Why do we have such high-stakes exams for students?

Simon Burgess

A report published on Saturday 4^th July raised serious concerns about pupils developing stress-related conditions linked to taking high-stakes tests. It was reported that “The National Union of Teachers’ report says … most teachers surveyed for the report agreed pupils became “very stressed/anxious in the time leading up to Sats/public examinations”.”

Obviously, reports of mental illness and stress amongst pupils are very concerning and need to be taken seriously.

There are two things here – the nature of the school accountability system and the nature of the tests that school children take. I have talked about school accountability before on this blog, and there is very good evidence that when Wales abolished school league tables, pupil attainment fell dramatically. That is not the topic here.

The tests that pupils take in England at age 16 are very high-stakes. Does this have to be the case? Are there alternatives? Why do we make kids sit these high-stakes high-pressure exams at 16?

I think that there are two main alternatives: use teacher assessments, or just have no exams and no marks and each pupil simply leaves with a certificate confirming that they have completed a certain number of years of schooling.

Teacher Assessments

What about teacher assessments? They would be based on a full year’s work and would be capable of giving a rounded view of each child’s capabilities. There are, however, three big objections.

The first objection is simply pragmatic: workload. At a time when teachers responded in such force to the Secretary of State’s enquiry about workload and over-stretch, adding another substantial and complex item would be impossible. It’s true of course that teachers provide assessments for their pupils now, but this would have to be a very different thing if it summed up a child’s time in school and was the only thing they left with.

Second, imagine the pressure that teachers would come under. Their personal assessment of your child’s maths ability might make all the difference to her getting into Cambridge – you’re not going to talk to the teacher? It would make the key parent-teacher relationship very fraught. And we can hardly ban parents from talking to teachers.

Third, there are two sources of bias in teacher assessments, both very clear in the data. There is centrality bias: teacher assessments of children’s performances are much more tightly clustered around the middle than are remotely-marked tests. For example, looking at Keystage 2 tests, the variance in test scores is around 20% higher than the variance in teacher assessments in English and maths. This happens in all jobs where people are evaluated, but particularly in ones where there is an on-going personal relationship. Second, there are gender, ethnic and social biases that seem to arise from stereo-typing, which we evidenced here and discussed more recently here. These arise simply because teachers – like everyone else – unconsciously use stereotypes to form judgements. There is evidence for this in England’s schools from us here and from IOE here. There has to be a way for children to shine, regardless of what their teacher might think of them.

Leaving Certificates

If using teacher assessments is out, what about just having a leaving certificate, no exams, no marks?

The reality is that in life beyond school, people will still need to be able to differentiate between candidates with different abilities. And to differentiate both types of ability – languages, maths, art, sports – as well as levels of ability. If all pupils left school with the same certificate, perhaps just differentiated by completed years of schooling, then firms and universities and other organisations would have to do the differentiating themselves. They would all have to run their own tests (or subcontract them).

This would be much worse than the present situation. There would potentially be very many tests. Perhaps with every job interview and every university application everyone would have to take tests in English, maths, language, science, … And the job would hang directly on the test, rather than slightly indirectly now. This would be much less efficient, much less fair and much less meritocratic.

Less meritocratic because there would be considerable scope for favouritism in each organisation.

Less fair because, removed from the impartial bureaucracy of the school exam system, there would be greater scope for extra help, cheating and non-neutral marking.

Less efficient because having (essentially) the same exams for everyone at the same time, marked at the same time and to the same standard is an incredibly efficient way of distinguishing different levels of ability. Without these, the testing and differentiating of abilities would be fragmented, cumbersome and very variable.

These local, amateur, idiosyncratic tests would reduce productivity by reducing economic (allocative) efficiency: the wrong people would be in the wrong jobs and courses, and would spend a lot of time working their way back to the right ones.

And finally, if everyone’s school days did culminate in just an ungraded leaving certificate, it is hard to believe that there would be as much motivation for pupils to try hard at school. The love of learning cannot take you all the way and if you get the same certificate for working on your homework as you get for watching TV, then it’s likely that less learning will get done.

This is obviously not at all to say that our current exam system is perfect; far from it. But one can imagine, if we were in that highly inefficient world of very many independent bespoke tests, that it would evolve to something like we have now.

Compared with the alternatives, our system of high-stakes exams is more meritocratic and more efficient and, in the end, probably less stressful than any alternatives. This is of course not to say that there might be valuable ways of making the experience less stressful.