The research design of a PBE should have demonstrated that the datagathering strategies fit their settings. Similarly, this planning principle ought to have led in the direction of systems of organization of data to make its retrieval for analysis as easy as possible. It is generally not a good idea to collect large amounts of data and then look for ways of deciphering it at a later date. Data analysis is, along with the formulation of conclusions, the culminating feature of the research process. It is designed to bring order, coherent patterns and meaning to data accumulated. The data analysis process can be thought of in terms of steps. First, raw data collected from questionnaires, survey instruments, interviews, observations, etc., has to be tabulated on charts, spreadsheets or other aggregating devices. The data is then subject to a process of reduction into meaningful bits. Categories are sought and allocated, relationships are tested, qualitative data is examined for meaning, statistics are computed, and so on. In reality, if the data-gathering instruments have been well prepared, the data should sort itself or at least point towards a sorting path. The classroom observation featured in Chapter 4 is both a recording and an explanatory device. The series of observations makes up the dataset and is available for analysis and interpretation by reference to the underlying theory.
As data is reduced and analysed the process of interpretation occurs. This is the process whereby warranted conclusions about the meaning of the data are drawn. Interpretation may be thought of as a reading, thinking, inferring and concluding process. It may involve the technical editing of field notes, decisions to use verbatim statements from interview transcripts, the application of a particularly searching inferential statistic, the matching of data to a hypothesis, and so on. The process of interpretation is an active intellectual one, and is now considerably aided by the availability of sophisticated computer software; more will be said about this later. Processes central to data analysis are coding and categorization. This is especially true of qualitative datasets such as extensive interview