Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
in qualitative research
1 GENERAL INFORMATION
Now that you have explained HOW you will collect your data, you must also explain to the reader how you will ANALYZE all those words! Most of you will also do more than one case study, so what is written below is based on that assumption. As explained, analysis means making sense of the data collected so you can present findings and draw specific conclusions at the end of your thesis (i.e. answer your research questions). Remind the reader what you brought up in methodology on how you will handle the analysis of all these WORDS and how this will provide a foundation on which you can draw conclusions. Again, you just want to explain here that Miles and Huberman (1994) refer to analysis of qualitative data as being organized into three stages: References brought up in this document are available at the end. NOTE: Some of this may be used in your analysis chapter, but most of how you will analyze your data should be presented already in Chapter Three and then just remind the reader in your analysis chapter (Chapter Five). Yin (1994) states that every case study should start with a general analytical strategy. These general analytical strategies with regards to case studies provide the researcher with a system by which s/he can set priorities for what it is they need to analyse and why. As Herriott and Firestone (1983) stated in Yin (1994), The evidence from multiple cases is often considered more compelling, and the overall study is therefore regarded as being more robust (p. 45). The way in which the data will be analysed is very important for any research study. For many of you in your work, it will involve the analysis of the interviews you conducted (i.e. words you collected), as well as the review and use of any documentation that the firms feel they are able to hand over (i.e. annual reports, brochures, etc.). As previously stated, these multiple sources of evidence (i.e. triangulation) are what add to a studys validity (Yin, 1994). Yin (1994) adds that before data can actually be analyzed, a researcher using case studies can choose from two general analytical strategies: Relying on theoretical propositions is the most common (and advised) strategy (according to Yin, p. 103-104). The result of this is the collection of data based on research questions taken from previous studiesand here you are comparing your findings with these previous studies (or those you used in your frame of reference). (MOST OF YOU USE THIS ONE)! Developing a case description can be used as a strategy as well, according to Yin, but this is less favorable and should only used when little previous research has been done (this is NOT the case for most if not all of you).
Yin (ibid.) then explains that specific techniques (and here he refers to Miles & Huberman, 1984) can be used to actually analyze the data. We use more recent thinking from these authors: Many of you will rely on two main forms of analysis for your case study data: 1. Within-case analysis: Comparing your data against the theory you are using. 2. Cross-case analysis: Data in one case compared to data in the other cases. In writing about qualitative data analysis, Miles and Huberman (1994) state that the focus is on data in the form of words in our case words that emanate from the interviews conducted. These words require processing, according to these authors, and this processing is itself a form of analysis.
2. Data display is the second major activity which the researcher should go
through, and this means taking the reduced data and displaying it in an organized, compressed way so that conclusions can be more easily drawn. These authors explain that, humans are not powerful processors of large amounts of information, and that extended text can overload humans information-processing capabilities (op. cit., p. 11, with reference to Faust, 1982). It is further explained that good displays are, a major avenue to valid qualitative analysis (op. cit., p. 11). In conclusion, they state that, as with data reduction, the creation and use of displays is not separate from analysis, but is a part of it. USING CROSS-CASE ANALYSIS (with matrices) WORKS BEST HERE!
3. Conclusion drawing and verification is the final analytical activity for the
qualitative researcher. It is here that the researcher begins to decide what things mean. They do this by noting regularities, patterns (differences/similarities), explanations, possible configurations, causal flows, and propositions. However, Miles and Huberman also add that the competent researcher should hold such conclusions lightly, while maintaining both openness and a degree of skepticism. FOR YOUR THESIS, THIS STEP IS YOUR FINAL CHAPTER (CONCLUSIONS). IMPORTANT: You should analysis in detail (and especially discuss these 3 steps) in your methodology chapter, explaining to the reader how you plan to analyze your data. In your actual analysis chapter, you will simply remind them briefly of using this approach.
as well. This is what makes doing more than one case so worthwhile -- you are making your data a little richer and more generalizable. BUT be careful with the word generalizable, as it can be rather dangerous in case study research. Your overall purpose (research problem), remember, is to gain a better understanding of some phenomena (i.e. the Internet, event marketing, relationships, advertisingwhatever). An EXAMPLE of how your cross-case analysis might be presented is given below: Research Question One: Write out your research questionor it's topic area. Then present a matrix with all the variables you collected data on in your interviews within that research question. Then compare each variable (or set of variables) across each case. Table X: Write out RQ topic here Variable/Case Case 1 Briefly write Variable X Variable Y Variable Z
data in here Briefly write data in here Briefly write data in here
Case 2
Briefly write data in here Briefly write data in here Briefly write data in here
Comparing case 1 with case 2 for this variable in RQ1 = one paragraph of writing! OR use a coding system
SOURCE: NOT NEEDED (IF no source, then it is authors own work!) You can also learn to CODE your responses (contact Tim if you need info or a reference on coding your data). Coding would be, for example, the types of responses given. EXAMPLE: Use of promotional tools in a study I did presented four types of responses when I looked across the cases: (++) = They thought TV advertising was important and they used it; (+-) = they thought it was important but did not use it (-+) = They did not think it was important but they used it (--) = they did not use it and did not think it was important.