Learning to Let Go While Trusting Your Data

Congratulations! You’ve just become a faculty member, a principal investigator (PI) at an institution with serious research expectations. But even as your responsibility for your research increases, new responsibilities, such as teaching, committee work, accounting, and writing grants, mean less time to spend in the laboratory. The problem has only one solution, and it’s the one every new PI must choose: Bring in graduate students, technicians, and postdocs, in a proportion that suits you and your research, and trust them to get the lab work done.

“Being a faculty member is as much about managing people as it is about conducting scientific research. And I think that’s the transition that shocks most people,” says Jo Handelsman, the chair of the department of bacteriology at the University of Wisconsin, Madison. Every transition is different–yours will be unique to your personality and work style–but every researcher has to adjust to relying on other people’s hands and minds. Making those new circumstances work means learning how to hire and train trustworthy researchers, building productive relationships with your staff, and establishing a system that ensures quality, allowing you to remain in control of a lab in which other people are doing the actual experiments.

Shift of control

The transition away from bench work is usually gradual, researchers say, but the change begins almost immediately after you accept your first PI post. Once you’re a PI, even the time you do get to spend in the lab–far less time than you’re used to spending there–is likely spent training researchers, observing, and discussing results rather than getting your gloves dirty.

That loss of hands-on control can be unnerving. “That never goes away. You just get used to it,” says Marianne Manchester, a cell biologist at the Scripps Research Institute in San Diego, California. Particularly at first, she says, it’s common to think that you could do the experiment both faster and better than the staffer who’s actually doing it. “At some point, I had to just say, ‘Okay, it’s not necessarily going to be done exactly how I might do it. But it might be done better than I might do it.’ ” Manchester’s first full-time technician was meticulous, she says, which put her at ease immediately.

After 22 years running his own research group at the University of Washington, Seattle, fisheries biologist Thomas Quinn looks back on the shift from researcher to adviser as a fairly natural progression. Although he came from his postdoc with ideas he pursued himself, advancing his field as a PI also meant bringing up the next generation of scientists. His greater mentoring and training responsibilities dovetailed naturally with the need to cede control to his students. Today, he says, “I have a few pet projects that I pursue. It’s more my mode to toss the football to students, and if they fumble it, try to pick it up myself and keep it on the field and find someone in the future.”

Of course, mistakes in the lab may mean that you do get your gloves dirty on occasion. Manchester recalls two situations in her lab that required that she redo experiments. In one case, an employee hadn’t understood how to standardize a set of animal experiments. In another, a postdoctoral researcher in a collaborating group hadn’t kept sufficient records to allow experiments to be repeated. “You just feel sick to your stomach about having to do these things over, but it’s a good (painful) learning experience,” she writes by e-mail, and she says that they’ve helped her develop better management policies.

Hiring and training: the foundation of trust

The first step in building trust is staffing your laboratory with enthusiastic, hard-working people who work well with you and with their other colleagues.

When making a new hire, Manchester and Handelsman both ask group members or other colleagues to meet with candidates. “My group was the best filter of all,” Handelsman says. In particular, group members would often see potential negatives that she hadn’t noticed. When talking with group members, one candidate spoke both negatively and sarcastically about a major professor. Because her group found the behavior unprofessional, Handelsman called the professor and found larger problems that she would have missed otherwise. Manchester always calls a prospective employee’s references for a thorough conversation about a candidate’s strengths and weaknesses in experimental skills and how they work with others.

Once you’ve hired good candidates, the next step is training them to produce work that meets your standards. Particularly in the early stages of building a new research group, it’s important to instill ethical and quality standards in a hands-on way, Handelsman says. With less experienced scientists, your example will show them how to carry out a careful experiment and avoid sloppiness. Your actions–particularly your attention to scientific rigor and attitude toward making sure that results are reproducible—are the most powerful ways to communicate your ethical standards, according to the Howard Hughes Medical Institute’s report Making the Right Moves. Talk with group members formally and informally about your standards and expectations. “I think it gets easier over time as your group matures,” she says. With guidance and structure, “those people pass on what they’ve learned to the next generation of people.”

Careful communication

As you train your staff, you’ll also be learning how to communicate with the individual members of your group. “Telling someone else what the experiment is, how it should be done, and how to analyze the data,” Manchester says, “–that’s just a skill that comes with practice.” Just because you say something doesn’t necessarily mean that the other person understood. What works for one person might not work with another. Some technicians in Manchester’s lab work well with verbal instructions. With others, she might need to write them down.

Just as you have to make sure that your directions are clear to a staff member, you can’t rely solely on a staff member’s interpretation of data. As she began building her own laboratory 10 years ago, Manchester recalls a staff member mentioning that he got a particular result, but when she reviewed the primary data, Manchester reached a different conclusion. “What I realized was that I really had to be very disciplined about going back to primary data all the time,” she says. It was up to her, she realized, to make sure her lab’s conclusions matched the primary results.

Trust but verify

That discipline–of going back to the primary data–is an essential piece of quality control. A combination of regular lab meetings, smaller group meetings, and one-on-one meetings offers opportunities for you to check data and record keeping. At the same time, meetings present scheduled times to guide staff members on their next steps and teach them how to carefully analyze their primary data.

As you examine data, it’s important to use the same common sense as a faculty member that you did as an experimentalist, says Richard Register, chair of the chemical engineering department at Princeton University. It can be frustrating when a graduate student comes in with data that don’t make sense and don’t seem reproducible, he says. “Because I’m not there when they’re generating it, I don’t necessarily know why.” At that point, he’ll often come back into the lab so that he can observe the experiment and see where the inconsistency came from.

Keep an eye on the lab informally, coming by regularly to check in on day-to-day successes and setbacks. Jeffrey Bode, an organic chemist at the University of Pennsylvania, strives for an environment in which lab members feel that they can discuss all experiments–those that worked and those that didn’t. “I’m not going to judge the outcome,” Bode says. “I’m trying not to focus so much on the success as much as the observation.”

Matthew Huber, who uses computer models to examine climate dynamics at Purdue University in West Lafayette, Indiana, has his new students work on test projects to prove their skills before they move on to more advanced problems. Because mistakes in climate modeling can be as discrete as a misplaced minus sign in thousands of lines of computer code, that initial training time–from 1 to 3 years–allows Huber to help a student learn how to find and fix mistakes.

Once a student demonstrates those abilities, Huber says “All right, it’s up to you now, and you tell me how this is going to go.” At that point, they begin to function more independently but they are “still subject to paranoid review on a regular basis.” In Huber’s line of work, as in mathematics, the only way to ensure correctness is to work carefully and verify each step. He has them work on their new climate questions in small chunks, making sure that their results are consistent with an idealized version of that piece of the puzzle. “That’s the only way that I have any faith, in fact, that anybody–including myself–is doing it correctly,” he says.

None of the researchers interviewed for this article admitted to observing dishonesty in their research groups. However, Quinn did have a case in his group for which a combination of sample mislabeling and the crash of a computer hard drive led to a perfect storm of mistakes that resulted in a student withdrawing a paper already in review. Withdrawing the paper was embarrassing for the student, he says, but “it was all a set of honest mistakes.”

Be an intellectual sounding board

If you want to have a lab that grows, you have to learn to delegate. Manchester has colleagues in their 50s who “have a really teeny lab because they can’t get to point where they can trust other people to do the work properly.” Even if you keep your gloves on part of the time, your role as an intellectual sounding board can and should take priority over your hands-on skills. After 10 years with her own group, Handelsman found it didn’t make sense to do technical experiments herself because she’d have to shut out other lab demands to concentrate. “It was more important to be available to my students and to talk with them about their research or the papers that they were reading,” she says. That intellectual availability–along with trusting relationships and solid communication–can help you keep control of your laboratory and manage any bumps along the way.

, sponsored by the Howard Hughes Medical Institute and the Burroughs Wellcome Fund, offers tips and advice about all kinds of management issues from faculty who’ve run research labs for a long time. , by Kathy Barker, Cold Spring Harbor Laboratory Press, 2002, 352 pages, $45 (ISBN 0-87969-583-8) From the National Academy of Sciences, . San Diego area institutions offer an annualin February.

Photos, top to bottom: Courtesy, Jo Handelsman. Courtesy, Thomas Quinn. Courtesy, Richard Register. Courtesy, Jeffrey Bode. Courtesy, Matthew Huber

For PhysicianScientists, ConflictofInterest Issues Are Complex

Finance’s Quant(um) Mechanics