Teaching intelligence - It is possible to avoid the negative mass effects
In the first of a series surveying research evidence about teaching and learning, Graham Gibbs considers the findings on large class sizes
As an undergraduate in the 1960s, my lectures at a low-ranking university contained about 20 students and my seminars about six. My teachers knew me, and their doors were open for discussions of my essays and lab reports. As resources in higher education have declined, class sizes have grown … and grown … and grown. One side-effect of the “rationalising” of course provision at many institutions is that the same number of students on the fewer remaining courses will inevitably find themselves in larger classes. But does this actually matter?
Well, it certainly matters in schools, where, as class sizes rise, teachers behave differently, learners behave differently, attitudes to learning change - and attainment declines markedly.
Average school class sizes are used in international league tables as indicators of national commitment to education. And school classes of a similar size to those in UK higher education are rarely found outside developing countries.
The effects of class size are greatest for younger pupils and are least, but still substantial, for those aged 18 or over. Studies of what goes on in higher education discussion classes as they get bigger still reveal a predictable pattern of fewer students saying anything, and the little they do say being at a lower cognitive level (checking facts, for example, rather than discussing ideas).
Students in larger classes have been found to take a surface approach (attempting to memorise) to a greater extent and a deep approach (attempting to make sense) to a lesser extent. Higher education students judge teaching to be less good in large classes - even those led by teachers who gain good ratings when they teach smaller classes. So if managers hope to improve National Student Survey scores by rationalising course provision, they have their work cut out.
It might be argued that classes, and their size, are less influential in higher education. After all, students are supposed to spend most of their time learning independently. However, even 25 years ago when the range of class sizes was much narrower, tripling the numbers in a class could halve the proportion of students gaining the top grade.
For some disciplines, class size was found to have more impact on performance than students’ A-level achievement. These are not small effects, and classes are now much bigger than they were 25 years ago.
Logistics also plays a key part in class size effects. Two of the best predictors of student engagement (and hence learning) - close contact and prompt feedback - are hard to arrange in large classes. Assessment becomes mechanised, and the intellectual level of assignments and tests drops to allow quick and easy marking. As marking loads and the number of parallel discussion classes expand beyond the capacity of a single teacher, postgraduates or part-time staff are drafted in. Student performance and retention are known to be lower when more part-time teachers are used.
Access to resources becomes competitive with larger groups of students. My daughter, studying at a Russell Group university, did not tackle a single essay on a topic she was interested in during her final two years because by the time she got to the library, all the key books for the popular topics had been checked out.
With more students, facilities such as design studios stop being personal spaces that individual students “own” for the duration of a design project and become places they visit occasionally. For every 12 students added to a course cohort in art and design, average marks fall by 1 per cent, it has been found.
These latter problems are cohort size effects rather than class size effects: they are about how students change the way they study on large-enrolment courses. But they still show that size matters in higher education, and cohort size is strongly negatively correlated with student performance.
Another issue with very large classes is that social processes start breaking down and students can become alienated. This matters because social engagement improves retention. Cheating, hiding library books and antisocial behaviour at the back of lecture theatres can all proliferate.
Although these negative “size” effects are not inevitable, they generally follow from the scaling-up and thinning-down of conventional teaching and assessment practices without regard to the likely consequences for how students go about their studying. For example, “formative-only” assessment - assignments with feedback but no marks, which are known to be important to learning - has disappeared from most large courses.
In contrast, The Open University regularly tops NSS league tables and yet has much the largest cohorts, with some course enrolments of more than 10,000 and few under 500. However, its students have high-quality learning resources at their fingertips (and not just in lecture notes), have tutor groups of only 20 (and as each tutor usually has only one group, he or she gets to know the students), and have far more assignments and feedback than is usual. The university’s pedagogic system works very well to support student learning despite huge enrolments.
The US National Center for Academic Transformation has helped scores of institutions to redesign large-enrolment, first-year courses. It has shown that it is possible to improve student outcomes while reducing teaching contact time - and spending less - but not by simply scaling up and thinning down conventional methods.
So as with most educational research evidence, findings about class size describe what normally happens given the way teachers normally behave and the way courses are normally designed. But it is possible to buck the trend by doing things differently.
Graham Gibbs is professor of higher education at the University of Winchester.