Action Research: Method, Cycle, and When to Use It

Action research is a research approach in which researchers and practitioners collaborate to diagnose a real, local problem, implement a change intended to address it, and systematically study the effects of that change — then repeat the cycle based on what was learned. Unlike most research designs, which aim only to describe or explain a phenomenon, action research explicitly aims to produce practical improvement in a specific setting and generate transferable knowledge at the same time. It traces to psychologist Kurt Lewin’s work in the 1940s, who coined the term to describe research that proceeds through a spiral of planning, acting, observing, and reflecting.

This guide explains the action research cycle, its major traditions, how it differs from program evaluation and quality improvement work, what a rigorous action research write-up needs to contain, and its common critiques and limitations.

The Action Research Cycle

Action research is organized around an iterative cycle rather than a single linear procedure. The most widely cited formulation, derived from Lewin and elaborated by later methodologists, has four recurring stages:

1. Plan. The researcher(s) and the practitioners or stakeholders affected by the problem jointly diagnose the issue and design an intervention or change intended to address it, grounded in existing evidence and the group’s own knowledge of the local context.
2. Act. The planned change is implemented in the real setting — a classroom, a clinical unit, an organization, a community program — under normal operating conditions rather than an artificially controlled one.
3. Observe. Data are collected on the effects of the change using whatever mix of methods fits the question: field notes, interviews, surveys, routine performance or outcome records, and reflective journals are all common.
4. Reflect. The group interprets the data together, decides what has been learned, and revises the plan for the next cycle. Reflection is treated as a substantive analytic step, not an afterword.

A single action research project typically runs through several of these cycles, each one refining the intervention and the researchers’ understanding of it. The spiral, not the single pass, is what distinguishes action research methodologically from a one-off pre/post evaluation of an intervention.

Major Traditions Within Action Research

"Action research" is not one method but an umbrella covering several traditions that differ mainly in how much power-sharing and social-change intent they build in:

• Technical action research is the most instrumentalist form: an outside researcher or expert introduces a pre-specified intervention and measures whether it improves practice, with practitioners largely implementing rather than co-designing.
• Practical (or practical-deliberative) action research foregrounds practitioner reflection and professional judgment; the researcher works alongside practitioners as a facilitator, and the practitioners’ own interpretation of the problem shapes the intervention as much as external theory does.
• Emancipatory (or critical) action research explicitly aims at social transformation and challenging structural power imbalances, treating the research process itself as a vehicle for collective empowerment, not just problem-solving.
• Participatory action research (PAR) overlaps heavily with the emancipatory tradition and is often treated as a closely related sibling approach: it emphasizes that community members or practitioners are genuine co-researchers with power over the research agenda, not merely subjects or data sources, throughout every cycle.

Action research is used extensively in education (teacher inquiry into their own classroom practice), nursing and health services (practice-improvement studies conducted by clinicians), organizational development, and international development and community-based work, where the goal of durable local change is as important as the goal of publishable findings.

Action Research vs. Related Approaches

Action research is frequently confused with several adjacent activities that share surface features but differ in intent or rigor:

• Program evaluation typically assesses a program that is already fixed in design, asking whether it achieved its intended outcomes. Action research, by contrast, deliberately changes the intervention across cycles based on what is learned.
• Quality improvement (QI) — for example, the Plan-Do-Study-Act (PDSA) cycle common in healthcare — shares action research’s iterative structure and is sometimes classified as a close relative or applied form of it, but QI is usually oriented purely toward local operational improvement and may not be designed, from the outset, to produce generalizable knowledge or undergo the same methodological scrutiny as a formal research study.
• Case study research studies a bounded case in depth but does not require the researcher to intervene and change conditions within it; action research is inherently interventionist.
• Grounded theory is an analytic method for building theory inductively from qualitative data (see CASRAI’s Grounded Theory entry) and can be used as the analytic approach within an action research cycle’s reflection stage, but on its own it does not require an intervention or a change agenda the way action research does.

Because action research sits at the boundary between research and practice change, institutions sometimes need to determine whether a specific project requires IRB/ethics review as human subjects research or falls under a quality-improvement exemption — that determination depends on local regulatory guidance and the project’s specific intent (generalizable knowledge vs. purely local improvement), not on the "action research" label alone.

Data Collection and Analysis in Action Research

Action research is methodologically pluralistic — it does not prescribe a single data-collection technique. Common sources include:

• Field notes and reflective journals kept by the researcher-practitioners themselves
• Semi-structured interviews or focus groups with stakeholders
• Surveys or structured observation instruments administered before and after each cycle
• Routinely collected administrative, clinical, or performance data already generated by the setting

Analysis is often qualitative (thematic coding of field notes and interviews) but quantitative and mixed-methods designs are common too, particularly in health services and education research, where standardized outcome measures are tracked across cycles. What is distinctive is not the analytic technique itself but that analysis feeds directly back into the plan for the next cycle, and that the practitioners who will act on the findings are typically involved in interpreting them.

Writing Up an Action Research Study

A methodologically rigorous action research report or manuscript typically needs to document:

• The problem and context — who identified it, and why, in enough institutional and situational detail that a reader can judge transferability
• Who the co-researchers/stakeholders were and what role they played in each cycle (planning, acting, observing, reflecting) — this is central to establishing the study’s action research credentials rather than framing it as a conventional evaluation after the fact
• Each cycle — the plan, the change implemented, the data collected, and what was learned and changed as a result
• The researcher’s own positionality — since the researcher is typically an active participant rather than a detached observer, reflexivity about that dual role is expected, not optional
• Evidence of rigor — triangulation across data sources, member-checking findings with participants, and an audit trail of how the intervention evolved across cycles

Common Critiques and Limitations

Reviewers and methodologists most often raise the following concerns about action research, and a strong study anticipates them directly:

• Generalizability. Because action research is deliberately local and context-specific, findings from one setting may not transfer cleanly to another — authors should be explicit about what is and is not likely to transfer, rather than overclaiming.
• Researcher role conflict. The researcher is simultaneously an intervener and an evaluator of their own intervention, which creates a risk of biased interpretation that needs to be actively managed (through triangulation, external audit, or co-analysis with stakeholders) rather than assumed away.
• Rigor and publication bias. Because action research reports are often written primarily for a practice audience, they can under-document methodological detail relative to conventional empirical papers; journals and reviewers increasingly expect the same level of methodological transparency (sampling, data collection, analysis procedures) as any other qualitative or mixed-methods design.
• Ethical complexity. Because participants are also collaborators, informed consent and the boundaries of their role need careful, ongoing attention rather than a single up-front consent event.

Frequently Asked Questions

Is action research qualitative or quantitative?

Either, or both. Action research is defined by its cyclical, interventionist, practice-embedded structure, not by a specific data type — projects commonly use qualitative methods (interviews, field notes), quantitative outcome measures, or a mixed-methods combination of both across cycles.

What is the difference between action research and participatory action research (PAR)?

PAR is best understood as a tradition within action research that places particular emphasis on shared power and co-ownership of the research agenda with the community or practitioner group involved, rather than a researcher directing the process and inviting participation. Some methodologists treat the terms as near-synonyms; others reserve "action research" for the broader family and "PAR" specifically for its most participatory, power-sharing variant.

Does action research require IRB or ethics committee approval?

It depends on the project’s intent and the local regulatory framework. Action research aimed at generating generalizable knowledge intended for publication generally requires the same human-subjects ethics review as other research involving those participants; purely local quality-improvement work that is not intended to produce generalizable findings may fall under a different institutional pathway. This determination should be made with the institution’s IRB or research ethics office directly, not assumed from the project’s label.

Can action research be used for a doctoral dissertation?

Yes — action research is a well-established and accepted methodology in doctoral programs, particularly in education, nursing, and organizational studies (often within professional doctorate, or EdD, programs), provided the dissertation documents the cycles, data, and reflexivity rigorously enough to meet the program’s standards for a defensible original contribution to knowledge.

How many cycles does an action research project need?

There is no fixed number. Studies range from two or three cycles to considerably more; the appropriate number depends on when the group judges the intervention has stabilized and the learning objectives have been sufficiently addressed, which is itself a judgment made and justified within the write-up rather than set by convention.