A Review of Adventure Programming Literature
QUALITATIVE METHODOLOGICAL PARADIGM
4.5. Data Analysis Process
4.5.3. The Use of Computer Software in Phenomenographic Data Analysis
There exist over 25 different software programmes designed to make qualitative analysis faster and easier, but they may not be appropriate for all forms of analysis (Willig, 2008). Most phenomenographic studies have utilised a manual analysis process, whereby utterances, notes, or transcripts are arranged and rearranged in piles on the floor in the construction of categories. Their relational qualities are represented by their positioning in relation to one another on the floor. Using this approach to analysis, there are four aspects of data analysis that pose a challenge in phenomenography: (a) the sheer volume of data, (b) the challenge of keeping in mind the meaning of an utterance in the context of the transcript as well as its contribution to the meaning of a category and differences between categories, (c) the long period of time it takes to iteratively
state, and (d) ensuring that categories emerge as a function of the meanings contained in the text, and not due to preconceived theories or biases of the researcher (Åkerlind et al., 2005; Bowden, 1994, 2005; Bowden & Walsh, 2000; Marton & Booth, 1997). Computer software has been available since 1990 to alleviate the challenge posed by these factors in phenomenographic research (Booth, 1993). Computer software can serve different functions during analysis. At the most basic level it can help with the storage and organisation of material, even if just for typing in and storing transcripts in a word processing package. At the next level, it can make data handling more streamlined by allowing selection and storage of quotes, saving notes and meanings assigned to such utterances, and reducing the cognitive load of keeping in mind all the data whenever looking at a single utterance by facilitating easier access to data in various contexts (Booth, 1993). At a more sophisticated level, computer software can automatically identifying concepts in the text using semantic and relational informational extraction. By utilising 'objective' parameters such as rank, percentage and frequency, computer software can provide a validity check against subjective researcher bias (Penn-Edwards, 2010).
Researchers have identified a number of strategies to facilitate manageability of the data, including the use of notes and transcripts as a whole as analysis strategies, and reducing the volume used at any one time (Trigwell, 2000). Computer software can provide a “fast, efficient method of sorting large amounts of transcripted data and identifying expressed concepts” (Penn-Edwards, 2010, p. 253). Computer software can also facilitate seamless switching between the original context of the transcript, and that of the collective meaning within and between categories. Penn-Edwards (2010) tested the ability of computer software to streamline the traditional manual phenomenographic process. She used the Leximancer package, a qualitative data analysis software using semantic and relational data extraction to automatically code and identify relationships between categories, while simultaneously performing a traditional manual phenomenographic
using the software, and reported that the results “validated, to me as a phenomenographer, that Leximancer is an acceptable phenomenographic tool” (p. 259). Penn-Edwards (2010) suggests that Leximancer provides a more expedient way of generating initial categories during the first stage of analysis when dealing with large volumes of data. Another role such software can play is as a check of bias in the development of categories. Such software provides “a clear bracketing process in identifying the concepts embedded in the responses” (Penn-Edwards, 2010, p. 263). Using computer software for analysis makes pragmatic sense, as it provides a more efficient process for comparing and grouping of utterances, allows you to assign reflexive comments to selected excerpts, and note the contextual factors relating to selected quotations (Cousin, 2009). Computer software becomes another tool in the researcher's belt, simplifying the management, coding, locating, control and review of data; “it does not eliminate the need for the researcher to think”
(Jemmott, 2002, p. 7). Computer software does not replace the embedded role of the researcher, which is a critical part of reflexive phenomenographical practice (Penn-Edwards, 2010).
Booth (1993) suggests ten criteria any computer-assisted qualitative analysis software used for phenomenographical analysis should meet: (a) it should simplify the researcher’s management of transcribed interviews, while enabling easy access to the original data; (b) it should allow quote fragments in the transcript to be selected and assigned a theme; (c) these theme names should allow for strings long enough to be immediately meaningful to the researcher; (d) the themes should not need to be pre-specified, it should be possible for the researcher to create, change, add, and delete themes throughout the analysis process; (e) the quote fragments should be saved together with the themes assigned, general information about where the quote originated, and a researcher’s note; (f) it should allow easy access to the original interview context for any selected quote; (g) the researcher should be able to see a quote in the context of quotes with the same or similar themes;
(h) it should provide a facility for maintaining memos on the evolution of the themes themselves; (i)
different stages of a study; and (j) the software should be ergonomic and user-friendly to those with low computer literacy skills (Booth, 1993).
One little-known open-source package that meets all of these requirements is RQDA (Huang, 2009). This computer-assisted qualitative data analysis package is free, and ironically, runs within the R statistical programming environment (R Development Core Team, 2009). It allows for coding of themes, linking quotes to themes and access to the original transcript, writing of memos attached to specific categories or the project as a whole, creating plots of the sociogram of available categories, and can be used to interface with various text-mining packages available in R. Despite running from R, it is fairly user-friendly and runs within a graphical user interface environment.
While I am not aware of RQDA having being used for phenomenographic analysis before, it has been used in various other qualitative research contexts. For example, Van Windekens, Stilmant, and Baret (2013) used RQDA in developing an inductive cognitive mapping approach for analysing systems of practices and decision making processes linked to grassland management in a Belgian socio-ecological system. Wu and Yip (2010) used RQDA to understand the characteristics and substance of urban and rural grassroots homeowners' resistances in China using qualitative analysis of news clippings.