The first area of criticism concerns the extent to which the results of ‘scientific’ educational research are valid. It has often been argued that, although the numerical evidence produced by such research has the appearance of being ‘hard data’ of the kind used in the natural sciences, there are, in fact, fundamental doubts about its validity; about whether it represents accurately what it claims to represent. We can get a sense of these criticisms by looking briefly at the work of Piaget, mentioned earlier. Interestingly, this was not strongly quantitative in character, and it has been criticized by some for being insufficiently rigorous from an experimental point of view; reflecting, at least in part, a difference between Piaget and commentators on his work about the requirements of scientific research. This highlights the point we made earlier: that although it is convenient to refer to the ‘scientific method’, there is, in fact, a variety of interpretations of what is involved in a scientific approach to research and of how it should be applied to the study of human beings and their behaviours.

A. Similar criticisms have been applied to psychological and educational tests. For example, Mehan points out how test questions may be interpreted in ways different from those intended by the researcher. In one language development test, children are presented with a picture of a medieval fortress, complete with moat, drawbridge, and parapets and three initial consonants: D, C, and G. The child is supposed to circle the correct initial consonant C for ‘castle’, but many children choose D. After the test, when I asked those children what the name of the building was, they responded ‘Disneyland’. These children used the same line of reasoning intended by the tester, but they arrived at the wrong substantive answer. The score sheet showing a wrong answer does not document a child’s lack of reasoning ability; it only documents that the child indicated an answer different from the one the tester expected.

B. Here we have questions being raised about the validity of the sort of measurements on which the findings of quantitative research are typically based. Some, including for example Donaldson, regard these as technical problems that can be overcome by more rigorous experimentation. Others, however, including Mehan, believe them to be not simply problems with particular experiments or tests, but serious threats to validity that potentially affect all research of this kind.

C. At the same time, questions have also been raised about the assumption built into the logic of quantitative educational research that causes can be identified by physical and/or statistical manipulation of variables. Critics suggest that this fails to take account of the very nature of human social life, assuming it to consist of fixed, mechanical causal relationships, whereas in fact it involves complex processes of interpretation and negotiation that do not have determinate outcomes. From this point of view, it is not clear that we can understand why people do what they do in terms of the simple sorts of causal relationships on which quantitative research focuses. Social life, it is suggested, is much more contextually variable and complex.

D. Such criticisms of quantitative educational research have been the stimulus for an increasing number of educational researchers, over the past thirty or forty years, to adopt more qualitative approaches. These researchers have generally rejected attempts to measure and control variables experimentally or statistically. Qualitative research can take many forms; loosely indicated by such terms as ‘ethnography’, ‘case study’, ‘participant observation’, ‘life history’, ‘unstructured interviewing’, ‘discourse analysis’, etc. In general, though, it has the following characteristics:

E. A strong emphasis on exploring the nature of particular educational phenomena, rather than setting out to test hypotheses about them. A tendency to work with ‘unstructured data’: that is, data that have not been coded at the point of collection in terms of a closed set of analytical categories. When engaging in observation, qualitative researchers therefore audio-or video-record what happens or write detailed open-ended field-notes, rather than coding behaviour in terms of a predefined set of categories, as would a quantitative researcher employing ‘systematic observation’. Similarly, when interviewing, open-ended questions will be asked rather than questions requiring predefined answers of the kind typical, for example, of postal questionnaires. In fact, qualitative interviews are often designed to be close in character to casual conversations.

F. Typically, a small number of cases will be investigated in detail, rather than any attempt being made to cover a large number, as would be the case in most quantitative research, such as systematic observational studies or social surveys. The analysis of the data involves explicit interpretations of the meanings and functions of human actions, and mainly takes the form of verbal descriptions and explanations. Quantification and statistical analysis play a subordinate role at most. The two areas of educational research where criticism of quantitative research and the development of qualitative approaches initially emerged most strongly were the sociology of education and evaluation studies. The trend towards qualitative research in the sociology of education began in the UK in the 1960s with studies of a boys’ grammar school, a boys’ secondary modern school, and a girls’ grammar school by Lacey, Hargreaves and Lambart. They employed an ethnographic or participant observation approach, though they also collected some quantitative data on, for example, friendship patterns among the pupils. These researchers observed lessons, interviewed teachers and pupils, and drew on school records. They studied the schools for relatively long periods, spending many months collecting data and tracing changes over time.