Faculty of Education, University of Cambridge

University of Cambridge Local Examinations Syndicate (UCLES)

Monday 27th May 2002, 10.30-12.30, Furness Lodge, Regent Street, Cambridge

Keith S. Taber Ó 2002

Dr. Keith Taber
University of Cambridge Faculty of Education
Homerton Site
Hills Road
Cambridge CB2 2PH
http://www.educ.cam.ac.uk/staff/taber.html
kst24@cam.ac.uk
keith.taber@physics.org

I was very pleased to be invited here today to talk to the Research and Evaluation Division about my work. I was also somewhat apprehensive on two counts: firstly that what I have to say about learning in science may be too parochial to be of value to colleagues here, and secondly (and somewhat in contrast) that anything I say about assessment may be seen as rather naive and ill-informed!

However, the theoretical background I use in my own work is somewhat eclectic, drawing on ideas from a range of fields, and as you have been kind enough to invite me, I will trust that (in the same spirit) some of what I say may usefully connect with your work here in UCLES.

My background is in science education, and I am aware that not everyone here today is likely to be a scientist. The area of research I will be talking about is (not exclusively, but largely) from within science education, as are my examples; but I believe that the relevance of the ideas is far wider, and so I hope the significant science bias in the presentation will be acceptable.

I will begin by considering a perspective on learning in science which has been very influential in recent years, both in some aspects of teaching practice, and in particular in terms of research. This approach, often called constructivism, is not well-defined (despite much literature on the topic) and so the version I will present has my own flavour and should not be seen as an authoritative definition of the field.

One of the main practical outcomes of this area of work is the recognition of the importance of eliciting learners' ideas, and this clearly connects with assessment, and - in particular - diagnostic assessment. I will discuss this topic, and I will refer in particular to a Royal Society of Chemistry funded project to support teachers in this aspect of their work.

However, I also feel that there are two other aspects of this area of research which are of particular significance for assessment. The first of these relates to notions of 'scaffolding' learning, and the concept known as the 'zone of proximal development'. I think this topic is especially relevant to the ways that formal (summative) assessment questions have changed over recent decades.

The other theme relates to much of the criticism that the examinations system currently faces from teachers (as anyone who subscribes to teachers' discussions lists will be well aware), and that is the way that science teaching in this country is often considered to be 'assessment-led', or at least 'highly assessment-constrained'. Research suggests that the image of science that learners develop during their schooling is very distorted, and part of the blame is sometimes considered to relate to the way we assess students.

At this point in time the constructivist view of learning seems little more than common sense, so perhaps I should begin by briefly considering the alternative. In this view (which now seems ridiculous, but seems to have been the tacit theory on which education operated for a long time, and which is still evidenced in much of the behaviour of many teachers) the learner starts as a tabula rasa, a blank slate on which impressions may be made. The role of the teacher was to transfer accepted knowledge to the learner through teaching (e.g. dictating or presenting on the board a set of accurate and comprehensive notes). Providing the learner was present, attentive, and able to receive the communication (e.g. within hearing range, able to see the board, given time to write everything down neatly), then learning should be possible - presumably by the learner reading through their notes a sufficient number of times before the examination.

This seems such a naive approach that as I write these words I even begin to doubt it was ever this way: but then on reflection I suspect there is still much teaching which in practice is still built upon such a model. Certainly the National Curriculum (NC) implies that there is a great amount of knowledge to be acquired by students: surely too much content for teachers to encourage students to take a reflective approach to developing understanding in science?

What is missing in such a description of learning is any notion of the active role of the learner. Perhaps this reflects a behaviourist influence, where anything inside the head that could not be observed and measured should be ignored.

Yet if this model had any validity teaching would be easy, and motivated students should all score close to full marks on formal assessments. In practice we know that even the keenest learners do not always easily understand difficult concepts, and that much of what is 'taught' is not 'remembered' at the point of assessment.

I put the word 'taught' in inverted commas as there is an adage in education that teaching has only taken place when learning occurs! On that standard much of what teachers do needs a different name: presenting perhaps? So, much of what is presented by teachers is not 'remembered' at the point of assessment.

I put the word 'remembered' in inverted commas for a different reason, which is that this must here be seen as a generic term from a number of processes. At one level there is the simple distinction between 'recall' and 'recognise' - i.e. it is easily to acknowledge that something is familiar than it is to access it from memory when it is not being perceived.

This is just part of a much bigger question of what is involved in remembering. The distinction between recognition and recall illustrates the significance of the contextual cues provided when eliciting information (an important theme that will be returned to later in this paper), but also touches upon the question of the extent to which memory is a process of knowledge reconstruction rather than straight retrieval of stored information.

In other words there is no doubt that we store information in memory, but there is much uncertainty about how we achieve this. We do not know at what 'grain size' the information is stored, or how associated knowledge is connected within cognitive structure. It could be that some complex information is stored as a unified complex, or that it is stored in modular form, and needs to be reconstructed when memory is accessed. As the 'associations' between items [sic] in memory may be of different and variable strengths such a distinction may suggest a dichotomy between the two possibilities, when the actual situation may be better seen as a continuum.

As this may be an important question for those working in assessment it is perhaps only fair that I acknowledge at this point that I do not have a definitive answer to that question, but my own view based on my reading and my own experience as a teacher and researcher is that the real question should be about the extent to which any particular act of memory is a process of knowledge reconstruction rather than retrieval of stored information.

Where data is habitually presented together and perceived to have a strong and constant relationship it seems likely that the memory trace reflects this (so that young people today are more likely to associate Nelson with Mandela than with Horatio). But we are also able to solve problems by forming novel associations between different stored items, so that a student can use memory to work out the shape of the PCl₃ molecule even if that is not one of the standard examples studied, by accessing separate stored information about the model used to assign shapes to molecules, and about the specific molecule. (And if the student has never thought about the structure of that particular molecule then that may first have to be compiled from other knowledge about the periodic table etc.)

I should point out that although these examples suggest that memory is being used to construct a new association, there is much evidence that even when we 'remember' familiar events we are actually reconstructing the memory from stored fragments, and filling-in missing details to give the memory overall coherence. (It is no wonder that witness testimony in court cases is so often found to unreliable: we often seem to have clear memories of events, quite unaware of the extent to which missing details have been compensated for with best guesses.)

This consideration of remembering provides a useful link with the problem of how we form memories of course material in the first place. The constructivist view sees the learner as having to construct knowledge (thus the term constructivism!) rather than absorb knowledge wholesale. This does not mean that a student will not come to an acceptable understanding of what is meant by capacitance, oxidation or photosynthesis (or watershed, verb, reformation, diminished fifth, or unemploydestagflation, etc.) - however, the student will not acquire an acceptable meaning of some concepts wholesale, ab initio, by a discrete act of learning.

Learning is restricted in at least three ways:

a) only a limited amount of material can be kept in mind at once - information that 'seems' complex may exceed this 'processing capacity';

b) for new information to be 'meaningful' it has to be understood to be relevant to prior learning;

c) permanent memories are only laid down hours after the initial presentation, and substantial restructuring of knowledge may take place over much longer time-scales (months and years).

If it seems (sic) odd for me to refer to how information 'seems complex' then this is because what is significant is not how complex information may seem to the expert (e.g. the teacher, much more familiar with the ideas, and already aware how they fit into the wider scheme), but the complexity of the material as it appears at 'the resolution' of the particular learner.

Given such a perspective it seems less surprising that teaching is such a skilled craft, and that failures of learning/teaching are so common (and thus the need for exam boards to set pass marks and grade boundaries!)

The other consequence of this view that has been the centre of much interest within science education is the importance of learners' ideas about a subject. The constructivist view on learning emphasises how meaningful learning (cf. rote learning) occurs when the learner associates the new information with something seen as relevant within their existing knowledge. The learner's existing cognitive structure acts as the substrate or substratum for new learning. Meaningful learning occurs when the learner is able to anchor new information to that bedrock of existing knowledge. As the body of existing knowledge may include various alternative conceptions, and as new information may be judged to be relevant to prior learning in unintended ways, such learning need not be appropriate just because it is meaningful.

For one thing, people are learners per se, not just when placed in classrooms. We all learn all the time. Young children actively explore and interpret their worlds and develop knowledge from a very young age.

For example, children commonly develop a view of force and motion which is heavily based on experience, but unfortunately not well matched to scientific knowledge. Whereas experience of the world says a force is needed to maintain motion, school physics says otherwise, and the physics teacher has to somehow overcome the students' tendency to rely on their long-standing, and well evidenced, view of the world.

'Common sense' and interpreting the world around them, 'teaches' children that plants grow in the soil, and are sometimes 'fed' by having plant food or fertiliser added to the soil. The soil also needs to be watered if the plant is not to die. Clearly then the growth of plants is due to them drawing materials from the soil. Unfortunately the school science version is a little different, with a key role given to the absorption of carbon from the atmosphere - something for which there is no support from everyday experience.

As a child's life is not a carefully planned set of experiences designed to bring them to curriculum knowledge it is hardly surprising that they exhibit a wide range of alternative conceptions such as 'constant velocity implies constant force' and 'the matter in a tree comes from the soil'.

Things do not necessarily get easier when the child reaches school age, as they are still subject to many other sources of information (including family, peers, media etc.) which need not be fully accurate, and they are now making sense of new information by interpreting it in terms of what they already think they know!

Considering what has been outlined it would seem that effective teaching requires the teacher to:

a) analyse the logical structure of a topic and to identify all the prerequisite knowledge that is required;

b) check that the learners have already learnt conventional versions of that prerequisite knowledge;

c) introduce the new knowledge (i) in ways that clearly build upon the prerequisite knowledge, and (ii) at a rate that does not overload students.

This is a tall order when working with a single learner, and becomes close to an impossible task with a class where each individual student has distinct background knowledge (each including some idiosyncratic alternative conceptions) and where learners are able to process information at different rates and in different sized aliquots!

However, this prescription does emphasise that it is important for teachers to have a good knowledge of what prior learning can be assumed when teaching any topic. (And you will recall my own comments that presaged this paper: I have had to make assumptions about this audience's prior learning. To the extent that I have guessed wrong you will either be confused or bored, or you may even be experiencing a mixture of confusion and boredom!)

During initial training science teachers are now taught the importance of pupils' conceptions, and are encouraged to actively check out prior learning at the start of a topic. It is considered that time spent on this activity should be well-spent, hopefully avoiding some of the worst inefficiencies of repeating familiar material (without developing it further) or talking over the heads of the class.

In research idiographic methods of exploring learners' ideas have often been found to be more effective than survey-type approaches. The semi-structured interview is the technique of choice for the alternative conceptions researcher, but such approaches are too time consuming to be used formally by classroom teachers. (Of course, much informal formative assessment in the classroom is through teacher questioning which is similar in nature to much research interviewing - with one very important distinction which will be considered later.)

Classroom teachers need to have available other techniques to elicit pupils' ideas - techniques that are workable with classes of 30 or so. I think there are two classes of response to this problem. I say this because there are some common clusters of alternative concepts that are likely to be present in most classes, and for these it is possible to have centrally produced diagnostic instruments that can be made available to teachers.

However, it is known that there is a seemingly inexhaustible creativity in the manner in which students are able to present conceptions not previously uncovered, and some of these ideas do seem to be idiosyncratic. So teachers need approaches they can apply, as well as prepared instruments.

My own research into developing understanding of the chemical bonding concept revealed a range of alternative conceptions, some of which seemed to be very common.

For two of the most common areas of misunderstanding (ionic bonding and ionisation energies) I prepared written diagnostic instruments that could be used by classroom teachers as simple paper-and-pencil instruments. These were designed by taking examples of students' comments from research, and using them to write statements about the concept area from the perspective of common alternative 'frameworks' as well as from the perspective of curriculum science. Students are asked to judge the various statements as being true or false (with an option of responding 'do not know') - an exercise which can be followed by discussion and appropriate feedback.

These materials were designed for a particular purpose - to survey the prevalence of alternative conceptions previously elicited from interview research. Had these instruments been primarily intended as research tools then it would have been appropriate to have used a more sophisticated design (perhaps asking students to explain their responses in their own words). The diagnostic instruments could be used as summative assessment tools at the end of a topic, but they would not be particularly suitable. Besides the presence of the 'do not know' option which could (depending on the scoring system used) bias results towards the more confident over the more knowing, there is also the presence of distracters which are incorrect but known to be seen as particularly feasible - and so such a test might be seen as a kind of intellectual entrapment!

My experience in this area led to me working for the RSC (Royal Society of Chemistry) on a funded project to help teachers find ways to respond to students' alternative conceptions in their lessons. Part of the outcome of this was a text about learning in science to present the background to the work in some detail.

However, the main product desired by the RSC was a set of classroom probes which could be used by teachers. These were designed to relate to a range of key areas of school and sixth form chemistry where research suggested there were common alternative conceptions. Most of the probes were at the level of paper-and-pencil exercises, provided with teachers' notes (and suggested answers).

In order to help teachers develop general strategies for diagnosing common alternative conceptions into their teaching repertoire (as it was not possible to cover all combinations of topics and levels), and to allow teachers to elicit less common alternative ideas, some of the activities provided were suitable for more open-ended use. So, for example, concept mapping was modelled. Concept map activities were provided for two topics ('acids' at KS3 and 'the periodic table' at KS4) but the intention was that this would empower teachers who were not already familiar with the technique to adopt concept mapping with other topics.

What is clear about the method used in the RSC project (and in other related approaches such as using concept cartoons, or word-association techniques) is that the purpose is very much assessment-for-learning. Such materials are intended to elicit aspects of pupil thinking, so that these ideas can be challenged, or used as the starting point for developing curriculum models of knowledge.

Traditionally exam boards are more concerned with summative assessment - assessment which is designed to provide a measure of achievement at the end of a course of teaching/learning. At one time such assessment tended to be largely limited to the end of formal schooling (e.g. at age sixteen) and at university entrance level. It was seen important for there to be some kind of grading of student achievement at these points, and no doubt much teaching was directed at the examination during the lead up to these two points. In principle there could be a much more relaxed attitude to teaching lower down in the school, where wider aims of teaching and a long-term view were appropriate.

The system in this country has now changed so much, so that we seem to have an assessment-oriented culture in education. It seems that anything that can be measured and rank-ordered, is. National testing of pupils now takes place from a young age, and much of this assessment seems very ambiguous to me now. At one level the assignment of NC grades at KS2 or KS3 is intended to provide information for teachers to help them match work to students, and judge their progress. However, if SATs are to be seen as formative assessment tools then we do not want teachers 'teaching to the test' and it is totally inappropriate that such information is used to rank-order schools. By using assessment data in this way it ensures that pupils are subjected to formal, effectively formative, assessment at these points, with all that entails!

A content-filled curriculum, and the imposition of tests that are used to judge teachers and schools, are inevitably going to focus the minds of teachers (which is presumably part of the intention). Unfortunately they focus teachers on trying to transfer the most knowledge into pupils heads, or - given that we know that there is just too much content to be meaningfully learnt by many pupils - at least to transfer as much as possible of the knowledge that is expected to be in the test. No matter how conscientious a teacher is, and no matter how well informed about the way in which students learn (slowly, incrementally, organically), she is very brave if she decides to teach for the deepest, widest understanding of scientific idea rather than try to optimise SATs scores. We now seem to have a culture where anything other than year-on-year improvement (in scores) is considered a failure.

A similar point can be made about the assessment of practical work in science. The science curriculum contains four areas, that approximate to biology, chemistry, physics and the scientific process. Practical work in science can perform many functions, and illustrating the scientific process is an important one as there is good research evidence that this is a poorly understood area. Students have 'alternative conceptions' of science process as much as of science content, and these misconceptions also need to be challenged and developed.

Yet when the NC includes a prescribed, reductionist, way of assessing this area of science education, then the inevitable happens. How much school science practical work is undertaken with at least 'one eye' on the need to train students to jump through assessment hoops in 'investigations'?

The exam boards do not set the agenda here - but they are part of the mechanism by which the central agenda is converted into - very dubious - classroom practice.

Another key notion is that of scaffolding learning. This derives from the work of Vygotsky who wrote about the 'zone of proximal development' (ZPD). The ZPD describes the part of an abstract problem space where a learner could achieve with help from an adult or more advanced learner, but could not succeed unaided. Vygotsky believed that assessments made within the ZPD were more significant than more traditional tests (such as IQ tests) where the student had to work unaided. Vygotsky thought that it was more informative for the teacher to know about what a student was in the process of striving to achieve than to just know about what the student could already do. We seem to be back to the difference between summative and formative assessment again, and this is significant - so I will return to this theme.

First however I would like to point out that the concept of the ZPD fits nicely with some of our other ideas about learning. For one thing we have noted that the human brain has a very limited processing capacity (the 'scratch pad' of working memory). Many tasks are beyond students because by the time they have set about organising how they are going to carry out the task, they have already overloaded their working memory!

Yet if the teacher is able to support the learner by taking some of the responsibility for organising the task for her, then the student may well be able to make new logical connections and develop further understanding than before. As part of my RSC project I tried to explore some ways in which teachers could put these ideas into practice in terms of helping students organise their knowledge base and progress to new insights (through devices labelled as scaffolding PLANKS and POLES). Following Vygotsky's insight, it seems that there is little point just getting students to undertake work that is clearly within their current competence - that may provide practice and lead to a more efficient performance (greater speed, lower error rate), but it will not extend the bounds of the learner's competence. If we wish to do that (and surely that is exactly what we do wish to do), then we need to have our students working 'in the zone'.

Now this may seem to be of limited value if the student loses the ability to demonstrate this knowledge as soon as the teacher's support is removed. But this need not be so: as the student explores the domain with the teacher's support they should be making new connections, and developing a better integrated representation of the topic. In other words - they are learning!

Now we may use some simple metaphors to consider what this might mean. So new knowledge may be fragile (and even labile), but when the student has the chance to explore new ideas, and relate them to existing knowledge in more substantial ways, it will become robust and eventually make suitable foundations to support further learning. As the learning becomes more robust it develops sufficient inherent structural integrity for the scaffold to become superfluous.

The other metaphor relates to the notion of seeing subject matter 'at the learner's resolution'. Although we all have very limited 'mental scratch pads' they must be measured in relative terms - for new information cannot be judged in absolute terms (so if the brain uses something like binary or hexadecimal code, then our references to 'scratch pads' is modelling the information processing at a higher level of analysis).

For example, something like CH₃CH₂CH₂COOH or 1s²2s²2p⁶3s¹ may appear to be 'complex' data, and indeed either would place significant cognitive load on a novice who is trying to hold them in mind. Any attempt to operate on or with the information is likely to overload capacity.

However, for someone who has the necessary prior knowledge CH₃CH₂CH₂COOH is butanoic acid and 1s²2s²2p⁶3s¹ is a way of representing a sodium atom. For someone who recognises the symbols holistically in this way the information will not overload processing capacity and can be related to other information as needed. The information has a different complexity and makes different demands at the resolution of the novice (high) and the expert (low).

So during the learning process - and I'm not talking about during the 'teaching episode', but over a much longer time-scale - the learner becomes able to 'chunk' information so that it can be organised and processed more efficiently. So, rather than using all of the available 'working space' rehearsing C-H-₃-C-H-₂-C-H-₂-C-O-O-H she is able to keep [CH₃CH₂CH₂COOH] in mind and operate on it in relation to other information. At this point she no longer needs the teacher's scaffold which was designed to act as an 'add-on' memory buffer and extend processing capacity.

The teacher scaffolds learning to enable the student to achieve with support what they cannot yet achieve without, and then gradually reduces the support as the learner is able to operate without it.

These days such questions no longer tend to feature in examinations. It may well be that an examination question may be set which in principle requires the same knowledge. However, the question will be structured - it will be broken down into, perhaps, 5 parts, each of which provides significant cues as to what is required. The candidate will probably be able to proceed to later parts even if unable to answer earlier parts. If examinations have changed in this way, then in one sense they are clearly easier because they make less demands on candidates. It is argued that this is not so, simply that the examination questions are now designed to find out what the candidates can do, rather than what they cannot do.

I sympathise with this aim, and I am not suggesting that either the aim, or the way in which questions are structured, is inappropriate. However, we should not conclude that the questions are of the same level of difficulty as the old-fashioned questions (when an O level biology paper fitted on one side of an A5 sheet). In GCSE papers, for instance, there are often many marks that could be gained by any intelligent person invited in from the street who has never studied the course. The nature of the questions, and some experience of the culture of answering examination questions, provide sufficient cues for an intelligent guess. Even in the 'gold standard' A level examination the questions help candidates considerably more than twenty years ago. (And I have seen A level chemistry questions about spectra that can be answered by anyone who is familiar with interpreting graphs without any knowledge of what the question is about!)

I do not intend this as a criticism, but as an observation. In teaching we are often concerned with providing support as part of a process of helping the learner move beyond what they can currently do. Teachers' questions are often designed to teach (unlike the researcher's interview questions which are designed to elicit), and often our worksheets are designed to lead students through a process to get a 'correct' answer.

So what are we intending to do in examinations? I do not think the response 'letting students show what they know and can do' is entirely satisfactory - because what they know and can do clearly depends upon the context of the way we structure and present the question.

This brings me back to an earlier point. A question that asks a candidate to recall the electronic configuration of silicon, or draw the circuit symbol for a thermistor, or label the components of the eye, or suggest a procedure for separating sand and salt is very different from one that presents some options and asks the candidate which configuration is correct for silicon, or asks which symbol stands for a thermistor, or which labels represent which eye components, or perhaps to sequence the steps in separating sand and salt. The latter questions are inherently easier because they ask for recognition rather than recall, and because they narrow the scope of the answer to a modest number of options to be considered rather than a potentially infinite set of possible responses.

So what is the purpose of the assessment? Perhaps recognising the correct answer is a more realistic assessment context in terms of some real life situations - but much of the knowledge we test is readily available in reference books, so asking candidates to look up the answers might be even more appropriate? Again I do not have the answers - but I do feel we should be aware that different formats of question make different demands upon our candidates.

This brings me to briefly comment on another aspect of the context of assessment. It was reported above that students often develop alternative conceptions from their experiences of the world, and these contradict the models presented in science teaching. It has been suggested that some students manage to learn, and 'store' the school science version separately so that they have both everyday 'life-world' and distinct 'scientific' versions of the concept. These two alternative versions are stored in different domains, and are activated by different cues. So perhaps when kicking a football around the playground with friends the student operates with the 'a force is needed to keep the ball moving' principle, but in the physics lesson applies the 'a force will cause the ball to accelerate' scheme.

I have no doubts that students do store information in separate domains to some extent (so that, for example, something that they know and use in physics is not brought to mind in a context which is judged to be chemistry!) Indeed I have found when interviewing science graduates for teacher training that many have considerable difficulty explaining how the carbon in charcoal got into the wood in the first place: although these same graduates are able to explain about photosynthesis when that is the question. The technical term 'photosynthesis' acts as a cue that activates stored knowledge that is not often activated when the question is in the more concrete context of the stick of charcoal.

This would seem to be a significant area for examination boards to consider and explore. It is often considered that candidates are helped if a question about some aspect of scientific theory is set in the context of an everyday example. However, this may actually cue some candidates to activate an everyday scheme for the topic, where a more abstract formal question could well activate the target knowledge. If students are taught about the concept in terms of abstract uniform bodies and frictionless surfaces, and then asked about a steel ball bearing rolling over a polished wooden floor they are provided with the additional task of decoding the context to give the abstract form of the problem.

Providing a context may help some students who need a concrete situation to visualise, but could disadvantage others who are able to operate with the abstract problems. Again, I do not suggest what is or is not appropriate for examination boards, but I do feel this issue needs some in-depth investigation.

However, I would like to make one plea to close this paper. It was explained above that learners' alternative conceptions do not all derive from intuitive interpretations of experience, but may often be due to things they have read or been told. The source of this misinformation may be peers or family or the media - but often it can be textbooks or teachers. For an example I will briefly mention the case of ionic bonding. Many students hold an alternative model that is not just a simplified version of the acceptable science, but rather is simply wrong.

The incorrect ideas about ionic bonding that I identified in my own research were found to reflect the presentations in many school text books. I will not go into detail here, but as a single exemplar ionic bonding is not the transfer of an electron between two atoms, and should not be represented by a diagram showing an electron being transferred between two atoms! Yet I have seen examination questions that ask pupils to represent ionic bonding in terms of electron transfer, and award marks accordingly.

This false idea is part of an alternative framework for conceptualising the ionic bond which is very common (no doubt in part due to the efforts of authors, teachers and examiners!) and which acts as an impediment to progression in understanding the topic at A level. To my mind, examination boards should be looking to award credit for appropriate scientific ideas, and not perpetuating incorrect ideas that confuse students and impede the development of scientific understanding. Research shows that once acquired, such over-simplified models (and in this particular case, blatantly poor science) may be difficult to overcome.

Many of the alternative conceptions held by students are tenacious. Where they have the status of being validated by examination board mark schemes then they are likely to be very stable, and to act as significant blocks to further (more appropriate) learning. Teachers will feel that they can best help students by teaching to the test, even when they know the science is poor, and that any students taking the subject further will then have to try and 'unlearn' established stable knowledge. I would like to see examination boards take the high ground in such situations and always set questions based on acceptable scientific models.

Taber, K. S. (2000) Chemistry lessons for universities?: a review of constructivist ideas, University Chemistry Education, 4 (2), pp.26-35, available at http://www.rsc.org/uchemed/uchemed.htm

Taber, K. S. (2001) Constructing chemical concepts in the classroom?: using research to inform practice, Chemistry Education: Research and Practice in Europe, 2 (1), pp.43-51, available at http://www.uoi.gr/conf_sem/cerapie/

Taber, K. S. (2002) Chemical misconceptions - prevention, diagnosis and cure, 2 volumes, London: Royal Society of Chemistry (ISBN 0-85404-390-X)