Introduction It
has been the experience of a good number of those working
towards a higher degree which involves a major research
element that little or no guidance is given on the
research process itself. Frequently it seems to be
expected that knowledge of research is something to
be ''picked up'' as one progresses through the activity.
Whilst this is no doubt an important part of the learning
that takes place it is an approach which can lead to
frustrations and unnecessary failures along the way.
In
the absence of any specific instruction, the higher degree
researcher may turn to the textbooks. A number now exist
but many are quite detailed and very specific. Few give
even a brief introduction. This guide is intended to
do that, and provide a basis for further reading and
development. It looks at the meaning of research and
the processes involved, and briefly describes the main
methods and techniques employed. A brief glossary of
basic terms helps sort out some of the semantics.
Finally, some guided reading is given with a few comments
on the contents.
The
reality of research It
may seem irrelevant to ask what research is all about;
since so many are doing it, then most people (at least
those carrying it out) must know what it is all about.
Some
people consider it to be a cosy and personal activity
that could be indulged in from time to time from the
safety of an armchair - and certainly not stretching
beyond a pile of books resting on the coffee-table. To
others, research is a rigorous activity aimed at developing
new bodies of knowledge and is normally ''acceptable''
only in a physical laboratory situation. Perhaps both
views are valid. For the former Central Training Council
(1973), research was seen as ''the discovery of fact
through a systematic process of survey, hypothesis and
experiment''. This is somewhat closer to the rigorous
scientific approach than the ''cosy, personal activity''.
For Rummell and Ballaine (1963) research is a ''careful
inquiry or examination to discover new information or
relationships and to expand and to verify existing knowledge''.
This immediately implies a vital role for research -
one of helping researchers to underline the effectiveness
of their approaches. It thus seems that research is an
inevitable element of the total professional process
- its absence leading to obsolescence, reduced effectiveness
and dissatisfaction.
But
research is not a ''careful enquiry'' for its own sake
- it always starts with some sort of problem, or at least
should do if it is to be of use to anyone. Whether the
research is carried out for personal or practical purposes,
some reason exists for it. The problem may be that little
is known about some phenomenon and it would be good to
have more knowledge about it (sometimes known as ''pure''
or ''fundamental'' research).
On
the other hand, the problem may have much more practical
significance in that the research may help us to do something
we could not do before (often referred to as ''applied
research''). This is the essence of the American school
of ''pragmatic philosophers'' for whom any theorising
or research is a waste of time unless it has ''cash value'',
i.e. helps us to solve problems or understand things
better than we did before. Some forms of research can
be much more practical than others. But, if the problem
is theoretical, then academic research is necessary.
The danger of falling between two stools lies in confusing
academic research with practical, i.e. operational, problems.
We
have more concepts currently than we can adequately cope
with. This is certainly true of the behavioural sciences.
Notions such as motivation, perception and learning are
highly developed and researched (if not agreed upon)
at the conceptual level yet lag far behind when it comes
to applying them to organisational situations. This point
seems to be missed because so much literature exists.
What is now required is a greater emphasis on ensuring
that these concepts can be usefully employed. This is
not to suggest that highly sophisticated academic research
is no longer required - it is, but at a considerably
reduced level. The poor image that research so often
appears to have is largely the result of inappropriate
approaches based on academic requirements - it would
be more fruitful to adopt problem-centred approaches
based on situational requirements.
It
is important at this juncture to refer briefly to a form
of research, known as action research. This form of research
is essentially ''applied'' in nature, even more so when
it is realised that in action research the researcher
actually gets involved in what he is researching. It
developed primarily from the need of organisational analysts
to explore thoroughly the organisation and at the same
time to ''change'' and ''develop'' it. The researcher
acts as a ''catalyst'', a ''facilitator'', or even a
''mirror'' for the organisation - a very far cry from
the ''objective impartiality'' of traditional research.
Action research therefore requires a joint approach -
a definite and agreed collaboration between organisation
and researcher.
The
research process
Understanding
research starts with knowing what, in essence, it is
all about. As we have seen, the process of research
starts, usually, with some form of problem or question.
The problem/question may be the researcher's - he may
wish to know which learning theory of several is most
relevant in explaining certain levels of performance
in different situations. The problem may, of course,
be initiated by a manager or someone else; perhaps
wanting to decide on the best technique for developing
greater participation. In either case, the requirement
is for some information that will shed light on the
problem and help make a decision to solve it. It may
be that solutions are not the end result of the research,
but rather the development of a new theory or body
of knowledge. Whatever the end result, the starting-point
is represented by an urge to find out, to explore,
to evaluate - in short, to do research. In between
these end points exist a number of other steps. Having
defined, or at least acknowledged, the problem or area
of interest, researchers may carry out a preliminary
study. This will enable them to set out the parameters
of the problem and to gain some idea of the essential
information to be sought. Such exploratory studies, free
from too much bias or preconceived ideas, can be of great
value in setting the research in the right direction.
For example, the problem being looked at may have been
concerned with inadequate bonus earnings related to immediate
post-training periods. The temptation here is to blame
the training. An exploratory study (usually much less
costly than the full treatment) might uncover poor supervision
during the first weeks on the job, or lack of understanding
of the bonus scheme, as possible alternative explanations.
If
this preliminary work is reasonably thorough, the next
stages can be less embracing than might otherwise be
the case. From this work the researcher may well set
up a hypothesis, or a series of hypotheses, which can
then be tested against reality. In simple terms a hypothesis
is an imagined answer to a real question. In the example
just given, the question would be ''What causes low levels
of bonus earnings in immediate post-training periods?''
The answer, as we have seen, might be based on guesswork,
theoretical inspiration, or an appreciation of the factors
involved, or indeed a combination of all three. In our
case, the hypothesis might be that, in immediate post-training
periods, operators will earn low levels of bonus if inadequate
supervision persists.
Having
framed this hypothesis, researchers then seek information,
or data, which will allow them to test its validity.
They might decide to check records for low earnings,
and see what situations led to this; or they could monitor
earnings and performance levels in two sections, one
of which had a high ratio of supervision, the other a
low ratio. The data collected would then be analysed
and subjected, possibly, to several statistical tests
to determine whether the proposed ''answer'' holds true
or not and with what degree of confidence or faith it
can be accepted. The results of this analysis and deliberation
would be interpreted and communicated - via reports,
seminars, planning groups or whatever - to the ''client''.
This phase can be a difficult one, but need not be so
inconclusive as so often is the case.
It
should be stressed that the research process may not
necessarily be geared to the testing of hypotheses. Often
a researcher will be more interested in the exploratory
stage, with a view to developing a number of alternative
hypotheses for later testing. If this proves successful,
a useful contribution will have been made to knowledge.
Levels
of research
Not
all research takes place at the same level of scientific
sophistication. The reason for this hinges on the state
of knowledge of the subject under investigation and
the hoped-for outcomes (and uses) of the research.
In general, most sciences follow a similar pattern
of development and progression, and the social and
behavioural sciences are no exception. In some disciplines
- such as, perhaps, biology and botany - the emphasis
is on the one level rather than another.
Perhaps
the most basic level of research is that connected with
describing what exists around us. For example, we may
not have enough knowledge about the different types of
training procedures in use - the first step in knowing
about them must be to describe them. Thus, job descriptions
are quite useful in telling us something about the work
of managers. Having obtained a description of these phenomena,
the researcher may be interested in comparing them for
differences or similarities, as we would with job descriptions,
in order to establish some form of job evaluation framework,
or training characteristics. This process of comparing
and grouping is known as classification (or categorisation).
The
next level of research, that of explanation, then becomes
possible. We can start to ask questions such as why?
and how? Our interest is in understanding what is happening
and seeking ways of representing this through theoretical
development, models, propositions and so on. You may
want to know, for example, why one student progresses
more quickly under the same conditions as someone else.
Hopefully, all this knowledge will lead to a stage of
development where prediction of events, circumstances,
behaviour, etc. is possible. In the physical and advanced
sciences, this is the level at which most researchers
are now operating. None of the space programmes would
have been possible if this were not so. In those disciplines
concerned with human behaviour, it is exceptional to
find some truly predictive theory based on adequate research.
Whilst the testing of hypotheses may take on this predictive
form, we are still very much concerned with understanding
and explaining human behaviour. In the field of training
and education, this must be so - with exceptions - until
the disciplines associated with our efforts (psychology,
sociology, neurology and so on) become more precise and
predictive themselves.
Research
methods
A
number of quite different methods can
be employed in establishing the acceptability or otherwise
of a hypothesis, or helping solve a problem, and in
some cases these can be used to complement each other.
Each has its advantages and drawbacks, knowledge of
which can aid you in assessing the feasibility of achieving
your objectives.
Experiment
The
classical method, used in the physical sciences for
many years, is the experiment. In most physical sciences,
if not all, the researcher aims to set up a situation
in which all variables can be controlled or varied
at will. The usual approach is to hold all variables
constant except one. By varying this one and monitoring
changes in the ''output'', the relationship between
variables can be carefully studied and documented.
In essence, the researcher seeks to vary one of several
independent (or input) variables whilst measuring the
effects on the dependent (or output) variable(s), keeping
intervening variables constant. For example, it would
be possible to vary the petrol mixture fed to an internal
combustion engine and note the difference in speed
or power achieved but, at the same time, keeping (say)
pressure or load constant and controlling room temperature
in the laboratory. When dealing with human behaviour,
it is not possible strictly to adhere to this approach,
although sometimes one can get reasonably close. It
might be possible to vary the instructional techniques
used for training managers and to measure their achievements.
Here,
however, control over intervening variables such as ability,
intelligence, attitude and the like would be complex
but the use of matched groups (e.g. different groups
of managers who had roughly the same IQ, etc.) undergoing
different approaches would take us a step nearer to the
''scientific'' method. We must not delude ourselves,
however, into thinking that this approach is ''foolproof''
- it is not. We cannot control, for example, the activities
of people outside work - their love-lives, drinking habits,
arguments with spouses - which may well affect their
performance. We can, nonetheless, attempt to recognise
and account for these factors. Experiments can broadly
be considered to be of two types - the laboratory experiment,
where the problem to be studied is divorced from the
other facets of the real world surrounding it, but not
connected to it; and the field experiment, where attempts
are made to study the problem in its real setting and
to minimise the influences of seemingly unconnected factors
or variables.
Most
experiments in training and education are likely to be
field experiments, although the existence of training
schools, simulators and so on, make laboratory experiments
quite attractive - even though the results may not have
much significance in the ''real'' setting.
Survey
This
is almost certainly the most widely adopted method
in the social sciences - and most aspects of training
and education are of a virtually ''social scientific''
nature.
Surveys
are usually cheaper, quicker and broader in coverage
than any experiment can hope to be but, on the other
hand, very often lack the control and in-depth exploration
of the experiment. Relying in the main on the techniques
of sampling, interviewing and/or the questionnaire, a
survey can provide useful information on many problems
or issues faced by the trainer or educator.
For
example, you may have wondered how people feel about
the training provided; what subject-matters people think
should be given priority treatment on courses; if members
of your organisation think participation is a good thing;
or maybe what young managers think about their career
prospects. These and other issues can be explored using
survey research methods involving research instruments
(e.g. questionnaires, checklists) which, if constructed
and tested adequately, can produce useful information.
By their very nature, surveys produce a lot of information
- or data, as researchers tend to call the basic responses
to questions. Thought must therefore be given to how
it can be analysed, preferably before the data are collected.
If this is not done, severe problems can arise causing
frustration, and even the abandonment of the project.
Many excellent techniques of analysis exist - from slogging
it out by hand to computer processing, and can be found
described in a number of sources (see, for example, Oppenheim,
1960 and Kerlinger, 1973). A survey, of course, is not
the answer to all research requirements. Used widely,
it can produce useful information in a short time, but
may suffer from problems associated with people not wanting
or bothering to respond to questions; giving false answers
where they do; treating it as a joke; misunderstanding
its purpose; and a host of others. Many of these problems
can be avoided or certainly reduced in terms of their
impact on the results, but only if care and attention
are applied throughout.
Carrying
out a survey is not so simple as some people would have
us believe, nor is it so difficult and scientifically
immoral as others obviously do believe. As with all things
in life, it has its place - as a planned collection of
information: no more, no less!
Field
(or case) study
Probably
falling between the experiment and the survey in terms
of scientific acceptability, usefulness to the practitioner,
and capacity to produce theoretical advances, the field
study (of which the case study is a particular example)
has considerable utility. Whilst the techniques adopted
(e.g. interviews, observations, questionnaires) are
similar to survey research techniques, breadth of coverage
is sacrificed for depth of probing and understanding.
Unlike the experiment, a field study does not normally
involve manipulating independent (or input, or causal)
variables, except possibly through statistical means.
Rather, the study involves measuring, looking at -
studying! - what is there, and how it got there, i.e.
it is historical. Two types of study can be carried
out.
Exploratory
studies seek to establish ''what is''; to discover significant
variables and relations between them and to lay the foundations
for perhaps more scientific work aimed at testing hypotheses.
For example, you may have wondered what variables have
the greatest influence in on-the-job learning: a field
study, probing through discussions with, and observations
of, the people involved, might throw some light on this
question. You might then be in a position to predict
how the variables would be related to each other in certain
situations, and set out to test this prediction. This
would be a different form of field study: hypothesis
testing rather than hypothesis generating. The point
to note about field studies is that they do not attempt
rigorous control - both a strength and a weakness. The
strength is that we obtain greater realism in the research;
the weakness is that things may get out of hand (sudden
incidents erupting) destroying the validity of the research.
Field studies are often costly and time-consuming, and
may, of course, not produce much in the way of earth-shattering
conclusions. For most of our requirements, however, the
results can be rewarding. In a more specific sense, studies
can be confined to particular persons or units or organisations,
and such case studies can produce illuminating information.
It must be recognised, though, that single cases may
have little value in explaining events outside the confines
of the case itself - it thus lacks ''generalisability''.
Techniques
of research
Whilst
many texts refer to instrumentation, measurement devices,
methods of data collection and the like to mean the
way in which the researcher goes about acquiring information
within one of the frameworks just described, it is
best to use the term ''technique''. This is because
some of the other terms are too precise (such as ''instrumentation'')
or involve the use of terms applied elsewhere (such
as in ''data collection method''). In essence, we are
talking about ''how'' we do it as opposed to ''what''
we do or ''why'' we do it. Only a brief description
of the most general techniques is given here - most
are well discussed (if not always jargon-free) in texts
on research methods.
Observation
This
is the most classical and natural of techniques. It
simply involves looking at what is going on - watching
and listening. We all do it, most of us badly because
we do not know what to look for or how to record it.
Work study practitioners are probably the most competent
of observers - after all, they have been trained to
do it.
So,
too, are most researchers and teachers. Important in
being a good observer is to have a wide scope, great
capacity for being alert, and the ability to pick up
significant events. Here, technology can aid us, offering
services ranging from simple pen and paper through tape
recorders and cameras to videotapes. If carried out quietly,
unobtrusively, and shrewdly, observation can be a useful,
if not powerful, technique. It does not allow much scope
for probing, exploring relationships further, unless
used in conjunction with other techniques. The combinational
use of techniques is now quite widespread and has much
to commend it. Since, however, observation is ''simple''
(if time-consuming) and opportunities for using it often
present themselves, it can be used quite effectively
for its purpose - enabling a general picture to be built
up.
Tutorial
Interviews
It
is quite tempting to suppose that the interview was
first ''created'' by the early observers who could
not resist asking people why they were doing what they
were doing. Whatever its origin, the interview has
a fundamental role in social and behavioural research.
It allows for exploration and probing in depth and,
if you have got the money and the time, in breadth
as well. The questions asked might stem from periods
of general observation - and this is to be preferred
to just dreaming up questions in the bath! Interviews
can be unstructured and free-ranging: a general discussion,
picking up points and issues as they emerge and pursuing
them in some depth; or they can be structured around
questions and issues determined in advance: based on
a literature search, preconceived ideas or prior investigation.
If the questioning is non-directive and free from biased
or loaded questions; if the interviewer is a good,
attentive listener (and adept recorder); and if the
interviewee is of a mind to ''tell
it like it is'', the results can be very effective.
However, problems of time, cost and sampling related
to your research objectives may mean that a full-scale
interview programme is not possible or necessary.
For
example, you may wish to gain ideas for the development
of a job appraisal form - for this, a small number of
''pilot'' interviews would be quite effective. If you
wanted detailed views on the attitudes of people on your
courses, a wider programme of in-depth interviews could
be of use. Remember, too, that for some purposes (e.g.
where a ''testing of views'' is required), group interviews
have a role to play. Whilst they can be a bit more difficult
to handle, the overall end results may provide more insights
than would the same people interviewed separately. Whatever
sort of interview is relevant, the means of recording
information must be thought through in advance: whether
to tape record unstructured group interviews or take
notes; how to design an interview schedule (a ''questionnaire''
completed by the interviewer) for structured interviews
with maximum ease of recording and information capture
but minimum effect on interviewees - e.g. a feeling of
''not being listened to'' as you write copious notes.
As with all research matters, a little advance thinking
and planning can save a lot of later difficulties.
Questionnaires
Whilst
undoubtedly the most used technique - or, more correctly,
instrument - of researchers in the behavioural and
social sciences, questionnaires do pose problems. The
major difficulties are associated with response rates,
bias and flexibility. Since questionnaires are important
to the survey researcher (as are interviews) the effect
on the results of someone not responding must be considered.
Who are they, what are their characteristics, would
they share the views of those who did respond? are
questions that have to be faced. Even when reasonable
response rates are achieved (more than 40 per cent)
those problems still exist, and in any case the resulting
data may be biased. Bias might be due to respondents
anticipating the answers they think the researcher
wants, or putting down ''socially expected''
answers (on the basis of what is ''good'', or would
be the ''right sort of thing to say''), or simply as
a result of finding some form of pattern to, say, the
first ten questions and assuming the pattern must be
repeated.
These
and other difficulties can be minimised, if not overcome,
by careful design and piloting of the questionnaire.
Flexibility, however, is not so much a design problem
(although it can be considerably reduced by poor design)
- it is much more a function of the nature of the research
questions being asked. Answers might range from factual
information (e.g. date of birth) - through simple ''yes''/''no''
replies (e.g. do you smoke?), to scale-type responses
of the agree/disagree form (e.g. training is a waste
of time!) with a number of possible responses in between.
Often,
however, the person filling in the questionnaire would
like to say ''yes - but!'' and has no opportunity to
do so. It is the qualifying ''but'' that may be important
and an interview would allow it to be explored. For information
of a somewhat broad and superficial nature (detail can
be obtained, of course, but mostly factual), involving
large numbers of people, the questionnaire is a useful
technique and is relatively easy and cheap to use. If
thought is given to the major drawbacks and to the way
in which the data are to be analysed, there is every
reason to expect fairly reliable and valid results. If
preceded or backed-up by interviews or observations,
many additional benefits can be derived as well as difficulties
minimised.
Other
techniques
Many
other techniques exist, some of them variations on
those briefly described here, others developed for
specific purposes. They will not be discussed here
since most of them require considerable experience
in their design and use. They can be found in many
of the texts, for example Helmstadter (1970), from
which special references (e.g. to sociometry; testing;
scaling and projective techniques in psychology) can
be obtained. Often such techniques are limited.
Model: A
pictorial representation of concepts and relations between
concepts, e.g. graph or flow diagram. Not to be confused
with the use of ''model'' which implies ''perfect'' -
as in ''a model job''!
Paradigm: Another
word for model, but without the latter's value connotations.
Proposition: A
statement or assertion concerning the problem or topic
being researched: origins and use mainly in philosophy,
logic and mathematics.
Reliability: A
term used mainly in connection with measurements (as
via a questionnaire, or test) and refers to repeatability,
i.e. getting the same results on different occasions
when measuring the same entity which has not changed
in dimensions since it was first measured.
Sample: A
number of people, objects or events chosen from a larger
''population'' on the basis of representing (or being
representative of) that population. Sampling, and sampling
theory, are important facets of survey research.
Theory: A
set of general laws (interrelated concepts) that specifies
relations among variables. A theory thus represents,
in a systematic way, the phenomena in the world around
us, explaining them and allowing predictions to be made
or, to borrow a phrase, ''there's nothing so practical
as a good theory''!
Validity:
A partner of ''reliability'', expressing the extent to
which a test, say, actually measures what it is supposed
to measure, i.e. does it do the job for which it was
designed? Various types of validity are looked for as
evidence of this.
Variable: In
the strictest sense, a variable is a symbol to which
a number is assigned. Constructs such as intelligence
are also referred to as variables. The terms ''factor''
and ''variables'' are sometimes used interchangeably.
Variables may be continuous (time, age) or dichotomous
(sex, marital status).
Recommended
Texts
Resources
|
