Education and debate

Qualitative research in health care

Analysing qualitative data
Catherine Pope, Sue Ziebland, Nicholas Mays

This is the Contrary to popular perception, qualitative research

second in a can produce vast amounts of data. These may include Summary points
series of three verbatim notes or transcribed recordings of interviews
articles or focus groups, jotted notes and more detailed “field-
Qualitative research produces large amounts of
notes” of observational research, a diary or chronologi-
textual data in the form of transcripts and
cal account, and the researcher’s reflective notes made
Department of observational fieldnotes
Social Medicine, during the research. These data are not necessarily
University of small scale: transcribing a typical single interview takes
Bristol, Bristol The systematic and rigorous preparation and
BS8 2PR several hours and can generate 20-40 pages of single analysis of these data is time consuming and
Catherine Pope spaced text. Transcripts and notes are the raw data of labour intensive
lecturer in medical the research. They provide a descriptive record of the
sociology
research, but they cannot provide explanations. The Data analysis often takes place alongside data
ICRF General researcher has to make sense of the data by sifting and
Practice Research collection to allow questions to be refined and
Group, University interpreting them. new avenues of inquiry to develop
of Oxford, Institute
of Health Sciences,
Textual data are typically explored inductively
Oxford OX3 7LS Relation between analysis and qualitative
Sue Ziebland using content analysis to generate categories and
senior research fellow
data explanations; software packages can help with
Social Policy In much qualitative research the analytical process analysis but should not be viewed as short cuts to
Branch, The begins during data collection as the data already gath- rigorous and systematic analysis
Treasury, PO Box
3724, Wellington, ered are analysed and shape the ongoing data
New Zealand collection. This sequential analysis1 or interim analysis2 High quality analysis of qualitative data depends
Nicholas Mays has the advantage of allowing the researcher to go back on the skill, vision, and integrity of the researcher;
health adviser it should not be left to the novice
and refine questions, develop hypotheses, and pursue
Correspondence to: emerging avenues of inquiry in further depth.
C Pope
c.pope@bristol. Crucially, it also enables the researcher to look for
ac.uk deviant or negative cases; that is, examples of talk or described below. The term grounded theory is used to
Series editors: events that run counter to the emerging propositions describe the inductive process of identifying analytical
Catherine Pope and or hypotheses and can be used to refine them. Such categories as they emerge from the data (developing
Nicholas Mays
continuous analysis is almost inevitable in qualitative hypotheses from the ground or research field upwards
research: because the researcher is “in the field” collect- rather defining them a priori).3 Initially the data are
BMJ 2000;320:114–6
ing the data, it is impossible not to start thinking about read and reread to identify and index themes and cat-
what is being heard and seen. egories: these may centre on particular phrases,
incidents, or types of behaviour. Sometimes interesting
or unfamiliar terms used by the group studied can
The analysis form the basis of analytical categories. Becker and
None the less there is still much analytical work to do Geer’s classic study of medical training uncovered the
once the researcher has left the field. Textual data (in specialist use of the term “crock” to denote patients
the form of fieldnotes or transcripts) are explored who were seen as less worthwhile to treat by medical
using some variant of content analysis. In general, staff and students.4
qualitative research does not seek to quantify data. All the data relevant to each category are identified
Qualitative sampling strategies do not aim to identify a and examined using a process called constant
statistically representative set of respondents, so comparison, in which each item is checked or
expressing results in relative frequencies may be compared with the rest of the data to establish analyti-
misleading. Simple counts are sometimes used and cal categories. This requires a coherent and systematic
may provide a useful summary of some aspects of the approach. The key point about this process is that it is
analysis. In most qualitative analyses the data are inclusive; categories are added to reflect as many of the
preserved in their textual form and “indexed” to nuances in the data as possible, rather than reducing
generate or develop analytical categories and theoreti- the data to a few numerical codes. Sections of the
cal explanations. data—such as discrete incidents—will typically include
Qualitative research uses analytical categories to multiple themes, so it is important to have some system
describe and explain social phenomena. These catego- of cross indexing to deal with this. A number of
ries may be derived inductively—that is, obtained computer software packages have been developed to
gradually from the data—or used deductively, either at assist with this process (see below).
the beginning or part way through the analysis as a way Indexing the data creates a large number of “fuzzy
of approaching the data. Deductive analysis is less categories” or units.5 Informed by the analytical and
common in qualitative research but is increasingly theoretical ideas developed during the research, these
being used, for example in the “framework approach” categories are further refined and reduced in number

114 BMJ VOLUME 320 8 JANUARY 2000 www.bmj.com

 Education and debate

cumbersome dataset. The sample size should be

directed by the research question and analytical require-
ments, such as data saturation, rather than by the
available software. In some circumstances, a single case
study design may be the most successful way of generat-
ing theory. Furthermore, using a computer package may
not make the analysis less time consuming,10 although it
may show that the process is systematic.

Taking the analysis forward—the role of

the researcher
A computer package may be a useful aid when gather-
ing, organising, and reorganising data and helping to
find exceptions, but no package is capable of
perceiving a link between theory and data or defining
an appropriate structure for the analysis. To take the
analysis beyond the most basic descriptive and

LIANE PAYNE
counting exercise requires the researcher’s analytical
skills in moving towards hypotheses or propositions
about the data.
One way of performing this next stage is called
analytic induction. This involves an iterative testing and
by grouping them together. It is then possible to select retesting of theoretical ideas using the data. Bloor
key themes or categories for further investigation— described his use of this procedure in some detail
typically by “cutting and pasting”—that is, selecting sec- (box).11 In essence, the researcher examines a set of
tions of data on like or related themes and putting cases, develops hypotheses or constructs, and examines
them together. Paper systems for this (using multiple further cases to test these propositions.
photocopies, cardex systems, matrices, or spread-
sheets), although considered somewhat old fashioned Inter-rater reliability
and laborious, can help the researcher to develop an Some researchers have found that the use of more
intimate knowledge of the data. Word processors can than one analyst can improve the consistency or
also facilitate data searching, and split screen functions reliability of analyses.5 12 13 However, the appropriate-
make this a particularly appealing method for sorting ness of the concept of inter-rater reliability in qualita-
and copying data into separate files. tive research is contested.14 None the less there may be
merit in involving more than one analyst in situations
Software packages designed to handle where researcher bias is especially likely to be
qualitative data perceived to be a problem—for example, where social
scientists are investigating the work of clinicians. In a
Several software packages designed for qualitative data study of diagnosis in cardiology, Daly et al developed a
analysis enable complex organisation and retrieval of modified form of qualitative analysis involving
data. Among the most widely used are qsr nud*ist external researchers and the cardiologists who had
and atlas.ti.6 7 This evolution has been welcomed as an managed the patients. The researchers identified the
important development with the potential to improve
the rigour of analysis.8 Such software can allow basic
“code and retrieval” of data, and more sophisticated
Analysis
analysis using algorithms to identify co-occurring
codes in a range of logically overlapping or nesting Stages in the analysis of fieldnotes in a qualitative study of ear, nose, and
possibilities, annotation of the text, or the creation and throat surgeons’ disposal decisions for children referred for possible
amalgamation of codes. Some packages can be used to tonsillectomy and adenoidectomy (with examples)11:
make theoretical links or search for “disconfirming evi- (1) Provisional classification—for each surgeon all cases categorised
according to disposal category used (tonsillectomy and adenoidectomy or
dence” (for example, by using boolean operators such
adenoidectomy alone)
as “or,” “and,” “not”). The Hypersoft package uses
(2) Identification of features of provisional cases—common features of cases
“hyperlinks” to capture the conceptual links which are in each disposal category identified (most tonsillectomy and adenoidectomy
observed between sections of the data; this can protect cases found to have three main clinical signs)
the narrative structure of the data to avoid the problem (3) Scrutiny of deviant cases—include in (2) or modify (1) to accommodate
of decontextualisation or data fragmentation.9 deviant cases (tonsillectomy and adenoidectomy performed when only two
Using software to help with the more laborious side of three signs present)
of analysis has many potential benefits, but some caution (4) Identification of shared features of cases—features common to other
is advisable. The prospect of computer assisted analysis disposal categories (history of several episodes of tonsillitis)
may persuade researchers (or those who fund them) (5) Derivation of surgeons’ decision rules—from the features common to
that they can manage much larger amounts of data and cases (case history more important than physical examination)
increase the apparent “power” of their study. However, (6) Derivation of surgeons’ search procedures (for each decision rule)—the
qualitative studies are not designed to be representative particular clinical signs looded for by each surgeon
in terms of statistical generalisability, and they may gain Repeat steps (2) to (6) for each disposal category
little from an expanded sample size except a more

BMJ VOLUME 320 8 JANUARY 2000 www.bmj.com 115

Education and debate

accounts and observations of the people studied (that

Five stages of data analysis in the framework approach is, “grounded” and inductive), it starts deductively from
• Familiarisation—immersion in the raw data (or typically a pragmatic pre-set aims and objectives. The data collection tends
selection from the data) by listening to tapes, reading transcripts, studying to be more structured than would be the norm for
notes and so on, in order to list key ideas and recurrent themes much other qualitative research and the analytical
• Identifying a thematic framework—identifying all the key issues, concepts, process tends to be more explicit and more strongly
and themes by which the data can be examined and referenced. This is informed by a priori reasoning (box).6 The analysis is
carried out by drawing on a priori issues and questions derived from the designed so that it can be viewed and assessed by
aims and objectives of the study as well as issues raised by the respondents
people other than the primary analyst.
themselves and views or experiences that recur in the data. The end product
of this stage is a detailed index of the data, which labels the data into
manageable chunks for subsequent retrieval and exploration Conclusions
• Indexing—applying the thematic framework or index systematically to all
the data in textual form by annotating the transcripts with numerical codes Analysing qualitative data is not a simple or quick task.
from the index, usually supported by short text descriptors to elaborate the Done properly, it is systematic and rigorous, and there-
index heading. Single passages of text can often encompass a large number fore labour-intensive and time-consuming. Fielding
of different themes, each of which has to be recorded, usually in the margin contends that “good qualitative analysis is able to
of the transcript
document its claim to reflect some of the truth of a
• Charting—rearranging the data according to the appropriate part of the
phenomenon by reference to systematically gathered
thematic framework to which they relate, and forming charts. For example,
there is likely to be a chart for each key subject area or theme with entries data,” in contrast, “poor qualitative analysis is
for several respondents. Unlike simple cut and paste methods that group anecdotal, unreflective, descriptive without being
verbatim text, the charts contain distilled summaries of views and focused on a coherent line of inquiry.”16 At its heart,
experiences. Thus the charting process involves a considerable amount of good qualitative analysis relies on the skill, vision and
abstraction and synthesis integrity of the researcher doing that analysis, and as
• Mapping and interpretation—using the charts to define concepts, map the Dingwall et al have pointed out, this requires trained,
range and nature of phenomena, create typologies and find associations and, crucially, experienced researchers.17
between themes with a view to providing explanations for the findings. The
process of mapping and interpretation is influenced by the original
research objectives as well as by the themes that have emerged from the
data themselves Further reading
Bryman A, Burgess R. eds. Analysing qualitative data.
main aspects of the consultations that seemed to be London: Routledge, 1993
related to the use of echocardiography, and they Miles M, Huberman A. Qualitative data analysis.
London: Sage, 1984
developed criteria which other analysts could use to
assess the raw data. The cardiologists then independ-
ently assessed each case using the raw data in order to
produce an account of how and why a test was or was The views expressed in this paper are those of the authors and
do not necessarily reflect the views of the New Zealand Treasury,
not ordered and with what consequences. The assess- in the case of NM. The Treasury takes no responsibility for any
ments of the cardiologists and researchers were com- errors or omissions in, or for the correctness of the information
pared statistically and the level of agreement was contained in this article.
shown to be good. Where there was disagreement
between the original researchers’ analysis and that of 1 Becker HS. Sociological work. London: Allen Lane, 1971.
the cardiologist, a further researcher repeated the 2 Miles M, Huberman A. Qualitative data analysis. London: Sage, 1984.
analysis and any remaining discrepancies were 3 Glaser BG, Strauss AL. The discovery of grounded theory. Chicago: Aldine,
1967.
resolved by discussion between the researchers and 4 Becker HS, Geer B. Participant observation: the analysis of qualitative
the cardiologists. Although there was an element of field data. In: Burgess RG, ed. Field research: a sourcebook and field manual.
London: Allen and Unwin, 1982.
circularity in part of this lengthy process (in that the 5 Perry S. Living with multiple sclerosis. Aldershot: Avebury, 1994.
formal criteria used by the cardiologists were derived 6 Richards T, Richards L. QSR NUD*IST, version 3.0. London: Sage, 1994.
7 Muhr T. ATLAS.ti for Windows. Berlin: Scientific Software Development,
from the initial researchers’ analysis) and it involved 1997.
the derivation of quantitative gradings and statistical 8 Kelle U, ed. Computer-aided qualitative data analysis: theory, methods and
practice. London: Sage, 1995.
analysis of inter-rater agreement that are unusual in a 9 Dey I. Qualitative data analysis: a user friendly guide for social scientists.
qualitative study, this process meant that clinical critics London: Routledge, 1993.
10 Lee R, Fielding N. User’s experiences of qualitative data analysis software.
could not argue that the findings were simply based In: Kelle U, ed. Computer aided qualitative data analysis: theory, methods and
on the subjective judgments of an individual practice. London: Sage, 1995.
11 Bloor M. On the analysis of observational data: a discussion of the worth
researcher. and uses of inductive techniques and respondent validation. Sociology
1978;12:545-52.
12 Daly J, McDonald I, Willis E. Why don’t you ask them? A qualitative
Applied qualitative research research framework for investigating the diagnosis of cardiac normality.
In: Daly J, McDonald I, Willis E, eds. Researching health care: designs, dilem-
The framework approach has been developed in Brit- mas, disciplines. London: Routledge, 1992:189-206.
13 Waitzkin H. The politics of medical encounters. New Haven: Yale University
ain specifically for applied or policy relevant qualitative Press, 1991.
This article is taken research in which the objectives of the investigation are 14 Armstrong D, Gosling A, Weinman J, Marteau T. The place of inter-rater
from the second reliability in qualitative research: an empirical study. Sociology
typically set in advance and shaped by the information 1997;31:597-606.
edition of
Qualitative Research requirements of the funding body (for example, a 15 Ritchie J, Spencer L. Qualitative data analysis for applied policy research.
In Bryman A, Burgess R, eds. Analysing qualitative data. London:
in Health Care, health authority).15 The timescales of applied research Routledge, 1993:173-94.
edited by Catherine tend to be short and there is often a need to link the 16 Fielding N. Ethnography. In: Fielding N, ed. Researching social life. London:
Pope and Nicholas Sage, 1993:155-71.
Mays, published by analysis with quantitative findings. For these reasons, 17 Dingwall R, Murphy E, Watson P, Greatbatch D, Parker S. Catching gold-
BMJ Books although the framework approach reflects the original fish: quality in qualitative research. J Health Serv Res Policy 1998;3:167-72.

116 BMJ VOLUME 320 8 JANUARY 2000 www.bmj.com