Introduction
As argued in Chapter 1, methods
are presented in research textbooks as pro- cedural rules for obtaining
reliable and objective knowledge. One kind of method concerns procedural rules
for collecting data, of which ethnography is an example. Ethnography tends to
rely on a number of particular data col- lection techniques, such as naturalistic
observation, documentary analysis and in-depth interviews. While these methods
are used on their own as well, what marks their ethnographic application is
that they are used to study a people in a naturally occurring setting or
'field', in which the researcher par- ticipates directly, and in which there is
an intent to explore the meanings of this setting and its behaviour and
activities from the inside. This is what 'ethnography-understood-as-fieldwork'
means. However, the procedural rules that lay down how this is properly done,
and which thereby certify the knowledge as reliable and objective, obtain their
legitimacy and authority as procedural rules because the community of scholars
and researchers endorses them. According to John Hughes's (1990) arguments in
developing what he calls the philosophy of social research, this endorsement
itself derives from the fact that the procedural rules 'fit' with and reflect a
broader theoretical and philosophical framework for which researchers and
scholars have a preference. This framework is called methodology, and, in
short, methods-as-procedural-rules are given the authority to certify knowledge
as reliable and objective because they are legitimated by a methodological
stance. Method and methodology are thus inextricably linked. This chapter
explores the different methodological frameworks in which ethnography is
located and which go towards explaining the particular pro- cedural rules that
are endorsed as the 'way to do' ethnography properly. The procedural rules
themselves are outlined in greater detail in Chapter 3. Here, we outline the
philosophy of social research, describing the methodological premises
underlying 'little' ethnography, the imperatives for social research which
follow on from this methodology, the typical techniques of data col- lection
used and the characteristic form of data. Differences in methodo- logical
preferences highlight the divisions between ethnographers about theory and
practice, and this leads on to a debate about the current con- tested terrain
among ethnographers. Two debates are addressed: that around 'thick
description', which was once seen as the central characteristic of ethnographic
data; and that around the accuracy, reliability, validity and relevance of
ethnographic representations of reality.
The philosophy of social research
The philosophy of social research
can be defined as the study of the theories of knowledge which validate
particular research methods. Conventionally, social research methods courses
offer examination of the data collection techniques by which research is
undertaken. That is, they offer practical training in how to do research.
However, developments in social theory and philosophy have made us realize that
these techniques or procedural rules exist within a broader philosophical and
theoretical framework, which can be called a 'methodology'. This was presented
in the Introduction, and can be reproduced again:
methodology + procedural rules =
methods + knowledge
These methodological positions
involve researchers in commitments whether or not they are aware of it, for
they entail assumptions about the nature of society (called 'ontological'
assumptions) and assumptions about the nature of knowledge (called
Gepistemological' assumptions). These methodological positions can also entail
different sorts of research practices, since they predispose the use of
different data collection techniques. They thus end up producing quite
different kinds of data. The study of this broader methodological context to research
methods has been called the philosophy of social research (Hughes 1990; see
also Ackroyd and Hughes 1981).
The most contentious claim in
this argument is not that research methods get their authority and legitimacy
from particular theories of knowledge, but that researchers choose the data
collection techniques to employ in any piece of research because of a prior
commitment to this methodological position rather than out of practical
expediency. It is possible to envisage that this preference can be scientifically
based - in that researchers believe one methodology and the set of methods and
techniques to be more scientific than another - or it can be subjective and
personal. The researcher may lack the competence to understand and apply one or
other technique: since we cannot count or are frightened by computers, or do
not like talking to people, we avoid those methods that involve our
shortcomings. But what- ever the reason, we have biases. According to Hughes:
the data collection methods used
to make the social world amenable are not neutral tools that somehow exist
within a vacuum, but operate within a given methodological position; since
methodologies lay down the procedural rules by which reliable and objective
knowledge is said to be obtained, the choice of data collection technique is
not dictated by the problem at hand, but largely by prior preferences in the
researcher for a given methodological position with which those techniques or
rules are associated; the differences in the kinds of data produced have to be
located in methodological choices by the researcher rather than decisions about
the problem at hand; at a technical level it may be desirable, even necessary,
to combine mul- tiple methods, but at an ontological and epistemological level
this can result in marrying incompatible methodological positions; in a state
where there are competing methodological positions, validating different
procedural rules for collecting data, there will be no consensus about the
value or merit of particular research methods and the use of particular methods
can become a source of contention.
In his explication of the
philosophy of social research, John Hughes out- lined two models of social
research which were premised on two different methodological positions, the
natural science model based on positivism and the humanistic model based on
naturalism. These were counterpoised as mutually exclusive (indeed, as if they
were in a 'paradigm war') and set up almost as ideal types. His argument can be
made more concrete by describ- ing the two models and their respective
methodological justifications in his ideal type terms, although most attention
here is naturally devoted to the humanistic model. Table 2.1 summarizes the
differences between the two.
The natural science model of social research
The natural science model of
social research is premised on positivism. The essential attributes of this
methodological position are summed up in the
Table 2.1
The two models of social research
Natural
science Humanistic
Methodology
Positivism Naturalism
Methods
Questionnaires, surveys, In-depth interviews, experiments ethnography, personal
documents
Style of
research Quantitative Qualitative
Type of
data Numerate, 'hard' Natural language, 'soft'
word 'positive', which in the
English language conjures up an image of 'cer- tainty', 'precision' and
'objectivity'. And its principal characteristic is that the methods, concepts
and procedural rules of the natural sciences can be applied to the study of
social life. This involves ontological assumptions about the nature of society,
for social life is perceived to comprise objective structures independent of
the people concerned, and to consist of wholes and systems which go beyond the
consciousness of individuals. There is thus a 'real world' out there
independent of people's perceptions of it: the social world is revealed to us,
not constructed by us. It thus follows that objective knowledge is possible,
for there is a fixed and unchanging reality which research can accurately access
and tap. Further epistemological assumptions follow: knowledge of social life
can reveal only that which is externally observable through the senses, and it
can disclose the causal relationships that exist within social life. From this
follows the epistemological assump- tion that it is possible and desirable to
develop law-like statements about the social world by means of the
hypothetico-deductive method and using nomological-deductive explanations.
These phrases essentially mean the deduction of general statements from a
theory or law, from which hypoth- eses are formed, which are then tested
against prediction and observation. The best example remains Durkheim's theory
of suicide (Durkheim [I9051 1951). His general statement was that suicide varied
inversely with the degree to which individuals were integrated with the group.
From this he deduced less general statements, to the effect, for example, that
Catholics have lower suicide rates than Protestants because Catholicism is a
more communal religion and integrates believers into a more collective group.
Factual state- ments could be deduced from this which could be tested against
prediction and observation, to the effect that the suicide rate will be lower,
for example, in Catholic countries than Protestant ones. Suicide statistics for
Italy com- pared to those for Holland could then confirm or refute the original
general law-like statement. It is the original law-like general statement or
theory that is the explanatory variable, below which come descriptive data that
are revealed through sense-experience observation. Confirmation or refutation
cannot be achieved by data revealed through people's interpretative or mean-
ing-endowing capacities (in Durkheim's case by studying the meanings of vic-
tims as revealed, say, in their suicide notes; see Jacobs 1979) but only data
revealed externally through the way the world is observed and experienced via
our senses (in this case 'objective', 'official' statistics). Data for the
natural science model of social research are thus called 'hard', wishing to
imply that they are untainted by the interpretative and meaning- endowing
processes of people, whether these people are the subjects of the research or
the researchers themselves. And such data are numerate, seeking to measure and
describe social phenomena by the attribution of numbers. This gives an elective
affinity, as Weber would say, between the natural sci- ence model of social
research and those data collection techniques which give best access to sense-experience
data, notably questionnaires, surveys and experiments. Positivism believes the
world to be an external, knowable entity, existing 'out there' independent of
what people believe or perceive it to be. In a world made known to us through
our sense experience, people contribute very little to knowledge in this way,
simply receiving the sensory stimuli and recounting the response.
Questionnaires and surveys are exemp- lary at doing this. They collect numerate
data that supposedly render social phenomena 'objective' and untouched by
people's interpretative and reality- constructing capacities. Hence, for
example, textbooks identify the pro- cedural rules for, say, constructing and
applying a standardized interview schedule (advice on prompting and probing by
means of standardized phrases to be used by the interviewer, the elimination of
the 'interviewer effect' and practices to standardize the instrument), the
following of which supposedly allows researchers to eliminate personal and
interpersonal vari- ables that distort what is seen as a simple and
unproblematic relationship between stimulus (the question) and response (the
answer). Since the stimu- lus takes the same form for everyone, if respondents
give different responses the differences are assumed to be 'real', not
artificially created by variations in the way the question was asked. The data
thus become 'real', 'hard' and 'objective', since they are seen as untainted by
the personal considerations of the interviewer or the respondent (see Box 2.1).
Box
2.1 Standardization of the interview
Standardization
of the interviewer efea This relates to all kinds of interview.
I
Have one person do all the interviews so that respondents are subject to a
constant interviewer effect (or at least use a homogeneous set of
interviewers). 2 Randomize the effect by picking a random sample of
interviewers. 3 Minimize any inequalities between interviewerlrespondent (such
as sex, age, class, social status, religion). 4 Institute control by
assessing/supervising the interviewlinterviewer.
Standardization
of the interview as an instrument This particularly relates to formal
interviews with a rigid schedule.
I
Ensure the respondent is nominated by the sampling procedure, not the
interviewer (except with quota sampling). 2 Standardize introductory statements
about the purposeslsponsors of the research -there should be a standard reply
to any query for further information. 3 The interviewer should not try to
persuadelinfluence respondents either in what to say or whether to say anything
at all. 4 There should be precise instructions about when and how to prompt/
probe. 5 Try to keep recording by the interviewer to a minimum. 6 Ensure there
is no modification or variation in the instructions given to respondents. 7
Leave controversiallsensitive questions until the end. 8 Wording of the
questions. Ten basic rules: (a) use familiar words which are classlculture
fair; (b) use simple words free from jargonltechnical phrases; (c) be specific
and unambiguous; (d) be concise and to the point; (e) be precise, especially
avoid double negatives; (0 keep it short; (g) avoid leading questions which
suggest a response; (h) avoid hypothetical questions; (i) avoid presumptuous
questions which assume a response; 0) avoid double-headed questions (two
questions in one).
The humanistic model of social research
From the 1960s onwards there has
been an intellectual attack on positivism, from people like Thomas Kuhn, Karl
Popper and various kinds of interpre- tative sociologies, such as phenomenology
and ethnomethodology. This has resulted in an intellectual attack on the
natural science model of social research and also on its procedural rules for
certifying knowledge as reliable and objective. So there was an attack on the
very idea of questionnaires, for example, as reliable methods for collecting
data, no matter how well they were operated. Familiar and old methods were
impugned to their core, with a level of vitriol that approached what Pawson
(1999: 29-32) called a paradigm war. A new tradition emerged, or, more
properly, was rediscov- ered, since its ideas were longstanding, which gave
legitimacy to new pro- cedural rules and thus new methods for collecting and
analysing data (such as conversation analysis) or reinvented and repopularized
underused ones from an earlier period (ethnography, documentary analysis,
in-depth inter- views and participant observation). Hughes calls this the
humanistic model of social research (a phrase also used by Berger 1963, and
Bruyn 1966), and it is premised on the methodology of naturalism (this
methodology is also sometimes called the interpretative or hermeneutical
paradigm). Naturalism is an orientation concerned with the study of social life
in real, naturally occurring settings; the experiencing, observing, describing,
under- standing and analysing of the features of social life in concrete
situations as they occur independently of scientific manipulation. The focus on
natural situations leads to this orientation being described as 'naturalism',
and it is signified by attention to what human beings feel, perceive, think and
do in natural situations that are not experimentally contrived or controlled
(the emphasis upon interpretation also explains why it is called the
hermeneuti- cal paradigm). These naturally occurring situations are also
sometimes called 'face-to-face' situations, mundane interaction,
micro-interaction or everyday life. Stress is laid on experiencing and
observing what is happening naturally rather than hypothesizing about it
beforehand, mostly by achiev- ing first-hand contact with it, although
researchers minimize their effect on the setting as much as possible. Stress is
also laid on the analysis of people's 'meanings' from their own standpoint: the
feelings, perceptions, emotions, thoughts, moods, ideas, beliefs and
interpretative processes of members of society as they themselves understand
and articulate them (see Box 2.2), at least initially. Naturalism presents this
as 'being true to the natural phenom- ena' (Douglas 1980: 2).
Box
2.2
Interview
with a member of Ian Paisley's Free Presbyterian Church, for research published
in J. D. Brewer, Anti-Catholicism in Northern Ireland 160&1998 (London:
Macmillan, 1998), with G. Higgins.
I
believe that the Roman Catholic Church is the whore of Babylon, a system which
is contrary to the word of God. I honestly have no time for the ecumenical
movement. To me there can be no reconciliation between what Rome teaches and
the Scriptures. I don't believe light can have any fellowship with darkness . .
. I would have no problem with a united Ireland if it was under British rule.
United Ireland as it means today would be a Roman Catholic dominated thing . .
.The British way of life would be what I would favour.
There are ontological and
epistemological assumptions within this stance, which further highlight its
contrast with positivism as a methodo- logical position. Central to naturalism
is the argument, going back to German philosophy in the nineteenth century (the
Geisteswissenschaften tradition), that human beings and social behaviour are
different from the behaviour of physical and inanimate objects. People are
meaning-endow- ing, in that they have the capacity to interpret and construct
their social world and setting rather than responding in a simplistic and
automatic way to any particular stimuli. Moreover, people are discursive, in
that they have the capacity for language and the linguistic formulation of
their ideas, and possess sufficient knowledge about discourse in order to
articulate their meanings. Society, thus, is seen as either wholly or partially
constructed and reconstructed on the basis of these interpretative processes,
and people are seen as having the ability to tell others what they mean by some
behaviour, idea or remark and to offer their own explanation of it or motive
for it. Society is not presented as a fixed and unchanging entity, 'out there'
somewhere and external to the person, but is a shifting, changing entity that
is constructed or reconstructed by people themselves. People live in material
and bounded structures and locations, and these contexts shape their
interpretative processes, so that we are not free to define the social world as
if we existed as islands, each one inhabited by ourselves alone. All social
life is partially interdependent on the concrete situations and structures in
which it exists, so 'society' is not a complete invention (or reinvention)
every time. But knowledge of the social world, in this methodological position,
is inadequate if we do not also document, observe, describe and analyse the
'meanings' of the people who live in it. This must be the starting point of any
study of society according to natu- ralism, although it may clearly not be the
end point, in that the researcher may want to extend the analysis beyond
people's own accounts, expla- nations and meanings. The theory of knowledge
within naturalism thus sees it as essential to understand 'the freely
constructed character of human actions and institutions' (Hammersley 1990: 7)
in the natural settings and contexts which influence and shape people's
meanings. Thus, knowledge must be inductive (the reverse of deductive), in
which researchers begin with particular observations, from which empirical
statements are made, which may, or may not, lead on to statements of more
generality. It is dis- covery-based rather than hypothesis testing. The three
essential tenets of naturalism are thus:
the social world is not reducible
to that which can be externally observed, but is something created or
recreated, perceived and interpreted by people themselves; knowledge of the
social world must give access to actors' own accounts of it, among other
things, at least as a starting point, and sometimes as the sole point; people
live in a bounded social context, and are best studied in, and their meanings
are best revealed in, the natural settings of the real world in which they
live.
Four imperatives or requirements
for social research follow from this methodological position. Social
researchers in the humanistic model of social research need:
to ask people for their views,
meanings and constructions; to ask people in such a way that they can tell them
in their own words; to ask them in depth because these meanings are often
complex, taken for granted and problematic; to address the social context which
gives meaning and substance to their views and constructions.
The implications of these
imperatives are significant and they go towards defining the attitude and
approach of humanistic researchers. There are three implications. First, they
predispose the humanistic researcher to study certain sorts of topics above
others, ones that lend themselves readily to the study of people's views,
beliefs and meanings. The reverse of this is that the researcher is cut off
from studying those topics that are not appropriate to being approached in
terms of people's beliefs and meanings. Second, while it is the case that most
topics can be addressed in various ways, researchers with a preference for the
humanistic model are predisposed to ask certain sorts of questions about that
topic, approaching the topic in terms of people's meanings, attitudes, beliefs
and interpretations. Let me illustrate this with the example of my research on
crime in Ireland, North and South, between 1945 and 1995 (Brewer et al. 1997).
It combined quantitative and qualitative research. In part it examined trends
in crime statistics in this time period at the national level for Northern
Ireland and the Irish Repub- lic, and at city level for Dublin and Belfast, as
well as trends in the official statistics for specific crimes, such as murder,
robbery, theft, rape and sched- uled offences under terrorism legislation.
However, official statistics have profound limitations, and as a qualitative
researcher first and foremost, I needed to approach the topic by asking other
sorts of questions as well. An ethnographic part of the study therefore sought
to supplement the quanti- tative analysis of crime by addressing a whole range
of issues raised by crime, to permit the expression of these concerns in the
actors' own terms and to capture the richness and depth of the crime problem.
Thus, we also focused on two local communities in Belfast, one each in Catholic
West Belfast and Protestant East Belfast, and addressed issues such as people's
perceptions of the crime problem in their locality, levels of fear of crime,
people's reporting behaviour, local crime management in the absence of reliance
on the police, the frames of reference through which people approach crime,
such as perceived levels of crime in other societies or his- torical
comparisons with the past, and people's fears about future crime in their areas
after the ceasefire by paramilitary groups. Such data captured the richness of
people's experiences in their own terms, proffering a coun- terweight to the
breadth and geographical coverage of official statistics. Actors' accounts take
on added value with respect to crime statistics because of the well known
limitations in official statistics on crime. There- fore, the topic itself was
defined in such a way as to permit study of people's meaning-endowing
capacities. A third implication of these imperatives is that they predispose a
prefer- ence for certain data collection techniques. Methods of data collection
in the humanistic model of research must access people's views and meanings,
and do so in depth without imposing views upon people. They must permit people
to speak in their own terms and in the context of the natural settings which
give meaning and substance to their views. The popular methods of data
collection in qualitative research are therefore techniques such as in- depth
or informal interviews, personal documents, like diaries, letters and
autobiographies, participant observation and methods for the study of nat- ural
language, like conversation analysis (these data collection techniques are
discussed in Chapter 3). These methods are sometimes summed up in what has here
been termed 'big ethnography' or 'ethnography-understood-
as-qualitative-research', or summed up in the term 'unobtrusive methods'. One
of the most important is 'little ethnography', what is here called
'ethnography-understood-as-fieldwork'. The data collected by these methods come
in a particular form. Quali- tative data come in the form of extracts of
natural language, such as quo- tations obtained from in-depth interviews, notes
from personal documents or records of participant observation, providing
actors' own accounts. Such data capture the richness of people's experiences in
their own terms. As Schwartz and Jacobs (1979: 4) write:
quantitative sociologists assign
numbers to observations. They produce data by counting and 'measuring' things.
The things measured can be individual persons, groups, whole societies, speech
acts and so on. Qualitative sociologists report observations in the natural
language at large [see Box 2.31. It is intrinsically important to develop ways
of gain- ing access to the life-world of other individuals in the context of
[their] daily life.
Subtitling their book 'A method
to the madness', they contend that such methods are the only way for the social
researcher to chart a way through the chaos and complexity of social life.
Box
2.3
Extract
from J. D. Brewer, B. Lockhart and F? Rodgers, Crime in Ireland, 1945-95
(Oxford: Clarendon Press, 1997), p. 128.
As
one resident remarked after being asked whether ordinary crime was a problem in
Poleglass [Catholic West Belfast] 'definitely . . .being totally honest with
you, you would probably hear of most crimes. They probably wouldn't be on a big
scale but you would get them - joyriding, thieving, thieving of anything;
anything you leave sitting about here is going to walk, whether your name is on
it or not. If it's not nailed down they'll have it'. In the Bloomfield district
of East Belfast a resident recounted her experiences:
There
is a lot of crime in the area. The most common is breaking-in and vandalism.
Broke into Elmgrove primary school this morning, they took what they wanted but
they vandalised it as well. Last Saturday the bakery was held up and all the
money taken. The day before I was in the post office and some fella was playing
with the blind [collection] box, then he lifted it. I know the travel agent has
been broken into. Basically all the shops around here have been broken into. I
have been broken into three times. I had stuff taken from the [washing] line
three or four times. The bikes have been taken umpteen times.
The methodological bases of ethnography
It is clear that the principal
methodological justification for ethnography comes from naturalism and the
humanistic model of social research. This gives us what we earlier called the
'humanistic' type of ethnography. It is what most people think of when they
reflect on ethnography and it is what most ethnographers do when they practise
fieldwork: 'getting close to the inside', 'telling it like it is', 'giving an
insider's account', 'being true to the natural phenomena', giving 'thick
description' and 'deeply rich' data. It is research that abandons natural
science models of research practice, such as hypothesis testing, deductive
analysis, description and measurement by means of assigning numbers; it
abandons even the rhetoric and ambitions of natural science in favour of
understanding naturally occurring behaviour in its own terms. However, the type
of ethnography we called 'scientific' or 'positivist' draws on elements of both
methodologies and is associated with both models of research. It does so by
accepting the orthodox consensus that natural science is the standard by which
research should be judged and by arguing that there is a 'real' world which it
is possible to access. Yet it believes that ethnography is, perhaps, more
scientific than quantitative methods because it enables researchers to get
closer and better access to this 'real' world. 'Critical realist' ethnographies
are like this, as was the practice of early advocates (Bruyn 1966; Blumer
1969), who argued that ethnogra- phy had a scientific character precisely
because it was better suited than experimental and survey research to
understanding human behaviour (see Hammersley 1990: 6). In one important
respect the two types of ethnography are identical in what they see as the
proper purpose of ethnography, even though they come to this from diametrically
opposed methodological positions. Both have a naive notion that there are
objective truth statements that can be made about the phenomena under study,
that ethnography best permits these truth statements and that these truth
statements reflect the 'real' understanding of the phenomena. Both believe that
it is possible to 'tell it like it is', and, further, that there is only one
'true' telling. Ethnography thus uniquely ren- ders a problematic social world
unproblematic, for it alone has the capacity to disclose the social world as it
truly is. It is for this reason that Silverman (1985: 116) observed that
naturalistic and positivistic types of ethnography both sought the elimination
of the effects of the researcher in order better to represent the 'real'
picture and the 'true' understanding of the phenomena, the former by
recommending that ethnographers embrace the culture and the setting to become
an 'insider', the latter by recommending the standard- ization of all research
procedures and instruments. In this sense, both types of ethnography subscribe
to what is known as 'naive realism'. Both types of 'postmodern reflexive'
ethnography challenge them on this, attacking the grounds on which they claim
to represent 'reality' and the criteria by which they claim legitimacy for the
validity and accuracy of their data. This chal- lenge makes ethnography 'the
most hotly contested site in qualitative research today' (Denzin and Lincoln
1998: xvi); and it describes what Denzin and Lincoln (1998: 21) call the
'double crisis' of contemporary ethnography, where its representational claims
are questioned by other eth- nographers and the validity of its data impugned.
Hence Atkinson and Hammersley (1998: 129) rightly point out that there is not a
single philo- sophical or theoretical orientation that can lay unique claim to
supply the rationale for ethnography, each endorsing a version of ethnographic
work. It is to these crises that we now turn.
The double crisis of ethnography
The 'crisis of representation'
The 'crisis of representation'
describes the disillusionment surrounding the ethnographer's claim to provide a
privileged and special access to 'reality' by means of 'thick description'. As
Dey (1993: 31) makes clear, to describe something is to recite its
characteristics in either numbers or natural lan- guage. In natural science
models of social research, description has low status, which is ironic, since
description in the form of numbers permeates the natural science model. It is
description by means of extracts of natural language that is problematic to
positivism. Yet for both 'scientific' and 'humanistic' forms of ethnography,
such description is central to the ethno- graphic enterprise, although they must
be what is called 'thick' as opposed to 'thin' descriptions. 'Thick'
description was a term first used by the anthropologist Clifford Geertz in
1973, and popularized in sociology by Norman Denzin, although its origins lie
within the British anthropological tradition of Malinowski, where researchers
were enjoined to describe phenomena from the natives' point of view. Thin
description is mere gloss, a bare report of the 'facts' independent of
intentions or circumstances, whereas thick description represents a thorough
account (see Box 2.4), taking in the context of the phenomena described, the
intentions and mean- ings that organize them, and their subsequent evolution or
processing (see Denzin 1989: 31,83ff). It is a form of 'subjective soaking'
according to Ellen (1984), in which researchers attempt to merge with the
phenomena being described. It is a 'written representation of a culture' (van
Maanen 1988: I), or what Fetterman (1998: 20) calls 'the emic perspective',
where phenom- ena are described from the insider's perspective, which is
'instrumental to understanding and accurately describing situations and
behaviours'. Fetter- man (1998: 29) outlines thick description thus: 'it
involves detailed descrip- tion, detailed frame-by-frame analysis of events or
parts of events.' Or in Denzin's own words, it:
presents detail,
context, emotion, and the webs of social relationships. Thick description
invokes emotionality and self-feelings. It establishes the significance of an
experience or the sequence of events. In thick description, the voices,
feelings, actions, and meanings of interacting individuals are heard. It
captures and records the voices of 'lived experi- ence'. (Denzin 1989: 83)
For many 'humanistic'
ethnographers, such thick description is an end in itself, since their task is
solely to engage in 'cultural description' as anthro- pologists might say, or,
as some sociologists might say, it facilitates the goal of demonstrating the
reality construction done by ordinary people. If reality reconstruction is the
goal, by which Schwartz and Jacobs (1979: 2) mean that 'messy, tortuous
business of learning to see the world of an individual or group from the
inside', then 'there is something vital that one does not know if one has no
access to the inside - that is, if [we are] unable to recon- struct the world
as it looks, sounds, and smells to those within it.' This is what thick
description achieves, a 'realist' narrative of the social world from
Box
2.4
Extract
from J. D. Brewer, Inside the RUC (Oxford: Clarendon Press, 199 1 ), with K.
Magee, pp. 60-2.
Another
type of work which is disliked is emotionally demanding work. When they talk
about work of this type, policemen stress the importance of remaining detached
and emotionally cold. Young probationers are instructed to follow the 'police
pattern', what Schutz would call the 'recipe knowledge'. An elderly sergeant
once remarked on his experience of attending a cot death for the first time,
'You just have to say to yourself, the next time I will be better equipped to
cope with this type of situation . . .There's a pattern police follow in every
situation. You lay the pattern down whatever you're dealing with, and you
follow it through'. If she could look at meat hanging in a butcher's, the
field-worker was told when she was accompanying policemen to a post-mortem, she
could look at dead bodies. The tendency to render horrific incidents of this
sort into funny tales or 'atrocity stories', told ritualistically within the
occupational culture of the station, is a further attempt to strip them of
their emotional hold. In the midst of passing on advice to the field- worker on
how to cope with her imminent attendance at a post- mortem, one policeman said,
'You get used to them. I don't mind them any more'. But after a pause he went
on to add, 'Except for kids, I hate going to post-mortems for kids'. Below is
an extract of conversation between two policemen who are telling a third about
a cot death the two of them had recently attended.
PC.
I: Jesus, it was awful, and the worst thing about it was, when we arrived the
baby was still warm, so we tried to revive it with mouth to mouth. Now the
couple had expected it was dead and we gave them false hope. When we couldn't
revive it, it made the whole thing worse for them. Then when the ambulance men
arrived they also tried to revive it. God it was awful. PC. 2: 1 always feel
like saying, 'Look it's OK, I'll come back in a couple of weeks. But you never
do like. You'd get the balls chewed off you if you returned without all the
details. PC. I: But it was awful. God, the couple were really upset, it was
their first baby, too.
It is
not that policemen and women fail to achieve emotional detachment. Primarily
what makes this type of work unpopular is the ever-present danger that work of
this sort will break the veneer of coldness, exposing them as emotionally
involved, which is something they dislike because it is considered
unprofessional.
the inside. But 'scientific'
ethnographers claim thick description to be of immense value too, for it is a
form of explanation. Thick descriptions are not a preliminary to explanation,
David Silverman (1985: 95) once wrote, 'but are in themselves adequate
scientific explanations'. If the ambition is to achieve a positivist account of
some phenomenon, to capture its 'real' fea- tures accurately and objectively,
thick description can be seen as an aid to science through the achievement of
'realism'. By 1997 Silverman was parodying the ambition to 'tell it like it is'
as the equivalent of the television chat show (1997a: 248-9). In the
intervening period, ethnography had become infected by what Hammersley (1990:
5; see also 1992) calls anti-realism. In the mid-1980s in cultural anthropology
and then in sociology, ethnographers used ideas from several sources to examine
critically their craft and criticized both 'human- istic' and 'scientific'
ethnographers for their realist assumptions that an objective reality exists
and that it is possible to represent it accurately in the ethnographic text. These
assumptions are naive - hence 'naive realism'. Anthropology's critique of
ethnography is longer established than soci- ology's, partly because
ethnography is so central to anthropology, but also because of the greater
demands made of the method in cultural and social anthropology. Much
anthropological knowledge depends upon the capacity of the ethnographic method
to represent reliably the dynamics of cultures which are strange, and some
anthropologists questioned the capacity of eth- nographers to represent foreign
cultures objectively (Marcus 1980; Clifford 1981, 1983; Marcus and Cushman
1982; Stocking 1983; Clifford and Marcus 1986; Spencer 1989). Although drawing
partly on this material to mount a challenge to ethnography in sociology,
ethnographers (for example, Woolgar 1988a: 24-9; Atkinson 1990: 25-8) also draw
on work in social studies of science. Naturalistic critiques of scientific
knowledge (and texts) were turned upon themselves and applied to social
scientific knowledge generally, and ethnography in particular (Anderson 1978;
Woolgar 1988a, b). Postmodernism's rejection of the meta-narrative of science,
in which the realist ambition to 'objective truth' is deconstructed to language
games involving competing truth claims, was also an impulse to anti-realism.
Several issues follow from the anti-realist attack on naive realism. The first
and most serious is a challenge to ethnographic representations or tellings 'of
it like it is'. In naive realism, the representation of social reality is seen
as unproblematic as long as the researcher follows the procedural rules and
gets sufficiently close to what it is like 'on the inside'. The researcher nar-
rates the ethnography, providing thick descriptions that give readers the
impression that they are in the field along with the ethnographer. Ethnogra-
phers must thus absent themselves from the text, trying to act as a mere con-
duit in which the insider's account is simplistically represented in the text.
As van Maanen (1988: 47) writes, 'the narrator of realist tales poses as an
impersonal conduit who passes on more-or-less objective data in a measured
intellectual style that is uncontaminated by personal bias, political goals or moral
judgements.' Realist ethnographers thought they obtained a privi- leged gaze by
means of their closeness and insider status, and this is what must be
represented in the text via a form of ethnographic reportage which uses various
rhetorical devices to construct the text as an accurate portrayal based on
close association in the field and the successful development of insider status
(ethnographic texts and writing are explored further in Chap- ter 4). The
problem, according to the anti-realists, is that there is no inde- pendent and
external reality, and the ethnographer's representation is not privileged; it
is just as much a partial account as the insiders', and claims to realist-like
objectivity, accuracy and truth are spurious. Thick descriptions, therefore, do
not represent 'reality as it is' because such descriptions are selective from
the various competing versions of reality that could have been produced and end
up presenting a partial picture: if naive realist ethnogra- phers see
themselves as cameras, the picture is blurred because there is more than one
image on the lens. 'The doctrine of immaculate perception', as van Maanen
(1988: 23) termed naive realism, is undermined by the opaque nature of
'reality' and the ethnographer's selection processes. Keeping 'an open mind' is
not the same as having 'an empty head' (Dey 1993: 63), and unadulterated
observation is impossible. As Fielding (1993: 163) says, 'objective'
observation is impossible because the observer is involved, not detached. One
of the factors that naive realist ethnographers fail to recognize as impinging
on their observation is theoretical bias. Hughes's arguments within the
philosophy of social research contend that their conception of ethnography in
naive realist terms is itself a theoretical preference, but beyond this, naive
realist ethnographers tended to present themselves as theoretically neutral,
building up theories in a grounded fashion from the data themselves.
Genuflection in the direction of Glaser and Strauss's (1967) explication of
grounded theory was routinely made, whereby theory is sup- posed to be based on
observation of data not deduced from prior assump- tions. Anti-realists argue
that ethnography's descriptions are theoretically naive and no different from
those produced by ordinary people as part of their everyday life (Hammersley
1990: 60-5). They are not specialized 'theoretical' descriptions. That is, they
neither adequately test nor generate theory; even the theoretical inferences
made from the data are often unsub- stantiated. Hammersley is particularly
critical of ethnographers who do not identify the theoretical assumptions and
wider values that they bring to their work, which often condition their
interpretation of the data and the theor- etical inferences made. Ethnographers
who imply that their accounts are accurate representations of the social world
'as it is', beyond the influence of theoretical presumption or prejudice, are
both ignorant of the effect of their values upon research and simplistic in
suggesting that there is only one objective description which they have reliably
captured. The final criticism made by anti-realist ethnographers is that the naïve
realist emphasis on thick description limits the ethnographer's task to that of
cultural description (as anthropologists might say) or reality reconstruc- tion
(as sociologists would say). Anti-realist ethnographers recognize both the
impossibility of 'telling it like it is' (since there is more than one
'telling' and more than one 'is') and the desirability of going beyond people's
words. Thus, Altheide and Johnson (1998: 297) write:
capturing members' words alone is
not enough for ethnography. If it were, ethnographies would be replaced by
interviews. Good ethnogra- phies reflect tacit knowledge, the largely
unarticulated, contextual understanding that is often manifested in nods,
silences, humour and naughty nuances. [And] it is necessary to give an
accounting of how we know things, what we regard and treat as empirical
materials - the experiences - from which we produce our second (or third) hand
accounts of 'what is happening'.
Altheide and Johnson close this
remark by alluding to the chief solution to the crisis of representation
offered by anti-realist ethnographers, which is reflexivity. As the postmodern
ethnographer's response to the crisis of representation, 'reflexivity' is something
of a buzz word in contemporary ethnography (and tiresome to a few
ethnographers, for Silverman (1997a: 239-40) has writ- ten that we play the
reflexivity card too often and risk being perceived as navel-gazing). The
problem is that realist ethnographers (like realist social scientists
generally) are unreflexive, in that they give no attention to the social
processes that impinge upon and influence their data. They do not adopt a
critical attitude towards their data, and even deny the influence of such factors
as the location of the setting, the sensitivity of the topic or the nature of
the social interaction between the researcher and researched. Thus, the
strengths of the data are exaggerated andlor the weaknesses underem- phasized,
further undermining the reliability of ethnographic thick descrip- tions.
Therefore, Hammersley and Atkinson (1983: 17) argue that instead of trying to
eliminate the effects of the ethnographer, we should be reflexive in trying to
set the data against this context (Bowden 1989 argues similarly for
quantitative approaches). Although Woolgar (1988a: 21-4) means some- thing
different when he urges us to be reflexive, the injunction remains: eth-
nographers (like social scientists generally) must locate their data in the
context of the social processes that brought them about, and recognize the
limits of their representation of reality (Woolgar 1988a: 26-7). This does not
mean that the ethnographer has to try to construct the relevant coun-
terfactual by engaging in the impossible task of imagining what the data would
have been like had circumstances been different. Rather, it implies that
ethnographers be explicit and open about the circumstances which pro- duced the
extant data, recognizing that ethnographers (like all researchers) are within
the social world they seek to analyse. Since there is no perfectly transparent
or neutral way to represent the social world (or the natural one), reflexivity
on the part of the researcher assists in identifying the contingen- cies that
produced his or her portrayal of it, so we should claim no more for the account
than what it is, a partial, selective and personal version (see Box 2.5). (The
question of reflexivity is addressed further in Chapter 4.) The 'crisis of
representation' thus describes a situation where ethnography is unsure about
the status of its descriptions and observations because its claim to privileged
access to 'reality' by means of thick description is impugned. Postmodernists
argue that there is no one 'reality' and ethnogra- phy captures only the
version that the researcher selects. This is a general complaint about all
methods, however: 'there is doubt that any discourse has a privileged place,
any method or theory a universal and general claim to authoritative knowledge' (Richardson
1991: 173). Realist ethnography, like all research methods, is stripped naked
under the impulse of anti-realism and postmodernism, an emperor without
clothes, whose claims to authority are illusionary. The problem thereby created
is how to judge good ethnography, for the postmodern condition is one that
undermines all criteria by which to judge and evaluate the products of
ethnographic research: all criteria are doubted, none are privileged and
everything goes. This is the crisis of legiti- mation.
Box
2.5
Extract
from J. D. Brewer, B. Lockhart and I? Rodgers, Crime in Ireland 1945-95
(Oxford: Clarendon Press, 1997), pp. 123-7, by permission of Oxford University
Press.
Ethnographic
findings can be easily misunderstood. For some, ethnography represents the only
research method because it alone captures people's experiences in their own
words but others denigrate it. . . However, it is necessary for ethnographers
to be reflexive and identify the contingencies that helped to produce the
extant data. This is our intention . . . Ethnographic research on crime in
Belfast, which touches on issues such as policing and the role of paramilitary
organisations in local crime management, fits the template of sensitive
research. It is important to identify what bearing this sensitivity had on the
research. The first was with respect to the sample. Because of the sensitivity
of the topic of the research, we felt it necessary to work through local
community- based agencies in order to access general members of the public; the
public were not accessed by means of unsolicited knocks on front doors but
through their involvement with local community groups and organisations.
Initial contact with the organisations was facilitated by the network of
contacts possessed by the investigators and by the snowball technique. These
community organisations acted as a buffer or gatekeeper between the fieldworker
and the public, giving each some reassurance and security when addressing
controversial and deeply sensitive questions. lnterviews also took place in the
familiar surroundings of the organisation's premises. Fieldwork took place over
twelve months between 1994 and 1995, with six months spent in each study area,
and the fact that the ceasefire pertained for most of the fieldwork, and for
all of that which took place in West Belfast, encouraged openness amongst
respondents. People's frankness about the paramilitaries was no doubt
facilitated by their thought that peace had arrived. The use of community
organisations as gatekeepers also facilitated a measure of representativeness,
a problem which hinders the reliability of much ethnographic research because
of the small numbers of people studied. Our research design allowed us to
ensure that the organisations selected were an accurate political and social
representation of the locality, as well as covering a cross-section of key
social groups, such as women, youth and the elderly; this social and political
representativeness could not have been so readily achieved by means of unsolicited
access to the general public, which can overlook members of minority groups.
Some community groups, however, are often politically aligned to the mainline
and fringe political groups in Northern Ireland but the wide variety of views
obtained seems to show that we were not hijacked by representatives or
supporters of any one political organisation . . . Our research design involved
exclusively the use of indepth interviews. lnterviews were arranged and
conducted solely by one of the authors. In total, 1 15 interviews were carried
out with individuals and ten with groups. They were recorded on tape and then
transcribed verbatim, except where respondents objected to the interview being
recorded, when notes were taken during the interview. There are ninety-two
hours of tape recordings. Two areas were selected for study, one each in East
and West Belfast, in order to reflect the spatial location of Belfast's
communal divide. To aid comparison it was important to select closely matching
sub- divisions, and ones which provided a cross-section of social classes and
housing styles, large council estates, areas of inner city deprivation and
suburban housing, providing a mix of community types and social classes.
Fieldwork deliberately covered organisations based in most of the localities
within each broad area in order to provide some geographical spread and social
representativeness. Each area contained pockets where members of the other
religious community live. In fieldwork we made sure that we covered organisations
in these enclaves.
The crisis of legitimation
'Humanistic' and 'scientific'
types of ethnography were both 'realist' in their different ways, in believing
that there was a knowable world which the proper procedural rules, faithfully
followed, could accurately tap. These procedural rules not only defined how
ethnography should be practised as a data collection technique, they outlined
the criteria by which the resulting data could be evaluated. Terms like
'validity', 'reliability' and 'generaliz- ability' were suggested as the
criteria. 'Validity y refers to the extent to which the data accurately reflect
the phenomenon under study (also sometimes called 'internal validity'),
'reliability' to the extent to which measurements of it are consistent and 'generalizability'
to the applicability of the data to other like cases (also sometimes called
'external validity'). These are terms 'owned' by positivism and appropriated
enthusiastically by 'scientific' ethnography, but even 'humanistic' ethnography
paid attention to the ways in which its data had validity and could be made
more generally applicable (see, for example, LeCompte and Goetz 1982; Kirk and
Miller 1986). 'Humanistic' ethnography's commitment to naturalism sometimes
ensured that practitioners thought the sole criterion should be whether the
data accu- rately captured the phenomenon (validity), taking solace in the
notion that while ethnography had high validity but low reliability, the
reverse was the case in the natural science model of social research because
quantitative methods could replicate data constantly but at the expense of an
accurate description of social life. Such 'validity' was achieved, Fielding
(1993: 164) wrote, when the observer knew the members' rules of action
sufficiently well to be able to tell others how to pass as ordinary members in
the same field. But even with the stress within humanistic ethnography on
validity as the sole evaluator, sampling methods were introduced into
ethnography (see Chapter 3), and fields were sought where the processes being
studied were most likely to occur but which were seen as single instances of
more general social experiences and processes. Constant comparisons were made
as the researcher was urged to develop an understanding that encompassed other
instances of the process, and a focus on negative cases reiterated the intent
to study the particular in order to examine the general. Thus, Denzin and
Lincoln (1998: xiv) argued that any single case bore traces of the universal,
and 'telling it like it is' was always associated with the idea that the
'telling' should encompass the general features of the 'it' (see Box 2.6). The
anti-realist challenge to the nature of knowledge (that there is no objective
and knowable 'real' world that can be accurately described) under- mines the
traditional criteria to evaluate ethnographic data since they are based on
'realist' assumptions. Hence the crisis of legitimation.
Box
2.6
Extract
from J. D. Brewer, Inside the RUC (Oxford: Clarendon Press, 199 I), with K. Magee,
pp. 30-3.
The
familiar adage is that ethnographic research provides depth by sacrificing
breadth, but it is possible to build an element of generality by constructing
individual projects in the mould of similar ones in different settings so that
comparisons can be made and a body of cumulative knowledge established. Our
project was designed deliberately to follow the pattern of ethnographic studies
of routine policing, so as to add to this tradition the dimension provided by
studying this kind of policing in a divided society. In a strategy pioneered by
the affluent-worker study in Luton, which one might call the optimal case
approach, a site was chosen for the research which was not representative but
was particularly germane to the topic of the investigation. 'Easton' was
purposely selected because it is in an area of Belfast where routine policing
is possible. If we are to establish how and to what extent routine policing is
affected by Northern Ireland's divisions, it would be useless to base our research
where there is only militarised political policing, for it is necessary to
explore the extent to which policing in so-called 'soft' areas is contaminated
by wider societal divisions. Given the nature of crime in 'Easton', the
problems Northern Ireland's divisions create for routine policing are as well
studied there as anywhere else . . .The social structure of the district of
'Easton' is worth noting.
It is over the issue of
legitimation and the criteria left to evaluate ethno- graphic data that postmodern,
reflexive ethnographers divide. Less extreme postmodern, reflexive
ethnographers accept that some criteria need to be applied or ethnographic data
cannot be vouched for and evaluated. We would then be in a state of utter
relativism, the epitome of the postmodern dissolution into nothingness. We
would all be ethnographers then - at least we could not distinguish between
good and bad ethnography - and prac- titioners could properly be parodied as
chat show hosts or hack journalists, for the criteria to distinguish lay and
professional ethnography, or good and bad ethnographic research, would be
unknown or uncertain. This is pre- cisely what extreme postmodernist
ethnographers claim, but ethnographers like Hammersley (1990) have outlined new
criteria, validated by a method- ology he calls 'subtle realism'. Altheide and
Johnson (1998) have done like- wise for a position they call 'analytical
realism', and I have outlined a set of guidelines emerging from the postmodern,
reflexive critique of ethnography to strengthen rather than undermine
ethnographic practice (Brewer 1994). The rupture of postmodernism is thus
lessened to a considerable degree (see also Lincoln and Guba 1985, for a
discussion of new criteria to assess 'nat- uralistic' research by; for a review
of what he calls 'criteriology', see Seale 1999: 32-52). This type of
ethnography constitutes a kind of post post- modern kind.
Post postmodern ethnography
The type of ethnography that
embraces the anti-realist critique of ethnogra- phy but reflects only the
representational concerns of postmodernism is reflexive and only loosely
postmodern. This is clear from the response anti- realist ethnographers have
made to their own complaints. Whether it be 'subtle realism', 'analytical
realism' or my own guidelines for the 'ethno- graphic imagination', this kind
of anti-realist ethnography advocates the possibility and desirability of
systematic ethnography. The criticisms of naive realism still fall short of
postmodernism's abandonment of the idea of rigorous, disciplined and systematic
research practice. Thus, post post- modern ethnography remains rooted in weaker
versions of realism (for an example in cultural studies see Jenks and Neves
2000).
Subtle realism
Martyn Hammersley's (1990: 61,
73ff; 1992) account of subtle realism makes it clear that he believes in
independent truth claims which can be judged by their correspondence to an
independent reality. No knowledge is certain, but there are phenomena that
exist independent of us as researchers or readers, and knowledge claims about
them can be judged 'reasonably accurately' in terms of their 'likely' truth
(Hammersley 1990: 61). It shares with naive realism the idea that research
investigates independently know- able phenomena but breaks with it in denying
that we have direct access to these phenomena. It shares with anti-realism a
recognition that all know- ledge is based on assumptions and human
constructions, but rejects the notion that we have to abandon the idea of truth
itself (Hammersley 1992: 52). If the idea of truth itself is not abandoned,
what need to change are the criteria by which we judge truth claims. Gone
should be naive realist notions, and in their place should be a new construct
of validity, as well as other criteria within subtle realism, such as
relevance. Validity is understood by Hammersley (1990: 61-106) to describe
three processes that extend beyond mere 'accuracy': plausibility (whether any
truth claim is likely to be true given our existing knowledge); credibility
(whether any truth claim is likely to be accurate given the nature of the
phenomenon, the circumstances of the research and the characteristics of the
researcher); and evidence tests (where truth claims are not immediately
plausible or credible, the evidence to substantiate them will need to be tested
for its plausibility and credibility). This is a weak basis for evaluating
ethno- graphic data compared to the idea that we can assess claims directly by
their correspondence to 'reality', but this weakness must be accepted given
that postmodernism makes it impossible to envisage any direct correspondence.
Yet assessment of ethnographic data is not by their validity alone, for valid-
ity is joined by relevance (see Hammersley 1990: 64-70, 107-17). Ethnographic
findings must be not only valid but also relevant to issues of public concern.
All 'thick descriptions' should be for some purpose beyond simply 'telling it
like it is', and the descriptions can be evaluated against this agenda. As
Seale (1999: 12) notes, this is a less dramatic version of the claim by extreme
postmodernists that the quality of research should now be judged only by its
political effects (a claim made, for example, by Lincoln and Denzin 1994).
Ethnographic research could be judged on whether and how well it resolves some
social problem, or achieves emancipation for some oppressed group (such as
women) or release from some constraining situ- ation or setting (such as
discrimination experienced by ethnic minorities). Many feminist ethnographers
are particularly concerned to ensure that their practice ends up with the
emancipation of women rather than the produc- tion of valid knowledge for its
own sake (Miles 1983; Harding 1987; Williams 1990). Such praxis echoes that of
Marxist and critical realist eth- nographers. Hammersley (1990: 107) defines
two aspects of public relevance: the importance of the topic in terms of public
issues, and the con- tributions of the findings to existing knowledge. Again
this ensures that the relevance of ethnographic data is uncertain - reflecting
the uncertainty of the post-modern moment - because there will be disagreement
on these two dimensions, but this does not stop reasonable judgements being
made.
Analytical realism
Altheide and Johnson (1998:
291-4) argue that analytical realism is based on the view that the social world
is an interpreted world, not a literal one, always under symbolic construction
and reconstruction by people and by the eth- nographers who study them. While
the ethnographer's commitment is still to obtain people's perspectives on
social reality, analytical realism recognizes that most fields have multiple
perspectives and voices, which means that the ethnographer must faithfully
report this multivocality and show where his or her voice is located in relation
to these. All knowledge is perspectival (is rela- tive to the perspective of
the knower), so the ethnographer's perspective must be specified as much as
that of the subjects of the research. They call this
'validity-as-reflexive-accounting', and distinguish it from other forms of
validity (validity as relevance, as culture, as ideology and as language), all
of which are inadequate as the criteria to assess ethnographic data in the
contemporary postmodern, reflexive moment. In this way, representation and
legitimation are part of the same problem and solved in the same manner. If
ethnography represents the social world faithfully, evaluating its data becomes
unproblematic. This is what 'validity-as-reflexive-accounting' achieves,
placing the researcher, the topic, the subjects, the field, the sense- making
process and the written text at the heart of ethnography. Five pro- cesses are
seen as critical for the post postmodern and reflexive ethnographer to address
(Altheide and Johnson 1998: 291-2):
the relationship between what is
observed (behaviour, rituals, meanings) and the larger cultural, historical and
organizational contexts within which the observations are made; the
relationship between the observed, the observer and the setting or field; the issue
of the perspective or point of view used to render an interpre- tation of
ethnographic data, whether the observer's or the members'; the role of the
reader or audience in the final written product; the issue of the
representational, rhetorical or authorial style used by the ethnographer(s) to
render the description or interpretation.
Analytical realism thus calls for
a particular kind of validity, which requires that ethnographers substantiate
their findings with a reflexive account of themselves and the process of their
research. In this way, ethnographic research is privileged, or 'disciplined' as
Altheide and Johnson (1998: 293) write, compared to everyday thinking and
observing, allowing ethnography to rise above the morass and meaninglessness of
postmodern relativism and scepticism.
Critical realism
Critical realism is an attempt to
explain the relationship of social structure and social action and is grounded
in the work of Roy Bhasker. Bhasker (1989: 3-4) explains that social reality is
not created by people (the error of natural- ism), yet the structures that
pre-exist us do not occur independent of human agency (the error of
structuralism) but are reproduced and transformed by our action and everyday
activities. Structures are 'real'; their effects can be demonstrated in causal
connections in the material world even if such struc- tures cannot be perceived
outside of their effects. These structures also constrain agency. But they also
simultaneously enable agency by providing the framework within which people
act, and such agency reproduces (and occasionally transforms) the structure it
occurs within. The persistence of such structures across time and space
requires their continual reproduction by people in everyday activity. Critical
realism is thus very similar to Giddens's structuration theory (on which see
Giddens 1984), although it has a stronger empirical thrust compared to
Giddens's theory. Giddens has used the critical realist ethnography of Willis,
which was about young working-class school children in Birmingham, to support
his theoretical claims about structuration (1984: 289-93), and it is possible
to use ethnographic research to demon- strate some of the claims of
structuration theory (see Box 2.7). Likewise, Porter (1993, 1995) has used
ethnography to explore the dimensions and claims of critical theory, and has
appropriated critical real- ism to defend ethnography from its postmodern
critics. He presents this critique as imposing upon aspiring ethnographers four
obligations (Porter 1995: 16). These are: to make apparent the assumptions and
values that underlie the investigation; particularly to identify its
methodological basis; to make explicit the theoretical issues which the
research is designed to illuminate; and to make explicit the ontological status
that social structures are given. He contends that critical realism answers all
four queries, such that the point of ethnography is not to describe small-scale
social events but to examine human agency in order to shed light on the relationship
between social action and social structure. The imperative for the ethnographer
is to be reflexive; the agenda of the research is to focus upon 'generative
struc- tures' through close examination of human agency. While critical realist
ethnography keeps to the use of ethnography-as-fieldwork as a data collec- tion
technique, it abandons the naive realism of naturalism as the method- ology
associated with this research practice. Understanding the actor's viewpoint may
be a necessary condition for social knowledge, Porter writes (1995: 21), but it
is not a sufficient one; there are more than individual inter- actions and
interpretations that we need to know about. Davies's (1999) account of critical
realist ethnography also attacks naturalism.
The ethnographic imagination
In earlier work, I have also
responded to the postmodern critique of ethnog- raphy (Brewer 1994), arguing
that it does not rule out the possibility of sys- tematic and rigorous practice
but instead can be used to develop a set of guidelines for good ethnographic
practice in this reflexive, postmodern moment. Ethnographers need to be mindful
of an important requirement if their data are to be recognized as having
authority. No matter how good their practice, and irrespective of their reflexivity,
ethnographers need to deploy, and encourage readers to adopt, what I call the
'ethnographic imagination'. Atkinson (1990) uses the same term to describe the
artful and creative rhetorical abilities of writers of ethnographic texts, but
here it is used to describe the imaginative leap necessary to recognize the
authority of ethnographic data. This is not suggested as a means to ensure that
readers overlook bad practice or weaknesses in data, or that they make
allowances for this kind of research which they would not otherwise. Instead,
it is a call to openness in people's attitudes towards ethnographic data, in
which their validity, usefulness and import is not immediately dismissed out of
hand. The ethnographic imagination has three dimensions:
The belief that fragments of
recorded talk, extracts from field notes and reports of observed actions can
reliably represent a social world which cannot be completely described in the
restricted spatial confines of an ethnographic text, as long as the ethnographer
has been reflexive and thereby established his or her integrity and the
authority of the data. The belief that small-scale, micro events in everyday
life have at least common features with the broader social world, such that
general pro- cesses permeate down to and are in part reproduced at the level of
people's everyday lives. Thus, microscopic events can illustrate features of
broader social processes, as long as the ethnographer sets out the grounds on
which these empirical generalizations are made. The belief that people make
sense of their everyday lives, and offer descriptions and accounts thereof,
involving a complex reasoning process, which must be analysed if that social
world is to be understood in the round, although members' accounts should not
be taken at face value.
Box
2.7
Extract
from J. D. Brewer, Micro-sociology and the 'duality of structure': former
fascists 'doing' life history, in N. Fielding (ed.) Actions and Structure
(London: Sage, 1988), pp. 152-6.
[Elx-Fascists
are aware of the modern connotations of their membership. In asking the former
Fascists to look back on their membership and the reasons lying behind it, they
were forced to confront these typifications. In this way they became concerned
to present themselves as rational beings in face of the irrationality
common-sensically associated with their membership and support. This was
achieved through the notion of crisis: they presented their personal biography
as involving a tremendous crisis which made their membership of the BUF
[British Union of Fascists] a rational pragmatic act. The connection between
their support for the BUF and the perception of a crisis was mentioned by all
respondents. The respondents' emphasis on crisis presents Fascism as the last
chance, the only means of hope for themselves and for Britain. As knowledgeable
agents the former Fascists were able to monitor reflexively their actions
across time-space, to monitor reflexively the unintended consequences of past
conduct by Fascists and monitor how this past agency had become transformed and
reproduced into a series of pejorative common-sense typifications and
idealisations [about Fascists]. The common-sense typifications of Fascists as
'killers', 'mad', 'irrational' and so on, represent the objectification of the
past human agency of Fascists, and they become embodied as rules, recipes,
formulae and institutional practices for the behaviour towards and assessment
of Fascists. This objectification reflects and reinforces the typifications.
This objectification of past agency into a series of pejorative typifications
constitutes a constraint upon Fascists when they accomplish life history. They
are forced to confront and challenge them. This was achieved through the notion
of crisis. But simultaneously the constraints embedded in this objectification
are a medium through which the accomplishment of the life history is organised.
The constraints became an enablement because they provided the experiences
which the former Fascists had to confront and were the principle by which the
life history is organised as a practical achievement. This accomplishment of
life history leads to structuration over time-space. By transforming the common
sense typifications of Fascists into the theme of crisis, the former members of
the BUF reproduced the very characteristics of the wider common sense world
they drew on in their accomplishment, reproducing the view that Fascists really
are 'mad', 'crazy' and 'irrational'.
This 'ethnographic imagination'
is predicated on a set of guidelines for good practice which are integral to
it. These guidelines attempt to embody the reflexive, postmodern moment which
contemporary ethnography confronts, yet also to go beyond postmodernism by
re-establishing the grounds for reli- able, rigorous and systematic
ethnographic practice. Thus, in doing and writing up ethnographic research,
ethnographers should:
1 Establish the wider relevance
of the setting and the topic, and clearly identify the grounds on which
empirical generalizations are made, such as by establishing the
representativeness of the setting, its general features or its function as a
special case study with a broader bearing. 2 Identify the features of the topic
that they are addressing in the study and those left unresearched, and discuss
why these choices have been made and what implications follow from these
decisions for the research findings. 3 Identify the theoretical framework they
are operating within, and the broader values and commitments (political,
religious, theoretical and so on) they bring to their work. 4 Establish their
integrity as researcher and author, by outlining: the grounds on which
knowledge claims are being justified (length of fieldwork, the special access
negotiated, discussing the extent of the trust and rapport developed with the
respondents and so on); their background and experiences in the setting and
topic; their experiences during all stages of the research, especially mention-
ing the constraints imposed therein;
the strengths and weaknesses of
their research design and strategy. 5 Establish the authority of the data by:
discussing the problems that arose during all stages of the research; outlining
the grounds on which they developed the categorization system used to interpret
the data, identifying clearly whether this is an indigenous one used by
respondents themselves or an analyst- constructed one, and, if the latter, the
grounds which support this; discussing rival explanations and alternative ways
of organizing the data; providing sufficient data extracts in the text to allow
readers to evalu- ate the inferences drawn from them and the interpretations
made of them; discussing power relations within the research, between
researcher(s) and subjects and within the research team, in order to establish
the effects of class, gender, race and religion on the practice and writing up
of the research. 6 Show the complexity of the data, avoiding the suggestion
that there is a simple fit between the social world under scrutiny and the
ethnographic representation of it, by: discussing negative cases which fall
outside the general patterns and categories employed to structure the
ethnographic description, which often serve to exemplify and support positive
cases; showing the multiple and often contradictory descriptions proffered by
the respondents themselves; stressing the contextual nature of respondents'
accounts and descrip- tions, and identifying the features which help to
structure them.
Conclusion
Until very recently, ethnography
was conceived of as both a method (data collection technique) and a methodology
(a theoretical and philosophical framework). The philosophy of social research
suggests that the two were interlocked, with methodological preferences
predicting the employment of the method. For a long time, ethnographers saw
this interpolation of method and methodology as unproblematic; they were
fooling themselves. The ethnographic method became narrowly associated in the
social sciences with one methodological stance (naturalism or naive realism),
within which it was treated as a privileged technique, superior to all others.
Its weaknesses as a method were thus overlooked by proponents of the
methodology or set aside amid exaggerated claims for its utility. This was
particularly associated with 'humanistic ethnography'. Conversely, opponents of
naturalism within the natural science model of social research dismissed the
method more or less out of hand. If they accorded ethnography a role at all, it
was merely as a sensitizing tool for collecting the preliminary data necessary
to pursue the topic quantitatively. 'Positivist-scientific ethnography' sought
to accommo- date itself to complaint and developed what it thought was
objective scien- tific practice in ethnography. But ethnography has been
challenged more effectively recently by ethnographers who are reflecting the
anti-realism of this postmodern moment. They established a kind of 'postmodern,
reflexive ethnography', which abandoned both the claim that 'reality' could be
accu- rately represented ethnographically and the criteria by which
ethnography's truth claims could be assessed. However, 'post postmodern ethnography'
rescues it from the complete relativism and scepticism of postmodernism and
seeks to ground good practice of the method in a surer methodological
foundation than naturalism. They find this in a combination of naturalist- like
realism and postmodernism, expressed differently as 'subtle realism',
'analytical realism', 'critical realism' or the 'ethnographic imagination'. All
versions of post postmodern ethnography outline criteria for good practice in
order to distinguish systematic ethnography from lay persons' obser- vation. It
is to the question of good practice that we now turn.
Suggested further reading
For accounts of the contested
terrain in ethnography see:
Denzin, N. and Lincoln, Y. (1994)
The Handbook of Qualitative Research. London: Sage. Hamrnersley, M. (1992)
What's Wrong with Ethnography? London: Routledge. van Maanen, J. (1988) Tales
from the Field. Chicago: University of Chicago Press. Seale, C. (1999) The
Quality of Qualitative Research. London: Sage.
Comments
Post a Comment