It’s a match! – On the allocation of thesis students to supervisors

How do we match thesis students with supervisors? Surprisingly enough, there seems to be no university-wide source of know-how on this matter here at Utrecht University. Speaking to other people both inside and outside our institution, it seems that matching thesis students to supervisors is almost an afterthought. Thesis coordinators do their best to carry forward a matching system that they themselves inherited from their predecessors, often without much awareness of its pros and cons or its alternatives. We realized that this is a bit of a blind spot for our university, although having a good match between student and supervisor can potentially have a large impact on the success of the student, the supervisor, and the educational programs that our university offers.

Therefore, we decided to look into this issue to gather (and share) information on alternative matching systems. As a start, we studied some academic literature on this topic. Then we did a presentation at on this issue for educational professionals at the “That’s US” seminar series, a bi-monthly meeting for all USE staff to discuss work-related topics, in which we asked them about their experience with matching systems for matching up students and thesis supervisors. Finally, we spoke to people from our network outside of the university to get their perspective on this matter. In this journey, we gained plenty of valuable insights regarding the pros and cons of different matching systems being used nowadays. This article summarizes the main findings.

Why matching is important

The motivation for our study is that the quality of matching matters for thesis students, supervisors, and their educational program. For students, a good match between research interest and style between the student and supervisor has a direct bearing on their motivation to write a good thesis, leading them to put more effort in it and to finish on time. A good match is also important for their skill development, because a student that is matched with a suitable supervisor can be better challenged to walk the extra-mile and acquire new abilities they would not master otherwise. Together, these effects can positively contribute to both the process and the quality of the thesis project. Finally, a good match of research interest and style between students and supervisors might positively impact their future career in at least two ways. First, it prepares them well to a future career that is in line with those interests. Second, a high-quality thesis is a great opportunity for a student to showcase their abilities to a prospective employer.

For supervisors, on the other hand, a high-quality matching with thesis students grants three important advantages. First, students that have a connection with a supervisor or their research interests and style are much easier to supervise. This is because it costs less effort to introduce them to a research topic and motivate them to do well. When supervisors and students are well-matched, it fosters a more productive and collaborative relationship that can lead to better communication, guidance, and support throughout the thesis process. Second, committed thesis students can directly contribute to a supervisor’s research agenda by exploring valuable pilots in their thesis research. A good match between supervisor and student increases the likelihood of producing high-quality research output. Third, and related, supervisors can use master’s thesis supervision as a way to search for potential PhD candidates.

Finally, the educational program can benefit in several ways from a good match between prospective thesis students and their supervisors. First, an ideal matching system should not demand too much time and effort from the coordinator and supervisors, leaving them space for other activities. Second, if the matching procedure is sound, it leaves little room for error, stress, and disappointment (e.g., having enough students per supervisor, ensuring that each student has a supervisor, ensuring that the matching procedure is finished on time). Third, it could bolster a sense of community by including considerations within the matching procedure that adheres to what the educational institute stands for such as student inclusivity (into the research community), equity of outcome, no discrimination of students and/or supervisors, and clarity for students on what to expect from the thesis project.

Two of the main hurdles that a matching system needs to solve are the “rural hospitals theorem” and students’ demotivation. The first issue refers to the fact that some supervisors are inherently less popular and/or visible than others, which leads to little to no demand for their supervision in a similar way to how rural hospitals are systematically shunned by doctors seeking residency in urban areas. This issue leaves thesis coordinators faced with a dilemma in which students must be appointed to a supervisor that they do not prefer or asking supervisors to change the number of students to be supervised.  The second issue arises when students form strong preferences for one or more specific supervisors but are matched to someone else. Specifically, not being able to study with their supervisor of choice might lead to disappointment and low motivation, thereby affecting student attrition, weak development of research skills, and lower thesis quality. Consequently, these negative effects can be a real challenge for supervisors to manage.  It is easy to see how these two hurdles are connected, as some of the students still need to be allocated to the supervisors that do not make anyone’s top preferences.

These challenges underscore the importance of implementing effective matching systems. Our findings can be summarized as follows. The five main types of student allocation we identified are 1) surveys, 2) pre-proposals, 3) sign-up sheets, 4) random allocation, and 5) KonJoin. We briefly present each of them in turn, along with their strengths and weaknesses, and then we conclude by offering some reflections on how to choose the best matching method.

Surveys

A survey allows students to express their preferences for one or more supervisors and/or research topics. The pros of this system are that it is relatively easy to set up and works especially well for large groups, which are notoriously difficult to manage in terms of supervisor allocation due to the sheer size of the task. The cons are that students can still get matched with a random supervisor if their top picks run out of slots, and there are many degrees of freedom on how to ration popular supervisors (each with pros and cons). The system can be improved by asking supervisors to increase their visibility through the provision of sample thesis topics, and by asking students to enlarge their set of preferences, for example providing a minimum number of picks and mentioning their preferred research fields. The system can be further improved by setting up a survey in terms of preferences for research areas rather than supervisors. As we argue below, the latter setup can effectively tick all the boxes for an optimal system.

Pre-proposals

In a pre-proposal matching system, students write a short research proposal to indicate not only a supervisor preference but also a specific research project they would like to work on. Supervisors then need to evaluate all proposals and pick out a subset of students they will supervise. The pros of this system are that this is a great tool for supervisors to evaluate student commitment and skills, and the system can be further refined by allocating different degrees of priority to different student cohorts (e.g., higher GPAs get to send pre-proposals first). The cons are that students may end up making significant effort in vain if their pre-proposals are not picked by any of their indicated choices, and supervisors need to make significant effort in carefully reading and evaluating all pre-proposals. For these reasons, this system may be appealing for small student groups but cumbersome for large groups.

Sign-up sheets

A sign-up sheet allows students to directly pick a supervisor. Therefore, they have 100% agency in their supervisor choice. This is a fully transparent system because students get to see all available slots in real time, which is especially important when their top picks run out of slots. The issues with this system, however, can be substantial. Students feel severe time pressure to grab a spot quickly with their supervisor of choice, especially under the binding capacity constraints that typically characterize the most popular supervisors. For example, this time pressure may lead to choosing a supervisor without thoughtful consideration. In addition, this system does not warrant an optimal match between student and supervisor, as merely the student’s preference and speed to voice that preference is used. Also, online spreadsheets are not typically made for supporting a lot of people entering data all at once. As a result, the system might crash (with associated loss of data) when too many students log on at the same time.

Random allocation

Students can also be allocated to supervisors through random allocation, which involves a simple draw. Its strength lies in its simplicity and fairness, with the additional advantage that students do not get to form strong preferences for specific supervisors. As a result, they do not get disappointed in the same way as in the other systems when top picks run out of slots. However, students lack agency in their choice, which can have a potential demotivating effect. Also, this can be an important missed opportunity for students with specific research interests. Symmetrically, supervisors also potentially miss an opportunity to recruit talented and motivated students in their own field. This is costly especially for graduate studies, as top students can be motivated to enroll in a doctoral program. These issues are mitigated or absent altogether in the presence of high homogeneity of research topics across supervisors, such as for example in a small-scale program in a specialized field.

KonJoin

Finally, KonJoin is an official university-wide tool for student-supervisor matching at Utrecht University. The remarkable feature of this system is that it is little known, as most of the people we spoke to were not familiar with it. It is essentially a partly automated survey system. However, the matching only takes place based on simple preference and availability. Any further nuances need to be accounted for manually by the thesis coordinator, thereby partly offsetting the benefit of automation. Also, everyone needs to sign up (within a deadline) to create an account. This is especially cumbersome for programs with many supervisors and students.

Is there a best method?

The best method should trivially satisfy a feasibility constraint, and effectively address the two main problems with matching mentioned above. If properly structured, a survey-based system might tick all these boxes. Surveys can largely mitigate or even eliminate the rural hospital issue altogether. To this end, students can make their choice in terms of research topics rather than supervisors, while supervisors commit to supervising multiple research topics. Also, asking students to choose a topic instead of a supervisor makes sure that they do not form strong preferences over specific supervisors, thereby largely avoiding disappointment. As for feasibility, surveys are suitable for student bodies of any size.

Of course, a survey-based system also entails some costs in terms of setup and processing capacity. For this reason, there are two special cases in which we recommend alternatives. The first case is when there is high homogeneity of research topics across supervisors. In this instance, a system based on random allocation would essentially achieve the same goals as a survey but without the associated costs. The second case is when the size of the student body is small, so that the processing costs of all the alternative systems (especially pre-proposals) become manageable. It is important to note that these two special cases often coincide, as small-scale programs typically focus on a particular subfield of research.

Finally, professors and students can sometimes be matched outside of the system. For example, a professor may offer to supervise a student that does exceptionally well on their exam and shows active interest in their research area. By definition, this is a type of matching that only works for a small minority of students. We believe nonetheless that it is important to make this possibility explicit, stressing that the professor does the picking (not the student), to incentivize even better performance among students without gaming the thesis matching system.

We hope that this discussion provides enough elements and context to collectively improve our decision-making when it comes to thesis coordination. If you have benefited from this article, or just wish to share your own experiences and solutions, we would be more than happy to hear from you!

dr Maurizio Montone (Utrecht School of Economics, Utrecht University, m.montone@uu.nl)

dr Pieter de Bordes (Educational Development & Training, Utrecht University,  p.f.debordes@uu.nl)

Marijn van Ellen