What are elements of a successful research program?
First, some quick parameters:
Research generates new knowledge for the scientific community. There are lots of things that are worthwhile and important to education that are not research. In particular, I exclude program development from research. I also exclude class projects which teach their participants new things, but don’t generate new knowledge for the broader community.
We gauge the success of research on whether other scientists think it’s novel and well-executed. Practically speaking, I’m going to operationalize success as “publishable”. This is a horribly reductionist definition and you can feel free to disagree with me outside the bounds of this article.
Research is fun, mostly. There are parts that are tedious and boring, like transcription or washing the glassware. There are parts that are frustrating, like when your participants don’t participate or your code is broken. But, in a deep and fundamental sense, the practice of research is fun and exciting.
Research is creative, mostly. There are parts which are routine and repetitive, like checking your journal formatting or applying a developed protocol repeatedly. Unlike some other creative endeavors, it’s important to document well what you do and why. But in a deep and fundamental sense, the act of doing research is about imagining something new and creatively seeking a solution or discovery.
Research is a human endeavor. People do research collaboratively in groups. Research is more fun when you’re doing it with people you want to work with. They can be local to you or remote, but either way you should communicate often.
So, a successful research program generates new knowledge, brings together multiple people in different roles, and is usually fun for the research team.
A research program is made up of a bunch of inter-related research projects. While each project might be long or short, the program last over many years. Most research programs have a lot of projects which never succeed: maybe the data were flawed, or the theory was insufficient, or the person doing the analysis kept bad notes and graduated early, or whatever. You should seek to minimize the number of unsuccessful projects, but recognize they will exist. About half of mine are unsuccessful.
A research program that involves undergraduate students as researchers needs to have a bunch of projects which are more-or-less independent from each other (encapsulated). They need to be independent from each other so that if one of your students disappears midstream, their work isn’t holding up the rest of the team. But the projects (at least some of them) should be related thematically so that your students can form a sense of community around what they’re doing.
It’s ok if you want to pursue only one project at a time; however, I am happiest pursuing a lot of programs simultaneously. This article takes the perspective that you want to work with 2-4 (or more) undergraduate research students at once, on multiple projects which build to at least one research program.
Researchers who work with undergraduates often use the phrase “undergraduate research experience” to center the idea that the undergraduates involved are gaining experience in doing research, and to focus our attention on those undergraduates. That’s a useful framing for a lot of work, but it’s not my focus here. The focus of this article is how faculty members can design research programs which include undergraduates as meaningful and central members of the research team.
Short pieces of advice:
- Undergrads work better in pairs. No really. Solo students are more work for you.
- Practice encapsulation of research projects.
- You’re going to do most of the writing and synthesizing.
- It will (almost) always be faster to do it yourself, but you should resist this impulse.
No story here.
The Goldilocks problem: What makes a good undergrad project?
A good research project for an undergrad is an interesting one that they can do in the time they have, with the skills they have. Undergrads can be good at all kinds of research. You need to match your expectations to their skills, interests, and availability. In the past, my undergraduates have worked on projects both qualitative and quantitative; in teams distributed across three universities, in small groups locally, and solo with me; and for projects as short as one semester and as long as 2 years. My students come from diverse disciplines – physics, education, computer science, psychology, mathematics – and do diverse things after graduation. About a third of my students go on to physics graduate school and about a third teach k12. I have started research projects with first semester students and final semester seniors, though generally I tend to prefer starting students as sophomores and juniors.
That said, a good research project for an undergrad is:
- appealing to them, you, and the scientific community.
- encapsulated, with defined ending points and insulation from projects that might depend on it
- likely to succeed
- flexible to their developing interests and preliminary results
Getting a student started in research
Each research project has elements of design, data collection, data analysis, and writing. While these phases should overlap – especially the design and writing phases – they tend to generally drift in order, especially for the shorter projects that undergrads engage in.
You might think that a new undergrad should start with a design phase: reading lots of papers, planning what kinds of data to take, picking a theory, etc. Starting with the design phase is a bad plan for several reasons:
- Reading papers is hard and boring for new researchers. When undergrads are bored, particularly at the beginning stages of a project, they ghost away.
- Undergrads don’t have a lot of time or expertise, so occupying all of that time with high-minded design activities won’t leave any time left over for actually carrying out the research.
- Learning theories and methods in the abstract pales in comparison to actually using those methods and theories in practice.
In contrast, starting a research experience with data analysis is a great idea. This is how I usually start my students. I wrote up an article about how we do it for video-based projects. The advice generalizes to my quantitative, statistical, or simulation-based projects: replace “find interesting data and catalog it” with “run simple counting statistics”, for example.
If you start your students with analysis, you need to have previously gotten ethics approval, collected the data and done some preliminary sanity checks on it (is the video audible? do the spreadsheets have expected fields? etc). You need to do these sanity checks yourself, but they shouldn’t take very much of your time. After your students are conversant with analysis methods and what constitutes good data, they’re ready to go out into the field and collect some new data. This is where having a research program is beautiful: you have data “in the can” for each new project because your previous students collected it.
If your research program tends to keep students overlapping for multiple semesters, then it’s possible for new students to start with that more experienced students are already working on.
This is a good idea for the new students so that they can orient to the data corpus and the lab. It’s a good idea for the experienced students so that they can feel like experts, practice talking about their work, and learn more deeply about what they’re doing.
Some research programs which center undergraduates can do this really well, but other programs (e.g. if they happen only in the summer) cannot. It’s not necessary, but it’s great if it works for you.
How do I match projects and students?
It’s important that your students are interested in their projects and develop a sense of ownership over their analysis and results.
You could be open to whatever your students are interested in, but students don’t have a good feel for the field and sometimes their ideas aren’t novel or possible. Also, with this approach, your research won’t be well-focused and you’ll drive yourself crazy managing a lot of disparate ideas. This approach is more common in graduate schools of education.
At the other extreme, you could identify detailed projects first and offer specific ones to specific students, but that runs the risk of not being able to accommodate interested students and/or not being able to find enough students for your very specific needs. This approach is more common in graduate physics departments.
The Goldilocks way is to have a flexible menu of projects that you can offer to students. If I want to hire N more students, then I want to have between N+2 and 2N potential projects available for them to join. Sometimes this looks like adding them to existing projects and pairing them with more senior students, sometimes the projects extend prior projects from former students, and sometimes they are wholly standalone. In a practical sense, that means I am perpetually on the lookout for new projects: new potential collaborators, new data repositories I could join, new methods I might want to try out.
When a new student approaches me, I tell them about the projects that are available, pitching them in terms of research questions, methods, and who they’d be working with. They don’t need to decide immediately; they should decide within a few days, at least for what broad program they want to join. I encourage them to seek out other advisors who might meet their interests, because good pastoral care means helping students find their passion (not keeping all the students for myself). I encourage them to talk to my current students about what it’s like to work with me (I’m not the right advisor for everyone).
Early in their tenure with me, I have them write a statement of research interests, then use that statement to match them more carefully with a project within a program.
What makes a bad project and how can I avoid badness?
As you’re managing your research program with undergraduates, you need to think about how their projects interact with each other. If one student’s project depends on someone else’s ongoing research – say you’re analyzing data taken at another institution – then if the other person takes more time than anticipated, your student just sits around doing nothing, bored. You can mitigate this problem with good encapsulation:
- Each ongoing project shouldn’t depend on another contemporaneous project.
- You can adequately separate the data collection and data analysis portions.
- If some aspect of this project goes horribly wrong, everyone else should still be able to get their work done.
This doesn’t mean that each project must be completely separate intellectually from the others.
You might have multiple different kinds of analysis on the same data set, or similar analyses on related data sets. Or subsequent students might pick up projects that previous students are rotating out of. Most people have a 1:1 ratio of students to projects, but I prefer to pair my students on projects whenever possible to manage my own workload and increase their chances for success. This also allows me to maintain projects over time as students enter and leave the lab.
Some research programs require a lot of tedium: you might need to transcribe a lot of interviews, for example, before you’re ready for computational linguistics. Or you might need to collect a lot of survey data in a lot of classrooms, shuffling papers and coordinating many instructors. This work enables research, but it isn’t research directly. It’s boring and time-consuming. Don’t make your research students do it.
Instead of research students, you have three choices for these tasks:
- Hire more students as lab techs. They are undergrads, they are cheap, and their primary qualifications are that they are reliable and want money. Be very explicit with them: they’re not research students; they are doing campus jobs. Pull these students from a separate population than the population of students who want to do research with you, because the students who want to do research with you should be able to do research projects, not just jobs.
- Hire professionals. Secretarial support can manage surveys; professional transcription services can transcribe your data or caption your videos (I used to use AI transcription, but now I work with a fabulous human transcriptionist).
- Do different research. There’s plenty of great video-based research which uses the video as primary data and never transcribes. Or consider online surveys instead of paper, observations instead of interviews, or collaboration with different institutions.
Another kind of project that’s difficult for undergraduates is curriculum evaluation. It’s difficult in two ways.
- Students can’t evaluate curriculum until they’ve covered that material, so you’re automatically excluding potential research students who are too young for the content.
- Knowing what makes curricula good requires a lot of higher-level expertise than most students can develop as undergraduates.
If curriculum evaluation is what you want to do, you need to think carefully about what you’d be asking your students to do, and how they can meaningfully and creatively contribute to analysis.
Sometimes, when you get started on a project, it seems like it will be tractable and appealing to undergraduates, but as you progress in the project, it isn’t. Perhaps the analysis is too complicated, or the data are taking too long to arrive, or the research-enabling tasks are more extensive than anticipated. Perhaps your students thought they might like it, but they don’t.
It’s ok to pull the plug on a project where nobody’s having fun. You can save it for later or abandon it altogether.
Getting students finished in research
A typical undergraduate will do research with you for one summer or one academic year. Some students last longer, and some only one semester. You need to pick projects for them which have a reasonable chance of having encapsulated results within the first 60% of their estimated tenure.
Why that short? Because research sometimes takes longer than anticipated, and it’s always easy to extend analysis but rarely possible to compress it into less time.
As students are finishing, you need to make sure that their lab notebooks, generative writing, slides, posters, and analysis files are all available to you. If they’re writing papers, have them gather as much of their writing as they can into drafts. There are lots of good tools out there for documentation and collaboration; pick ones that work for you. Personally, I use github for code and overleaf or google docs for writing, but this is something I struggle with a lot.
If the research is going well and you are generating a lot of exciting new knowledge, you might want to bring in a new student to work together with your finishing student, extending the project even more. This is great for keeping a research program going. You should also find a way to articulate how the new student is doing a slightly different thing so that they can build a sense of ownership over the project.
Additional topics to consider
History
This article was first written on August 1, 2017, and last modified on March 13, 2024.