Research philosophy: five types, ontology and the research onion

Research philosophy is the set of assumptions a study makes about the nature of reality (ontology), what counts as knowledge (epistemology) and the part the researcher's values play (axiology). Those assumptions shape what follows: whether you test a hypothesis or build theory from data, and whether you count or interpret. Saunders, Lewis and Thornhill's Research Methods for Business Students, the source of the “research onion”, names five philosophies: positivism, critical realism, interpretivism, postmodernism and pragmatism.

Start with the sample section below, which shows one study's assumptions deciding its methods. The five-philosophy scheme and the onion come from one business methods textbook, so check which framework your department expects before you adopt its labels.

A philosophy section in practice

Each stated assumption earns a method decision. Treating the records as a real pattern is why all three years are used; treating the interviews as accounts of meaning is why interviewees are chosen for contrast and their words analyzed rather than tallied. That link, from assumption to choice, is what a philosophy section is for, and the rest of this page explains each part of it.

Ontology, epistemology and axiology

In the eighth edition of their book (Pearson, 2019), Saunders and colleagues treat ontology as your assumptions about the nature of reality, epistemology as your assumptions about knowledge, meaning what counts as valid knowledge and how it can be passed on, and axiology as the role of values and ethics in the research. Each one is a question with a range of answers:

In the four-day-week study above, the three terms come apart like this:

• Ontology. Is staff turnover a real pattern that exists whether or not anyone studies it, or something made by how people talk about leaving? The study treats the pattern as real and the reasons people give as their own accounts of it.
• Epistemology. What counts as evidence about it: the firms' records of who left, or what staff say about why? The study accepts both kinds and judges each by whether it helps explain the change.
• Axiology. What part do the researcher's own values play? A researcher who already favors shorter working weeks says so in the methodology chapter, so readers can weigh the interpretation with that in mind.

The five research philosophies

Positivism

Auguste Comte coined the word in his Cours de philosophie positive, published in six volumes from 1830 to 1842. In research, positivism treats social reality as external and observable and looks for law-like regularities with the methods of natural science: a hypothesis derived from theory, measured variables, a large sample, statistics. Many quantitative researchers hold a softer version, often called post-positivism, that treats every finding as provisional. Its best-known idea is Karl Popper's falsifiability, from The Logic of Scientific Discovery (1959): a scientific theory is one that observation could prove wrong.

Critical realism

Critical realism grew from Roy Bhaskar's A Realist Theory of Science (1975). It holds that reality exists independently of us but is layered: the events we observe are only part of it, produced by structures and mechanisms we can't see directly. A critical realist studying why some hospital wards adopt a safety checklist and others don't would look past the adoption rates to the mechanisms behind them, such as staffing and hierarchy, and would use whatever data can reveal them.

Interpretivism

Interpretivism starts from the view that people create meaning, so the social world can't be studied like falling objects. Its roots run back to Max Weber's Verstehen, the understanding of action through the meaning the actors give it. Interpretivist research is usually inductive and qualitative: small samples, interviews and fieldwork, and findings that describe how different people experience the same situation.

Postmodernism

Jean-François Lyotard introduced the term to philosophy with La Condition postmoderne (1979), which describes the postmodern as a loss of faith in metanarratives, the grand stories that claim to explain history and knowledge. In research, postmodernism emphasizes language and power: it takes apart the dominant account of a subject, asks whose voices it leaves out, and reads texts and practices for their silences.

Pragmatism

Pragmatism judges ideas by their practical consequences. Charles Sanders Peirce stated its founding maxim in “How to Make Our Ideas Clear” (1878), William James spread it, and John Dewey turned it toward education and politics. In research it means starting from the problem and choosing whatever methods answer it, which is why it is often paired with mixed methods (see qualitative vs quantitative research).

The research onion, layer by layer

The research onion is a diagram by Mark Saunders, Philip Lewis and Adrian Thornhill in Research Methods for Business Students, and the layers below follow chapter 4 of its eighth edition. It comes from a business and management textbook, so programs in other fields may expect a different framework. Its point is order: the outer layers have to be settled and explained before you reach the choice of data collection techniques at the center. From the outside in:

1. Philosophy: the five positions above.
2. Approach to theory development: deductive, abductive or inductive.
3. Methodological choice: one method or several, qualitative, quantitative or mixed.
4. Strategy: the overall plan, such as a case study, an ethnography or action research.
5. Time horizon: one snapshot, or change followed over time.
6. Techniques and procedures: how the data are collected and analyzed.

Here is one question taken down two routes: how has remote work changed employees' engagement at a mid-sized software firm?

The three approaches in the second layer, in plain terms: deduction starts from theory and tests it, so its conclusion follows logically from premises drawn from the theory; induction builds theory from data, with conclusions the observations support but can't guarantee; abduction starts from a surprising observation and works back to the explanation that would account for it, then tests that explanation with more data. For the techniques at the core, see the types of research methods.

How to write your philosophy section

1. Start from the question

   Write the research question first and ask what kind of answer it needs: a measured effect, an explanation of causes, an account of experience. The philosophy follows from that. The guide on how to write a research question covers this step.

2. State your assumptions in plain words

   One or two sentences each on reality, knowledge and your own values. If you can't say it without jargon, you don't hold the position yet.

3. Name the philosophy and cite it

   Name the position that matches and cite a methods text for it, with the edition you read.

4. Connect it to your design

   Walk down the layers: approach, methodological choice, strategy, time horizon, techniques. Each choice should follow from the one above it.

5. Say why not the obvious alternative

   One or two sentences on the position you didn't take, and why it fits your question less well.

The sample at the top of this page follows these steps in about 160 words. In a dissertation the section sits at the start of the methodology chapter, before the design; the dissertation outline template shows where that chapter fits.

Common mistakes

• Surveying all five philosophies as if the section were a literature review. Explain yours; mention one alternative.
• A mismatch with the methods: claiming positivism, then running six open interviews, or claiming interpretivism and reporting only significance tests.
• Mixing incompatible claims, such as “an objective truth” and “meanings co-created with participants” in the same paragraph.
• Pragmatism as permission for anything. It still needs the question to justify each method.
• Confusing epistemology with methodology. The first is what counts as knowledge; the second is your argument for the design.

A mismatch is easier to spot in someone else's chapter than in your own. You can ask the Strokes Assistant to review your methodology chapter for places where the stated position and the design disagree, such as an interpretivist position followed by a purely statistical design.

Test yourself

Which philosophy does each researcher hold?

1. “I will test whether class size predicts exam scores across 300 schools.”
2. “I want to understand what grief means to the widowers I interview.”
3. “Falling recycling rates are the surface; I want the structures that produce them.”
4. “The council needs to know which outreach works, so I'll use whatever answers that.”
5. “I read the official history of the strike for the workers it never mentions.”