Saunders’ (2019) Research Onion

Your methodology chapter has to explain why you chose your methods, not just what they were. That’s where most students come unstuck: they pick a method that sounds right, then can’t say what assumptions sit underneath it when their advisor (supervisor, in the UK) asks.

Saunders’ research onion is the tool built for exactly that problem, because it lays out the key decisions you’ll need to justify, in order. Having worked 1-on-1 with thousands of students on their methodology, we’ve found it’s most useful as a checklist of choices to defend. This post walks through all six layers of the onion, every option inside each one, with an example for each, and then shows you how to turn your choices into a methods chapter.

---

What is the research onion?

Saunders’ research onion is a framework from Saunders, Lewis and Thornhill’s textbook Research Methods for Business Students that sets out the six layers of decisions behind a research methodology: research philosophy, approach to theory development, methodological choice, research strategy, time horizon, and techniques and procedures. You work from the outside in, and each outer decision narrows the ones inside it.

The onion has appeared in several editions of the book and has grown along the way. This post follows the figure in the 2019 (8th) edition, which Mark Saunders has shared openly. The 2023 (9th) edition keeps the same philosophies and adds retroduction as a fourth approach to theory development, which we’ll cover briefly in layer 2.

The point of the onion-based shape is that your data collection sits at the very center. Most students start there (“Should I do interviews or a survey?”), but the authors argue that the outer layers are ones you need to “understand and explain rather than just peel and throw away”. Your committee (examiners, in the UK) wants to see that your interviews follow from your view of knowledge, not that you happen to like interviews.

Saunders’ research onion helps you achieve methodological alignment, where every choice follows from the ones before it and from your research question. That’s also why the onion mirrors the general flow of a methodology chapter, and why it’s a handy way to think about your research design as a whole.

---

The layers of Saunders’ research onion

The onion has six layers, peeled back here one at a time:

1. Research philosophy
2. Approach to theory development
3. Methodological choice
4. Research strategy
5. Time horizon
6. Techniques and procedures

Onion layer 1: Research philosophy

The outer layer is your research philosophy: the set of beliefs and assumptions your study is built on. Saunders and colleagues describe three kinds of assumption that make up a philosophy, and you’ll meet all three words in your reading:

• Ontology is about the nature of reality. Does reality exist as one objective thing, the same for everyone, or is it different for each person? Think of the simulated world in The Matrix.
• Epistemology is about knowledge: what counts as acceptable, valid knowledge, and how you can come to know something.
• Axiology is about the role of values, including your own, in the research.

That’s a gross oversimplification, but it’s enough to work with (there’s a full explainer on ontology vs epistemology). The research onion names five philosophies. Positivism sits at the objective end, interpretivism and postmodernism toward the subjective end, critical realism between them, and pragmatism tries to reconcile both:

• Positivism
• Critical realism
• Interpretivism
• Postmodernism
• Pragmatism

Research philosophy 1: Positivism

Positivism is the stance of the natural scientist. It assumes one real, external reality that can be observed and measured, and it aims to produce law-like generalizations, the way physics does. The researcher stays detached and neutral, so their values don’t color the findings.

In practice, positivist studies are usually deductive, highly structured and quantitative, with large samples and a lot of measurement. In this business study, watch how little room the researcher leaves for their own views:

Notice what the researcher doesn’t ask: how employees feel about the change. If your question is about something you can count and you want a finding that generalizes, positivism is probably your home.

Research philosophy 2: Critical realism

Critical realism also believes in a real world that exists independently of us, but argues that we can’t observe it directly. What we experience are sensations and events, and the researcher’s job is to reason backwards from them to the underlying structures that produced them. The philosophy grew out of the work of Roy Bhaskar, and it sits between positivism and the subjective philosophies below.

Saunders’ illustration is the sponsor’s logo you see “standing” on a sports field on TV, which is actually painted flat on the grass. What you see is real, but it isn’t the whole story. In a study, that thinking looks for causes behind the numbers:

The lesson for your own project: if you’re trying to explain why something happens, and you suspect the answer lies in structures people can’t see from where they stand, critical realism gives you a way to say so.

Research philosophy 3: Interpretivism

Interpretivism developed as a critique of positivism. It argues that people aren’t like the objects physicists study, because we create meanings, and those meanings are shaped by culture and language. So there are multiple realities, and the researcher’s job is to understand them from the inside. You can’t stay detached, either: your own interpretation is part of the research.

Interpretivist studies are typically inductive, with small samples and in-depth, qualitative work. In this project, notice that the researcher is after each student’s meaning, not a measurement:

If your research question starts with “how do people experience…” or “what does it mean to…”, you’re probably working in interpretivist territory, and your methodology chapter should say so.

Research philosophy 4: Postmodernism

Postmodernism focuses on language and power. It holds that what counts as “truth” is decided by dominant ideas, and that some voices are routinely silenced by them. So a postmodernist researcher questions accepted ways of thinking and looks for what’s missing: the absences, silences and suppressed views in a text or an organization. It draws on French philosophers such as Michel Foucault and Jacques Derrida.

The typical method is deconstruction, reading a text against itself. For example, a researcher might analyze ten years of a company’s diversity reports and ask whose experiences never appear in them, and what the reports’ language makes it impossible to say. It asks a lot of reflexivity from you, so choose it deliberately rather than because it sounds radical.

Research philosophy 5: Pragmatism

Pragmatism cares about what works. Concepts only matter where they support action, so a pragmatist judges a theory by its practical consequences. Research starts with a problem and aims to produce a practical solution, using whatever mix of methods best answers the question: quantitative, qualitative, mixed or action research.

A health services project shows the mindset, because the problem picks the methods:

The authors add a warning worth taking seriously: you need to be sure you’re “not treating pragmatism as an escape route from the challenge of understanding other philosophies!” Choose it because your problem calls for mixed methods, not because philosophy feels hard.

To recap, your philosophy is the foundation of the whole project and reflects your assumptions about reality and knowledge. So when you’re designing your research methodology, it’s the first thing to settle, and everything below has to be consistent with it.

Onion layer 2: Research approach

Peel off another layer and you reach your approach to theory development: the reasoning your study uses to connect theory and data. It matters because it shapes what data you collect and how you analyze it. The onion names three approaches.

Deduction starts with theory, usually from the literature, and designs research to test it. You form a hypothesis and collect data to see if it holds. For example, a study testing whether the technology acceptance model predicts which nurses adopt a new e-prescribing system is deductive: the theory comes first.

Induction goes the other way, collecting data to explore something and building theory from the patterns you find. It suits topics where existing theory has little to say. For example, a study of how food delivery riders cope with being managed by an app would start with the riders’ accounts and work toward an explanation.

Abduction moves back and forth between the two. It starts with a “surprising fact”, works out a plausible theory that would explain it, and then tests that theory with more data. Saunders and colleagues note that this matches what many business and management researchers actually do in practice. Watch for the surprising fact in this retail case, because everything else follows from it:

Notice that abduction doesn’t pretend you started with a clean theory or a blank slate. If your project began with a finding that didn’t fit what you expected, abduction may describe your reasoning more honestly than either alternative.

The 2023 edition adds a fourth, retroduction: reasoning backwards from what you observe to the underlying causes that might explain it. It’s the reasoning critical realists use, so if you’ve chosen that philosophy, it’s worth reading up on.

How do the approaches line up with qualitative and quantitative research? Deduction tends to go with quantitative work and positivism, testing hypotheses on numbers. Induction tends to go with qualitative work and interpretivism. Those are tendencies, not rules (our post on qualitative vs quantitative research covers the difference in full), so name your approach and say why it fits your question.

Ask yourself: does my research build on a theory that already exists, start from something theory can’t yet explain, or start from a finding that surprised me? The cards below line up the three approaches against exactly those questions.

Onion layer 3: Methodological choice

The third layer is about how many methods you’ll use, and of what type. Saunders calls it methodological choice, and it offers three options.

A mono method study uses a single data collection technique with its matching analysis, either quantitative or qualitative. For example, if you were studying a community’s opinions of a new pizza restaurant, you might run one set of interviews and analyze them qualitatively.

A multi-method study uses more than one data collection technique, but stays on one side of the line: all quantitative or all qualitative. Keeping with the pizza example, you might run interviews and focus groups, both analyzed qualitatively. Plenty of guides use “multi-method” loosely for any mix, but in the onion it means several methods of the same type.

A mixed method study combines quantitative and qualitative data. Back to the pizza: you might add a survey asking how often people eat at the restaurant, analyzed statistically, alongside your interviews. The onion splits mixed methods further into simple and complex designs, depending on how far the two kinds of data are combined.

As with every layer, the right choice depends on your research aims and objectives. There’s also a practical question: what data can you actually get, in the time you have? Mixed methods can be powerful, but it’s two studies’ worth of work.

Onion layer 4: Research strategy

So far, the layers have been fairly conceptual. Now it’s time to get practical with your research strategy: the overall plan for how you’ll answer your research question. Many methods textbooks call these research designs. The research onion lists eight:

• Experimental research
• Survey research
• Archival and documentary research
• Case study research
• Ethnography
• Action research
• Grounded theory
• Narrative inquiry

These aren’t mutually exclusive. A case study can contain a survey, for instance, so pick the strategy that answers your question and say how any others support it.

Strategy 1: Experimental research

Experimental research changes one variable (the independent variable) to see its effect on another (the dependent variable). Its purpose is to support or refute a research hypothesis. It follows the principles of the scientific method and runs in a controlled setting, such as a laboratory, so that nothing else explains the result.

Because it tests existing theory rather than creating new theory, experimental research is deductive, and it fits a positivist philosophy. The textbook setup has two groups and one change:

Notice that the control group is what makes this an experiment. Without it, you couldn’t tell the brand’s effect from anything else that changed. If you can’t control conditions like this in your own project, a survey or case study is usually the more honest choice.

Strategy 2: Survey research

A survey collects standardized data from a sizeable group, usually through a questionnaire, so you can compare answers and generalize to a wider population. It’s one of the most common strategies in business research, and you’ll most often see it with a deductive, quantitative approach.

For example, if you wanted to know whether small-business owners who use accounting software report fewer cash-flow problems, you could send a questionnaire to a few hundred owners and test the relationship statistically. The strength is scale. The trade-off is depth: a survey tells you what and how many, rarely why. Our survey design guide covers how to build one that holds up.

Strategy 3: Archival research

An archival research strategy, which the onion calls archival and documentary research, uses materials that already exist rather than data you collect yourself. Meaning is built by reviewing those records. It suits historical research, and it’s common in business too, where board minutes, annual reports, emails and policy documents are all fair game.

For example, if you were interested in people’s beliefs about supernatural phenomena in the medieval period, you could use manuscripts and records from the time as your core data set. In a business setting, you might trace how a company’s stated strategy changed by reading fifteen years of its annual reports.

Strategy 4: Case study research

A case study is a detailed, in-depth study of a single subject: a person, a group, an organization, an event or an issue. You study the subject in its real-life setting to understand it deeply, not (necessarily) to generalize the findings.

Because context matters so much here, case studies are more often than not qualitative and inductive, and since your own understanding plays a role, they’re often interpretivist. They don’t have to be, though: a case study can include quantitative data too.

For example, a study of political views in one community would need to take the country’s political situation into account, along with anything else that might push people toward a particular view. If your question is about how something works in one particular setting, a case study lets you go deep without having to claim your findings hold everywhere.

Strategy 5: Ethnography

Ethnography means studying a group in its natural setting, usually by spending extended time with it, to understand its behavior, language and interactions from the inside. The aim is to see the world the way the group sees it.

For example, if you were interested in how people support each other on an online mental health forum, you could take part in the forum over several months and analyze its interactions to understand the members’ experiences.

Ethnography generally reflects an interpretivist philosophy and an inductive, qualitative approach. There are exceptions, though: quantitative ethnography, developed by David Williamson Shaffer, applies statistical tools to ethnographic data. Whichever kind you choose, ethnography takes time in the field, so check you can get sustained access to the group before you commit to it.

Strategy 6: Action research

Action research is learning through action. It’s conducted in practical settings, such as a classroom, a hospital or a workplace, rather than a lab, and it aims to solve a real problem while studying it. Participants are closely involved; in participatory action research (PAR), they’re full partners in the research.

A typical example is a community intervention. The researcher comes with an idea, the community helps implement it, and the results are discussed together to improve the intervention. The cycle repeats until the intervention works for that community.

Action research is common in education, health and management, where people want to improve their own practice. It often leans qualitative, but it can draw on whatever data the problem needs. If your project sits inside your own workplace and you’re in a position to change how things are done, action research lets you study that change as you make it.

Strategy 7: Grounded theory

Grounded theory is about letting the data inform the development of a new theory, model or framework. True to the name, the theory you develop is “grounded” in the data, which makes it useful for issues that are new or under-researched.

Grounded theory is typically qualitative and inductive, although it can use quantitative data. You collect and analyze data in rounds, comparing each new piece of data with what you’ve found so far, and you don’t try to fit the findings into a theory you chose in advance. For example, a study of how first-generation doctoral students decide whether to quit could start with interviews and build an explanation from what they say, rather than testing an existing model of dropout.

Strategy 8: Narrative inquiry

Narrative inquiry collects people’s stories and analyzes them as stories, keeping their sequence and context intact rather than cutting them into separate themes. It assumes that people make sense of their experiences by telling them as narratives.

For example, if you were researching why experienced nurses leave the profession, you could ask a small number of them to tell the story of their career, from training to the day they left. You’d then analyze how each account is built: the turning points, who they blame, and what they leave out. Narrative inquiry is qualitative and usually interpretivist, and it works best with a few rich accounts rather than many short ones.

As you can see, there’s a wide range of strategies. The right one depends largely on your research aims and objectives, as well as your choices of philosophy and approach. Here are all eight side by side, with what each one is best for and the data it usually uses.

Onion layer 5: Time horizon

Next is the time horizon. Thankfully, this one’s pretty straightforward. It describes how many points in time you’ll collect data at, and there are two options: cross-sectional and longitudinal.

Imagine you’re wasting time on social media and think, “Ooh! I want to study the language of memes and how it changes over time.” For that, you’d need to collect data at several points in time, perhaps over a few months or even years. That’s a longitudinal time horizon, and it’s the right choice when you’re studying change.

If instead you wanted to study the language of memes at one moment (say, a single month in 2020), you’d use a cross-sectional time horizon: data collected at one point in time, a snapshot rather than a story of change. The type of data can be qualitative, quantitative or both, because this layer is about when you collect it, not what kind it is.

As with the other choices, your research aims are the main deciding factor. Be realistic about time, though: a dissertation or thesis rarely leaves room for a true longitudinal study, which is one reason most student projects are cross-sectional.

Onion layer 6: Techniques and procedures

Finally, you reach the center of the onion: your techniques and procedures. This is where you get down to the practicalities. Specifically, this is where you’ll:

• Decide what data you’ll collect and how, for example through a survey or 1-on-1 interviews.
• Decide how you’ll sample the population, for example through snowball, random or convenience sampling.
• Choose the data analysis method you’ll use to answer your research questions, such as content analysis or a statistical test like correlation.
• Prepare your materials, such as the questions for a survey or an interview guide.

The important point is that these techniques need to align with every layer above them: your philosophy, approach, methodological choice, strategy and time horizon. For example, if you’ve taken a deductive, quantitative approach, you’re unlikely to use interviews to collect your data, because you’ll want numerical data from many people, which a survey suits far better. Check each layer against the others, and above all against your research aims and objectives.

---

How to use the research onion

Here’s the important part. The research onion is a planning tool, not a template for your methodology chapter. Don’t structure your chapter as Philosophy, Approach, Choice and so on unless your university specifically asks for it.

Instead, work through the onion early, while your advisor can still weigh in on your design, and write down a sentence or two for each layer: what you chose and why it fits your research question. Those notes become the justifications your chapter needs.

When it’s time to write, follow your university’s handbook and look at a few past dissertations from your department. Use the structure they expect, and drop your onion decisions into it. Our guide to writing the methodology chapter shows that structure with real extracts. And if you’d like someone to pressure-test your layers before you commit, our methodology coaches do exactly that.

Still have questions?

What are the six layers of Saunders’ research onion?

In the 2019 version, from the outside in: research philosophy, approach to theory development, methodological choice, research strategy, time horizon, and techniques and procedures. Each layer’s decision narrows the options in the layers inside it, so you work from philosophy down to data collection.

Do I have to use the research onion in my dissertation?

No. It’s a thinking tool, not a requirement, though many advisors (supervisors, in the UK) like it because it shows that your methodology decisions connect. If your university prescribes a methodology structure, follow that, and use the onion to check that your choices align.

Which research philosophy should I choose?

Choose the one that fits your research question, not the one that sounds best. Measuring something to produce generalizable findings points toward positivism; explaining the hidden causes of what you observe, toward critical realism; understanding people’s experiences, toward interpretivism; questioning dominant ideas and silenced voices, toward postmodernism; and solving a practical problem with mixed methods, toward pragmatism.

Is the research onion only for business students?

No, but it fits some fields better than others. It comes from a business research textbook, Research Methods for Business Students, and the decisions it maps (philosophy, approach, methods, strategy and timing) apply to most social science research, such as education, nursing, psychology and management. It’s a looser fit for fields that frame their methods differently, such as the lab sciences or the humanities.

How do I cite Saunders’ research onion?

Cite the edition you used. In APA 7, the 2019 edition is: Saunders, M. N. K., Lewis, P., & Thornhill, A. (2019). Research methods for business students (8th ed.). Pearson. The 2023 9th edition is cited the same way with its own year and edition number.

What are the limitations of the research onion?

It makes research look more linear than it is. In practice you’ll move back and forth between layers, and some studies don’t fit neatly into one philosophy or strategy. It also comes from business research, so fields outside the social sciences may frame these choices differently. Treat it as a checklist of decisions to justify, not a set of boxes to tick.

Is a research strategy the same as a research design?

Largely, yes. Many textbooks use the two terms interchangeably. What the onion calls a strategy (experiment, survey, case study, ethnography and so on) is what most methods guides call a research design: the overall plan for answering your research question.

Can’t find your answer here? Ask a Grad Coach directly – the initial chat is free.

Speak with a friendly coach →