More doctoral students emerge from small liberal arts colleges than from the undergraduate populace of research institutions.

This is a point of pride held by liberal arts colleges, that market themselves as the best place to go if you want to become a scientist receiving a Ph.D. from a big-name research institution. Demographically, they’re correct.

Are small teaching schools better equipped to train undergraduate researchers better than big research institutions? I don’t think so.

In practice, liberal arts schools are far better at producing high quality researchers, but it’s not because of any inherent property of liberal arts schools. Some could argue that the curriculum itself might matter – that’s a discussion for another time – I’ll spend the rest of this post thinking about the single reason that people identify about what makes liberal arts schools a special place for budding researchers.

Here is the standard reasoning: Teaching schools provide students with the opportunity to have close professional interactions with their professors. Students in labs in small teaching institutions benefit from direct mentorship from the PI, which will more likely result in a higher quality research experience, better insights into how to do research, and greater opportunities to own their own research projects, enabling them to present at major venues and eventual publication as undergraduates.

How true is the preceding paragraph? It’s a straight-up fact that students at small teaching campuses are more likely to do more original research of their own working with their PI. And, if an undergraduate arbitrarily selects a research lab to join, then they’d probably end up getting a better experience at a teaching institution.

But, though this trend is real, research institutions have tremendous potential for training undergraduates. Without providing any additional resources, any research institution can be a top-notch training ground for undergraduates. After all, there is nothing inherent about teaching institutions that makes them better at training researchers.

There is nothing magical about having the PI as your direct mentor that will make you a better researcher and help you get into a better grad school. Looking closely at what supposedly makes a teaching institution better for training undergraduate researchers – close involvement with the PI – I see a massive handicap.

All of the literature on research mentorship says that the relationship is most successful when the mentor is just a little above the mentee in research experience. Even though the PI is a better academic expert and has mentored more, the Ph.D. student and the postdoc are in a position to be more effective as mentors.

The best mentoring arrangement is a multi-level team, in which the early undergrad works with a senior undergraduate, who then works with a Ph.D. student, who works with a postdoc as well as the PI. The PI knows everyone personally, and spends some time with the undergrads, but the graduate students are the better formal mentors. (A colleague of mine at a research institution recently tried to kick one of her own undergrad researchers out of the lab, because she didn’t recognize her. That’s not good.)

I suppose a young PI can connect more easily with students, but as we get older, then the nature of the relationship evolves. Add on a few years, and the gap between the PI and the student grows. Even if the PI is affable, and might truly understand the perspectives and thoughts of the students, it would be silly to ignore the fact that our students can’t relate to us and that we can’t relate to our students, even if we were once in their position. No matter how much time I spend with my students, now matter how similar our backgrounds are, the fact of who I am limits my ability to serve as a model. I can do all the right things in the mentoring process, but if a grad student did all of the right things, it would be even better. (And for my students from underrepresented groups, having a mentor from the same group is particularly powerful.)

I really like most of my students. I enjoy their company, and over time some have become good friends of mine. But, let’s face it, there’s a big gap. I’m older, have a kid and am married, and we don’t have that many overlapping interests. While I try hard to be transparent, I recognize that I seem like an enigma in a bunch of ways. (For example, earlier this summer one of my students was totally surprised that I use torrents to watch a couple TV shows. He just thought this was outside my realm for some reason.) I didn’t go to grad school in the middle ages, but things have changed since I’ve been there, and this is true for anybody who is at least halfway to tenure. If I try to discuss grad school with my students, I’m not nearly as credible or powerful as the same information coming from a current graduate student.

My position of authority makes me a less influential mentor.

I don’t want to overgeneralize from my experience, but I doubt that I’m alone.

You might be thinking, “Do your students really have to relate to their mentor to have an excellent research experience, and move their career to the next level?” Not necessarily. But I think it really helps. Especially for students who aren’t able to visualize themselves as capable of excelling in graduate school, a proximate model is an essential part of the mentoring process. Having seen my undergrads interact with doctoral students on a regular basis, it’s clear to me that without this kind of opportunity, that my students would missing out, big time.

Having a student know that the path has been blazed in front of them by other students, like them, matters. If students see other students throw themselves into research with great passion, they are more likely to allow themselves to get that excited. Of course, the same was true for me. But now, I’m an old bald dude with kids, and I get really excited about research, but in a different way. I can’t serve as a model for my students, even if I tried.

While grad students might not have the same authority and skill set as the PI, they can offer things that the PI can’t. This is exactly why a multi-level mentoring scheme is the way to go. The PI can choose to become involved when it is wise, and step back and focus on other things when the grad student has things under control.

Research institutions have grad students, but this doesn’t mean that they deliver great research experiences for undergraduates. While the personnel are available for a multi-level mentoring system, in many labs the system is nonfunctional because undergrads are often treated as serfs. I know many R1 labs that that are exceptional for undergraduates who work with graduate student mentors. However, I’m aware of far more labs that do not focus on making sure that undergraduates have their own research experience and are able to focus on building their own academic identity. In general, undergraduates in research institutions that receive their own project (as a piece of their mentor’s work) are the exception rather than the norm.

As for the mass production of Ph.D. students from small liberal arts colleges, I would bet that the outcome is a done deal even before the students enroll in college. The social and economic class that produces doctoral students is the same caste that is able to send students to fancy private liberal arts schools. Yes, there are scholarships and financial aid. But even if you look at small liberal arts colleges that heavily emphasize economic and ethnic diversity, they simply can’t match the diversity of the nation’s populace because, simply, most people can’t afford it. As long as the average cost of a liberal arts college is more than average cost of research universities, of course a higher proportion of doctoral students will emerge from liberal arts colleges.

How do I get my own students a multi-level mentored experience? Well, I don’t have that happen inside my lab on a day-to-day basis. I may have Master’s students around, but I usually have undergrads that are more seasoned than my grad students. That experience helps, but the way I really bring in graduate student and postdoc mentors is by having my students conduct their research in a hub of collaborative activity during the summer at a field station: La Selva Biological Station, in Costa Rica. There, my students build strong relationships with scientists from all over with different levels of experience, and these bonds typically stay tight after they leave the field station. Sometimes their projects become collaborations with grad students and postdocs at other institutions. I like that a lot, for a bunch of reasons.

If multi-level mentoring is important for the success of undergraduates, then what does this mean for you?

If you’re in a research institution: Postdocs and grad students should become genuine mentors and give undergraduates the time and resources to have their own students, and supervise them properly.  Faculty at research institutions should support their lab members, not just in the process of research but also in the process of mentorship. Don’t exploit undergraduates as trained monkeys. If you want someone to be an unthinking data-generating machine, then hire a technician. If you take an undergraduate to do “research,” then do actual research with them. Your own research agenda is easily split up into several smaller questions. Hand one of those questions to your undergraduate researcher, and learn how to mentor them. Give them the same support that you expect to receive from your own research advisor. Yeah, it’s not easy, but it will pay off for both of you in the long run.

If you’re at a teaching institution: Seek routes for multi-level mentoring in the lab. At a minimum, the undergraduates with more than two years of experience in the lab should be given the chance to actively supervise new students. Ideally, you can develop relationships with colleagues in other institutions with graduate students and postdocs. Find a way for your undergrads to become friends with doctoral students. I don’t know how to make this happen, and it varies with institutional context and geography, but from where I sit, it’s an ingredient that really promotes success. (For starters, you can bring students to smaller national meetings where they can build relationships with the students of your colleagues.)

I don’t have a big specific solution to the problem, but recognizing the fact that we as faculty are inherently flawed mentors is a start, and recognizing that the lack of graduate students at teaching institutions isn’t a strength, but a weakness, of the mentorship process.