NATIONALSCIENCE FOUNDATION OFFICE OF lNSPECTOR GENERAL OFFICE OF INVESTIGATIONS CLOSEOUT MEMORANDUM 11 Page 1of 1 We learned fiom a university1 that it had completed an inquiry into an allegation of research misconduct by an ~ ~ F - f u n d postdoctoral ed~ fellow (the subje~t)~ and that it was proceeding with a detailed investigation. The allegation involved the fabrication andlor falsification of data in a published journal article.4 After we concurred and referred the matter to the university for a detailed investigation, the university reversed its opinion, doing so without sufficient reason to dissuade us fiom the need for a complete investigation. We conducted our investig&on and determined that the evidence showed that the subject had falsified the published data. We prepared the attached investigation report recommending that NSF: make a finding of research misconduct; send a letter of reprimand to the subject; debar the subject for a period of 2 years; require the subject to submit certifications and assurances by a responsible official of his employer that his submissions to NSF do not contain research misconduct for 3 . ' years after the debarment period; require the subject to certify completion of a course covering research misconduct before applying for NSF funding again; require the subject to certify retraction of the article containing the falsified data; and bar the subject fiom sewing as a reviewer of NSF proposals for a period of 3 years. The NSF Deputy Director made a finding of research misconduct in the attached letter and adopted our recommendations. Accordingly this case is closed. 1INSF OIG Form 2 (11/02) NATIONALSCIENCE FOUNDATION 4201 WILSON BOULEVARD ARLINGTON, VIRGINIA 22230 MAR 9 2007 OFFICE OF THE DEPUTY DIRECTOR CERTIFIED MAIL -RETURN RECEIPT REQUESTED Dr. Juan Carlos Jor~e-Rivera Re: Debarment Dear Dr. Jorge-Rivera: On January 11, 2007, the National Science Foundation ("NSF") sent you a Notice of Proposed Debarment in which NSF proposed to debar you from directly or indirectly obtaining the benefits of Federal grants for a period of two years. The Notice sets forth in detail the circumstances giving - - rise to NSF's decision to propose your debarment. Specifically, NSF indicated in the Notice that the proposed debarment is based upon your falsification of results from experiments conducted on the impact o-receptors, and the publication of these results i n a scientific journal. In that Notice, NSF provided you with thirty days to respond to the proposed debarment. Over thirty days have elapsed and NSF has not received a response. AccordingIy, you are debarred until January 11, 2009. Debarment precludes you from receiving Federal financial and non-financial assistance and benefits under non-procurement Federal programs and activities unless an agency head or authorized designee makes a determination to grant an exception in accordance with 45 CFR Section 620.215. Non-procurement transactions include grants, cooperative agreements, scholarships, fellowships, conhacts of assistance, loans, loan guarantees, subsidies, insurance, payments for specified use, and donation agreements. In addition, you are prohibited from receiving Federal contracts or approved subcontracts under the Federal Acquisition Regulations ("FAR") at 48 CFR Subpart 9.4for the period of this debarment. 45 CFR Section 620!110(c). During the debarment period, you may not have supervisory responsibility, primary management, substantive control over, or critical influence on, a grant, contract, or cooperative agreement with any agency of the Executive Branch of the Federal Government. Lastly, note that, in the Notice of Proposed Debarment, NSF also took the following actions against you, which continue to remain in effect: By June 30,2007, you must retract the publication containing the falsified data, and certify to the OIG that you have done so. For three years from the end of your debarment period, you are required to certify that proposals or reports you submit to NSF do not contain plagiarized, falsified, or fabricated material. For three years from the end of your debarment period, you are required to submit assurances.by a responsible official of your employer that any proposals or reports you submit to NSF donot contain plagiarized, falsified, or fabricated material. You are prohibited from serving as an NSF reviewer, advisor, or consultant through January 1,2010. You are required to complete an ethics training course on research misconduct by December 3 1,2007. You must certify in writing to the OIG that such training has been completed. If you have any questions regarding the foregoing, please contact , Assistant General Counsel, National Science Foundation, Office of the General Counsel, 4201 Wilson Boulevard, Room 1265, Arlington, Virginia, 22230. Sincerely, Kathie L. Olsen Deputy Director NATIONALSCIENCE FO1I.NDATION , - --- -- -- - . - -. . .. 4201 . . . - . - -. . . . . .. ARLINGTON, VIRGINIA 22230 JAN 1 1 2007 OFFICE OF THE DEPUTY DIRECTOR CERTIFIED MAIL --RETURN RECEIPT REQUESTED Dr. Juan Carlos Jorge-Rivera Re: Notice of Proposed Debarmerrt and Notice of Misconduct in Science Determination Dear Dr. Jorge-Rivera: As documented in the attached investigative report, you knowingly falsified the results of experiments conducted on the impact o v D n d published the falsified results in this article. In light of your misconduct, this letter serves as formal notice that the National Science Foundation ("NSF") is proposing to debar you from directly or indirectly obtaining the benefits of Federal grants for a period of two years. During your period of debarment, you will be precluded from receiving Federal financial and non-financial assistance and benefits under non- procurement Federal programs and activities. See 45 CFR Pad 620, Subpaits A, B and I. In addition, you will be prohibited from receiving any Federal contracts or approved subcontracts under the Federal Acquisition Regulations ("FAR"). See 45 CFR 620.125. Lastly, during your debarment period. you will be barred from having supervisory responsibility, primary management, substantive control over, or critical influence on, a grant, contract, or cooperative agreement with any agency of the Executive Branch of the Federal Government. See 45 CFR 620.1 15. . . ... . . . . . .-- ....-. .- . . . . .. . . . . . .- - -- .-- .. ,., , Page 2 In addition to proposing your debarment, I am requiring that, by June 30,2007, you retract the article containing the falsified data and certify in writing to NSF's Office of Inspector General ("OIG") that you have done so. I also am prohibiting you from serving as an NSF reviewer, advisor, or consultant until January 1,2010. Furthermore, for three years after the period of debarment expires, I am requiring you to certify that any proposals or reports that you submit to NSF do not contain plagiarized, falsified, or fabricated material. For this same period of time, you must submit assurances by a responsible official of your employer that any such proposals or reports do not contain plagiarized, falsified, or fabricated material. Lastly, by December 3 1,2007, you must complete an ethics training course on research misconduct, and certify in writing to the OIG that you have done so. ScientiJic Misconduct and Sanctions other thanDebarment Under NSF's regulations in effect at the time of your misconduct, "research misconduct" was defined as "fabrication, falsification, plagiarism or other serious deviation from accepted practices in proposing, carrying out, or reporting results from activities funded by NSF." 45 CFR 5 689.1(a). Falsification is defined as "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." 45 CFR tj 689.1(a)(2). 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 al.legationbe proven by a preponderance of evidence. 45 CFR 5 689.2(c). As the OIGYsreport demonstrates, you manually changed the gain setting on the- d e v i c e so that the Variant 1 cells appeared to exhibit results in support of the hypothesized change in obsented current. You engaged in similar misconduct with respect to the Native 2 cells. Thus, your conduct unquestionably constitutes falsification. I therefore conclude that your actions meet the applicable definition of "research misconduct" set forth in NSF's regulations. Pursuant to NSF's regulations, the Foundation must also determine whether to make a finding of misconduct based on a preponderance of the evidence. 45 CFR tj 689.2(c). After reviewing the Investigative Report, NSF has determined that, based on a preponderance of the evidence, your falsification 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. . . .. . ..,.. -. ... . . .... . -. .- .. . . -.... .- -, , . .. . - ..- . . . . . ... - - --.. - . . , - .... , . . . Page 3 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 a i institutional ~ representative certify as to the accuracy of reports or certificationsof 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 I11 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;the impact your misconduct had on the research record; and the determination that it was an isolated incident. I have also considered other relevant circumstances. 45 CFR 689.3(b). I, therefore, take the following actions: By June 30,2007, you must retract the publication containing the falsified data, and certify to the OIG that you have done so. For three years from the end of your debarment period, you are required to certify that proposals or reports you submit to NSF do not contain plagiarized, falsified, or fabricated material. For three years from the end of your debarment period, you are required to submit assurances by a responsible official of your.employer that any proposals or reports,you submit to NSF do not contain plagiarized, falsified, or fabricated material. From the date of this letter through January I , 2010, you are prohibited from serving as an NSF reviewer, advisor, or consultant. You are required to complete an ethics training course on research misconduct by December 31,2007. You must certify in writing to the OIG that such training has been completed. . .......... ....... ... -- _ .......... ...... _. . .. - . .. - ....... Page 4 Regulatory arbfor Debarment Pursuant to 45 CFR 620.800, debarment may be imposed for: @) Violation of the terms of a public agreement or transaction so serous as to affect the integrity of an agency program, such as - (1) A willful failure to perform in accordance with the terms of one or more public agreements or transactions; or (3) A willful violation of a statutory or regulatory provision or requirement applicable to a public agreement or transaction In any debarment action, the government must establish the cause for debarment by a preponderance of the evidence. 45 CFR 620.850. In this case, you knowingly falsified data in connection with research experiments and published the falsified data in a scientific journal. Thus, your actions support a cause for debarment under 45 CFR 620.800(b). Length of Debarment Debarment must be for a period commensurate with the seriousness of the causes upon which an individual's debarment is based. 45 CFR 620.865. Generally, a period of debarment should not exceed three years but, where circumstances warrant, a longer period may be imposed. 45 CFR 620.865. Having considered the seriousness of your actions, as well as the relevant aggravating and mitigating factors set forth in 5 CFR 620.860, we are proposing debarment for a period of two years. Procedures Governing Proposed Debarment The provisions of 45 CFR Sections 620.800 through 620.855 govern debarment procedures and decision-making. Under our regulations, you have 30 days after receipt of this notice to submit, in person or in writing, or through a representative, information and argument in opposition to this debarment. 45 CFR 620.860. Comments submitted within the 30-day period will receive full consideration and may lead to a revision of the recommended disposition. IfNSF does not receive a response to this notice within the 30-day period, this debamlent will become final. -.- . .-. . - .. . - ..-.,. , .... . ..., .-. .- - .. . .. . -.... - . ...,,.. .. ., . . . - . . . . .. ... --. .- - -..-. .. ... . , .. - .- ..-. , .. ., .- -.. . . Page 5 Any response should be addressed to Lawrence Rudolph, General Counsel, National Science Foundation, Office of the General Counsel, 4201 Wilson Boulevard, Room 1265, Arlington, Virginia 22230. For your information, we are attaching a copy of the Foundation's research misconduct regulations, its regulations on non-procurement debarment, and FAR Subpart 9.4. Sincerely, Kathie Olsen Deputy Director - Enclosures: hvestigative Report Nonprocurement Debarment Regulations 45 CFR Part 689 FAR Regulations cc: CONFIDENTIAL National Science Foundation 1 I Office of Inspector General I I I m I II a Confidential 1 Investigation Report I Case Number A03070041 1 Volume 1 of 2 a Report and Tabs 1-38 a 26 July 2006 CONFIDENTIAL a NSF OIG FORM228 (1103) Summary The Office of Inspector General (OIG) has concluded that the subject1 committed research misconduct2by knowingly falsifying data for a series of experiments published in a journal article (the As part of its procedures,4 the university5 conducted an inquiry6 and notified OIG and also the Department of Health and Human Services (DHHS), Office of Research Integrity ( 0 ~ 1 of) ~its intention to proceed with a detailed investigation.' However, the University, after accepting our investigation referral, re-opened its inquiry and reversed its decision to proceed with an i n ~ e s t i ~ a t i o nBased . ~ on the evidence from the inquiry, we did not agree with the University's decision to forego a detailed investigation. lo We conducted our own investigation1' , I 2 h m which we concluded that the evidence supports a finding of research misconduct. We recommend that NSF: send a letter of reprimand to the Subject informing him that NSF has made a finding of research misconduct; debar the Subject from receiving federal funds for a period of 2 years commencing on the date of NSFYsfinding of research misconduct; require the Subject to certify to NSF that the publication containing the falsified data has been retracted; n ' Dr. Juan Carlos Jor e-Rivera a/k/a Dr. Juan Carlos Jor e), presently Assistant Professor- o NSF-funded Minori Postdoctoral Fellow At the time of the alleged misconduct, the Subiect was an in the NIH-funded laboratory o The alleged conduct in this case occurred before April 17,2002. Therefore, we apply the definition of misconduct from the pre-April 17,2002, regulation and use the procedures in the current, post-April 17,2002, regulation. Misconduct under the pre-April 17,2002, regulation means: (1) Fabrication, falsification, plagiarism, or other serious deviation from accepted practices in proposing, canying 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. This is the institution where the alleged m~sconducttook place. University's First Inquiry Report, Tabs 3-16. 7 See footnote 1 for ORI's jurisdictional basis. 8 University's Letter to OIG and ORI, Tab 18. University's Second Inquiry Report, Tabs 21-23. l o OR1 concurred in our assessment of the evidence and the need to proceed with a detailed investigation. I I OR1 provided assistance to OIG with regard to the experimental technique involved in this case because OR1 staff had particular expenence with the fundamental technique. ORI's role was limited to an advisory capacity due to differences between OIG's and ORI's authorities. l 2 An NSF expert with relevant scientific experience read and commented on the sc~entificaspects of the Draft Investigation Report (Draft). ORI's experts also read and commented on the Draft, concumng with our assessment of the scientific-technical aspects of the case. 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), co-principal investigator (Co-PI), or otherwise a participant; require the Subject to certify each time he submits a proposal or report to NSF that the proposal or report does not contain fabricated or falsified material for 3 years after the debarment period; 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 fabricated or falsified material for 3 years after the debarment period; and bar the Subject from serving as a reviewer of NSF proposals for 3 years commencing on the date of NSF's finding of research misconduct. The University's First Inquirv ~ e p o r t ~ The University conducted an inquiry into the Subject's actions in response to an' allegation of data fabrication and falsificationI4in accordance with its policies and procedures.'5 The University's inquiry committee (the Panel) determined that a detailed investigation was warranted. The data in question'6 involved a series of experiments on cells isolated from tissue from different regions of rat brains (Native cells) and cells modified through genetic manipulation (Variant cells) to mimic the cells taken from the different rat brain regions. The data for one type of Variant cells appeared to have been manipulated to achieve a desired result. At the time of the data collection, NSF provided funding for the Subject's postdoctoral fellowship at the university.I7 Consistent with the NSF's Research Misconduct regulation,'* the University notified us of its determination to proceed with a detailed investigation and forwarded a copy of the inquiry report for our review. l 9 The Panel concluded: 1. The charges raised are serious and not frivolous. 2. It appears unlikely, and perhaps physically impossible, hta-t could have been conducted as quickly as indicated on some of the computer l3 See Tabs 3-16 (First Inquiry Report). 14 Tab 4, Initial Allegation Letter. lau L. nsisted of the following experiments:- ee Tab 7, "Appendix 3." I*45 C.F.R. 5 689.4@)(2). 19 Tab 1. ORI was also notified because the allegation involved work published with an acknowledgement to the NIH award funding the laboratory. changescod-uli occur without a change in the experimental conditions. 3. It is hi hl unlike1 that the changes in current reported to be secondary to p r e s e n t s physical phenomena. The Inquiry Panel could not rule out the possibility that changes were made in amplifier gain but not recorded in the lab notebook. However, the Panel felt this was unlikely. 4. The allegations warrant further investigation.[201 OIG's Assessment of the First Inquiry Report We reviewed the substance of the allegation and the inquiry report. We concurred that a detailed investigation was warranted to determine the facts surrounding the questioned data in the Paper. Consistent with NSF's Research Misconduct regulation,21 we referred the matter to the University for a detailed investigation, deferring our investigation pending completion of the ~niversity's.~~ The University's Second Inquirv ~ e p o r t ~ Following acceptance of our referral, the University engaged in a series of communications with the Subject's attorney.24 In response to objections the Subject's attorney raised, the University re-opened its inquiry to receive additional written comments from the Subject and to allow him to make a visit to the campus to address the Panel in person.25 We notified the University that our referral of investigation would continue until we had received its report, at which time we would review the matter in total and determine how to proceed.26 At the end of the reopened inquiry, the University notified us that it had concluded: "this matter does not warrant Detailed In its second inquiry report the Panel noted the following as reasons to reverse its previous determination: 1. Some of the experiments could not have been done correctly in the time interval noted in the computer logs. However, there are reasonable explanations for how this error could have occurred. 2. The changes in-current can be explained by variables in the experimental conditions other than deliberate manipulations in gain. 20 Tab 3, page 4. " 45 C.F.R. 4 689.5(d)(2). 22 Tab 17. 23 See Tabs 21 -23 (Second Inquiry Report). 26 Tab 19. 27 Tab 21, page 1. 3. The work flow in the laboratory in regards to data analysis and presentation was such that it would be very difficult for [the Subject] to single-handedly manipulate data used in the paper. 4. Replicating the exact recording conditions in which [the Subject] did his studies would be very difficult and would not likely resolve the issue of scientific fraud. 5. The allegations. . . were well intentioned and were serious enough to warrant this Inquiry. 6. Based on the review, it does appear that some of the data used in the [Plaper . . .were flawed since [the Sub'ect] noted that it would have been impossible to make the-changes as quickly as noted on the time stamps. It will be the responsibility of the authors to review all the data used in the paper to determine whether the results remain valid.[2s1 OIG's Assessment of the Second Inquirv Report We reviewed the University's second inquiry report and requested clarification from the University on several points.29 The University's response30provided no additional clarity to the rationale behind its decision to forego a detailed investigation. Thus, the second inquiry report did not dispel the need for a detailed investigation. While it is true that explanations other than data fabrication and falsification could account for the questioned data, the allegations remained unresolved by the evidence the University gathered during the inquiry, thereby necessitating an investigation. The purpose of an investigation is to gather information and evaluate that information in total to determine which of the possible explanations, including data fabrication or falsification, is supported by a preponderance of the evidence. The second inquiry report erroneously concluded that the existence of alternative explanations rendered unnecessary an assessment of the evidence related to the allegations. We determined that neither inquiry report provided a sufficient basis for dismissing the allegation. Because the University had not completed the referred investigation within the 180-day referral period and because it had expressed its intention not to proceed,31we notified both the University and the Subject that we would continue with our investigation.32 Tab 22, page 3. ''Tab 26. 30 Tabs 27 and 28. Tab 29. 32 Originally, this case was a joint effort between our office and ORI. ORI's Division of Investigative Oversight took the lead. When the University decided not to proceed with its investigation, ORI concurred with our assessment that issues remained warranting a detailed investigation. However, OR1 lacked the authority to investigate. In order to protect NSF's interests and the general federal interest, we assumed the lead. Because the Subject asserted that the experiments involved were particularly complex (see Tab 30, pages 1-2), and ORI staff had particular experience with the experimental techniques involved, OR1 staff participated in a limited advisory role with respect to the experimental techniques and DHHS's interests. OIG's Investigation Consistent with NSF's Research Misconduct regulation,33we proceeded with an investigation into the allegation that the Subject fabricated andlor falsified data that he published with two coauthors (CoAuthor 1" and CoAuthor 235). During the course of the investigation, we reviewed the evidence collected by the Panel, interviewed several individuals including the Subject, collected supplementary documentation, and visited the laboratory where the Subject collected the data. Basics of the Experiment are part of the receptor. When the channel is open, ions flow between the interior and exterior of the cell. The ion flow is observable as an electrical current that can be measured in the laboratory using electrophysiological methods. Such methods require attaching microelectrodes to the neurons whether the cell is an isolated individual or an integral part of a tissue sample. In this particular case, the Subject and his coauthors published electrical current data collected from tissue samples as well as isolated individual cells c o n t a i n i n g 3 ' The Subject initially collected data on tissue from two different regions of rat brains. The tissue slices from one region Tissue 1)39had a higher proportion o l e c e p t o r s with a particular to the tissue slices from Subject. 35 36Receptors are the structural part of a cell located in the cell membrane, often bridging between the interior and exterior of the cell. They are generally made up of protein subunits. Molecules referred to as ligands, such as drugs and neurotransmitters, can bind to receptors to tngger the receptor and cause a response (i.e.,change in the receptor). A receptor and a ligand are analogous to a lock and key, respectively. Other molecules, generally referred to as modulators, can facilitate or hamper the ligand's ability to bind to the receptor and produce the response. In the case at hand, the response of interest is the ability of the ligand to cause the receptor to open a channel in the cell membrane that allows charged particles (ions) to flow between the interior and exterior of the postsynaptic currents (sIPSCs). 39 These were tissue slices from t h e o f ars-t theother region-th issue 2)," which had a higher proportion o e c e p t o r s with a different subunit (Subunit 2).42 The Subiect observed different effects on the ain~litudes receptors in each tissue type were the underlying explanation for the different responses to the application of the modulator. 46 We refer to these isolated cells as the Native 1 and Native 2 cells, respectively, because they are cells containing the receptors as they are found in nature (i.e., in their native To further test the h othesis, the Subject and Coauthor 2 decided to use isolated cells that did not contai- in their native form.48 Through a process known as transfecti~n;~CoAuthor 1 genetically modified these cells to produce cells containing- cells. The DNA coded for each of collected data on isolated Native and a microelectrode to individual cells current^.'^ ~ h e t e c h n i ~ allowed ue the Subject to rapidly appl to a cell in either the presence or absence of a modulator in order -- - - - - to observe the electrical current and record it digitally by a computer. A s i n g l e x p e r i m e n t consisted of several successive data acquisitions (traces) on a single cell. Each trace spanned a two second time frame. Near the beginning of the trace a computer-controlled actuator toggled a pipette that was bathing the cell with one solution rapidly (2-3 milliseconds) to a second position and back to the first thereby briefly bathing the cell in a second solution. Exposure of the cell to the second solution for that short duration produced a change in electrical current that appeared in the trace as an abrupt change in current in the negative (downward) direction, as shown in Figure 1. Because the cell's exposure to the solution is brief, the observed electrical current returns (decays) back to the starting (holding) current over the two-second acquisition period. Figure 1. Basic Features of an Experimental race." The black portion represents the measured current collected over time. The blue line represents the baseline of zero current. We have added the red elements for illustrative purposes. 50 The Subject and CoAuthor 2 chose to use t h e m e t h o d . Other methods of attaching the cell to the electrode are used in this field of research, in fact when CoAuthor 2's lab attempted to re eat the results reported in the Pa er, the electrode was attached to the cells in a different manner. Regardless of t h e m e t h o d used,i s as much an art as science with respect to the skill involved m creating- a seal between the cell and the electrode that will remain intact during the course of the experiment. 5' Adapted from one of the Subject's experimental t r a c e s , w h i c h is not at issue in this case. known as curve-fitting provided the means of determining the TTP. Curve-fitting results in the calculation of a mathematical equation(s) or curve(s) thatbest represents the dat;collected. Not all of the data points collected will lie on the curve because of background noise that is inherent in the instrumentation andlor other conditions of the experiment will offset them from the curve, giving the plotted data its irregular appearance suggestive of static interference (Figure 2). Figure 2. Expanded view of the peak region of the trace in Figure 1.53 We have added the orange elements to illustrate the time-to-peak (TTP), which is the time from the beginning of the rise to the tip of the peak. The noise is the observed variation in the data not attributed to the signal (current) of interest. Portions of observed current are shown as line segments in red along with the associated noise level bracketed with the green line segments. It is customary to evaluate signals in terms of the signal-to-noise ratio with respect to a specific point in time. The noise level remains relatively consistent across the entire trace while the signal is strongest at the tip of the peak and approaches the holding current level by the end of the trace. Because the end of the trace provides (not shown) the lowest signal-to-noise ratio, it is a suitable and convenient measure for the changes in noise between traces in an experiment on a single cell. " Tab 33, page 6; Tab 35, page 3; and Tab 37, page 13. '3Adapted from one of the subject's experimental t r a c e s j w h i c h is not at issue in this case. A single completemexperiment involved only one cells4 and consisted of three phases55(Figure 3). In the first phase, the Subject usually acquired two traces fiom the cell under control conditions, i.e., without a modulator present (the Control Phase). The Subject then acquired a series of traces of that same cell exposed to a modulator (the Drug Phase). Finally, he acquired traces under wash conditions (the Wash Phase), wherein that cell was again exposed to the initial control conditions. A cell could only be used for a ~ i n ~ l e x ~ e r i m eThe nt.~~ mechanism for changing the conditions during a single experiment is known as solution- switching. Concerns about the Subject's data related to solution-switching and TTP are discussed below under their respective headings. - - - CT, 0 .-C .C L L w 0 0 * *'6 +d '6 Control Phase Drug Phase Wash Phase r r g- r r r -0g r r C c *0 - -V) Figure 3. Schematic Representation of a s i n g l e x p e r i m e n t (not to scale with respect to time). The traces were collected in 2 second periods while solution-switching times varied, based on timestamps, from 4 to more than 30 seconds. 54It appears that on at least two occasions the Subject reported data in the Paper for successive experiments on the same cell, when a cell could only have been used for one experiment casting further doubt as the actual exposure of the cells to the experimental conditions. According to the Subject, a cell could not be used more than once;- and it took at least ten to fifteen minutes t o ( ~ a 37,bpage 9.) was only 250, also an impossibly short time " Tab 37, page 9. CoAuthor 2 provided us with copies of the data traces and data logs for the Native 1, Native 2, Variant 1, and Variant 2 experiments along with a spreadsheet identifying the specific traces reported in the We have confirmed that the data and subsequent calculations correspond with the data reported in the Three figures in the Paper included data from the questioned experiments and are discussed below (Figures 8,9, and 10). 0 The Subject was particularly interested in the behavioral aspects of the research, which according to the Subject reflected his training and interests in psychology and biology.61 The second The allegation involves primarily the data for two particular sets of experiments: one set supporting the structure-function relationship; and the other supporting the behavioral conclusions of the Paper. The initial allegation had three basic facets: 1) apparent instrumental gain changes to produce the hypothesized results; 2) questionable solution-switching times; and 57 We received the data in its raw form and as screenshots archived in Powerpoint. Relevant experiments and traces are included in this report at Tabs 39,41,43,45, and 46. We have archived the remainder as part of the investigation case file. Those data logs and traces are available for review on request. 58 We correlated the aggregate results presented in the Paper with the spreadsheet calculations provided by CoAuthor 2. The data values from the spreadsheet were then correlated with the curve-fitting values that CoAuthor 1 wrote on the data logs for the experiments. ) ( (desu questioned data, the Subject Tab 37, page 4. 3) review of the data using time-to-peak (mP)selection criteria. Of these three, the instrumental gain changes are directly relevant to the data fabricatiodfalsification allegation, while the remaining two facets (as well as other facts) relate to the circumstances surrounding the data fabricatiodfalsification as well as the Subject's intent. 1. Gain Changes The gain setting on an instrument, such as t h e d e v i c e used in the Subject's data collections, functions much like the volume adjustment on a radio, with the same effect on the signal (i.e., current or sound volume) observed. For example, music playing on the radio increases or decreases in volume (amplitude) in response to and in proportion to the change in the volume setting in a particular direction. The amplitude of the signal transmitted from the radio station does not change, only the listener's perception of it based on the volume setting of the radio receiver. Similarly, the listener's perception of the background noise (static) accompanying the music would increase or decrease in proportion to the change in volume setting while the actual amplitude of the transmitted noise relative to the music would not change. Similarly in the ~ u b j e c t ' s e x ~ e r i m e n tts ,h e d e v i c e measures the current observed by the electrode, and the computer records the signal. The subject's- device had an articulated knob for adjusting the gain in discrete incremental units (Figure 4). Figure 4. Articulated gain knob on the-device that the Subject used, showing incremental settings.63 ~ h d e v i c ethe, gain knob, the perceived signal (i.e., recorded current and noise), and the computer are analogous to the radio receiver, the volume knob, the speaker output (i.e., music and static), and the listener, respectively. While the actual signal remains unchanged, an For larger views of the e n t i r e d e v i c e see Tab 33, Photos taken on OIG visit to the lab, images 4 and 5. adjustment to the gain knob will change the amplitude of the signal as perceived by the computer in proportion to the adjustment of the knob. In this case the gain knob has pre-defined incremental settings labeled 0.5, l , 2 , 5, 10, 20, 50, 100,200, 500, and 1000 ~ V / ~ A which ,@ required modest force to change. Therefore, an adjustment of a single increment results in an incremental change in the perceived signal by the same relative proportion.65 In the Subject's experiments, the gain knob was set at the beginning of each experiment66 and should have remained unchanged through the final trace. The actual setting used was dependant upon the cell and other experimental conditions such as the strength of the actual current being measured. The use of different gain settings between experiments (i.e.,for different cells) was accounted for during data processing, which determined the vertical scale bar displayed for each trace in the experiment. In this particular case, the instrumental configuration required that the Subject manually input the gain setting from t h e d e v i c e into the software on the computer as well as on the data log sheet because an automated system for recording the gain value from t h e d e v i c e on the computer was not a~ailable.~' In an ideal experiment, a constant electrical potential (V) is maintained across the cell membrane. In the Subject's experiments, this was denoted as eithe-on the computer on- the data log sheet. The basic relationship between electrical potential (V), current (I), and resistance (R), is expressed as V = IR. Generally, the resistance in these experiments is sufficient1 1ar e that any changes are relatively negligible. When the V is held c o n s t a n d the I consequently remains constant and is referred to as the holding current.68 As demonstrated in Figure 1, the holding current is reflected in the data as the distance between the baseline (I = 0) and the observed signal before the pulse or after the -. signal decayed. The example shown in Figure 5. '"'" 64 Each incremental change on the dial results in a perceived, not actual, change in the signal amplitude on the millivolt (mV) or picoampere (PA) scale. The data traces in the Subject's experiments were plotted with respect to pA against time. See Figure 1. 65 For example, changing the gain setting from 2 to 5 will result in an increase in the perceived amplitude by a factor of 2.5; or changing the gain setting from 1 to 0.5 will result in a decrease in the perceived amplitude by one-half. P6 Tab 37, page 15. 67 Tab 32, page 1; Tab 33, page 4; and Tab 37, pages 15-16. 68 The resistance, R, remains virtually constant in these experiments. 69 Biological vanability recognizes the fact that all biological systems are not uniform and measurements on such systems will demonstrate some inherent variability within certain limits. ~h-variability observed for the Subject's unquestioned experiments is much less than the-variability observed between traces where improper gain changes appear to have occurred. 70 Tab 37, page 17. " Based on the traces (Tabs 39 and 41) and the data log sheets (Tabs 40,42 and 46) there remains some question as e to whether -ctualljl remained constant across the traces in slngle experiments and across the experiments in eneral. According to the subject (Tab 37, page 18) data can only be compared to other data collected at the same the current scale bar overlain onto the traces did not change during the course of the experiment. The software generated this scale as a function of the gain setting, which the Subject manually entered into the software at the beginning of the experiment.72 Control Phase Drug Phase Wash Phase Figure 5: Representative traces from an experiment on -reported in the paper." Note that the-current and signal noise remain consistent across the series of traces. Control Phase Drug Phase Wash Phase a Figure 6. Representative traces from a single experiment74demonstrating the observed change in urrent and noise amplitude between phases of the experiments on a The current scale ba-remained constant across the series of traces indicating that no change in gain was recorded w5 in the software. The gain setting manually entered into the software was 72 Tab 37, page 15. 73 The traces are t r a c e s We selected these traces based on the Subject's own # 1,4,6." (See Tab 45.) 74 See Tab 39 75 See Tab 39, The combination of the change in-current amplitude was consistent with a manual change in the gain setting on device resulting in an artificial change in the perceived current amplitude in the In fact, in the course of the experiment that rovided the traces shown in Figure 6, the instrument captured what appears to be the change in¤t and noise amplitude in mid-trace (Figure 7). Figure 7. Experimental trace showing the apparent mid-trace change in holding current and noise amplitude resulting from an improper gain change. The trace is the last trace of the Drug Phase on the experiment.76 O f the twelve experiments in which gain changes between traces were originally questioned during the University's inquiry,77the one in Figure 7 was the only one that appeared to catch the manual gain change while the computer recorded the trace. The holding currents before and after the apparent gain change in this experiment were in a 2.5: 1 ratio, although the gain setting entered into the computer for all traces in this experiment was "2"; thus, the change in holding current is consistent with the Subject having changed the gain switch from "2" to "5" for the Drug Phase and then back to "2" for the Wash Phase. 76 Tab 39-, included traces reported in the Pape Tabs 41 and 42)) and cell experiments these traces during the University inquiry as "contain[ing] notable evidence of gain manipulation". We have independently evaluated the twelve as well as all of the data traces reported in the Paper for both Native and Variant cell experiments. Those cells showing indications of improper gain changes are identified in Tables in Tab 5 1. We showed the Subject the.traces from this experiment one at a time and in chronological order for his comments. When he saw the trace in Figure 7, the Subject said that it showed that he had lost the cell at the break in the line78and that the entire experiment should have been excluded from the published data set.79 The phrase lost the cell means that the cell is no longer viable for collecting data, yet the traces and data log showed that the Subject continued to collect traces in this experiment, which he could not have done if he had lost the cell. Furthermore, he wrote "Gorgeo;s! Gorgeous! on the data log for this experiment after collecting the traces: and used the data from the later traces in this experiment in making thA- ) In fact, the Sub'ect reported in the Paper a total oi All complete Variant 1 experiments and one ed symptoms of gain adjustments during the course of the experiment." Figure 8 ~orting agETega1 from the question nents. 83 78 Tab 37, page 19. "' Tab 37, page 19. We note that one of the five experiments (Tab 39 to be incomplete consisting only of traces collected unde experiments (Tab 4 1 , also appeared 83 Tab 5, page 3306. In another particular Variant 1 experiment,84the Subject not only appeared to have changed the gain setting between the Control and Drug Phases but also appeared to have changed the gain between each of the traces in the Drug The control traces showed very weak currents relative to the noise amplitude, such that the currents were almost indistinguishable from the the holding current and noise both increased dramatically in noise amplitude returned to control trace levels in the next the following t r a c e o n l y to decrease again in the next trace above in Figure 7 shows an experiment from the previous day in which the Subject appeared to have changed the gain in the middle of a drug trace, the pattern of apparent gain changes in this experiment suggests that the Subject observed the traces on the computer monitor directly as they were recorded. The pattern of alternating gain changes suggests that he increased the gain to view the weak current but returned to the lower gain setting in the next trace, knowing that the gain level should remain constant throughout the experiment.86 The Subject's comments on the data log sheetg7suggested that the data were not optimal, further indicating that he had reviewed the traces for quality before including them in the Paper. The Subject prepared all of the figures for the Paper including those that presented the Variant 1 and Variant 2 datag8 The highlighted traces shown in Figure 9 represent selected traces from a complete Variant 1 experiment. As described above, each of the Variant 1 experiments reported in the Paper appeared to contain improper gain changes based on changes in holding current and noise amplitude. To have produced the highlighted figure, the Subject must have aligned the traces vertically to achieve overlap in the holding current between the phases of the experiment.89 The Subject told us he edited the traces to align them horizontally to achieve overlap in the l T P but not to aligning them vertically.90 Although such editing appears .to be consistent with the practices of researchers in this field, the Subject would have needed to select the traces to use and in making the figure observed the differences in the holding current for the different phases of the experiment chosen for the figure. 84 ~ x ~ e r i m eTabs n t 39 and 40. The file designatio-indicates that this experiment was the third experiment on the da following the experiment shown in Figure 7. " E x p e r i m e n t ~ a e 2,3,4,5, s and 6 Tabs 39. 86 The data log indicates that there were 2(and & ( T a b 40.) Thus, we have t r e a t e a as if it were part of the-although it appears that the gain setting was at the lower, original level for this n ' m m e n t s on the log sheet included: "UGLY but effect on [illegible]"; "Serpentine stream!"; and -was -- dying!"; "last 2 were supposed to b e ( ? ) " , Tab 40. "'ab 37, page 19. 89 It was not necessary to a s t r a c e s from the eported in the current) betwee used one of the experiment reported in figure. 90 Tab 37, page 19. A [Native I ] m - [Native 21 I I [Variant 11 [Variant 21 Figure 9. -showing a composite of traces from the Variant experiments, with the questioned Variant 1 results The experiments with the Variant 1 and the Variant 2 cells were crucial to sugporting the structure-function relationship hypothesized from the Native cell and Tissue studies. The Native cells contained receptors of both types in differing pro ortions. Variant 1 and Variant 2 cells were prepared from cells, which did not have any of t h e b e e p t o r s of interest prior to transfection. Through the transfection process only one form of the receptor was conferred in each Variant ce11.93'94With the Variant 2 cells, the hypothesis indicated that the current amplitude would decrease or not change as observed for the Native 2 cells and Tissue 2. Therefore, it was the Variant 1 line of experiments that were the definitive experiments with regard to the conclusions reached in the Paper with regard to the structure-function relationship. Because all of the complete Variant 1 experiments show symptoms of improper gain changes -- Tab 5, p a g m between phases, it appears that the resulting structure-function conclusion relies 'entirely on falsified data. With regard to the Native cell experiments initially questioned of improper gain changes," six Native 2 cell experimentsg6appeared in aggregate -- - datafor the data point highlighted - - in yellow in Figure Five of those six Native 2 cell experiments show distinct changes in holding current aid changes in noise amplitude between phases characteristic of a change in gai; setting and consistent with the hypothesized decreasei n current amplitude.98 ~ h e s e experiments s appeared on the spreadsheet if calculations the calculations supporting the right-most point of the lower curve in This data point was the endpoint of the calculated S-shaped dose-res onse curve demonstrating the hypothesized That data point appears to have exerted a downward influence on the S-shaped -ring the curve-fitting procedure, thereby leaving the preceding data point the !utlier) above the curve. If the gain had not been manipulated for those experiments, thi n intensity would have been slight or nonexistent. Consequently the endpoint would lie much closer upper curve, which depicted current amplitudein the absence-of the modulator. The curve that would fit the data with the non-fabricated endpoint would likely have included the Outlier. The downward shift in the curve correlates with the Subject's statement in the original draft of the Paper: 95CoAuthor 2, Grad Student 1, and Grad Student 2 initially identified for the Panel t h e m a p p a r e n t l y e r changes (See Tab 7 and footnote 16). We have inde~endentlvassessed the data traces and showing i m ~ r o ~gain %remaining one-is more difficult to assess definitively; however, CoAuthor 2 andmcurrent graduate ''Tab 48, page 5. loo During curve-fitting, the points at either end of the data set tend 'to exert the most influence on the ultimate shape and slope of the curve calculated to fit the data. lo' Tab 13, page 13. This conclusion carried into the final published version of the Paper. 102,103 A mative 11 [Native 21 Figure 10. r e p o r t i n g aggregate data from. the questioned Native 2 experiments, highlighted in yellow.Io4 The Native 2 data highlighted in blue are not in question. - - - Paper, relied i n these figures in preparing the subsequent drafts bf the Paper. '04 ~ a 5-,b The remaining experiment initially questioned as showing changes in holding current and noise amplitude between phases105does not appear to have been reported in the Paper. We recognize that successful and unsuccessful experiments are recorded in general practice across scientific disciplines. We have treated this Variant 2 experiment as data recorded but discarded from use in the Paper. Based on our assessment of this experiment, we have determined that the evidence is insufficient to include this experiment among those we have identified with questionable gain manipulations. The proportion of traces ultimately reported as aggregate data in the Paper in relation to the total data published in the Paper is summarized in Table 1. Table 1. Relative Proportion of Questioned Traces. Total Number of % of Total Number Published Total Number Total Questioned Cell type of Experiments of Traces in Number of Traces in Experiments with Questioned Published Questioned Published Published Traces Io6 Data Traces lo7 Data Native 1 31 0 27 1 0 0 Native 2 28 7 227 12 5.3 108 Variant 1 5 39 13 33.3 Variant 2 7 1 62 The Subject said that he showed some traces with the increased noise level to Coauthor 2, who, after consultation with a researcher at another university,109told him that the increased noise resulted from channel opening.' lo Coauthor 2 remembered discussing the increased noise with the other researcher in order to posit what could cause such an observed increase. ' I ' ol- See Tab 46. 106 See Tables 1 and 2 at Tab 51 for specific details on the experiments containing questioned traces. "O Tab 8,page 8; and Tab 37, page 17. Ill Tab 33, page 6. ' Coauthor 2 and t w o L o m g r a d u a t estudents (Grad Student 1 l 2 and Grad Student 2' 13) have subsequently analyzed the data and have been unable to identify any biological cause consistent with the reported results. ' I 4 The Subject did not admit to changing the gain during the experiments. He attributed the changes in holding current and noise amplitude to factors such as the large size of the transfected cells,' l5 the cell's position in the solution stream,' l6 and other factors. 'I7 The Subject's explanations were inconsistent with the data. The factors the Subject cited would be random in effect rather than producing changes in the data that were consistent with incremental changes in the gain setting in each experiment and that were always consistent with the hypothesized ' result. lg Thus, the only remaining plausible explanation for the apparent changes in holding current and noise amplitude across the entire set of questioned ex eriments is a manual, unrecorded incremental change in the gain setting on thed e v i c e . The Subject collected the data, wrote the instrument settings on the data log, and entered ' the parameters, including the gain setting, into the computer. l 9 The Subject set up the experiments and made the necessary calibrations and adjustments. 120 Coauthor 1's primary role in the data collection was to observe and learn the technique.12' After limited attempts to collect data on her own, they determined that Coauthor 1 did not have the touch for it.'22 They reallocated the workload such that Coauthor 1 fit the curves after the Subject collected the data Tab 33, page 8; Tab 35, panes 1-2; and Tab 37, pane 8. 'IY and entered the curve-fitting parameters onto the data log sheets under the heading "Analyzed ~ a t a . " Although '~~ Coauthor 1 was physical1 in the room while the Subject collected some of the questioned data, she was involved int h e cells, preparing for coursework, and other projects. '24 It is unlikely that she was positioned to see the Subject's access to knob or at that time fully understood any adjustments she saw the Subject make to the e v i c e during the course of experiments. ' 2 5 Therefore, the Subject was the individual responsible for adjusting the gain to achieve the hypothesized results. 2. Anomalous Solution-Switching Times An essential component of t h e m t e c h n i q u e is the ability to expose the outside of a cell Iz3 The data collection and curve fitting were all done on a single, shared Macintosh computer; therefore, only one person could work at the workstation at any given time. Iz4 Tab 35. Dages 1-2. See also Tab 33. Dage 4. fulfilling all of the roles ascribed by the Subject while maintaining a clear view of any adjustments the Subject made to the instrumentation for the data collection. '26 The instrument t h a a s seen pictured in lower part of images 4 and 5 at Tab 33. '27 Gravity-fed solution transfer is similar to the process of siphoning a liquid from a tank or other reservoir with a piece of tubing. Once the tube is filled with the liquid, the tubing will draw liquid out of the reservoir when one end (the intake end) is placed in the reservoir and the other (the output end) is located somewhere below the level of the reservoir. The speed of liquid flowing through the tubing will increase as the output end is moved farther below the level of the reservoir. This increase in speed or flow rate occurs because of the longer distance through which gravity can accelerate the liquid's flow. 12' Tab 33, page 2. Figure 11. Diagram of thes y s t e m adapted from the diagram supplied by Coauthor 2. When the pulse was initiated, the computer began recording data including a for the trace. After two or three Control traces, the Subject would move the cell awa i t h a micromanipulator, apply a vacuum to the tubing to remove the solutions, and move the feeder tubes to 'the reservoirs for the Drug Phase.132 After collecting the final trace of the Drug Phase, the Subject moved the feeder tubes back to the original solutions for the Wash Phase. It was the time necessary to change the solutions between phases (i.e.,the solution-switching time) that the University Inquiry Panel addressed when it wrote: "It appears unlikely, and perhaps physically impossible, that solution-switching could have been conducted as quickly as indicated on some of the computer time stamps."133 The minimum amount of time required to execute all the necessary physical manipulations to change the solution between a trace at the end of a phase and the beginning of the next was 30-40 seconds.134,135 While the Subject explained at length why there was 13' Tab 37, pages 11-12. '33 page Tab 3, 4. '34 CoAuthor 2 a n d o t w o graduate students described their concerns after reviewing the time stamps on the m at Tab 4, pages 5-6. variability in the solution-switching times for his experiments, 136 he did not adequately explain how he was able to perform the physical actions necessary for solution-switching in less than 30 seconds and as quickly as 4 ~ e c 0 n d s . lRather, ~~ he addressed, in general terms, several "shortcuts" he developed such as shortening the solution delivery tubing and raising the reservoirs to increase solution flow rates.13* In addition to the physical actions necessary for solution-switching, the experiment -. required a,sufficient amount of time to purge the previous solutions fiom the tubing139and to reestablish a stable between the solution streams (which was observable through the microscope). Coauthor 1 indicated that waiting one minute would be necessary.O' contrast, the subject said that after moving the tubing and a 1 in the vacuum to clear the theta tube, he acquired the next trace as soon as he observed th He never measured how long it took for the solution to come through the tube, because the change was "Very spontaneous."141 He did not know how long to wait; and he never knew if the previous solution was coming out mixed with the new solution. The current practice in Coauthor 2's laboratory is to wait 2 minutes to ensure a full purge of the tubing with the new solution. Four of the Variant 1 and five of the Variant 2 experiments in the Paper involved solution-switchingtimes of less than 30 seconds duration. We also identified short solution- switching times in experiments that did not show changes in holding current and noise amplitude.145These observations suggest that the Subject rushed through the experiments and did not conduct the experiments as he reported in the Paper. The Subject's only explanation was that the problem would average itself out. 135 Tab 33, page 2; Tab 35, page 5; Tab 23page 5. 136 Tab 8, pages 3-4 and Tab 23, page 4 ("I became very proficient with the process and, as explained in my.previous letter to you, [Tab 81 I was able to shortcut some of the procedures that a less experienced and less confident researcher might not be willing to do.") 13' See Tab 5 1, Table 1. 138 Tab 8, pages 3-4 and Tab 23, pages 4-5. See also Tab 35, page 4; and footnote 127. '39 The Subject minimized this time to some degree by the application of a vacuum to the system, thereby pulling the solutions back through t h m t u b e . However, that process introduced solution from the dish holding the cell into the tubes. Thus, it was still necessary to allow the new solutions to flow in order to purge the tubing and establish the flow of the appropriate solutions for the particular phase of the experiment. I4O Tab 35, page 5. 141 Tab 37, page 20. 14' Tab 37, pages 12-13. '43 Tab 33, page 2. Of the remaining experiments, one Variant -1 and one Variant 2 x p e r i m e n t contained only Control traces therefore no solution-switchingtook place in those experiments. In the remaining Variant 2 experiment- solution-switching appeared to have taken place in 128 seconds, although we deduced this conclusion from conflicting information found on the data log sheet and the data traces with regard to which trace was the first Drug Phase trace. We have tentatively resolved the conflict in favor of the explanation most likely consistent with the published results and have not included it as one of the five experiments with solution-switching times < 30 seconds. 145 ~ a 51. b 146 Tab 37, page 13. Thus, the evidence shows that although the Subject collected data on cells, he did not do so according to the method reported in the Paper because he either did not actually perform the physical manipulations required to establish the conditions for those experiments or provide sufficient time for the solutions to flow such that the cells were exposed as reported. This is consistent with the alleged improper gain manipulations discussed above, because the Subject created the desired experimental results by changing the gain setting. Thus, it did not matter whether he exposed the cells to the correct experimental conditions. 3. The Subject's Review o f the Data The Subject asserted that he did not review the traces or perform the curve-fitting.147He has both denied and claimed responsibility for the data used in the publication.148He placed the responsibility for these activities on CoAuthor 1, a graduate student he was assigned to mentor.149Our reviews of the Subject's laboratory notes 150,151 and his assessment of data during our interview152demonstrated that he possesses the capability for all the functions he attributed to his mentee. Additionally, even if CoAuthor 1 was responsible for these data analyses and 147 Although he did h o w how to fit curves, he relied exclusively on CoAuthor 1 to fit the curves because she had gain knob during data collection and not through the mathematical analysis conducted by his inexperienced subordinate. 148 CoAuthor 2 assigned the Subject the responsibilities of supervising CoAuthor 1 as part of his training in her laboratory. Tab 37, pages 5-6. 149 The Subject explained to us that he used CoAuthor to d o a n d other parts of the experiments he did not want to do, particularly the the curve-fitting. (Tab 37, pages 13 and 19). I SO The Subject recorded his experhental notes on loose-leaf pages bound in three-ring binder, which was consistent with the practice of that laboratory. Is' On data logs for other experiments, the Subject made specific notations about baseline, curve shape, and TTP, sometimes rejecting the use of those data sets. For example_"nice TTP r n a y b e m n l y " (Tab 40); quality!!!" (Tab 42 ' (Tab 46); andG a r b a g e anyway baseline change for each trace !?!b9 "too bad, do not use. No wash. Nice kinetics and hu e effect though . . also Tab 35, page 3 and CoAuthor 1's annotations in green to the data sheet fo-dentifying (Tab 46). See the Subject's handwriting including comments such as "TAKE sweeps #1,9, 11, 13." Is2 The Subject and Coauthor 1 each told as the primary criterion for identifying the useful data but that-the criteria were relaxed with regard to the Tab 33, page 6; Tab 35, page 3; and Tab 37, page 13. When we showed the Subject the series of traces from the experiment shown in Figures 6 and 7 one at time and in sequence, he determined by visual inspection within the first four traces that the traces were "bad data" and that the experiment should not have been incorporated into the Paper (Tab 37, page 19). In explaining his rationale for that conclusion, he indicated that based on his experience ("you train you[r] eye to detect good and bad current'' (Tab 37, page lo)), the shape of the peak, including t h e was not acceptable and he would not have included the data in the Paper (Tab 37, page 19). He made this assessment visually before we showed him trace 5, which contained the mid-trace change in gain. Then the Subject contradicted this by answering affirmatively when we asked, hypothetically, if trace 5 did not exist, would he have included the experiment as good data for the Paper. The Subject appeared shocked to learn that the experiment had been included in the Paper, but also confirmed to us that he wrote "Gorgeous! Gorgeous! Gorgeous!" (Tab 40- on the data log. selection activities, it does not mitigate the Subject's manipulation of the articulated gain knob to gather data that conformed with the theory to be presented in the Paper. The Paper presented both structure-function and behavioral relevance of the observations. According to the Subject, his primary interests were in the behavioral aspects of the study, and the structure-function information was of primary importance to CoAuthor 2. 153 The Subject has asserted that he gathered the structure-function information to satisfy CoAuthor 2's interests. While this appears to be true, it again does not mitigate the Subject's manipulation of the gain knob to acquire data that conformed with the theory to be presented in the Paper. The Subject has asserted that both CoAuthor 1 and CoAuthor 2 154 had delineated roles in the data preparation and analysis,155and in writing the paper,156respectively. We accept his assertions. Any roles CoAuthor 1 or CoAuthor 2 may have had in these efforts or any differences the Subject may have had with CoAuthor 2 about research interests do not alter the Subject's manipulation of the gain knob to acquire data that conformed with the theory to be presented in the Paper and that the two coauthors were involved in the analysis of already altered data. OIG's Assessment The Subject's actions in this case took place before April 17,2002; therefore, the definition of misconduct that was contemporaneous with his actions applies. 157 In all other Although the Subject and CoAuthor 2 decided together that the experiments on-would be a good experimental approach to pursue, the Subject asserts that their interests ultimately diverged. (Tab 37, page 4.) This divergence manifested itself,.according to the Subject, when he refused to do the ( " I wasn't going to do it " which made CoAuthor 2 "upset.") (Tab 37, page 8.) The Subject has distanced himself from the Pa e r ' s ~ c o n c l u s i o n based s on the data that he collected by armbuting the interest in t h e e l a t i o n s h i p to CoAuthor 2: [Tlhis is a hypothesis developed by [CoAuthor 21. Thus, it is she who rightly receives the accolades when research such as that which I conducted in her Lab, confirms her previous hypothesis. Thus I am not an important player in the finding that as that was a known fact in the Lab by the time I arrived. [Tab 23, page 8.1 See also Tab 8, page 10 and Tab 37, page 5, particularly the discussion of CoAuthor 2 wanting the paper finished for inclusion in her tenure review package. CoAuthor 2 was the Subject's postdoctoral advisor. She treated the Subject as a colleague rather than as a student, as expected in the mentor-postdoctoral fellow relationship. (Tab 37, page 6. See also Tab 33, page 7.) As such CoAuthor 2 reviewed data only when the Subject showed it to her and did not have day-to-day review of the data as it was being collected. (Tabs 37, page 14. See also Tab 33, page 4.) 155 See footnotes 147 through 149. The final version of the Paper was a remarkably different document with respect to writing style reflecting CoAuthor 2's role as the lead in editing and revising the,drafts. (Tab 33, page 8.) CoAuthor 2's primary role was preparation of the draft for submission, in part because the Subject's first language was not English. CoAuthor 1 was involved only to the extent of reviewing for grammer and clarity. However, the pertinent experimental details and data remained consistent between the first draft as prepared by the Subject and final versions of the Paper. (Tab 5 and Tab 13.) . 157 See footnote 2. respects, NSF's current research misconduct regulation applies to this case. A finding of misconduct requires (1) an act that constitutes "fabrication, falsification, plagiarism, or other serious deviation from accepted practices in proposing, carrying out, or reporting results from activities funded by NSF"; (2) the act constitutes a significant departure from accepted practices of the relevant research community, (3) the act was committed intentionally, knowingly, or recklessly, and (4) the facts are proven by a preponderance of the evidence.158 The Act We have concluded that the Subject manually changed the gain setting on t h e m d e v i c e so that the Variant 1 cells appeared to exhibit results in support of the hypothesized change in observed current. We have concluded similarly for the Native 2 cells used to support the highlighted data point in Figure 10. The relative changes in amplitude of both the holding current and the noise were consistent with incremental adjustments in the gain setting. The timing between solution switches, in most instances, was unreasonably short for the steps necessary to execute not only the physical actions necessary for a solution-switch but also to allow sufficient time to ensure that the neurons were exposed to the solutions exiting t h e m a s reported. Fabrication and falsification of data fall under the definition of research misconduct. '51 Fabricating and falsifying data strike at the heart of research, and the extent of the Subject's falsification constitutes a significant departure from accepted practices. Intent A change in the gain setting would of necessity be a knowing act because the ain knob had discrete incremental settings that required more than negligible force to change. ' l The necessary gain settings to achieve the desired results also required changes between specific traces in the experiments in specific directions, because the Wash Phase required the traces to return to the conditions of the Control Phase. Such actions were necessarily calculated rather than random or accidental. We conclude that the Subject knowingly changed the gain on the -device during the data acquisition in order to support the desired hypothesis. The Subject reviewed, selected, and incorporated the data into the manuscript for the Paper. We conclude that the Subject knowingly reported falsified data in the original manuscript, which ultimately became the Paper. Standard o f Proof We conclude that a preponderance of the evidence shows that the Subject has knowingly committed data falsification in reporting results in the Paper. Is8 45 C.F.R.5 689.2(c). Is9 The falsification occurred prior to April 17,2002, therefore we use the definition of research misconduct as described in footnote 2. 160 OIG investigators manually adjusted the gain knob during their visit to the lab where the equipment is housed today. The laboratory has upgraded the equipment since the Subject has left the group. Tab 32, page I, and Tab 33, page 1. 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. [I611 Seriousness The Subject's actions were very serious because they demonstrated his disregard for the accurate reporting of experimental results, with a negative effect on the literature in this field. Although the Subject has admitted that there were "some flaws in the data," he has asserted that these flaws "averaged out" in the data analysis, attributing the idea of averaging out the flaws to CoAuthor 1. However the entire data set for Variant 1 was created by a mechanical manipulation of the-device, undermining the validity of the average of the data from the outset. The conclusions derived from these data were a significant achievement reported in the Paper. 16' Similarly, for the Native 2 cells whose data were aggregated as the endpoint which deflected the dose-response curve in Figure 4 exaggerating the effect of the modulator on isolated Native cells. Therefore, we conclude that the Subject's actions were very serious. Degree o f Intent The Subject's intent exceeded the reckless intent that the regulation requires for a finding of research misconduct. The Subject knowingly adjusted the gain during the experiments discussed above to produce results consistent with the hypothesis and knowingly incorporated those results into the Paper. 16' 45 C.F.R. $689.30). '62 Tab 37, page 13. 163 See footnote 62. CONFIDENTIAL Pattern We concluded that there was insufficient evidence to support a pattern of misconduct beyond the activities covered in this report. Impact on the research record The fabricated data have had a significant impact on the published research record. The data provided substantial support for the structure-function hypothesis that explained the observations in the Tissue studies. We identified twenty-seven papers in the Science Citation Index that cited the Paper, of which twenty were by authors other than the Subject and/or his coauthor^.'^^ We also note that Grad Student 1 and Grad Student 2 have had manuscripts with contrary results have been delayed in reaching the broader research community. 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; debar the Subject from receiving federal funds for a period of 2 years commencing on the date of NSF's finding of research misconduct; require the Subject to certify to NSF that the publication containing the fabricated and falsified data has been retracted; 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), co-principal investigator (Co-PI), or otherwise a participant; require the Subject to certify each time he submits a proposal or report to NSF that the proposal or report does not contain fabricated or falsified material for 3 years after the debarment period; 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 fabricated or falsified material for 3 years after the debarment period; and bar the Subject from serving as a reviewer of NSF proposals for 3 years commencing on the date of NSF's finding of research misconduct. The Subiect's Response to the Draft Investigation Report The Subject did not submit a response to the Draft Investigation Report directly nor did he respond through his attorney.
Data Tampering / Sabotage / Fabrication
Published by the National Science Foundation, Office of Inspector General on 2007-03-19.
Below is a raw (and likely hideous) rendition of the original report. (PDF)