Multi-source Non-participant Observation

Two sources is the practical minimum for triangulation, but three or more is recommended when resources allow, since two-source divergence cannot be resolved without a third reference point. The number should be justified by the research question — enough to capture the meaningful contextual variation relevant to the phenomenon, but not so many that fieldwork becomes unmanageable.

Yes, in most institutional contexts. Even in publicly accessible spaces, ethical review is required if the observation is systematic, if individuals could be identified from records, or if the data will be published. Many institutional review boards treat observational research in semi-public or institutional settings (workplaces, schools, hospitals) as requiring full consent procedures regardless of participation level.

The methods are identical at each source; the difference is strategic. Single-source designs produce rich depth at one site but cannot distinguish site-specific patterns from general ones. Adding a second or third source allows the researcher to test whether findings replicate and to map contextual variation — trading some depth at any given site for broader evidential scope.

Yes, and doing so is common in practice. Combining multi-source non-participant observation with interviews or surveys creates a mixed-methods design where observation captures behaviour and interviews capture meaning. The combined design should be planned explicitly, with a clear rationale for how the data streams will be integrated at the analysis stage.

Divergence is informative, not a failure. It signals that the phenomenon varies by context — a substantively important finding in itself. The researcher should investigate whether divergence is explained by site characteristics, observer variation, timing, or protocol inconsistency. Unresolved divergence should be reported transparently rather than suppressed in favour of a false convergence narrative.