Triangulation
The deliberate comparison of evidence from different methods, data sources, researchers, or theories to strengthen and complicate conclusions.
Download Markdown (.md)Soroush’s Product Wiki · https://wiki.soroush.uk/topic/triangulation/
01What it means
Triangulation examines a research question from more than one angle. A team may combine interviews with behavioral data, compare different participant groups or settings, involve multiple analysts, or interpret findings through different theoretical lenses. The aim is not to make every source agree; convergence, contradiction, and silence each reveal something about the claim and its boundaries.[1][2]
- Methods
- Interview, observation, behavioral or quantitative data
- Sources
- Participant groups, records, settings
- Analysts
- Different interpreters and reflexive comparison
- Theoretical lenses
- Alternative explanatory frames
- Convergence
- Where evidence supports a shared claim
- Contradiction + silence
- Boundaries, missing evidence, next research
Analyze each evidence stream on its own terms before combining it; variety without a reason is not triangulation.
02Why it matters
Every method has characteristic blind spots. Interviews capture meaning but not exact frequency; analytics show behavior but often not intent; fieldwork shows context but only in selected settings. Triangulation makes evidence more robust by testing whether a conclusion survives different perspectives and by exposing where context or measurement changes the story.
03When to use it
Use triangulation for consequential decisions, contested findings, complex services, and research programs with several available evidence streams. It is especially useful when one method produces a surprising result or when stakeholders are tempted to treat a single metric, quote, or study as definitive.
04How it works
- Write the claim or research question precisely enough that different evidence can address it.
- Choose complementary methods, sources, settings, analysts, or theories for clear reasons—not merely variety.
- Analyze each evidence stream on its own terms before combining results.
- Create a convergence matrix that records agreement, contradiction, missing evidence, and differences in scope.
- Revise the conclusion, confidence, and next research step based on the complete pattern.
05A practical example
A checkout team sees high mobile abandonment in analytics. Interviews suggest unexpected fees, while usability sessions reveal an address-validation loop on older devices. Support logs contain both complaints. The team avoids one grand explanation: it fixes the technical loop immediately, tests earlier fee disclosure, and tracks each mechanism with separate measures.
06Where it breaks
More data does not automatically mean better evidence. Methods can share the same sampling bias, weak construct, or flawed assumption, creating false confidence. Contradictions may reflect different populations or time periods rather than error. Combining incompatible measures carelessly can flatten nuance, and triangulation adds cost; the choice of sources needs a defensible purpose.
Common mistakes
- Counting several metrics from the same dataset as independent confirmation.
- Forcing contradictory findings into one neat story instead of investigating scope and context.
- Collecting many methods without defining the claim each one can support or challenge.
07Sources
- The Use of Triangulation in Qualitative ResearchPubMed · Peer-reviewed article recordThe Use of Triangulation in Qualitative Research
- Design Decisions in ResearchNIH Office of Behavioral and Social Sciences Research · Government research guidanceDesign Decisions in Research