2

Scientific Leadership for Ethical and Responsible Research

The first panel, moderated by planning committee member Susan Wolf, University of Minnesota, provided context for how the responsibilities and expectations for research leaders are changing with regard to responsible conduct of research (RCR), as well as the challenges these changes pose to research organizations and the broader research enterprise. Wolf noted the committee wanted to begin by asking scientific leaders for their views on ethical and responsible research and why it is important in directing laboratories, centers, and larger cross-institutional collaborations. The literature on team science, while important to the workshop themes, pays comparatively little attention to these issues and instead looks more at success in achieving scientific goals, she commented. But, she continued, a team that strives to meet its goals is headed for trouble if it does not pay attention to reproducibility and replicability, the duties to humans and animals involved in the research, ethical attribution, and respect and inclusion. The question of how scientific leaders can ensure ethics and RCR becomes more challenging with the rise of larger teams and more complex research networks that span multiple labs, institutions, and disciplines, she added, as they become “teams of teams” or “multiteam systems.” She also called attention to NetEthics, a project funded by the National Science Foundation (NSF) to study the challenges involved in complex engineering networks, of which she is principal investigator (PI).¹

___________________

¹ For more information about NetEthics, see https://consortium.umn.edu/grants/netethics-building-tools-training-advance-responsible-conduct-complex-research-networkspioneering/7226.

One contextual note: while the title of the workshop and this proceedings highlights the individual lab, the workshop also addressed collaborations ranging from labs to large-scale efforts crossing departments, countries, and scientific disciplines. Furthermore, the use of lab is meant to denote a working collaboration of researchers as opposed to a physical laboratory.

The panel included Maria Zuber, Massachusetts Institute of Technology (MIT); Michael Witherell, Lawrence Berkeley National Laboratory (LBNL); and Gerald Goodwin, U.S. Army Research Institute (ARI) for the Behavioral and Social Sciences.

PERSPECTIVES AS A RESEARCHER AND RESEARCH ADMINISTRATOR

Zuber began by stressing the importance of responsible scientific leadership because trust in science and technology is paramount to the acceptance of scientific findings. She noted she shared perspectives as a researcher who leads large teams and as a senior research officer for MIT.

To illustrate the key element of trust in research, Zuber drew from her own research, which uses observations from spacecraft, with gravity one of the quantities measured. The software has been developed over decades by teams of people under configuration control, and the observations must be corrected to a very precise level to produce the gravity field of a planet. Zuber explained that for decades, separate software systems at the Goddard Space Flight Center (GSFC) and at the Jet Propulsion Laboratory (JPL) have collected the data with independent analyses in both systems. Over the course of years, proposal and panel reviews commented that they saw this as duplicative work, but her group refused to stop making two independent assessments. She recounted that the agreement in the outputs from the two systems led to her selection as the PI for GRAIL (Gravity Recovery and Interior Laboratory), National Aeronautics and Space Administration’s (NASA’s) $500 million dedicated mission to map the moon. Selection was possible was because of the trust in the analysis and observations over decades. It took years of perseverance, and it would have been easy to back down from the two-system approach, she reflected.

As a research officer charged with maintaining integrity at the university level, she sees many accusations about researchers’ conduct, but very few of them are found to be willful misconduct. Most common are authorship disputes that are usually resolved locally. Other problems stem

from mistakes, sloppiness, or bad record keeping. What is clear from years of adjudicating these cases, she said, is that the tone at the top matters. When the PI is careful, keeps good records, and sets a good tone for safety, inclusion, and rigor, these practices are reflected throughout the lab. In contrast, when the PI is less attentive, integrity issues show up repeatedly. She noted the existence of certain labs where problems arise more frequently.

CARE FOR PEOPLE, RESEARCH, AND RESOURCES

As director of LBNL, Witherell said two of his primary roles are to formulate strategic research initiatives and manage operations. When he started his term in 2016, he saw an opportunity to develop a culture of ethics and responsibility across the lab’s twenty-one research and ten operational divisions. He wrote down what he saw as main elements of the desired culture, which he and a small team drew upon to create a statement on stewardship for the lab.² The statement is one that everyone in leadership positions at the lab could accept as central to their role and could be used to recruit employees as members of a community. He regularly communicates in messages to the lab that “as leaders and citizens of the lab, we are entrusted to care for the people, research, and resources that make our mission possible.” He expanded on what he meant by stewardship of people, research, and resources.

Taking care of people encompasses sustaining a safety culture, providing psychological and physical safety, committing to a welcoming and inclusive workplace environment in which everyone feels they can do their best work and contribute to collective success, and recruiting and retaining a diverse workforce. The second component of stewardship culture involves safeguarding the excellence and integrity of the research, somewhat akin to the role of a university’s office of research, to encompass ethics, integrity, security, and the management of conflicts of interest. He stressed this goes beyond mere compliance training modules to acting as a responsible citizen of the lab and the research community. He noted that postdocs and graduate students look to their PIs for guidance, and they need to hear this message consistently throughout the organization. The resource component involves stewarding facilities and infrastructure, for example, in managing biological and chemical waste, so that the lab is in the best condition possible to continue accomplishing its mission well into the future. In other

___________________

² For more information, see https://stewardship.lbl.gov/.

contexts, such as academic labs where students, postdocs, and early-career faculty are being supervised, security and stewardship might also include attention and care on the part of leaders to the funding support and career prospects of those who they supervise.

Witherell shared the lab’s five core values: team science, service, trust, innovation, and respect. Taking team science as an example, Witherell said it is important to not set up a hierarchy with two classes of employees: scientists and non-scientists. Instead, the goal is to operate as one laboratory with one culture. One of the largest initiatives has been an annual leadership development program. Now in its third year, the Director’s Stewardship Summit brings together 150 emerging and established leaders in both research and non-research positions. It includes an immersion workshop, follow-up sessions, mentoring and coaching, and small “home” groups. The goal is for participants to have the skills and confidence to function as responsible leaders to help their teams feel motivated, inspired, and deeply integrated into the broader lab community. He commented that he initially thought of the program as an experiment but, based on surveys and other inputs, is convinced that it is working well for leaders and the lab.

To drive decisions, the core values are now embedded into safety and performance management systems, the employee recognition system, and divisions’ strategic plans. He encourages early-career researchers writing their first proposals to consider how they will act on these values and communicate that information in their proposals. Managers are encouraged to talk to job candidates with this message. He commented that an engineer who received a strong offer from a tech company chose to stay at the lab because its values match his own. As Witherell concluded, “That is what I want our people to feel about the lab.”

CONSIDERATIONS IN A SMALLER RESEARCH SETTING

Goodwin noted that, compared with LBNL, the ARI is a small lab, but the issues are similar and, in some ways, more complex because of the focus on the social and behavioral science needs within the Army. Basic and applied research are conducted to support change and transformation of management personnel issues, such as selection and placement of individuals, officer training, and leadership development. Thus, he said, ARI deals with a lot of personnel data, and the findings affect people’s lives in real ways. This creates a layer of responsibility for the researchers in executing their job and fulfilling their mission.

ARI, he explained, incorporates management approaches similar to those used by NSF, where staff scientists work as program managers, as well as approaches similar to those employed by national labs, where researchers serve as PIs in guiding research teams. He noted that some staff researchers recently completed graduate school and are serving as co-PIs with more senior researchers external to the lab. That power dynamic creates significant challenges, and one of his roles is to prepare these researchers to meet these challenges with a goal to sustain the quality of their research. It is important to ensure that both researchers and the Army leadership as recipients of the research can trust and act on the findings. A core aspect of scientific integrity in this context is to take care of each other and, most importantly, to take care of the soldiers who are the research participants. Often very young, the soldiers are trained to follow orders and if the research is not executed carefully, they can be taken advantage of. Goodwin also discussed reproducibility and risk management. While some inherent risk is present, a systematic approach is needed to manage the risks as research moves from basic to applied. It is important to ensure that inappropriate risks are not being taken that could undermine the applicability or introduce questions about whether the findings can be trusted.

A key challenge is to take care of Army personnel data, which brings up issues related to individual privacy and national security. Exposing individuals’ data could put them in jeopardy. ARI researchers do not normally work in a classified environment, although it is sometimes controlled, and they must be mindful when the research is rising to a level that might need to be classified. Issues that might then arise include how to address reproducibility. Human research protection is core and “part of the air we breathe,” Goodwin continued. While everyone in the social and behavioral sciences requires training, he said the training is more rigorous for ARI internal and external scientists to understand the nature of the research and population. They go through the elements of population-protection training that are normally used for those who conduct research with prisoners and children because of the fundamental risk to soldiers.

The past 10 to 15 years have seen a growth and emphasis in interdisciplinary research at ARI. Most researchers are psychologists, but they need to work in teams with economists, data scientists, and others, and they must understand the differences in approaches and methods across disciplines. Being able to have a conversation about what is rigorous, reproducible research in this context is an ongoing challenge. Mechanisms

to address this include bringing in independent experts to serve as advisors and coaches to provide additional rigor.

DISCUSSION

Wolf began by asking about how aspiring scientific leaders can learn to lead and to foster research ethics and integrity, especially those who come out of labs that did not offer good examples. In addition to participating in training and other programs, Zuber urged that postdocs, graduate students, and other early-career researchers who feel they are in a lab where the highest standards are not followed should seek out other faculty members for additional mentoring. If they feel actions in their lab are egregious, they can report these actions. She cautioned about the slippery slope between a violation of conduct and standards of integrity, acknowledging that different PIs have different views. She counsels students that it takes years to develop a good reputation but not much time to lose it. If their name is on any lab output (i.e., papers), they must feel comfortable with it and should talk to another faculty member if they have concerns.

Witherell reflected he received no training on the culture of ethics earlier in his career. Traditionally, mentors and advisors were expected to provide this grounding, but, as Wolf and Zuber mentioned, sometimes this is not enough. Twenty years ago, he did not focus on establishing a strong ethics culture when he led other scientific institutions, but he commented that the workforce, society, and he himself have changed. Most younger researchers expect this training. Mentorship is critical throughout one’s trajectory in science, he added. At LBNL, everyone has access to a mentor who is not their supervisor.

Goodwin said the ARI approach to mentorship is very proactive and brings forth a variety of RCR issues. He meets monthly with team leaders in which they focus on one topic or example. He is outside their supervisory chain, and the conversations are open and wide-ranging, from very pragmatic to research focused. He also serves as a higher-order mentor to all on staff. He noted he had to seek out and craft his own relationships, but others may not be able to do this. He stressed that care for people and scientific rigor can be crafted to co-exist.

C. K. Gunsalus, University of Illinois, Urbana-Champaign, reiterated Goodwin’s earlier point about the power dynamic when more junior scientists serve as co-PIs with those who are more senior. Goodwin said that he has developed an approach that he calls supervised empowerment. They are

not “put on an island by themselves,” and he looks for ways to bolster, not undermine, their authority. As an example, two prominent researchers with an antagonistic relationship had to work together, and neither listened to the junior researcher managing the project. Goodwin stressed to the senior researchers that he supported the junior researcher in managing the relationship, then stepped back. This approach allowed her to retain authority without him being directly involved. He says he does not over-emphasize power relationships with junior researchers but does discuss responsible stewardship of funding, broad federal science policy, and different methods and approaches to guide them in responsible oversight of the research. Witherell commented that as people rise in an organization, their words carry more weight, and their opinions are taken more seriously. People must be coached that “as they rise higher in an organization, they must lower their voice,” he said. Zuber remarked that in the past, individuals facing difficulties would endure their challenges without expressing their concerns openly. Now, organizations offer mechanisms and processes to get help or support from other faculty members, department heads, or others in the organization. In an ideal situation, resolving an issue begins with informal discussion between a PI and a peer, so as not to put the junior researcher at risk because of the power dynamic discussed.

Michael O’Rourke, Michigan State University, commented that culture is an important but amorphous concept. He asked the presenters how they maintain their awareness of their organization’s culture and when it might be changing in ways that are not conducive to research integrity. Witherell recalled that he went from being a faculty member to being director of a large national laboratory with no training in these issues at all. “We need to do better, and that is what we are trying to do,” he asserted. That said, measuring culture is complicated. At LBNL, division directors are at the point of transition between the senior leadership and the rest of the lab. He expects them, as part of their jobs, to steward the people they manage, understand the culture, and communicate. Other mechanisms are a survey to serve as a baseline, as well as feedback received during the stewardship summit. Goodwin agreed it is difficult to measure culture. It is often only noticed when violations occur. The challenge is that it is easy to have a culture that does not prioritize integrity but not be able to really see that. He suggested bringing in someone from the outside who is not familiar with the culture but knows the organization’s goals. It is also essential to be sensitive to violations and react quickly, not with overwhelming force but with persistent correction.

Wolf observed that leadership occurs at different scales, often across labs or institutions, and through teams of teams. Witherell agreed that PI leaders of groups, programs, divisions, and larger groupings have different challenges, but they must have a common language, while recognizing that his lab is a more managed environment than a university. When dealing with multi-institutional efforts, Zuber said it is critical to be available for interactions to build a culture of trust. Everyone must feel comfortable bringing a problem forward. She related that when she ran the GRAIL mission, a coffee mug was presented each month to the person who brought the “best problem” forward. In addition, the chief engineer convened a session in which people could freely discuss their worries and concerns, in which neither she nor the program manager were present. The only way to get in trouble was not to bring a problem forward, she added.

A participant commented about trade-offs when everything cannot be optimized, for example, when people have life constraints, but the team is pushing against a deadline. Goodwin agreed this is a challenge for scientific leaders. In team science conducted at ARI, people with different levels of authority are embedded throughout the team, and depth is baked in. Rarely is there an individual who is a single point of failure related to an impending deadline. If so, they consider whether they can move or otherwise affect the deadline. In some cases, he conceded, the leader has to make hard decisions. He also noted the real-world demands on the soldiers and other people from whom they collect data. The soldiers have limited time, and rigorous research has to fit within their availability. He returned to culture: if people are doing the right thing, more formalized policies are not needed, but if there is variability, then practices that reinforce good decision-making around ethical issues must be instituted.

Fear of retaliation is a reality for early-career researchers (ECRs), a participant noted, to which Zuber added that even the perception of retaliation makes it a reality. As an example of one approach to address this specific concern, she highlighted an MIT process for instances in which something is not working out for an ECR in a lab for whatever reason, including different styles of working, and MIT provides bridge funding for the researcher to move to another lab. This program makes moving less problematic. These are not easy decisions for an ECR to make, and they may mean years of lost work. Further solutions are needed to mitigate the potential negative consequences. Witherell agreed that sometimes there is not a good match but also pointed to the more egregious situation of bullying. As part of research integrity, bullying must be addressed and dealt with, he said. When asked

to comment on this situation in the context of federally funded research, workshop participant Rhonda Davis (Head of the Office of Equity and Civil Rights at NSF) noted that NSF can look at these situations from a Title IX perspective. She pointed out if a person files a Title IX complaint and feels they did not get due process or were retaliated against, they can go to the federal agency funding the research and the agency will approach the university. France Córdova, Science Philanthropy Alliance, followed up that sometimes fear of retaliation prevents people from speaking up. In those cases, Davis clarified, they can come to NSF, which protects them while working with the university to address the issue. When the university knows that a funding agency is involved, this tends to up the ante, she commented. She added that NSF does not go any further than the individual would like. Lastly, if needed, NSF can go to a university to conduct Title IX compliance with no specific name or case.

When asked how the National Academies could help strengthen scientific leadership, Zuber said that conducting the workshop has already elevated the topic. Many organizations are putting programs in place, but she noted there would be value in National Academies–recommended best practices, as happened with the National Academies study on sexual harassment (NASEM, 2018). Witherell commented that in planning the stewardship summit at LBNL,³ they did not find many relevant materials. A community of practice would be helpful, and he noted that the National Academies have the convening power to do so. Goodwin pointed to the research around organizational behavior and leadership, which he said could help structure ideas and best practices to create a culture focused on responsible research. The National Academies could draw on this research to structure a set of best practices and ideas for lab directors to use with other resources, he suggested.

REFERENCE

___________________

³ For more information, see https://sites.google.com/lbl.gov/stewardship-summit/.

This page intentionally left blank.