Within-class grouping: a rejoinder.

Prais, S.J.

Let us briefly remind ourselves of the current policy-context of this issue in Britain. The need to raise children's schooling attainments to a very substantial extent has become widely accepted in the past fifteen years following international comparisons (many based on research at this Institute) of workforce vocational qualifications and school-leaving standards. The consequences are expressed today in interventionist public policy in terms of a National Curriculum laid down for all school-ages (adopted ten years ago), together with more recent detailed syllabuses in the core subjects of language and mathematics embodied in the National Literacy and Numeracy Strategies for primary schools (adopted in the past two years). Much of the need for such reforms in Britain can be traced to worries as to whether teaching time was well spent, particularly in primary schools using 'modern' teaching methods which required children within each classroom to be divided into small groups, each group sitting around its own small table, many children not facing the wallboard (many classrooms even having their wall-board removed) so as to promote less 'didactic' teaching and more 'discovery' learning by pupils. The frequently ensuing difficulties of teachers in dividing their time effectively among those groups, the consequential frustration of those children who awaited the teacher's attention, the slower general pace of learning, and the particular disadvantages suffered by slower-developing children, need not be spelled out here; they have been closely examined in research involving timed classroom observation, such as the 'Oracle' project of Professor Maurice Galton and his colleagues.

Direct teaching of the whole class, according to a detailed syllabus-sequence, forms the essential 'new' element of the National Literacy and Numeracy Strategies; but those strategies still include a provision that for a central part of each lesson the class is to be divided into ability-groups to be taught separately. In other words, public policy has adopted a debatable compromise one that hardly satisfies those who see within-classroom grouping as a source of substantial inefficiency. There was thus much policy-interest as to whether the summary of North American studies by Professor Abrami and his colleagues can be accepted as showing that within-classroom grouping can be relied upon to raise pupils' attainments.

His finding of a small average positive effect based on a very large number of studies has to be accepted very cautiously because of the wide underlying variability, with a large proportion of those studies showing substantial negative consequences of within-classroom grouping. For policy purposes, as I argued in my previous comment, the variability in results of those studies is so great that their average has to be treated as effectively meaningless. (In technical statistical terms: our concern must be not with the sampling error of that calculated average, but with the standard deviation of the underlying distribution - which turned out to be enormous, and to which they did not give sufficient emphasis). Many other factors must thus be at work, and a multivariate analysis needs to be carried out to determine the net independent contribution of within-classroom grouping. In their Response (printed above) Professor Abrami and his colleagues now tell us they have now carried out such an analysis, and have identified 'a combination of [other] factors which greatly enhance pupils' learning'. When those other factors are not contributing positively, they tell us, 'learning in small groups may be less [effective]'.(1) We shall have to wait patiently to learn what those factors are; and whether those other factors taken alone might provide a more effective and economical way forward of improving children's learning.(2)

In any event, we must not forget the wider context of associated organisational reforms in schooling that are on the agenda and against which within-classroom grouping would properly need to be compared. One method, for schools which are large enough, is to group children into parallel classes, either into streams according to their 'general ability', or into sets for each subject according to attainments in that subject; in that way the teacher is not under pressure to divide her time between separate groups within each class.(3) Another method, widespread in primary schools in the European Continent, is to aim for more through-going whole-class teaching based on a panoply of measures to keep the whole class together in its academic progress: intensive attention to help low-attaining pupils catch up with the majority of the class; a deliberately steady - if sometimes seemingly slow - rate of progress in teaching, which anticipates difficulties (irregularities and exceptions) rather than correcting mistakes afterwards; and, with a view to ensuring that a child is placed in a class in which he can best progress, encouraging parents of the occasional very forward child to allow him to jump a class, and the parents of the occasional very slow-developing child that he would develop better if he spent another year in Nursery/Kindergarten and delayed starting formal schooling by a year.

The latter measures seem to work well in high-achieving Continental countries, including countries covered in the tests administered by IEA and IAEP in recent years. Those high achievements and the associated teaching methods have provided much of the inspiration for experimental reforms in schools in the London Borough of Barking and Dagenham which are associated with this Institute's educational research programme. The rapid progress of the latter schools (as now evidenced in their results in nationwide SAT tests) have, in turn, provided much of the inspiration for the central ideas of the current National Strategies in Numeracy and Literacy.

NOTES

1 Presumably this means that within-classroom grouping on average is by itself a net disadvantage. In other words, it now appears that the absence of such propitiating factors completely overturns the gross (apparent) effect of within-classroom grouping!

2 They also say, curiously, that 'univariate analysis' is equivalent to examining each factor 'separate[ly] from the others' and 'eliminates the influence of collinearity'; as it stands, it would seem that the word 'not' has been inadvertently omitted at various places due to some typographical errors. Or could there be some fundamental difference of understanding on these matters?

3 The recent NFER survey - while arriving at no clear conclusion - is of interest: L Sukhnandan and B Lee, Streaming, Setting and Grouping by Ability: A Review of the Literature (NFER, 1999), and the summary in NFER News (Spring 1999), p. 6.