Research Methods

How to run a contribution analysis: build the contribution story, test it against evidence, and address rival explanations when no counterfactual exists.

What realist evaluation actually asks, how to build and test a CMO configuration, where the mechanism concept gets misused, and how realist quality is judged.

How process tracing establishes causation without a comparison case: the four evidentiary tests, what counts as diagnostic evidence, and where the method is misapplied.

Contribution tracing combines process tracing with explicit Bayesian updating to put a defensible confidence level on a contribution claim. How it works, and its limits.

How QCA finds the combinations of conditions that produce an outcome: calibration, truth tables, consistency and coverage, and the errors that discredit results.

Outcome mapping tracks behaviour change in the partners a programme works through, not the impact it cannot control. The three stages, progress markers, and real limits.

MSC collects stories of change and has stakeholders select the most significant ones. How the selection process creates the data, and why the technique is often misused.

A participatory group method for surfacing the secondary and unintended effects of a programme. How a REM session runs, how the map is coded, and what it cannot show.

Brinkerhoff’s method deliberately studies the best and worst cases rather than the average. How it works, what it can support, and the sampling objection it must answer.