Using Qualitative Methods
to Study Commonalities
Introduction
In some respects, qualitativeresearch does not seem as scientificas other
kinds of social research. Usually when we think of social science we
think of sweeping statements like: "People with more education tend to
get better jobs." "Poor countries tend to have more social conflict and
political instability than rich countries." These statements offer "big-pic-
ture" views that say nothing about individual cases. In these big-picture
views, a single statistic or percentage can summarize a vast amount of
information about countless cases.
But a lot may be missed in the big picture. Often, researchers do not
want these broad views of soeialphenomena because they believe that a
proper understanding can be achieved only through in-depth examina-
tion of specific cases. Jkdeed, qualitative researches often initiate re-
search with a conviction that big-picture representations seriously
misrepresent or fail to represent important socialphenomena.
Consider the researcher who wants to understand the fascinationthat
some people have with guns-for example, gun collectors, some military
personnel, hunters, and other enthusiasts. A big-pichue view might
show that certain categories of people (for example, lower middle-class
white males) are more likely to collect guns and subscribe to magazines
devoted to guns (Stinchcombe et al. 1980 study this question). But does
the big-picture view really say very much about the fascination with
guns? What's the best way to study and understand this fascination?
A lot can be learned simply by taking to gun enthusiasts. They can
be located in gun shops, gun clubs, and at practice ranges. The researcher
in this case might try to get to know as many as feasible and interview
them in depth. How did they get started with guns?How many guns do
they own? How often do they shoot them?How do they feel when they
are using them? How do they feelwhen they don't have easy access to a
gun? How many of their friends are gun enthusiasts? Do they feel that
law dorcement agenciesare effective?What do they think about capital
Určeno pouze pro studijní účely
punishment? What political organizations, if any, do they belong to?
How do they vote?
From these interviews it would be possible to build an image of at
least one major type of gun enthusiast, to craft a compositeimage based
on interviews of many individuals. This composite image could be
fleshed out further by studying the magazines and other literature that
the interviewees read and by observing what goes on at social gather-
ings of gun enthusiasts. The key would be to achieve as much in-depth
knowledge aspossible and look for common patterns among gun enthu-
siasts and their socialworlds.
Sometimes the emphasis of the qualitative approach on in-depth
knowledge means the researcher examines only a single case (for ex-
ample, the life history of a single individual or the history of a single
organization).Knowing as much as possible about one case is not easy
because every case potentially offers information that is infinite in its de-
tail. Much of this information is not useful because it is redundant or
irrelevant, given the researcher's questions and purposes. In the qualita-
tive approach, researchers must determine which information is useful
in the course of the in*estigation, and they becomemore selectiveas ad-
ditional knowledge about each case is gained. In the course of learning
more about the research subject, the investigator sharpens lus or her un-
derstanding of the case by refining and elaborating "images" of the re-
search subjectand relatingthese to analyticframes (seeChapter 3).These
emerging images serve to structure further inquiry by marking some
data collection paths as promising and others as dead ends.
Qualitative research often involves a process of reciprocal clnriicntioit
of the researcher's image of the research subject, on the one hand, and
the concepts that frame the investigation, on the other. Images are built
up from cases, sometimesby looking for similarities among several ex-
amples of the phenomenon that seem to be in the same general category.
These images, in turn, can be related to concepts. A concept is a general
idea that may apply to many specific instances. Concepts offer abstract
summaries of the properties shared by the members of a category of so-
cial phenomena. They are the key components of analytic frames, which,
in turn are derived from ideas--current theoretical thinking about social
life (see Chapter 3).
Consider a simple example first mentioned in Chapter 1: "Emotion
work" is a concept developed by Arlie Hochschild (1983)to describe the
conscious manipulation of feeling to create a publicly observable facial
and bodily display. This concept, in effect, summarizes a lot of what
flight attendants do because they often have to create certain appear-
Using QllnlitntiueMetllods toStirdy Coll~nlo~lnlities83
ances for passengers.Her study of flight attendants thus involved a mu-
tual clarification of the categoty "the work of flight attendants" and the
coilcept of "emotion-work." She refined the image of the flight attendant
(anempiricalcategory) as she clarified the concept of emotion work. This
process of reciprocal clarification is ongoing and culminatesin the repre-
sentation of the research that the investigator offers at the conclusion of
the study. Thenewly refined concepts-those that were elaborated in the
course of the study-are featured in the representation of the results of
qualitative research.
The Goals of Qualitative Research
Because of its emphases on in-depth knowledge and on the refinement
and elaborationof images and concepts, qualitative research is especially
appropriate for several of the central goals of social research. These in-
clude giving voice, interpreting historical or cultural signiicance, and
advancing theory.
Giving Voice t
There are many groups in society, called marginalized groups by social
scientists, who are outside of society's mainstream, for example-the
poor, sexual minorities, racial and ethnic minorities, immigrant groups,
and so on. Often, these groups lack voice in society. Their views are
rarely heard by mainstream audiencesbecause they are rarely published
or carried by the media. In fact, their lives are often misrepresented-if
they are represented at all.
Techniques that help uncover subtle aspects and features of these
groups can go a long way toward helping researd~ersconstruct better
representations of their experiences. By emphasizingdose, in-depth em-
pirical study, the qualitative approach is well suited for the difficult task
of representing groups that escape the grasp of other approaches.
I7tterpretiltg Historicnlly or Cz~ltumllySign$caltt Pheitolizetta
How we think about an important event or historic episode affects how
we understand ourselves as a society or as a nation. For example, in the
middle to late 1800sthe United Stateswas involved in a series of temto-
rial struggles with Mexico. These struggles can be interpreted as part of
the inevitablewestward expansion of European-Americansacross a vast,
Určeno pouze pro studijní účely
sparsely populated continent. Or perhaps they can be seen as part of a
pattern of unjust bullyingof a generallypeacefulneighbor.As the United
States gains an ever larger Hispanicpopulation, a revision of our under-
standing of these territorialstruggles may help us adjust our view of the
diverse collection of people who make up American society.
Methods that help us see things in new ways facilitate this goal of
interpreting and reinterpreting significant historical events. Of course, if
the evidencedoesnot stronglysupport a new image, or offersbetter sup-
port to existingimages, then new ways of understanding past events will
not gain wide acceptance. The important point is that the qualitativeap-
proach mandates close attention to historical detail in the effort to con-
struct new understandings of culturally or historically significant
phenomena.
Advn~zcingTheory
There are many ways to advance theory. New informationabout a broad
pattern that holds across many cases (for example, a strong correlation;
see Chapter 6) canstimulatenew theoretical thinking. However, in-depth
knowledge--the kind that comes from case studies-provides especially
rich raw material for advancing theoretical ideas. When much is known
about a case, it is easier to seehow the differentparts or aspects of a case
fit together.
For instance, it is difficult to know how the structure of a nun's daily
routines of prayer, work, and communitylifehelp her maintainher deep
religious commitments without collecting detailed observations of the
lives of nuns. Thisin-depth knowledge is useful for elaboratingconcepts
such as "commitment" and for direct examination of the connections
among the phenomena that the researcher believes illustrate and elabo-
rate the concept, for example--the daily routines of those with strong
commitments.
The value of qualitative research for advancing theory also follows
directly frompractical aspects of this type of research. It is impossible to
decide which bits of evidence about a case are relevant without clarify-
ing the concepts and ideas that frame the investigation.The initial goal
of knowing as much as possible about a case eventually gives way to an
attempt to identify the features of the case that seem most significant to
the researcher and his or her questions. This shiftrequires an elaboration
and refinement of the concepts that initially prompted the study or the
development of new concepts. Researchers cannot forever remain open
to all the information that their cases offer. If they do, they are quickly
Using QrralitatiueMetlwds to Shrdy Conrnronalities 85
overwhelmed by a mass of indecipherable and sometimesconhadictory
evidence.
Finally, qualitative research also advances theory in its emphasis on
the commonalities that exist across cases. In some studies cases may be
selected that at first glancemay seemvery different.Identifyingcommon-
alities across diverse cases requires that the investigator look at the cases
in a different way and perhaps discover new things about them. Diane
Vaughan's study Ui~cotrpliizg(1986), for example,focused not only on the
breakup of conventional relationships-heterosexual marriages-but also
onthe breakup of homosexual relationships.Despiteprofound differences
in the sexual orientations of her subjects, Vaughan found shiking simi-
larities in the process of "uncoupling" across these differentkinds of rela-
tionships. By looking for similarities in unexpected places, social
researchers developnew insights that advance theoretical thinking.
The Process of Qualitative Research
Qualitative research is often less structured than other kinds of socialre-
search. The investigator initiates a study with a certain degree of open-
ness to the research subject and what may be learned from it. QuaLitative
I
researchers rarely test theories. Instead, they usually seek to use one or
more cases or categories of cases to develop ideas. The qualitative re-
searcher starts out by selecting relevant research sites and cases, then
identifies "sensitizing concepts," clarifies major concepts and empirical
categories in the course of the investigation, and may end the project by
elaborating one or more analytic frames.
Selectilig Sites nizd Cnses
Qualitativeresearch is stronglyshaped by the choice of research subjects
and sites. When the goal of the researchis to give voice, a specific group
is chosen for study. When the goal is to assess historical or cultural sig-
nificance, a specific set of events or other slice of social life is selected.
When the goal is to advance theory, a case may be chosen because it is
unusual in some way and thus presents a special opportunity for the
elaboration of new ideas.
Sometimes,however, cases are chosen not because they are special or
unusual or significantin some way, but because they are typical or un-
distinguished. A researcher interested in medical schools in general, for
example, might select a school that is typical or average, not the best
Určeno pouze pro studijní účely
medical school in the country or the worst (see Becker et al. 1961). To
select a school at either extreme might limit the value of the study for
drawing conclusions about medical schools in general. In short, because
qualitativeresearchers oftenwork with a smallnumber of cases, they are
sometimes very concerned to establish the representativeness of the cases
they study (see Chapter 1).
In-depth knowledge is sometimes achieved through the study of a
single case. Often, however, it is best achieved by studying several in-
stances of the same thing because different aspects may be more visible
in different cases. Consider a study of a neighborhood with many new
immigrants.The researcher might find that in this neighborhood the cul-
tivation of interpersonalnetworks (that is, making connectionswith lots
of differentpeople)is the key to the successfuladjustment of immigrants
to the United States. Much can be learned fromstudying one suchneigh-
borhood in depth. In fact, it is only through in-depth study that immi-
grants' use of interpersonal networks could be thoroughly documented.
However, the study could be deepened further through the study of sev-
eral immigrant neighborhoods. There may be various ways of establish-
ing a reliable interpersonal network, depending on the cultural
backgrounds of the immigrants.Different ways of establishinginterper-
sonal networks might be more apparent in other neighborhoods.
When qualitative researchers collect data on many instances of the
phenomenon under study, they focus on what the different instances
have in common. Examining multiple instances of the same thing (for
example, interviewing thirty-five flight attendants) makes it possible to
deepen and enrich a representation (forexample, a representation of the
emotion work required in service jobs). A study of environmental activ-
ists might focus on the life experiences they share. A study of Catholic
priests might focus on how they maintain their religious commitments.
A study of immigrant neighborhoodsmight focus on the different ways
of establishing and using interpersonal networks to facilitate immi-
grants' adjustment to their new surroundings.
When many instances of the same thing are studied, researchers may
keep adding instances until the investigation reaches a point of satura-
tion. The researcher stops learning new things about the case and re-
cently collected evidence appears repetitious or redundant with
previously collected evidence. It is impossible to tell beforehand how
many instances the researcher will have to examine before the point of
saturation is reached. In general, if the researcher learns as much as pos-
sible about the research subject, he or she will be a good judge of when
this point has been reached.
Of course, if the cases selected for study are not sufficientlyrepresen-
tative of the category the qualitative researcher hopes to address, then
the point of saturation may be reached prematurely. A study that seeks
to represent the work of taxi drivers in New York City may reach satura-
tion (nonew things are being learned) after the researcherinterviews ten
taxi drivers who are recent immigrants from Romania. However, these
ten Romanian taxi drivers are probably not representative of all New
York taxi drivers. The researcher should seek out taxi drivers with differ-
ent backgrounds.
Even when qualitative researchers study many instances of the same
thing (as when fifty priests are intemiewed, for example), they often de-
scribe the case as singular ("the case of Catholicpriests") because the fo-
cus is on commonalities-features that the instances share. By contrast, a
quantitative researcher (see Chapter 6) interested in systematic differ-
ences (say, the covariation between age and strength of religious com-
mitments among these same priests) would emphasize the fact that the
research summarizes information on 111n11y cases (fiftypriests). State-
ments about patterns of covariation (for example, "older priests appear
to be more committed than younger priests") are more likely to be ac-
cepted if they are based on as many cases as possible.
This distinction is subtle bu; very important. The qualitative re-
searcher who interviews fifty priests seeks to construct a fullporbait of
"the priest" and how priests maintain their deep religious commitments.
It may be that the images that emerged changed very little, if at all, after
the tenth priest was interviewed, and not much was learned from the
remaining forty priests. The difference between ten and fifty is not im-
portant; what matters is the soundness of the portrayal of this case (the
Catholic priest). If a study is done properly and is based on a sufficient
number of interviews,it canbe used for comparisonwith other cases (for
example, comparingpriests with the ministers of a Protestant denomina-
tion). The important point is that even though many examples of the
same thing may be examined, research that emphasizes similaritiesseeks
to construct a single, composite portrait of the case.
Use of Selzsitiziilg Coizcepts
It is impossible to initiate a qualitativestudy without some sense of why
the subject is worth studying and what concepts might be used to guide
the investigation.These concepts are often drawn from half-formed,ten-
tative analytic frames, which typically reflect current theoretical ideas.
These initial, sensitizing concepts get the research started, but they do not
Určeno pouze pro studijní účely
straitjacket the research. The researcher expects that these initial con-
cepts, at a minimum,will be altered significantlyor even discarded in
the course of the research.
A researcher studyinghospital patients may bring "social class" as a
sensitizing concept to the research and expect to find that patients from
families with more income receive better care. However, the concept of
social class, as expressed in family income, might prove to be too limit-
ing as a frame for the research and be supplanted by an emphasis on
some other aspect of family social status, such as occupational prestige
of the head of the household. Sometimes concepts that seem important
or useful early in the study prove to be dead ends, and they are dis-
carded and replaced by new concepts drawn from different frames.
Armed with these new concepts, the researcher may decide that some of
the evidence that earlier seemed irrelevant needs to be reexamined.
For example,John Walton (1991,1992)studied the conflict over water
rights in Owens Valley, California, a struggle that pitted the residents of
Owens Valley against water-hungry Los Angeles. (This struggle pro-
vided the background for the movie Clzit?ntowtl,starringJack Nicholson.)
The battle over water rights dragged on for decades and generated so
much mass protest and collective violence that it became known as
"California's dirty little civil war." At first, Walton tried to use concepts
that centered on social class and class conflict to understand this
struggle.These were his initial, sensitizingconcepts. He found that these
concepts did not help him make sense of the evidence that he collected,
nor did they direct him down data collection paths that advanced the
study. Eventually he came to understand the struggle more in terms of
collective responses anchored in local conditions to changing govem-
mental structures,especiallythe growinginfluenceand power of the fed-
eral government. These new concepts directed him to important
historical evidence that he might have overlooked otherwise.
ClnrifiJi7zgConcepts ai~dCategories
Qualitative research clarifies concepts (the key components of analytic
frames)and empiricalcategories (whichgroup similarinstances of social
phenomena) in a reciprocal manner. These two activities, categorizing
and conceptualizing, go hand in hand because concepts define catego-
ries and the members of a category exemplify or illustrate the concepts
that unite them into a category.
Generally,the members of a category are expectedto be relativelyho-
mogeneous with respect to the concepts they exemplify. If a researcher
Using Qtlnlitntiue Methods to Shrdy Cotttntot~nlities 89
found that only some flight attendants engage in emotion work, then it
would be wrong to use the conceptto characterizeflight attendants. Sup-
pose a researcher studying flight attendants found that only those flight
attendants hired after a specificpoint in time engage in a lot of emotion
work. It might be possible to trace this to a change in the training of
flight attendants and perhaps to a conscious attempt by management to
alter how flight attendants interact with passengers. The lack of fit be-
tween the concept "emotion work" and the broad category "all flight at-
tendants" in this event would enrich the study, making it possible to
nanow the relevant category to a subset of flight attendants--those sub-
jected to a specifickind of training-and showing a direct connection to
management intervention.
This example shows the importance of examining the members of a
category to make sure that they all display the concepts they are thought
to exemplify. Researchers develop conceptsfrom the images that emerge
from the categories of phenomena they examine. They then test the lim-
its of the concepts they develop by closely examining the members of
relevant categories. In the example just presented, the concept of emo-
tion work emerged from images of flight attendants constructed by the
investigator. Subsequent examination of all flight attendants-to see if
they all engage in emotion work-would establish the limits of the rel-
evant category.
Consider a second example of the interachon of categories and con-
cepts,Howard Becker's (1953)early study of becoming a marijuana user.
Becker studied several marijuana users and found that each went
through a process of lennlitlg to become a user-of leaming lzozu to enjoy
marijuana. This led him to speculate that all lnnrijttnltn lisers (the cat-
egory) go through a socinl process of lennlillg (the concept) to enjoy mari-
juana. He elaborated the key steps in the process of becoming a user by
intemiewing more than fifty users in the Chicago area in the early 1950s.
He found that most, more or less, went through the same process of
learning how to enjoy marijuana.
However, Becker did encounter a few users who did not go through
this process, and, although they were users, they said that they did not
enjoy the drug. Becker described them as people who used marijuana for
the sake of appearance--in order to appear to be a certainkind of person
or to "fit in" with the people around them. Did this invalidate the idea
that all users go through the same leaming process? Becler solved the
problem by narrowing the relevant category. He argued that the social
process of learning how to enjoy marijuana applied only to those who
used marijuana for pleasure, a category that embraced most, but not all,
Určeno pouze pro studijní účely
users. This narrowing made it possible for him to establish a closer cor-
respondence between category (those who use marijuana for pleasure)
and concept (the socialprocess of learninghow to use marijuana).
These examples show that the core issue in the clarification and
elaborationof categories and concepts is the assessment of the degree to
which the members of a categoryexemplifythe relevant concept.Are the
same elements present in each instance in more or less the same way?
When encountering contradictory evidence (for example, flight atten-
dants who don't do emotion work or marijuana users who did not go
through the socialprocess of learninghow to enjoymarijuana),research-
ers have two choices. They can discard the concept they were develop-
ing and try to developnew ones--concepts that do a betterjob of uniting
the members of the category. Or they can narrow the category of phe-
nomena relevant to their concept and try to achieve a better fit with the
concept.
Elnbornfing Annlytic Frnllzes
Because categories and concepts are clarified in the course of qualitative
research, the researcher may not be certain what the research subjectis a
"case of" until all the evidence is collected and studied. Deciding that
the research subject is a case of something and then representing it that
way is often the very last phase of qualitativeresearch.
The open character of qualitative research can be seen clearly in the
role played by analytic frames in this strategy. In some research strate-
gies (for example, quantitative research; see Chapter 6), the main pur-
pose of the analyticframeis to express the theory to be tested in terms of
the relevant cases and variables.In qualitativeresearch, by contrast, there
is often only a tentative, vaguely formulated analytic frame at the outset
because it is developed in the course of the research.
As more is learned about the cases and as categories and concepts are
clarified, the researcher can address basic questions: What is this case a
case of?What are its relevant features? What makes the chosen research
subject or site valuable, interesting, or significant? As qualitative re-
searchers elaborate analytic frames, they also deepen their understand-
ing of their cases. To describe the work of flight attendants as a case of
emotion work (Hochschild 1983) suggests that there are other jobs that
also require emotion work (for example, tour guides, camp counselors,
waitresses, and so on) and that the emotion-work frame developed in
the study of flight attendants may be applied to these other people-ori-
ented service occupations.
Usb~gQtrnlitntiueMetlwds to Shrdy Cotttntonnlities 91
Not all qualitative researchers develop analytic frames. Sometimes
they leave this task to other researchers studying related cases. The de-
velopment of analytic frames is challenging because it requires the ex-
tension of the concepts elaborated in one case to other cases. Many
qualitative researchers are content to report detailed treatments of the
cases they study and leave their analytic frames implicit and unstated.
They feelthat their cases speak well enough for themselves.
This unwillingness to generalize is found in all types of qualitative
research, from observations of small groups to historical interpretations
of the international system. For this reason, qualitative researchers are
often accused of being "merely descriptive" and not "scientific" in their
research. As should be dear by now, however, the process of represent-
ing research subjects is heavily dependent on the interaction between
concepts and images, regardless of whether this interaction is made ex-
plicit by researchers when they represent their subjects. Without con-
cepts, it is impossible to select evidence, arrange facts, or make sense of
the infinite amount of informationthat canbe gleaned from a single case.
Like other forms of social research, qualitative research culminates in
theoretically structured representations of social life-representations
that reflect the regimen of socialresearch.
Using Qualitative Methods
There are many textbooks on qualitative methods, and they describe
qualitative methods in a variety of ways (see for example Denzin 1970,
1978; Glaser and Strauss 1967;McCall and Simmons 1969; Strauss 1987;
Schwartz and Jacobs 1979).In part, this diversity of views follows from
the emphasis on in-depth investigation and the fact that there are many
different ways to achieve in-depth knowledge. In sociology, anthropol-
ogy, and most other social sciences, qualitative methods are often identi-
fied with participant observation, in-depth intemiewing, fieldwork, and
ethnographic study. These methods emphasize the immersion of the re-
searcher in a research setting and the effort to uncover the meaning and
significanceof socialphenomena for people in those settings.These tech-
niques are best for studying social situations at the level of person-to-
person interaction.
For an anthropologist, this immersion might involve living in some
isolated village in some faraway part of the world. Consider,for example
Margaret Mead's work Coltring of Age ill Smnoa (1961).For a sociologist,
immersionmight involvelongperiods of observing and talkingto people
Určeno pouze pro studijní účely
in one setting,such asErving Goffman's research on the staff and patients
of a mental institution, reported in his classic study Asyltims (1961).In
both examples, the organizingprincipleof the researchis the idea that the
kind of in-depth knowledge needed for a proper representation of the
research subject must be based on the perspectives of the people being
studied-that their lives and their worlds must be understood "through
their eyes." In short, the emphasis is on immersion and empirical inti-
macy (liuzzi1974).
The goal of this presentation of qualitative methods, however, is to
address procedures that are relevant to all types of qualitative research,
not simply the work of those who seek to represent social life and as it
appears through the eyes of participants. Researcherswho seek to repre-
senthistorically signiicant events, for example, cannothope to see these
events through the eyes of the participants if these eventsoccurredin the
distant past (the French Revolution, for example, or slavery in the U.S.
South).Still, these historical researchers, like others who use qualitative
methods, value and seek in-depth knowledge about cases, and they at-
tempt to piece togethermeaningful images from evidence,with the help
of concepts and analytic frames.
The key features common to all qualitative methods can be seen
when they are contrasted with quantitative methods. Most quantitative
data techniques are datn condelzsers. They condense data in order to re-
veal the big picture. For example, calculating the percentage of union-
ized workers who vote for the Democraticparty condenses information
on thousands of individuals into a single number showing the link be-
tween these two attributes (union membership and party preference).
Qualitative methods, by contrast, are best understood as data e~zhancers.
When data are enhanced, it is possible to see key aspects of cases more
clearly, depending on how it is done.
In many ways, data enhancementis likephotographic enhancement.
When a photograph is enhanced, it is possible to see certain aspects of
the photographer's subject more clearly. When qualitative methods are
used to enhance social data, researchers see things about their subjects
that they might miss otherwise. Data enhancementis the key to in-depth
knowledge.
Almost all qualitative research seeks to construct representations
based on in-depth, detailed howledge of cases, often to correct misrep-
resentationsor to offernew representationsof the research subject. Thus,
qualitativeresearchers share an interest in procedures that clarify key as-
pects of research subjects-procedures that make it possible to see as-
pects of cases that might otherwise be missed. While there are many such
Usirlg Qr~nlitntiueMethods to Sh~dyCorrtmonnlities 93
proced~es,two that are common to most qualitative work are empha-
sized here: analytic induction and theoretical sampling. Both techniques
are data enhancers.
Annlytic l~zdz~ctio~z
Analytic induction means very different things to different researchers.
Originally, it had a very strict meaning and was identified with the
search for "universals" in social life (Lindesmith 1947; Cressey 1953;
Turner 1953;Robinson 1951).Universalsare properties that are invariant.
If all upper middle-classwhite males over the age of fifty in the United
States voted for the Republican party, then this would constitute a "uni-
versal." Ionly one person in this category voted for some other party,
then the pattern would not be universal and thus would not qualify as a
finding, according to a very strict,very simple-minded application of the
method of analytic induction. Today, however, analytic induction is of-
ten used to refer to any systematic examination of similarities that seeks
to develop concepts or ideas.
Rather than seeing analytic induction as a search for universals, a
search that is likely to fail, it is better to see it as a research strategy that
directs investigators to pay close attention to evidence that challenges or
disconfirmswhatever images they are developing. As researchers accu-
mulate evidence, they compare incidents or casesthat appear to be in the
same generalcategorywith each other.These comparisonsestablishsimi-
larities and differences among incidents and thus help to define catego-
ries and concepts. (SociologistsBarney Glaser and Anselm Strauss call
thisprocess the constant comparativemethod.) Evidence that challenges
or refutes images that the researcher is constructing from evidence pro-
vides important cluesforhow to alter concepts or shift categories.
A study in a hospital might examine the care given to dying patients.
By comparing cases of thistype, the resea;cher can identify common fea-
tures and the major dimensions of variation among incidents. Based on
hours of observing the care of dying patients, a researcher might find:
1. that nurses and other hospital personnelimplicitly evaluate the
potential "social loss" represented by each patient if the patient were
to die
2. that a small number of patient characteristicsenter into thisevalua-
tion (for example, the age and education of the patient)
3. that the quality of patient care depends on the potential social loss
inferred by the hospital personnel
Určeno pouze pro studijní účely
Incidents that challenge either the generality of the evaluation of the
social loss of dying patients or the impact of this evaluation on the care
patients receive would be especially important for refining these ideas.
In the next phase of the research, the investigator might seek out
disconfhming evidence (forexample, a patient who is judged to be not
much of a "social loss" but nevertheless receives excellent care) to test
these initial images and see how they need to be revised or limited. If,
for example, the researcher found that hospital personnel ignored the so-
cial loss representedby accident victims, then he or she would be forced
either to reformulate the image to accommodate accident victims or else
limit its applicability to nonaccident patients.
In effect, the method of analytic induction is used both to construct
images and to seek out contrary evidencebecause it sees such evidence
as the best raw material for improving initial images. As a data proce-
dure, this techniqueis less concerned with how much positive evidence
has been accumulated (forexample,how many cases corroborate the im-
age the researcher is developing), and more with the degree to which the
image of the research subject has been refined, sharpened, and elabo-
rated in response to both confirming and disconfirming evidence.
Analytic induction facilitates the reciprocal clarification of concepts
and categories, a key feature of qualitative research. When Howard
Becker narrowed his category from "all marijuana users" to "those who
use marijuana forpleasure," he used the techniqueof analytic induction.
Essentially, the technique involves looking for relevant similarities
among the instances of a category, and then linking these to refine an
image (forexample, the image of how one becomes a marijuana user). If
relevant similarities cannot be identified, then either the category is too
wide and heterogeneous and should be narrowed, or else the researcher
needs to take another look at the evidence and reconceptualize possible
similarities. Negative cases are especially important because they are ei-
ther excluded when the relevant category is narrowed, or they are the
main focus when the investigator attempts to reconceptualize common-
alities and thereby reconcile contradictoryevidence.
Consider a more detailed example: Jack Katz (1982)studied legal as-
sistance lawyersthose who help poor people. He found that many le-
gal assistance lawyers bum out quickly-in less than two years-and
abandon this kind of work, often for more lucrative legal careers. Katz
wanted to understand why by studying those who stayed with legal as-
sistancework despite its drawbacks. He assembled evidence on the legal
assistance lawyers in the group he studied and checked out several of
Using QtralitntiveMetlrods to Study Cotnntonnlities 95
his initial ideas by comparing those who had quit before two years of
service with those who had stayed on for more than two years.
One of the first ideas Katz examined was based on his initial impres-
sions of these attorneys.He speculatedthat legal assistancelawyers who
were former political activists did not bum out like the others. Asystem-
atic examination of the evidence on many lawyers provided some sup-
port for this speculation. However, the fit was far from perfect. There
were some who stayed with legal assistance work who were not former
political activists, and there were former political activists who left legal
assistance work before two years had elapsed.
Katz examined these negative cases closely and found someproblems
with his initial formulation. Some former activists left for obvious rea-
sons.They were offeredpositions that were clearly a step up, careerwise.
Some who were not former activists stayed because they lacked altema-
tives-they couldn't get better jobs as lawyers--or because they had po-
sitions in the organization that they liked (such as administrative
positions).
It was clear to Katz that his categories "staying versus leaving" had
to be refined and that his search for adequate explanatory concepts was
far from over. First,he narrowed the category that interestedhim most-
those who stayed. Clearly be was not interested in all stayers. Some
stayers, after all,had interestingwork within the legal assistance organi-
zation he studied. Rather, he was interested in people who stayed de-
spite being involved in frustrating or limiting work. He reshicted his
focus to this subset of stayers and searched for relevant similarities
within this group.
With this shift he became less interested in all stayers versus all
leavers and more interested in differences between categories of
stayers--those who stayed despitefrustratingwork versus other stayers.
In short, the focuswas on how people stayed,and he had straightforward
explanationsformany stayers (forexample, those with interestingwork).
As it turned out, this tighter category-stayers with frustrating work-
also proved to be too broad, and he later narrowed it further to legal
assistance lawyers who were involved in low-status work. After all,
some lawyers doing significant work, he discovered, were nevertheless
frustrated with their work.
The search for explanatoryfactorsbecame more focused as the main
category of interest narrowed. After rejecting "activist background as
an explanation for staying, Katz hied to distinguish lawyers who were
more oriented toward using the legal system for reform from those who
Určeno pouze pro studijní účely
96 Clrnpter 4
were less so. He also looked at the participation of lawyers in social ac-
tivities that celebrated reform work (for example, progressive political
groups). This search for important commonalities among stayers went
hand in hand with narrowing the relevant category of stayers from all
stayers to those who were involved in low-status work.
The process of narrowing and refiningis depicted in Table4.1, which
shows the process of analytic induction in tabular form based on Jack
Katz's description. The table reports hypothetical information on thirty
lawyers to illustrate the generalprocess he describes, not his specificcon-
clusions. The first three columns show the narrowing of the category of
stayers, from all stayers (column 1;18out of 30 lawyers) to stayers with
frustrating work (column2; 13out of 30 lawyers), to stayers involved in
work that carriedlow status (column3; 10out of 30 lawyers). Columns 4
through 6 show the various ways Katz tried to explainstaying-his vari-
ous images of the "stayet" As his focus shifted from column 1to column
2 and then to column 3, he became more interested in how and why
people stayed and less in the differencebetween stayers and the twelve
leavers at the bottom of the table. In other words, he came to view stay-
ing as an accomplishment for those doing low-status work and studied
how it was accomplished.
First, Katz tried to construct an image of staying as a continuation of
a commitmentto political activism (column4).As the hypothetical data
in Table 4.1 show, this image fails. Of the eighteen lawyers who stayed
more than two years, only seven were former activists, and of the twelve
who left the organization, four were former activists.Next, Katz studied
his negative cases closely (especially,nonactivists who stayed) and found
that his categorization of stayers versus leavers was too crude. He rea-
soned that what really interested him most was people who stayed de-
spite their involvement in frustrating work. He then tried to find
commonalities among this subset of stayers, looking at their reform ori-
entations and their participation in a social life supportive of reform
work. The fit was still not close enough. There were some lawyers who
did frustrating work, for example,who were not reform oriented.
Examination of negative cases led to a further narrowing of the cat-
egory-to lawyers involved in low-status work-and further refinement
of the image--to participation in a social environment that glorified re-
form work. These further refinements resulted in a good fit. The data in
the table suggest that legal aid lawyers will do low-status work if they
participate in a social environment that glorifies the idea that important
socialreforms can be achieved through the legal system.
Using Q~ralitatiueMethods to Study Cornmorrnities 97
T A B L E 4.1
Hypothetical Example of Analytic Induction
Categories E~lnr~otonjConcepts
1 7 3 4 5 6- -
Socinl Life
Stayed More Works in n b~uolnedill Supports
nrnf:Two Fnrstrntirrg Low-Stntra Actiuist Refornt Refonrr
Cnse Yenrs? Ploce? Work? Bnclrgmsed? Oriented? Orientofion?
1 yes Yes Yes Yes Yes Yes
2 yes
3 yes
4 yes
5 Ye=
6 yes
7 yes
8 yes
9 yes
10 yes
11 yes
12 yes
13 yes
14 yes
15 yes
16 yes
17 yes
18 yes
19 no
20 no
21 no
22 no
23 no
24 no
25 no
26 no
27 no
28 no
29 no
30 no
Určeno pouze pro studijní účely
Columns 3 and 6 correspond perfectly. In fact, most qualitative re-
searchers are satisfied with less than perfectfit. There is usually at least a
handful of extraneous evidence that neither fitsnor challengesa particu-
lar image. The goal is not perfect fit, per se, but a conceptual refinement
that provides a deeper understanding of the research subject. Basically,
the greater the effort to account for or understand negative cases or con-
trary evidence,the deeper the understanding of the researchsubject. The
technique of analytic induction thus facilitates the goal of in-depth
knowledge.
Katz comments that analytic induction is poorly labeled because it is
not a technique of pure induction. Researchers work back and forth be-
tween their ideas and their evidence, trying to achieve what Katz calls a
"double fitting" of explanationsand observations (thatis, ideas and evi-
dence). As discussed in Chapter 3, this process of double fitting is best
understood as retroduction, a term that describes the interplay of induc-
tion and deduction in the process of scientificdiscovery.
T\zeo~.eticnlSniilpliilg
Sometimes qualitativeresearchers conduct investigationsof related phe-
nomena in several different settings.Most often this interest in a broader
investigationfollowsfrom a deliberatestrategy of theoretical sampling,
a term coined by Barney Glaser and Anselm Strauss (1967) to describe
the process of choosing new research sites or cases to compare with one
that has already been studied. For example, a researcher interested in
how environmentalactivists in the United Statesmaintain their political
cornmihents might extend the study to (1)environmental activists in
another part of the world (forexample,EasternEurope)or perhaps to (2)
another type of activist (for example religious activists in the United
States).
This process of theoretical sampling occurs not only in the study of
social groups (forexample,environmentalactivists),but also in the study
of historical processes and episodes. General questions that arise in a
study of the Russian Revolution of 1917might be addressed by examin-
ing the Chinese Revolution of 1949or the recent Nicaraguan Revolution.
There may be questions about the role of peasants in the Russian Revo-
lution that could be answered by examining the Chinese case and com-
paring it to the Russian case.
The choice of the comparison group (comparingenvironmentalactiv-
ists in the United States with either environmental activists in Eastern
Europe or with people in the United States who maintain radical reli-
gious commitments)can vary widely depending on the nature and goals
of the investigation.Different comparisonshold different aspectsof cases
constant. Comparing environmentaland religious activists in the United
States holds some things constant such as the impact of national setting,
but allows the nature of the commitment to vary (environmentalversus
religious). Comparing environmentalactivists in the United States with
environmental activists in Eastern Europe highlights the impact of the
factor that varies most, national setting,but holds the nature of the com-
mitment, environmental,constant.
When a researcher employs a strategy of theoretical sampling, the se-
lection of additional cases is most often determined by questions and is-
sues raised in the first case studied. Selectionof new cases is not a matter
of convenience; the researcher's sampling strategy evolves as his or her
understanding of the research subject and the concepts it exemplifies
matures. The goal of theoretical sampling is not to sample in a way that
captures all possible variations, rather in one that aids the development
of conceptsand deepens the understanding of research subjects.
A researcher studying how hospital personnel evaluate the potential
social loss of dying patients and link the care they give to these evalua-
tions might believe that this practice is caused by limited resources in the
hospital studied. If the hospital had more resources (for example, more
nurses), it might be able to provide better and more uniform care to all
patients, regardless of their social value. To explore this idea, the re-
searcher might study two additional hospitals, one with more resources
and one with fewer resources than the first hospital. If the reasoning
based on the first hospital is correct, then the staff of the hospital with
more resources should spend less time evaluatingthe socialloss of dying
patients and provide more uniform care, while the staff of the hospital
with fewer resources should spend more time evaluating social loss and
should adjust their carein more strict accordancewith these evaluations.
This expansion of the study to two new sites is a straightforward
implementation of the idea of theoretical sampling. The selection of the
new sites follows directlyfrom ideas developed in the first site and pro-
vides an opportunity to confirm and deepen the insights developed in
that setting. Of course, if research in these new settings were to contra-
dict expectations based on research in the first hospital, then the re-
searcher would be compelled to develop a different understanding of
how and why hospital personnel varied their care of dying patients.
This example of theoretical samplingalso shows that it is a technique
of data triangulation (Denzin 1978). Triangulation is a tenn that origi-
nally described how sailors use stars and simple trigonometry to locate
Určeno pouze pro studijní účely
100 Chnpter4
their position on earth. More generally,hiangulation can be understood
as a way of using independent pieces of information to get a better fix on
something that is only partially known or understood. In the example
just presented, the researcher used evidence from two other hospitals,
one with more resources and one with fewer, to get a better fix on the
irst hospital. By comparing the three hospitals, arrayed along a single
continuum of resources, the researcher could assess the validity and gen-
erality of findings from the firsthospital.
Theoretical sampling is also a powerful technique for building ana-
lytic frames. Helen Rose Fuchs Ebaugh (1977)studied ex-nuns-women
who left Catholicreligious orders-and used thisgroup of women to de-
velop the concept of "role exit," in much the same way that Arlie
Hochschild used her study of flight attendants to develop the concept of
emotion work. Ebaugh became interested in people whose current self-
identitieswere stronglyinfluencedby the roles they had left behind. This
interest led her to develop a deliberate strategy of sampling different
kinds of "exs" in addition to ex-nuns: ex-doctors, mother's without cus-
tody, transsexuals, and so on. Each group offered evidenceon a different
type of role exit, the most dramaticbeing an exit from one sex to another.
The end product of Ebaugh's strategy of theoretical sampling was a fully
developed analytic framefor role exit (Ebaugh1988).
Howard Becker (1963)studied a variety of groups classified as "devi-
ant" in addition to marijuana users. He joined these different cases to-
gether in a single analytic frame and called all these groups "outsiders."
His frame emphasized a dual process of socinl lenn~ing(peoplelearn "de-
viant" behaviorsfrom others in socialsettings)and lnbelii~g(society's ten-
dency to label some groups deviant furthers their isolation from the
larger society). His work challenged conventional thinking that certain
types of people were at a greater risk of becoming deviant and focused
subsequent research on social processes. In a similar manner, Erving
G o h a n (1963)studied a wide variety of stigmatized people, from those
with physical handicaps to homosexuals. From a consideration of many
different types, he developed a powerful analytic frame forunderstand-
ing how stigmatizedindividuals deal with their discreditedidentities.
While the strategy of theoretical sampling is an excellent device for
gaininga deeper understanding of cases and for advancingtheory (oneof
the main goals of social research), many qualitative researchers consider
the representation of even a single case sufficient for their goals. Some
consider the addition of new cases-using the strategy of theoretical sam-
pling-to be a useless detour from the important task of understanding
one case well. They are content to leave the comparison of cases and the
Using Q~rnlitatiueMethods to Shrdy Commonalities 101
development of broad analytic frames to researchers more interested in
general questions.
While this reluctance to broaden an investigationis common among
qualitative researchers, the strategy of theoreticalsampling offers a pow-
erful research tool. As Glaser and Strauss (1967) argue, theoretical sam-
pling offers the opportunity to construct generalizations and to deepen
understanding of research subjectsat the same time.
The Study of a Single Case
The techniques of analyticinduction and theoretical sampling work best
when there are multiple instances of the phenomenon the researcher is
studying. The study of the care of dying patients, for instance, involves
obseming how patients are treated. Each patient provides another in-
stance to examine.What techniquescan researchersuse when they study
only a single instance--for example, one person's life or a single histori-
cal event? While it is true that most data procedures are designed for
multiple instances, the study of a single case is not haphazard and un-
structured (Feagin et al. 1991).In fact, the single-case study is structured
in ways that parallel analytioinduction.
For illustration, consider a researcher who seeks to evaluate the his-
torical significance of the resignation of President Richard Nixon, who
left office in the middle of his second term. Suppose the goal of the re-
searcher in this investigationis to try to interpret this episode as a seri-
ous blow to the authority of the U.S. government, at least in the eyes of
the Americanpeople. Because of what transpired, according to thisinter-
pretation, the American people could never again see their politicians as
statesmen or trust governmentleaders and officials to tell them the truth.
Of course, there are many different ways to interpret each historical
episode, and eachinterpretationis anchoredin a different analytichame.
The researcher's interpretation sees the events surrounding the resigna-
tion of President Nixon in tenns of the authority and legitimacy of gov-
ernments. What kinds of conditions and events enhance a government's
authority?What kinds undermine its authority?
In order to evaluate this interpretation, the researcher would have to
assemble facts relevant to the analytic frame (which emphasizes factors
iduencing a government's authority) and see if they can be assembled
into an image that supports thisinterpretation.Of course, there are many
facts,and not all will necessarilybe consistent with the initial interpreta-
tion. The key question is: among the relevant facts, which are consistent
Určeno pouze pro studijní účely
102 Clinpter4
and which are not? Analytic frames play an important part in this pro-
cess because they define some facts as relevant and others as irrelevant,
and different frames define different sets of facts as relevant.
Inmany ways, this evaluation of facts is like analytic induction. In
analytic induction the goal is to see if all the relevant instances are the
same with respect to some cause or characteristic, as in Jack Katz's re-
search on legal assistance attorneys. In the study of a single case, the
problem is to see if all the facts that are relevant in some way to the sug-
gested frame agree with or support an interpretation. Thus, the different
facts in the study of a single case are like the different instances in ma-
lytic induction.
Often the facts relevant to a particular frame, once assembled, do not
provide strongsupport for the initialinterpretation.As in analyticinduc-
tion, the interpretation and the facts are "double fitted." That is, there is
an interplay between the researcher's interpretation and the facts, an in-
teraction that moves either toward some sort of fit or toward a stalemate.
As in the study of many instances (forexample, the care of many differ-
ent patients in a hospital), the interplay between evidence-based images
and theoretical ideas expressed through analytic frames leads to a pro-
gressive refinement ofboth.
It is important to remember that each different interpretation is an-
chored in a different frame.Thus, the facts relevant to one framewillnot
overlap perfectly with the facts relevant to another. Thus, there can be
many different ways to frame a single case, and each interpretationmay
be valid because of thisimperfect overlap. Cases that can be interpreted
in a variety of different ways are considered "rich" because they help
researchers explore the interconnection of the ideas expressed through
different frames.
Conclusion
Researchers use qualitativemethods when they believe that the best way
to construct a proper representation is through in-depth study of phe-
nomena. Often they address phenomena that they believe have been se-
riously misrepresented, sometimes by social researchers using other
approaches, or perhaps not represented at all. This in-depth investiga-
tion often focuses on a primary case, on the commonalities among sepa-
rate instances of the same phenomenon, or on parallel phenomena
identified through a deliberatestrategy of theoretical sampling.
Usiiz~Q~~nlitnfiveMetlrods to Shrdy Con~inonalities 103
Qualitative methods are holistic, meaning that aspects of cases are
viewed in the context of the whole case, and researchers often must tri-
angulate information about a number of cases in order to make sense of
one case. Qualitativemethods are used to uncover essential features of a
case and then illuminate key relationships among these features. Often,
a qualitative researcher will argue that his or her cases exeinpl$j one or
more key theoretical processes or categories. Finally, as qualitative re-
search progresses, there is a reciprocal clarification of the underlying
character of the phenomena under investigation and the theoretical con-
cepts that they are believed to exemplify.
Určeno pouze pro studijní účely