By Gerald Coles, Heinemann pound;12.25

Anyone interested in evidence-based practice should read this book. It catalogues the manipulation and deception that can happen in the process of compiling an evidence base - in this case, an evidence base about how to teach reading.

Gerald Coles exposes the process by which researchers and politicians get together to concoct what is taken to be hard scientific evidence. The recipe for such a process is simple: gather some committed researchers who have been marinating in their own ideas for a decade or so, mix with one or two highly flavoured politicians who want to be popular, pour in a dash of corporate alcohol and a few publishers - and hey presto! You have produced a) a body of incontrovertible knowledge about how to teach reading; b) popular politicians; c) rich publishers; and, d) researchers confident that their research will receive funding into the next millennium.

The backdrop for all of this is the United States, but have no doubt that the same process occurs the world over. The central players in this scenario are George W Bush, the National Reading Panel (NRP) and the National Institute of Child and Human Development, all of whom have contrived to discover that phonics-based, skills-oriented teaching of reading is best.

Coles documents in his characteristically forensic way how these players have got together to package the facts, exclude dissent and promote their version of the truth. For example, the NRP boldly claimed to have reviewed a massive 100,000 studies on how children learn to read and to have come to their conclusions about the primacy of phonics on the basis of this huge database. But Coles reports how 99.9 per cent of these studies weren’t examined, having been screened out as unsuitable. Ultimately, only a meagre 104 studies remained, on precisely the kinds of instruction that the panel deemed to be important before the process began.

And not a single “beginning reading” teacher sat on the panel. This exclusion highlights an extraordinary view - prevalent, sadly, among politicians, public and even some academics - about what constitutes “evidence”. That view says that evidence about practice in a field such as education must come from research of a particular type, not from practitioners who understand the multifaceted nature of success and failure.

Once those researchers have got themselves together, they produce and favour the kinds of study and the kinds of finding with which they are familiar. The philosopher of science, Thomas Kuhn, described how the process of garnering scientific knowledge, far from being disinterestedly rational, is a social one - it’s more about clubs than reason. Kuhn’s focus was physics, but Coles reveals how the process operates even more acutely in the educational community.

Once the “facts” have been established by clever researchers, it seems churlish, simple-minded, unscientific or just plain nutty to question them.

So classroom teachers are forced to swallow the received discourse, patiently ticking boxes for inspectors, while in reality quietly doing their own sensible thing behind their classroom doors. (Naturally, such a thing could never happen in this country.) This book is shocking in its revelation of the flaws in design and reasoning behind supposedly key studies on teaching reading, and the cabalistic quality of the panels and scientific institutes that arbitrate on the place of that research. Everyone should read it.

Gary Thomas is professor in education, Oxford Brookes University