Data management

At the beginning of the analytic process, the researcher is faced by a mass of unwieldy,
tangled data and so the first task is to sort and reduce the data to make them more
manageable. This stage involves generating a set of themes and concepts according to which
the data are labelled, sorted and synthesised. Initially,these themes and concepts should
remain close to partici pants' own language and understandings, though later these may be
replaced by more abstract analytical constructions.Data management may be carried out
manually or, more commonly nowadays, using one of the many CAQDAS packages
available.
Descriptive accounts
Qualitative analysis asks such questions as: what kinds of things are going on here? What are
the forms of this phenomenon? What variations do we find in this phenomenon? That is,
qualitative analysis is addressed to the task of delin eating forms, kinds and types of social
phenomena; of documenting in loving detail the range of things that exist. (Lofland, 1971:
13)
Having generated and applied a set of themes and concepts at the data management stage, the
analyst makes use of the synthesised data to prepare descriptive accounts, identifying key
dimensions and mapping the range and diversity of each phenomenon. In this context, it is
important to empha sise two features of qualitative data which are central to descriptive
analy ses. The first is language - the actual words used by study participants. It is these that
portray how a phenomenon is conceived, how important it is and about the richness or
'colour' it holds. Second, the substantive content of people's accounts, in terms of both
descriptive coverage and assigned meaning, forms the nucleus of qualitative evidence. This
needs to be sensitively reviewed and captured so that the fineness of detail in different
perspectives or descriptions is understood.
Once the nature of phenomena have been clarified and the data classified according to a set of
substantive dimensions, refined categories or more abstract classes, the analyst may go on to
develop typologies. Typologies are specific forms of classification that help to describe and
explain the segmen tation of the social world or the way that phenomena can be characterised
or differentiated. They may apply to groups of people within the population or to sets of
phenomena like beliefs, circumstances or behaviours. Patton describes typologies as
'classification systems made up of categories that divide some aspect of the world into parts
along a continuum' (2002: 457).
Writers have distinguished between different forms of typologies. Patton (2002), for example,
differentiates between two forms that may arise from an analysis, which he terms 'indigenous'
and 'analyst constructed'. The former ANALYSIS: PRACTICES, PRINCIPLES AND
PROCESSES 215 are classification systems devised by participants themselves, for
example, terms such as 'chronics' or 'borderlines' which have been used by teachers to label
and distinguish between different kinds of truanting and lateness among pupils. The latter are
created during the analytical process, and clas sify patterns, categories or themes emerging
from the data. Lofland (1971) distinguishes between typologies based on static analysis (at a
particular time) or phase analysis (a process over time).
Explanatory accounts
Explanatory accounts tend to be developed at the later (or higher) stages of analysis when
most of the descriptive and typological work has already been undertaken. In order to move
from descriptive to explanatory accounts, the analyst will usually try to find patterns of
association within the data and then attempt to account for why those patterns occur. Again,
there are different ways in which linkages may be found. There will be explicit associa tions
that occur in the text or notes; linkages between sets of phenomena; and associations between
experiences, behaviours and perspectives and certain characteristics of the study population.
The ability to explain, or build explanations, lies at the heart of qualitative research. Most
qualitative data sets are rich in the levels of explanation theycan offer. They enable the
analyst to explain why the data take the forms that have been identified, to account for why
patterns, recurrent linkages, processes or apparent contradictions are found in the data.
When people explain their behaviour overtly or when a researcher infers an explanation,
however, what is the status of the explanation that is offered? While some qualitative
researchers seek explanations in terms of universal deterministic causes, for example this was
an aim within analytic induction, others increasingly reject the possibility of identifying these
kinds of causes, arguing that the social world is not governed by laws in the way that the
physical world is thought to be. However, if human behaviour is not law-like, neither is it
chaotic; it displays regularities which can be identified through careful analysis. How can
these regularities be explained? Can any kind of causal explanations be developed in
qualitative research? What is meant by the idea of cause within a qualitative context?
Because qualitative research is particularly concerned with the way in which people
understand and give meaning to their social world, some writers believe that the concept of
cause is not necessarily helpful. For example, Hughes and Sharrock (1997) argue in favour of
explanations at the level of meaning rather than explanations at the level of cause (in the
narrow deter ministic sense). Giving the example of traffic behaviour in the vicinity of traf fic
lights, they argue that an explanation of this behaviour can be developed by understanding
the meaning the lights have within a particular setting, group or culture, rather than by
attempting to specify the necessary and sufficient conditions and causal mechanisms which
produce a given pattern.
Other qualitative writers, for example Patton (2002), suggest that causal explanations may be
developed within qualitative research but use the term cause in a loose sense to refer to
conjectures, rather than narrowly deterministic laws. They claim that qualitative researchers
use a different kind of logic: rather than specifying isolated variables which are mechanically
linked, in qualitative analysis the analyst tries to build an explanation based on the way in
which different meanings and understandings within a situa tion come together to influence
the outcome.
Some writers maintain that universal deterministic causes are not achiev able in either
qualitative or quantitative social research. Giddens (1984) argues that causes may be sought
for social phenomena, and that reasons are causes, but not in a Humean (X always follows Y)
sense. He distinguishes between doing things for a reason, where the actor has an
'understanding of "what is called" for in a given set of circumstances in such a way as to
shape whatever is done in those circumstances' (1984: 345), and reasons for things happening
which may include a range of situational factors over which the actor has no control.
According to Giddens, social research will never be able to specify invariant, deterministic
causal relations because any causal mech anism will be inherently unstable: people differ in
their ability to make things happen and actions often have unpredictable and unintended
consequences.
It is our view that qualitative explanations attempt to say why patterns and outcomes in the
data have occurred. These explanations may use a causal logic in a loose, non-universal, non-
deterministic sense, but the logic is not based on linear variable analysis. They rarely cite a
single cause or reason, but set out to clarify the nature and interrelationship of different con
tributory factors or influences - such as personal intentions, patterns of understanding, norms
and situational influences. Sometimes explanations will be offered with some certainty
because of the strength of the evidential base. Other times they will be suggested as
hypotheses which need to be tested in further research. But essentially the 'building blocks' of
qualitative explanations need to be made clear so that others can view the sources and logic
of the construction, judging for themselves the 'validity' or 'credibility' of the findings (see
Chapter 10).
In the following chapter we describe how all the different steps involved in data management
and descriptive and explanatory analyses are carried out in practice. But it is important to
reiterate here that qualitative analysis, albeit exciting, is not without its challenges. Popper
once likened theory development to 'building on piles driven into a bottomless bog' (cited in
Campbell, 1977) and such a description could well be applied to the process of qualitative
data analysis. It is therefore important to have a strong analytic structure within which to
carry out all the investigative and creative tasks that are required. With this, there is some
hope that what initially appears to be a muddy field or 'bog' will begin to transform into firm
pasture with clear borders, landscape and rich colour.
• There are many different traditions and approaches for analysing qualitative data which
vary with epistemological assumptions about the nature of qualitative enquiry, the status of
researchers' accounts and the main focus and aims of the analytic process.
• There are a number of different 'tools' available for analysing quali tative data. Originally
these were manual methods but the latter part of the twentieth century saw a rapid growth in
computerassisted qualitative data analysis software (CAQDAS). These offer a variety of
functions. There is strong advice that these should be seen only as an 'analytic support' to aid
the process of analysis and not as a replacement for the intellectual role that is required of the
researcher.
• Whatever method of analysis is used, it needs to offer certain facili ties to the researcher in
order to maximise the potential for a full and reflective analysis. These include features that
facilitate and display ordering; permit within and between case searches; allow flexibility and
transparency to others; and allow emergent ideas, concepts, patterns etc. to remain rooted
within the original data.
• Analysis requires an organisational and conceptual structure that allows the researcher to
gain an overview of the data, carry out dif ferent analytic tasks and make sense of the
evidence collected. This structure needs to permit continuing and iterative movement
between the original data and the conceptualisation, abstraction and interpretation derived
from them.

The analytic hierarchy refers to the process through which qualita tive 'findings' are built
from the original raw data. It is described as a form of conceptual scaffolding within which
the structure of the analy sis is formed. The process is iterative and thus constant movement
up and down the hierarchy is needed.
The analytic process requires three forms of activity: data management in which the raw
data are reviewed, labelled, sorted and synthesised; descriptive accounts in which the
analyst makes use of the ordered data to identifying key dimensions, map the range and
diver sity of each phenomenon and develop classifications and typologies; and explanatory
accounts in which the analyst builds explanations about why the data take the forms that are
found and presented.
Further reading
Bryman, A. and Burgess, R. (eds) (1994) Analyzing Qualitative Data, London: Routledge
Coffey, A. and Atkinson, P. (1996) Making Sense of Qualitative Data, Thousand Oaks, CA:
Sage
Fielding, N.G. and Lee, R.M. (1998) Computer Analysis and Qualitative Research, London,
Sage Seale, C. (ed.) (2000) Researching Society and Culture, London: Sage Strauss, A.L.
and Corbin, J. (1998) Basics of Qualitative Research: Grounded
Theory Procedures and Techniques, 2nd edition, Thousand Oaks, CA: Sage Weitzman, E.A.
(2000) 'Software and qualitative research' in N.K. Denzin and Y.S. Lincoln (eds) Handbook
of Qualitative Research, 2nd edition, Thousand Oaks, CA: Sage