• No results found

3.10 Qualitative data analysis process

3.10.2 Data analysis: the tools used

The process of examining the data set and breaking it down into meaningful parts – analysis (Gonzalez, 2002) - is not easy. As a researcher, I am aware of the proposition that if the analytical processes adopted within the study are not articulated, it is not possible to make judgements about the appropriateness or value of the findings presented (Froggatt, 2001, p. 436). In this section and the next ones, I speak to how I arrived at the answers to the question this research addresses.

After establishing ways and means to store massive data, the next hiccup was processing it. Concerning the tools to use to process the data, I decided to do it manually. I am aware of the many qualitative data analysis computer programs available on the market today but to gain a deeper understanding of the data, I engaged in working with it manually over and over again. After all, computer programmes essentially are simply an aid to sort and organise sets of qualitative data, (Creswell, 2012) but none are capable of the intellectual and conceptualising processes required to transform data into meaningful findings (Thorne, 2000, p. 69). This requires the researcher’s capacity to think deeply.

From a phenomenological view point there is no hard and fast rule in analysing data. Data analysis process is largely intuitive (Merriam, 2001) and the learning is in the doing (Crewswell, 2012; Merriam, 2001). The researcher is at liberty to employ an analytical procedure she considers easy to follow (Osborn, 1990). Phenomenological researchers are usually reluctant to focus on prescribed steps and they are justified since that would not do justice to the integrity of the phenomenon (Hycner, 1985). The implication is that the guidelines usually proposed are just a possible way of analysing the data. Thus I am not glued to them but I am flexible. I welcome this stance considering the complexity of my data (it was collected using one collated interview schedule for the main research project). This gives me room to do some preliminary analysis first in the way I see fit.

93 The complexity of the collated interview schedule has ripple effects, both positive and negative. It spills into the data collection which in turn spills into the data analysis process. This made the process more complicated. Since there is no hard and fast rule to qualitative data analysis, especially within the phenomenological tradition, I decided to engage in what I coin the two-staged multiple level data analysis process.

First, I used Susan van Zyl’s ‘vertical and horizontal’ preliminary data analysis processing by (Personal communication: Research Workshop notes 21 August, 2011). According to van Zyl, soon after transcriptions some preliminary data analysis needs to be done. This involves the process of reducing the large volume of data in a way that makes it manageable at the same time not losing the minutiae which includes maintaining the presence of the voice of the participants which is a major strength of qualitative research data. This plummeting of data could be done in two ways and levels: vertical level analysis and horizontal level analysis.

Van Zyl proposes that vertical level analysis entails reducing the volume of data by way of summarising and paraphrasing interview by interview. Furthermore, at this stage one also does preliminary selection of what to keep in and what to leave out. This is informed by what the study set out to achieve: the aim and the nature of the research question5 since in the final product, the data functions as evidence. To some extent this implies bringing in some level of preliminary interpretation or thematic analysis.

Horizontal level analysis involves making comparisons - relating the data from the different interviews to each other usually by way of vertical analyses. Here the data that was identified from the vertical level for each transcript is compared with each other and this involves for instance, looking for recurring themes, categories, discourses that could then be understood as experiences, perceptions, attitudes and beliefs. Again this is guided by the research question, the topic and methods. In other words, here I consider phenomenology since the two analysis processes or tools should speak to each other.

5 In this case “the research questions serve as navigational tools that can help a researcher map possible

94 Second, for the main analytical strategy or tool, I employed Osborn’s (1990) phenomenological data analysis procedure consisting practically in interactive levels of analysis from the specific to the general and from simple (actual statements) to complex (abstraction) as illustrated in figure 3.1 that follows.

Figure 3.1: An illustration of Osborne’s phenomenological data analysis process frame

(Ideas adopted from Wiart, Dorrah, Hollis, Cook & May, 2004, p. 9)

As illustrated in Figure 3.1, Osborn’s (1990) data analysis procedure has the following steps:

The researcher transcribes the data then reads through all the transcriptions extracting or identifying units of meaning relevant to the research questions.

Once the units of general meaning have been established, they are carefully scrutinised to eliminate the redundant units by checking the literal content for number of times it was mentioned and how it was stated thus, reducing to units relevant to the issue asked.

The units of meaning are then paraphrased to facilitate the creation of first order themes.

95 The first order themes are then clustered into few second order themes. Finally, the second order themes are then consolidated into fewer and

more general higher level thematic abstractions that form a pattern or structure of the phenomenon.

The procedure culminates into the final thematic synthesis which then could be presented to each participant for validation (goodness of fit). Every aspect of the common experience should fit with every participant’s experience (Osborne, 1990).

The thematic high level abstractions are used as a framework to guide the presentation of the results (Osborne, 1990). Finally, the researcher writes a composite summary which captures the participants’ campus experiences as experienced by the participants (Lester, 1999). In the discussion part, a contemporary phenomenological approach allows the researcher to interpret and conceptualise the qualitative data and make reference to existing literature (Thomas et al., 2007).