navbar 4Resource papers in action research

A beginner's guide to action research  1


This is a resource file which supports the regular public program "areol" (action research and evaluation on line) offered twice a year beginning in mid-February and mid-July.  For details email Bob Dick  or

...  in which action research is briefly described, and the simultaneous achievement of action (that is, change) and research (that is, understanding) is discussed




Action research consists of a family of research methodologies which pursue action and research outcomes at the same time.  It therefore has some components which resemble consultancy or change agency, and some which resemble field research.

Conventional experimental research, for good reason, has developed certain principles to guide its conduct.  These principles are appropriate for certain types of research; but they can actually inhibit effective change.  Action research has had to develop a different set of principles.  It also has some characteristic differences from most other qualitative methods.

Action research tends to be...

  • cyclic -- similar steps tend to recur, in a similar sequence;
  • participative -- the clients and informants are involved as partners, or at least active participants, in the research process;
  • qualitative -- it deals more often with language than with numbers; and
  • reflective -- critical reflection upon the process and outcomes are important parts of each cycle.

In fact, some writers insist on those characteristics.

To achieve action, action research is responsive.  It has to be able to respond to the emerging needs of the situation.  It must be flexible in a way that some research methods cannot be.

Action research is emergent.  The process takes place gradually.  Its cyclic nature helps responsiveness.  It also aids rigour.  The early cycles are used to help decide how to conduct the later cycles.  In the later cycles, the interpretations developed in the early cycles can be tested and challenged and refined.

In most instances the use of qualitative information increases responsiveness.  It is possible to work in natural language, which is easier for informants.  There is no need to develop a metric (which may have to be abandoned later if it doesn't fit the emerging situation).

The use of language also makes the whole process more accessible to participants.  They can develop enough understanding to become co-researchers in many situations.

One crucial step in each cycle consists of critical reflection.  The researcher and others involved first recollect and then critique what has already happened.  The increased understanding which emerges from the critical reflection is then put to good use in designing the later steps.

The cycle best known in Australia is probably that of Stephen Kemmis and his colleagues at Deakin University.  The steps are:

plan --> act --> observe --> reflect (and then --> plan etc.)

The reflection leads on to the next stage of planning.  The "planning" isn't a separate and prior step; it is embedded in the action and reflection.  Short, multiple cycles allow greater rigour to be achieved.

As change is intended to result, effective action research depends upon the agreement and commitment of those affected by it.  This is usually generated by involving them directly in the research process.  In many instances, researchers try to involve them as equal partners.

Action research in more detail

I regard action research as a methodology which is intended to have both action outcomes and research outcomes.  I recognise, too, that in some action research the research component mostly takes the form of understanding on the part of those involved.  The action is primary.  In distinction, there are some forms of action research where research is the main emphasis and the action is almost a fringe benefit.

I regard all of these as action research.  This definition is capable of encompassing a variety of research and intervention methods.  It is broad enough to include, as examples, the critical action research approach of Carr and Kemmis (1986), the soft systems methodology of Checkland (1981), and perhaps even the evaluation of Guba and Lincoln (1989), to name just a few.

The responsiveness of action research allows it to be used to develop hypotheses from the data, "on the run" as it were.  It can therefore also be used as a research tool for investigative or pilot research, and generally for diagnosis or evaluation.

Cyclic, participative, qualitative

Most writers on the topic state or assume that action research is cyclic, or at least spiral in structure.  To put this differently, certain more-or-less similar steps tend to recur, in more-or-less similar order, at different phases of an action research study.  At the same time (so the action researcher hopes) progress is made towards appropriate action and research outcomes.

A commonly known cycle is that of the influential model of Kemmis and McTaggart (1988) mentioned earlier -- plan, act, observe, reflect; then, in the light of this, plan for the next cycle.

It is also generally held that action research is participative, though writers differ on how participative it is.  My own preference is to use participative methods.  On the other hand I don't see why action research must be limited to this.

So, the extent of participation may vary.  In some instances there may be a genuine partnership between researcher and others.  The distinction between researcher and others may disappear.

On other occasions the researcher may choose for whatever reason to maintain a separate role.  Participation may be limited to being involved as an informant.  The participants, too, may choose something less than full partnership for themselves under some circumstances.

Most action research is qualitative.  Some is a mix of qualitative and quantitative.  All else being equal, numbers do offer advantages.  In field settings, though, one often has to make other sacrifices to be able to use them.  Most importantly, sometimes numbers are not easily applied to some features of a study.  If these include features of particular interest or importance, the choice is between qualitative research or omitting important features.

In addition, developing a suitable quantitative measure is often difficult and time-consuming.  It may be more time-efficient to use qualitative data.  As I mentioned before, it is also easier to be flexible and responsive to the situation if you are using qualitative methods.

In short, it is my view that action research more often than not exhibits certain features.  It tends to be, in some sense of the terms, cyclic, participative, qualitative and reflective.

I see all of these features except the last as choices to be made by the researcher and the other participants.  In my view, good action research (and good research of any variety) is research where, among other features, appropriate choices are made.  Perhaps even critical reflection might be abandoned for sufficient reason.

"Good" action research

Whatever action research is, I suspect it is mostly or always emergent and responsive.  In fact, I think that the choices made about its cyclic and qualitative nature are mostly to be justified in terms of the responsiveness which they allow.  This may be true of decisions about participation too. 

In many field settings it is not possible to use more traditional quasi-experimental research methods.  2  They can't readily be adjusted to the demands of the situation.  If you do alter them in midstream you may have to abandon the data collected up to that point.  (This is because you have probably altered the odds under the null hypothesis.)

But to achieve both action and research outcomes requires responsiveness -- to the situation, and the people, and the growing understanding on the part of those involved.  Using a cyclic process in most circumstances enhances responsiveness.  It makes sense to design the later stages of an action research activity in such a way that you capitalise on the understanding developed in the early stages.

It is the cyclic nature of action research which allows responsiveness.  It is often difficult to know just where a field intervention will end.  Precise research questions at the beginning of a project may mislead researcher and clients.

Imprecise questions and methods can be expected to yield imprecise answers initially.  But if those imprecise answers can help to refine questions and methods, then each cycle can be a step in the direction of better action and research.

In other words, there are times when the initial use of fuzzy methods to answer fuzzy questions is the only appropriate choice.  Action research provides enough flexibility to allow fuzzy beginnings while progressing towards appropriate endings.

To my mind, a cyclic process is important.  It gives more chances to learn from experience provided that there is real reflection on the process and on the outcomes, intended and unintended.  Qualitative information is less constraining of the process.

Participation is a somewhat different issue, more to do with action than research.  Action outcomes can usually be achieved only with some commitment from those most affected.  One of the most important ways of securing that commitment is through involving those affected.

There may well be other reasons, too.  For instance, for some researchers it is more ethical to use participative methods (in general, this is my position in the action research I do).  On some occasions the eventual interpretation of information is richer if involvement is greater.

So far, I have taken the view that action research can take many forms.  There are some conditions, however, that I believe are more important.  As a starting assumption I assume that good action research is empirical: responsive to the evidence.  I also think it is important that the evidence is used critically rather than uncritically.

Again, a cyclic process allows this to happen more easily.  If each step is preceded by planning and followed by review, learning by researcher and client is greater.

The quality of evidence can also be increased by the use of multiple sources of evidence within all or most cycles.  Differences between data sources, used critically, can then lead the researchers and the participants towards a deeper and more accurate understanding.  Literature can be such an alternative data source.

I would sum up my recommendations for good action research in this way:

  • Use multiple cycles, with planning before action and critical analysis after it.
  • Within each cycle --
    • use multiple data sources;
    • and try to disprove the interpretations arising from earlier cycles.

Action research is a family of research processes whose flexibility allows learning and responsiveness.  Vague beginnings can move towards better understanding and practical improvement through the critical analysis of the information, the interpretation of it, and the methods used.

Good action researchers, I think, critique what they do and how they do it, the better to learn from the experience.  It is the balance between critical reflection and flexibility which allows adequate rigour to be achieved even in confused field settings.


In summary, I assume that action research is true to label: it pursues action and research outcomes.  It is most effective when the end result emerges from the data.  The conclusions drawn are data-based, preferably drawing the data from multiple sources.  The conclusions emerge slowly over the course of the study.  At each cycle the researchers challenge the emerging conclusions by vigorously pursuing disconfirming evidence.

I think that the major justification for action research methods is that they can be responsive to the situation in a way that many other research methods can not be, at least in the short term.  On these grounds I think action research will usually, though perhaps not always, be cyclic in nature.  In the interests of rigour, each cycle will include critical reflection.  In most instances it will also be qualitative and participative to some extent.


  1. Modified from _Arcs Newsletter_, Vol 1.  No.  1, May 1993, pages 5-9.  I am pleased to acknowledge the help of Ron Passfield and Paul Wildman with the earlier version of this document.  back ]

  2. I invite you to choose some simple quasi-experimental field study, and study how well it meets the threats to the validity of its results.  Then compare how easily action research might fare in the same setting.  Some quasi-experimental research is superb.  But I think sometimes the difficulties of field research are used to justify poor quasi-experimental designs.  This appears to be because they are traditional, because the same liberty is less often extended to other research paradigms.
    back ]


Carr, W.  and Kemmis, S.  (1986) Becoming critical: education knowledge and action research.  London: Falmer Press.

Checkland, P.  (1981) Systems thinking, systems practice.  Chichester: Wiley.

Guba, E.G.  and Lincoln, Y.S.  (1989) Fourth generation evaluation.  Newbury Park, Ca.: Sage.

Kemmis, S.  and McTaggart, R., eds.  (1988) The action research planner, third edition.  Victoria: Deakin University.



Copyright (c) Bob Dick, Ron Passfield, Paul Wildman 1995-2000.  This document may be copied if it is not included in documents sold at a profit, and this and the following notice are included.

This document can be cited as follows:

Dick, B.  (2000) A beginner's guide to action research [On line].  Available at



navbar 4  

Maintained by
Bob Dick; this version 1.08w last revised 20000429

A text version is available at URL