How data makes student–supervisor matching meaningful

Why supervisor matching matters

Dissertation supervision is one of the most developmental experiences in higher education. The quality of the student–supervisor relationship influences outcomes such as engagement, confidence, satisfaction, and academic achievement. Yet, in many large programmes, allocating students to supervisors is still treated as an administrative task rather than an educational design decision.

When hundreds of students and supervisors are involved, traditional methods such as sign-up sheets, preference surveys, or pre-proposals often become inconsistent, time-consuming, and inequitable. These approaches can work well in small groups but struggle to ensure fairness and transparency at scale.

A data-driven approach to measuring topic alignment

In my recent study, “Improving Student–Supervisor Matching: Educational Gains from a Data-Driven Approach,” I explore how topic alignment can be systematically measured and used to guide dissertation allocations. The project, based on the 2024/25 postgraduate Economics cohort at the University of Glasgow,[1] combines survey data from 102 students with a matching algorithm designed to make allocations more systematic, transparent, and educationally meaningful.

The idea is simple: measure how closely a student’s dissertation topic aligns with a supervisor’s expertise using a structured research classification system. In Economics, this system is the Journal of Economic Literature (JEL) classification. Each JEL code represents a specific field, ranging from broad categories to narrowly defined topics (see Figure 1). By comparing the overlap between student and supervisor JEL codes, I construct a match score between zero and one. Exact three-digit matches receive the highest weight, while broader overlaps still contribute to the score.

[1] Ethical approval for this study was granted by the Committee for Scholarship of Teaching and Learning at the University of Glasgow (application #26).

Figure 1: Hierarchical structure of the JEL classification system.

Evidence from student experience

To examine whether topic alignment influences the student experience, I surveyed students after submitting their dissertations. Among those supervised by the same person, students with higher topic alignment were more likely to report an excellent supervision experience (see Figure 2). This finding aligns with previous research showing that a good supervisor–student fit is associated with higher satisfaction and improved outcomes (Armstrong et al., 2004; Ives and Rowley, 2005; de Kleijn et al., 2012; Orellana et al., 2016; Kushwah and Navrouzoglou, 2022; Wang et al., 2022).

Figure 2: Predicted probability of reporting an excellent supervision experience as a function of the match score, estimated using a fixed-effects logit model with supervisor fixed effects. Error bars represent 95% confidence intervals.

The matching algorithm

To translate this measure into practice, I developed a transparent, rule-based algorithm that systematically allocates students to supervisors in three steps:

1. Compute match scores between all possible student–supervisor pairs.

2. Rank students, prioritising those with fewer possible matches.

3. Assign each student to the supervisor with the highest match score.

When applied to the sample, the algorithm raised the share of three-digit JEL overlaps from 20% under random allocation to 80%, and the average match score tripled.

Educational and institutional benefits

From an educational perspective, systematic matching can:

• Improve the quality of dissertations process by pairing students with supervisors who specialise in their chosen area.

• Enhance student experience.

• Support transparency in the allocation process.

• Reduce workload for staff and anxiety for students.

  
  

Lessons for programme leaders

For programme conveners and administrators, the message is clear: supervisor allocation is a matter of educational design.

Practical suggestions for implementation:

• Train students to choose a concise set of appropriate classification codes.

• Use structured classification systems already available in your discipline, such as Medical Subject Headings in Medicine or ACM Computing Classifications in Computer Science.

• Adapt the match score and algorithm to suit institutional needs and context.

• Begin on a small scale and refine the process before expanding.

  
  

From allocation to educational design

This project contributes to showing that data-driven matching is not only feasible but also educationally valuable. It helps bridge the gap between computational tools and the practical needs of large programmes. By treating allocation as an educational design challenge rather than an administrative process, universities can improve both efficiency and the quality of the learning experience.

References

Armstrong, S. J., Allinson, C. W., and Hayes, J. (2004). The effects of cognitive style on research supervision: A study of student–supervisor dyads in management education. Academy of Management Learning & Education, 3(1):41–63.

de Kleijn, R. A., Mainhard, M. T., Meijer, P. C., Pilot, A., and Brekelmans, M. (2012). Master’s thesis supervision: relations between perceptions of the supervisor–student relationship, final grade, perceived supervisor contribution to learning and student satisfaction. Studies in Higher Education, 37(8):925–939

Ives, G. and Rowley, G. (2005). Supervisor selection or allocation and continuity of supervision: Ph.d. students’ progress and outcomes. Studies in Higher Education, 30(5):535–555.

Kushwah, L. and Navrouzoglou, P. (2022). Enhanced student satisfaction through effective supervisor–supervisee allocation: A case study. Journal of Perspectives in Applied Academic Practice, 10(1):54–65.

Orellana, M. L., Darder, A., Pérez, A., and Salinas, J. (2016). Improving doctoral success by matching phd students with supervisors. International Journal of Doctoral Studies, 11:87–103.

Santos, Ana (2025). Improving Student–Supervisor Matching: Educational Gains from a Data-Driven Approach. Working Paper.

Wang, F., King, R. B., Zeng, L. M., Zhu, Y., and Leung, S. O. (2022). The research experience of postgraduate students: a mixed methods study. Studies in Higher Education, 48(4):616–629.