Why Use Triangulation in Qualitative Research?

Triangulation borrowed its name from surveying, where you fix an unknown position by taking bearings on it from two known points. The metaphor is doing a lot of work, and not all of it is honest. In surveying there is a single true position waiting to be found. In qualitative research, whether there is a single truth waiting to be found is exactly the question people disagree about.

That tension is the most interesting thing about triangulation and the thing almost every guide leaves out. Below: what the four types actually are, what triangulation can and cannot buy you, what to do when your sources disagree, and how to write it into a methods section without overclaiming.

<CTA title="Keep Your Evidence and Your Claims in Line" description="Jenni helps you draft, cite, and check that every claim matches what your data can support" buttonLabel="Try Jenni Free" link="https://app.jenni.ai/register" />

What Is Triangulation?

Triangulation is the practice of examining a phenomenon from more than one vantage point, so that your conclusion does not rest on a single source, a single method, a single analyst, or a single theory. The term entered social research through Norman Denzin, whose The Research Act set out the typology everyone still uses.

The core idea is straightforward: any single way of gathering evidence has characteristic blind spots, and those blind spots are not random. Interviews capture what people are willing to say. Observation captures what they do while being watched. Documents capture what an organization chose to write down. Each of those distortions is systematic, so adding a second angle does not just add data, it adds a different set of distortions that may not overlap with the first.

Denzin's Four Types of Triangulation

Denzin distinguished four, and the distinction between them is which single thing you are varying.

Data triangulation

Vary the source. Denzin gave this three subtypes: time, space, and persons. You might interview the same group in March and again in September, run the same protocol at two different sites, or speak to nurses, doctors, and patients about the same episode. The method stays constant; who or when or where changes.

Investigator triangulation

Vary the researcher. Two or more analysts collect or code independently and then compare. This is the type most likely to be mistaken for a reliability check, and treating it purely as an inter-rater agreement exercise misses the point: the interesting output is often where two experienced readers saw different things in the same transcript, not where they agreed.

Theory triangulation

Vary the lens. Interpret the same dataset through more than one theoretical framework and see what each one brings into focus. This is the least used of the four and often the most productive, because it exposes how much of a finding was in the data and how much was in the theory you brought to it.

Methodological triangulation

Vary the method. Denzin split this into within-method and between-method variants, a distinction covered in more depth in the companion piece on data versus methodological triangulation.

A fifth type, environmental triangulation, appears in a University of Florida extension publication, defined as using different locations, settings, and other key factors related to the environment in which the study took place. Treat it as a useful extension rather than as one of Denzin's original four.

<ProTip title="🔺 Angle it:" description="Before adding a second source, write down what your first one systematically cannot see. If the second source shares that same blind spot, you have added volume rather than triangulation" />

Confirmation or Completeness?

Here is the argument the popular guides skip entirely. Triangulation is usually sold as a validity booster: check the same thing two ways, and if the answers match, you can be more confident. That framing is contested, and it has been for decades. The classic statement of the tension is Breitmayer, Ayres and Knafl's 1993 paper in Image: Journal of Nursing Scholarship, titled Triangulation in Qualitative Research: Evaluation of Completeness and Confirmation Purposes, which names the two purposes in its title.

The confirmation reading says convergence between sources increases confidence that a finding is real. It assumes there is a single account to converge on.

The completeness reading says different methods reveal different aspects of a phenomenon, so the point is a fuller picture rather than a verified one. Denzin himself is quoted to this effect: different methods reveal aspects of a phenomenon that the use of a different method may not uncover.

Which one you claim has consequences. If you write that triangulation established the validity of your findings, you have committed to the confirmation reading, and a constructivist examiner may well ask you to defend the assumption that a single fixed account exists to be validated. If you write that it produced a more complete account, you have made a smaller claim that is much easier to support.

<ProTip title="🤔 Not proof:" description="Two sources agreeing may mean the finding is robust, or it may mean both sources share the same bias. Convergence is evidence, not confirmation, and saying so in your methods section signals that you understand the difference" />

What to Do When Your Sources Disagree

Most students treat disagreement as a failure. It is not, and there is a recognized vocabulary for handling it properly.

O'Cathain, Murphy and Nicholl set out three techniques for integrating data in the BMJ, one of which is a triangulation protocol built around a convergence coding matrix. It sorts findings into four categories, and their key line is worth quoting to your supervisor: disagreement is not a sign that something is wrong.

Silence is the category nobody teaches and the one that produces the sharpest insights. If a concern is raised repeatedly in interviews and appears nowhere in the organization's own documents, that absence is a finding about the organization, not a gap in your data.

An older and more detailed protocol exists in Farmer and colleagues' triangulation protocol for qualitative health research, which argues that convergence, complementarity and dissonance all have to be addressed and notes that there was, at the time, little guidance on operationalizing the process.

<ProTip title="🗣️ Disagreement:" description="When two sources conflict, resist the urge to decide which one is right. Ask instead what each method was positioned to see, because the explanation for the conflict is usually more informative than either finding on its own" />

The Critique Worth Knowing

If you are working in a constructivist or interpretivist tradition, someone will eventually ask whether triangulation is compatible with your epistemology. Have an answer ready.

The objection is that triangulation as validation presupposes a single objective reality that multiple methods converge on. Braun and Clarke, quoted in a 2024 Qualitative Report review, put it directly: triangulation strategies cannot take researchers closer to a single, fixed truth because for them, this type of thing is nonexistent. A related worry is that using triangulation to eliminate bias discards researcher subjectivity and reflexivity, which many qualitative traditions regard as resources rather than contaminants. The canonical formal critique is Blaikie's 1991 paper in Quality and Quantity, A critique of the use of triangulation in social research.

The most striking piece of evidence is that Denzin himself moved. A 2018 review in the Journal of Social Change notes that he later proposed reframing triangulation as crystal refraction, many points of light, and became critical of naive postpositivist readings of mixed methods.

That points toward crystallization, Laurel Richardson's alternative. Her formulation: in postmodern texts we do not triangulate, we crystallize, and the central image is not the rigid two-dimensional triangle but the crystal, which combines symmetry and substance with an infinite variety of shapes, transmutations, multidimensionalities, and angles of approach. Triangulation converges toward one account. Crystallization multiplies partial, situated ones.

How to Report Triangulation in Your Methods Section

Two practical points, one of which is checkable and almost never mentioned.

The SRQR reporting standard for qualitative research names triangulation twice, once under data collection methods and once as a technique to enhance trustworthiness. COREQ, the other widely used qualitative checklist, does not mention triangulation anywhere in its 32 items; its nearest equivalents are the number of data coders, participant checking, and data saturation. If your target journal asks for COREQ, triangulation is not a box you are ticking, which means you have to justify it on its own terms rather than by pointing at a checklist.

What a good methods paragraph contains: which type you used and why that type, what specifically you varied, at what stage of the analysis you compared the sources, what you did when they disagreed, and what you are claiming as a result. That last item is the one to get right, because it is where confirmation and completeness quietly diverge.

<ProTip title="📄 Write it up:" description="Name the type explicitly. Writing that you used data triangulation across three participant groups is checkable. Writing that you triangulated your findings is not, and examiners read the vague version as decoration" />

Use It, Then Claim Less Than You Want To

Triangulation earns its place when it is planned rather than added afterwards, when the second angle has genuinely different blind spots from the first, and when disagreement is reported instead of quietly smoothed away. It does not turn a qualitative study into a verified one, and the researchers who get the most out of it are usually the ones claiming the least from it.

<CTA title="Write Methods Sections You Can Defend" description="Structure your chapter, cite the methodologists you rely on, and check every claim against your design" buttonLabel="Try Jenni Free" link="https://app.jenni.ai/register" />

If you are choosing between types, the companion piece on data versus methodological triangulation walks through which axis to vary and when. And whichever you use, the underlying habit is the same one that governs any methodology section: say exactly what you did, in words a reader could use to repeat it, and make sure every source you lean on can carry the weight you put on it.