Topic
Action research
Moderate Moderate
Action research is the research logic underneath organizational development. Kurt Lewin proposed it in the 1940s: knowledge about a social system is produced by trying to change it, with the members of the system as co-researchers rather than as subjects.
Lewin's summary: "No action without research; no research without action."
Evidence: Moderate. The participation mechanism is well supported: people implement what they helped design, and participation in decision-making is meta-analytically associated with acceptance of change. The full cycle has rarely been tested against controls, because its logic makes controlled comparison awkward. Strong on mechanism, weak on outcome trials.
Use when: the problem is not well understood, the people affected have knowledge you lack, and the fix must be owned locally to survive.
Do not use when: the answer is already known and the work is implementation, or when the situation is an emergency requiring command decisions.
The cycle
- Diagnosing: identify the problem with the people who have it.
- Action planning: design a change, and state what you expect to happen.
- Taking action: implement it, at a scale small enough to learn from.
- Evaluating: collect data on what actually happened.
- Specifying learning: name what you now believe, and feed it back into the next cycle.
The cycle is a spiral, not a loop: each pass should leave the system more able to run the next pass without you.
Why the participation matters
Participation is not politeness. It is a method claim:
- People in the system hold data no outsider can obtain.
- People implement what they helped design and resist what they did not.
- The act of collecting and discussing data is itself an intervention; the system changes because it is being examined.
That last point is why "just do a survey first" is never neutral. Asking a question raises an expectation. See survey-feedback-method.
Variants
- Participatory action research: members of the system lead the research.
- Appreciative inquiry: an asset-based variant that studies what works rather than what is broken. See appreciative-inquiry.
- Improvement science / PDSA: the health care and manufacturing lineage of the same idea, with tighter cycles. See plan-do-study-act.
- Improvement kata: the Toyota lineage, with an explicit target condition. See improvement-kata.
Examples by audience
Health care: a ward team measures its own interruption rate during drug rounds, tries a "do not disturb" tabard for two weeks, measures again, and decides. Clinicians accept data they collected themselves and reject data handed down.
Software: a team forms a hypothesis that review latency, not build time, dominates lead time; instruments both; changes review assignment; and re-measures. See dora-metrics.
Executive: a leadership team treats its own meeting as the unit of study, records decisions made vs. revisited over a quarter, and redesigns the forum based on the count.
Limitations
- Slow relative to directive change.
- Vulnerable to the observer effect, which is also its mechanism.
- Weak external validity: what worked on one ward may not transfer.
- Requires a sponsor who tolerates learning that contradicts their prior view.