3 RESEARCH METHODS AND DESIGN 54
3.4 Instrument design, data collection and analysis techniques 71
3.4.7 Content analysis 76
Content analysis is a well-established method for making meaningful inferences from text by categorising data into clusters to identify patterns and relationships (Flick, 2009, p. 323ff.; Given, 2008, pp. 121-123; Schamber, 2000, p. 735; Williamson & Johanson, 2013). The method is “a way of reducing data and making sense of them” (Given, 2008). The method has been described (Babbie, 2010, p. 121) as “essentially a coding operation”, in which the textual data is coded or classified according to some conceptual framework. In inductive content analysis the data is reviewed, categories or labels are assigned to “chunks” of varying size, such as words, phrases, sentences or whole paragraphs, and typically the labels on review
generate more abstract categories (Miles & Huberman, 1994, p. 56). Schamber (2000) explains:
The analytic process requires the use of a coding scheme, which consists of categories and operational definitions for specific variables (e.g., images of a certain societal group). Content-bearing units are identified in the texts and coded for appropriate categories. Categories can be derived inductively from the texts being analyzed, adapted from previous studies, or adopted unchanged from previous studies. Inductive content analysis is particularly appropriate for research that takes a grounded theory approach, or which derives theory from data rather than verifies existing theory. The development of new schemes entails decisions about units of analysis, category construction, and coding procedures. (p. 735)
The technique has been used widely in LIS research for both qualitative and quantitative analysis (White & Marsh, 2006).
The use of interview guides for the interviews may suggest that a deductive or “a priori” approach is adopted for the coding framework, but while the guides are useful for structuring interviews, participants are able to respond as they choose. The subsequent analysis of responses identifies themes emerging from the data through a process referred to as “open coding” (Strauss & Corbin, 1998). In this process conceptual categories are identified and grouped to create a framework. The coding scheme is generated from close examination of the data and the creation of codes that most closely describe the content.
A content analysis software program (QSR N6), into which data from the interviews and photo analyses are entered, supports the analysis, identification and coding of code content-bearing units in the text. The use of computer-assisted qualitative data analysis provides quick and accurate processing and a reliable general picture of data, although it may guide the direction of research or distance the researcher from the data (Welsh, 2002). Use of software can assist with validity as it provides a record of how data are analysed and may help reduce errors stemming from coding inconsistencies.
Content analysis is used in both stages of the research, with the data from the interview and photo analysis sessions, and with the survey responses provided
during the tagging studies (C to E). During content analysis a codebook is developed to control terms and ensure consistency as the coding scheme goes through various stages of development and refinement. Development and coding varies from the generally straightforward, because of manifest content or concrete terms found in the text, to more challenging analysis, based in part on latent or underlying meaning identified in interpreting the text (Babbie, 2010).
3.4.7.1 Content analysis workflow
The identification of categories and themes is proposed to follow a series of steps which are similar to those outlined in the literature (Williamson & Johanson, 2013). For example, each of the interviews in Studies A and B, are to be transcribed by the researcher and the transcription closely reviewed. Prior to coding, the researcher is to read through the interview transcripts noting any issues of key interest or significance. Throughout this process, the researcher is to note possible coding terms. At the next reading, the researcher is to begin to develop a list of key terms to be used in the coding and further developed through notations, which include keywords and themes, in the content analysis software QSR N6.
Beginning coding early in the data collection process allows for growth in understanding, which informs subsequent data gathering. As interviews are conducted and transcribed, the researcher is proposing to add new terms as necessary and to modify the coding list as appropriate. Some text units may be coded to more than one category.
The researcher is then to review this list of codes. This review is to take into account the research purpose, the research questions and the transcripts. The coding terms be to be considered against the terms and categories used in previous studies to determine the most suitable terminology. Links or relationships between the codes are to be identified, leading to the emergence of concepts and themes.
Once the list of codes is refined the researcher is to re-examine, reduce and code the data. Data reduction is “the process of selecting, focusing, simplifying, abstracting and transforming the data that appears in written-up notes or transcriptions” (Miles & Huberman, 1994, pp. 10-11) enabling a more focussed analysis and revealing of
further connections, patterns and emergent themes. Data reduction is an iterative process that continues until the final report is written (Miles & Huberman, 1994, pp. 10-11; Patton, 2002, pp. 436-437). According to Given (2008, p. 121) iterative analyses helps to improve trustworthiness and credibility.
The list of codes developed working with the interview transcripts is to be used and tested against the survey response data from the tagging studies C to E.
The themes and an example of the process of the content analysis are provided in Appendix 6 - Themes from the Content Analysis and a Detailed Example of the Process of Data Analysis.