A topic in the Open Knowledge Graph — a free, open map of 15,290 topics and the order to learn them in.
Triangulation uses multiple methods, data sources, or investigators to examine questions from different angles. Convergent findings increase confidence; divergence prompts investigation. Divergence often proves more informative than convergence.
From your mixed-methods work, you know that each individual method carries its own vulnerabilities — surveys may suffer from social desirability bias, interviews from recall error, administrative records from bureaucratic categorization artifacts. Triangulation is the strategy of using multiple methods so that the weaknesses of one are compensated by the strengths of another. The underlying logic borrows from measurement theory: if two independent measures of the same construct reach the same conclusion, the probability that both are wrong in the same way is lower than the probability that either is wrong alone. This is convergent validity — confidence in a finding grows when methods with different error structures agree.
There are four distinct forms triangulation can take. Data triangulation uses multiple data sources — interviewing different stakeholders, sampling from different time periods, or drawing on records from different agencies. Investigator triangulation uses multiple researchers to code, analyze, or interpret the same material, testing whether findings are specific to one analyst's perspective. Theory triangulation applies multiple theoretical frameworks to the same data, asking what each lens makes visible and obscures. Methodological triangulation — the most common — combines qualitatively different methods such as a survey and an ethnography or a regression and a case study. Each type increases confidence for different reasons.
The methodologically surprising insight, however, is that divergence is not failure. When a survey shows that 60% of employees report high job satisfaction while ethnographic fieldwork reveals chronic workplace frustration, the divergence is a finding, not a problem. It prompts deeper questions: perhaps the survey elicits socially acceptable answers that respondents feel unsafe revising in an anonymous questionnaire; perhaps the ethnographic sample over-represents a particular department; perhaps satisfaction and frustration genuinely coexist for different aspects of work life. Divergence forces the researcher to interrogate the assumptions each method smuggles in, and those interrogations often produce more sophisticated theoretical accounts than either method alone would have generated.
The practical discipline is to plan triangulation prospectively, not retrospectively. Retroactively assembling convergent evidence after seeing results is a form of confirmation bias — you select the sources that agree and quietly set aside the ones that diverge. Good triangulation specifies in advance what counts as convergence, what counts as divergence, and what inference each pattern will support. It also requires honest reporting: when your survey and your interviews disagree, you must report both and theorize the disagreement rather than privileging one source because it aligns with your hypothesis. This connects triangulation to the broader norms of mixed-methods design: methodological pluralism is a commitment to letting reality surprise you.
Topics in reflective domains aren't scored by quiz answers. Read, reflect, and mark when you've thought it through.