plagiarismavoiding plagiarism (ii) guideline 9: when in do btdoubt, pro ideprovide a citation....

PlagiarismPlagiarism

NSF D fi iti f Pl i iNSF Definition of Plagiarism

“The appropriation of another person’sid l dideas, processes, results or wordswithout giving appropriate credit.”

i.e., it is literary and intellectual theft.e., s e y d e ec u e

http://nsf.gov/oig/resmisreg.pdf ; 45 C.F.R. 689.1

UR Definition of Plagiarismg

“The use, whether deliberate or unintentional, of an idea,h h i l f i hphrase, or other materials from a source without proper

acknowledgment of that source in a work for which thestudent claims authorship; the misrepresentation ofp psources used in a work for which the student claimsauthorship; the improper use of course materials in a workfor which the student claims authorship; the use of papersfor which the student claims authorship; the use of paperspurchased online and turned in as one's own work;submitting written work, such as laboratory reports,

hi h h b i dcomputer programs, or papers, which have been copiedfrom the work of other students, with or without theirknowledge and consent.”g

http://www.rochester.edu/College/CCAS/AdviserHandbook/AcadHonesty.html#plagiarism

Common Types of PlagiarismCommon Types of Plagiarism

• Copying & pasting text from websites or transcribingCopying & pasting text from websites, or transcribingtext from printed material is plagiarism. So is simpletranslation of text from one language to another.

• Unpermitted or unacknowledged use of videos,photos, slides or images is plagiarism.

• Unacknowledged modification of text is plagiarism(e.g., replacing a few words, by using a thesaurus, orrearranging the text slightly is not originalrearranging the text slightly is not originalscholarship).

http://www.ehhs.cmich.edu/~mspears/whatis.html (no longer online)http://uwf.edu/dupserv/plagbroch.pdf

An Example of PlagiarismAn Example of Plagiarism

Reasons Honest People Plagiarizep g

Lack of understanding of the concept of plagiarism. You can help!Some junior trainees may not understand what plagiarism is. By explaining it, you’ll be doing them a big favor.

l l k i i i i h diffi lWriting is hard. People may lack experience in writing & have difficulty expressing themselves well. So they lift or rearrange sentences. There is no magic solution; only practise and effort.

Lack of scientific knowledge. If you do not fully understand the science, you cannot tell which words/phrases are important; this may cause you to lift sentences or phrases verbatim The solution is to ask questions; knowlift sentences or phrases verbatim. The solution is to ask questions; know the science - then write your report.

Detection of Plagiarismg

• A piece of text is just too good: A student who ordinarily does notp j g ywrite well turns in a paper in which some sections are grammaticallyincorrect and have spelling errors (the ones they wrote themselves)while other sections are fabulous and read just like a textbookwhile other sections are fabulous and read just like a textbook.

• A piece of text seems awfully familiar: A student turns in a paper thatit is somehow very familiar (in whole or in part) to the readerit is somehow very familiar (in whole or in part) to the reader.

• Text search/plagiarism detection software: Readily available and usedby some medical journal publishers. One can search the PMC databaseeasily. Deja vu is an example: http://spore.swmed.edu/dejavu/browse

Why is this plagiarism?y p g

• Significant chunks of text are copied verbatim

• The structure and organization of the abstracts areidentical. This exemplifies “patchwriting”, which iswhen the scientific design of a paper is plagiarized(along with much of the text) and the author’s data areessentially plugged in, replacing the original data.

• 21 of 36 references are identical

• There are no common authors between the two papers.The original article IS cited in the duplicate article,g p ,which is unusual for a plagiarized article.

Evaluating the Seriousness ofEvaluating the Seriousness of Plagiarism

Mark Wiser (Tulane U.) suggests one should consider:

1 Extent of the plagiarism1. Extent of the plagiarism2. Intent (malicious or not?)3 Previous history of plagiarism by the author3. Previous history of plagiarism by the author4. Position and training of the author5 Did the plagiarism occur from notes or original5. Did the plagiarism occur from notes or original

material?

Benos D. et al. Adv. Physiol. Educ. 29:59, 2005

Th Th iThe Thesis

“The doctoral thesis is expected to be an originalcritical or synthetic treatment of an appropriatecritical or synthetic treatment of an appropriatesubject, an original work of creative art, or anextended report of independent research, formulatedin a scholarly manner and of a general excellenceconsistent with publication as a book or in scholarlyjournals of quality.”

http://www.rochester.edu/Theses/index.html

C I i ThCommon Issues in Theses

• Inclusion of the author’s previously published papers:This is acceptable and even routine, with properacknowledgement.

• Inclusion of data from experiments performed byInclusion of data from experiments performed byothers: Acceptable, with suitable acknowledgement.

• Incl sion of information from a grant application:• Inclusion of information from a grant application:Acceptable, with suitable acknowledgement.

Avoiding Plagiarism (I)g g ( )

Guideline 1: ALWAYS acknowledge the contributions of others andgthe source of your ideas.

Guideline 2: Any verbatim text taken from another author must beGuideline 2: Any verbatim text taken from another author must beenclosed in quotation marks.

Guideline 4: When we summarize, we condense, in our own words, aGuideline 4: When we summarize, we condense, in our own words, asubstantial amount of material into a shorter form.

Guideline 7: In order to make a proper paraphrase we must have aGuideline 7: In order to make a proper paraphrase, we must have athorough understanding of the ideas and terminology being used.

http://facpub.stjohns.edu/~roigm/plagiarism/Plagiarism.html

Avoiding Plagiarism (II)

Guideline 9: When in do bt pro ide a citationGuideline 9: When in doubt, provide a citation.

Guideline 10: When we submit a manuscript for publicationcontaining data that have already been disseminated in somecontaining data that have already been disseminated in somesignificant manner (e.g., published in another journal, presentedat a conference, posted on the web) we must clearly indicate this.

htt //f b tj h d / i / l i i /Pl i i ht lhttp://facpub.stjohns.edu/~roigm/plagiarism/Plagiarism.html

P ibl P ltiPossible Penalties

Possible penalties at UR include:

• Requirement to repeat the assignment• A request to write formal letter of apology• An official letter of reprimand from the Dean

Di i li b ti f d fi d i d• Disciplinary probation, for a defined period• A change in grade and/or issuance of a failing grade• Academic suspensionAcademic suspension• Explusion

http://www.rochester.edu/GradBulletin/ -> Regulations

Wh t i NOT l i i ?What is NOT plagiarism?

NIH Office of Research Integrity (ORI):

ORI generally does not pursue the limited use of identical or nearly-identical phrases which describe a commonly-used methodology…..because ORI does not consider such use as substantially

i l di h d i i imisleading to the reader or of great significance.”

http://ori.dhhs.gov/policies/plagiarism.shtml

Scientific MisconductScientific Misconduct

Wh t i Mi d t?What is Misconduct?

Federal policy on misconduct in science defines research misconduct as “fabrication, falsification, or plagiarism in f , f f , p gproposing, performing, or reviewing research, or in reporting research results”.

http://www.ostp.gov/html/001207_3.html

H i Mi d t D fi d?How is Misconduct Defined?

Fabrication is making up data or results and recording or Fabrication is making up data or results and recording orreporting them.

Falsification is manipulating research materials,Falsification is manipulating research materials,equipment, or processes, or changing or omitting data orresults such that the research is not accurately representedin the research recordin the research record.

Plagiarism is the appropriation of another person’s ideas,processes, results, or words without giving appropriateprocesses, results, or words without giving appropriatecredit.

http://www.ostp.gov/html/001207_3.html

Wh H i C f Mi d ?What Happens in Cases of Misconduct?

Loss of grant support: banned from receipt of federalgrants.

Loss of job: Betsy Goodwin (U-Wisconsin) is alleged tohave falsified data in grant applications. In 2006, herstudents turned her in. UW returned the grants andgGoodwin resigned. Science 313:1222, 2006. The Scientist, Sept.2007.

L f f d E i P hl (f l f UVM) Loss of freedom: Eric Poehlman (formerly of UVM) wassentenced to federal prison for a year and a day in June of2006, for using false data to obtain federal grants.

Formal Procedures (I)( )

Initial Reporting and Inquiryp g q y

• A written allegation is provided to the person’s supervisor, who then informs the dean.

• The dean informs the provost and the senior VP and vice provost for health affairs (if this occurs in SMD or SON), who determines whether the allegations merit further scrutiny.

• If yes, an initial enquiry is carried out with input from the person h i b i i d Th d h d id ifwhose actions are being questioned. The dean then decides if a

formal investigation is warranted.

http://www.rochester.edu/provost/FacultyHandbook/-> Misconduct

Formal Procedures (II)( )

Investigation

• The dean notifies the person being investigated and the director of ORPA, and appoints a fact-finding committee of at least 3 faculty

b ff d b h i imembers unaffected by the inquiry.

• The person(s) whose conduct is being investigated is given a written summary of all allegations and the opportunity to respondwritten summary of all allegations and the opportunity to respond in writing.

• The committee files a report whether the dean stating whether orThe committee files a report whether the dean, stating whether or not the preponderance of the evidence indicates misconduct.


Formal Procedures (III)( )Further Action

If t t i i l d ORPA ill i f th• If grant support is involved, ORPA will inform the sponsor.

• If publications are involved, the dept. chair will be asked to contact relevant journal editorscontact relevant journal editors.

• Disciplinary action will be taken, possibly including terminationtermination.

• If misconduct did not occur, reasonable efforts will be undertaken to restore the reputation of the individual whoseundertaken to restore the reputation of the individual whose conduct was questioned and to protect the standing of those who raised the questions – unless they acted in bad faith.


How Scientists Regard Misconductg

In 1998 survey of 606 NSF grantees, scientists ratedIn 1998 survey of 606 NSF grantees, scientists ratedvarious acts for their “malfeasance rating” and unethicalnature. The most unethical acts of all were:

1. Fabricating data2. Throwing out negative data3. Copying ideas or text without attribution4 U i id f fid ti l t li ti4. Using ideas from a confidential grant application5. Using data without attribution of its source

Korenman et al. JAMA 279:41, 1998

(Self-Reported) Major Misconduct is Rare

Martinson, B.C. et al. Nature 435:737, 2005..

…But Other Behaviours are not


…and Bad Science is Common


Misconduct may also be under-reported

NAS: Scientists have "an obligation toNAS: Scientists have "an obligation to act" when suspected acts of scientific misconduct are observed…..Yet, the number of investigations reported to the ORI is low (only 24/year).

Titus et al. surveyed >2,000 scientists holding NIH grants and concluded that there may be as many as 2 300 observedthere may be as many as 2,300 observed acts of misconduct among all NIH/DHHS funded scientists per year.nation). Titus et al. Nature 453:980982, 2008..

New Approaches May be NeededNew Approaches May be Needed

It may be unhelpful to regard misconduct as something weIt may be unhelpful to regard misconduct as something we can eliminate by getting rid of a few “bad apples”.

“N ll f i d ll i d“Not all cases of misconduct are equally egregious, and not all perpetrators deserve to be branded as cheaters for the rest of their careers. There is often room for honest mistakes and differences of opinion.….. Within individual labs, airing complex matters — such as decisions about when data can be justifiably excluded from analysis may reduce thebe justifiably excluded from analysis…. may reduce the chance that any single investigator’s decision will later lead to accusations of misconduct.”

N t Edit i l (2008)- Nature Editorial (2008)

Statistics and PhotoshopStatistics and Photoshop

Two key sources of potential data misrepresentation arestatistical analyses and image manipulationstatistical analyses and image manipulation(“Photoshopping”). For example, data that appearstatistically “too good to be true” probably are exactly that.Other times, a failure to use the wrong statistical test ormixing up standard error and standard deviation can lead toerroneous and even misleading conclusions.

Photoshopping & Its ConsequencesPhotoshopping & Its Consequences

Nature 453:969, 2008

Journal Policy on Image ManipulationJournal Policy on Image Manipulation

"No specific feature within an image may be enhancedNo specific feature within an image may be enhanced,obscured, moved, removed, or introduced.”

“The grouping of images from different parts of the same gelThe grouping of images from different parts of the same gel,or from different gels, fields, or exposures must be madeexplicit.”

“Adjustments of brightness, contrast, or color balance areacceptable if they are applied to the whole image and as longp y pp g gas they do not obscure or eliminate any information.”

- J. Cell Biol.

Common Forms of Image ManipulationCommon Forms of Image Manipulation

1 G i t ti ( d l ti l tti d1. Gross misrepresentation (e.g., deleting lanes, cutting-and-pasting the same control lanes onto multiple gels)

2. Selective removal of background bands or blemishes

3. Selective enhancement of specific regions of an image

4. Juxtaposition of data from different gels (or p g (photographs) onto the same final figure (the gel should be re-run!)

Rossner M & Yamada KM. J Cell Biol 166:11, 2004.

Examples of Image Manipulation (I)Examples of Image Manipulation (I)


Examples of Image Manipulation (II)Examples of Image Manipulation (II)


Statistical MisconductStatistical Misconduct

Science 321:1144, 2008

Are these data too good to be true?

C St ti ti l E (I)Common Statistical Errors (I)

38% of papers in Nature in 2001 contained at least one statistical error! Common errors were:

• Overprecise p values (e.g., p = 0.002387)

• Inaccurate rounding. g

• Failure to state sample sizes

• Reporting p values without stating the test used

Edit i l N t M di i 11 1 2005

Reporting p values without stating the test used

Editorial. Nature Medicine 11:1, 2005Garcia-Berthou, E & Alcaraz C. BMC Medical Research Methodology 4:13, 2004

C St ti ti l E (II)Common Statistical Errors (II)

Its not just Nature! Papers published in Infection andIts not just Nature! Papers published in Infection andImmunity in 2002 contained a similar error rate (54% had atleast one statistical error). The errors here were:

• Reporting errors (22%): Unlabeled or inappropriate error bars ormeasures of variability; failure to describe statistical tests

• Errors of analysis (20%): failure to adjust or account for multiplecomparisons; making conclusions without a statistical test; use ofstatistical tests that assume a normal distribution on data that follow askewed distribution

• Both types of errors (12%)

Olsen CH. Infect Immun 71:6689, 2003.

P d d S iPropaganda and Spin

P d I d M i S iPropaganda Index: Measuring Spin

A major concern in clinical trial reports

• Our results approached but did not reach significance

• Our results would have been statistically significant if wey g f fhave a bigger sample

VS

• Our results were not statistically significant

Nature Medicine 15:1100-01, 2009

C l i (I)Conclusions (I)

• Scientific misconduct includes fabrication, falsification andplagiarism.

• Plagiarism is easily detected and includes duplicatebli tipublication.

• Fabrication and falsification commonly involve statisticaldata manipulation or “photoshopping”

C l i (II)Conclusions (II)

• Bad science is much more more common than outrightmisconduct - and also damaging to the fieldg g

• Scientists have an ethical obligation to act when theyScientists have an ethical obligation to act when theyobserve potential misconduct

• Many accusations of misconduct could be avoided thrubetter communication

plagiarismavoiding plagiarism (ii) guideline 9: when in do btdoubt, pro ideprovide a citation....

Documents