Triangulation is the deliberate use of more than one method, data source, investigator or theoretical perspective in the study of a single research problem, so that the conclusion does not rest on the peculiarities of any one of them. The metaphor is borrowed from surveying and navigation, where a position is fixed by taking bearings from two known points rather than trusting a single sighting. Its sociological premise is that every technique carries a method-specific bias: a questionnaire captures what respondents are willing to state, an interview what they can articulate to a stranger, observation what happens while the observer is present, and a document what an official thought worth recording. If findings produced by instruments with different biases nevertheless agree, the agreement is unlikely to be an artefact of the instrument.
The idea was formalised by Donald Campbell and Donald Fiske as convergent validation through a multitrait-multimethod matrix, extended by Webb and colleagues in their argument for unobtrusive measures, and named and classified for sociology by Norman Denzin. It belongs to the family of ideas around validity rather than reliability: repeating the same flawed instrument yields consistency, not truth, and only a differently biased instrument can expose the flaw.
Denzin's four types
Data triangulation varies the sources while holding the method constant — studying the same phenomenon at different times, in different settings, and among different categories of person. Investigator triangulation puts more than one researcher into the field, so that the personal equation of a single fieldworker does not silently shape the record; discrepancies between fieldnotes become evidence rather than embarrassment. Methodological triangulation is the type most often meant, and Denzin divided it into within-method triangulation, which combines several techniques of the same family, such as multiple scales inside one questionnaire, and between-method triangulation, which combines survey, observation and documentary analysis. Theoretical triangulation approaches the same data with rival frameworks — reading agrarian change through both a structural-functional and a Marxian lens — to see which residues each leaves unexplained.
The two logics
Discussions of triangulation slide between two distinct justifications, and it is worth keeping them apart.
The convergent logic treats triangulation as a test. Two methods with independent weaknesses point at the same object; agreement raises confidence because the probability that both errors run in the same direction is low. This is the logic of validation, and it presumes that the methods are measuring the same thing.
The completeness or complementarity logic treats triangulation as elaboration. Methods do not measure the same thing at all: a survey establishes distribution and magnitude across a population, ethnography establishes meaning, process and mechanism within a setting. Combining them produces a fuller account rather than a stronger proof. On this reading, divergence is not failure but information — each method illuminates a different facet, and the researcher's task is to explain why the facets look different.
Mixed-methods designs
Contemporary practice organises these logics into recognisable designs. Sequential designs order the methods: exploratory sequencing begins with qualitative fieldwork to discover the categories and vocabulary a valid questionnaire will need, while explanatory sequencing begins with a survey and follows the anomalous cases with interviews. Concurrent designs run both at once and compare results at the analysis stage. Embedded designs nest a small qualitative component inside a large survey or trial to interpret its outcome. The choice is not merely technical: sequencing decides which method sets the research questions and which is reduced to servicing them.
When the methods disagree
The difficult case, and the one candidates usually avoid, is contradiction. Suppose a household survey reports that a large proportion of women take decisions about their own healthcare, while a year of village observation shows those decisions being taken by mothers-in-law. There are at least four possible readings: one instrument is invalid; the two are measuring different constructs, namely stated norm against enacted practice; the samples or settings are not comparable; or the phenomenon genuinely varies by context and both records are locally correct. Triangulation cannot itself adjudicate between these. It supplies the discrepancy and obliges the researcher to reason about it — which is a substantial gain, but not the arbitration that the surveying metaphor promises.
The critique
The strongest objection, pressed by Norman Blaikie among others, is epistemological. The convergent version of triangulation presupposes a single knowable reality against which several bearings can be taken. Interpretive and constructionist positions deny this: if the survey and the ethnography constitute their objects differently, then there is no common target for the bearings to converge upon, and agreement between them would be as puzzling as disagreement. A second objection is practical — triangulation is expensive in time, money and skill, and often delivers a shallow second method bolted onto a serious first one, gaining rhetorical legitimacy rather than analytical purchase. A third is that combining methods can multiply errors instead of cancelling them, since a weak study added to a strong one does not average out. The defensible position is the completeness logic, held without the promise that convergence proves anything.
An Indian illustration
Indian sociology has long practised triangulation without the label. Village studies have been most persuasive where enumeration and ethnography were combined: a census of households, holdings and caste composition establishing the structural facts, and prolonged residence establishing how dominance, patronage and ritual rank actually operate — the combination M. N. Srinivas and his colleagues reflected on in The Fieldworker and the Field. The long-term study of Palanpur, revisited by successive investigators over decades, joined panel survey data on landholding and income to resident observation of tenancy and caste, and several of its central findings emerged precisely from the gap between what the schedules recorded and what the fieldworkers saw. In the same spirit, national survey estimates of women's autonomy, dowry or land ownership are best read alongside localised fieldwork, because the survey captures what is reportable and the fieldwork what is customary — and in India the distance between the two is itself the sociological finding.
For the UPSC answer
Define triangulation as multiple methods, sources, investigators or theories brought to one problem, and attribute the fourfold classification to Denzin while noting Campbell and Fiske's earlier convergent validation. The mark-earning move is to separate the two logics — validation through convergence, and completeness through complementarity — and to say plainly that triangulation strengthens validity without guaranteeing truth. Deal honestly with disagreement between methods, and register the constructionist critique that convergence presumes one knowable reality. For the Indian illustration, pair large-sample survey data with village ethnography and point out that the divergence between the reportable and the customary is itself data.
References & further reading
- Campbell, D. T. and Fiske, D. W. (1959). Convergent and Discriminant Validation by the Multitrait-Multimethod Matrix. Psychological Bulletin, 56(2).
- Webb, E. J., Campbell, D. T., Schwartz, R. D. and Sechrest, L. (1966). Unobtrusive Measures. Chicago: Rand McNally.
- Denzin, N. K. (1970). The Research Act: A Theoretical Introduction to Sociological Methods. Chicago: Aldine.
- Jick, T. D. (1979). Mixing Qualitative and Quantitative Methods: Triangulation in Action. Administrative Science Quarterly, 24(4).
- Blaikie, N. W. H. (1991). A Critique of the Use of Triangulation in Social Research. Quality and Quantity, 25(2).
- Srinivas, M. N., Shah, A. M. and Ramaswamy, E. A. (eds.) (1979). The Fieldworker and the Field. Delhi: Oxford University Press.