What is qualitative data analysis (QDA)?
QDA is the search for patterns in textual, visual or artefact data to uncover associations, identify relationships and propose explanations about the studied phenomenon.
The process of analysis involves several steps, often described in sequence although much intertwined in practice, known as getting familiar with the material, coding the data, identifying patterns and generating results and presenting findings.
A range of methods of qualitative analysis exists (thematic analysis, grounded theory, content analysis, etc.) all of which come with their epistemological standpoint, sampling requirements, coding procedures, techniques to generate findings and quality criteria.
This course does not focus on a method of QDA in particular; rather, it explores the common denominators all methods share and provides tools and food for thought to implement them.
Why is this course relevant?
Historically, QDA has been criticised for being opaque and subjective given that it is sometimes difficult to see how researchers went from hundreds of interview pages to a handful of conclusions since no discussion is provided about what the researchers actually did when they analysed their data. For this reason, QDA has a ‘black box’ association problem: we are told about the data, how they were collected and what results they yielded, but nothing in between. There is also a misconception that QDA merely involves the identification of themes in data and the reporting of themes in the form of quotes, often lots of them.
The analytical task researchers should be concerned about – that is, the examination of commonalities and differences across units of analysis, discovering patterns and relationships across the data, and synthesising these in a storyline, model, or schema – is usually absent in published qualitative studies. Conversely, the widespread myths that the ‘method will emerge’ from the data, or that the researcher’s tacit knowledge will lead him/her to ‘make sense’ of the data, account for opaque reporting since no audit trail accompanies the analytic process.
What will I learn on this course?
How to plan, conduct and report QDA in a transparent, traceable and auditable way in your research. You will learn
- key design issues to consider when planning the analysis phase in qualitative study
- different approaches to code qualitative data
- strategies to transform qualitative data into findings
- debates about the quality of qualitative research and techniques to confirm the trustworthiness of qualitative findings
- best and worst practices when presenting qualitative findings in doctoral theses and journal articles.
The different stages of QDA will be demonstrated with NVivo software so you can put concepts into practice and learn the advantages and pitfalls of QDA software. To that end, please use NVivo 12 Pro for Windows, NVivo 12 for Mac or other QDA software you are familiar with. See Software and Hardware Requirements.
- To demonstrate the influence that methodology has on the choice of data analysis.
- To describe the different approaches to coding qualitative data.
- To distinguish the strategies for transforming qualitative data into findings.
- To situate the competing philosophies regarding validity in qualitative research.
- To present qualitative results evocatively and effectively.
At the end of this course, you will be able to
- develop an analytical plan of your study
- choose the right approach to code your data
- identify what strategy to use to transform data into findings
- reflect on your position regarding validity in qualitative research
- present qualitative findings convincingly and transparently.
Day 1 – Foundations of QDA
We open with a lecture on the foundations of qualitative analysis with definitions, historical problems and challenges ahead. The qualitative analysis cycle is introduced as a heuristic device to understand that qualitative analysis, often if not always, occurs iteratively between the phases of coding data, generating meaning, confirming findings and presenting results. We situate the phase of qualitative analysis within the research design and consider the influence that research questions, study purpose and type of data collected have on the choice of data analysis method. We then look at some popular approaches to QDA; I present in turn the aim, sampling requirements and analytical procedures of qualitative content analysis, thematic analysis, cross-case analysis, grounded theory and analytic induction. During the second half of the class you will set up an NVivo project and import and classify data in preparation for Day 2.
Day 2 – Concepts and approaches for coding qualitative data
In qualitative research, coding is defined as the process by which data are segmented in coding units and assigned a code that represents a concept, construct or idea. Coding is a core task in qualitative analysis, so knowing how to code one’s data meaningfully and efficiently is key.
We discuss the key concepts that shape the coding process – meaning and coding units, codes, codebook and coding schemes – alongside situations where descriptive, interpretive, and pattern codes can be used to capture distinct levels of abstraction in the data. We then examine the specificities of theme- versus design-based coding schemes and review their respective strengths and requirements. We conclude by using visualisations to map the coding process and the emerging relationships between codes. Finally, in a coding workshop with NVivo, you will code your own data and reflect on the process.
Day 3 – Transforming data into findings
‘I have a mountain of information here. Which bits go together?’ (Miles & Huberman, 1994: 256). Coding is a first step in the transformation process because it reduces the amount of data in codes and organises them in thematic families. However, coding alone is no analysis.
Qualitative analysis involves the researcher detecting the structure that lies behind the data and explaining how the structure connects people, places, processes or practices in one coherent storyline. To do so, we propose a range of strategies for transforming data, and follow with a discussion on the logic of scientific reasoning: in turn, I present examples of inductive, deductive, abductive and retroductive analyses. I then introduce the nuts and bolts of retrieval procedures that operate on data concurrence, overlap, sequence, proximity, precedence and exclusion in NVivo.
We conclude with a discussion on the ladder of abstraction in qualitative analysis.
Day 4 – Presenting qualitative findings
Beyond using quotes to illustrate participant views, in what ways can qualitative findings be presented? What are the best and worst practices when presenting qualitative findings in journal articles and theses? And what about the ubiquitous phrase ‘themes were identified in the data’, when themes were part of the interview guide?
We explore these questions so you gain a critical understanding of how visual display(s) can tell the story of their findings effectively and evocatively. To that end, we learn that models are best to illustrate conceptual integration, matrices are good for cross-tabs, tables work for typology, and diagrams suit the depiction of structure.
We move on with a discussion and examples of the reporting of analysis, results and interpretation in a qualitative report.
The second part of the class is a workshop during which you assess the the reporting of qualitative analysis in a qualitative study.
Day 5 – Validating qualitative results
Validity of qualitative research is a debated issue among qualitative scholars because it brings together divergent perceptions and antagonistic practices, depending on whether you espouse empiricism, critical realism, constructionism or subjectivism as your anchored epistemology.
We review the current debates on the quality of qualitative research, examining what the conventional scientific paradigm, the naturalistic paradigm, and other complementary strategies, have put forward since the 1980s.
I then ask you to reflect on your own epistemological positionality, individually, then in small-group discussions. We look at different techniques to confirm the trustworthiness of qualitative results, among which we discuss member checking, researcher effects, triangulation and using deviant cases, and comment on their appropriateness depending on the research context and the nature of data collected.
We conclude with a masterclass in which teams present examples of qualitative analysis of a research project.