# Action Research: Method and a Two-Cycle Example

Learn action research through a two-cycle training example: act, observe, revise, and keep conclusions local to the evidence.

- Canonical URL: https://www.harmate.com/en/blog/action-research-method-and-a-two-cycle-example
- Author: Harmate Team
- Published: 2026-10-04
- Updated: 2026-10-04T00:20:22.498105+00:00
- Language: en

## Content

# Action Research: Method and a Two-Cycle Example

Action research is an inquiry carried out alongside deliberate action in a real setting: people affected by the practice observe what happens, test their first explanation, then adjust the action and the question for another cycle. In adult training, it can help practitioners document why a learning arrangement does—or does not—address a situated difficulty. On its own, it does not turn a local improvement into general causal proof.

Consider this fictional two-cycle example: a team of trainers supports technicians who hand work over between day and night shifts. The first trial changes the team's interpretation of the problem; the second examines a more contextual response.

## What is action research?

Action research joins a question about practice with deliberate changes to that practice. It proceeds through loops: plan an action, put it into practice, observe what happened and where it differed from expectations, reflect with the people involved, and decide what to do next. The cycle is not a universal four-box recipe; its stages inform one another and can overlap.

An inquiry is participatory when people affected by it genuinely help shape the investigation and its interpretation—not merely when they complete a survey. [INRAE describes several levels of involvement in participatory research](https://www.inrae.fr/sciences-recherches-participatives-inrae), from collecting data to co-constructing a question, analysis, and dissemination. That definition concerns participatory research; it does not automatically certify every workplace improvement project as scientific research.

A study at a rural hospital in South Africa illustrates participatory action research applied to workplace learning: a group of students and educators completed two cycles, documented their discussions, then designed and reflected on an intervention tailored to the site. This is one situated case, not proof that the same arrangement will produce the same result elsewhere ([study in *BMC Medical Education*](https://link.springer.com/article/10.1186/s12909-016-0607-3)).

## How it differs from a project review or delayed evaluation

Action research is not simply a lessons-learned review after a project ends: observations in one cycle inform the next. Nor is it the same as a delayed assessment of workplace application after training. A [project lessons-learned review](/en/blog/project-lessons-learned-method-questions-and-an-adaptable-template) looks back at a completed project; [post-training evaluation](/en/blog/post-training-evaluation-questions-and-method-for-assessing-workplace-application) examines later opportunities and conditions for applying learning. Either can inform an action-research cycle, but neither replaces its iterative inquiry and revision.

“Action” also does not mean an intervention the team has already decided to defend. If it only looks for evidence that confirms a plan, it loses the chance to discover that the problem was framed incorrectly. Contradiction is working evidence: it may lead the team to adjust training, revise its explanation, or acknowledge that the current data cannot settle the question.

## A two-cycle example: handover instructions

A working group of the trainers and volunteer technicians from both shifts notices that requests for clarification recur after some handovers. Before creating a new form, it separates what it knows from what it assumes. The people concerned confirm the scope; the group uses only authorized work situations and removes details that could identify a colleague or service user.

| Step | Cycle 1: test the first explanation | Cycle 2: revise the interpretation |
|---|---|---|
| Practical question | What is missing from notes that would help the next team resume an intervention? | In which situations must the decision, its context, and what remains to be checked be made explicit? |
| Starting hypothesis | Teams forget important fields; a short template may help. | Fields are filled, but a detail may remain ambiguous depending on role and urgency. |
| Action | Trainers have teams test a handover template in practice scenarios, then—with the teams' agreement—in a small work setting. | The adjusted template separates “observed fact,” “decision made,” “next check,” and “useful context.” Teams discuss fictional examples and authorized cases. |
| Evidence reviewed | Anonymized handover notes and the questions the next team asked afterwards. | Notes written with the adjusted template and team comments on the examples discussed. |
| Contradictory observation | Several notes include the expected fields, yet recipients still ask questions. Forgetting is not the only explanation. | Some comments show the “decision” field is hard to complete when a technician is not authorized to decide alone. |
| Revision | The group drops forgotten fields as the sole explanation and also examines handover conditions, roles, and available time. | It keeps the template but clarifies who can decide, who should be consulted, and when to escalate uncertainty. The team plans to check whether the issue appears in other intervention types. |

This example does not prove that the revised template reduces errors. It shows how an observation that contradicts the initial explanation changes what the team examines in the next cycle. In a real inquiry, specify the sources used, situations not represented, disagreements, and other changes that occurred at the same time.

## Keep an evidence log that shows revisions

A lightweight table can prevent the final account from turning a hypothesis into an established fact. For each cycle, record:

| Record | Question to ask |
|---|---|
| Situation and participants | What precise context is included, and who could not contribute? |
| Inquiry question | What does the team want to understand, apart from what it hopes to improve? |
| Action and date | What changed, with whom, and under what conditions? |
| Expected trace | What could show a difference without selecting only favorable evidence? |
| Observation | What was seen or heard, and in what context? |
| Counterexample | Which situation does not fit the dominant explanation? |
| Interpretation and revision | What changes in understanding or the next action? |
| Limit | What remains unknown, uncollected, or impossible to compare? |

Separate observation from interpretation: “the note does not say who approves the change” is checkable; “the author cannot make decisions” is an attribution that would require other evidence and may be unfair. Open responses can surface participants' own words, context, exceptions, and unexpected explanations. A closed measure can track a defined option over time. They answer different questions.

If you begin with a questionnaire, shape it around a decision the team can actually make. The [training placement questionnaire guide](/en/blog/training-placement-questionnaire-how-to-keep-it-short-without-losing-quality) explains why a recent situation, context, and desired outcome are more informative than quickly converting an answer into a level.

## Keep the research question separate from the improvement target

An improvement target describes the change participants want to try: “make handovers clear enough for the next shift to pick up the work.” A research question specifies what they will examine: “in what circumstances does a note let the next team distinguish a fact, a decision, and an uncertainty?”

This distinction helps even in a modest project. It makes three things visible:

- the practical change sought, chosen with the people affected;
- the information needed to understand the situation;
- the criteria for reviewing the interpretation, including when observations contradict expectations.

The question may change if observations show that the original problem was misframed. Record what changed and why rather than rewriting the initial objective as if it had always been the same. If a management decision, safety duty, or immediate risk is involved, use the appropriate process; action research should not delay a protective measure.

## Set a scope and a stopping rule

Before the first cycle, agree with participants on a realistic scope, a timeline, and a review rule. For example, review on the agreed date—or sooner if an observation reveals a risk, unauthorized data, or an unintended consequence. At each review, ask whether another action can still inform the question, whether the necessary people can participate, and whether the next observation could change the decision.

A small inquiry may stop when participants have enough evidence to choose what to do next in this context and the remaining limits are explicit. It may instead conclude that a perspective is missing, comparison is not possible, or a specialist should examine the issue. Do not impose a universal number of cycles: scope the work to the question, participants, and risks.

Do not use the word “research” to make an internal preference sound like a scientific finding. State who participated, what changed, what evidence was reviewed, which cases contradict the interpretation, and what remains uncertain. A change after an action does not establish that the action caused it: other events, team changes, or work conditions may have contributed. A local conclusion can guide a local next step without being generalized to every training provider or learner group.

## Where a response tool can help

When several people need to describe their situations, a tool can support response collection or import, surface themes with the verbatims that support them, and prepare a report for review. The trainer and participants remain responsible for interpreting those traces in context, examining divergent responses, and choosing what to try in the next cycle. A theme is not an individual diagnosis, an automatic recommendation, or proof of a causal effect.

Action research is useful when learning from practice requires being able to change one's mind. Its most concrete sign is not a spiral diagram: it is an honest record of what was tried, what contradicted the first idea, and how the people involved chose what came next.