Here's the American academic, L. Shepherd, writing about the problems connected with 'test-leveraged' reform in 1992:
- When test results are given high stakes by political pressure, and media attention, scores can become inflated, thus giving a false impression of student achievement.
- High-stakes tests narrow the curriculum. Tested content is taught to the exclusion of non-tested content.
- High-stakes testing misdirects instruction, even for the basic skills.
- The kind of drill-and-practice instruction that tests reinforce is based on outmoded learning theory. Rather than improve learning, it actually denies students opportunities to develop thinking and problem-solving skills.
- Because of the pressure on test scores, more hard-to-teach children are rejected by the system.
- The dictates of externally mandated tests reduce both the professional knowledge and status of teachers.
(Shepherd, L. 'Will National Tests Improve Student Learning?’
CSE Technical Report 342, CRESST, University of Colorado, Boulder. 1992)
Shepherd's points are even more pertinent today over twenty years later, on the day that Nick Clegg announced plans to rank every student in primary school according to their test scores. I'm reading 'Beyond Testing: Towards A Theory of Educational Assessment' by Caroline Gipps (1994) and finding that Gipps' research is just as relevant now as it was back then. Gipps' central problem with external exams is that they are often not 'valid' or 'reliable'. Many standardized tests today -- particularly those which are multiple choice -- are based on psychometric testing developing by psychologists to test innate intelligence in the 1920s and 1930s. Crucial to the development of a good test was that it must have 'construct validity'. Gipps writes:
"In test development the construct being assessed is defined before the test is developed: this is to make sure that the test assesses the attribute that it is supposed to, that it is ‘valid’. In the case of reading a detailed definition of the ‘construct’ reading would include accuracy and fluency in reading both aloud and silently, comprehension of material, interest in reading etc. Thus a test which had a high construct validity (i.e. which actually assesses reading adequately) should address each of these aspects of the skill. In fact, standardized tests of reading tend to assess only one aspect of the reading skills, for example, comprehension of simple sentences. This means that such a standardized reading test score does not represent the individual’s ability to read in the widest sense, and therefore that the meaning of the score cannot be universally understood (since the user of the score would need to know which aspect of reading had been tested)." (p. 6)
We can see then that a standardized test on reading which is based on a pen and paper exam can NEVER have 'high construct validity' because of the many skills it fails to assess: the ability to read out aloud, the ability to read longer texts, the ability to discuss a text in a spoken situation and so on. This sort of research needs to be read by Gove and Clegg because they just don't understand how they are setting up YET ANOTHER situation where invalid tests will be delivered to our children. And we haven't even got onto the issue of 'reliability'; many reading tests, particularly those with essay style answers are unreliably marked.
We need to move to a situation where children are being measured regarding 'how well' they are doing rather than 'how many' points they are scoring on a test. Gipps writes:
To find out ‘How well’ rather than ‘How many’ requires a quite different approach to test construction. Wood’s ((Wood, R. (1986) The agenda for educational measurement’ in Nuttall, D (Ed) Assessing Educational Achievement, London Falmer Press) definition of educational measurement there is that it:
- Deals with the individual’s achievement relative to himself rather than to others;
- Seeks to test for competence (NOTE: means attainment/achievement) rather than for intelligence;
- Takes place in relatively uncontrolled conditions and so does not produce ‘well-behaved’ data;
- Looks for ‘best’ rather than ‘typical’ performance
- Is most effective when rules and regulations characteristic of standardized testing are relaxed;
- Embodies a constructive outlook on assessment where the aim is to help rather than sentence the individual."
Rather than getting bogged down in thinking about one-day strikes that will just alienate parents, the WHOLE profession needs to start considering a serious boycott of these wretched high-stakes tests which will lower standards and worsen the life-chances of our students. I spoke about this to the Tower Hamlets' Youth Parliament last week. Some young people were interested in pursuing this. My thought is that if a boycott is threatened ahead of time -- ie two years before Gove's new GCSEs are introduced -- then maybe they'll think again about introducing them. There's no doubt in my mind that these exams are unfair and divisive; they'll be far worse than our current GCSEs, which far from perfect, at least give students a chance to show a wider range of skills in their coursework. Gove's proposals to make exams the sole arbiter of student attainment, without the possibility of re-sitting, are really quite scary in their implications.