"Long before it's in the papers"
October 27, 2008

RETURN TO THE WORLD SCIENCE HOME PAGE

Faked research data surprisingly common, survey suggests

June 19, 2008
World Science staff

Sci­en­tif­ic mis­con­duct, no­tably in­clud­ing fal­sifica­t­ion of da­ta, may be far more com­mon than sus­pected, ac­cord­ing to the au­thors of a new sur­vey of more than 2,000 sci­en­tists.

San­dra L. Ti­tus and col­leagues at the Of­fice of Re­search In­tegr­ity of the U.S. De­part­ment of Health and Hu­man Ser­vic­es in Rock­ville, Md., sur­veyed 2,212 sci­en­tists at 605 in­sti­tu­tions. They found that nearly 9 per­cent be­lieved they had seen po­ten­tial re­search mis­con­duct in the pre­vious three years.

The find­ings are pub­lished in a com­men­tary in June 19 is­sue of the re­search jour­nal Na­ture.

The results suggest as many as 2,300 ob­serva­t­ions of mis­con­duct, 1,000 of them un­re­ported, oc­cur each year in the larg­er re­search com­mun­ity funded by the U.S. Na­tional In­sti­tutes of Health, Ti­tus and col­leagues wrote. They added that it’s un­likely such be­hav­ior is con­fined to the Un­ited States.

Sur­vey par­ti­ci­pants de­scribed misbe­hav­ior rang­ing from sci­en­tists’ chang­ing num­bers to make re­sults look more def­i­nite than they really were, to more cre­a­tive fab­rica­t­ions. One par­ti­ci­pant told of a col­league us­ing Pho­to­shop to tweak re­sults of chem­i­cal tests that ap­pear as blots on sheets of pa­pe­r.

Sus­pected mis­con­duct was seen “at all sci­en­tif­ic ranks in­clud­ing post­docs, stu­dents, and tenured fac­ul­ty mem­bers,” the au­thors wrote. Six­ty per­­cent of the cases in­volved fab­rica­t­ion or fal­sifica­t­ion, and 36 per­­cent pla­gia­rism “on­ly,” Ti­tus and col­leagues added.

The au­thors wrote that the prob­lem arises partly be­cause sci­en­tists are re­luc­tant to turn in cheat­ing col­leagues, and com­monly face ill con­se­quenc­es for do­ing so. They cit­ed ev­i­dence that in­sti­tu­tions of­ten en­cour­age whistle­blow­ers to drop al­lega­t­ions.

“In­sti­tu­tions must es­tab­lish the cul­ture that pro­motes safe­guards for whistle­blow­ers and es­tab­lishes ze­ro tol­er­ance both for those who com­mit mis­con­duct and for those who turn a blind eye to it,” Ti­tus and col­leagues wrote. In­sti­tu­tions may al­so want to con­sid­er au­dit­ing re­search records as part of re­newed ef­forts to root out mis­con­duct, they added.

The num­ber of cases re­ported to the Of­fice of Re­search Mis­con­duct is very low—about 24 in­ves­ti­ga­t­ions per year from in­sti­tu­tions for cases that in­volve Na­tional In­sti­tutes of Health fund­ing, Ti­tus and col­leagues wrote.

* * *

Scientific misconduct, notably the falsification of research data, may be far more prevalent than suspected, according to the authors of a new survey of more than 2,000 scientists. The findings are published in a commentary in June 19 issue of the research journal Nature. Sandra L. Titus and colleagues at the Office of Research Integrity of the U.S. Department of Health and Human Services in Rockville, Md., surveyed 2,212 scientists at 605 institutions. They found that nearly 9% believed they had witnessed potential research misconduct in the preceding three years. This suggests that as many as 2,300 observations of misconduct, 1,000 of them unreported, occur each year in the larger research community supported by the U.S. National Institutes of Health, Titus and colleagues wrote. They added that it’s unlikely such behavior is confined to the United States. Survey participants described misbehavior ranging from scientists’ changing numbers to make results look more definite than they really were, to more creative fabrications. One participant told of a colleague using Photoshop to tweak results of chemical tests that appear as blots on sheets of paper. Suspected misconduct was seen “at all scientific ranks including postdocs, students, and tenured faculty members,” the authors wrote. Sixty percent of the cases involved fabrication or falsification, and 36 percent plagiarism “only,” Titus and colleagues added. The authors wrote that the problem arises partly because scientists are reluctant to turn in cheating colleagues, and commonly face negative consequences for doing so. They cited evidence that institutions often encourage whistleblowers to drop allegations. “Institutions must establish the culture that promotes safeguards for whistleblowers and establishes zero tolerance both for those who commit misconduct and for those who turn a blind eye to it,” Titus and colleagues wrote. Institutions may also want to consider auditing research records as part of renewed efforts to root out misconduct, they added. The number of cases reported to the Office of Research Misconduct is very low—about 24 investigations per year from institutions for cases that involve National Institutes of Health funding, Titus and colleagues wrote.