Develop and refine hypotheses
This is particularly appropriate if your qualitative analysis is part of some mixed methods research. A very common research design in mixed methods is to start with some qualitative research aimed at mapping out the landscape so that quantitative research questions can be identified and addressed in the later stages of the research. Typically, this design will be used if the researchers don’t know enough about the research domain at the start to specify clear hypotheses to be addressed in quantitative terms. The aim of this approach is to establish the range of phenomena to be found in the research area and the ways these phenomena might interact. This will then enable later quantitative research to establish how frequently these phenomena occur, how much impact they have and what the most likely (and least likely) relationships are between them and other circumstances. However, this kind of approach does not need to be followed by quantitative research. Although the qualitative analysis won’t be able to give accurate numeric estimates of frequency, etc., your analysis could at least establish the most likely situations and their relationships with other phenomena in the field you are investigating.
Create a model that explains the determinants of the research phenomenon
This kind of approach is most pertinent if your research is addressing some policy or practice needs or even is aimed at evaluating some programme of activity. The model you develop will show all the significant aspects of the phenomenon you are investigating and how they affect each other. This will give those wanting to change their practice in dealing with these phenomena some idea of what they can change and what impact they might have. Your analysis will focus on establishing the main components of this model (the key phenomena) and how they affect each other (what causes what) and even what strategies people adopt in order to achieve certain outcomes.
Develop a theory
This is an approach favoured by those following grounded theory and it overlaps a little with the previous approach, creating a model. The key point to recognize here is that there are different kinds and levels of theory. You don’t need to come up with a grand theory of the kind we associate with Goffman, Foucault or Bourdieu – though it would be great if you did. The theory can be much more localized or smaller scale. It may focus on a key phenomenon and explain how it has all kinds of impact on people’s actions. It might even be a version of an existing theory but modified for the particular circumstances you are investigating. The main characteristics of a theory in qualitative research are that it has the power to explain some phenomena and outcomes and commonly it uses terminology that is not being used by the participants you are investigating. It is not that they are not aware of what is going on but rather that they won’t talk in those terms or explain their actions that way.
These approaches are not mutually exclusive, actual analysis might combine two or more of them, and they are not the only ways of thinking about your analysis. But they do reflect the differences in methodology and philosophy that I discussed in the previous section. It is not that a particular methodological stance inevitably means that you are committed to a particular research aim, but rather that it will tend to lead you to certain aims in the first instance. However, one of the big benefits of qualitative analysis is that you can be flexible, and although you may start off (by inclination or by design) with a certain methodology and research aim, as your analysis develops and your understanding grows, other outcomes and other aims may begin to present themselves. That is a good thing. That is the way you can come up with original, insightful and useful results from your analysis.
Ethics
Ethical issues bear upon qualitative research like any other research. However, they mostly affect the stages of planning and data collection. For example, the principle of fully informed consent means that participants in research should know exactly what they are letting themselves in for, what will happen to them during the research and what will happen to the data they provide after the research is completed. They should be made aware of this before research on them starts and they should be given the option to withdraw from the research at any time, and usually, if they request it, any data that has been collected from them will be returned or destroyed. All of this happens well before the data are analyzed.
However, there are some special aspects of qualitative data and their collection that raise ethical issues. Perhaps the most significant is that qualitative data are usually very personal and individual. The identification of individuals cannot be hidden behind aggregated statistics when data are analyzed and reported on. Unless special steps are taken, reporting on qualitative data, and especially the use of direct quotations from respondents, will commonly identify specific participants and/or settings. Sometimes this is not an issue, and especially when it is with the agreement of participants, their real identity and that of the settings and organizations they are operating in can be revealed. But usually this is not the case. We are normally required to go to some lengths to protect the identity of those involved in our research. Chapter 2 discusses some of the aspects of anonymization of transcripts that are required in qualitative analysis.
The personal nature of much qualitative research means that researchers need to be very sensitive to the possible harm and upset their work might cause to participants. Again, mostly these issues arise at the stage of data collection, when, for example, the nature of depth interviews might allow people to talk at length and in depth about issues they would not normally address. Researchers have to be aware of the distress this might cause participants and make provisions for dealing with it. By the time the data are analyzed, these issues should have been dealt with, although there might still be some remaining issues connected with the publication of the results of the analysis. These issues will be discussed further in Chapter 7.
Key Points
Qualitative data are very varied, but all have in common that they are examples of human meaningful communication. For reasons of convenience most such data are converted to written (or typed) text. The analysis of what is often a large amount of material reflects two characteristics. First, the data are voluminous and there need to be methods for dealing with this in a practical and consistent way. Second, the data need to be interpreted.
There are some practical issues that make qualitative data analysis distinctive. These include starting data analysis before the sampling is decided and the data collection is complete, and the fact that the analysis of the data tends to increase its volume (at least to start with) rather than reduce it.
There is a tendency to see qualitative research as constructionist, inductive and idiographic. That is to say, to see it as concerned with the interpretation of new explanations about the unique features of individual cases. However, this is a gross simplification. Much qualitative research is concerned with explaining what people and situations have in common and doing this with reference to existing theories and concepts. To that extent it is nomothetic and deductive/abductive. In addition, although all researchers are sensitive to the way that even their descriptions are interpretations, they are sufficiently realist to believe that it is important to represent the views of participants and respondents as faithfully and accurately as possible.
These differences in methodology have an impact on the overall aims of your qualitative analysis, at least to start with. But qualitative analysis is very flexible and your final analysis might examine phenomena you hadn’t imagined when you started out.
Because of its individual and personal nature, qualitative research raises a host of ethical issues. However, most of these should have been dealt with before data analysis starts. Nevertheless, it is important to ensure that anonymity is preserved (if the assurance has been given) and that respondents know