication access to Source I, the Committee found a source that predated Source I containing the same copied text.66 With regard to Source Q, we have had no difficulty accessing the page repeatedly during our investigation. The Subject provided us with examples of similar text in the literature to demonstrate that the questioned language appeared in other sources.67 In those instances, we determined that his proffered examples were less similar to the source text than the text he copied into his proposals was to the text from the source documents. The Subject's examples negated any claim that the text was technically constrained so as to require the use of those specific words. The Committee's statement regarding the library search supported this conclusion in this regard when it wrote: . . . the Library search does not support the contention that . . apparent instances of copying are the result of using common technical vocabulary or modes of expression.[681 With the exception of Source ~ , the6 Committee ~ also found these instances to be violations of the "institutional standard of scholarly integrity."'' Therefore with respect to the text from Sources D, E, F, G, H, I, N, 0," and Q copied without proper attribution, we concluded that the Subject significantly departed from accepted practices. Descriptions of instruments and commercial software packages (2 Proposals, 21 lines). The Subject included under this subgroup Sources K, P, R, S, and T. These passages involved text describing software and instrumentation. The Subject explained that he used the text from these sources: 64 AS we stated in footnote 44, we have accepted the Subject's earlier proposal to another agency as evidence that he had authorship in the text from Source J. 65 Tab 17, page 4. Tab 16, page 2268. 67 Tab 18-27, the Subject's Appendices A-D.2. Tab 7, page 7. 69 We note that the Committee's one member most aligned with the Subject's own discipline (i.e.,the Committee chairman), found that the Source N text "involves a specific methodological research result that was copied and used without attribution." (See Tab 8, page 6, explanatory note (n).) We agree with the Committee chairman with regard to Source N. 70 Tab 8, pages 2-4. See also footnote 33. 7' The Committee's email to the Subject's faculty mentor (Tab 37) contained the text the Subject copied from Source 0 as it appeared in the mentor's publication rather than as it appeared in Proposal 3. The mentor addressed the technical constraints on the language in his response to the Committee but did so outside of the context of Proposal 3. The Subject did not offer technical constraints of the language as an explanation for his copying and instead pointed us to reference I03 in Proposal 3. Although he did cite reference 103, another of his mentor's publications, he did not distinguish the copied text with quotation marks or indentation. To avoid confusion due to ambiguous descriptions that a complete rewrite of.the features of the commercial products would have caused (I am not an expert in the development of any of these instruments or software packages, nor do I claim to be one), I tried to adhere to the product descriptions supplied by the company - where appropriate.[721 Thus, the Subject admitted to copying text from the suppliers. While it can be beneficial to stay close to the developer's description of a particular product, our concern was and still is that the Subject copied text from the documentation, in several instances from manufacturers' literature, without attribution. For example fiom Source P , the ~ Subject ~ copied verbatim six lines of text74and provided a citation to sources that do not contain the text copied.75 Source R is a manual for a fiom which the Subject copied a partial list of common problems.76 With regard to Sources S and T, which we found in the budget justification of Proposal 4, the Subject wrote these sections "clearly refer to the attached quotations fiom the companies and products."77 The Subject explained Source K in a similar manner, pointing out that he referenced the name of the software package in the first sentence of that discussion "to ensure that the producer of the software receives all the credit he deserves.'778While we do agree that the Subject's text established a context in which he was discussing the particular instrumentation or software, the Subject did not offset the copied text or provide citation to the source of the text or the ideas contained therein. The Subject seems to argue simultaneously that the intellectual content of the text is ubiquitous such that citation is not warranted while at the same time he is free to use another author's text that embodies the ideas much better than he can state them himself. We concluded that the text was not technically constrained. Thus, we concluded that the Subject copied the text without proper attribution from Sources K, P, R, S, and T. The Committee found that the copied material from these sections of text were violations of the "institutional standard of scholarly integrity.7y79Therefore with respect to the text in these passages, we conclude that the Subject significantly departed from accepted practices. Pattern of Copying Although the Subject continued to submit proposals to NSF through the University during the course of the investigation, the Committee's conclusion that the Subject's actions constituted a pattern of copying was limited to Proposals 1 through 4." The Committee did not ..lookbeyond the scope of the four proposals we referred for investigation. As part of our 72 Tab 17, pages 2-3. 73 Tab 32.P, pages 1-2. 1, page 14. 74 ~ a 3 b 75 Tab 35. 76 Tab 32.R, page 2- 17. 77 Tab 17, page 3. 78 Tab 17, page 3. 79 Tab 8, pages 3-4. See also footnote 33 investigation, we have reviewed Award l a ' and the six proposals the Subject submitted since the beginning of our inquiry.82 In our review, we found that the Subject continued to copy verbatim text from sources without distinguishing that text from his own with quotation marks or indentation in three of those proposals, two of which he submitted after November 2004 when we referred the investigation to the University. Notably, we did not find copied text within his funded proposal (Award In the declined proposals, the text copied tended to be instrumentation or software descriptions taken from the manufacturer's literature. In Award 1,84we identified 20 lines of text copied from an instrument manufacturer's product literature (Source v).~' The Subject was the PI on this award with four Co-PIS, all of whom were senior faculty members at the University including his senior faculty mentor. The copied text appeared in a section of the proposal discussing the generalities of an instrument under consideration, rather than in a section describing a particular PIICo-PI'S contribution to the project. The copying was consistent with the Subject's other copying. Although the section of copied text introduced the instrument by name, the following 20 lines of text were verbatim with only minor exceptions and without citation, quotation marks, or other mechanism to signify that the words originated with another author. In Proposal 6,86the Subject copied text not only fiom the software manufacturer's literature (5 lines) but also from a review of that software on the Internet (3 lines, Source w).~' The Subject did provide a reference to the manufacturer's website's homepage at the end of the entire passage of copied text; however, it was clear that the original text on the manufacturer's site was part of a white paper by a named author several levels removed fiom the homepage. In fact, that white paper was Source K, which was also copied into Proposal 2. Furthermore, the largest string of text copied in this passage came from Source W, which was an independent website with reviews of similar software packages. We did not find the Source W text on the manufacturer's website, therefore, the Subject's attempt at citation did not encompass this part of the copied text. The Subject used no means of offsetting the copied text fiom his own. We concluded that the Subject copied this text into Proposal 6, providing only a token reference. -'submited o n i . e . , two months after the Subject responded to our inquiry letter and more than two weeks after we referred the investigation to the University.) 84 Tab 39. 85 Tab 40. 86 ~ a 41.b " Tab 42. We also found one line of text from Source K copied verbatim into Proposal 9, submitted on 24 August 2005. The copied line contains a citation to the manufacturer's website as described for the same text in Proposal 6. Although this involves a single line, it is a line of text that we questioned in Proposal 2 in our inquiry. The Subject continued to use the line verbatim in a proposal he submitted during the course of the investigation. When the additional instances of copying found in Award 1, Proposal 6, and Proposal 9 are aggregated with the copying found in Proposals 1 through 4, the total number of lines copied is 156 as summarized in Table 2. Table 2. Summary of Proposals Containing Plagiarized Material and Associated Source Documents Date Proposal Type Source Total Lines Citations Submitted . Documents of Text copiedg8 Copied Proposal 1 Research A, B7C, D, E 40 14 Proposal 2 Research A, D, E, F, G , H, 54 17 K, L, M YW Proposal 3 Research N, 0 . 11 Proposal 4 Research P7Q, R, S7T 22 Award 1 Instrumentation V 20 Proposal 6 Research K, W 8 Proposal 9 Research K 1 Totals 156 14 OIG Assessment A finding of misconduct requires that (1) there be a significant departure from accepted practices of the relevant research community, (2) the research misconduct be committed intentionally, or knowingly, or recklessly, and (3) the allegation be proven by a preponderance of the evidence.g9 The Act As described above, we identified 124 lines of text copied into Proposals 1 through 4 from nineteen sources. We also identified copied material in Award 1 (20 lines), Proposal 6 (8 lines), and Proposal 9 (1 line) drawn from two additional sources as well as one of the original nineteen we identified. The preponderance of the evidence supports the conclusion that the In this case, each citation was equivalent to one line of text and has been included under "Total Lines of Text Copied." ''45 C.F.R. $ 689.2(c). Subject's copying without appropriate attribution to the sources violated the "institutional standard of scholarly integrityMg0 and as such was a significant departure from accepted practice. Intent The Subject's actions could be considered reckless if we, like the Committee, treated each "instance" of text as a discrete and separate allegation giving rise to 37 allegations.91192In contrast, we considered the totality of the text copied in each proposal and the circumstances surrounding the Subject's preparation of the proposals. The evidence supports a finding that the Subject acted knowingly with regard to copying from the sources we identified.93 The Subject's actions in this case involved finding text in numerous sources and placing that text into his proposals. The act of copying is a knowing act by its very nature. There was no evidence to indicate that any other person was involved in this process beyond the Subject's faculty mentor suggesting some references for con~ideration.~~ In some instances, the Subject edited the copied text creating de minimis differences from the source documents. Such editing .is by its nature a knowing act. Thus, the extent of the copying from numerous sources with selective editing demonstrated a knowing intent. Pressing submission deadlines do not appear to be a motivating factor behind the Subject's actions. Proposals 1 through 3 had the same timestamp indicating that the three were submitted simultaneously.g5 Although the Subject may have been pressured by an impending deadline for Proposal 1y6the Subject submitted Proposal 2 and Proposal 3 as unsolicited proposals (i.e.,not in response to a specific program announcement or soli~itation).~'Therefore, he had no NSF imposed deadlines for completing Proposal 2 and Proposal 3. The Subject also stated in his initial response to us: "The fact that my ideas matured.over several years enabled me to submit three proposals . . . within a relatively short period after my starting date'' at the University. Regardless, it appeared that the Subject failed to prepare each proposal according to the scholarly standards expected by N S F . ~ ~ Tab 8, pages 2-4. See also foomote 33. The Committee concluded that the Subject demonstrated knowing intent with regard to the text copied from Source A and reckless intent with regard to the remaining instances in which it found plagiarism. See Tab 2, page 3. 92 See footnote 28. 93 This excluded the material copied from Source J, as the Subject ultimately produced a proposal to another funding source that showed he shared the text as a coauthor at some point with an author of Source J. 94 Tab 37, page 2. The Subject's mentor wrote to the Committee: I have indeed been mentoring [the Subject], and have recommended that he readand cite important previous papers from our laboratory to help justify and support the feasibility of his own research plans. [emphasis added] 95 The timestamp reflects the time the Authorized Organizational Representative (AOR) certified the proposals, it 96 The Subject submitted Proposal 1 in response to which designated a submission time for unsolicited proposals. The Subject's submission was three days after that deadline. 97 AS designated on the proposal Cover Pages as NSF 04-2 at the top of the page. See Tabs 29 and 30. 98 NSF Grant Proposal Guide 1.B (October 2003). With respect to the manufacturer's product literature copied into Proposals 2,4, 6, and 9, and Award 1, we concluded that the Subject's copying was knowing. He admitted that he used the words from manufacturer's product literature because he did not feel capable of providing a better description.99 The preponderance of the evidence supports our conclusion that the Subject's actions were knowing. Standard o f Proof We conclude that the preponderance of the evidence shows that the Subject knowingly copied 156 lines of text, including 14 references, from twenty-one references into seven NSF proposals, one of which was awarded. Therefore, the Subject's violation of the "institutional standard of scholarly integrity" is research misconduct. OIG's Recommended Disposition When deciding what appropriate action to take upon a finding of misconduct, NSF must consider: (1)How serious the misconduct was; (2) The degree to which the misconduct was knowing, intentional, or reckless; (3) Whether it was an isolated event or part of a pattern; (4)'Whether it had a significant impact on the research record, research subjects, other researchers, institutions or the public welfare; and (5) Other relevant circumstances.['001 Seriousizess The pattern of the Subject's actions elevated the seriousness of his misconduct. The Subject's copying was not an isolated event. Furthermore, the Subject's conduct continues as demonstrated in proposals submitted during the course of this investigation. In submitting these latter proposals, the Subject may have relied on the Committee's assertion of the nebulous concept of "low-level copying" as permissible because of its perceived prevalence. These more recent proposals the Subject submitted do include some citations to the source material; however, the Subject continues to copy text verbatim without offsetting the text to distinguish it fiom his own original text or his own paraphrase of the source. The three lines of Source W text also appeared in Proposal 2, although we had not identified it at the time of our initial inquiry.''' 99 Tab 17, pages 2-3. loo 45 C.F.R. 5 689.3(b). 101 Tab 44, Excerpt fiom Proposal 2 containing Source W text 16 Degree o f Intent The Subject's degree of intent was knowing and consistent with NSF's past findings of research misconduct in plagiarism cases. Pattern In addition to the four NSF proposals we referred to the University for investigation, three of the Subject's other six NSF proposals contained plagiarized material. The number of lines of plagiarized material in those submissions decreased over time (Table 2) suggesting improvement in the Subject's citation practices. However the most recent submission, Proposal 9, contained one line of text that we havequestioned since the beginning of our inquiry. The Subject provided citation for the intellectual content of the text copied, but he did not distinguish the copied text from the his text with quotation marks or indentation, thereby failing to provide appropriate credit to the source material. Both Proposal 1 and Proposal 2 each individually contain an amount of knowing copying for a finding of research misconduct consistent with past NSF findings. While the copying contained in the other five proposals may or may not rise to a finding of research misconduct individually, the five proposals demonstrate a continuing pattern of copying spanning two years during which the Subject has submitted NSF proposals. Thus, we concluded that the Subject has exhibited and continues, although to a lesser degree, to exhibit a pattern of plagiarizing the words of others. Impact on the research record Based on the evidence in this case, the Subject's actions in his NSF proposals did not appear to impact the research record. Other Relevant Circumstances The Subject acknowledged that he had some difficulty with English because it is not his native language. Io2,lo3 However under the circumstances in this case, we concluded that the Subject had sufficient skill with the English language as demonstrated by his publication record in ~ n ~ l i s h .Many " ~ of those publications appeared to be collaborative efforts during his Tab 7, page 11. lo' The Committee used the language difference as a factor mitigating against the Subject's knowing intent. In fact, the Committee went further using it to mitigate against a finding of research misconduct. While such an approach is at the University's discretion, we found n o basis for such an approach in our investigation or past NSF findings. Rather the Subject's facility with the English language is a more appropriate consideration when determining what action to take after reaching a finding of research misconduct. '04 Tab 38, the Subject's publication record as reported online in the Science Citation Index. The Subject has published in English), 4 of which he published between 1998 and 2003. Of those , he was the first author on ( in English). While there is no definitive rule or practice in the Subject's discipline as to which postdoctoral training and research positions at U.S. institution^,'^^ most likely involving some editorial collaboration to smooth out any of his difficulties with the language. The Subject acknowledged to us his perceptions of his own limitations specifically in describing instrumentation and software.'06 But he appeared to have sought out little or no editorial assistance with the preparation of his proposals despite having a senior faculty mentor advising him on substantive content. Recommendation Based on the evidence, OIG recommends that NSF: send a letter of reprimand to the Subject informing him that NSF has made a finding of research misconduct; require the Subject to certify completion of an ethics course covering research misconduct before submitting any proposals to NSF on which he is a principal investigator (PI) or co-principal investigator (Co-PI); require the Subject to certify each time he submits a proposal or report to NSF that the proposal or report does not contain plagiarized, falsified, or fabricated material for 5 years commencing with the date of NSF's finding of research misconduct; require the Subject to submit assurances by a responsible official of his employer each time he submits a proposal or report to NSF that the proposal or report does not contain plagiarized, falsified, or fabricated material for 5 years commencing with the date of NSF's finding of research misconduct; and bar the Subject from serving as a reviewer of NSF proposals for 3 years commencing from the date of NSF's finding of research misconduct. The Subiect's Response to Draft Investigation ~ e p o r t ' ~ ' The Subject provided comments to our Draft Investigation Report (the Draft). Those comments focused primarily on our decision to recommend a finding of research misconduct in lieu of deferring to the University's decision and his perception that we have not made our recommendations objectively. With specific regard to the text of the Draft, the Subject objected to the content of several quoted passages and to the objectivity of phrases characterizing his author's name appears first, one of the most common practices in that discipline is to list the author who drafted the . - '07 Tab 45. actions. The Draft accurately reflected the content of the quoted sources and our assessment of the facts. We have nevertheless modified the text to accommodate the Subject's expressed concerns. The quotes have been expanded to provide the context that the Subject stated was necessary to evaluate their content. In some quotes, we have expanded the quotations beyond the text that the Subject requested to insure that the reader has sufficient information to discern the context of the statements. We have also removed the phrases that he specifically found objectionable and provided additional explanatory information where necessary. We have also provided additional line counts and other metrics in response to his comments. In light of these changes, our analysis and recommendations based on the facts in this case remain the same. The Subject provided an email dated 23 March 2006 from the instrument manufacturer fiom which he copied the Source V text into Award 1. The manufacturer's representative stated: "The quote and description of the [instrument] that I provided you was intended for use in the NSF proposal."'08 It is common for NSF PIS to include materials from product vendors as supplementary materials in a proposal. In this case, the Subject copied text fiom the product literature directly into Award 1 rather than appending the materials as supplemental information. The Subject referred to the instrument by name and manufacturer in the text of Award 1 but used the manufacturer's words (20 lines) without quotation marks or other distinguishing treatment. It was clear that the copied text was a description of the instrument, but it was not clear that the description was authored by someone other than the Subject. We analyzed Award 1 with respect to the Subject's pattern of copying and included those lines in the final total of lines copied. Our line totals continue to include those lines in light of the Subject's previous statement regarding his abilities to describe the instrumentation, as quoted above.Iog Whether the Source V text is included in the totals has no bearing on our recommendations because it is a relatively small portion of the text copied without attribution. In addition to his concerns with the text of the Draft, the Subject wrote: Your report implies that a middle ground, poor practice, but not misconduct, is a theoretically impossible state, seriously distorting the real world.[' lo] We agree that a "middle ground" exists. However, the Subject's actions in this case fall squarely ' within the applicable definition of research misconduct, specifically plagiarism. ' I The extent of his plagiarism well exceeds the amount for which NSF has previously made findings of research misconduct. The Subject's rationale for classifying his actions within the "middle ground," which we interpret to mean a questionable research practice (QW),appears to rely extensively on the University's determination that his actions were not "a significant departure from accepted practices of the relevant research c~mmunity.""~We agree that QRPs are part of the "real world" of scientific research, but the rationale upon which the University and the Subject '08 Tab 45. 109 See page 2. l lo Tab 45, page 3. "' 45 C.F.R. 689.1. 'I2 45 C.F.R. 689.2(~)(1). have characterized the copying in this case as a QRP appears premised on the pervasiveness of copying, not the acceptability of it, in the relevant research community. Our recommendation for a finding of research misconduct in this case is consistent with the policy statements of those communities in which the Subject has published the majority of his work. We have reviewed the policy statements from professional societies and the journals in which the Subject has published and found that regardless of the University's perception of the prevalence of. copying . within the community, the community appears to reject copying According to the Guidelines: " We acknowledge that assessing what is "common knowledge" is an exercise in discretion, but it would be difficult to rationalize that full sentences or paragraphs such as those copied from Sources A or B would qualify as "common knowledge" when the subject matter is not technically constrained to those particular words or a unique set of circumstances. As stated above,'16 the University elected to analyze the questioned text as individual discrete allegations corresponding to our numbering scheme which correlated proposal text with the source text. NSF reviews a proposal as a single submission unit when determining whether to fund a project. Therefore NSF assesses the whole proposal as a unit in relation to the source document(s) in making a determination of research misconduct rather than assessing the individual line, word, or phrase corresponding to our correlation scheme. In this manner, NSF assesses the text in the aggregate much like a chemist analyzes the molecule as an aggregate of bonded atoms. The University's approach reduced the text to its elemental form by focusing on its constituent parts (the atoms) rather than their connection and relation to the whole (the molecule). The University's assessment may be consistent with its own policies but it is inconsistent with past NSF cases in this area.'17 Some of the copying we identified could be considered an example of a QRP except that is was within a much larger body of unattributed copied text. Our analysis of the copied text in the aggregate is consistent with our recommendation for a finding of research misconduct, although some of the materials copied, such as those from Source B, are likely sufficient on their own for a misconduct finding. The Subject proposed three "voluntary sanctions" in his response, in lieu of our recommendations to NSF. Two are remedial education courses which are consistent with our this statement. See footnote 28. "'See discussion on pages 4-5, particularly footnote 33. recommendations. The third is a certification requirement requiring him to certify to the University rather than to NSF that his proposals and reports to NSF "fully comport with the research conduct standards of the University and of the research funding agency.""8 We agree that certification is a necessary and useful means of helping the Subject "avoid any kind of bad practice in the future." ' I 9 Our recommendations for certifications and assurances to NSF in conjunction with a finding of research misconduct are appropriate for protecting the interests of NSF and are consistent with past cases. 'I8 Tab 45. page 6. 119 Tab 45. page 6.
Published by the National Science Foundation, Office of Inspector General on 2006-10-30.
Below is a raw (and likely hideous) rendition of the original report. (PDF)