A note on how to do coding in qualitative data analysis: for MBA students

A note on how to do coding in qualitative data analysis: for MBA students

 

As a part-time MBA student learning to do my dissertation project, I feel that the coding method(s) is an important topic in qualitative data analysis (mainly for the inductive approach), e.g. to study research interview transcript, e.g. on the research theme of "strategic leadership in business model innovation" in case study research. Please describe 3 straightforward ways to do coding and indicate when to preferably use which one, so that I can employ the method to do my dissertation project.

A practical distinction

A code is a short label attached to a meaningful segment of an interview transcript; codes are then compared, grouped, and developed into broader themes or explanations. For an inductive study, begin with the transcript rather than forcing passages into a pre-set theoretical framework.

For a case study on strategic leadership in business model innovation, these are three straightforward and useful options.

Coding approach	What you do	Prefer it when	Illustrative code from a manager’s interview
1. Descriptive / initial coding	Label each meaningful passage with a brief noun phrase that says what it is about.	You are new to qualitative analysis, have a modest number of interviews, and want a transparent first-cycle method.	“Customer-data capability”; “Board resistance”; “Experimentation budget”
2. In vivo coding	Use a participant’s own striking word or short phrase as the initial code.	Participants use meaningful business language, metaphors, or local terms that you do not want to dilute through academic wording.	“Fail fast, learn faster”; “protecting the core”; “selling outcomes, not products”
3. Process coding	Code actions and interactions, normally using a verb ending in “-ing.”	Your research question is about how leaders make innovation happen over time, rather than only what barriers or enablers they mention.	“Reframing the value proposition”; “mobilising cross-unit support”; “testing revenue logic”

Descriptive coding provides a quick, manageable overview of the main topics in interviews and is suitable across qualitative designs, including case studies. In vivo coding preserves participants’ own language and is especially helpful for novice coders and studies that prioritise participant voice. Process coding captures actions and interactions, making it particularly apt for strategic leadership and innovation processes.

A sensible MBA choice

For your topic, a strong and practical combination is:

1.    Start with descriptive and in vivo codes for every interview transcript.

2.    Add process codes wherever interviewees describe leadership actions, decisions, negotiations, or change over time.

3.    In a second cycle, combine related codes into themes, such as building strategic legitimacy, orchestrating organisational experimentation, and redesigning the value-capture model. Thematic analysis normally moves from familiarisation, through systematic coding, to generating, reviewing, defining, and reporting themes.

For instance, imagine a participant says that senior leaders persuaded finance and operations to support a subscription-based offering after a small pilot showed promise. Rather than assigning only one code, you might code the same passage as “persuading internal stakeholders” (process), “pilot evidence” (descriptive), and perhaps an in-vivo expression such as “winning hearts and minds.” Multiple coding is often analytically valuable because one passage can serve more than one purpose.

When to use another route

If the explicit goal is to develop a new explanatory theory of how strategic leadership produces business model innovation, consider grounded-theory-style coding: open coding to identify concepts, axial coding to examine relationships among categories, and selective coding to integrate them around a core category. This is more demanding than a thematic case-study analysis and requires continual comparison between data, codes, and emerging categories.

For an MBA dissertation, it is usually more defensible to state something like: “The interview data were analysed inductively using reflexive thematic analysis. Initial descriptive, in-vivo, and process codes were applied to meaningful transcript segments; related codes were then iteratively clustered into themes addressing the research questions.” Maintain a codebook with each code’s definition, inclusion and exclusion rules, example extracts, and revisions, while also writing analytic memos explaining why you combined or reinterpreted codes. A code is a descriptive or conceptual label assigned to an excerpt, and coding frameworks can organise those codes into categories for systematic analysis.

 

I am mainly interested in theory-building that has actionable value on situational-specific managerial practice, e.g. leading change to introduce business model innovation for SME in the retail sector in Hong Kong and to do so via semi-structured research interview with SME managers. In this case, is your suggested coding practice is highly relevant?

Yes—but adapt it for theory-building

Yes. The earlier coding practice remains highly relevant, but for your purpose it should be organised as a theory-building sequence, rather than ending with a list of descriptive themes. Your study aims to explain how and under what Hong Kong SME retail conditions strategic leaders introduce business model innovation; that calls for a context-sensitive, mid-range explanation with practical propositions. Multiple-case research can strengthen such an explanation by comparing whether emerging relationships recur or differ across deliberately selected SME cases.

A suitable approach is a Gioia-inspired inductive analysis, which progresses from informant-centred concepts to researcher-centred themes and then higher-level dimensions, making the evidence-to-theory chain visible.

How to code interviews

1.    First-order concepts: code close to managers’ words.
Use descriptive and in-vivo codes, without prematurely imposing leadership or innovation theories. For example, managers might describe “the landlord will not wait,” “staff fear losing commission,” or “we have to try it in one shop first.” Preserve such phrases because they may represent locally important mechanisms rather than generic barriers.

2.    Second-order themes: interpret actions and mechanisms.
Now use process coding and analytical comparison: ask what leaders are doing, why it matters, and what consequence it produces. The illustrative first-order concepts above might contribute to themes such as managing immediate cash-flow exposure, protecting employee identity and incentives, and de-risking change through bounded experimentation. This is comparable to axial coding: relating categories and identifying patterns rather than merely listing topics.

3.    Aggregate dimensions and a process model: build the explanation.
Cluster second-order themes into a small number of overarching dimensions—for example, contextual pressures, leadership change practices, and business-model adaptation outcomes. Then develop a tentative process relationship, such as: Hong Kong retail constraints → leadership sensemaking and risk containment → staged business-model experimentation → adoption or adjustment. The resulting data structure should show the traceable movement from interview evidence to constructs; it is evidence for the theory, not the theory itself.

What makes it actionable

Make each emergent theme answer four questions:

·        Condition: Under what retail-SME circumstance does it occur?

·        Leadership action: What specifically does the manager do?

·        Mechanism: Why does that action influence employees, partners, customers, or resource allocation?

·        Outcome: What business-model element changes—value proposition, value delivery, or revenue/cost logic?

For example: where a small retailer has limited cash reserves and uncertain customer uptake, a leader may run a limited pilot, which reduces perceived downside risk and produces local evidence, enabling a revised digital or service-based offering. This can eventually become a practice-oriented proposition: When financial slack is low and uncertainty is high, staged pilots help SME leaders mobilise commitment to business-model change. Cross-case comparison should also seek disconfirming cases—firms where pilots did not build commitment—to refine the boundary conditions.

Recommended position

Do not claim a completely theory-free analysis: use prior literature on strategic leadership, change, and business model innovation as a sensitising lens for interview design and later interpretation, while allowing your codes and mechanisms to emerge inductively. This is commonly described as an iterative or abductive stance, and it prevents generic findings from being presented as locally grounded theory.

Your next coding decision is to define one unit of analysis: will each case be an individual SME’s entire business-model-change episode, or a particular innovation initiative within an SME?

 

I am not familiar with the Gioia-inspired inductive analysis topic. Could you provide 3 academic references on this topic (in Harvard reference format)?

Three useful references

These three sources provide a sound starting set. The first is the core methodological article; the second supports the theory-building logic behind a practically useful contribution; the third helps you position case-study theory building alongside a Gioia-inspired analysis.

1.    Gioia, D.A., Corley, K.G. and Hamilton, A.L. (2013) ‘Seeking qualitative rigor in inductive research: Notes on the Gioia methodology’, Organizational Research Methods, 16(1), pp. 15–31. Available at: https://doi.org/10.1177/1094428112452151.journals.sagepub+1

2.    Corley, K.G. and Gioia, D.A. (2011) ‘Building theory about theory building: What constitutes a theoretical contribution?’, Academy of Management Review, 36(1), pp. 12–32. Available at: https://doi.org/10.5465/amr.2009.0486.aom+1

3.    Eisenhardt, K.M. (1989) ‘Building theories from case study research’, Academy of Management Review, 14(4), pp. 532–550. Available at: https://doi.org/10.5465/amr.1989.4308385.mackenzie

For your dissertation, read Gioia, Corley and Hamilton (2013) first: it explains the progression from first-order, informant-centred terms to second-order, theory-centred themes and aggregate theoretical dimensions. Corley and Gioia (2011) is especially relevant to your aim of producing an original contribution with pragmatic managerial value.

 

Highlight 5 main ideas of the Gioia-inspired inductive analysis?

Five main ideas

The Gioia-inspired approach is a systematic way to develop concepts and theory from qualitative interview data while preserving participants’ perspectives. It is particularly useful when your goal is theory-building rather than merely summarising interview topics.

1. Treat participants as knowledgeable agents

Interviewees are treated as people who understand their own organisational realities and can explain what they are trying to achieve, why they act, and how they interpret events. Early analysis therefore gives priority to managers’ own words rather than immediately imposing established theories.

For your study, a Hong Kong retail SME manager’s phrase such as “we tested it in one shop first” should initially be retained rather than immediately translated into an abstract concept such as incremental innovation.

2. Begin with first-order concepts

First-order coding stays close to the informants’ language and experience. You identify meaningful phrases, actions, problems, and interpretations from the transcripts, often using descriptive, in-vivo, and process codes.

Possible first-order concepts include:

• “Testing in one shop first.”
• “Staff fear losing commission.”
• “Customers still want to see the product.”
• “The landlord will not wait.”

The objective is not to reduce the data too quickly, but to preserve potentially important distinctions.

3. Move to second-order themes

After developing first-order concepts, you ask more analytical questions: What is happening here? Why does it matter? What broader process might explain several related codes?

For example:

First-order concepts	Possible second-order theme
Testing in one shop first; starting with a small budget	Containing innovation risk
Staff fear losing commission; explaining the new reward system	Reconfiguring employee incentives
Customers still want to see the product; combining online and physical selling	Integrating digital and physical value delivery

Second-order themes use more researcher-centred and theoretical language, but they should remain demonstrably grounded in the first-order evidence.

4. Develop aggregate dimensions

You then combine related second-order themes into a small number of broader theoretical dimensions. These dimensions represent the main building blocks of your emerging explanation.

For example:

• Contextual pressures: financial constraints, competition, rental costs, changing customer expectations.
• Leadership change practices: sensemaking, stakeholder mobilisation, risk containment, incentive redesign.
• Business-model innovation outcomes: new value propositions, new delivery channels, new revenue mechanisms.

Aggregate dimensions should not simply be broad headings; they should help explain the phenomenon being studied.

5. Turn the data structure into a process model

The well-known data structure displays the movement from:

First-order concepts→ Second-order themes→ Aggregate dimensions 

It demonstrates how interview evidence was transformed into theoretical constructs. However, the data structure is not itself the final theory: the theory is the explanatory model showing how the dimensions relate and unfold over time.

For your research, a tentative model might be:

Retail-sector pressures → leadership sensemaking → stakeholder mobilisation and risk containment → staged business-model experimentation → revised retail business model

You would then investigate whether this sequence is supported, modified, or contradicted across different Hong Kong SME cases.

Important qualification

Although it is commonly called an inductive approach, Gioia-inspired analysis is usually better understood as iterative and partly abductive. You move repeatedly between interview data, emerging concepts, relevant literature, and further questions, rather than pretending that theory is produced from data without prior knowledge.

For an MBA dissertation, you can describe your approach as: “a Gioia-inspired, inductive-abductive analysis involving first-order concepts, second-order themes, aggregate dimensions, and an emerging process model.” This is more accurate than claiming that you used the full Gioia methodology as a rigid formula.