Trend report

Etat de la question

Mira Crouch and Heather McKenzie

The logic of small samples in interview-based

qualitative research

Abstract. In a qualitative framework, research based on interviews often seeks to

penetrate social life beyond appearance and manifest meanings. This requires the
researcher to be immersed in the research field, to establish continuing, fruitful
relationships with respondents and through theoretical contemplation to address the
research problem in depth. Therefore a small number of cases (less than 20, say) will
facilitate the researcher’s close association with the respondents, and enhance the
validity of fine-grained, in-depth inquiry in naturalistic settings. Epistemologically
prior to these considerations, however, is the explanatory status of such research.
From a realist standpoint, here concept formation through induction and analysis
aims to clarify the nature of some specific situations in the social world, to discover
what features there are in them and to account, however partially, for those features
being as they are. Since such a research project scrutinizes the dynamic qualities of a
situation (rather than elucidating the proportionate relationships among its
constituents), the issue of sample size – as well as representativeness – has little
bearing on the project’s basic logic. This article presents this argument in detail, with
an example drawn from a study of persons with a past history of cancer diagnosis and
treatment.

Key words. Analytic induction – Exploratory research – Interviews – Qualitative

methodology – Realism – Sample size

Résumé. Dans un cadre qualitatif, la recherche basée sur des entretiens s’efforce
souvent de décrypter la vie sociale au delà des apparences et des significations évidentes.
Le chercheur se doit alors de s’immerger dans son domaine de recherche, d’établir des
relations suivies et riches avec ses répondants et d’appréhender le problème de recherche
en ayant recours à l’investigation théorique. Il s’ensuit qu’un nombre de cas limité

Social Science Information & 2006 SAGE Publications (London, Thousand Oaks, CA and New
Delhi), 0539-0184
DOI: 10.1177/0539018406069584 Vol 45(4), pp. 483–499; 069584
484 Social Science Information Vol 45 – no 4

(moins d’une vingtaine par exemple) facilitera la proximité de la relation. Etant

donné qu’une telle approche de recherche scrute la dynamique d’une situation (plutôt
que d’élucider les relations entre ses éléments constitutifs), le problème de la taille de
l’échantillon – et de sa représentativité – n’a que très peu d’impact sur la logique
fondamentale du projet. Cet article présente cette argumentation en détail en
s’appuyant sur l’exemple d’une étude qui porte sur des individus ayant un passé de
diagnostic et de traitement du cancer.

Mots-clés. Entretiens – Méthodologie qualitative – Pensée réaliste – Raisonnement

analytique – Recherches exploratoires – Taille de l’échantillon

Against the background of increased interest in qualitative methods

(Flick, 2002), interview-based studies involving a small number of
respondents are becoming more common in social science. While
proponents of this research mode may take its value for granted,
often such studies have limited acceptability in the broader research
community, since it appears to be widely held that only sizeable
samples can hold promise of validity. Justification of small-sample
studies hinges most frequently on phenomenological assumptions
(broadly speaking) which underwrite investigations of personal
experience in a largely subjectivist framework. From a more empiri-
cal perspective, the labour-intensive nature of research focused on
depth (including, sometimes, ‘‘reflexivity’’) can be evoked to justify
a small sample size. By contrast to these positions, this article argues
the positive case for small samples in research where in-depth inter-
viewing is the method of choice and realism the epistemological
foundation. For this argument to make sense, some points about
‘‘interviewing’’ and the analysis/interpretation of interview material
need to be elucidated first.

Telling tales, or saying how it is?

Interviewing is one of the most frequently employed qualitative

methods. Indeed, the term ‘‘qualitative methods’’ commonly denotes
data collection techniques based on various types of conversations
between researchers and respondents. Of these, the one-to-one inter-
view is arguably the primary form; it is used in many research settings
and can be quite variable in style (including such things as duration,
Crouch and McKenzie Trend report 485

the role of the interviewer and the degree of ‘‘structure’’ of the con-
versation which constitutes it).
This article is concerned with interviewing of the in-depth, rela-
tively free-flowing (though still focused) kind, mostly employed in
investigations of subjective feelings and reactions, and commonly in
relation to ‘‘sensitive’’ topics (Renzetti and Lee, 1993). Respondents
in such studies are usually selected on the basis of being, or having
been, in an unsettling or disturbing situation – for instance, a diag-
nosis and subsequent treatment for cancer. (The argument presented
in this article is an explication of the methodology of our own
research into this particular problem [Crouch and McKenzie, 2000;
McKenzie and Crouch, 2004].)
Many researchers working with in-depth interviewing think of
interview protocols as ‘‘constructed’’ versions of the social world,
sometimes even intersubjectively cobbled together in the very course
of the interview (see, for example, Gubrium and Holstein, 1999).
The radical extreme of this position (its logical conclusion, at any
rate) represents a denial of the possibility of deriving any objective
knowledge from the interview material. There are inherent contra-
dictions in any statements putting forward the ‘‘hard’’ constructivist
position,1 but even setting these aside it is possible to advance a
different view of the epistemological status of in-depth interview
protocols.
Our position shares with radical constructivism the understanding
of the purpose of in-depth interviews. By definition, these interviews
target the respondents’ perceptions and feelings rather than the
social conditions surrounding those experiences; at least, the collec-
tion of the interview material and its interpretation and analysis are
not primarily directed towards establishing ‘‘objective facts’’ con-
cerning these conditions. Thus for us the primary aim of in-depth
interviewing is to ‘‘generate data which give an authentic insight
into people’s experiences’’ (Silverman, 1993: 91). From a realist per-
spective, here the term ‘‘authentic’’ implies that the respondents’
point of view (as lodged in the interview protocols) can be granted
the ‘‘culturally honoured status of ‘reality’’’ (Miller and Glassner,
1997: 99).
However, in order fully to understand that reality, we need to take
into account its social context. This means, in effect, that the inter-
view material is ultimately comprehended within a frame of a situa-
tion assumed to exist independently from experience. For this reason,
it is essential that this situation – theoretically elaborated, of course –
486 Social Science Information Vol 45 – no 4

be taken into account in interpretation and analysis. Most crucially,

therefore, we see this process as a realist attempt to formulate socio-
logically both the subjective and the social meanings in the respon-
dents’ accounts. In other words, not only do we accept these accounts
as facets of the respondents’ experience overall, but, equally impor-
tantly, we understand that this experience is imbricated in a social
milieu which is ontologically prior to both the respondents’ and
the interviewers’ actions, and therefore causally related to them.
From this perspective, both ‘‘researchers’’ and ‘‘subjects’’ share
the same epistemological position: in both cases, their ‘‘conscious-
ness is neither passive nor purely or irresponsibly active; rather, it
is reactive in the sense that it reacts creatively to the possibilities of
the context’’ (Levy, 1981: 19). This position is consistent, at least in
research practice, with the ‘‘middle way’’ (between objectivism and
relativism, and similar dualities) model for social science inquiry.
Here it is accepted that ‘‘although we (researchers) always perceive
the world from a particular standpoint, the world acts back on us
to constrain the points of view that are possible’’ (Seale, 1999: 27).

Engaging conversations

Interviewing is an obtrusive method; that is to say, it elicits the

statements to be analysed. This is so even where respondents are
encouraged freely to ‘‘tell their story’’, because that story might
not have been told, or told in a particular way, without the inter-
viewer’s intervention, both through the initial prompt and sub-
sequently throughout the course of the interview. It is for this
reason that the literature on interviewing contains extensive discus-
sions of problems associated with the interactive aspect of the inter-
viewing process. Most commonly mentioned are interviewer bias,
variability of rapport and, especially from an empiricist perspective,
validity issues relating to the interpretation of the interview material.
(References for these points are too numerous to mention; most of
the important arguments are summarized in Minichiello et al., 1995.)
However, there are also – much less frequently mentioned –
advantages in spontaneous interviewer–interviewee interaction.
(While these may be taken for granted in a constructivist framework,
it is important here to consider them from a realist point of view.)
In an in-depth interview, the researcher’s discretion with respect to
the conduct of the interview is part of an open-ended mode of
Crouch and McKenzie Trend report 487

inquiry which can produce great richness of material if the researcher

is responsive to cues as they occur in the course of the interview.
Furthermore, without the constraint of a pre-determined grid of
specific questions or issues to be discussed, the very scope of the
inquiry can broaden or even shift in response to the emergent inter-
view material. As Van Maanen (1983) has pointed out, the role of
surprise in research cannot be ignored. Important, too, is the
researcher’s primed sensitivity as an appropriately schooled social
scientist – and as a member of society with a stock of personal
knowledge which can be an invaluable research resource (Phillips,
1971: 158–60). (Outside the scope of this article, but well worth
noting, are the psychological benefits respondents may obtain from
the opportunity to ‘‘tell their story’’ to a most attentive listener.)
None of this need imply that interview protocols thus obtained
lack veracity. On the contrary, we propose that complex reactions
and feelings are best given meaning and are optimally articulated
– to the respondents’ satisfaction (i.e. their sense of ‘‘closure’’) –
through a dialogue which encourages reflection on, rather than
mere reporting of, experience. This is particularly the case where
experiences under investigation involve pain and suffering, expres-
sions of which are largely suppressed in everyday life, however
avidly we may consume them in their ‘‘mediatised’’ form (Kleinman
and Kleinman, 1996). The dialogue about experience can, of course,
be an internal one, with oneself, as it were; but in the reflective inter-
view a skilled and, most importantly, theoretically as well as emo-
tionally informed interviewer is invaluable as a sounding board
and an aide to veridical expression of thoughts and feelings.2

The knowledge we seek

It has already been suggested that in-depth interviews can generate

new knowledge, or at least understanding, for the respondents. In
addition to improved self-understanding, this knowledge may also
include better insight into the social conditions of their lives. As
researchers, we too seek insight into those conditions as they are
reflected in our respondents’ experience, but as investigators we
need an incisive view of the terms in which our new knowledge is
to be framed, which in turn depends on our understanding of the
relationship between interviewing and our particular research ques-
tion. In other words, it is necessary to have clarity with respect to the
488 Social Science Information Vol 45 – no 4

theoretical grounds on which we choose such a research strategy. At

stake here is the methodology of in-depth interviewing.
To consider this matter briefly: in-depth interviews might con-
ceivably be used in a hypothesis-testing study, though in most
cases the open-endedness and fuzzy validity (on empiricist criteria)
of unstructured in-depth interviewing would undermine this strategy.
For similar reasons, the method would not always suit research
aimed at identifying attitudes and/or behaviours of particular
groups of individuals with a view to establishing patterned relation-
ships among variables uncovered in the research. Nonetheless, the
in-depth unstructured interview can be – and has been – employed
in this way (e.g. in studies by Wiener [1975] and Oakley [1980]),
usually as part of an overall plan which rests on purposive selection
of respondents and leads to a structured analysis which transforms
interview protocols into chunks of data through coding and count-
ing, and produces results which show cross-references among these
data categories. Most commonly, however, in-depth interviews
feature in research where respondents’ experience is analysed with
the uncovering of its thematic dimensions in view. Interpretations
of the interview yield are guided by various approaches. Commonly
employed is grounded theory, either in its flexible mode as originally
put forward by Glaser and Strauss (1967) or in the more structured
revision by Strauss and Corbin (1990). Phenomenology and its off-
shoots (most recent ones veering towards subjectivism) also feature
prominently. (For a comprehensive review of this genre of research,
see Crotty, 1996.)
This discussion implies an important point: data collection from
interviews is in principle both distinct and independent from analysis
and interpretation. The distinction between methodology and methods
is relevant here. Whereas the term ‘‘qualitative methods’’ refers to a
category of ways of collecting data, ‘‘qualitative methodology’’
denotes theoretically underpinned research strategies which are con-
ceptual and hermeneutic throughout the research process (Walker,
1985). It is with such an understanding of ‘‘qualitative methodo-
logy’’ that our article examines the place of small samples in research
based on in-depth interviews.
We argue that interview protocols in such research are best
analysed in ways which do not depend on delineated categories
and the numbers of ‘‘hits’’ in them, but rather on thematic strands
extracted from the material by dint of the researchers’ interpretive
and conceptual efforts. Thus we agree with Walker that qualitative
Crouch and McKenzie Trend report 489

methodology is aimed at finding out ‘‘what things ‘exist’ [rather]

than determin[ing] how many such things there are’’ (1985: 3). This
statement of course refers primarily not to the number of respondents
(or, more generally, ‘‘cases’’) but rather to dimensions and aspects of
the situation under investigation. That is to say, qualitative research
scrutinizes the dynamic qualities of a situation rather than its consti-
tuents and the proportionate relationships among them. ‘‘The situa-
tion’’ is for us a realist concept in that we hold that it exists
independently of us and our view of it (Archer, 1995). As realists,
we assume that we have at least partial access to social reality, and
that the validity of our knowledge is therefore linked to the way the
world is (Brante, 2001: 187).
An important feature of situations or, more precisely, states of
affairs (in the social as well as material/natural realms) is the sub-
merged nature of their crucial determining dimensions which often
cannot be directly observed. Since through our interviews we seek
to uncover – get to ‘‘know’’ – precisely these dimensions, the terms
of our knowledge have to be theoretically grounded; the status of
that knowledge is therefore explicitly provisional. We may indeed
be in error in some or all of what we conceptualize and put forward.
However, the possibility of erroneous conclusions is logically inde-
pendent from the objective existence of the state of affairs under
investigation, and does not in principle negate any knowledge we
may obtain of them.

How to ground interview data?

If, then, theory – presuppositions, propositions, etc. – is an inevit-

able aspect of knowledge sought from qualitative data, how does
it arise from in-depth interview material? Although in some respects
our view of analysis and interpretation echoes grounded theory in its
earlier formulations (Glaser and Strauss, 1967; but it is not in accord
with the later versions of this position [Strauss and Corbin, 1990]),
there is a fundamental difference. We hold that it is not possible
for data interpretation and theory development to be accomplished
– indeed, even attempted – purely ‘‘from the ground up’’.
Respondents’ accounts are located in domains of social life which
contain (in both senses of that term) the experiences under investiga-
tion. Themes and dimensions of these experiences – Walker’s (1985)
‘‘things’’ (see above) – must therefore be understood, sociologically
490 Social Science Information Vol 45 – no 4

speaking – and, especially, in realist terms – to have systematic con-

nections with sundry sociocultural circumstances. However subjec-
tive, these ‘‘things’’ are lodged in the fabric of respondents’ daily
lives which are, axiomatically for us, profoundly social, by which
we mean not merely ‘‘interactive’’ and ‘‘intersubjective’’, but also
inescapably shaped by impersonal, objective social conditions. To
give those experiences sociological meaning is to comprehend
them in the context of the social conditions within which they
arise; in other words, it is to attempt to explain those experiences
‘‘vertically’’ by addressing crucial questions about the necessary con-
ditions under which experience is possible at all (Williams, 1999:
808). As the ‘‘system elements’’ of social life are rarely more than
hinted at in respondents’ accounts, the emergent analysis and inter-
pretation of these accounts require reference to relevant social
circumstances to be grounded in extant sociological (including, of
course, theoretical) knowledge.
Hence the limitations of the grounded theory approach. Clearly
interpretation and analysis do not treat each respondent as an
unique case, self-contained in its own solipsistic bubble. But just
as soon as significant themes are identified, there is at least an impli-
cit assumption of a common social world in which these manifest
and make sense. In other words, respondents express social rela-
tions, even though they are not necessarily consciously aware of
them (Bhaskar, 1979). At the same time, the social context of the
respondents’ situation cannot be pieced together solely from
accounts of personal experiences which arise out of it: first, because
not all relevant social circumstances come to the fore in those
accounts (unless one falls into the teleological trap of asserting
that only what is expressed is relevant); and second, because agglom-
erations of individual cases, even with considerable commonalities
among them, cannot be taken as fully constituting structural dimen-
sions of social life.
When employed on empirically and theoretically well-trodden
ground, interview-based investigations can, and do, extend their
findings from the individual to the structural level (for example,
regarding family life, the classic studies by Willmott and Young,
1960; Bell, 1968; Oakley, 1974). The research strategy of such studies
usually includes the expectation that samples will be large enough to
support generalizations from research findings to a broader social
spectrum. While dangers of methodological individualism may lurk
in this strategy,3 a critique of it is not implied by the present discus-
Crouch and McKenzie Trend report 491

sion. Rather, our purpose is to make a distinction between, on the one

hand, research objectives which need to be mindful of sample size
and, on the other, exploratory, concept-generating studies for
which it is not only reasonable to have a relatively small number of
respondents, but may even be positively advantageous (for example,
research reported by Salander et al., 1996; Little et al., 1998; Crouch
and McKenzie, 2000). Indeed, Dreher holds that in inductive research
in naturalistic settings, small samples, which permit repeated contacts
with respondents and greater involvement of the investigator,
enhance validity and reliability (1994: 286).

The sociological imagination writ small

We can now bring together aspects of qualitative analysis of in-

depth interview material (as already discussed) with an exploratory
research strategy. First, such an analysis is informed by knowledge
and understanding of pertinent social circumstances – both theo-
retical and factual – from sources other than the interview material.
Second, it continuously carries out ‘‘recontextualisation’’ (Morse,
1994: 34) so that throughout the investigation each ‘‘case’’ is
attended to both in its own terms and in relation to other
‘‘cases’’ – as well as within a more general framework of disciplinary
knowledge (pre-existing and sourced during the research). Third, it
aims ultimately to establish, conceptually/theoretically, points of
contact (adhesion or friction, as the case may be) between individual
experience/action and the social context.4 In the course of such
research, then, in addition to the interview material and its extant
disciplinary/conceptual background, a new entity is enticed to come
to light as a third force in the proceedings – the emergent theoretical
frame (Layder, 1998: 170) which eventually envelops the ‘‘findings’’
of the research.
This research strategy harks back to some procedural precepts of
analytic induction (Robinson, 1951; Turner, 1953; Znaniecki, 1963;
Jones, 1979; Crouch and Manderson, 1993) where the investigator
re-searches the material within a spiral of conceptual development,
rather than across a plane of data. Rather than looking for enduring
structures in a particular field or system – ‘‘uniformities of coexis-
tence’’, as J. S. Mill would have put it (Michell, 2003: 531), the
researchers seeks to uncover dynamic patterns – Mill’s ‘‘uniformities
of succession’’. In the spiral of interpretation the conceptualizing
492 Social Science Information Vol 45 – no 4

movement oscillates between individual cases and sociocultural

circumstances in an effort to weave together these disparate strands
of the social fabric into a sociologically meaningful pattern. Perhaps,
then, this is an instance of C. Wright Mills’s ‘‘sociological imagina-
tion’’ at work, attempting to ‘‘grasp history and biography, and the
relation of the two within society’’ (Mills, 1959: 6). To refer more
explicitly to the structure/agency dialectic, through this spiral move-
ment the researcher endeavours to scale the slippery slope between
individual and collective levels of action (Crouch and Manderson,
1993: 188).
If, then, we accept that small-scale interview-based – and radically
qualitative – research is intentionally conceptually generative, its
findings (in the sense of last, rather than definitive) cannot escape
the charge of speculation. Nor should they have to do so. It is in
the nature of exploratory studies to indicate rather than conclude.
More strictly put, such studies formulate propositions rather than
set out to verify them – or, at least, convincingly demonstrate them
(through reliance, for example, on ‘‘representativeness’’ and the per-
suasive weight of large samples). It should be noted here that rigour
in procedure and argument is as crucially important for exploratory
research as it is for any other kind of investigation – perhaps even
more so, since it generally lacks the familiar and reassuring trap-
pings of research oriented towards verification. In any case, within
a realist framework at least, there is in principle little difference
between the hypothetical nature of concepts arising out of explora-
tory studies and results obtained from research that aspires to truth-
claiming conclusions. In science – using the term in its most general
sense – all knowledge is contingent on further developments, as it
is always a matter of provisional designation of symbol, myth or
theory (Levy, 1981), as the case may be.

Small is beautiful

The term ‘‘small sample size’’ has been used in this discussion as a
shorthand expression to denote a small number of respondents
(often less than 20). Strictly speaking, however, the whole notion
of ‘‘sample’’ is not appropriate here since in research of this kind
respondents are not drawn (i.e. sampled) from a ‘‘target popula-
tion’’. This is because a particular shared experience or life-situation
is hardly a sufficient basis on which to establish boundaries that
Crouch and McKenzie Trend report 493

define such a ‘‘group’’ of respondents (as a fraction of the hypo-

thetical larger group to which they may in some sense belong). In
the kind of research discussed here, if anything is being ‘‘sampled’’,
it is not so much individual persons ‘‘of a kind’’, but rather variants
of a particular social setting (the real object of the research in ques-
tion) and of the experiences arising in it. Rather than being system-
atically selected instances of specific categories of attitudes and
responses, here respondents embody and represent meaningful
experience–structure links. Put differently, our respondents are
‘‘cases’’, or instances of states, rather than (just) individuals who
are bearers of certain designated properties (or ‘‘variables’’). We
conceptualize these ‘‘cases’’ as states arising within a field of a par-
ticular set of circumstances, which casts them as ‘‘engaged in perpe-
tual dialogue with their environment’’, doing or enduring a variety
of things, ‘‘each of which may be seen as an event arising either in
agency (what they do) or in structure (what they endure)’’ (Abbott,
1992: 64–5).
The interaction of the doing with the enduring is the process under
scrutiny in small-sample research. Its approach is therefore clinical,
involving as it does careful history-taking, cross-case comparisons,
intuitive judgments and reference to extant theoretical knowledge.
This is not something that can reasonably be done with a large
number of cases. Affirmatively restated, such research positively
calls for a collection of respondents’ ‘‘states’’, the size of which
can be kept in the researcher’s mind as a totality under investigation
at all stages of the research. The implicit requirement that (in analy-
tic induction) every case should be taken into account puts the
emphasis not on the individuality of each, but ‘‘on the unimportance
of the number of cases to theoretical explanation’’ (Platt, 1992: 24;
emphasis added).
As already outlined above, this mode of research rests on the
continuing monitoring of the interview material in relation to theo-
retical developments. The labour involved in conceptualization is an
integral part of the research, rather than a ‘‘framing’’ or ‘‘contextua-
lization’’ of it. Arguably, therefore, it is its most ‘‘qualitative’’ fea-
ture and, as such, its most important aspect as it simultaneously
deepens understanding and builds breadth into the investigation
through mindfulness of other work in the field. Thus, in principle,
just one ‘‘case’’ can lead to new insights (as a number of studies
has shown; see, for example, Frank [1995]) if it is recognized that
any such case is an instance of social reality.5 Nonetheless, with most
494 Social Science Information Vol 45 – no 4

research topics, some variety in the sources of the data does facilitate
and enhance the dialectic inherent in the search for depth of meaning
– and for appropriate ways of formulating it – that is the essence of
qualitative research methodology. For this depth to be achieved, it
is much more important for the research to be intensive, and thus
persuasive at the conceptual level, rather than aim to be extensive
with intent to be convincing, at least in part, through enumeration.
The mode of research produces concepts and propositions that have
construct validity because they make sense as pivotal points in a
matrix where interview yield intersects with pre-existing theoretical
knowledge.

An illustrative example

Our own research projects exploring the experiences of cancer sur-

vivors (Crouch and McKenzie, 2000; McKenzie, 2004; McKenzie
and Crouch, 2004) illustrate this approach. Both projects involved
in-depth interviews with small numbers of people who had been
treated for cancer, and who had hopes – though no assurances –
of a cancer-free future. We have found that people in this situation
can be profoundly preoccupied – even many years after diagnosis
and treatment – by the fear of the recurrence of cancer. However,
the suffering associated with that fear tends not to be meaningfully
accommodated in contemporary social life where a ‘‘positive out-
look’’ is a significant aspect of the contemporary collective con-
sciousness. Our respondents reveal that even those closest to them
may reject their continuing need for emotional support.
Thus many people in this situation remain isolated in their pain,
unable to share their existential fears. Our analysis here has drawn
on the concept of liminality, derived from the work of Van
Gennep (1960) and developed in the context of the experience of
cancer by Little et al. (1998). Confined to a borderline condition
between well and unwell, surviving and being threatened, cancer sur-
vivors must endure the indeterminacy of both their lives and their
social personae. It is from within this liminal space that our respon-
dents interact with their social environment, and it is this interaction
– of the doing with the enduring – that has been the focus of our
attention. In order to arrive at this point, our analysis has had to
build on our intense engagement with respondents and be supported
by intensive study of extant literature. This process precluded –
Crouch and McKenzie Trend report 495

and, indeed, would have been undermined by – a large number of

respondents.
Liminality, then, emerged, in due course, as a meaningful theo-
retical conceptualization of our respondents’ social situation –
what they had to endure (it did not account for the way in which
they managed that situation) and what they did about it. However,
our ‘‘vertical’’ analysis of the interview material has also led to new
insights about liminality, which, in turn, have contributed to pre-
existing theoretical knowledge concerning the liminality of cancer
survivors. Little et al. suggest that cancer patients experience two
liminal phases. The first of these, identified as ‘‘acute liminality’’,
involves the initial diagnosis and treatment stage, during which
the individual first ‘‘experiences the existential threat and its fear
and dread’’ (1998: 1492). This phase is followed by ‘‘sustained limin-
ality . . . in which each patient constructs and reconstructs meaning
for their experience by means of narrative’’ (1998: 1485). Little et al.
argue that, because of the uncertainty associated with cancer, this
phase extends, in some sense, for the term of the individual’s life.
However, our research suggests that cancer survivors experience
three liminal phases rather than two, the last one (coming into the
picture somewhere between 5 and 12 years after diagnosis) trans-
cending, for most individuals, sustained liminality. At the very
least, cancer survivors yearn for a third phase in which they can
experience some security and become more like ‘‘ordinary’’ people.
Most cancer survivors eventually imagine themselves inhabiting a
different state in the future, and they expend considerable effort
working towards that end. Thus, although the concepts of acute
and sustained liminality seem to be applicable to all cancer survi-
vors, the idea of a third phase is also very important because it facil-
itates conceptualization of a very significant dimension of experience
for many people.
The notion of liminality,6 then, conceptually bridges the gap
between personal exigencies and the social world that envelops
them. We argue here that studies which seek to provide such theo-
retical couplings of individual experience and social context are by
their very nature intensive. The work of linking interview accounts
– continually analysed – and conceptual frameworks – under con-
struction throughout the research – clearly requires small sample
sizes so that all the emerging material can be kept in the researcher’s
mind as a totality under investigation at all stages of the research.
496 Social Science Information Vol 45 – no 4

Conclusion

In sum, then, a small number of respondents is in no way an approx-

imation of the manner in which ‘‘ideally’’ research is to be done,
given the excuse, as it were, of the laborious nature of the activities
involved. On the contrary, we have argued that this is the way in
which analytic, inductive, exploratory studies are best done. There
is, however, an important qualifier of this conclusion. It has to be
borne in mind that exploratory research has little value if it is
restricted to stand-alone acts – ‘‘owned’’, as it were – by individual
researchers, rather than embedded in fields of relevance that are
tended by communal knowledge-building labour. Put differently,
let us follow up colleagues’ inquiries, not just cite them, approvingly
or critically, as the case may be, in our own work which, too, aims to
be unique. A more collectivist and generous attitude to research as
our fundamental property – in both senses of that word – might
then lead to ‘‘good work [which could] enrich and enliven the con-
versation. There should be a sense that it is moving forward, that
we seem to be getting somewhere’’ (Eisner, 1997: 259). If we
accept Blumer’s (1999) contention that the concepts of social science
are intrinsically sensitizing rather than definitive, we can move for-
ward in our understanding of social reality only by continuing,
and taking further, good work already done.

Mira Crouch is Visiting Senior Research Fellow at the University of New South
Wales. She has taught sociology of health and illness and social research methods
for many years. Her research interests are located in the broad domain of transi-
tions, e.g. early motherhood, menarche, youth more generally, and narratives of
illness and suffering. She is currently working on a memoir about loss and depri-
vation during the Second World War. Author’s address: School of Sociology and
Social Anthropology, University of New South Wales, Sydney 2052, Australia.
[email: [email protected]]

Heather McKenzie is a lecturer in the Faculty of Nursing, University of Sydney.

Her research interests include cancer survivorship, emotions and health, social
theory, health care politics, and the interface between institutions of health care
and consumers. She is currently involved in two research projects exploring the
nurse–patient relationship in community cancer care. Author’s address: Faculty
of Nursing and Midwifery, University of Sydney, Sydney 2006, Australia.
[email: [email protected]]
Crouch and McKenzie Trend report 497

Notes
1. For example, are these statements themselves ‘‘constructed’’ – again and again,
in different ways each time – when read or heard? As well, the ‘‘constructions’’ which
arise in interaction do not emerge from a vacuum, there has to be something initially
‘‘unconstructed’’ upon which the ‘‘constructions’’ arise. Where, then, is the bound-
ary between them?
2. ‘‘Interview’’ is a rather flat and inadequate term for conversations of this kind,
and we continue to use it here only for want of a better word.
3. We are referring to the problems of ‘‘methodological individualism’’, a discus-
sion of which is outside the scope of this article. The classical treatment of this issue
can be found in Lukes, 1977: 177–87.
4. The term ‘‘social context’’ is used here in its most general sense, to denote
aspects of social life which are supra-individual, and in that sense it also includes,
for the purposes of this article, ‘‘cultural’’ as well as ‘‘social’’ circumstances.
5. After all, how many times does one have to see a two-headed calf to be able to
say that it exists?
6. Liminality was not the only concept we developed in the course of our research.
We discuss it here in isolation for the purpose of illustrating some methodological
points made in our article. A comprehensive summary of our research as a whole
would have been inappropriate in this context.

References
Abbott, A. (1992) ‘‘What Do Cases Do? Some Notes on Activity in Social Analysis’’,
in C.C. Ragin and H.S. Becker (eds) What is a Case? Exploring the Foundations of
Social Inquiry, pp. 53–82. Cambridge: Cambridge University Press.
Archer, M. (1995) Knowing the Social World: the Morphogenic Approach. Cambridge:
Cambridge University Press.
Bell, C. (1968) Middle-class Families: Social and Geographical Mobility. London:
Routledge and Kegan Paul.
Bhaskar, R. (1979) The Possibility of Naturalism. Brighton: Harvester.
Blumer, H. (1999) ‘‘What is Wrong with Social Theory?’’, in A. Bryman and
R.G. Burgess (eds) Qualitative Research, vol. I, pp. 25–33. London: Sage. (Orig.
published 1954.)
Brante, T. (2001) ‘‘Consequences of Realism for Sociological Theory Building’’,
Journal for the Theory of Social Behaviour 31(2): 167–95.
Crotty, M. (1996) Phenomenology and Nursing Research. Melbourne: Churchill
Livingstone.
Crouch, M. and McKenzie, H. (2000) ‘‘Social Realities of Loss and Suffering Follow-
ing Mastectomy’’, Health 4(2): 196–215.
Crouch, M. and Manderson, L. (1993) New Motherhood: Personal and Cultural
Transitions in the 1980s. Yverton, Switzerland: Gordon and Breach.
Dreher, M. (1994) ‘‘Qualitative Research Methods From the Reviewer’s Perspective’’,
in J. Morse (ed.) Critical Issues in Qualitative Research Methods, pp. 281–97.
Thousand Oaks, CA: Sage.
498 Social Science Information Vol 45 – no 4

Eisner, E. (1997) ‘‘The New Frontier in Qualitative Research Methodology’’, Quali-

tative Inquiry 3(3): 259–73.
Flick, U. (2002) ‘‘Qualitative Research – State of the Art’’, Social Science Information
41(1): 5–24.
Frank, A. (1995) The Wounded Storyteller: Body, Illness and Ethics. Chicago, IL:
University of Chicago Press.
Glaser, B. and Strauss, A. (1967) The Discovery of Grounded Theory. Chicago, IL:
Aldine.
Gubrium, J.F. and Holstein, J.A. (1999) ‘‘Method Talk’’, in A. Bryman and
R.G. Burgess (eds) Qualitative Research, vol. I, pp. 119–35. London: Sage.
Jones, W.L. (1979) ‘‘The Investigation of Counter-predictive Research Outcomes
through the Method of Analytic Induction’’, in J.A. Martin (ed.) Counter-predictive
Research Outcomes, pp. 18–35. Canberra: Research School of Social Sciences,
Australian National University.
Kleinman, A. and Kleinman, J. (1996) ‘‘The Appeal of Experience; the Dismay of
Images: Cultural Appropriation of Suffering in our Times’’, Daedalus 125(1): 1–23.
Layder, D. (1998) Sociological Practice: Linking Theory and Social Research. London:
Sage.
Levy, D. J. (1981) Realism; an Essay in Interpretation and Social Theory. Manchester:
Carcanet Press.
Little, M., Jordens, C., Paul, K. and Phillipson, B. (1998) ‘‘Liminality: a Major Cate-
gory of the Experience of Cancer Illness’’, Social Science and Medicine 47(10):
1485–94.
Lukes, S. (1977) Essays in Social Theory. London: Macmillan.
McKenzie, H. (2004) ‘‘Personal and Collective Fears of Death: A Complex Inter-
section’’, in A. Fagan (ed.) Making Sense of Dying and Death, pp. 107–24. Amster-
dam: Rodopi.
McKenzie, H. and Crouch, M. (2004) ‘‘Discordant Feelings in the Lifeworld of
Cancer Survivors’’, Health 8(2): 139–57.
Michell, J. (2003) ‘‘Epistemology of Measurement: the Relevance of its History for
Quantification in the Social Sciences’’, Social Science Information 42(4): 515–34.
Miller, J. and Glassner, B. (1997) ‘‘The ‘Inside’ and ‘Outside’: Finding Realities
in Interviews’’, in D. Silverman (ed.) Qualitative Research: Theory, Method and
Practice, pp. 99–112. London: Sage.
Mills, C. Wright (1959) The Sociological Imagination. London: Oxford University
Press.
Minichiello, V., Aroni, R., Timewell, E. and Alexander, L. (1995) In-depth Interview-
ing: Principles, Techniques, Analysis, 2nd edn. Melbourne: Longman.
Morse, J. (1994) ‘‘Emerging from the Data: the Cognitive Processes of Analysis
in Cognitive Inquiry’’, in J. Morse (ed.) Critical Issues in Qualitative Research,
pp. 22–43. Thousand Oaks, CA: Sage.
Oakley, A. (1974) The Sociology of Housework. Oxford: Martin Robertson.
Oakley, A. (1980) Women Confined: Towards a Sociology of Childbirth. Oxford:
Martin Robertson.
Phillips, D. L. (1971) Knowledge from What? Theories and Methods in Social Research.
Chicago, IL: Rand McNally.
Platt, J. (1992) ‘‘Cases of Cases...of Cases’’, in C.C. Ragin and H.S. Becker (eds) What
is a Case? Exploring the Foundations of Social Inquiry, pp. 21–52. Cambridge:
Cambridge University Press.
Crouch and McKenzie Trend report 499

Renzetti, C.M. and Lee, R.M., eds (1993) Researching Sensitive Topics. Newbury
Park, CA: Sage.
Robinson, W. S. (1951) ‘‘The Logical Structure of Analytic Induction’’, American
Sociological Review 16: 812–18.
Salander, P., Bergenheim, T. and Henriksson, R. (1996) ‘‘The Creation of Protection
and Hope in Patients with Malignant Brain Tumours’’, Social Science and Medicine
42(7): 985–96.
Seale, C. (1999) The Quality of Qualitative Research. London: Sage.
Silverman, D. (1993) Interpreting Qualitative Data: Methods for Analysing Talk, Text
and Interaction. London: Sage.
Strauss, A. and Corbin, J. (1990) Basics of Qualitative Research. Grounded Theory,
Procedures and Techniques. Newbury Park, CA: Sage.
Turner, R.H. (1953) ‘‘The Quest for Universals in Sociological Research’’, American
Sociological Review 18: 604–11.
Van Gennep, A. (1960) The Rites of Passage. London: Routledge and Kegan Paul.
Van Maanen, J. (1983) Qualitative Methods. Beverly Hills, CA: Sage.
Walker, R. (1985) ‘‘An Introduction to Qualitative Research’’, in R. Walker (ed.)
Applied Qualitative Research, pp. 3–26. Aldershot: Gower.
Wiener, C.L. (1975) ‘‘The Burden of Rheumatoid Arthritis: Tolerating the Uncer-
tainty’’, Social Science and Medicine 9: 97–104.
Williams, S. (1999) ‘‘Is Anybody There? Critical Realism and the Disability Debate’’,
Sociology of Health and Illness 21 (6): 797–819.
Willmott, P. and Young, M. (1960) Family and Class in a London Suburb. London:
Routledge and Kegan Paul.
Znaniecki, F. (1963) Cultural Sciences: their Origin and Development. Urbana, IL:
Illinois University Press. (Orig. published 1934.)