What can be done to ensure scientific integrity?

Over the past dec­ade, sci­entif­ic integ­rity has become a pub­lic policy issue. This insti­tu­tion­al­isa­tion high­lights the increas­ingly cent­ral and pre­script­ive role that sci­ence plays in our soci­et­ies and coin­cides with both a change in the scale of risks and an aware­ness of the sys­tem­ic nature of the prob­lem. We dis­cuss this with Michel Dubois, Dir­ect­or of the French Office for Sci­entif­ic Integ­rity (Office français de l’intégrité sci­en­ti­fique, OFIS).

Michel Dubois. In France, the Research Pro­gram­ming Law (Loi de Pro­gram­ma­tion de la Recher­che, LPR) of Decem­ber 2020 defined sci­entif­ic integ­rity as “the set of rules and val­ues that must gov­ern research activ­it­ies in order to guar­an­tee their hon­esty and sci­entif­ic rigour¹.” The insti­tu­tion­al­isa­tion of integ­rity in France dates back to the early 2000s. But its inclu­sion in the law is not unre­lated to the Cov­id crisis, which served as a full-scale stress test for both the func­tion­ing of the sci­entif­ic com­munity and its rela­tion­ship with soci­ety and the polit­ic­al world. Des­pite the emer­gency, the legis­lat­or wisely left it to the sci­entif­ic com­munity to define these rules and val­ues for itself.

With the phe­nomen­on of paper mills and pred­at­ory journ­als, there is a lot of talk at the moment about large-scale fraud. How­ever, accur­ately meas­ur­ing mis­con­duct remains a sig­ni­fic­ant chal­lenge. The few exist­ing sur­veys sug­gest that the clas­sic trio of fab­ric­a­tion, falsi­fic­a­tion and pla­gi­ar­ism affects only a very small frac­tion of the sci­entif­ic com­munity, less than 3 to 5% of the sci­ent­ists sur­veyed. The main issue lies else­where: the notion of “ques­tion­able prac­tices”, mean­ing con­duct that is det­ri­ment­al to the research pro­cess, con­cerns a much lar­ger pop­u­la­tion, just over 30% of research­ers sur­veyed in these same surveys.

The core of the integ­rity issue lies not only in delib­er­ate trans­gres­sion, but also in the norm­at­ive uncer­tainty sur­round­ing dif­fuse and widely shared behaviors.

These prob­lem­at­ic prac­tices are almost com­mon­place: adding an author who does not meet the require­ments to be an author, dis­reg­ard­ing res­ults that are intu­it­ively deemed irrel­ev­ant, avoid­ing present­ing data that could con­tra­dict hypo­theses, select­ively cit­ing sources to rein­force own con­clu­sions, com­mu­nic­at­ing res­ults to the gen­er­al pub­lic even before a peer-reviewed pub­lic­a­tion is avail­able, etc. 

While remain­ing vigil­ant about sci­entif­ic fraud is vital, the core of the integ­rity prob­lem lies not only in delib­er­ate trans­gres­sions, but equally in the norm­at­ive uncer­tainty sur­round­ing wide­spread and widely accep­ted beha­viours. This obser­va­tion calls for a genu­ine intel­lec­tu­al revolu­tion in the way we think about integ­rity, but also in the design of roadmaps for struc­tures such as the OFIS.

There is now a sub­stan­tial body of lit­er­at­ure on the diversity of factors asso­ci­ated with dif­fer­ent types of mis­con­duct. The study by Roje et al. pub­lished in 2023² iden­ti­fies no few­er than 70 factors that can have a neg­at­ive impact on sci­entif­ic con­duct, ran­ging from per­son­al­ity traits to work­ing envir­on­ments, skills and train­ing to research policies. 

As a soci­olo­gist, I am par­tic­u­larly attent­ive to the qual­ity and effects of “research envir­on­ments”, i.e. the con­di­tions under which sci­entif­ic work is car­ried out, which are sub­ject to numer­ous ten­sions: sta­bil­ity vs. pre­cari­ous­ness, solid­ar­ity vs. com­pet­i­tion, organ­isa­tion­al sup­port vs. aban­don­ment, clar­ity vs. opa­city, sense of justice vs. sense of injustice, etc. 

Sci­entif­ic mis­con­duct is not solely the res­ult of indi­vidu­al fail­ings. It is very often a prag­mat­ic response, an adapt­a­tion to work­ing envir­on­ments that are con­sidered prob­lem­at­ic, such as pres­sure to per­form, con­tra­dict­ory instruc­tions and norm­at­ive uncertainties.

Of course, we must con­tin­ue to edu­cate and raise aware­ness, but we can­not change prac­tices without ques­tion­ing the sys­tems that pro­duce them. This task falls to sci­ent­ists, of course, but also to the heads of research organ­isa­tions, fund­ing agen­cies, high­er edu­ca­tion insti­tu­tions, private research and devel­op­ment act­ors, sci­entif­ic pub­lish­ers, and sci­ence policy stake­hold­ers. This involves examin­ing their prac­tices and decision-mak­ing pro­cesses to identi­fy the extent to which they con­trib­ute to cre­at­ing these pres­sures, injunc­tions and uncertainties.

In my opin­ion, the OFIS should pro­mote and sup­port this organ­isa­tion­al dia­gnost­ic work. It can also count on sci­entif­ic integ­rity advisors, of whom there are now nearly 200. They are respons­ible for rais­ing aware­ness and train­ing staff, but also for invest­ig­at­ing cases of mis­con­duct. They are also in the best pos­i­tion to identi­fy con­tra­dic­tions at work in their insti­tu­tions. For example, insti­tu­tions may be very pro­act­ive in set­ting up struc­tures ded­ic­ated to integ­rity, while at the same time over­em­phas­ising quant­it­at­ive indic­at­ors in recruit­ment or staff eval­u­ation. It is up to the advisors to identi­fy these con­tra­dict­ory injunc­tions to high­light them more clearly.

I had the oppor­tun­ity to present my roadmap for the next five years to the OFIS steer­ing com­mit­tee last Novem­ber. It dis­tin­guishes between recur­ring actions, such as pro­du­cing a bian­nu­al report on integ­rity breaches, and tar­geted actions. 

OFIS intends to fully play its role as an obser­vat­ory, pro­du­cing an in-depth assess­ment of the situ­ation in France and help­ing to identi­fy the neces­sary changes

One of the tar­geted actions con­cerns private research and the trans­fer­ab­il­ity of integ­rity issues to R&D. Con­trary to pop­u­lar belief, the intro­duc­tion of integ­rity into law con­cerns all research activ­it­ies. How­ever, today, most research efforts are car­ried out by the private sec­tor. The ques­tion is there­fore how integ­rity can be made vis­ible and assessed in this sec­tor, in line with the government’s fin­an­cial com­mit­ment, par­tic­u­larly through research tax credits.

A second action con­cerns sci­entif­ic pub­lish­ing. Bey­ond announce­ments about the cre­ation of ded­ic­ated ser­vices, what are major inter­na­tion­al groups and nation­al pub­lish­ers actu­ally doing to address integ­rity issues? How do they identi­fy prob­lem­at­ic journ­als? How do they handle requests for cor­rec­tions or retrac­tions? How do they train their teams? What tools do they use? OFIS is com­mit­ted to ful­filling its role as an observing body by pro­du­cing an in-depth assess­ment of the situ­ation in France and help­ing to identi­fy the neces­sary changes.

The Amer­ic­an Restor­ing Gold Stand­ard Sci­ence decree con­trasts with what has been done in France: the rules are not left to the dis­cre­tion of the sci­entif­ic com­munity but are dic­tated by a pres­id­en­tial decree. This is a form of polit­ic­al inter­fer­ence, adding to all those recor­ded since the begin­ning of the year in the United States. 

Fur­ther­more, while most of the rules appear reas­on­able at first glance, their accu­mu­la­tion is prob­lem­at­ic. Many areas of research will not be able to meet all the cri­ter­ia, par­tic­u­larly because not all of them have the same intrins­ic rela­tion­ship to repro­du­cib­il­ity or exper­i­ment­al research. By mak­ing integ­rity an unat­tain­able ideal, Restor­ing Gold Stand­ard Sci­ence is part of an effort to dis­cred­it and dis­qual­i­fy a large part of the aca­dem­ic community.

Finally, it is import­ant to high­light the core of the ini­ti­at­ive, which Don­ald Trump’s sci­ence advisor, Michael Krat­si­os, calls “informed dis­sent”, or “a dis­trust of blind con­sensus and a valu­ing of informed dis­agree­ment”. To be con­sidered hon­est accord­ing to this stand­ard, one must there­fore be able to give equal weight to con­tra­dict­ory opin­ions. Applied to cli­mate research, this means that half of the argu­ments should be giv­en to those who deny the effect of human activ­it­ies on the cli­mate. Integ­rity then becomes a rhet­or­ic­al tool: under the guise of bal­ance and open­ness, it serves to rein­tro­duce mar­gin­al or dis­cred­ited the­or­ies into the sci­entif­ic arena, not for their sci­entif­ic value, but for their polit­ic­al usefulness.

Interview by Anne Orliac