SCMP Saturday, November 25, 2000


Global debate over whether size matters


Teachers have consistently seen large classes as a barrier to effective learning. In the Western world, cutting class sizes has been a rallying cry of teachers' unions, and parents.
The importance of the debate is reflected in hundreds of studies into class size and the effect on learning and achievement. In the United States 1,100 studies had been published by 1998.
Only a tiny percentage of studies unambiguously show that smaller is better. Some found class size has little effect on outcomes. Several said larger classes were actually better. Most studies link class size to test scores, ignoring issues such as teacher quality, student effort and intellectual ability, cultural factors and classroom interaction.
The definitions of "large" and "small" vary, with some regarding "small classes" as 25 to 30 pupils and others under 15 pupils. Some studies regard 25 pupils as a "large" class.
US researchers note that while average class sizes fell from 30 in 1961 to 24 in 1986 at primary level and 27 in 1961 to 22 in 1986 for secondary, achievement scores have dropped. "This presents the paradox of the smallest classes in the history of US schools producing the lowest test scores," say researchers Lou Ann Dickson and Gail Sexton.
Various explanations are offered, including the racial mix and greater mobility of pupils. Studies on large classes (above 35 pupils), including one done in Taiwan in the mid-1990s, showed that highly competent teachers could handle classes of up to 50 children and still produce good results while less competent teachers could not get the best out of a class of 12 children. This translated into big class equals good, small class equals bad in the research.
The Taiwanese research is more relevant than at first glance. Western governments argue that to cut class sizes, more teachers would have to be recruited. Already there is a severe worldwide shortage of teachers in certain subjects, particularly mathematics and science, so some children could be taught by poorly qualified or unqualified teachers, the argument goes.
Culture cannot be ignored. Japan, with some of the largest average class sizes (37 to 41 at secondary) comes first in maths and science test scores (Third International Maths and Science Study - TIMSS). Luxembourg, with the smallest average class size (19 at secondary), ranks 18th. Hungary has similar class sizes to the US but Hungarian pupils perform far better.
Parental support, the importance of education in the national culture, and status of teachers are all factors that can override class size. In an analysis of TIMSS data, Dr Suet-Ling Pong, associate professor of Education and Sociology at Penn State University, noted that in Australia and France large classes do better than small ones, even taking socio-economic status into account.
In Canada, Germany, Iceland, Korea and Singapore, class size does not matter. Socio-economic data was not available for other countries. Taipei's Hsyoulang primary was until recently in the Guinness Book of Records as the world's largest primary school - 8,000 pupils, averaging 40 to a class. Class sizes were as high as 45 before the birth rate fell. The school is huge because parents actively choose it. Some of the largest classes are led by teachers considered to be the "best".
Evidence in Japan also shows parents were determined to get their children into the best performing schools, irrespective of class sizes. Yet the Taiwanese or Japanese experience where chalk-and-talk is the norm may not apply elsewhere. Teachers' unions say teaching a class of mixed ability where the children discuss and do their own work in groups of four to six is more difficult to manage than conventional methods. This change in teaching style, now common in Britain, the US and other English-speaking countries, has fuelled calls for smaller classes.
When the teacher is with one group, the other groups must be "working independently". In reality, other groups may be off-task in a large class where a teacher can only give a few minutes to each in one lesson.
In an analysis of class size studies in the late 1980s, US-based researcher Gene Glass and his associates examined dozens of papers. They found a huge difference between pupils taught individually and those taught in classes of 40. However improvement in scores levels off at about 20 to 25 pupils per class. A small difference in scores was noted in classes of 20 to 40. They say the best results are gained in classes of 15 to 20.
With more than 30 pupils, it appears to make little difference how big the class is. The Tennessee STAR (Student-Teacher Achievement Ratio) study is the most widely quoted on class size. An experimental school was established in a Tennessee district in the 1980s. In Grades One to Three classes were cut to 15 pupils with control groups of 25 to compare the effects. The control group matched the experimental group in social background and scores.
This was followed by a longitudinal project studying classes of 13 to 17, 22 to 25 and classes of 22 to 25 with a full-time classroom assistant. There were about 100 classes of each type.
Results showed that class size had little effect on motivation and self-confidence, but students in smaller classes performed better in all locations at every grade level, particularly in reading scores, less so in maths.
The small class effect diminished after Grade One (age 6) indicating that the most cost-effective cuts occur prior to Grade Two (age 7). Classroom assistants were less effective than smaller classes in enhancing achievement.
The students were followed in subsequent years after returning to regular-sized classes. In this Lasting Benefits Study the effects of small classes carried through to Grade Seven (age 12) although it diminishes in effect after Grade One.
In its analysis the STAR project found that smaller classes allowed more individual instruction, and that co-operation between pupils increased while discipline problems were reduced.
Subsequent analyses have tried to further pin-point the effects. Small classes lead to a drop in disruptive behaviour with a rise in participation, including completing tasks and asking questions to get more information.
Researchers now argue whether it is teaching standards (which can be increased through training) or actual numbers in the class which produce these effects.
Eric Hanushek, professor of economics and public policy at the University of Rochester, New York, told a recent seminar organised by the Centre for the Economics of Education in Britain that teaching standards had the biggest impact on pupil performance with reducing class sizes resulting in only a marginal academic improvement.