Education ministers argue that performance-related pay (PRP) rewards good teachers, makes the profession more attractive to good-quality graduates and raises standards of student achievement. There is one major flaw with this argument: it is not true.

There is no evidence, from any education system in the world, that performance-based pay improves educational standards.

There is evidence, however, that performance-based pay narrows the school curriculum because assessment of teacher performance is overly dominated by pupil performance in timed written tests. This results in a rote-learning approach to education - because if teachers’ jobs and pay rely on pupil test performance, they will, unsurprisingly, do all they can to prepare pupils for the tests.

Too often, what is lost in this equation, as President Obama acknowledged recently, is a love of learning, an open, enquiring approach to new knowledge, and a broad and balanced curriculum.

Even shakier is the politicians’ proposition that PRP rewards good teachers and keeps them in the profession, because it is simply not the case that good teaching is something that can be easily observed and measured. The truth is that teaching does not lend itself easily to categorical statements of “good” and “bad”.

Research produced by the Sutton Trust, What makes great teaching?, concludes that there is no framework for effective teaching that is not open to interpretation. Every recommended teaching and learning strategy could be done well or badly; all could be done appropriately or inappropriately; and “none should be treated as a recipe or a formula”.

Ofsted inspectors used to observe teachers - often for 20 minutes at a time - and then come to apparently confident judgements of teacher quality, upon which careers could be made or ruined.

Professor Robert Coe, director of the Centre for Evaluation and Monitoring, Durham University, demonstrated that Ofsted’s approach was deeply flawed. He highlighted research findings, which show that if lesson observation is used to identify “above average” and “below average” teachers and their impact on student learning, the judgements would be right about 60 per cent of the time. (This is a best-case scenario. When lesson observations are conducted by untrained teachers or school leaders, using unvalidated observation protocols, with no moderation of judgements or quality-assurance processes, the correlation between judgements and teacher quality is much lower.)

Assessment of teacher quality should be based on a wide evidence base, be undertaken by trained evaluators, and their judgements should be carefully calibrated and moderated. How often, honestly, can one say that this is the case?

Unable to answer the weight of the critique of lesson-observation data as a means to judge teacher quality, Ofsted has changed tack. Inspectors continue to make judgements about teacher quality partly based on lesson observation (though they no longer give individual grades to teachers), but these judgements are supplemented by a wider range of evidence, including looking at pupils’ work, and talking to them about their learning and progress. Ofsted also requires school leaders to justify their pay decisions, with teachers’ pay linked to leaders’ assessment of the quality of teaching.

The trouble is, if lesson observation data is deeply dodgy as a marker of teacher quality, then these other sources of evidence are even worse, because there is even less research evidence about whether they are valid measures of teacher performance. The ways of Ofsted are a mystery to us all, but one thing is very clear: Ofsted has no reliable or valid way to hold senior leaders to account for their pay decisions - and it should not be doing so.

All of which leaves the intellectual case for performance-related pay in tatters. And yet, when research tells us that pupils’ progression rates naturally vary greatly, hundreds of thousands of teachers are being given, as one of their appraisal objectives, hard targets for pupil achievement. We also know that the evidence upon which teacher quality is measured is, at best, deeply inadequate. Improved rates of teacher retention are absolutely essential if the teacher-shortage crisis is to be tackled.

How teacher quality is measured is not only a matter of justice, it is also an issue that is driving good teachers out of the profession.

Performance-related pay is in special measures.

Mary Bousted is general secretary of the ATL education union

*Read about our exclusive YouGov poll showing that many teachers believe their school has failed introduced PRP in Friday’s TES magazine.

Available in all good newsagents. To download the digital edition, Android users can click here and iOS users can click here