4 Developing indicators for
concepts
To be useful, concepts must have empirical indicators: if we cannot
say what behaviours, attitudes or characteristics reflect conservatism,
authoritarianism or social status, for example, then for the purposes
of research the concept is useless. The difficulty is in developing good
indicators for concepts. This chapter examines how to translate
concepts into indicators by looking at three steps:
1 clarifying the concepts;
2 developing initial indicators;
3 evaluating the indicators.
i
Before looking at these steps it is helpful to look at an example
which illustrates the process of developing indicators and highlights
the difficulties. Suppose we are interested in the theory which argues
that religiousness is a response to deprivation: that religious faith
serves to compensate people for their frustrations and disappoint-
ments. We might propose that the more deprived people are, the
more religious they will be. To test this we must work out who is
deprived and who is not, who is religious and who is not. We might
use income to distinguish between the deprived and non-deprived:
I
those earning $10000 or less a year being classified as deprived and
those earning $20000 a year as being non-deprived. We could use
church attendance to indicate religiousness- monthly and more
regular attenders could be defined as religious with less regular
attenders being classified as non-religious. Suppose, contrary to
expectations, we find only 15 per cent of those with low incomes
attend church regularly, while 50 per cent of those with higher
incomes do so. Can we then reject the theory that deprivation leads
to religiousness? There would be two fundamental problems with this
1 'research'.
1 We have not clarified the meaning of 'deprivation' and 'religious-
ness'. Unless we are clear about the meaning of the concepts we
cannot develop measures of them.
47
Určeno pouze pro studijní účely
48 Surveys in social research
2 We do not know whether the indicators we have used are
adequate. Does church attendance adequately measure religious-
ness? Does income indicate deprivation? Before answering these
questions we need to know what the concepts mean.
Clarifying concepts
Concepts are simply tools which fulfil a useful shorthand function:
they are abstract summaries of a whole set of behaviours, attitudes
and characteristics which we see as having something in common.
Concepts do not have some sort of independent existence 'out there':
they do not have any fixed meaning. Concepts are terms which
people create for the purpose of communication and efficiency.
When developing indicators for concepts, the task is not to find
indicators which match some concept which has a set definition. It is
up to us to first define what we mean by the concept and then develop
indicators for the concept as it has been defined. By their very nature
definitions are neither true nor false: they are only more useful or less
useful.
There is a problem here. If concepts have no set meaning then
anyone can define a concept any way they wish. The result would be
that the concept would become useless; unless people mean the same
thing by a word, communication is impossible. In sociology lack of
agreement about how words are defined leads to confusion and
pointless debates. For example, debates about the extent to which a
country is secularised, equal or democratic depend substantially on
definitions of religion, equality and democracy respectively.
The view that concepts do not have real or set meanings can lead to
conceptual anarchy, a problem with no entirely satisfactory solution.
The most practical action is to make it very clear how we have defined
a concept and to keep this definition clearly in mind when drawing
conclusions and comparing the findings with those of other research-
ers. Although we can define a word any way we wish, there seems to
be little value in developing entirely idiosyncratic definitions. Since
concepts are used to communicate, it makes most sense to use the
word in its most commonly understood sense. If the definition of the
concept is idiosyncratic this should be made very clear. Where a
concept takes on a number of widely held but different meanings, we
will need either to decide on (and justify) one, or to design the
research so that we have indicators of each of the different meanings.
Developing indicators for concepts 49
ow to clarify concepts
ce concepts have no set meanings yet it is crucial that the concepts
d in research be defined, how do we go about clarifying them? In
ice people use different approaches. I will describe three steps
help in the process.
Obtain a range of definitions of the concept: Before adopting one
definition of a concept look for the ways in which sociologists use
the concept then look at lay definitions. People do not always
provide formal definitions so we may need to work out what they
mean by the way they have used the term (i.e. their definition may
be implicit rather than explicit).
Once we have an idea of the range of ways in which the term is
used, we may find that we can classify definitions into a number of
categories. Developing a definition of religion can serve as an
example. There are many sociological and lay definitions but they
can be grouped into two main categories:
a Inclusive or functional definitions: this approach, which de-
rives from Durkheimian and functionalist theorising, defines
beliefs and behaviour as religious, not by their content, but by
their function for either individuals or society. Thus any set of
beliefs which provides people with meaning in life may be
defined as religious. There are a large number of diverse
definitions which fit into this category.
b Exclusive or substantive definitions: these definitions are
based on the content of belief and typically specify that the
beliefs must include some notion of a supernatural being.
An alternative approach is to look at common elements of
definitions and develop a definition based on these. Hillery (1955)
listed 94 definitions of 'community' and Bell and Newby (1971)
note that the majority of definitions include three elements: area,
common ties and social interaction. This approach could then
form the basis of a definition which incorporates the generally
understood meaning of the concept.
2 Decide on a definition: Having listed types of definitions or
delineated the most common elements of definitions, we need to
decide on which definition to use. We might opt for an existing
one, create a new one, choose a classic definition or use a more
contemporary one. Regardless of which we do, we need to justify
the decision.
In practice, the process of conceptual clarification continues as
data are analysed. Clarification is not a once-and-for-all process
which precedes research. It is an ongoing process: there is an
Určeno pouze pro studijní účely
50 Surveys in social research
interaction between analysing data and clarifying concepts. As a
result of analysing data we are often in a better position to say
what we mean by a concept than before we began. Nevertheless,
this process must begin before data collection.
To assign a definition to a concept is to give it a nominal
definition: it is a working definition which is used in the research.
It provides a focus for research and guidance about the type
of information to collect, but does not tell us precisely which
information to collect. For example, we might define religious
beliefs as those with a supernatural element. This helps focus on
the range of beliefs to examine but does not specify which beliefs
to examine. This is the task of an operational definition which will
be dealt with shortly.
3 Delineate the dimerzsiorzs of the concept: Many concepts have a
number of different aspects or dimensions. When clarifying
concepts it is often helpful to distinguish between those dimen-
sions. This may result in using only one of the dimensions in the
study or it may lead to a more systematic development of
indicators for each dimension. Distinguishing between dimen-
sions can lead to more sophisticated theorising and more useful
analysis.
Deprivation is an example of a multidimensional concept:
social, economic, political, psychic or physical deprivation can be
delineated. Distinguishing between these dimensions can force us
to clarify what our theory is about and in so doing ensure that we
develop measures relevant to that theory. Earlier we noted the
model which says deprivation causes religiousness. By delineating
dimensions we are forced to ask, do we mean any sort of
deprivation will lead to any sort of religiousness? Perhaps we
might become more specific and state that social deprivation leads
to religious beliefs (one of the dimensions of religiousness). If we
decided this was the theory we were testing, we would need to
include questions relating only to one type of deprivation and one
aspect of religiousness.
We might want to develop measures of each type of deprivation
and each aspect of religiousness. If so, delineating the separate
dimensions helps in choosing indicators systematically.
Developing indicators
The process of moving from abstract concepts to the point where we
can develop questionnaire items to tap the concept is called 'descend-
ing the ladder of abstraction'. It involves moving from the broad to
Developing indicators for concepts 51
the specific, from the abstract to the concrete. In clarifying concepts
we begin to descend this ladder. A further step is taken when
dimensions are specified. Sometimes these dimensions themselves
can be further subdivided into some more specific categories. For
example, there may be different types or aspects of social depriva-
tion: it may involve social isolation or it may mean the absence of
socially valued roles or social skills. These dimensions of social
deprivation are more specific and give more clues about which
questions to ask in a questionnaire (Figure 4.1).
When delineating dimensions and dimensions of dimensions it is
helpful to define the terms on the way. If one aspect of social
deprivation is social isolation, what do we mean by this? Does it
mean lack of friends, not seeing your family, not belonging to
organisations and so on? Before concepts can be measured we must
descend from the lofty and often vague heights of some theories and
deal with these more mundane issues. The process of descending the
ladder of abstraction is summarised in Figure 4.1.
When we get to the point where we can develop indicators there
are three broad problems to deal with:
1 how many indicators to use;
2 how to develop the indicators;
3 how to form items into a questionnaire (this will be dealt with in
Chapter 6).
4
How many irzdicators to use
There is no simple answer to this problem but the following guide-
lines are useful.
1 When there is no agreed way of measuring a concept it may be
helpful to develop indicators for a range of definitions and see
what difference this makes to the results and interpretations.
2 If the concept is multidimensional, consider whether you are
really interested in all dimensions. Are they all relevant to the
theory?
3 Ensure that the key concepts are thoroughly measured. The
behaviour and attitudes that we are trying to explain and the
theorised causes must be carefully measured using several
indicators.
4 Typically attitudes and opinions are complex and are best mea-
sured with a number of questions to capture the scope of the
concept (see Chapter 15).
5 Pilot testing indicators is a way of eliminating unnecessary
Určeno pouze pro studijní účely
Developing indicatorsfor concepts 53
questions. Initially we might have 50 questions to measure
authoritarianism but find that we need only 10 of these items:
c ht not add anything to our index (see
The number of items is affected by practical considerations such
as overall length of the questionnaire and method of administra-
tion (see Chapter 7).
ing an indicator is simple and the
(e.g. marital status, education level),
more abstract concepts, it is more
ree main approaches to developing
nnaire research. In each case these initial
uring the evaluation stage.
in previous research can be used. There
many well-established and tested scales which we ignore at our
il. Ideally we should try to evaluate these measures: they may
ng to fit a particular context or a
11-established indicators has the enorm-
mparison of results with those of other
ful in building up a cumulative body of
n carrying out their own idiosyncra-
easures (see references at the end of
h topics, especially those where we are
. migrants, Aborigines, young people,
childless couples) it is very helpful to use a less structured approach to
data collection first (e.g. observation, unstructured interview). This
can help us understand things through the eyes of these people, learn
of their concerns and ways of thinking, and this can be extremely
nd appropriately worded questions
A third alternative is to use 'informants' from the group to be
surveyed. Such people can provide useful clues about meaningful
questions. For example, if surveying a union it would be helpful to
talk to key types of people in the union to get their ideas and
comments on questions.
In the end we have to decide which indicators to use and how
to word them. In doing so we need to be as informed as possible
about the study population, to be clear about what we want to
measure, to look at other people's efforts and to evaluate our own
Určeno pouze pro studijní účely