CHAPTER THREE: Methods of data collection and analysis 3.1 Introduction
3.6 Techniques of data analysis
In ethnography, “analysis of data involves interpretation of the meaning and functions of human action, and mainly takes the form of verbal description and explanation”
(Hammersley, 1990, p.2). This is what I did using the grounded theory method to identify
and describe patterns in the data. In addition to the initial field analysis of data undertaken to guide the research process, the data obtained (the interview transcripts) were coded and analysed thematically using N-vivo software for qualitative data analysis.
3.6.1 Grounded theory method of data analysis
The method of data analysis used was grounded theory. According to Barton and Hamilton (1998, p.68), “This is not a methodology in itself,” but a way of approaching data analysis and collection. In this approach, data is collected and analysed and the analysis is used to decide what data to collect in the next round of collection in a circle of repetition. This process of data analysis moves back and forth between data collection, analysis, and conclusion (See Barton & Hamilton, 1998; Bryman, 2001; Glaser & Strauss, 1967).
Generally, grounded theory is more popular as a method used for generating theory.
However, it is also used to validate and extend an existing theory (Glaser & Strauss, 1967; Strauss & Corbin, 1994). In spite of that, “Grounded theory has become by far the most widely used framework for analysing qualitative data” (Bryman, 2001, p.390).
Ethnography generates qualitative data through unstructured interviews, observations, and collecting documents of all kinds like diaries, letters, and newspapers. All these data are of a qualitative nature that is suitably analysed using grounded theory (Strauss & Corbin, 1994). Furthermore, my study was focusing on the social uses of literacy for which grounded theory is best suited. Therefore, I used grounded theory as a method of data analysis in the ethnographic research process. Barton and Hamilton (1998) used grounded theory as a method of data analysis to approach their research data on local literacies in Lancaster.
In using the grounded theory framework to look at my data, the process of data analysis started as soon as the first initial data were in. The use of the grounded theory method of data analyses involved the use of the following grounded theory tools of analysis.
3.6.1.1 Theoretical sampling
Theoretical sampling is a process of collecting data, in which data is collected, coded, analysed and used to decide the next data to collect (Bryman, 2001). For example, the initial data about the community was collected through general observations and interviews. From this first round of data collection and analysis, some codes and categories were generated about the socio-economic structure of the community. The conclusion from this first round of data collection and analysis was used to guide the next round of data collection that went through the same process (Strauss & Corbin, 1994;
Glaser & Strauss, 1967). This theoretical sampling was to go on until a point of
theoretical saturation was attained (see section 3.6.1.3 below for details of the concept of theoretical saturation). However, as discussed in section 1.8 above, because of the scale of the research, this process generated more issues than were possible to continue pursuing within the time available for the study.
3.6.1.2 Coding
In grounded theory, coding is the process of grouping data into categories (Glaser &
Strauss, 1967). The codes or groupings are not preconceived but derived from the data as they come in. Accordingly, this process started as soon as the first data were collected and it continued as more data came in. The data collected were broken down into component parts or emerging categories. The categorisations were not conclusive, as re-
categorisation of data continued. This was planned to go on until a point of theoretical saturation was achieved (see Bryman, 2001; Strauss & Corbin, 1994; Tuchman, 1994).
3.6.1.3 Theoretical saturation
As mentioned above, theoretical saturation is a point where no more new categories are emerging from the data. It is a point where further classification emerging from the new data is not possible. In other words, new data is no longer bringing in new information that could change the existing categories that have been set out by earlier data analysis.
The final and long surviving categories constitute the different findings on the literacy practices in the lives of rural people and/or the different things that give meaning to literacy for people in Bweyale (Bryman, 2001; Glaser & Strauss, 1967).
In this study of community literacy practices in Bweyale, this process of following up leads from an exhaustive analysis of data from one context to the next became difficult to conclude. This was because most of the categories generated from analysis of the data being collected after the initial general study of the social and economic structure in the community generated more categories whose further investigation continued to open up more new areas still requiring further and in-depth investigation in their own right. In other words, the process became endless. For example, literacy practices within the family as a social unit led to the following categories that all have influence on the literacy practice that goes on in the family:
• religion,
• school,
• socio-economic life in Bweyale,
• educational background and literacy status of family members,
• gender relations within the family,
• composition of the family (e.g., age range of both children and parents, presence of grandchildren etc)
These and many more were categories generated from the initial identification of the family as an important literacy domain in the community that demanded further investigations. In this case, investigating all those dynamics within the family literacy practices would have required an independent study it its own right. Because these factors have a bearing on the literacy practices in the community, they demanded an independent investigation within the context of home life.
Notwithstanding the fact that these were also major domains of literacy in rural
community life, this realisation led to a strategy of anchoring the investigation at the next appropriate convenient point and the use of other methods that would end the apparently endless depth required to investigate most of the categories that were being generated.
Whatever the case, this study provides a general description and understanding of local literacy practices in Bweyale.
I would like to state that for most of the data collected analysis was complete. That is, the point of theoretical saturation was achieved. What became a runaway situation was the continued follow up of the themes generated from analysing the data being collected.
Each of those themes constituted new phenomenon requiring further examination.
3.6.1.4 Constant comparisons
The other tool of data analysis in grounded theory is constant comparison (Bryman, 2001) or comparative analysis (Glaser & Strauss, 1967; Strauss & Corbin, 1994). In this
process, I studied the data in one category to see the connections. This helped to find out if all the data put together in one category really belong together. It also helped to come up with the characteristics that define a particular category. The characteristics facilitated the process of coming up with a generalization for the data falling under that category.
The different categories were then compared and refined through critical examination. In doing that, new categories were emerging from splitting or merging existing categories (See Bryman, 2001). As I have already explained above, this process of constant comparison, like theoretical saturation, could not be brought to a meaningful conclusion for most categories of data generated from this study. Likewise, the constant comparison had to be deliberately anchored at the next most convenient point in analysing a particular
category. This deliberate restriction in theoretical analysis (theoretical saturation and constant comparisons) of the data had to be applied because this study was time bound.
For example, the analysis of data related to religious literacy practices in the community generated the following categories: The Sunday church service, private Bible or religious reading, Bible study reading, Catholic and Anglican literacy practices etc. When the church service as a category was further analysed, the following came up: the use of reading and writing by the priest or the reading and writing that goes on around the altar, the reading and writing done by the congregation, the reading and writing used in organising the Sunday service like prayer and songbooks. This breakdown can go on endlessly into investigating the meanings of certain literacy practices, the church doctrines, how they inform each practice in different church processes and the local appropriation of the universal church literacy practices. For that reason, I decided to deliberately confine the investigation to the observable present practices and describe them as they appear.
3.6.2 Textual analysis of interview transcripts and documents
Some collected documents were subjected to a textual analysis to gain some insight into the respondents’ understanding of and attitudes toward reading and writing (Cacciapouti, 1998; Harper, 1994). The interview transcripts were read and reread to identify and follow different themes using content and theme analysis procedures used in textual analysis. Later the transcripts were coded and analysed thematically using the N-vivo software for qualitative data analysis as stated above.
This was not only about finding out what people say about literacy, but also how and why they say what they are saying. Thus, the textual analysis of the transcripts helped to identify the dominant discourse that influences local literacy practices in rural community everyday life. The transcription and analysis were carried out in the language in which the interviews were conducted and recorded. Only those sections of the interviews selected for use in the thesis were translated into English. This translation was done by me as a first language speaker of the Luo language in which all the non-English interviews were conducted. Those interviews conducted in English were reported as stated, with limited adjustments in the language to make it understandable in written form. The adjustments were only made if the statements were not very clear in writing. Since I conducted the interviews myself, I was in the position to understand what was being said from the context of the discussion at that particular time.
3.6.3 Photographic analysis
Photographs of some literacy events and some aspects of people’s everyday life were taken to provide concrete supplementary evidence of literacy in rural community life. The photos were not subjected to any serious form of analysis except studying them and selecting the most relevant aspects needed to support the presentation of different aspects of community literacy practices in Bweyale (Harper, 1994).