In other words, falsification of a theory or hypothesis is not the end of an enquiry, but a prompt for further investigation and more refined explanation. While it is wrong to put poorly evidenced empirical theories into practice, it cannot be an objection to EER as such that it is fallible and often falsified. In this respect it is much like other forms of scientific enquiry. As we shall see, however, in later chapters, there is a need to have a clearer account for what counts as criteria for truth in educational research and also for the standard of warrant needed (and the necessary qualifications) for accepting that research has a bearing on educational practice.
A further issue of great importance for our investigation concerns the scope of empirical theories, or the contexts in which they are true or false. Why is this an issue? The brief reason is that social practices are not governed by regularities which have law-like features in the way in which natural phenomena are. Adopting the language of efficient causality, we could say, following Mackie (1965, pp. 248–249) that social reality is not a uniform causal field and that it is necessary to carefully specify the background factors which set the scene for an explanation in terms of causes and conditions. Thus one might reasonably assume that non-biological physical reality forms a unified causal field in which the same laws apply irrespective of local circumstance. As Mackie points out, we need to be pragmatic about what constitutes a causal field relative to any particular enquiry. We would need to be more circumspect in terms of biological reality; we cannot assume that the same range of causal factors is operative in animal as in human biological systems for example. In relation to seeking an explanation of the causes of a house fire, the causal field would be constituted by the history of the house and which features constituted the background and which potentially causally active conditions would be a matter for the investigators to determine in accordance with their own explanatory priorities. The same point applies with even more force for reason-type as opposed to causal explanations (see Ch. 4).
Before we leave the issue of falsity we need to make clear the distinction between contrary and contradictory theories.10 Two theories contradict each other if the truth of one implies the falsity of another and vice versa. Thus for any true theory P its negation not-P is its contradictory and false and not-P is the contradictory of P. However, a very common situation is that various two (or more) explanatory theories are all false, even if the truth of one is incompatible with the truth of the others. Thus both P and not-P are inconsistent with each other and both can be false. In these cases, the falsification of an educational theory need not tell us anything of interest about competing theories which are incompatible unless it is possible to discern common elements within them. It might, for example turn out that two false and incompatible theories both contained elements that were true. The presence of contrary theories is common in EER, but that should not be a reason to dismiss them entirely unless we are convinced that there are no true elements in either, or in what they hold in common.
The Problem of Context and How to Interpret It
It may be the case that an explanatory theory P turns out to be true in one context but not in another. Given what we have just said about causal fields this is often likely to be the case. So we have to qualify what has been said above to take account of context. If P is false in context C1 then not-P is true in C1. But it may turn out that P is true in C2 in which case not-P is false in C2.11 All explanations in educational research need to have a contextual limit placed on them. In order to explain this, it will be useful to introduce the idea of an explanatory field. An explanatory field is a causal field which admits a broader range of explanations than those allowed by efficient causality; in particular it allows for reason explanations as well. It is one of the great challenges of EER to determine what the relevant explanatory field is in determining the scope of an educational explanation. There can, unfortunately, be no simple answer to this question.
An explanatory field covers a range of situations in which explanations of a certain kind are valid. What we mean by a valid explanation will become clearer in the following two chapters. Causal fields are explanation relative. They are specified in terms of the phenomena to be explained. Thus, to use Mackie’s examples, a house and its history will be the relevant causal field when explaining the cause of the fire in the house. Exposure to a virus may have the causal field of human beings when investigating the conditions in which the viruses are present. However, when investigating the conditions under which the virus is contracted, we may restrict the field to those human beings who contract the virus. Sometimes the causal field can be very broad: explaining the influence of gravity on physical entities will take the known universe as its causal field, on the background assumption of uniform influence of gravity across it.
Explanatory fields of considerable generality cannot, however, be assumed when investigating educational practices and institutions. Not just the intensional nature of the investigation is important (the field is determined according to the scope of the investigation and hence the purposes and interests of the investigator), but there are also likely to be many different factors influencing outcomes in different contexts, including for example, cultural traditions, the labour market, attitudes towards education, income differentials, class or caste structure and political configurations.12 It may be that there are factors which could influence outcomes of which researchers are not even aware. We have to deal with the inescapable fact that contextual factors, often hidden, will limit the scope of explanatory fields when seeking to understand the nature and operation of educational practices and institutions, the main focus of EER.13 This does not mean, pace Barrow, that there can be no broad explanatory fields in EER, nor that explanations valid in one context may not be valid in another, but it does mean that there has to be great caution when postulating such fields and great care taken in providing valid explanations so as to eliminate hidden factors that could upset results. An alternative approach could be iterated investigations in a variety of different contexts, leading to the inductive building up of a relatively broad explanatory field.14
What are the issues that need to be taken into account when delineating explanatory fields? The first is researcher intention, which will specify the range of phenomena of interest. Thus an investigation of the efficacy of a method of teaching reading for young children may be quite general, and apply to all practices which use an alphabetical script. In such a case the explanatory field covers the teaching reading practices (and, probably, associated factors) in societies which use alphabetic scripts. It is more likely, however, that concern will be focused on one particular writing system, say English, in which case the causal field will be those teaching reading practices that involve the English-spelling system. It is also quite possible that the explanatory field will be teaching reading practices in a particular local authority, as in the West Dunbartonshire study already mentioned (MacKay 2006). This is by no means to say that other studies in other explanatory fields may not be drawn on in constructing an explanation in this case, but other results and explanations require interpretation in the context of a new explanatory field.
At the start of any investigation researchers usually have a good idea of what factors will be relevant. In the case of intervention studies this will involve determination of those known factors whose influence is likely to be important, but which is not yet known. These factors can be incorporated into a study in different ways depending on the methodological strategy employed. For example, a quantitative study of teacher effectiveness15 will take into account factors that are known to be likely to influence effectiveness (e.g. qualifications, training, class size, and characteristics of pupils) to an unknown degree, and these can be incorporated into the design. However, factors that are not known to be likely to influence teacher effectiveness are, by their nature, difficult to include in a study. The disturbing fact is, however, that they may be present and active. To some extent such worries can be alleviated by a successful empirical study.
One could argue that this should not be possible. Since the work of Fisher, R.A. (1935) experimentation has involved randomisation of a population sample prior to assignment to treatment and control groups. This supersedes the earlier procedure of controlling for all known