Anthony Crosland once commented in an interview that “you don’t want your priorities set by research”. He was not the first to think so. Lady Plowden, whose committee on primary schools was set up by the Conservative Edward Boyle, but most of whose work was carried out under Crosland, had already performed mental gymnastics to ignore research evidence that grouping pupils according to their abilities led to improved academic performance. Her reasoning closely reflects that of our opponents today. Here are some examples:
Schools which treat children individually will accept unstreaming throughout the whole school. When such an organisation is established with conviction and put into effect with skill, it produces a happy school and an atmosphere conducive to learning. (819)
Streaming can be wounding to children… No more certain way can be found of alienating children from school or of creating irresponsibility. (823)
Compare this with a comment on mixed ability teaching in languages from 2008.
“Grouping pupils according to their ability would seem a logical way to allow all pupils to achieve their potential. However it makes very transparent the differences in the academic ability of pupils and is therefore not a very inclusive approach. Setting can lead to stigmatisation, low self-esteem and disruptive behaviour in pupils in lower sets. A mixed ability class allows for more of a social mix…”
This line of thought has clear consequences for research. Research generates evidence, the bulk of it in terms of advances, or lack of them, in children’s knowledge, skills and understanding.
Most of us think of this as achievement or standards, but in educational circles it is usually called attainment, a term with fewer positive implications. If the central goal of education, in the context Sir Anthony Seldon’s ideas of balance and aptitudes, is to advance standards and achievement, then research is an important tool.
If the goals are those of Lady Plowden, Anthony Crosland, and the London Institute of Education, research is dangerous. I have been unable to locate any research whatever into the effects of mixed ability teaching in modern languages, despite its being imposed on languages teachers, against the will of most of them, since the days of Crosland and Plowden. Mossbourne and some other schools, such as Dereham Neatherd, generated vastly improved results in languages by setting and fast-tracking – the latter taking advantage of the dumbed-down GCSE, though in a way that was not in the long-term interest of the pupils. Mixed ability teaching brought modern languages close to meltdown until Michael Gove’s Ebacc stopped the rot at GCSE, though they are still in intensive care at A level, about which more next week.
One response to unwelcome research evidence has been to throw the academic kitchen sink at any piece of research that does not fit the progressive book, and I’ve described the ill-informed attacks on phonics research here.
Central to the argument is the notion that the only way to find anything out is by randomised controlled trials, which play a key role in the introduction of new drugs, but which are by no means the whole story in scientific research, any more than in education. Their main limitations are cost and scope. Cost is important, as you need a very large sample to produced a result in an educational context – the excellent ESRC-funded study that showed the value of teaching grammar to secondary pupils cost £750,000.
Scope is important, as these trials can only test one element against another. Unless both are carefully chosen, they can yield no useful evidence at all, and the study carried out by Professors Carol Torgerson and Greg Brooks in the link above is a prime example.
A fully effective programme of educational research should include the following:
1. Case studies. Properly conducted case studies have played a crucial role in the development of surgery. They allow very close observation of individuals, and highly detailed tracking of progress over time. It is fashionable to dismiss them as “anecdotal”, which avoids considering them on their merits. Major medical developments, including vaccination, anaesthesia and antisepsis began with close consideration of cases.
2. Small-scale school-based studies. Careful sampling allows the progress of different groups to be tracked in ways similar to those used with the pupil premium. Control groups can be selected with confidence that unknown factors are not going to appear – such as different teachers – and historical control groups can be used with previous year-groups. As with case studies, the cost is minimal beyond the time of the people carrying out the research.
3. Studies with matched control groups, followed up over time. “Washout” effects are a permanent problem in educational research, and bedevilled early work on Reading Recovery, where children not receiving the extra teaching were found to have caught up over the next year with those that had received it. Matched control groups ensure that we are comparing like with like, and that groups receiving extra teaching do not start at a higher point than comparison group. Research into children’s awareness of sounds in the eighties used an additional comparison group, using another form of extra support (grouping words according to their categories) to guard against the possibility that any improvement was just due to extra personal attention to the pupils. The Clackmannanshire study is the largest model using matched control groups.
4. Long-term statistical studies and trends. Not all research evidence comes from experiment, and national statistical evidence can be very valuable, provided we look at its basis. The national reading test, NS6, abolished by the Bullock committee after it showed a fall in reading standards (to make sure that such a thing did not happen again!) should be restored.
5. The place of randomised controlled trials is to investigate big issues that have been clarified using the forms of research described above. They are the equivalent of what General Patton called “a reconnaissance in force”, and using them to investigate ideas that might or might not work is a waste of time and money. So far, the grammar study described above is the only RCT to have yielded useful evidence, and RCTs with completely inadequate research design have been given too much credence just because of the randomisation factor – a link to Professor Rhona Johnston’s view on the subject is here.