Educational Explanations. Christopher Winch

Explanations which relate to the functional role of institutions and practices will be very important for educational explanations, together with the closely related concept of a teleological explanation which relies on goals and purposes as an organising principle (see also von Wright 1971 for more on this). We will also be looking at the concept of a good explanation, making use of some of the insights of Lipton (2004). Although Lipton is primarily concerned with explanations which involve efficient causation his work has wider applicability. Finally, we will also be looking at the important role that context plays in both shaping and limiting the scope of educational explanations.

      Chapter 5 takes these concerns further, focussing on what a good educational explanation might look like. The importance of plausible methodological assumptions constituting a research tradition is emphasised, along with the purposive nature of research which entails the posing of questions that need to be answered by the research. Further issues dealt with include the following:

      Explanatory adequacy. How can we tell whether an explanation does actually address the question posed which gave rise to it? An example would include Bernsteinian educational sociolinguistics, discussed in Chapter 9.

      Coherence. To what extent is an explanation not only internally consistent, but also whether it is well-articulated through sound inference and appropriate transition from evidence to conclusion. We will also be concerned with the extent to which explanations can be externally consistent, with those offered for the same or related phenomena. Closely related to this is the property of economy – is the explanation overly complicated and does it contain features that do not really add to understanding of the phenomenon? See the discussion of dyslexia research in Chapter 12.

      Methodological probity. To what extent is there a strategy for answering a research question or line of inquiry and the appropriateness of the techniques used for answering it and the explanatory concepts deployed to do so? See the discussion of School Effectiveness Research in Chapter 11.

      Appropriateness of methods. Closely related to this is the question of whether the array of techniques of data gathering and analysis employed actually work together effectively to produce explanations that can cohere. Some of the problems encountered by school effectiveness research illustrate this need.

      Evaluability and comparability. A very current concern is the extent to which the explanations offered can be properly evaluated for their quality and whether or not meaningful comparisons can be made between explanations of different but related phenomena. Questions of methodology, technique, purpose and context are all relevant here. Comparative research on vocational education, discussed in Chapter 10, is a useful example.

Knowledge cumulation. How well does educational research fare in building up knowledge? This vital issue needs to be addressed in an honest way which avoids the Scylla of excessive scepticism about the ability to build up a corpus of educational knowledge and the Charybdis of excessive optimism and careless overgeneralisation that has been the bane of much EER. Both Chapters 11 and 12 provide case studies relevant to this issue.

      Chapter 6 takes up the issue of knowledge cumulation and seeks to make a measured response. Issues taken up include the fact that different paradigms – in Kuhn’s sense (1962) – are often used, sometimes relating to the same phenomena. Here we need to consider both the possible incommensurability and incompatibility of different paradigms. The chapter also addresses the question as to whether there can ever be a common conceptual framework for the conduct of systematic educational enquiry. I go on to consider what makes particular research programmes progressive or degenerating – in Lakatos’ sense (1970) – and the relevance of Lakatos’ account for EER. This leads on to questions of the replication of research and the cumulation of findings within metasurveys and metaevaluations, together with the extent to which it is safe to recommend policy and practice arising directly from the research. In this context we also have to take into account Kuhnian-style revolutions in which large-scale conceptual revision and change become a feature of the educational research landscape and the extent to which such changes compromise cumulation.

      Chapter 7 broaches the broad divide so often made in EER between quantitative and qualitative methods in data gathering and analysis. The distinction is often made wrongly in terms of two different paradigms in the Kuhnian sense, but it is argued in this chapter that this is a misleading way of framing the issue. It would be more helpful to say that ‘quantitative’ and ‘qualitative’ refer to two more or less contrasting families of techniques used in hypothesis formation, data gathering and data analysis. It is important to see them both as primarily belonging in the ‘methods’ category rather than the paradigmatic or even the methodological category. Some paradigms will favour quantitative methods, particularly those that seek a close identification of the aims and approaches of EER with those of the physical sciences.

      It is important to note, however, that a refusal to abide by the canons of natural scientific research does not imply rejection of quantitative methods. By the same token, paradigms that reject close identification of EER with natural science cannot be taken to automatically reject quantitative techniques. The argument of Chapter 7 will be that research questions should generate a research strategy or methodology that in turn prescribes appropriate methods, without prejudice as to whether or not these fall into the ‘quantitative’ or the ‘qualitative’ camp. The role of both families of methods in the context of hermeneutic inquiries will be discussed there.

Chapter 8 focuses on a currently influential paradigm with a close relationship to approaches in natural science. This paradigm, which valorises intervention studies as a way of understanding educational practices, consists of a family of approaches, including natural experiments, quasi experiments and RCTs (Smith, H.W. 1975). Because the RCT has become so influential and has been held up as the gold standard of scientific probity in educational policymaking circles, it is appropriate to pay particular attention to its use, appropriateness and strengths and limitations. Particular attention will be paid to the structure of inference that underpins RCTs, to the causal mechanisms that can be assumed to underlie them, to the types of explanation that they sanction and to the relationship between significance testing for an intervention variable and a claim for the causal efficacy of that variable. Mackie’s (1965) introduction of the concept of a causal field will be introduced as a way of understanding some of the problematic features of the use of RCTs and will be discussed with some examples. Conclusions about the advantages and limitations of the use of RCTs and associated methods will be drawn, together with a reflection on the use of these in metasurveys and metaevaluations.

      Chapter 9 is the first substantial case study chapter and examines the rise and fall of the Bernsteinian sociolinguistic research programme (Bernstein 1973a, 1973b). This research programme has been chosen to illustrate the ambitions of EER in a time of considerable optimism about its potential to change educational practice, its relatively innovative approach to empirical research and the way in which it appeared to tie together various factors involved in educational achievement in a convincing looking explanatory framework. At the same time, the weaknesses of the programme, which eventually led to its abandonment, will be looked at. The included: inadequate conceptualisation of the key organising concept of linguistic code; ignoring of factors such as acquaintance with literacy; inappropriate intervention methods of data collection and a weak empirical basis for large-scale generalisations.