Too much reliance on random trials in education would be 'bad science', says Keith S Taber

spoon full of medicine

Source: Shutterstock

The Department for Education in England has just published a paper on educational research by Ben Goldacre, a medical doctor who is known for his campaigns against various forms of 'bad science'.1 Goldacre's paper is aspirational and supportive of education, which is to be welcomed. He makes several key points, which are considered here:

Education should become an evidence-based profession, informed by research

As the author of a textbook for teachers on Classroom-based research and evidence-based practice2 my only question here is why the Department for Education did not make Goldacre aware that this is something that has been widely accepted for decades. Many teachers already carry out their own evidence-based practice as well as collaborate on larger studies with academic researchers.

Teacher education should include learning about the processes of research

Understanding research is indeed an essential part of the professional preparation of teachers. This is especially important as different classrooms, teachers and groups of learners are so diverse, that what works in one context does not always apply to another: so teachers need to not only be informed by published research but also skilled to test out 'what works here' through their own enquiries.

Traditionally, a major route into teaching in England has been through postgraduate courses taught by university-school partnerships. These Masters level courses are required to engage trainees directly with the processes and outcomes of educational research (eg the Journal of Trainee Teacher Educational Research3). Unfortunately this route is under threat as the Department for Education seeks to shift more teacher education from postgraduate routes to on-the-job training.

Teachers should be given better access to the results of research

Goldacre correctly points out that most teachers seldom access primary research literature. However, the practitioner journals and digests he suggests already exist (for example Education in Chemistry and School Science Review) and are widely accessed by teachers. Organisations such as the Royal Society of Chemistry commonly offer courses based on findings from major research initiatives, and university-based components of initial teacher education programmes (whilst they continue to exist) inform new teachers of major research findings.

Teaching can be informed by adopting an evidence-based model, as in medicine

Goldacre makes the critical point that research does not tell the practitioner what will work in a particular case, but rather provides a knowledge base for making informed decisions about which options seem most promising. This has been recognised as the right medical model for education: one that recognises the complexity of teaching and learning, and the practitioner as a 'learning doctor' who can diagnose problems and prescribe 'treatment' on a case-by-case basis.4

Infrastructure and resources should be developed to support random trialling of educational innovations

Goldacre acknowledges that what trial results suggest is effective is not necessarily the right option in all contexts. However, there is a danger that such caveats may be ignored when the 'headline' message of Goldacre's paper seems to be the importance of carrying out random trials. Teachers only develop expertise in applying new approaches or teaching materials over time, undermining any trial where teachers are expected to adopt a novel approach or teaching resource. Testing the application of a new educational approach or resource is not like testing the efficacy of a tablet that simply has to be swallowed.

People respond to the novelty of innovations regardless of their inherent merits. Some teachers and students feel uncomfortable and threatened when they first experience something new. Conversely, anything that makes lessons seem different may motivate students, sparking interest and engagement that would soon fade once the novel became routine. Moreover, people's expectations of outcomes can be especially powerful - as has been recognised since studies found that progress in school could be increased for those children who researchers (randomly) suggested were most likely to make good progress.

Goldacre's analogy with medical trials is questionable as they are often double-blind studies, so neither the patients nor the researchers know who is getting the treatment being tested rather than a placebo. If we could test educational treatments in this way, without the researchers, or the teachers or students, being aware who was experiencing the innovation, then randomised trials would become a lot more informative in educational settings. In most cases, this is just not possible.

A strong bias towards random trials in education in the US has been strongly criticised by academics,5 and resulted in the National Research Council issuing a report arguing for a much broader range of educational research.6 Too much reliance on random trials in education would be 'bad science'.

Conclusion: prescribing the right medicine

Teacher preparation certainly should include a strong research element, as currently found in university-based initial teacher education. Being involved in educational research is part of being a fully professional teacher.2 This can certainly include involvement in large-scale projects such as randomised trials of new innovations. But it must also include testing out the applicability of research findings in the specific conditions of particular classrooms. Without local validation, the outcomes of trials can only ever be suggestive, and such findings need to be taken with a large pinch of salt.

Keith S Taber is a reader in science education at the University of Cambridge (where he teaches educational research methods), and editor of the research journal Chemistry Education Research and Practice.7 The second edition of his book 'Classroom-based research and evidence-based practice' will be published later this year.

Updated 17/05/13: This article has generated some discussion on Twitter. Take a look at the links to Feedback and Storify at the end of this article.


Feedback - May 2013

Work to school transition and building evidence into education

 Storify - The right medicine for educational research? 

A conversation on Twitter between @jdhowgego and @teachingofsci