Now is the time of year when we work with students on designing summer research projects. How do you decide exactly what their project is, and how the experimental design is structured? This is something I struggle with.
Image by T.McGlynn
In theory, quality mentorship (involving time, patience and skill) can lead a student towards working very independently and still have a successful project. Oftentimes, though, the time constraints involved in a summer project don’t allow for a comprehensive mentoring scheme that facilitates a high level of student independence. Should the goal of a student research project be training of an independently-thinking scientist or the production of publishable research? I think you can have both, but when push comes to shove, which way do you have to lean? I’ve written about this already. (Shorter: without the pubs, my lab would run out of dough and then no students would have any experiences. As is said, your mileage may vary.)
A well-designed project will require a familiarity with prior literature, experimental design, relevant statistical approaches and the ability to anticipate the objections that reviewers will have once the final product goes out for review. Undergraduates are typically lacking in most, if not all, of these traits. Sometimes you just gotta tell the student what will work and what will not, and what is important to the scientific community and what is not. And sometimes you can’t send the student home to read fifteen papers before reconsidering a certain technique or hypothesis.
When students in the lab are particularly excited about a project beyond my mentorable expertise, or beyond the realm of publishability, I don’t hesitate to advise a new course. I let them know what I hope students get out a summer research experience:
- a diverse social network of biologists from many subfields and universities
-
experience designing and running an experiment
-
a pub
All three of those things take different kinds of effort, but all three are within reach, and I make decisions with an effort to maximize these three things for the students. Which means that, what happens in my lab inhabits the right side of the continuum, sometimes on the edge of the ‘zone of no mentorship’ if I take on too many students.
You might notice one thing is missing from my list: conceive an experiment and develop the hypotheses being tested.
Students can do that in grad school if they want. Or in the lab of a different PI. I would rather have a students design experiments on hypotheses connected to my lab that I am confident can be converted into papers, rather than work on an experiments of the students’ own personal interest. (Most of my students become enamored of their experimental subnets pretty quickly, though.)
This approach is in the interest of myself to maintain a productive lab, but I also think that being handed a menu of hypotheses instead of a blank slate is also in the long-term interest of most students. I’m not keen on mentoring a gaggle of students who design their own projects when these projects are only for their edification, and not for sharing with the scientific community. That kind of thing is wonderful for the curriculum, but not for my research lab.
Other people have other approaches, and that is a Good Thing. We need many kinds of PIs, including those that give students so much latitude that they will have an opportunity to learn from failure. And also those that take on 1-2 students at a time and work with them very carefully. I like the idea of thinking about my approach to avoid falling into a default mode of mentorship. Does this scheme make sense, and if it does, where do you fit in and how have you made your choices? I would imagine the nature of your institution and the nature of your subfield — and how much funding is available — structures these choices.
Small Pond Science 
My department has just shifted from a final year project model where the students designed their own studies in 2nd year to one where we we can offer specific projects for them to run (my preferred model). I think pedagogically the latter design is better; the projects they come up with are generally fairly pointless because they’re UG students and they don’t know the literature in depth. They end up running pointless studies that never work and I don’t think they learn much. I’ve always previously had students working on my own research and they spend time doing real science and always get a kick out of it – and sometimes even a publication.
UG students are not up to the challenge of designing real science experiments just because it takes depth and breadth of knowledge of a field that you only get with time. They can still have a valuable experience in which they learn a lot without having to come up with the study themselves
Although I have no firsthand experience (yet, but hopefully soon!), my idea is to guide students through an initial project. Then if they meet expectations and complete the project, I would be willing to allow them much more latitude in a subsequent project.
I think there’s immense value for undergrads in simply being part of scientific research before being given a chance to influence how that science occurs. My first few research experiences as an undergrad, in both lab and field, involved doing exactly what I was told to do, but being encouraged to ask questions and having mentors who took the time to explain what was going on was enough to feel like I was participating. During my first field experience, I was told that I could conduct an independent project only if I came up with an idea that was good enough. I didn’t end up conducting an independent project that summer, but bounced many many ideas off my mentor–that was the perfect way in which to begin to think like a scientist and receive excellent mentorship, while still collecting publishable data. Of course, actually following through with any of those ideas would have brought other challenges with it, but as you say, the experience of dealing with those challenges need not occur in an undergrad lab.