More formally
described, action research is a research paradigm: a family of
research methods united by a certain set of principles and a
certain style. Examples of other research paradigms are
experimental (including quasi-experimental), and
ethnographic. And there are others. There are many
varieties of action research within the paradigm, at least as the
term "action research" is used here. Other views may
differ.
As its name implies,
action research is intended to achieve both action and
research. It is suited to situations where you wish to bring
about action in the form of change, and at the same time develop
an understanding which informs the change and is an addition to
what is known.
Action research is
typically cyclic. The later cycles are used to challenge and
refine the results of the earlier cycles. In most of its
forms it also tends to be qualitative, and participative.
Some would say that neither of these is strictly necessary.
Some insist, strongly, on high participation as a necessary
feature. Some have defined publication of results as a
defining characteristic.
Action research is also
critically reflective. The researchers (and probably clients
acting as co-researchers) regularly and systematically critique
what they are doing. They refine the questions they are
asking and the methods they are using and the understanding and
subsequent action plans they are developing.
Action research commonly
proceeds like this. The researcher, again often involving
clients as co-researchers, plans the first or next step.
This is then carried out. Researchers meet to recollect and
critique their experience. In the light of this, they decide
what to do for the next step: what information do they need or
what outcome to pursue, and what method to use.
In short, action
research alternates between action and critical reflection.
The reflection consists first of analysing what has already
happened in previous steps, and then of planning what next step to
take.
Is action research
intended as a substitute for
experimental or quasi-experimental research?
Emphatically not.
Different research paradigms serve different purposes. It
may with benefit be used in some situations where
quasi-experimental designs are presently used. But there are
other situations where experimentation or quasi-experimentation
may be far better suited.
For instance, you may
wish to find out about a few variables, and the causal
relationships between them. If so, experimental or
quasi-experimental research will serve you much better than action
research. Alternatively, you may wish to explore some
organisation or group or culture in depth, and with low
impact. For this, you may do better to use ethnographic or
other qualitative methods.
Action research methods
are most likely to be appropriate when you do not know where to
start, and do not have a lot of time to invest in the study.
It is useful for exploratory research, where you do not yet have a
very precise research question.
But it is most valuable
when you have to be responsive to the changing demands of a
situation. For example, this may be when you wish to build a
research component into some change program or the like. For
this reason it can also be used for evaluation of an ongoing
program.
Good research, it can be
argued, is research which uses a methodology which fits the
situation and the goals you are pursuing. Sometimes, that is
action research. Sometimes it is something else.
What is the
relationship of action research to science?
By some definitions
action research is science. By other definitions it
complements science.
Those who think it is
science recognise that it shares with other approaches a pursuit
of understanding. It values scepticism and empiricism.
That is, in pursuing knowledge it strives vigorously to disconfirm
present views, and it uses evidence to do that.
If science is systematic
and sceptical empiricism, then action research is
scientific.
Those who think it is
not science point to a number of ways in which it differs from
some other scientific paradigms. For example, action
research:
- does not usually
provide causal explanations of what is studied
- usually does not
attempt to answer questions which are as precise as those
addressed by e.g. experimental research
- mostly uses
qualitative data
- is done by
researchers who do not make the same effort to distance
themselves from what they are researching; in fact, they often
set out to build close relationships with the people within the
system studied
- uses a research
process which, rather than being standardised, is modified on
the run in response to what happens
- does not necessarily
seek explanations at a more specific level than the phenomena
it is dealing with -- that is, it tends not to be
"reductionist".
In addition, action
research is often regarded as giving answers which are specific to
the particular situation, and which cannot be generalised to other
situations. This is partly true, but a more complex issue
than this criticism recognises. It can be said that the
generalisation which experimental science provides is difficult to
relate to social situations. I believe its value has
therefore been overstated by those who criticise action research
for its lack of generalisation.
Because of the
differences between experimental science and action research, some
would not regard action research as sufficiently rigorous to be
regarded as science. In reply it can be said that this
depends upon particular definitions of rigour.
In any event, action
research can provide a useful contribution to knowledge, both in
its own right and as a complement to more traditional approaches
to research. It can provide answers to fuzzy and general
questions. It can provide more "realistic"
explanations. If necessary, more traditional methods can
then be used to develop causal explanations, and to check how well
the explanations can be generalised to other
situations.
Further, action research
is _designed_ to allow simultaneous change and
understanding. You could say that
- action research =
action + research
It allows for systematic
understanding to arise from activities which are oriented towards
change. It has a capacity to respond to the demands of the
informants and the situation in a way which most other paradigms
can not.
What is the difference
between action research and
consulting, or action research and other professional
practice? Am I already doing action research?
It is true that many
practitioners, on first hearing about action research, reply: "I
am already doing that."
It is true that action
research and some forms of practice are in some ways
similar. Both are often directed towards the achievement of
change. Both are qualitative and often participative.
Both tend to be flexible and cyclic. In both instances,
there is a desire to base planned changes in the situation on
understanding, and to derive that understanding from
evidence.
Because of these and
other similarities, less formal approaches to action research may
be very similar to some forms of practice.
However, most forms of
action research are more deliberate in their pursuit of
understanding. Most importantly, frequent critical
reflection is a formal and central part of most action
research. Most practice is much less reflective than action
research. Most practice, if it does use reflection, is
neither as deliberate nor as critical in its use.
There are other
processes which do use regular and systematic critical
reflection. Action learning and experiential learning
provide two example. These and action research have much in
common, and shade into one another at the boundaries.
Can action research
yield causal explanations?
In the sense of
explanations given in terms of causal relationships between
variables, generally not. On occasions, the understanding
developed may be causal -- people may come to recognise that event
A precedes and probably causes event B. But this isn't a
central interest.
A different type of
causal explanation often features in action research. A
causal connection between certain actions and certain outcomes may
be identified. Such a connection is then tested when the
planned actions are carried out. Action research can give
causal explanations in _this_ sense. In fact, as action
research is intended to produce action, this is usually an
emphasis.
How can action research
justify its conclusions
when it provide give causal explanations?
It is true that action
research does not usually try to provide causal explanations --
explanations of causal relationships between limited numbers of
variables.
It shares a belief with
some "systems" approaches: in some situations, causal explanations
are either not possible, or too cumbersome to be very
useful. When there are many variables, and they interact
(often bi-directionally) in complex ways, causal explanations are
themselves likely to be very complex.
To put it differently,
when almost everything has an effect on almost everything else, it
may not be much practical help to know that "a" influences
"b".
As discussed above, it
is usually helpful to know that I can often achieve outcome
Z by doing action X. In that sense, causal understanding
actively pursued.
Can action research be
quantitative?
Yes, though it is
usually qualitative.
Most of the time action
research uses natural language rather than numbers: the use of
natural language suits a paradigm which is participative and
responsive to the situation. People communicate naturally in
language. Communication is an important part of action
research, especially in its more participative
varieties.
Quantitative measures
can be valuable. But developing them requires a substantial
investment in time. This may not be warranted if you are
likely to change your mind about the measures that you need.
When suitable quantitative measures are available, there is no
reason why they may not be used.
Qualitative and
quantitative approaches can often complement each other
well. For example, you might carry out a quantitative survey
or other activity. These data might then be interpreted
qualitatively by the people within the system being
analysed.
I should probably
mention the view held in some quarters that qualitative and
quantitative data ought not to be used in conjunction. This
is based on the argument that they depend upon different
philosophical assumptions. I won't go into detail here, but
in my view this is impractical. It places philosophy as an
unavoidable foundation for all research. On this view, we
should give up all research right away. Philosophers don't
presently agree on the foundations of research, and I see no
reason to think they ever will.
Can you generalise from
action research? If not,
how can action research add to knowledge?
Often, you can't
generalise from action research. In other words, you can
often make claims only about the people and/or systems actually
studied. It may not be safe to assume that other people or
other systems are the same. For that matter, it may be
hazardous to assume that what is true of this person (or
organisation or community) today will be true next
year.
This is often held to be
one of the major disadvantages of action research.
Experimental research, done well, does allow generalisation.
An experimental claim can often be taken to be applicable
universally.
The issue can also be
looked at in a different way. Generalisability might be
regarded as global relevance -- the ability to apply a finding
from one experimental setting into other settings. Action
research then pursues local relevance, if necessary at the expense
of global relevance. Experimentation often achieves global
relevance, but at the cost of being difficult to apply practically
to local situations.
There are ways in
which some generalisability can be claimed for the findings of
action research. For example, if several studies in diverse
settings give similar findings, this allows greater
generalisability than a single study typically does. Similar
actions may produce similar outcomes in different situations; this
implies generalisability. One can also use
indirectly-relevant literatures to test the relevance of
findings.
Action research and
experimentation value different outcomes. Unless something
is done about it, action research is often limited by the local
relevance of its findings. Unless something is done about
it, experimental research is often limited by the narrowly defined
set of variables it investigates. It might be argued that
the generalisability of experimental research, when done within
social systems, is a very narrow form of generalisability.
Knowing that it is true that "a" causes "b" may not be much help
when "a" and "b" are immersed in a system of very many other,
interconnected, variables.
Are social sciences
really science?
I don't know. Are
they? It depends on your definitions of science, and what
you include as social science. This is an example of a
definitional problem. While I think it is worthwhile to
define terms, I find it less useful to debate the
definitions. For me, action research is useful in settings
where other paradigms work less well. It works. I'm
not concerned whether or not it's science that I'm doing.
How can you do action
research when you
don't know enough to develop a hypothesis
to start with?
This is actually another
definitional issue in disguise. It assumes that research is
necessarily a test of a hypothesis.
If you believe this,
then you will conclude that action research is not really
research. As long as you don't try to persuade others of
your conclusion, I doubt that you will do much harm. If you
wish to change the terminological habits of action researchers you
may be in some difficulty. You will then have to devise a
different term to describe it. As it has been known as
action research for over half a century, I don't hold much hope
that you will change the verbal habits of those who do action
research.
Copyright (c) Bob Dick
and Pam Swepson 1997-2013. 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. and Swepson,
P. (2013) Action research FAQ: "frequently asked
questions" file [On line]. Available at
http://www.aral.com.au/resources/arfaq.html
