· By Thesisroom
Triangulation in Research: Four Types, Steps and a Thesis Example
The four types of triangulation, what convergence can and cannot show, eight planning steps, what to do when sources disagree, and a printable convergence matrix.
This guide has a printable version: open the printable triangulation plan and convergence matrix
Triangulation in research means studying one question through more than one data source, method, investigator or theory, so each can be read against the others. Carter and colleagues (2014), citing Patton (1999), describe it as a way to build a comprehensive understanding. They note it has also been viewed as a way to test validity through convergence.
That second idea is where theses get into trouble, because the sources below disagree on whether agreement between your data proves anything. This guide covers the four types, what triangulation can show, how to plan it, what to do when sources disagree, and how to write it up.
What are the four types of triangulation?
The four types are data, investigator, theory and methodological triangulation. Carter and colleagues trace them to Denzin (1978) and Patton (1999), under the labels data source, investigator, theory and method triangulation. The descriptions below follow Arias Valencia (2022), who reports Denzin’s classification and later operational definitions in detail.
| Type | What it compares | Thesis example | How to report it |
|---|---|---|---|
| Data | The same question across people, times or places | Teachers, mentors and principals asked about one mentoring scheme | Name each source and the sub-question it answers |
| Investigator | Two or more researchers’ coding or interpretation | A second person codes a sample of transcripts independently | Say who coded what, when, and how differences were settled |
| Theory | Two or more theories applied to the same data | Two rival explanations of teacher retention set out before analysis | Introduce both theories and say which data would favor each |
| Methodological | Two or more methods aimed at the same question | Interviews plus observation of meetings plus policy documents | Say whether you mean methods or designs, and within or between methods |
You can combine types. Arias Valencia reports this as multiple triangulation, and Johnson and colleagues (2017) give a full example. Across three sites, they combined document review, interviews, observations, digital diaries and focus groups, with investigators representing clinical, research and service user perspectives.
Data triangulation
Data triangulation uses more than one source of data on the same question. Arias Valencia reports that Denzin split it into three subtypes: time, space and person. In a thesis, that might mean interviewing teachers, mentors and principals about the same scheme, or collecting the same data at two sites.
The operational definitions by Kimchi and colleagues, as Arias Valencia reports them, set limits. A longitudinal study is not temporal triangulation, because it documents change rather than checking the same phenomenon at different points. Data from several sites counts as spatial triangulation only when you compare across the sites.
Investigator triangulation
Investigator triangulation means more than one researcher collects or analyzes the data. In an open textbook chapter, Tsindos (2023) notes it often shows up in a positionality or reflexivity statement. Tsindos also notes that some forms of thematic analysis require more than one coder, while reflexive thematic analysis does not.
Arias Valencia adds a reporting warning: investigator triangulation is hard to recognize unless authors describe explicitly how they achieved it. So “the data were triangulated by the research team” tells an examiner nothing. Say who coded what, independently or together, and how you settled differences.
Theory triangulation
Theory triangulation uses more than one theory to interpret the same data. Arias Valencia reports Denzin’s definition as testing rival theories or hypotheses, and Arias Valencia writes that few researchers manage to reach it. Tsindos suggests writing it up like a literature review: describe each theory, then compare them before you interpret the data.
The practical test is simple. If you did not set two theories against each other before analysis, do not claim theory triangulation. Using one framework and citing a second in passing is not the same thing.
Methodological triangulation
Methodological triangulation uses two or more methods to study the same question. Arias Valencia reports Denzin’s split between triangulation within methods (for example, observation and open interviews, or two quantitative measures of one phenomenon) and between methods (dissimilar methods aimed at the same phenomenon). Arias Valencia also reports a blunt rule: five variations on the same method do not make five triangulations.
The label itself is slippery. A scoping review by Schlunegger and colleagues (2024) notes that methodological triangulation can refer to either data-collection methods or research designs. State which one you mean the first time you use the term.
Does triangulation in research prove your findings are valid?
Not on its own, and the sources disagree about whether it is a validity check at all. Your thesis should say which purpose you claim, and then stay consistent with that claim.
On one side, triangulation is treated as a validity strategy. Carter and colleagues say it has been viewed as testing validity through convergence. Tonkin-Crine and colleagues (2016) write that a triangulation protocol can enhance the validity of findings. Arias Valencia quotes Denzin’s view that its greatest goal is to control researcher bias and so increase validity.
On the other side, triangulation is treated as a route to completeness. Arias Valencia states that methodological triangulation is not a concurrent validation technique, and that the purpose of simultaneous triangulation is different but complementary data rather than replicated results. Tsindos writes that the purpose is not to confirm results but to understand differences, though the same chapter says triangulation can be used to demonstrate rigor, validity and credibility.
For a thesis, this leaves two honest options:
- Convergence: you claim that agreement across sources strengthens confidence in a finding. Then you must also report where sources did not agree.
- Completeness: you claim that each source shows a part of the picture the others cannot. Then you must show what each source added.
Avoid “triangulation validated the findings” in either case. Johnson and colleagues use the more defensible phrase that triangulation “enhanced the credibility of findings”. In the trustworthiness table of our guide to defending qualitative research in a viva, data and researcher triangulation sit under credibility, with viva scripts for the trust question.
How do you plan triangulation for a thesis?
Plan it before you collect data. Tonkin-Crine and colleagues ran their triangulation after the fact, and wrote that their study would have benefited from a prospective design, so that interview guides and questionnaires complemented each other.
- Start from the research question. Tonkin-Crine and colleagues write that the types you use should reflect the individual research question. If your questions are still loose, tighten them before you choose any type.
- Choose the purpose. Write one sentence: convergence, completeness, or both. Everything else follows from it.
- Pick the types that serve that purpose, and name the ones you leave out. Tonkin-Crine and colleagues left out theory triangulation, since all their research took a realist perspective.
- Map each source to a sub-question. A source that answers nothing the others address cannot converge with them or complete them.
- Analyze each data set on its own first. In the protocol Tonkin-Crine and colleagues followed, data are integrated only after every data set has been analyzed individually.
- Build a convergence coding matrix. List your key findings, compare each pair of data sets, and mark each comparison as agreement, partial agreement, dissonance or silence.
- Compare independently if you can. In that study, three researchers built the comparison separately, then discussed their versions to reach consensus.
- Keep a decision record. Note every judgment you make about a comparison, because the methodology chapter will need it.
What should you do when your sources disagree?
Treat the disagreement as a finding to explain, not an error to hide. Tonkin-Crine and colleagues write that dissonance does not indicate a failure in the study, and can be constructive if it leads to new findings or a richer understanding.
Their own study shows how. They found three instances of dissonance in 39 independent findings. Two of them set patient questionnaires against patient interviews on satisfaction with the consultation. The authors reasoned that the interviews may have captured mainly satisfied patients, and that the larger questionnaire group was likely more representative.
Tsindos is candid that inconsistent results can be hard to explain in a write-up. A fixed routine helps:
- Check it is a real disagreement. Were both sources asked about the same thing, at the same time, by the same definition?
- Ask whose voices each source holds. Who took part, who chose to, and who was missing?
- Ask whether role explains it. Some gaps are not disagreements at all. In the Tonkin-Crine study, silences between patient and doctor data often arose because data from one group did not apply to the other.
- Ask whether the method shapes it. What people report in an interview and what an observer records can differ for reasons that are themselves worth reporting.
- Go back to the raw data. Reread the passages behind each side before you decide anything.
- Report it either way. An explained disagreement goes in the discussion. One you cannot explain goes in the discussion and in the limitations section of your study.
Never drop the source that disagrees, and never average two accounts into one that neither source supports.
What does triangulation look like in a thesis? A worked example (invented)
This example is invented for illustration. A Master’s candidate in education asks: “How do first-year teachers experience the mentoring scheme in two rural schools?” The purpose is completeness, with convergence noted where it occurs.
The plan:
- Data triangulation: eight first-year teachers, four mentors and two principals.
- Methodological triangulation (between methods): interviews, observation of six mentoring meetings, and each school’s written mentoring policy.
- Investigator triangulation: the supervisor codes two teacher transcripts independently, and the two sets of codes are compared.
- Theory triangulation: not used, because the study is exploratory and does not test rival theories.
After analyzing each data set separately, the candidate builds this matrix excerpt:
| Key finding | Teachers vs mentors | Interviews vs observations | Interviews vs policy |
|---|---|---|---|
| Meetings are mostly about paperwork | Agreement | Agreement | Silence (policy does not describe content) |
| Mentors give feedback on teaching every week | Dissonance (mentors say weekly, teachers say rarely) | Partial agreement (meetings observed every two to three weeks) | Silence |
| Advice from colleagues in passing matters more than formal meetings | Silence (mentors never raised it) | Not applicable | Silence |
The dissonance in row two is the interesting result. Rereading the mentor transcripts, the candidate finds that mentors counted short hallway conversations as feedback, while teachers counted only planned meetings. The two groups were not contradicting each other. They were using different definitions of feedback, and that difference becomes a finding in its own right.
How do you write up triangulation in the methodology chapter?
State the purpose, each type used, what it compared, and how the comparison was done, in a short paragraph. Put the full matrix in an appendix and point to it.
In a thesis that follows the usual chapter-by-chapter structure, triangulation shows up twice: the methodology chapter says what you compared and how, and the discussion reports what the comparison showed.
In the case studies Schlunegger and colleagues reviewed, qualitative and quantitative data were reported separately, with little connection between them, so neither the comparisons nor the contradictions could be clearly determined. They recommend existing tools, such as a triangulation protocol or a mixed-methods matrix, for more transparent reporting. Your paragraph should name:
- the purpose (convergence, completeness, or both);
- each type, with what it compared;
- any type you did not use, and why;
- when the comparison happened (after separate analyses);
- who compared the data, and how differences between analysts were settled;
- where the matrix is.
For the invented mentoring study, it might read:
To build a fuller account than any single source could give, I used data, methodological and investigator triangulation. Teachers, mentors and principals were interviewed, six mentoring meetings were observed, and both schools’ mentoring policies were analyzed. Each data set was analyzed separately before key findings were compared pairwise and coded as agreement, partial agreement, dissonance or silence (Appendix D). My supervisor coded two transcripts independently, and we resolved differences by discussion; theory triangulation was not used, as the study did not test rival theories.
For where this paragraph sits among the chapter’s other justifications, see how to write a methodology chapter that survives a viva.
How do you write up triangulation in the discussion chapter?
Report each triangulated finding by what the comparison showed. Agreement is stronger support, partial agreement shows complementary angles, dissonance needs your explanation, and silence marks where only one source could speak.
For the invented study:
Mentors and first-year teachers described the frequency of feedback differently. Mentors reported weekly feedback, while teachers described it as rare, and observed meetings took place every two to three weeks. Mentor transcripts suggest the gap reflects different definitions: mentors counted brief hallway exchanges as feedback, while teachers counted only scheduled meetings. This suggests the scheme’s written policy, which does not define feedback, may leave each group with its own measure of whether mentoring is happening.
The paragraph names the disagreement, gives the evidence on each side, offers an explanation grounded in the data, and draws an implication. Our guide to writing a discussion chapter works through that paragraph pattern step by step.
Once both chapters are drafted, check that “triangulation” means the same thing in each; Thesisroom’s Argument Map finds contradictions between chapters and drifting definitions.
What mistakes weaken triangulation in a thesis?
Beyond the labeling limits in the types above (variations of one method, theory in name only, sites never compared), three patterns stand out:
- Parallel reporting: each data set reported separately, with little connection between them, the pattern Schlunegger and colleagues describe.
- “Validated” with no dissonance reported: if everything agreed, show the matrix that proves you checked.
- Triangulating by habit: Tsindos notes that when the purpose is to develop a new theory, more than one method, data point or theoretical foundation may not be needed.
Printable triangulation plan and convergence matrix
Fill this in before data collection, then complete the matrix after you have analyzed each data set separately.
- My research question is written as one sentence: ______
- My purpose is: convergence / completeness / both
- Data triangulation: sources (people, times, places) and the sub-question each answers: ______
- Investigator triangulation: who codes what, independently or together: ______
- Theory triangulation: the rival theories, and what data would favor each: ______
- Methodological triangulation: methods, and whether within or between methods: ______
- Types I will not use, and why: ______
- Each data set analyzed separately before comparison
- Matrix placed in an appendix and referred to in the methodology chapter
Mark each comparison as agreement, partial agreement, dissonance or silence (categories as used by Tonkin-Crine and colleagues).
| Key finding | Source pair compared | Category | Evidence (page or transcript line) | What I will do with it |
|---|---|---|---|---|
Frequently asked questions
What is triangulation in research, in simple terms?
It means looking at one research question from more than one angle: different data sources, methods, researchers or theories. You then compare what each angle shows. Agreement, partial agreement and disagreement between them are all results you report.What are the four types of triangulation?
Data triangulation (several sources, times or places), investigator triangulation (more than one researcher), theory triangulation (rival theories applied to the same data) and methodological triangulation (more than one method). Carter and colleagues trace the four to Denzin and Patton.Does triangulation make my findings valid?
The sources disagree. Some treat convergence across sources as a validity check; others, such as Arias Valencia, say methodological triangulation is not a concurrent validation technique. Say which purpose you claim, and avoid writing that triangulation validated your findings.Is triangulation only used in qualitative research?
No. Arias Valencia reports within-method triangulation using two quantitative measures of the same phenomenon, and Tonkin-Crine and colleagues triangulated questionnaire and interview data from a trial.What if my data sources contradict each other?
Report the contradiction as a finding. Check that both sources addressed the same thing, ask whose voices each holds, return to the raw data, and explain the gap in your discussion. If you cannot explain it, also name it as a limitation.How many types of triangulation does a thesis need?
None of these sources sets a number. Tonkin-Crine and colleagues say the types should reflect your research question, and they used three of four. Tsindos notes that combining types is common in mixed methods research. Name the ones you use and give a reason for each one you leave out.Sources
- The use of triangulation in qualitative research (Carter et al., 2014, Oncology Nursing Forum)
- Principles, Scope, and Limitations of the Methodological Triangulation (Arias Valencia, 2022, Investigación y Educación en Enfermería)
- Chapter 28: Triangulation (Tsindos, in Qualitative Research: a practical guide for health and social care researchers and practitioners, 2023)
- Methodologic and Data-Analysis Triangulation in Case Studies: A Scoping Review (Schlunegger et al., 2024, Western Journal of Nursing Research)
- Multiple triangulation and collaborative research using qualitative methods to explore decision making in pre-hospital emergency care (Johnson et al., 2017, BMC Medical Research Methodology)
- Discrepancies between qualitative and quantitative evaluation of randomised controlled trial results: achieving clarity through mixed methods triangulation (Tonkin-Crine et al., 2016, Implementation Science)
How this guide was made: Thesisroom drafted this guide with the help of AI writing tools. Every fact, figure and source in this post was checked against its original source before publishing.