The evidence base for exam results is becoming less valid as the system moves back to wholly end-of-course testing, says Harry Torrance
There are good reasons, rooted in traditional assessment concerns for validity and reliability, to involve teachers in setting and marking national test work in their own schools (coursework, project work and so forth).
Validity demands that the pursuit of broader curriculum goals such as analysing data, applying knowledge and developing practical skills be underpinned by broader methods of assessment. These wider skills and abilities cannot be tested by written final papers alone. For example, final papers can test knowledge of how to conduct an experiment, but not the actual practical skills involved or the collecting and recording of data over time.
Equally, reliability demands that these and other skills and abilities should not simply be measured by a one-off test, but assessed on several occasions over a longer period: the larger the sample of assessed work, undertaken under a variety of conditions, the more reliable the result is likely to be.
Now, however, history, experience and good educational practice are being set aside as the Conservative government moves back to an entirely final exam-based system. The argument is that the previous Labour administration allowed too many flexible teacher-assessed elements into school exams, lowering educational standards and inflating pass rates.
Final papers can test knowledge of how to conduct an experiment, but not the actual practical skills involved
Yet teacher assessment has been a key element of education for many years under Conservative and Labour, while pass rates at GCSE and A-level have risen consistently under both parties since the 1970s.
Given that these upward trends have extended over so many years, there is likely to be some element of a genuine rise in standards driven by the better socio-economic conditions of students, higher expectations of educational outcomes by students, parents and teachers, and better teaching underpinned by better training and resources.
More recently, however, this trend has been combined with and compounded by an increased focus on passing exams because of the perceived importance of educational success for school accountability, teacher career progression and student life chances.
Research evidence indicates that the pressure to raise results at almost any (educational) cost is a key driver of grade inflation. Thus, identifying a possible problem of grade inflation is one thing; assuming that eliminating coursework and teacher involvement in assessment is the only solution is quite another.
Pursuing new curriculum goals demands new forms of assessment to report grades with validity and reliability – coursework, fieldwork, oral work and so forth can capture different outcomes from end-of-course written tests.
The evidence base for published results is becoming narrower and less valid as the system moves back to wholly end-of-course testing
When we also add in ideas about formative assessment and changes in pedagogy – including students drafting work, receiving feedback on it, and then redrafting it for final submission – we produce a potentially positive situation in which students can be supported to develop their knowledge and understanding of subject matter over time and produce their best possible work for exams.
In principle this should constitute the core of any attempt to broaden and raise “educational standards”. However, in a context of intense accountability such practices can lead to little more than coaching students to meet exam criteria, thus undermining the validity and credibility of results. Yet improving the validity and reliability of teacher assessment is possible – this is what assessment policy, research and development should be trying to achieve.
As accountability pressures increase, the evidence base for published results is becoming narrower and less valid as the system moves back to wholly end-of-course testing. Instead, policy should:
(i) decouple accountability measures from routine student assessment and address the monitoring of standards over time by use of specifically designed tests with small national samples;
(ii) re-conceptualise the development of educational standards by starting from the perspective of the curriculum: ie, put resources and support into rethinking curriculum goals for the 21st century and developing illustrative examples of high-quality assessment tasks that underpin and reinforce these goals, for teachers to use and adapt as appropriate.
A longer version of this article is available in the British Journal of Educational Studies, 2017.
Harry Torrance is director of the Education and Social Research Institute, Manchester Metropolitan University