I am a fan of knowledge transfer between peers, teaching what I know to others and learning back from them. At University I frequently helped my fellow course mates with the material, so I was very interested in formally mentoring students when I started my PhD. Luckily my supervisor, who is really talented at this, agreed to let me help him with supervising some Master’s theses. In this article, also published as a Nature Career Column, I present eight lessons that I learned by watching him at work and trying on my own.

I supervised three Master’s students in the past year. One of them was quite good and independent, did not need a lot of guidance and could take care of most things on his own, while the other two required a fair amount of help from us, one of them even coming close to not graduating successfully. Dealing with the difficult situations is when I learned the most important lessons, but regardless of the ability of the students a common thread soon appeared.

But first, here’s a brief digression on how that happened. While I was writing a draft for this blog, I noticed an interesting article on Nature’s newsletter. While I was reading it, I felt its style was quite similar to what I usually aim for in this blog: use headlines to highlight the important points, and elaborate on those with a few paragraphs. I then noticed the author of that column was a PhD student, and I thought: “how comes she has an article there? Why can she do that? Can I do that?”. I quickly found how to do it, finished the draft and sent it to them, and, after eight rounds of review in the course of two months, the article was finally up! The editor was very responsive and we could iterate quickly on the manuscript, and the quality of the writing is so much better than what I had originally sent in. On the other hand, I sometimes felt the message was being warped a bit too much. After the editing process was finished I had to agree to an Embargo Period of six months during which Nature had the exclusive right of publishing the final version on their website. As those six months are now over, I am finally allowed to publish the final version here, too. Enjoy!

This is a post-peer-review, pre-copyedit version of an article published in Nature Career Column. The final authenticated version is available online at: https://doi.org/10.1038/d41586-021-02028-1.

The lessons I learnt supervising master’s students for the first time

PhD student Emilio Dorigatti supported three junior colleagues during their degrees.

I started my PhD wanting to improve not only my scientific abilities, but also ‘soft skills’ such as communication, mentoring and project management. To this end, I joined as many social academic activities as I could find, including journal clubs, seminars, teaching assistance, hackathons, presentations and collaborations.

I am a bioinformatics PhD student at the Munich School for Data Science in Germany, jointly supervised by Bernd Bischl at the Ludwig Maximilian University Munich and Benjamin Schubert at the Helmholtz Centre Munich, the German Research Center for Environmental Health. When I went to them asking to gain some experience in communication and mentoring soft skills, they suggested that I co-supervise three of Benjamin’s master’s students.

At first, I felt out of my depth, so I simply sat in on their meetings and listened. After a few months, I began offering technical advice on programming. I then started proposing new analyses and contributions. Eventually I became comfortable enough to propose a new master’s project based on part of my PhD research; Benjamin and I are now interviewing candidates.

I gained a great deal from this experience and I am grateful to both of my supervisors for supporting me, as well as to the students for staying motivated, determined and friendly throughout. Here are some of the things I learnt about how to ensure smooth collaboration and a happy outcome for all of us.

Draft a project plan

With Benjamin and Bernd, I put together a project plan for each of the master’s students. Drafting a two-page plan that ended up resembling an extended abstract for a conference forced us to consider each project in detail and helped to ensure that it was feasible for a student to carry out in their last semester of study.

If you’re a PhD student supervising others, sit down with your own supervisor and agree on your respective responsibilities as part of the project plan. At first, you might want your supervisor to follow you closely to help keep the project on the right path, but as you gain more experience and trust, you might request more autonomy and independence.

Use the project plan to advertise the position and find a suitable student: share it online on the group’s website or on Twitter, as well as on the job board at your department. Advertise it to your students if you are teaching a related topic, and sit back and wait for applicants.

We structured the plans to include a general introduction to the research subject as well as a few key publications. We described the gap in the literature that the project aimed to close, with the proposed methodology and a breakdown of four or five tasks to be achieved during the project. My supervisors and I also agreed on and included specific qualifications that candidates should have, and formalities such as contact information, starting dates and whether a publication was expected at the end.

Benjamin and I decided to propose publishable projects, sometimes as part of a larger paper. We always list the student as one of the authors.

Meet your student regularly

I found that I met with most students for less than an hour per week, but some might require more attention. Most of the time, Benjamin joined the meeting, too. We started with the students summarizing what they had done the previous week and any issues they had encountered. We then had a discussion and brainstorming session, and agreed on possible next steps. I learnt that I do not need to solve all the student’s problems (it is their thesis, after all). Instead, Benjamin and I tried to focus on suggesting a couple of things they could try out. At the end of the meeting, we made sure it was clear what was expected for the next week.

We used the first few weeks to get the students up to speed with the topic, encouraging them to read publications listed in the plan, and a few others, to familiarize themselves with the specific methods that they would be working with. We also addressed administrative matters such as making sure that the students had accounts to access computational resources: networks, e-mail, Wi-Fi, private GitHub repositories and so on.

Encourage regular writing

Good writing takes time, especially for students who are not used to it, or who are writing in a foreign language. It is important to encourage them to write regularly, and to keep detailed notes of what should be included in the manuscript, to avoid missing key details later on. We tried to remind our students frequently how the manuscript should be structured, what chapters should be included, how long each should be, what writing style was expected, what template to use, and other specifics. We used our meetings to provide continuous feedback on the manuscript.

The first two to four weeks of the project are a good time to start writing the first chapters, including an introduction to the topic and the background knowledge. We suggested allocating the last three or four weeks to writing the remaining chapters — results and conclusions — ensuring that the manuscript forms a coherent whole, and preparing and rehearsing the presentation for the oral examination.

Probe for correct understanding

In our weekly meetings, or at other times when I was teaching, I quickly realized that asking ‘did you understand?’ or ‘is that OK?’ every five minutes is not enough. It can even be counterproductive, scaring away less-assertive students.

I learnt to relax a little and take a different approach: when I explained something, I encouraged the students to explain it back in their own words, providing detailed breakdowns of a certain task, anticipating possible problems, and so on.

Ultimately, this came down to probing for understanding of the science, rather than delivering a lecture or grilling an interviewee. Sometimes this approach helps when a student thinks they fully understand something but actually don’t. For example, one of our students was less experienced in programming than others, so for more difficult tasks, we broke the problem down and wrote a sketch of the computer code that they would fill in on their own during the week.

Adapt supervision to the student

Each student requires a different type of supervision, and we tried to adapt our styles to accommodate that. That could mean using Trello project-management boards or a shared Google Doc to record tasks; defining tasks in detail and walking through them carefully; or taking extra time to explain and to fill knowledge gaps. I tried to be supportive by reminding students that they could always send an e-mail if they were stuck on a problem for too long. One of the students found it very helpful to text brief updates outside of scheduled meetings, as a way to hold themselves accountable.

Sometimes, if we felt a student needed to be challenged, we proposed new tasks that were not in the original plan or encouraged them to follow their interest, be it diving into the literature or coming up with further experiments and research questions.

One student conducted a literature review and summarized the pros and cons of the state-of-the-art technology for a follow-up idea we had. That saved some time when we picked up the project after the student left; they learnt lots of interesting things; and the discussion section of the manuscript was much more interesting as a result.

When things go badly, make another plan

Not all projects can be successful, despite your (and your student’s) best efforts. So, as part of each project, my supervisors and I prepared a plan B (and C), working out which tasks were essential and which were just a nice addition. This included a simpler research question that required less work than the original. The initial plan for one of our projects was to compare a newly proposed method with the usual way of doing things, but the new method turned out to be much more difficult than anticipated, so we decided not to do the comparison, and just showed how the new method performed.

Halfway through the project is a good time to evaluate how likely it is that the thesis will be handed in on time and as originally planned. The top priority is to help the student graduate. That might entail either forgoing some of the tasks planned at the beginning, or obtaining an extension of a few months if possible.

Have a final feedback round

After the oral examinations, Benjamin and I met to decide the students’ final grades on the basis of the university’s rubric. We then met the students one last time to tell them our decision, going through each item in the rubric and explaining the motivation for the score we had given. We tried to recall relevant events from the past months to make each student feel the grading was fair.

We also remembered to ask the student for feedback on our supervision and to suggest things they thought we could do better.

Lastly, I encouraged those students to apply for open positions in our lab, and offered to write recommendation letters for them.