As teachers, people often dream of research results which will tell them, with scientific accuracy, what is going to work in their classroom. However, as researchers, people know how elusive such "recipes" or "best methods" are. Conventional quantitative research methods have been designed to identify the effect of "factor X" in a given context, as predicted by a given theory, and to allow generalisations to be made to similar contexts. But learning environments are infinitely complex and their variation is hard to control; it is therefore difficult to predict that what works in one case will necessarily work in another albeit similar one, no matter what the theory suggests. Exhortations to "evidence-based practice", then, may be met with scepticism. However, to develop as teachers, people constantly need to evaluate their practice and explore what works and what doesn't in their own context. For this reason, "action research" has become a common research approach in educational research. In the process of encouraging learners to reflect on their learning, teachers need to ensure that they in turn continue to reflect on their teaching.