Do uni class sizes matter?

Should we worry that uni student numbers have grown more quickly than academic staff numbers? The history of school research on student:staff ratios suggest that we should be cautious about making lower SSRs a policy priority. Most analysis has found little or no educational benefit in reducing school student:staff ratios. Increased SSRs may be a sign of higher productivity rather than a problem.

Given that uni students are supposed to be more independent learners than schools students, universities (with rare exceptions like the Oxbridge tutorial system) have always used large groups methods of instruction such as lectures. Unis can teach many more people at the same time than schools.

On interesting question is whether if the class expands beyond the size where personal interaction is feasible, does it matter whether there are 100, 200, 300 or more students in the room? If, as the school research suggests, it is teacher quality that matters most to learning it might be better to have the best lecturer teach a class of 300 than to have six not-so-good lecturers take six classes of 50.

On the other hand, bigger classes may mean more scope for disruption or distraction of students by misbehaving students, and complications from teaching students with a wide range of abilities if opportunities to stream are lost (in my own limited lecturing experience 20 years ago teaching first-year politics, range of abilities was one of the biggest problems as I had some students who were quite well-informed and others whose knowledge of politics was close to zero).

In tutorials, I can see much stronger theoretical reasons for believing that bigger is not better than for lectures. This is a chance for students to ask questions, and for the tutor to monitor the students. If there are too many in the class, students may be too shy to ask questions, be too easily dominated by loud students, or simply shirk. And it is harder for tutors to keep track of all the students.

Though SSRs don’t seem to be much used in analysis of university performance, I have found some papers on class sizes.

In research based on an unamed UK university the analysis used numbers of students enrolled to take the final exam; a perhaps less than ideal measure given that it is consistent with a wide variety of actual classroom practices. It found that large classes were associated with lower marks, though not in the 30-100 class size range (counter-intuitive, as I would have thought that this is precisely the range in which it becames hard for the lecturer to know and attend to everyone).

A study of an Italian university with regulated class sizes found that class size (as in people in the room) had a largish effect on pass rates. ‘An increase of 50 students in class size reduces the student’s probability of passing an exam by about 8.7 percentage points.’ They were comparing students with the same teachers, so that was not a confounding factor.

The most interesting study I have seen to date, also on an Italian university, was able to measure both academic results and subsequent salaries. The academic effects were statistically significant but of modest apparent importance. An increase of class size by 20 from the mean of 131 resulted in a mean reduction in grades by about 1/3 of a grade point. But an increase of class size by 20 was associated with a lower monthly starting salary of about 80 Euros, or 6%. It’s not obvious to me why such a small difference in grades could translate into a non-trivial difference in salaries. But if this finding could be replicated, it could provide a financial justification for reducing class sizes.

I’d want to see more research in this area, but unlike the schools research the persistent pattern of findings is that class sizes do matter.

7 thoughts on “Do uni class sizes matter?

  1. I think that range of ability is generally solved these days by targeting the lowest common denominator. This is often enforced explicitly by universities (and indeed the government) by having maximum possible percentages of students that can fail and implicitly by student satisfaction surveys.
    I also think that one of the main effects that having ever greater numbers of students per teaching academic has had is to change courses in the long term — if you know you are going to have large numbers of students, you simply design courses to handle it — don’t run labs that require equipment, tasks that students may need to consult you about, etc. . Some courses do not not even get started because of these constraints (i.e., everyone is looking for high student number, low resource courses to run), and no doubt some good ones get closed down because of resource use. This is one reason less and less practical stuff is done in many courses and the reason for the proliferation of certain types of course. It’s also one reason there are less and less “open” tasks given to students in many areas (but most students don’t really appreciate these anyway — many will complain that the task was not well specified enough).
    I think there’s also an expectation difference between now and when you would have gone to university, especially at the undergraduate level. Many students now don’t expect to participate or do a whole lot in tutorials (a growing percentage some never turn up to any)– most just want to be told how to do the assignments and what is on the exam versus actually do stuff.


  2. Conrad – Indeed, I did target the lowest common denominator, but I could tell that some of the others were bored.

    None of these studies look at lab subjects, which is an obvious omission that would need to be rectified in a more comprehensive study of the issue.

    My understanding is that many courses grade attendance at and participation in tutorials as a remedy for the problems you identify. On the other hand in the online stats subjects I have been doing via your uni, I have been surprised by how few people use the discussion board – even when, as recently, there was an explicit though not graded request to put work there.


  3. “unlike the schools research ”
    Actually, school research results vary, as reductions from “large” class sizes do have an effect. It seems that reductions below the low 20s have little effect, but reductions to about 22 do have an impact.

    In Victorian Catholic schools, the current limits are:
    29 in Years P to 10;
    27 in Years 11 to 12 classes.
    Many independent schools also have similar ratios.

    There seems to be scope in some schools for improving potential outcomes by reducing class sizes. Teaching Year 12 Business Management to a class of 27 means I don’t have the time to keep track of all individuals as much as would like. The range of abilities is also increased, as at university.

    I do get annoyed when all research on schools is summarised with “no effect”.

    Sorry to post off topic.


  4. “Indeed, I did target the lowest common denominator, but I could tell that some of the others were bored. ”
    It’s a problem for everyone, since most people would like to give those that are interested the best possible experience, but the university wants you to target the whole range, including the “here for no reason bunch”.
    “My understanding is that many courses grade attendance at and participation in tutorials as a remedy for the problems you identify. On the other hand in the online stats subjects I have been doing via your uni, I have been surprised by how few people use the discussion board – even when, as recently, there was an explicit though not graded request to put work there”
    Online subjects are exceptionally difficult to get running in terms of discussion — if you don’t get people participating early on, then no-one will participate later, and it works better if the participants already know each other. This is especially so if you want people to ask interesting questions because they want to, versus ask questions because they are forced to via some grading requirement, which almost never works because people then don’t reply and continue the discussion.
    I’ve actually experimented with this, and at least with students that know each other, if you put up a number of directed questions to people early on (versus just ask people to discuss things), then it basically breaks the ice and interested people will ask each other questions later. I’m not sure if that translates to groups where people don’t know each other.


  5. They way we are getting around the lack of discussion, in a secondary school, is to require at least one question _and_ three answers to others’ questions per week. While this still a tentative approach, it seems to have engaged the students to the level where participation becomes a habit. Of course, with a larger class to keep track of, this could also be a problematic approach.


  6. Conrad – I’m on my third Swinburne subject, and yet to meet anyone involved or even to interact except with the lecturer. So probably it’s not conducive to discussion. The situation is not ideal, but if I had to go to classes I would not be able to study at all, so given my actual alternatives online is best. I’m learning quite a lot, even if not as much as I would in better study conditions.

    Michael C – Thanks for the clarification on school class sizes.


  7. My experience was certain lecturers was that some improved greatly when teaching a smaller class (150->15 or so).

    One of my worst undergrad lecturers (several people wrote letters to complain) was actually ok at honours level when in a room small enough. This was because he recieved continuous feedback on whether he made sense or not.

    Those who were good in a large class context could also handle the smaller group as well.

    My theory is that most people have a maximum group size they are comfortable with and adept at interacting with. For many people that maximum size is less than 20. For some people that size is less than 5. Above that size we have trouble relating to individuals and reading non-verbal ques. Only very gifted public speakers can interact and engage with a large audience and sense the mood/vibe and adjust.

    The optimum class size is actually lecturer specific.

    The reason why school class sizes matter is because as they become smaller and increasing number of teachers are shifted into their comfortable size class.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s