Innovating Backwards?

Why You Might Not Want and "Innovative" School for Your Kids

If Hospitals Were Like Schools...

Imagine a visit to the emergency room that went something like this. Worried you might be having a heart attack, you complain of chest pains. Instead of using the usual protocols, the attending physician says, “Yes, thousands of other doctors have had good results using what’s tried and true, but it’s not my style. I’ve developed my own way.” A savvy patient would be worried; while this doctor’s approach might be better than the existing protocol, it is far more likely to be inferior. Adopting a new approach in the absence of evidence is dangerous. Yet, this is exactly what we force teachers to do in our schools - adopt idiosyncratic and untested ideas.

Or rather, it is what many would have us do. If we are to keep improving our schooling outcomes, we need to keep what is working. The imperative to innovate in our schools is seductive, but we run the risk of changing what already works in favour of untested hypotheses. Looking back over the history of schooling, we can see a lot of under-evidenced reforms. Multiple intelligences come to mind. As do learning styles. The history of schooling is too often the frequent adoption of fads unsupported by evidence. This amounts to the worst of all worlds: ignoring the systematic evidence in favour of the gleaming one-off study. Or adopting teaching models that have no empirical evidence at all.

Teachers who resist these ideas are often labelled as lazy or troublesome, or worse of all, acting without their students’ best interest in mind. But educational reformers, consultants, and political actors often label experimental ideas as certain truths. In fact, teachers are often wise to approach educational innovation with some skepticism.


Isn't Educational Research Bad? Hardly.

Part of the blame rests in the common wisdom that educational research is bad. Any field has its share of studies with spurious conclusions and shoddy methodology. But we should not allow the bad to obscure the research that is productive and helpful. The best of it examines studies in the same way medical researchers do; while achieving a double-blind study in schooling research is likely impossible, thorough meta-analyses of decades of research approaches the kind of systematic understanding we rely on in health sciences.

This is the view of Canadian-trained New Zealander John Hattie, whose work - most notably, his Visible Learning project - systematically analyzes educational research and provides welcome insight into the sometimes confusing results of studies. Not unlike nutrition research, individual studies need to be understood as part of a longer story. Some studies say one thing; others disagree. Meta-analyses like Hattie’s bring together thousands of research results involving millions of students to provide a good, if provisional, answer to the question: what works best in the classroom?

That language is important. The question is not: will students learn if a teacher uses a particular model of teaching, but will they learn more using this model than if we had used another? If there are no gains to be made over existing practice, we harm our students. By way of small example: Hattie’s work suggests direct instruction - a teacher-centred, traditional version of teaching - is more effective than many other teaching practices. And yet teachers have been told for at least a generation or two that direct instruction does not work.


Schools Are Already Good, But They Can Get Better

Canadian schooling has excellent outcomes - the OECD PISA results are a testament to that. This is not to say that improvements cannot be made; they can and should. (Schooling outcomes for aboriginal Canadians, for example, demand urgent reform.) But if schooling is to improve further, the answer cannot lie in adopting just any notion, no matter how interesting it seems. We can hardly afford to scale up ideas that, while different, are not improvements over current practice. In schooling, as in medicine, what is different should only be adopted if it is demonstrably better.



Image credit: Robb north via Flickr