4.7 Data Analysis, Coding and Categorization

The purpose of analysis is to "make sense of the data" and to "make meaning" (Merriam, 1998, p.178). It is an "ongoing cyclical process integrated into all phases of the research" that begins with the data collection ( McMillian Schumacher, 1997, p. 502). As Merriam (1988) explains:

Data collection and analysis is a simultaneous activity in qualitative research. Analysis begins with the first interview, the first observation, the first document read. Emerging insights, hunches, and tentative hypotheses direct the next phase of data collection, which in turn leads to refinement or reformulation of one’s questions, and so on. (p. 119)

In the case of this study, there were seven stages of data analysis. In each stage, the data underwent a certain reduction and transformation. The stages which are each described separately in this section of the chapter, are as follows:

1. On-going analysis during data collection;
2. organization of the data;
3. reading of and conversation with the data;
4. data reduction;
5. descriptive coding leading to the presentation of the data;
6. pattern coding leading to the interpretation of the data;
7. thematic coding leading to conclusions and recommendations about the data.

The first stage was conducted during the gathering of the data. In the context of the discussion lists, ongoing analysis of participants’ postings led to the researcher’s posting of particular questions or comments. The analysis of the postings led, as well, to the formulation of questions for the dialogues with discussion list participants. A preliminary analysis of the responses in the questionnaire led to the formulation of particular questions in the discussion list. Finally, analysis of the questionnaires led to the formulation of questions for the dialogue with the five questionnaire respondents.

The second stage of the analysis began once all the data had been collected from all sources. The aim of this stage was to organize the data into manageable formats. All the postings from the discussion list multilogues and dialogues were first organized chronologically and included the postings of the researcher. Such organization was necessary in order to follow the "train of thought" of the discussion. The total postings amounted to approximately 300 pages and filled two large binders. The questionnaires were assembled together into a binder but not according to any classification system.

The third stage of the research involved becoming familiar with the data that had been collected. Unlike the ongoing analysis where the data are viewed in isolated bits, this stage allowed for viewing of the data from a holistic perspective. The multilogue and dialogue postings as well as the questionnaires were read through several times. The tapes were played through numerous times. Merriam (1988) describes this stage of the analysis as one of "holding a conversation with the data, asking questions of it (sic), making comments, and so on" (p.131). Notes taken during this process comprised a preliminary organization and sorting which led to the following stage: that of data reduction.

The data collection yielded several hundred pages of postings, 176 questionnaire pages as well as approximately two hours of recorded telephone dialogues with the five questionnaire participants. Some data could not be used because of risk to anonymity of informants. Reporting of certain comments made would have possibly revealed the identity of some of the participants. Many pages of data could be compressed because of the nature of discussion list postings. On occasion, an individual’s response may have constituted only two or three lines. However, their message may have comprised as many as two or three pages because of the identifying headers, signature files, and, most of all, because, they were replying to previous comments made and these comments were included in their message. Thus, once the postings were printed, it became obvious that there was considerable repetition of postings. Other postings were not relevant to the issue. For example, at one point in the discussion, a long posting was made by one participant in relation to an ongoing political dispute between the United States government and a middle-eastern country. Other postings were of a personal nature or simply unrelated to the discussion whereby individuals may have related anecdotes, described their education, or asked questions related to personal issues. Finally, the postings of the researcher were removed from the collection of data.

Reduction or elimination of material from the discussion list postings could be easily accomplished. The entire collection of postings from CREDO and CREO could be retrieved as one file from an archive created on the computer of the service provider. The files were imported into a word processor where cutting and pasting allowed for easy organization of the data. Any data not relevant to the study were cut from this file. Headers were removed, repeated postings were deleted, signature files removed and comments from the researcher were removed as well. The dialogue postings could not be retrieved from the archive but were saved in a "mailbox" from which they could be easily retrieved. For each message, the headers and repeated or second postings were removed. The remaining postings were combined or added onto those of the multilogue to create one large file which comprised, at the end of reduction, approximately, 70, single spaced pages of data.

In terms of the questionnaires, it was possible once again to eliminate some of the data however, there was far less data reduction than with the discussion list postings. This is not surprising given that the questionnaires were far more structured and responses were directed by the questions in the questionnaire. The questionnaire comments were read and reread. Using a highlighter, it was possible to isolate all parts of the data that were relevant. Once these comments were highlighted, they were then typed into a word processor. While there were 88 questionnaires each two pages long, many respondents did not write for the full two pages. The reduction in the data resulted from abbreviating the respondents’ comments in order to highlight the essential points.

A similar procedure was followed for the cassette recordings of the dialogues with the five participants from the questionnaires. A certain portion of these dialogues could be eliminated or reduced because the discussion centered around developing rapport, getting to know the individual, making him/her feel comfortable, explaining the purpose of the research and answering questions about the research. The tapes were listened to repeatedly, notes were taken on the comments made to assist in the later development of themes and patterns. The final product of merging all the data into one text resulted in approximately one hundred single-spaced pages of typed text.

The next important step in the process of analysis was to code the data. "Codes are tags or labels for assigning units of meaning to the descriptive or inferential information compiled during a study" (Miles Huberman, 1994, p.56). They are astringent in the sense that they pull together a lot of material (Ibid.). The initial coding simply involved writing a descriptive word next to a given excerpt in order to summarize what the excerpt was about: for example, resources, training, time, teaching, learning, Internet, equipment etc. Once all the data had been coded in this way, categories and sub-categories could be established.

All of the data were coded manually with the categories written in the margins alongside each excerpt. Then, through the use of the cut and paste features of a word-processor, all excerpts could be moved into the categories into which they had been coded. At this stage, the data could be reported or described. Chapter 5 thus presents the data grouped into categories. This presentation of the data does not include any interpretation. The presentation provides an answer to the first research question which was: What are some of the beliefs of teachers about teaching and learning FSFL in OLEs?

Once the beliefs had all been identified, it was possible to move to the next level of analysis, stage six, the aim of which was to interpret the data. The data were interpreted in light of the conceptual and historical framework which was outlined in Chapter 2 of this study and the aim was to provide an answer to research question 2: What do these beliefs reflect in terms of the evolution of approaches and use of technology in the teaching of FSFL? The interpretation was accomplished through a search for patterns in the data which related to elements identified in Chapter 2. In the descriptive coding of the data, key words were used in order to group the data according to topical categories. In order to identify patterns, similar types of beliefs were grouped analyzed, explained and interpreted.

The final stage in the analysis of the data involved identifying overall themes. These themes allow us to draw conclusions from the data, to determine the implications of the findings and to make recommendations for educational practice. The conclusions, implications and recommendations are presented in Chapter 7.