CHAPTER THIRTEEN ANALYSING DATA II QUALITATIVE DATA ANALYSIS

  • Slides: 21
Download presentation
CHAPTER THIRTEEN ANALYSING DATA II: QUALITATIVE DATA ANALYSIS

CHAPTER THIRTEEN ANALYSING DATA II: QUALITATIVE DATA ANALYSIS

STAGES OF QUALITATIVE ANALYSIS Miles and Huberman (1994) suggest that qualitative data analysis consists

STAGES OF QUALITATIVE ANALYSIS Miles and Huberman (1994) suggest that qualitative data analysis consists of three procedures: 1. Data reduction. This refers to the process whereby the mass of qualitative data you may obtain – interview transcripts, field notes, observations etc. – is reduced and organised, for example coding, writing summaries, discarding irrelevant data and so on. At this stage, try and discard all irrelevant information, but do ensure that you have access to it later if required, as unexpected findings may need you to re-examine some data previously considered unnecessary.

2. Data display. To draw conclusions from the mass of data, Miles and Huberman

2. Data display. To draw conclusions from the mass of data, Miles and Huberman suggest that a good display of data, in the form of tables, charts, networks and other graphical formats is essential. This is a continual process, rather than just one to be carried out at the end of the data collection.

3. Conclusion drawing/verification. Your analysis should allow you to begin to develop conclusions regarding

3. Conclusion drawing/verification. Your analysis should allow you to begin to develop conclusions regarding your study. These initial conclusions can then be verified, that is their validity examined through reference to your existing field notes or further data collection.

CODING QUALITATIVE DATA Coding is the organisation of raw data into conceptual categories. Each

CODING QUALITATIVE DATA Coding is the organisation of raw data into conceptual categories. Each code is effectively a category or ‘bin’ into which a piece of data is placed. As Miles and Huberman (1994, p. 56) note: Codes are tags or labels for assigning units of meaning to the descriptive or inferential information compiled during a study. Codes are usually attached to ‘chunks’ of varying size – words, phrases, sentences or whole paragraphs.

Codes should be: • Valid, that is they should accurately reflect what is being

Codes should be: • Valid, that is they should accurately reflect what is being researched. • Mutually exclusive, in that codes should be distinct, with no overlap. • Exhaustive, that is all relevant data should fit into a code.

STAGES OF DATA CODING 1. The data is carefully read, all statements relating to

STAGES OF DATA CODING 1. The data is carefully read, all statements relating to the research question are identified, and each is assigned a code, or category. These codes are then noted, and each relevant statement is organised under its appropriate code. This is referred to as open coding.

2. Using the codes developed in stage 1, the researcher rereads the qualitative data,

2. Using the codes developed in stage 1, the researcher rereads the qualitative data, and searches for statements that may fit into any of the categories. Further codes may also be developed in this stage. This is also referred to as axial coding.

3. Once the first two stages of coding have been completed, the researcher should

3. Once the first two stages of coding have been completed, the researcher should become more analytical, and look for patterns and explanation in the codes. Questions should be asked such as: • Can I relate certain codes together under a more general code? • Can I organise codes sequentially (for example does code A happen before code B)? • Can I identify any causal relationships (does code A cause code B)?

4. The fourth stage is that of selective coding. This involves reading through the

4. The fourth stage is that of selective coding. This involves reading through the raw data for cases that illustrate the analysis, or explain the concepts. The researcher should also look for data that is contradictory, as well as confirmatory, as it is important not to be selective in choosing data. You must avoid what is referred to as confirmation bias, or the tendency to seek out and report data that supports your own ideas about the key findings of the study.

ORGANISING YOUR DATA Coded data may then be organised as suggested by Biddle et

ORGANISING YOUR DATA Coded data may then be organised as suggested by Biddle et al. (2001) whereby the data units (statements, sentences, etc. ) are clustered into common themes (essentially the same as codes), so that similar units are grouped together into first order themes, and separated away from units with different meaning. The same process is then repeated with the first order themes, which are grouped together into second order themes. This is repeated as far as possible as shown…

Raw data themes Higher order themes General dimensions The ordinary wood… has the ultimate

Raw data themes Higher order themes General dimensions The ordinary wood… has the ultimate feel, it feels like it’s a golf club that you're very much in control of, rather than its in control of you. The whole club swung very well, it felt nice. You felt as if you were in control. Controllable feel … just feels as though I'm in control of the clubhead right throughout the shot. Club control I feel that I've no control over that clubhead at all. This feels much more difficult to control… …but I could not control it due to the length and the flex of the shaft. Uncontrollable feel

At no stage are numbers assigned to any category. As Krane et al. (1997,

At no stage are numbers assigned to any category. As Krane et al. (1997, p. 214) suggest: Placing a frequency count after a category of experiences is tantamount to saying how important it is; thus value is derived by number. In many cases, rare experiences are no less meaningful, useful, or important than common ones. In some cases, the rare experience may be the most enlightening one.

WHAT SHOULD I LOOK FOR WHEN I HAVE CODED MY DATA? • You should

WHAT SHOULD I LOOK FOR WHEN I HAVE CODED MY DATA? • You should look for patterns or regularities that occur. • Within each code, look for data units that illustrate or describe the situation you are interested in. • Try to identify key words or phrases, such as ‘because’, ‘despite’, ‘in order to’, ‘otherwise’ and so on and try to make sense of the data. • Look for statements that not only support your theories, but also refute them. • Try to build a comprehensive picture of the topic.

Frankfort-Nachimas and Nachimas (1996) suggest that you ask yourself a number of questions to

Frankfort-Nachimas and Nachimas (1996) suggest that you ask yourself a number of questions to assist in your analysis: 1. What type of behaviour is being demonstrated? 2. What is its structure? 3. How frequent is it? 4. What are its causes? 5. What are its processes? 6. What are its consequences? 7. What are people’s strategies for dealing with the behaviour?

USING RAW DATA TO SUPPORT YOUR ANALYSIS You should resist the temptation to over-use

USING RAW DATA TO SUPPORT YOUR ANALYSIS You should resist the temptation to over-use quotes. However, as a rule of thumb, you should use direct quotes or observations: • When they describe a phenomenon particularly well. • To show cases or instances that are unusual. • To show data that is unexpected. You should also avoid including quotes without making clear reference to how such quotes refer to your analysis.

ENSURING THE TRUSTWORTHINESS OF YOUR ANALYSIS Holloway and Wheeler (2009) summarise the means by

ENSURING THE TRUSTWORTHINESS OF YOUR ANALYSIS Holloway and Wheeler (2009) summarise the means by which you can try to ensure the trustworthiness of your data. These include: Member Validation − One particular method of note is to ask those being investigated to judge the analysis and interpretation themselves, by providing them with a summary of the analysis, and asking them to critically comment upon the adequacy of the findings.

Searching for negative cases and alternative explanations – Interpretation should not focus on identifying

Searching for negative cases and alternative explanations – Interpretation should not focus on identifying only cases to support the researcher’s ideas or explanations, but to also identify and explain cases that contradict. Triangulation – Combining the analysis with findings from different data sources is useful as a means to demonstrate trustworthiness in the analysis.

The audit trail – To ensure reliability all research should have an audit trail

The audit trail – To ensure reliability all research should have an audit trail by which others are able to judge the process through which the research has been conducted, and the key decisions that have informed the research process. Reflexivity – Reflexivity means that researchers critically reflect on their own role within the whole of the data collection process, and demonstrate an awareness of this, and how it may have influenced findings, to the reader.

SUMMARY 1. Although qualitative and quantitative data are different in nature, the analysis of

SUMMARY 1. Although qualitative and quantitative data are different in nature, the analysis of both involves inference, systematic analysis and comparison. Both try to seek valid conclusions and avoid errors. 2. There a number of ways of approaching qualitative analysis. 3. Analysing qualitative data should be an ongoing process throughout, as well as after the collection of data.

4. There are three key stages to qualitative data analysis: data reduction, data display

4. There are three key stages to qualitative data analysis: data reduction, data display and conclusion drawing/verification. 5. Data reduction takes place through the process of coding. Coding involves assigning units of meaning to data chunks, and can be open, axial or selective. These codes can then be displayed or organised to allow the drawing of conclusions.