Participatory Action Research: The Cycle, Power-Sharing, and Co-Analysis

Participatory action research (PAR) is a specific tradition within action research: one where the people affected by the problem under study are treated as co-researchers with real authority over the research agenda, not as participants a researcher invites to comment on a plan already set. That authority is what separates PAR from the wider family of action research and from qualitative studies that simply gather community input. PAR asks a further, harder question than “did we consult the community” — it asks who set the research question, who controls the method, who interprets the data, and who is named as an author when it’s published.

This guide focuses on the mechanics that actually make a project participatory rather than merely community-informed: how the plan-act-observe-reflect cycle runs in PAR specifically, what power-sharing looks like in practice at each stage, and how co-analysis of data with community co-researchers differs from the member-checking and advisory-board consultation that many “participatory” studies substitute for it.

Where PAR comes from

PAR’s methodological roots trace to Kurt Lewin’s 1940s formulation of action research as a “spiral of steps,” each composed of planning, action, and fact-finding about the results of that action — research conducted to solve a real problem, in the setting where it occurs, with the people affected by it. Lewin’s spiral gave action research its iterative shape, but not yet its participatory commitment.

That commitment came from a distinct, largely Latin American tradition associated with the sociologist Orlando Fals Borda, who helped organize the first World Symposium on Action Research in Cartagena, Colombia, in 1977, and who argued that researchers should treat the knowledge of grassroots communities as a full partner in the research process rather than raw material for expert analysis. Paulo Freire’s work on dialogical education and conscientização — the process by which people become critically aware of the social conditions shaping their lives, as a precursor to acting on them — supplied PAR’s emancipatory framing: research is not just a way of generating knowledge, it’s a vehicle through which the people studied build the capacity to change their own circumstances. Education researchers Stephen Kemmis and Robin McTaggart later gave the Lewinian spiral its now-standard four-part vocabulary — plan, act, observe, reflect — and folded Fals Borda’s and Freire’s participatory commitments directly into that cycle rather than treating them as a separate add-on.

CASRAI’s action research guide covers the broader method, its major traditions, and when to use it. This page goes deeper on the specifically participatory variant: the cycle as PAR practices it, and the power-sharing and co-analysis mechanics that distinguish it from action research led by a researcher who simply invites community comment.

The plan-act-observe-reflect cycle, as PAR actually runs it

Every action research project moves through some version of plan, act, observe, reflect, repeated as a spiral of cycles rather than a single pass. What makes a cycle participatory is not the four labels — it’s who is in the room, and who has final say, at each of them.

• Plan. In researcher-led qualitative work with community input, the researcher typically arrives with the research question and study design already set, and invites feedback on wording or logistics. In PAR, the problem definition and the research question itself are negotiated with community co-researchers from the start — sometimes the initial “cycle zero” is spent entirely on agreeing what the actual problem is, which is frequently not the same as the problem the funder or outside researcher assumed going in.
• Act. The intervention or action being studied is carried out by, or jointly with, the co-researchers — not administered to them as subjects. In a workplace-safety PAR project, for example, the workers who identified the hazard are also the ones testing the proposed fix on the floor, not a study population a researcher observes trying it.
• Observe. Data collection roles are shared: community co-researchers are commonly trained to conduct interviews, keep field notes, run surveys, or generate their own data (photovoice images, video, oral histories) rather than being exclusively the source of data collected by an outside team. This is a genuine methodological choice with trade-offs — see CASRAI’s photovoice guide and participant observation guide for two data-generation methods PAR projects draw on heavily.
• Reflect. The cycle closes with joint interpretation of what happened and what it means for the next cycle — not a researcher writing up findings and presenting them to the community afterward. This is where co-analysis (below) does its real work, and where a project either commits to shared interpretive authority or quietly reverts to a researcher-led model at the one stage most likely to be skipped.

Each turn of the cycle feeds the next: the “reflect” stage of one cycle becomes the raw material for the next “plan” stage, which is why PAR projects are usually described as running through several cycles over months or years rather than a single planned intervention. There is no fixed number of cycles a PAR project needs — most run until the co-researchers judge the problem sufficiently understood or addressed, which is itself a participatory decision rather than one set unilaterally by a funding timeline.

Power-sharing: what actually has to change hands

“Power-sharing” is often asserted rather than specified. In practice, a PAR project’s participatory claim rests on a small number of concrete, checkable transfers of control:

• Agenda-setting. Who defines the research question and decides it’s worth answering? In PAR, this sits with the affected community from the outset, or is negotiated jointly in an early cycle — not pre-written into a grant proposal before any co-researcher is involved.
• Instrument and method design. Interview guides, survey items, and observation protocols are drafted or substantively reviewed by co-researchers, not handed to them as finished tools to administer.
• Recruitment and access. Co-researchers frequently control who is approached and how, since they hold the trust relationships an outside researcher doesn’t — this changes sampling in ways that need to be documented, not treated as a threat to validity.
• Compensation and roles. Are co-researchers paid for their time in the same way research staff are, or expected to volunteer their labor while the funded researcher is salaried? This is one of the more concrete, auditable markers of whether power-sharing is real or rhetorical, and it’s increasingly a line item funders expect to see budgeted explicitly.
• Interpretive authority. Do co-researchers have a genuine vote — not just an advisory comment — on what the data means? This is the co-analysis question, covered next.
• Authorship and dissemination decisions. Are co-researchers credited as authors, and do they have a say in where and how findings are published, including the right to withhold or reframe findings that would harm the community? CASRAI’s CRediT taxonomy gives a structured way to record a co-researcher’s actual contribution role (e.g. investigation, data curation, writing) rather than defaulting to an acknowledgments footnote.

A project that shares power on some of these and not others is not automatically illegitimate — full power-sharing on every dimension is rare and not always appropriate to the setting — but it should say plainly which dimensions were shared and which weren’t, rather than claiming “participatory” as an unqualified label. This kind of disclosure is exactly what critiques of “participation-washing” ask projects to be honest about (see Pitfalls, below).

Co-analysis: interpreting data with participants, not for them

Co-analysis is the mechanic most often skipped in projects that otherwise describe themselves as participatory, largely because it’s the most time-consuming stage to genuinely share. It is not the same as member-checking, in which a researcher analyzes the data independently and then shows participants a summary to confirm accuracy — member-checking validates a researcher’s interpretation; co-analysis builds the interpretation jointly in the first place. Common mechanics PAR projects use:

• Joint coding sessions. Co-researchers and academic researchers code a shared set of transcripts or field notes together, developing the codebook collaboratively rather than the researcher coding alone and presenting a finished scheme. CASRAI’s thematic analysis and interview coding guides cover the underlying coding mechanics that a PAR co-analysis session still relies on — the difference in PAR is who is in the room doing the coding.
• Participatory analysis workshops. Structured sessions where community co-researchers review raw or lightly-processed data (transcripts, photographs, field notes) together and generate themes directly, sometimes using visual or ranking exercises rather than formal qualitative software, so participation isn’t gated by software fluency.
• Negotiated disagreement. Co-analysis surfaces real interpretive conflict between academic and community co-researchers more often than a tidy consensus — a community reading of a finding can differ sharply from an academic one trained to look for different patterns. Well-documented PAR projects record where interpretations diverged and how (or whether) they were reconciled, rather than presenting a single smoothed-over reading as if it were unanimous.
• Feeding back into the next cycle. Because co-analysis happens inside the reflect stage of an ongoing cycle, its output is usually a decision about the next round of action, not only a written finding — this is a structural difference from a standard qualitative study, where analysis typically happens after data collection has fully concluded.

Indigenous-led research has been especially influential in formalizing what genuine co-analysis and data control require, including who retains rights over the data itself after the project ends. See CASRAI’s indigenous data sovereignty and indigenous data governance entries, and the Global Indigenous Data Alliance (GIDA)‘s CARE Principles, for a framework that treats data governance — not just data collection — as part of what “participatory” has to mean.

PAR vs. action research vs. CBPR vs. community-informed qualitative research

These four are frequently used loosely as synonyms. The distinctions that matter in practice:

• Action research is the broad family: a cyclical, practitioner-oriented method aimed at solving a real problem in the setting where it occurs. It does not require community co-researchers with shared decision authority — a teacher running solo action-research cycles on their own classroom practice is doing action research without it being participatory in the PAR sense.
• Participatory action research (PAR) is the tradition within action research that makes co-researcher power-sharing the defining feature, not an optional enhancement, at every stage described above.
• Community-based participatory research (CBPR) overlaps heavily with PAR and is often used near-interchangeably with it, particularly in public health. Where a distinction is drawn, CBPR is typically framed around an equitable, ongoing partnership between researchers and a defined community throughout the entire research process (not necessarily organized as an explicit plan-act-observe-reflect action cycle), while PAR is defined by that cyclical action-and-reflection structure specifically. See CASRAI’s CBPR dictionary entry for the fuller definition.
• Researcher-led qualitative research with community input — an advisory board that reviews a protocol, a community consultation before recruitment, participants who are interviewed and thanked — is not PAR unless that input rises to actual decision authority over the question, method, or interpretation. This is the most common point of overstatement: describing meaningful but bounded consultation as “participatory research” when the researcher retained control of the analysis throughout.

Ethics and IRB review specific to PAR

PAR complicates several assumptions built into standard human-subjects review, because co-researchers occupy two roles at once — they are both people the research is about and people conducting it.

• Consent is layered, not one-time. Beyond the standard consent to participate, PAR projects typically need a separate agreement covering co-researchers’ role, compensation, data access, and authorship expectations before the project starts — and often revisit it as the project’s direction shifts across cycles, since the “plan” a participant consented to may look different by cycle three. See CASRAI’s informed consent and principles of informed consent material for the baseline requirements this builds on.
• Co-researchers reviewing their own data raises confidentiality questions a standard protocol doesn’t anticipate — if community co-researchers are coding transcripts, they may recognize other participants’ identities even where an outside researcher wouldn’t. This needs explicit handling in the protocol, not an assumption that standard de-identification covers it.
• IRBs reviewing an evolving, multi-cycle design sometimes require amendments between cycles as the plan changes based on what the community co-researchers decide — build this into the project timeline rather than treating IRB review as a single up-front gate. CASRAI’s IRB entry and protocol amendment guide cover the mechanics.
• Indigenous and other sovereignty-asserting communities may require review or approval that runs alongside, not instead of, institutional IRB review — see free, prior and informed consent (FPIC) for the distinct standard that applies when a project involves an Indigenous community’s own governance processes.

Reporting and authorship

A PAR write-up should make its power-sharing and co-analysis choices explicit rather than implicit — reviewers and readers otherwise have no way to distinguish a genuinely co-analyzed study from one that used “participatory” as a description of recruitment alone. In practice this means:

• Naming who held authority at each stage (plan, act, observe, reflect) and where that authority was shared versus retained by the academic team.
• Describing the co-analysis process itself — how coding or theme development happened, and how disagreement between academic and community readings was handled — not just stating that analysis was “participatory.”
• Crediting co-researchers accurately. A structured contribution taxonomy such as CASRAI’s CRediT roles gives co-researchers a real, citable record of what they did (e.g. investigation, data curation, formal analysis, writing – review and editing) rather than folding substantial intellectual contribution into a generic acknowledgment.

Common pitfalls and critiques

• Participation-washing. Labeling a project “participatory” on the strength of community consultation or recruitment involvement alone, without any of the power-sharing described above actually occurring in analysis or decision-making. This is the single most common critique of PAR-labeled work in the methodological literature.
• Elite capture. Power-sharing with “the community” in practice often means power-sharing with whichever community members are already positioned to participate — organizational leaders, English speakers, people with flexible schedules — while the people most affected by the problem remain unrepresented among the co-researchers.
• Extractive co-analysis in form only. Convening a community workshop to “validate” findings the research team has already effectively finalized, timed and structured so there’s little real room to change the interpretation, functions as member-checking with participatory branding rather than genuine co-analysis.
• Uncompensated labor. Treating co-researchers’ time as volunteer contribution while academic staff are paid undermines the power-sharing claim in a very concrete, budget-line way, and is increasingly flagged by funders and ethics reviewers as an equity issue in its own right.
• Cycle fatigue and timeline mismatch. Genuine multi-cycle PAR takes longer than a fixed grant period often allows, creating pressure to compress or skip cycles — usually at the expense of the reflect/co-analysis stage specifically, since it’s the stage a rushed team can most easily do alone and write up afterward.

Frequently asked questions

Is participatory action research qualitative or quantitative?

Either, or both. PAR is defined by who holds decision authority over the research and how the cycle is run, not by the type of data collected — a PAR project can use structured surveys and quantitative outcome tracking (e.g., co-researchers monitoring an intervention’s measurable effects across cycles) as readily as interviews, photovoice, or field notes. In practice, qualitative and mixed-methods designs are more common, largely because co-analysis of open-ended data lends itself more naturally to joint interpretation than statistical modeling does.

How is PAR different from just having a community advisory board?

An advisory board typically reviews and comments on decisions made elsewhere; PAR co-researchers hold decision authority over the question, method, and interpretation directly. A project can have both — an advisory board for broader oversight and a smaller group of co-researchers embedded in the actual plan-act-observe-reflect cycle — but the two roles aren’t interchangeable, and describing advisory input alone as “participatory research” overstates what happened.

Does PAR still need IRB approval?

Yes, in essentially all cases where it meets the standard definition of human-subjects research — co-researcher involvement doesn’t exempt a project from institutional review, and in practice it usually adds review complexity rather than reducing it, for the consent, confidentiality, and multi-cycle-amendment reasons covered above.

How many cycles does a PAR project typically involve?

There’s no fixed number — PAR projects commonly run three to six cycles over the course of a year or more, though this varies widely by scope, and the decision to continue or close out the cycle spiral is itself meant to be made jointly with co-researchers rather than set unilaterally by the academic timeline or funding period.

Can PAR be used for a doctoral dissertation?

Yes, and it’s an established design in fields like education, public health, nursing, and social work — but it requires planning for the extended, multi-cycle timeline and the co-researcher relationship-building PAR needs well before candidacy, since compressing a genuinely participatory cycle to fit a standard dissertation timeline is one of the more common ways PAR gets diluted in doctoral work specifically.

See also CASRAI’s research methods pillar for the wider set of study-design and analysis guides this page sits within, including qualitative research methods and citizen science and human-subjects regulation, a related model where volunteer contributors play a different, non-decision-authority role in the research process.