Who is this course for?
This course is designed for participants who wish to acquire methodological expertise in qualitative data analysis generally and, more specifically, widen their understanding and applied skills in conducting QCA, TA, CCA and GT in NVivo. The course will benefit to participants who plan to conduct one of the above method in their PhD or postdoctoral research, and to those wanting to generally broaden their area of methodological expertise in qualitative research. The course responds well to participants that have collected their data and want to apply the methods’ coding and analytic procedures on their dataset as well as those who don’t have data yet. For the latter, sample data will be provided.
Contribution of this course
Amongst the methods available to analyse qualitative data qualitatively, the four methods which this course is based on have been widely used across the social sciences. Their procedures to carry out analysis are straightforward and this makes the analytic journey transparent, traceable, and auditable on the part of the researcher. Each method is unique in its own right in that it suits a particular type of research questions, responds to specific objectives, requires a distinct sampling strategy, implements specific coding procedures, and generates tailored findings. The course also sheds light on some of the malpractices and misrepresentations that the four methods suffer in the qualitative literature both as a result of the lack of standardised training in qualitative analysis and researchers’ sloppy analytic practices and obscured reporting. To this end, the course’s daily assignment involve that participants appraise the quality of studies that used each method in the light of the content seen. The course thus promotes the development of critical skills in assessing other researchers' work as well as self-awareness to accurately report one's analysis.
At the end of this course, participants will be able to:
- Describe the aim, objectives, and expected outcomes of QCA, TA, CCA and GT
- Demonstrate how each method suits a given research design
- Implement each method’s analytic procedures in NVivo
- Generate graphic displays that match each method’s findings
- Appraise the quality of studies that used the four methods
- Propose situations where methods integration is feasible
Day 1 Qualitative content analysis (Schreier, 2012): Day 1 opens with qualitative content analysis as proposed by Schreier (2012). Qualitative content analysis is a method that is particularly suited for studies that aim to explore and then describe the manifest and latent meaning of categories in text, multimedia, pictures, and social media data. In the first part of the class, we review the methodological tenets that distinguish the quantitative from the qualitative approach to content analysis and proceed with looking at sampling requirements, coding units vs unit of analysis, and the building of a coding frame where categories are organised. This leads us to conduct the initial phase of data coding and conduct a preliminary reliability check to assess the categories adequacy to capture meaning in the data. In the second part of the class, we move to NVivo where we aggregate categories in sets and cross-tabulate them in the search of coding co-occurrence. We display the results in models where we use both qualitative and quantitative indicators to show the coding occurrence across categories.
Day 2 Thematic analysis (Boyatzis, 1998): Thematic analysis is indisputably a popular method used by qualitative researchers in the social sciences. However, when looking at the different approaches to thematic analysis, Boyatzis' approach is one of the very few that has formalised its procedures in a series of clearly-defined stages known as seeing and encoding themes, codes development, and scoring / clustering of themes. We open the class by looking at the concepts of pattern recognition and labelling consistency which are fundamental in Boyatzis’ understanding of how a theme is first seen, recognised, and then consistently ascribed the same meaning by the researcher. We then review the different ways that codes may be generated using theory-driven, research-driven, data-driven, or an hybrid-driven approach. In the second part of the class, we move to NVivo where we cross-tabulate codes in matrices to find out where coding across themes overlaps. Instances of coding co-occurrence are examined and conceptual associations are formalised in relationship nodes, NVivo's unique feature to put forward propositions, and formulate / falsify hypotheses. Theoretical sets are created as second level constructs and the patterns identified are displayed in the model.
Day 3 Cross-case analysis (Miles & Huberman, 1994): Amongst the different schools of case study research, the strategies proposed by Miles and Huberman for within- and cross-case analysis have had a tremendous impact in the way qualitative researchers examine similarities and differences across cases, so to make generalisable claims and promoting theoretical elaboration. The first part of the class centres on the first stage of cross-case analysis, that is, a description of what is going on in each case and explanations about why the phenomenon occurs the way it does. We then move on with identifying the overall pattern that gives explanation to the overall phenomenon and we formulate propositions about what could happened if similar circumstances would be met elsewhere. We reproduce these stages in NVivo using matrix queries, memos, see also links, relationship nodes and the model. In the second half of the class, we apply some of the strategies to describe the similarities and differences across cases. Finally, we examine the conditions under which a phenomenon may be generalised using NVivo's framework matrix.
Day 4 Grounded theory (Strauss & Corbin, 1998): Grounded theory is often claimed to be the method of choice by many qualitative researchers when conducting qualitative analysis. However, under scrutiny, only a scarce amount of studies actually implement the tenets proposed by the different schools of GT. The malpractice of labelling a study "a grounded theory" to legitimise one’s work while none of the methodology's tenets have been implemented, and the negative impact that this malpractice has had on the GT representation in academia, opens the first part of the class. We then locate both epistemologically and philosophically the origins of grounded theory as it was initially conceived by Glaser and Strauss in 1967, and look at the methodological developments of what was coined the postpositivist school of GT by Strauss & Corbin. In NVivo we examine the association between open coding and theoretical sampling in the generation of categories until saturation is reached. In the second part of the class, the categories of axial coding are applied onto the data and patterns of relationships between categories are identified. Emerging propositions are sought using matrix queries and those showing depth and width across respondent groups are contrasted with group queries. We conclude the class with the phase of selective coding, where a core category is identified and theoretical hypotheses are formalised using relationship nodes.
Day 5 Integration & quality appraisal: Day 5 addresses the similarities, differences, and possibilities for methods integration and proposes some tools to assess the quality of qualitative analysis. The class opens with a comparative overview of the similarities and differences of the four methods along the epistemological spectrum. This overview brings us to assess how different stances regarding knowledge creation inevitably influence the type of research questions asked, the type of analytic devices each method uses, and the level of abstraction reached in the results they generate. We then look at how, in the analytic process, some of the methods’ features – i.e. approaches to codes generation, sampling strategy, means to validate findings - may be combined in a single study only and when this is methodologically justified. In the second half of the class, we review some appraisal tools that have been proposed in the qualitative literature to assess the quality of qualitative analysis. This leads us to examine the criteria that shape these assessment tools and formulate a critique on how some tools succeed, while others and fail, to capture the process of analysis from the stages of data coding, seeking patterns, validating claims and presenting findings.
Teaching & data
Teaching methods include lectures, individual exercises with NVivo, and quality appraisal outside class hours. All four methods will be taught using sample data provided by the instructor. Participants that have their own data are welcomed to use them during the hands-on exercises. After class, compulsory readings are required for the next day. For those wishing to develop or polish their appraisal skills, quality appraisals are proposed where participants assess the implementation fidelity of each method in published qualitative studies.