The Discovery of Grounded Theory Strategies for Qualitative Research
BARNEY G GLASER AND ANSELM L STRAUSS University of Califoinia San Francisco Medical Center
ALDINE PUBLISHING COMPANY / Chicago The Discovery of Grounded Theory
mainl
sis
of
strategy dures
slight
positions
this
of
which
as
standable
w
terpretations
show,
major
social
of
of
selves
thereby
with
)rks—provides
grounded
improved
Previous
yet
As
theory
in
theory
Most
is
how
and
knowledge
research—can
current
only task
sociologists
we
to
such
few
on
be
that
and
the
writing
conceive
to
how
from
accurate
definitions,
from
confronting
more
a theories
books
theory
methods
a
and
sociologists
we equally
beginning,
examples.
them-v
to
data—systematically
us
data—which
rigorously
shall
applications.
makes
on
serifv
engaged
on
is
as
facts
with
of
be
a important
for
sociological
methods
because
a
fits
emphasize
this
general
furthered.
sociology
beginning
we
theories.
on
any
rather
and
discovering
can
relevant
empirical
in
tested.
nature,
shall
the
be
formulation
layman
research
we
1
at
of
method
enterprise
than
verification
obtained
today,
often
This
We
many
social
method
for
predictions,
In
call
venture
And
grounded obtained
situations,
Grounded
this
believe
offering
ahke,
furthering
The
suggests
of soon
state
grounded
pomts
research
for,
so
comparative
book
of
premature.
has
and
in
we
Most
discover,
positions.
as
of
how
Discovery
that
the
and
explanations,
clear-cut
theory.
been
we hots
we
offer
an
we
theory,
and
the
have
the
the
important,
development
theory—is
overempha
address
believe
analyzed
shall
theory
this
I
concerned
Theory
discovery
is
there
discovery
discovery
analysis.
4
counter-
Because
focused
and
under
proce
major
try
book.
our
our
can
are
in
of
to
in
a
it a
he ter
crating
Press, data
purposefully
ery
how
ing
wish
in
mandate
choose
flict
existence
Grounded
because
cussion on
resolution if
flict
scene,
it.
is
For
between
have their
concepts resultant
hut research wishes
task
discussing
was
not
deals
logically
of
1.
The
gives
Our
Since
Surely
verification—will
verifications
“serendipity”;
is
many
systematically
many
Nlerton
concerning
1949),
a
confronting
zeal
barely
of
concerned
totally
cingle
created
the
which
basic
rise
should
book—especially to
with
theory
of
for
a
generating
of
verification
and
de-emphasis
of
Theory
discovering
sociologists,
no to
the
desire
research.
desire
sociologists
to
Chapter
never
too
necessary
hypothesis
such
excellent
started
our
theme this
lost,
a
test
svill
through
“theoretic
conflict
when
Social h
not
with ness
great
that
reached
primacy
beginning
to
potheses
conflict
sc
a
to
either
in
receive
III.
obtained
hypothesis, be
C
grounded
conffiet.
in generate
‘s.
theor\
theory
to
Theory
Testing
generate
sociologists
research indicate
specific
on
olog
an
has
research.
during
between
our
en
however,
functions taken
have
on
generate.
the
a
existing
when
adherence
ur,ar.ticipated,
takes
surprise
of
relative
primacy
through
book
notion
are
the
formulation
goes
and
from
W
Rather,
modifying
feed
been
theory
as
This
purpose.
researches,
theory
the
many
theor
of
we
among
erih
Social relevant
prior
indicating
is
THE
uridouhtcdls
hand
basic.
do
of
back
theories
research,”
vould
Concept
social course
social
emphasis
the
diverted
discuss
the
on
nd
not
to
DISCOVEaY
our
facets
is.
ii
Structure
into
often
step
of
anomalous,
Nlerton
in
discovers’
discovery
reflecting
sociologists,
g
research.
verification
of
the
al
of
clearly
theory, research, 1
a
does
for
position
haud
and
31
and
some
or
trained
The
the
that
of
course
and
becomes
current
any
from
grc
in
was
a
the
not
modify
there
OF
generating
I
closest
discovering
current
with
Glenc’oe,
given
not
of
It theory
of
and
and
we
of
forms
preoccupied
given GROUNDEO
catch
the
grounded
puts
area
is
this
need
the
that
grounded
Every
but
as
theory
exists
also
strategic
endorse
theory,
that
verifyIng
consciously
he sociological
researches
secondary,
opposition
processes
the
the
the
truism
that
emphasis
research.
that
Ill
came
that to
sn
this
a
a
idea
discov
chap
theory
a
verify THEOR1
theory
from
social
Thus,
chief
basic
what
Free con
gen with
find
con
they
dis
was
the
the
one
of
it:
in
this
lIc
stract hut
area
ahzat
that
discovered
the
From
under
meaningfully and
by meet
gories
point,
must
in
so
search;
theory
modes
research
ology
ations: tical
br to
in
enable
compare
generating
of
to
generated
ChapLer
The
3.
2.
must
qi.tantitative
To
issue the
the
give
that
taken
theoretical its
best
ion.
The research
the
work
to
In
Of
significant
the
Discovery
these
also
applications—prediction
must
sociologist
study.
generate
to
supposed
is
data
principic
which
of
should
have
course,
more
they
prediction
(4)
the
theory
crucial
approach
data
interrelated
on
toward
students
a
be
II
it
when
conceptualization
by
requirements
be
by
grounded
under
practitioner
strategy
a
fully
to
with
must
particular
readily
for
we
relevant
of
categories
peispective
the of
the
examination
readily
advance
provide
any
studies
logical
will
provide
should
theory
ones
social
Grounded
uses.
put
in
data;
researcher
generating
hall
theory
an be
study; other
and
concept
and
Chapters
fit
understandable
into
for
can
clear
initial,
theoiy
develop
jobs
We
(not
research.
to
from
and
when
that deduction
explanation
and
clear
areas discuss
to
in
understanding
a
that
must
conceptions
by
applies
handling
Theory
anti
use.
be
can
shall
perspective
does
of
sociology;
significant
forcibly)
enough
of
work. 3
his
II fills
ç
will
for
“work”
systematic 5)
is
these
his theory.
enough
verified
be
the
of
and
fit
be
theory
By
scrutiny
and
not
a
contrast
what
Then
to
this
operationalized
help
concepts
describing
behavior.
will
the
way
from
data,
XI.
able
“fit”
And
approach
guide
are
to
of
data
explanation
we
to
large
applicable
him
situation
of
Our
usually categories
one
of
in
be beha
(3)
laymen.
of
we
in
on
appropriate. 2
and
we
to
sociologists
since mean
a
laymen
discovers’
this
to
theory.
arriving
the
present
see
readily
and
in
sociology
can
behavior—a
priori
explain
mean
basic
order, to
facilitate
reality
Thus
mean
some
and
ior;
data.
reles
research,
position
he
the
he
provide
in
that
be being
Theory
to
involved
quanbtative
(2)
explaining.
ant and
should
able
position
operationalized assumptions.
usable
at
We
theory by
as
we and
relatively
control
categories
that
of
and
they
the
this
of
a
theory
data
are:
researched,
to
The
shall
the
suggest
hypotheses
theory with
tabula
to
a
any
future
indicated
the
providing
operation behavior
stance
that
be
must
style
be
in
and in
under-
theory
in of
discuss
theory
(1)
is
view
theory
useful
suited
cate
ways,
sure
prac
rasa.
soci
situ
the
The able
are
can
ab
that
and
be
re
for
as
to
In to 4 THE DiSCOVERY OF GROUNDED THEORY The Discocery of Grounded Theory 5 stand it, while sociologists who work in other areas will recog ouslv connected, omitting of many other possible explanations. nize an understandable theory linked with the data of a given s a tacked-on explanation so often is, area, Another opportunistic use of theory that cannot occur with Theory based on data ca sually rot be conipletcly refuted groinded theory is uhat may bc termed “exampling.” A by more data r replaced by another theory. Since it is too searcher can easily find examples for dreamed-up, speculative, intimately linked to data, it is destined to last despite its inevi or logically deduced theory after the idea has occurred. But table modification and reformulation. The most striking exam since the idea has not been derived from the example, seldom ples are Weher’s theory of bureaucracy and Durkheim’s theory can the example correct or change it (even if the author is of suicide. These theories have endured for decades, stimulating willing), since the example was selectively chosen for its con a variety of research and study, constantly exciting students and firming power. Therefore, one receives the image of a proof professors alike to try to modify them by clever ways of testing when there is none, and the theory obtains a richness of detail and reformulation In contrast, logically deduced theories based that it did not cam, on ungrounded assumptions, such as some well-known ones on There is also a middle zone between grounded and logico the “social system” and on “social action” can lead their folloss- deductive theorizing, in which the sociologist chooses examples ers far astray in tring to advance sociology. However, systematically and then allows them to feed hack to give theo grounded theories—which take hard study4 of much data—are retical control over his formulations: but often it is hard to fig worth the precious time and focus of all of us in our research, ure out when this is happening, even when we are clearly told. study and teaching, Much of C. Wright Mills’ work, we believe, is exampled with Grounded theory can help to forestall the opportunistic use only little theoretical control, though he claimed that data dis of theories that have dubious fit and working capacity. So ciplined his theory. In contrast, grounded theory is derived often in journals we read a highly empirical study which at its from data and then illustrated by characteristic examples of conclusion has a tacked-on explanation taken from a logically 5data. deduced theory. The author tries to give his data a more gen In contrasting grounded theory with logico-deductive theory eral sociological meaning, as well as to account for or interpret and discussing and assessing their relatixe merits in ability to what he found. He uses this strategy because he has not been fit and work (predict. explain, and be relevant), we have taken trained to generate a theory from the data he is reporting so the position that the adequacy of a theory for sociology today that it will help interpret or explain the data in a general man cannot he divorced from the process by which it is generated. ner, lie does this also because he has been trained only to Thus one canon for judging the usefulness of a theory is how research and verify his facts, not also to research and generate it was generated—and we suggest that it is likely to be a better his explanation of them. The explanation is added afterward. theory to the degree that it has been inductively developed Far instance. many papers dealing with deviance conclude with from social research, We also believe that other canons for an interpretation based on Merton’s anomie theory, a classic assessing a theory, such as logical consistency, clarity. parsi example of this use of logically deduced theory. An author monv, density, scope, integration, as well as its fit and its ability could, of course, borrow the grounded theory of another soci to work, are also significantly dependent on how the theory was ologist for its general relevance, but—since this kind of theory generated. They are not, as some theorists of a logico-deductive fits and works—it would readily he seen whether it is clearly persuasion would claim, completely independent of the proc applicable and relevant in this new situation. It cannot he tenu esses of generation. This notion of independence too often ends up being taken as a license to generate theory from any source— 4. And also in trYing to advance their personal careers, for one cannot empirically dissociate the need to generate theore from the need to advance 5.See. for example. Howard S Becker et al., Boys in lVhite (Chicago: careers in sociology. University of Chicago Press. 1961 6 THE DISCOVERY OF GROUNDED THEORY The Discovery of Grounded Theory happenstance. fantasy, dream life, common sense, or conjecture u ork’. Only sociologists are trained to want it, to look for it. — and then dress it up as a bit of logical deduction. and to generate it Probably we need to emphasize here hat we shall discuss Besides reminding colleagues of a somewhat slighted task, later moic explicitl . Cencr’iting a theory from data means that also are trying through this hook to strengthen the man most hypotheses and concepts not only come from the data, but date for generating theory, to help provide a defense against are systematically worked out in relation to the data during doctrinaire approaches to verification, and to reawaken and the course of the research Generating a theory incolves a broaden the picture of what sociologists can do with their process of research. By contrast, the source of certain ideas, or time and efforts. It should also help students to defend them even “models,” can come from sources other than the data. The selves against verifiers who would teach them to deny the biographies of scientists are replete with stories of occasional s aliditv of their own scientific intelligence. By making genera flashes of insight, of seminal ideas, garnered from sources out tion a legitimate enterprise, and suggesting methods for it, we side the data. But the generation of theory from such insights hope to provide the ingredients of a defense against internalized must then be brought into relation to the data, or there is great professional mandates dictating that sociologists research and danger that theory and empirical world will mismatch. We shall write in the verification rhetoric, and against the protests of discuss this issue again more fully, particularly in Chapter XI colleagues who object to their freedom in research from the on ‘Insight, Theory Development, and Reality.” rigorous rules of verification (so stifling to the creative energies For many colleagues, our position will be at best a hypotlie required for discovering theory). us, to be tested in the years to come; while for many others it In trying to stimulate all sociologists to discover grounded is proven fact, and for still others an article of faith, However theory—from those who are only at the dissertation stage of colleagues max’ respond, our position is not logical; it is phe their careers to those who are already “retired” professors-—we nomenological. We could not suggest a process of generating hope to contribute toward the equalizing of efforts in gen theory if we did not believe that people who might use it erating theory, which are now often limited to the earlier would arrive at results that potentially may be judged as suc stages of a sociological career. Foi’ example, Hammon. in cessful. Furthermore, we believe that grounded theory will be presenting us with chronicles of some of the best sociological more successful than theories logically deduced from a priori research (those with the highest theoretical yield), has chosen assumptions. Our position. we hasten to add, does not at all mainly chronicles of dissertations or studies done as soon as the imply that the generation of new theory should proceed in dissertation was finished.’ Similar studies could be done by isolation from existing grounded theory. (We shall discuss this mature sociologists, and with more speed (less fumbling, clearer in Chapter II.) purpose) and more sophisticated theoretical yields. Indeed, that the growth of a theorist is linked to the increasing sophistication of his output is clearly seen in the work of men like Goffman, Purposes of This Book Lipset and Wilbert Moore, Yet many sociologists as they mature disiegard whate er fledgling potential for generating This book is intended to underscore the basic sociological theory they showed in their dissertations and early monographs. activity that only sociologists van do: generating sociological They cease or slow up their research and writing of monographs theory Description. ethnographv, fact-finding, verification (call and turn to scholarship and the mastery of others’ works, par them what you will are all done well by professionals in other ticularlv earlier “great man” theories. One respected scholar, by fields and by layman in various But investigatory agencies. 6. Philip E. Hammond (Ed.). Sociologists at Work (New York: Basic these people cannot generate sociological theory from their Books 1964).
verification.
to
discovering
(luire that
in and
ination
and
for
room the
of contrast
Our
integrated ries be
should ing;
already
any
dcsc’ribing,
give
theory.
criteria grounded
any
of
to graphs
would
by
virtue
8
Hammond,
generate
8.
explicitness
in
Our
try
students
What
able
Throughout
systematizing
theory’s
they
publish beating
saying,
ssill
strategies
for
leasing
theon
Compare
Mehille
theory.
colleagues
for
our
to
of
leading
like
not
principal
This
for
and
until
are
to
to
will
[end
including
his
generate
about
own
(see theory
grounded theory
ont
op.
judging
the
Though use
explaining,
in
that
curb
their
him
to
Dalton,
acceptance deeply We
start
theme
position
join
to The
they
to
and
do
cit.,
the
effect,
see
to
attempt
Merton’s
ways
Chapter it
this this
an
urge
into
they
end
us
own
aim
the
not
anyone’s
worrying
them, 8
pp
“irrelevant” effectively
that general
that theory!
overdrawn
proceed
“Preconceptions
both
older
effective
pervades
the
involved
in book’s
book
difficult,
and
intellectual
up
theory
insist
of
rendition
5’-.38
them
at
will
is
methods
telling
it
predicting,
worth
strictures
will
by
to
VI).
propositions
verification,
to
thin, a
sociologists
prominence,
is
we
teach,
comparative
creativity usefulness
recent
discuss
that
with to
“drugging
about
not
stimulate
means
richness
the bc
we
it
in
to
those
Our
call
explanation of
unclear continue
for
of
is
on
a
THE
the
help
generating
a
all
submission.”
in
shall
discussed
whole
apply
residual
for
sociological
an suggestions
teaching
theory
bit
codification
generating
interpreting
who
methods
Methods
for
theory, DISCOVERY
of analyst
they
and
cease
for
for
systematize
more
has
exciting
other
in
more
the
connotation!
for
writing
evaluating
book,
have
method
or
purpose,
those
generating
the
encouraged
allow,
in
should
chore reader’s
the
in
use
writing
as in
theory,
engage
the
theory?
theorists
method
and
an
not
of
OF
Men
richness It
theory.
for
well
meeting,
adventure.
most
Our Part
monographs
sociologists
in
theory,
and
effort
next
GROUNDED
is
even
for
in
yet
processes
Op.
encourage
systematizing their
Who
and
the
research,
our
as
imagination
but
in
suggestions
this
their
Mans’
part,
their I
attempted
testing
theory;
generating
generation
to
cit.,
this
We
to
grounded
a
demand,
of
which
worth
not intent
Manage,”
provides
theoriz
not
that
age
degree
codify
coerce
infor
mono
THEORY
p.
keep
trend.
trust
theo
well
who
may
for
and
just
14.
re for
of
it.
in
he
of
to
tion
logico-deductive
verification,
tive
contemporary
Chapter
relations We
ter
we
Vii
flieon 1 —sve
important
a
/
designed
analysis
sampling—the
comparative
IV
-nd
groups.
volves
a
grounded
—and
published
theme work—and
examples
suggested
purpose—though spersed the
thinking
The
cpeciallv
number
strategy
Before
In
In
IX)
consider we
In
and
of
close and
Discovery
in discussion
the
the
the
theory.
take
discovering
of
the
and
of
Xl
quantitative
Chapter
In
of
ssith
VII
to
about
second
with
this
questions. from
theory,
moving
first
third
of
product;
procedures
qualitative
theory
whereby
and (locumentarv)
systematic
we
Chapter
up
for
generate
consider
in
its
analysis.
of
previous
emphasis
respectively).
occasional
book,
process
section—Comparative
detail
the
discuss
research,
an
the
the
Grounded
a
theorizing,
practical
open-minded,
part
part
V
both
very
to
not
on
theon 1
transition
epilogue
topic.
we
distinction
we
data—a
sociologists
II
research,
the
substantive
the
of
of
data.
of
to
or
choice
at
comparative In
insight,
substantive
good
know
on
we
offer
the
collecting
credibility
the
frank
Theory
we
these
implications
the
generation
Our
as
the
and
Chapter
which
verification,
In
discuss
book—The
summarizing
have
a
distinction
from
book—Implications
reason.
logic
our
our
and
expense
Chapter
suggestions
usually theory
to
process
polemic—always,
quantitative
chapters.
can
and
own
encourages
method
stimulate
drawn
substantive
data study
and
lying
of
studies
the
III
of
Anal
formal facilitate We
development
grounded
(Chapter
drawn
better
is,
VI
of
formal.
for
theory
the
useless Flexible
we
purpose
behind
of
heavily
we ysis—we
know
of
are
our
we
stopping
for
comparative
in
influential
data
several
theory.
discuss
rather
course,
as
the shall
between
clarify
deliberately
to
terms
from
the
position the
This
for
others’
work-in-process
theory
X).
Use
them.
we
formal
of
upon
of
drive
(in
shall
comparative
the
discovery
and
discuss
than
the
strategy
In
comparison
qualitative
our
Grounded theoretical
the
hope,
and
Lastly,
of
of
Chapters
style
qualita
genera Chapter
our
In
work
analysis
toward
(Chap
on Data—
reality.
several
present
theory.
flow
central
use
assess
freeze
using
inter
with
the own
the
of
in in
of
of
as
of 9 THE DISCOVERY OF GROUNDED THEORY The Discovery of Grounded Theory 10 11 by training vosng sociologists to test their teachers’ work but not to imitate 0it) Verification and “Grand” Theory In the face of this prevalent attitude, we contend, however, that the masters have not provided enough theories to cover all Vcrificat r of Leo y is the kevncte of urrent sociology the areas of sociel life that sociologists have only begun to Some three lecades aO it was felt that we had plenty of explore. Further, some theories of our predecessors. because of theories hut fev C nhimations of them—a position made sun their lack of grounding in data, do not fit, or do not work, or feasible b ths greatly increased sophistication of quantito e are nor sufficiently understandable to be used and are there methods, As this shift in emphasis took hold, the discos ei o fore useless in research, theoretical advance and practical appli nyu- theories h came slighted and, at some universities, s ix cation, On the other hand, the great theorists have indeed given ally negiected. Those who still wished to generate theory hxd us models and guidelines for generating theory, so that with to brook the negative, sometimes punitive. attitude ot thc:i recent advances in data collection, conceptual systematization cohcagues or rore5SOrS. and analytic procedunes. many of us can follow in their paths: Part of the trend toward euphasixing .ifiunm wes the from social research we can generate theories for new’ areas, assun pti )n hr many sociologists that ‘“ ‘ ‘,i eat ben” frn_ as w’ell as better theories for areas where previous ones do not i x Cooley fathers (\\ eher. i)urkheirn, Simmel, \ ehicu, sork)’ Mead Park etc had generated a suffici it umber of out We contend also that it does not take a “genius” to generate standing t seories on enough areas ) sal life to last for a a useful grounded theory. It does take some codification of the long while, Although we their soc ol gicel offspring, could method of doing it, as well as recognition of its legitimacy for never equ Feir genius, we d Li s’ hc is to modify and re student training and academic careers, Our hook provides some formulat I cir theories with ear ass .found abilities in yen of both. It is well known that in science the highest rewards ficatior —aad so that was tF next job of sociology. As a sesult, have always gone to those who generate an important new manr of our teachers a ‘ris’cldepai tments of sociology into repositories of , eat-man” theories and taught these mere 10. The following are the words of a voug theoretical capitalist theories s itha chaniso’ _nc finality that students could seldom modestly asking the proletariat testers to correct his conjectured ory-: “Whereas the sbudcnts trained to master great-man empirical tests would undoubtedly prove a good proportion resist. Cnrresitlr are of the inferred predictions to be incorrect, these negative findings would theories and to tut them in small ways, but hardly to question provide a basis for refining the theory, whereas as no such refinements are the thecw a a is isole in terms of its position or manner of gen possible if a theory fails to yield operational hypotheses that he can be negated empirical es idence .“ Thus to encourage the testers he eration As a result many potentially creative students have his carefully writes theory’ so it can be readily operationalized and proven niseives to pu7zhng out small problems bequeathed ways—a wrong in several limited t temptation for those who like to prove the theorist proletariat wrong. These to tI er I ‘g theo ew men (like Parsons and Merton) testers do not realize that allowing themselves to be tempted sim ‘sly puts the refined theory and s charismatic v of the eat mer suffi thc thsorist on firmer ground, while have :ee ough tb’s ue g ase soon forgotten. they Sec Peter Blau, Exchange and Power in Social Life cientlv to g nerate grand’ theories on their own. But even (New York. John Wiley and Sons, 1964), p 9. We can only say that it is [no our position that theorists these te I ‘ e 1nc I- d methods tor gen sa ig theo y n data, lie responsible for the grounding from the start, of their theories or n their methods, They have at rate base not writt r about For another attempt at theore’tje’al capitalism and request for colleagues plan e ‘ 1woletical capitalist to the mass of ‘pi iletatiat” testers to test him out, see Phonias J. Seheff, Being Publishing Mentally III (Chicago: Airline Co., 1966, especially p. 101, 11. For example, this is happening in the study of deviance. shall 13. See Mar 0. r Hans L. letterberg, Co 7 Jour, ad Vs:-ibc0ti0u in Soc inlo Clinard (Ed.). Anomie’ and Deviant Behavior Press of (New York: Free Toto’. , \ .E - Bcclniinster Press. 1963 Glcneoe, 1964). 12 THE DrSCOSERY OF GROUNDED THEORY The Discouery of Grounded Theory 13 )I2 theory çsociology is like physics in this regard historical simpler elements “ In contrast, the monograph was directed at reasons then, account for the paradox that more sociologists furthering general sociological theory and giving a very detailed do not try their hand at generating theory and publishing it, interpretation of Polish peasant society in Europe and America. thus achieving high rewcrds We ssish to help alleviate this Bluiner’s principal criticism of The Polish Peasant was di condition by encouraging able sociologists to generate more rected at what he believed was an important methodological and better theory ssith the type of comparative method dis flaw in it—one that needed to be discussed as an issue basic to cussed in our book, and, in turn, to start developing methods of sociological research rather than as pertinent merely to this par their msn for all of us to uce, ticular monograph. The authors claimed that their analyses rested largely on numerous “human ocurnent.s”: letters, agency records, life histories, court records. Blumer noted first that not Verification or Generation? all—perhaps not even the major—theoretical conceptions used by Thomas and Znaniecki were grounded on thoce documents, The following account is an example of the kind of historical Indeed, “the major outlines are foreshadowed in the previous circimtance that put the generation of grounded theory into writings of Thomas,” and even “their particular interpretations second place. and made verification the dominant orientation of Polish peasant life were not formed solely from the materials in virtually all sociological ssork: they present; sse have to assume that the familiarity with Polish During 1.938 the Social Science Research Council struck peasant life which enabled their interpretations was made in a upon the idea of subjecting to critical appraisal a series of sig wide variety of ways.” nificant contributions to social science, In sociology, Herbert But this was only a minor criticism. Blumer’s major concern Blumer as assigned the task of appraising l homas and Znani was this: “the important question is whether the materials ade ecki’s great monograph, The Polish Peasant in Poland ansi quately test the generalizations (regardless of their source) America. A year later Blumer’s critique was published by the which are being applied to the materials But “the answer Council. The volume included comments on Blumer’s analysis is very inconclusive.” Some interpretations seemed to him to 1by Thomas and by Znaniecki, as well as a reprinting of the pi’o be borne out by the materials; some did not. W’orse vet, usually ceedings of a conference that discussed the analysis (the con one could not say that “the interpretation is either true or not, ference included such participants as Murdoek, Wirth. Bain, even though it is distinctly plausible.” (pp. 74-75). Blumer Wiley and Wailer). agreed that these plausible interpretations made the materials Blumer noted that Thomas and Znaniecki had been much more significant and made “theoretical interpretation more concerned with methodological issues and had taken a stand understandable,” Yet the very puzzling issue of plausible inter against several types of knowledge then much advocated, These pretation versus genuine verification remained latter included “common sense generalization,” ‘planless empiri Therefore BIn sicr concluded, first, that the materials were cism,” “mere statements of uniformities of social behavior in not a decisive test of theoretical interpretations, although they response to social irfluences,” “statements of causal influences did more than simply illustrate them second, that a test of which hold true ‘on the average,’ or ‘in a majority of cases,’” “theory would have to come in other ways, such as in its and a type of misleading oversimplification in which “effort is internal consistency, in the character of its assumptions, in its made to resolve what must be taken as a primary relation into relation to other theories, in its consistency with what seems to he ‘human,’ or than by 12. F’or esanplc. ix of the eight Maclver Assards have gone to sociolo in other kinds of data those provided gists for generatiic ground”d theor. human documents”: and, third, that the authors’ use of human 13. Thomas and F. Znaniecki (‘w York: Alfred A. Knopf. 1918). documents would seemingly imply that their essential fimction 14 Appraisal of IIionsa’ and Ziianiecki’s ilie Polish Peasant in Europe “svoulcl be to yield to a sensitive and inquiring mind and America 1 \CW York: Social Science Research Council, 11.39). ...
718-19. better American
being he
this conceptual lations.
better
the als issue
ploying
qualitative The
the by
and and that left
dominance,
ing cal himself
analysis see
theorize emphasizing
and pened
toward enormous
theory,
critique
less exert
Stouffer,
data spectives, hunches,
better
14
was
Znaniecki’s
studies a
15.
future,
did
meaning
instance the
Blumer’s
intimate
improvement
that
it,
the
stimulating
the
studied.
fortunate
more
n
is
theory
theory,
poised
great
A
to
if
gap.
whatever
crc
not
hut
(quantitative
also whatever
With
the
result
Journal
on
most
latter
from
year
Blumer
the toss
unguided
data,
Chapin
as
and
insights,
influence b
and
useful
familiarity
Closing
always
of
framework he
on
except verification,
,
instructive.
question
and
See
in
and
influence
gap
ard
the
later, critique
hindsight,
he
the
rejoinder one of
attributed
data too
combination—an
we
problem
new
generation.
relevant
of
in
his
in
too
in
between
data)
attacked
its
had
for
an
theory-data
by
his
b
sharp
latter,
agree judgment, it,
the
Sociology
I
provide
close
could
“The
Blumer
ready
questions
azarsfeld
rather
demonstrable
of
with
understandings”
he
the any
theorizing
focused
of
examination
past,
intent,
in
measures
The
n
imagination a
with
relation
believed,
with
Problem
He
largely
to
ungrounded
it
then,
general
how
theories. we
to
the
as posture
this
say
since
sociology.
published
effectively
a
than
give
a
Blumer’s
(1940),
in
agreed
He
Polish
kind
gap
slow
can
written
thoroughly.
good
which
was
to
Blumer
less
to
observation,
suitable
Guttman
fit
to
up
he but
of formula
untouched,
against
would
inquiring
“the
did,
Operationalism
TH
and
from
of
maturing
ssonder
generate
for
data,
the
findings
the
observers
that
on
on
test
Peasant
707-19,
theories
conduct
that
of raised
not
as
an
we
DiSCOVERY
on
critique
the The
inadequacy
checking
Concept
the
of
during
mood
(pp.
depend
not
threw
the
verification
was—again—blunted
verification,
admirable
of good
of
Blumer’s
for
the adequate
his
problem
and
approached
course,
and
process.”
problem what
offering
sticking
the
mind,
emphasis
the
than
and
being
the
armchair,
grounded
may
75-76).
for
requisite
apparently
monograph’s
reflection,
of
theory
in
on
other
in
the
on
was
theoretical OF
the
quotes
the
issue
two
the
might
theory
fortunately of
Social
solution
of studied
concept
“developing
rich
article,,
(operationaiism
the
a
GROUNDED
close
of
come
for
His
then
countless weight
its
how
of
solution
period
In
need
day.
advocates
decades
that
rather
is
in
are
verificatiOn
verification experience,
But,
theory.
Psychology,”
emphases verification
of
stimulating
our
to
have
coming
to
produced short,
and
new
assuming
began
will
to
addressing
Blumer’s
close
from
for how
general
the
Yet
materi
generate
formu
because
possess.
to
of
data,”
empin as
getting
THEORY
in
when
a
be
than
hap
good
data
clos
was
per
rich
em into
the
pp. He
his
we
the
on
to
to
in
to of
in
groundbreaking
by
gang.
research facts
reproducible
search,
to
begin concepts
World
in
on
theories
graphs
small
sense,
sisted
data
data
of
from
advocates
grounded
tualization;
generation
Qualitative
qualitative
‘system”
would theoretical of
anhow
The
B’
18.
men
test
producing
‘Thus,
Meanwhile,
theory
qualitative
Historically
the
the
about
in
Discoeery
at
For
generation
framework,”
the
amounts
of
In
because
War
was
like
unconfirmed
ghetto,
a
all),
was were
‘attitude-value’
substitute
into
example
lengthy,
in
advances
addition,
nonsvstematic
in
challenge
and
of
theory
social
Stouffer
to
IT.
of
conception
data the
he
vs.
fact,
to
the
in
research
quantitative
accurate
too
the
of
“get
of
theory
data
“pattern”
quantitative beginning
was
of
Quantitative
work
provide
monograph
conjunction
linked
Grounded
to
late
see
taxi-dance
its
of
structures
“impressionistic.”
detailed
and
Zaniecki
theory,
the
monographs
verification
in a
and
still
pool’
of
was
various
theories
the
l930’s,
for
largely
more
operations.
quantitative
story
evidence
that
its
testing
conceptual
various Lazarsfeld. not
quantitative
and
(which
showing
either
in
ith
Theory
getting
hail,
and
if
sensitivity
descriptions
researchers
and
emphasizing
organized
the
with
kinds.
Data
by
straight.”
referred
sophisticated
nonrigorous
which
any. 16
with
the
in
the
studies
theory
of
qualitative
late and
based
not
terms
social
and
he
in
their
theory
hoboes,
The
change
surveys
the
He
combinations”—the
to
methods
would
The
producing
1930’s,
theoretical large, believed
research
of
to
in
in
rigorously.
preliminar),
In the
on
of
result
made
systems.
was
own
“facts.”
which
the
the
translating methods
etc. was
picking
effort
way
short,
a
qualitative
monograph.
data. in
have
started. had
conception
Chicago
and
pre-existent still
“excessive
logic
initiated
great
was
the
with relative
had
(when
the
resulted
used
in
the
Qualitative
especially demanded
was
enough
The
thinking
clash
up
for
an
and
scientifically
these
been
exploratory.
school
strides
a
Qualitative
work
theoretical
relegated.
qualitative
generators
generating
few ability they
everyday
data
the
emphasis
Znaniecki
common
simplicity
between
involving
principal
in
concep
or
implicit
mono
on
based
of
sub-
after
both
used
zeal
very
fuller
con
the
re
the
to
the 15
in
interviews qualitative assumption to
opment
what accomplished fications ress and
the terials. in
complished. ways to tributions,
path qualitative sampling,
ogists quantitative tative test
so tinue and stantive
systemization can
codify example. research
given
that
Organization
16
sociology
their
the
assemble
A
17.
forth).
The
pressure
towards rules
current
sociology,
sociology
parsimonious
to
in
the
today—to
they
smaller
and
data
to
For
point
Sometimes
th
1949
systematization strength
categories
research
were
of
would
type
clarify
of
clarify
coding,
clarifications
or
articles
One
researchers
collected,
data,
Research
conceptual
hypotheses.
(testing,
behind,
modeled
the
by
that
accurate theory.
evidence
becoming
data.”
where
observation
according
number
was
of
for
was
Robert
was
only
use
position
all
take
of
and
quantitative
data
in let’s
and
reliability
and
the
they
far
presentation
methods,
embarked
Richard this
(Homewood,
These
they
based
the
and
proving,
qualitative
analyses
codify
and
assembled
K.
evidence
over,
of
teaching,
verify
or
of
hypotheses.
broader. on
a
quantification
was
formulization,
position,
used
verification
Merton,
to
quantitative
was
the
science,
offered
were
because
codifications
sociologists
these
N.
on
advocates
precise
including
explore
guided
validity,
all
content
with
Adams
verifications
quantifying
on
tentativeness,
“since
done,
th
recorded,
and
Ill..’
op.
but
Virtually
research
methods issues
of
and
which
a
any
by
of,
such
Then,
cit.,
emerging
patterns—for
straight-line
Dorsey
and
rhetoric evidence,
(as
to
further,
and they
qualitative
of
virtue
we
of
aualysis’
presented
indicators,
assurance
of
did
tried
these
had
hypothesis
p.
Jack test
qualitative
soon
this
the
data,
concepts
are
coding
of
qualitative
receiving
390.
operations,
began—and
Press,
had techniques,
every
take
J.
of
not
hypotheses.
research
course,
demonstrating,
to
book
in
discover
developments
swept
Preiss
so
systematic
research
quantitative
The
as
put
on
talking
systematize
been
1960).
other
awsnn
course
qualitative
of
procedures
accustomed
maneuver
frequency
methods
example,
they
has such
clarified.
on
(Eds.),
construction,
much
their
over quantitative
methods
report. 17
data!
no
developed
all
been)
The facts
was
positions,
but
still
of
issues
do
of
Indeed,
Ameri
canons
matter
sociol
ability
see,
devel
Human
rnEcrnY
quali
prog
call
earlier
their
yen
with
how
con
The
The
was ma
was
and
and
dis
the
ac
by
for
to
of
as
to
the
data to
data
es mental tested
tween
and own
tati
trend
remarked their subordinated
generating use
station,
their
“we
what
eases.
rhetoric
nology, been generated
finds
work
tinued
importance,
been
techniques the
about
to
social
The
stably
which
18.
Our
primacy
obtain
The
was
Another
data
e
quantitative
checked
have
best
Discovery
theories
the
theories.
theories
that
use
that
in
too
In
verifications
as
and
people
they
systems.
clash
exploratory
or
to
position
such
position
we of
the
antagonists,
sociology
for
heated
and
generating
been the
earlier, data
what
they
concerned
generate
these
theory
only
theory,
leaned
verification!
for
of
need
position
1930’s,
between and
this
generating
as
by
to
of
theories
who
richest
were
emphasis
This
collecting
on
linked
Also,
make
“the
slightly Grounded
procedures
the
in
methods
data
of
toward other
out
discussions
in
yet
they
more
or
many
since
as
that
men
have
using
this
function,
theories
the
meant,
generated,
hypothesis
rhetoric
many
taken
theory
qualitative
how
they
for
the
qualifications
with
the
were
historically. 18
are
men.
the like
In
never
on
pointed
hook
theory
modified
logico-deductive
denite
both
perhaps
areas
taken
Theory
quantitative theorizing
purposes
or
best
reading
in
it
verification
Stouffer-Lazarsfeld
times.”
have
by
thus
E.
formulating
their
of
led
data,
from
(or
on They
of
types
was
prediction,
is
mentioned
W.
of
advocates
course, this
way
is
anyway,
because
proofs
neutralizing
to
verification.
qualitative
as toward
not
method
have
media
they
social
tentative,”
rather
Burgess
We
What
relevant)
of
qualitative
their
suggested
could
and
stand
follows:
using
We
to
about
data
explicitly
data.
or
cannot
wanted
in
that
not
reformulate
capacities
life
writings,
and
we
believe
they
the
their
clash
than
generation
application
attempted
of
still
not
in
theorists
is
the
the
its
future
social
position
described
not
The
they
versua’
their
that
qualitative
are “we
generative
there
Since
perfection
because
therefore
hypotheses.
supported evaluate
lose,
was
verification
data,
replaced.
lost
there
ideas
their
referred
amenable
that
seldom
fascinating
had
research,
they
research,”
supported
one
structures
of
that
is
to the
they
emphasis
also
As
quantitative
and
of realizing
work
is
qualitative
within
each and
no only
mediate
possibilities.
they
constantly
how
have
how
qualitative
theory—
concerns
were
we
only
data
of
did became
to
modify
quanti
told
funda
But
expla
termi
to
to
a
form
have
their
have
their
well
they
con
and
few way
fact
not
the
and
still
the
be
has
the
be
on
in
its
of 17
rhetoric
with
tion
quired
quate” area
move often
patterns because
structural
he
i cit.. found on
also ‘Research
thr
research
data, :=t ments,
to different whatever
the of
the
the quantitative
theory.
pared,
18
Sd
t t is
alitative
19. data
used
Wc
get
theoretical To
benefits
In
be
pp.
quar
kind
circumstances
researcher,
beyond
‘Ac
Thor
a
book,
since
See
obtained
and
and
best
mane
and
further
will
focus
as
‘90-193.
library
titatis
set
is
by
of
the
Chronich’.
forms
and
of
and
James
conditions,
the
that
vish useful
oct,
mutual
because
to
data
“efficient”
those
generation, the
each
with
of
data
£ used
crucial
on
which
instances,
contend
with
systems
tk
primacy
he
sampling
resear
the
this
206. Colemar
categories of
also
etn0
and
process
and
erehy
qualitative
i.
for
suggests
who
generate
used.
a
a data to
The
of
end
verification
strong
qualitative
view,
Coleman
“coinparatise
qualitative
few
I
field on
througl
elements
both
consequences,
test
way
research,
19 wish with
‘
kdolescent
in
encourage
product
of
to on
(See
of
both
the
and
discussion
for
research
analyzing
because
qualitative,
we
research
in verification
to
for
emphasis.
balance
theory
the
generating
the to
data
agrees
a
kinds
particularly
quantitative
obtain our
forms
this
“comparatise
seek
generate
of
writing
same
research
and,
difficulties
on
qualitatise
method, of
THE
Society”
for
o
book, it
sociological
nith
(see
the
sociologists
book
qualitative
researr’h
of
the
out
in
deviances, off
DISCOVERY
the
of
the a
most
subject,
Although
Primacy
material
but
and
number theory
‘working
this
u.
interests
Chapter
their
data
theory
relative
the
that
is
in
to type
nost
that
analysis,”
Chapters
e
quantitative
but
data.)
of
both
important
often
Philip
generation
within book
provide
defensive of
,ire
an
which,
theories OF of
is
he
chapters
research
is,
are
merits
theory ssith
norms,
depends
parts
of
he
the
used
verification.
empirical
III).
independent and
GROUNDED
E. information
the is
necessary—not
to
as
other
from
needs motivated
a
not
emphasis
Hamm
II
quantitative
we
of
of sociologists further
substantive
when
analysis”
most
training
as
for
processes.
of
and
are
within
aware
in
hail
is
lualitative aio
reasons.
data
only
supple
system’
theory
for srid.
THEORY
us,
situa
social
“ade
more often
VIII,
com
show
can
that
the
re
his
can
on
op
on
on
to
of
as
of
a
CO1IPARA
GENERA
PART
I
TIZi1G
TIVE
THEOR
ANALYSIS
Y
B Y
of
parative
which
parative
assigns
sociologists
“comparative
and
general
other
and
scale
parison.) parative
experimental
describe
—the
this
clear
comparative
‘,arious
ologists
ings
anthropology—has
any
1.
Comparative
distinguishing
The
large
method
In
and
social
at
methods,
generation
it
size,
the
method
Chapter
purposes.
term
analysis analysis.
the
analysis
and
has
what
thereby
regions
Furthermore,
method”
method
and
units,
large
with
outset
frequently and
analysis,
comparative
anthropologists,
kind
VI
them
to
anthropologists
be
analysis
as
to
statistical
of
particularly
of
or
to
certain
To
was
use
we
grown
its
as
used
from
of
refer
a the
carry
theory.
small,
diccuss
avoid
to
used,
fullest
strategic with
have
theory
been
world,
comparative
our
is
other
our
for
analysis—often
to
only
several
ranging
methods.
examining
a
one
own
We
in
confusion,
encompass
employed
generality
21
own
organizations,
applied.
recognizing
social
general
can
uses. 1
detail
to
But
customarily
method
suggested
specific
shall
different
comparisons
use
be
from
such
Generating
them
a
units (All
Then
analysis
method,
first
number
generated
for
Our
we
for
it
class several
for
purpose.
a
men
for
used
use
comparative
the
burdens.
reference
for
we
contrast
of
must,
use
discussion
nation,
use
their
generating
of
the
of
or
any
in
can,
shall
great
just
between
achieving
on
different
the
studies
social
specific
their
through
logic
sociology
therefore,
social
institutions,
II
size.
as
our
term
Many
define
like
restricts
Theory
power
of
roles
units
in
of
purposes
are analysis
theory
use
large-
mean
Some
units those
com
which
com
com
com
their
soci
and
and
the
to
to be
of
of a 22 THE DISCOSERY OF GISOR FORD THEORY Generating Theory 23 nations or world regions Our own recent experience has Accurate Es idence demonstrated the usefulness of this method for small organiza tional units, such as wards in hospitals or classes in a school. Q the factual level, evidence collected from other com Before 2 distinguishing our purpose in using comparative parative groups—whether nations. organizations, counties, or anaJvis from other pLrnoc-’s we should mention one unfor h.ocpital wards—is used to check out ss hether the snstial cvi— tunate use of comoarisons: to debunk, disprove, or discount the lnct’ was correct Is the fact a b.cta Thus, facts are replicated work of colleagues From his own readings a sociologist can ‘a ith comparative evidence, either internally (within a study>. almost always find if he uants to, some piece of data that dis externally (outside a study) or both, Sociologists generally proves the fact on sshich his colleague has based a theoretical igree that replications are the best means for validating facts, notion, Many sociologists do! If each debunker thought about Although this use of compa ative analysis is not of itself, the potential value of comparative analysis, instead of satisfying our goal, it is definitely subsumed under our goal. Naturally we his urge to “put down a colleague he realize that he would wish to be as sure of our evidence as possible, and will therefore ha me clv posed anothci comparatisc datum for generati -Ig heck on it as often as we can. Howerer, even if some of our another theoretical property or categor. That is all he has idence e is not entirely accurate this ‘a ill not be too trouble done Nothing is disproved or debunked, despite what those some: for in generating theory it is not the fact upon which we who are overly concerned with evidence constantly believe. stand. but the conceptual category (or a conceptual pro port Kinder colleagues. who present a sociologist with one or more Of the category) that was generated from it. A concept may be negative case hut are afraid of impairing his motivation. usually generated from one fact, which then becomes mereh’ one of a will suggest that some qualification in his theoretical assertion universe of mans possible diverse indicators for, and data on, may be advisable. Their comparative analysis aids him in round the concept. These indicators are then sought for the compara ing out his own comparative analysis and further generating his tive analysis. (See Chapters III and IV.) theory In discovering theory, one generates conceptual categories We also intend to hold a dialogue with those who “put or their properties from evidence; then the evidence from dow the comparative as strategy “not especially original.” which the category emerged is used to illustrate the concept. True, the general notion of compalative analysis was developed The evidence may not necessarily he accurate beyond a doubt by our sociological forclathcrs—Webcr, Durkhem. Mannheim— nor is it even in studies concerned only with accuracy). but and by social anthropologists. We can only trust that our read the concept is undoubtedly a relevant theoretical abstraction ers will absorb enough details of comparatis e analysis as about ren what is going on in the area studied. Furthermore, the dered in this hook to be able to spot the advances in the concept itself sviii not change. while even the most accurate strategy that should make a world of difference in its use. facts change. Concepts only have their meanings respecified at times because other theoretical and research purposes have evolved. Purposes of Comparative Analyses For example, one theoretical category related to the care of dying patients is their social loss—loss to family and occupa The distinction made earlier between relative emphasis on tion.’ This category clearly affects how nurses care for dying generating and verifyi ig can he illuminated further by consider 3. ‘Ac are applying ing the typical uses of evidence obtained through comparative here I,azarsfcld’s rule of ‘interchangeability of in dices” in a new connection Sec Paul F Lazarsfcld and Wagner studies Jiw Thielens, Acarleinic Iliad (Ncw York: Free Press of Glencoe, 1958), pp 402-407. 4. For an explication and theoretical discussion of the category of social 2 Barney C. (Pacer and Anseim L. strauss Awareness of Dying (Chi loss, see Barney G. Glaser and Anseim L. Strauss, “The Social Loss of c ag): Aldine Publehing Co , 1965 Dying Patients, American Journal of Nursing, 64 (June 1964), pp 119-22. 24 THE DISCOVERYOF GROUNDED THEORY Ger crating 7 heory patients The category of ‘social loss” can he generated from .r d pati nts find that cues tha rr gi alert thcm that Lie arc either the ohseration that VIP’s receive special care on inten thing are sague and hard to read until the last stages of their sive care units or that lower-class Negroes often are neglected dying. In a Japanese hospital ‘ae once visited. cancer patients on city hospital emergency wards. Even if the evidence changes tvpicath know they are dying ( an open awarcue’a context” . or is different in other hospitab for various other reasons), we Wln? Because the hospital ward is openly labeled “Cancer.” can he sure that social loss is a category related to nursing The patient entering the ward reads a clear cue that rnakas him care, and we can make predictions on its basis. We can predict an, are that he is dying. \Vhile in America the cue’. tend to be that patients who have high social loss will receive better care ‘aiue and fleeting, we discovered through the Japanesc example than those who have low social loss. If that prediction proves that they can be clear even at the beginning stage o a long incorrect, then we are likely to find out next what structural term of dying, Until then, we had not realized that ‘.u’ ‘. ca,1 conditions have tended to negate this relationship; for example, vary in clarity at the beginning of such a disease as cancer how the medical staff has overcome this socially induced tend We had thought that clear cues emerged only during the final ency in one tYpe of hospital. short, t sgcs for In the discovered theoretical example. when the priest arrives, or the patier t s category lives on until proven theoretically defunct for any sin is beyord endurance or massive bodily degeneration class of data, while the life of the accurate cvidcncc that indi occurs. cated the category may he short. ‘This comparative data from Japan stimulated us te find oca ions in America where clear cues are provided at the start of dying. Empirical Generalizations We found that in a veterans’ hospital and in a prison medical ward, patients from the outset were given clear cues Another standard use of comparative studies is to establish that thex had cancer, Thus we discovered that under the struc the generality of a fact. Does the incest taboo exist in all soci t iral condition of being a captive patient in a government hos eties? Arc almost all nurses women? Is basic research the most pital. one tends to die in an open awareness context, But most revered goal of scientists in all research organizations? Accuracy patients in America do not die under such circumstances. is not at stake so much as establishing the structural boundaries of a fact: where is the fact an accurate description? For some sociologists and anthropologists this purpose becomes a quest Specifying a Concept for “universals”—faets and their explanations by other facts— that apply to all men irrespective of their society or culture. Another (usually detailed and painstaking) use of compara Our goal of generating theory also subsumes this establish tive data is to specify a unit of analysis for a one-case study. ing of empirical generalizations, for the generalizations not only This is done by specifying the dimensions of the concept desig help delimit a grounded theor-’s boundaries of applicability; natin the unit. To make certain the reader understands what more important, tlie help us broaden the theory so that it is a given monograph will be about, in comparison w’ith seemingly more generally applicable and has greater explanatory and similar units, the author compares his unit for analysis with predictive power. By comparing where the facts are similar or these other units, His comparison brings out the distinctive ele different, we can generate properties of categories that increase ments or nature of the case he has studied. For instance, tlic categories’ generality and explanatory power. Cressey painstakingly’ compared taxi-dance halls with all other For example, dying of cancer in America can he character forms of dance halls before proceeding with his analysis.6 Iipset, ized as occurring in a “closed awareness context”—while the 5. Glaser hospital staff does, the patient does not know he is dying. Most and Strauss, Awareness . . , op. cit., Chapters 3 and 8. 6. Paul Cressey, The Taxi-Dance Hall (Chicago: University of Chicago doctois do not tell their patients that their illness is terminal, Press, 1932), 26 THE Dii( OVERT OF GROUNDED IHEORT Generating Theory 27 Trow and Coleman compared the distinctive political nature ties and universals to strategic s r atior f thcory under dif conditions,10 of the flU with the characteristic political structure of other ferent an I to grounded modifications of they11 unions to establish their “deviant” case study. Wirth compared A touch of generation may be includcd but the researcher’s the Chicago with the is ghetto European to establish distinctive focus on verifving, he generates theory only in the service of changes in the new 8 modifying world ghetto Coleman with the aid of his original the as a r suit f he tests, And most IBM equipment, carefully of of this work distiriauished between types high is done with existing theorie , f r example, Blauner’s schools on dimensions, three themselves checked out empirically work with Marxian theory or Iipsct’s ssork with Michel’s to assure us that they are different in rriore than 9script. theory.’ This standard, required use of comparative analysis is accom 2 Some analysts focus on serifying the new theory that pushed early in the presentation of a study of emerges in for the purpose their data Thus, ie their work, theory is generated, getting the ensuing story but its straight. This use is, of course, sub emergence is taken for granted: what is intentionally sumed under the purpose of generating theory. However, when worked for is the verification of this emergent theory. The ana the analyst’s is lvstc are purpose only the specifying of a unit of anal sis, preoccupied with “checking out” the “emergent set ot he stifles his chances for generating to a greater degree than propositions.” Their favorite technique is looking for negative with any other use eases or of comparative analysis. The distinctive setting out deliberately to accumulate positive ones to empirical elements distinguishing of gain further the units comparison are evidence for their hypotheses And while, as in kept on the level Dalton’s of data, to insure clear understanding of dif research, great trouble may be taken in actively seeking ferential definitions. comparative As a consequence, the units’ general proper groups, other analysts may use comparative groups ties in common, which might occur to the analyst as he com incidentally or even implicitly. pares, are carefully No These unattended. ambiguity of similarity, researchers in specific studies do not seem to have such as a general underlsing property focused pervading all of them, directly on how their theory emerged; as a result, they is aliou ed between the competing have not units. Comparative analysis, explored how the’ could have generated more of it then, is carefully more put out of the picture, never to “disrupt” the systematically, and with more conceptual generality and monologue again. scope. A focus on testing can thus easily block the generation of a more rounded and more dense theory (see Chapter \‘I,). Ordinarily, we are presented with u elI-tested theory fragments, VerifyingTheory which can only partially account for what is happening in the researched situation. When the analyst turns to theoretical concerns, evidence is Also, we are presented with plenty of evidence coupled invariably used as a test of his hvpotheses—and thereby of the with at least implicit assurances that there were mountains relevance of his categories: comparative data give the best test more for verificatjon,because evidence is still most important Both implicitly and explicitly, the analyst continually checks to the analyst as the means for testing how he out his theory as the data pour in. Explicit verification beyond 10. For testing his hypotheses may lead to establishing major uniformi example, Robert Blauner, Alienation and Freedom (Chicago’ Uiiisersitv of Chicago Press, 1964) 11. See Robert K. Merton, Social Theory and Social Structure (New 7. S. M Lipset, Martin Trow and James S. Coleman, Union Democracy lork: Frce Press of Glencoe, 1957) Chapter Ill. (\ew York. Free Prei, of Clencoc, 1956). 12. See Blauner, op “it. and Lipset et al., op. cit. 8. Louis Wirth, The Ghetto (new ed.) (Chicago: University of Chicago 13. See, for example, Melville Dalton Men Who Manage (New York: Press, 1962) John Wiley and Sons, 1959 , and Howard S. Becker Blanche Geer, Everett 9. James Coleman, 7 he Adolescent Society (New York: Hughes and Free Press of Anseim L Strauss, Boys In White (Chicago: University of Glencoe, 1961), Chicago Press, 1961),
ferent knew
pooh’ ing doxically
eration. possible
while
never point and the
theory, service sis for
t suie, descriptions, everyone cially
terification
accurate 28
given to
t
generating
egins.
10
.1
1)P”
Generating
While
When
I
gis [4
professor. existing
new
Rcuvlmti
both
find.n
data
professors
a
c
the where
his
destror
research,
when
‘[Ia one e
his
tes
comparative
Beck
iv
of
o
This
Thus,
he
theories
themselves
evidence
with
urge
ar
an’
allows
of
theory
generation.
irs nvolved subsumes verifying
theory,
he
discovers
s
1
e o a
ii it
are
is
theories,
the
ni
verification
social
but
Theory ‘(
to
0
idenc
is
I
a
generation
not alike, 0 to
fhe e
vtal
as
not (I,’eago:
it
theon’
i
s’ critiques
especially 1
r
cantankerous
is
was
only
r
can
generate
wholly
.
ph
oWS
fails and
accurate e
clearly
so
ibid
evidence
research. sod
their
analyst’s is
nen from
job if
enthusiastically we
and
and
dmnc
cemer
Otherwise
be
“
crucial
“right”
the
to
they
(o
verified
p.
to
a
mountains
becomes
suggest
rm’vrsmty
Jr
of
purposeful
quickly
tells
of
generates
or
i
30);
the
t
are
‘a
realize
assumes
reseatcher’s
any
.
is
recognized
mc
-
true
hick t
are
cstrrsg theory
evidence I
or,-
The
in
or
when confidence
Of of normal;
is
somas
made
extent
and
colleagues
generality
THE
hypotheses.
part.
than
not
‘5000
personal
a they
because
of
it This
that
killed
course,
of
imp 1 so
that
Dahon
be
was
Chic
DISCOS
one S
through
time
“hooked” his
b cerifLations
a systematic
paramount
a
to
that
his
are
ten
K
lonco-oc
and single-spaced
focus
as accurate
as igo principal
S generated
a by
s
an theory—but
amm’i
generating.
(,hid.
FRY
purpose
main
wiy I, verifying
n
u
t sure
theory experience ev,dence
possible the
the
Presn
with
or”
sociologists,
the
influential
the
This
evidenec.
comparatiVe
destroyed
on
OF
he
dutis
on
gierated
to
‘inh1O
latter
twin
main goal
tells
generation
(;ROUNDED
Sac description as
1962’•.
evidence
their
only
verifying, and
happens stifle
and
is
is
typed
c
to
of
as
is
and
in
critiques
to
tsr
quite
or.
‘ riot
are
But goal
to modify
curb
isis
vital
colleague
We
own
it.
much
requisite
develop
generate
accurate
students
because
psges’
“pooh
testing
research THEORY
in to
theosa
analy To
be1ev” likeb
when
para
espe
from
S
tend ( of
gen
task
and
dif
the the
ic
i
be
as it.
of
f
a
and
Needless
there
ensure
York:
A
the
take
inal
extremely
of
taught
thinkers
source
York:
is
by “The verify parison
with logico-deductive
fortable
that
eralk and
plus
cOnnection.
has
the
sense, on
behavior This deductive
in theories
Fuithermore,
Case
dence
to
Generating
studr
17. a
data
hypotheses
16.
the
him,’ working
ought Generating
their
Basic
be
idea the
previous
an
erudite.’°
John
his
an
us). does
nothing
to
in
As
The
works
a Study
leaves
is of situation
to
peppered
“ not
forgotten
cautious
of
that think
unconfirmed
the
few
additional
promotion one’
and
completels
feeling
basis
long
warns
to
Wiley
Books,
are
careers
sa’..
A
it
anals
hypotheses, not
both
,“
insights
he
of
Theory
to
theory,
eXampIe,
to
analysis purports
grounded will
propositions
that
certain ago.”
us
in
we
speculative other
Iciree
to
such
data,
st
one The
of
sociologists
and
seem
theory
he
19fi4),
Philip
is
that
theories
with
theory
with
be
science may,
believe
in
or
in
a
Peter
do
is
opposed
researcher
into
Sons,
one,
a
in
presented
social
is
See
verifier
in ‘a priori
quite
an
used order
to
ignored.
relation,
set
since
to
some
to
hich
to
pp
it
a
with his
Hammond indeed,
carries
contrast
social
theory’
Blau,
be Stanley
organization
Verify
is 1964).
few
that
stray
fit
pertaining
of
explain,
scientists, best
book likely
than
would
174-75.
to
for
applying
he
theory, highly
will
to assumption
may speculations;
is
about
or
“thinker”
in
our
Exchange life
because his old
forcing
from
is
force verifying
Now, test
the
ing
p.
a
of
Fl.
this
can
dubiously
to
work
an
find
obliged
emphasis
to
presented
reformulated
(Ed.
theoretical
vii.
find Or conjecture—based
provide.
declining
evidence,
Udy,
since find
applicable and
a same
book
science.
an
grounded
discovery accurate staflc
be
this
the
the
he
a it
logico_deductive to
the
I
and
himself analytic
(and
that
“great
thes’
might
worthwhile
Jr.. used
mas’ to
have
d Sociologists
the
it
connection connection
only and
risk
benefit
on
focus
Power
by
with
and,
to
move
related
was
“Cross
“Unless
Whether
often
the
a
force
been
perhaps
man”
same speculations
force
as social
straying
unless
description
a
to
fit
of
a framework
have
theories,
gives
colleagues
hypothesis
merely
at
is
speculative
in part
touch
a
‘wlmc
between
areas
as
have theory.
Cultural
testing
strongly
the
in
said
fuller
the Social worst,
on
philosophers
between because
at
to
area
to
testing
been rc’.
favor
us
from or
of
connection
hr
their
Work
the
the researcher
risk
under
theory
one
the
Ci
of
thought
Blau
since
not to
the them,
Life
a
by
test
influenced
ss
Analysis:
Joi.ces
a of
the a
used
made
his
common
playing
existed new
uncom
ho
ultimate
an
a he
theory or
theory theory. area
theory
(New
fitting
theory
states: logico theory,
there
(New
com period
study.
orig
data
of
feel
area
gen
these
was
two
and
to
to
evi
is
by
up
a
of
a
29
31
I 4: 30 THE DISCOVERY OF GROUNDED tHEORY Generating 7 heory that ‘fits or works” in a substantiye or formal area çthough theory not only researched description.a Indeed, the market further testing, clarification, or reformulation is still necessary), ate, corpol and government fact-finding agencies can since the theory has been derived from data, not deduced from easiln outdo any sociologist in researched descriptions through shee logical assumptions. resources, if 1 they care to. V/here the sociologist can help these Since accurate evidence is not so crucial for generating agencies is by providing them with theory that will make their theory, the kind of evidence, as well as the number of cases, is research relevant And, as a brief reading of typical fact-finding also not so crucial. A single case can indicate a general con and market-research reports indicates, sociological relevance is ceptual category or property; a few more cases can confirm the sorely needed both for understanding the “dust heap” of data indication. As we note in the next chapter on theoretical sam piled up by agencies and for correcting the conventional pling. generation by coinparati e analysis requires a multitude ideology that guides this piling up of data,21 of carefully selected cases, but the pressure is not on the soci ologist to “knoss the whole field’ or to have all the facts “from a careful random sample. His job is not to provide a perfect What Theory Is Generated description of an area, hut to deelop a theory that accounts for mu h of the relevant behavior. The sociologist with theoreti This book is about the process of generating grounded theory cal generation as his major aim need not know the concrete so our nd polemic is with oth°r processes of arriving at theory, situation better than the people involved in it (an impossible particularly the logico-deductive, Grounded theory, it should task anyway). His job and his training are to do what these be mentioned, may take different forms, And although we con laymen cannot do—generate general categories and their prop sider the process of generating theory as related to its subse erties for general and specific situations and problems. These queut use and effectiveness the form in which the theory is can provide theoretical guides to the layman’s action (see Chap presented can he independent of this process by which it was ter X on practical applications). The sociologist thereby brings generated, Grounded theory can be presented either as a well- sociological theory, and so a different perspective, into the situa codified set of propositions or in a running theoretical discussion, tion of the layman. This new perspective can be very helpful using conceptual categories and their properties. to the latter. 22 Sociologists who conceive of this task as their job are not 20. We are in complete agreement with Zetterberg on this issue of whether sociology will plagued (as are those who attempt to report precise descrip advance more by cuncentrating on theory or on niethodologv. But we tion) by thoughts such as “eervbody knows it, why bother to feel that a methodology of generating it is needed for theoretical advance, See Hans L. 18; Zetterberg, On lheorzj und Verification in Sociology ( write a book” or feelings that description is not enough: a Totowa. NJ.: Ilednsinster Press, 1963). Preface. 21. A good instance is good sociologist from Chicago must do more, but what” Soci the sociological relevance of vast amounts of govenimentl statistics or, the differential medical ologists who set themselves the task of generating theory from strata care of socioeconomic in America. The common-sense meaning of these statistics the data of social self evident, hut is almost research have a job that can he done only by deeper sociological significance neither guides these ernmental surveys gov the sociologist, and that offers a significant product to laymen nor much affects agency policies. What sociologists about socioecononuc’ know life st”les arid about the organization of medical facili and colleagues alike, Research sociologists in their driving ties can easily be brought to hear upon government data. See policy paper efforts to get the facts tend to forget that, besides methodology on medical care by Anselm Strauss, written for the Studies Institute for Policy th distinctive offering of sociology to our society is sociological (Washington l).C.. July, 1965), 22. This choice is not news. since most theory is written whether grounded this way, oi logsco-cleductive. But we have noted this decision, 18. Bla schc Geer, ‘F rst Days in thc Fi Id,’ in I{ammond, op. cit, the request of on several colleagues, to fend off the critique that the only true p. 322, theory is the one written, by the numbers as an integrated set of 19. David Reisman and Watson, Project: A tions, proposi Jeanne “The Sociability The form in v,hich a theory is presented does not it make it a theory’ Chronicle of Frustration and A hi scment” n Hammond, op. cit., p 292. is a theory because it explains or predicts something. 32 THF DISCOVERY OF GROUNDED THEORY Generating Theory 33 We have choscn the disussional form for several reasons hes fall et veer F ‘m’ or s ‘orking Fypotheses f e crydav Our strategy of compalative analysis for generating theory puts life and the ‘all-inclusiv’ grand theories.23 high emphasis on theory as process, that is, theory as an Substantive and formal theories exist on distinguishable ever-developing entity not as a perfected product. (The reader levels of generality, which differ only in terms of degree There s’ll Sec furthe what e mean in CI apters HI and IV.) To be fr ‘i iny one study ach type ar shade at points into the sure as process can in as a e theory be presented publications other The analyst, ho houlJ füua Llc’stllyon one level momentary product, hut it is written sith the assumption that or other or on a specific combinat’on, because the strategies is still Theory as believe, renders it developing process, we vary for arriving at each one, For example, in our analysis of reality -I as quite well the of social interaction and its structural ing a nonscheduled status passage, the focus was on the context, ubstantive area ‘f dying not on the formal ‘rca of status pa of gives a feeling age 2 The discussional form formulating theory With the to us on a substantive area such as this, the of “ever-developing” to the theory, allows it to become quite generation of theory can be achieved by a comparative analysis rich, complex, and dense, and inaks its fit and relevance easy b ssccn or among groups within the same substantive area In to comprehend. On the other hand, to state a theory in propo this instance we compared hospital wards nhere patients sitional form, except perhaps for a few scattered core proposi haracteristically died at different rates 1 he substantive theory tions, would make it less complex. dense, and rich, and also could be generated b comparing dying as a status passage more laborious implication to read, It would also tend by ssith other substantive cases within the formal area of status to “freeze” the theory instead of giving the feeling of a need passage with other substantive cases within the formal area of for continued development. If necessary for verificational studies, status passage, whether scheduled or not, such as studenthood parts of the theoretical discussion can at any point be rephrased or engagement for marriage. The comparison would illuminate as a set of propositions. This rephrasing is simply a formal the substantive theory about dying as a status passage. exercise, though, since the concepts are already related in the However if the focus were on formal theory, then the com discussion, Also, with either a propositional or discussional parative analysis would be made among different kinds of sub grounded theory, the sociologist can then logically deduce stantis e cases which fall within the formal area, without relating further hypotheses. Indeed, deductions from grounded theory, them to any one substantive area, The focus of comparisons is as it develops. are the method by which the researcher directs now on generatirg a theory of status passage, not on generating his theoretical sampling (see Chapter III) theory about a single substantive case of status passage. Both substantive and formal theories must be grounded in data, Substantive and Formal Theory Substantive theory faithful to the empirical situation can not, we believe, be formulated merely by applying a few ideas is Cowparative analysis can be used to two basic from ‘I4 generate an established formal theory to the substantive area, To k nds of theory: substantie formal. By substantive theory, be and sure one goes out and studies an area with a particular we mean that developed for a substantive, or empirical, area of sociological perspective, and with a focus, a general question Ii sociological inquiry, such as patient care, race relations, r pro a problem in mind, But he can (and we believe should) also fessional education, delinquency, organizations. By or research study an area without any preconceived theory that dictates, formal theory. mean a we that developed for formal, or con prior to the research, “relevancies” in concepts and hypotheses. ceptual, area of sociological inquiry such as stigma, deviant behavior, formal organization, socialization, status congruency, 23. See Merton, op. cit, pp 5-10 24. Barney authority and power, reward systems, or social mobility. Both C Claser and Anselm L Strauss, “Temporal Aspects of Dying as a Non-Sch duled Status types of theory may be considered as “middle range.” That is, Passage,” A,nerican Jow’nal of Sociology, LXXI (July, 1965), pp 48-59. 34 THE DISCOVERY OF GROUNDED FHEORY Generating Theory
Indred it is presumptuus to assume that one begins to know that is generated from logical assumptions and speculations the relevant categories and hypotheses until the “first days in about the “oughts” of social life. the held,” at least, are over, A substantive theory generated Within these 25 relations existing among social research, sub from the data must first be formulated, in order to see which of stantive theory and formal theory is a design for the cumulative diverse formal theories are, perhaps. applicable for furthering nature of knowledge and theory. The design involves a pro additional substantive formulations. gressive building up from facts, through substantive to grounded Ignoring this first task—discovering substantive theory rele fonnal theory. To generate substantive theory, we need many vant to a given substantive area—is the result, in most instances, facts for the necessary comparative analysis; ethnographic stud of believing that formal theories can be applied directly to a ies, as well as direct gathering of data, are immensely useful for ‘ubstantive area, and will supply most or all of the necessary this purpose. Ethnographic studies substantive thcories and concepts and hypotheses. rhe conseqence is often a forcing of direct data collection are all, in turn, necessary for building up data, as well as a neglect of relevant concepts and hypotheses bs comparative analysis to formal theory. This design then, that may emerge. Our approach, allowing substantive concepts locates the place of each level of work within the cumulation of and hypotheses to emerge first, on their own, enables the analyst know ledge and theory, and thereb suggests a division of labor to ascertain which, if any, existing formal theory may help him in sociological work. generate his substantive theories, lie can then be more faithful This design also suggests that many ethnographic studies and to his data, rather than forcing it to fit a theory. He can be multiple theories are needed so that various substantive and more objective and less theoretically biased. Of course, this also formal areas of inquir can continue to build up to more inclu means that he cannot merely apply Parsonian or Mertonian sise formal theories. Such a call for multiple theories is in con categories at the start, but must wait to see whether they are trast to the directly monopolistic implications of logico-deductive linked to the emergent substantive theory concerning the issue theories, whose formulators claim there is only one theory for in focus. an area, or perhaps even one sociological theory for all areas. Substantive theory in turn helps to generate new grounded The need for multiple theories on the substantive level may be formal theories and to reformulate previously established ones. obvious, but it is not so obvious on the formal level, Yet multiple lius it becomes a strategic link in the formulation and develop formal theories are also necessary, since one theory never ment of formal theory based on data. For example, in our handles all relevancies, and because by comparing many theories theory bearing on “awareness contexts” relevant to dying, two we can begin to arrive at more inclusive, parsimonious levels. important properties are cues leading to awareness and the The logico-deductix-e theorist, proceeding under the license and personal stakes involved in the various parties’ becoming aware. mandate of analytic abstraction, engages in premature parsimony Currently, in generating a formal theory of awareness contexts, when arriving at his theory. (In Chapters III, IV and V we we are developing the generalities related to stakes and cues by shall discuss in more detail the relations of research to studying such groups as spies and building subcontractors. A the gen eration of substantive and formal theory.) dying patient or a spy has a great stake in any type of aware ness context, and a subcontractor has a quantifiable or monetary Elements stake. In Chapter IV, we shall discuss more fully the generation of the Theory of grounded formal theory. Suffice it to say that we use the As we shall discuss and use them, the elements of theory word grounded here to underline the point that the formal theory that are generated by comparative analysis are, first, conceptual we are talking about must be contrasted with “grand” theory categories and their conceptual properties; and second, hypothe ses or generalized relations among the categories and their 25. Geer, op. cit. properties.
provided
collection
ties
userriding
clearh
quickly their
Dying ir sidering
gories
areas—as It
g
that a
also
are tionship
of
related’
“loss conceptual
to their of
uch sure and
both
category
as
their
36 ween
xc
takes
26.
ategory
Although
themselves
a
It
a
The
nursing
Categories
urde
that
concepts
occupation. 2
indicated
that
patient
changes conceptual
rise categories
when
pr
Patients,”
must view
attention
ratioriales”—tliat
See
and ‘perc,ptions
these
be
during
constant
much
perties,
low
helps
elaborate
that hese
to
both
Barney
and
well
coding
important
and
of
cinervation.
be
properties
facing
or
care
aspect
them,
w
two
categories
ratzcnale
in indicated
ss
Amcricue
the
more
their
hors
kept
property
property
it
the
the
leads
integrating,
vary
as hat
and
to
and
the
C.
element
which,
comparing
will
are
element
and
data
their
his One
the
nurses
early
them—tend degree
or
Glaser
perceptions
they
in of
evidence—usually
their
original
prope
in
to
him
the
not
have
death.
Jourual
mind
creation
analysis eiem”
social
property
are
arise
s,
degree
can
by
is
many
a
Lower
since indica
phases
of
uroerties.
and
cc
necessarily
to
theory
conceived,
to of
of
the
tie.
uria
conceptualizations—and
continually
the
a
the be
of
category.
that
as
maintain among
of
loss
Anseim ti-
loss’
life
generate
they
similarities
s’
many
of
of
rationales
Nursing, of
of
borrowed
a
data
of
of
level
to theory
HF
Making
of
both
explaining
high
his
“professional
apart
social
a
the conceptual
y. data
a
a
come
the
of
emerge
For nurses
DISCOVERY
L.
systematic
better
A
death
(and
alter,
category.
groups
categories
a
from
In .A
their
data,
categories
social a
64
studied
Strauss,
category
abstract category
collection.
change
from
loss. example.
short,
a
later
dying
and
property,
(june, from
nurses
not
clarify to
will
different
category
from
professional
distinction
the
All
draws loss,
OF
the
abstraction.
explain
“The
differences.
during
the
to conceptual
composure’
We
existing in
1964),
he
relationship
patient kind
GROU’DED
categories
of
stands
three
emerge
and
the
two
use
evidence
make
and
Higher
or
the
to
social
Social
data
in
the
the
have,
substantive
to
destroy
data,
of
properties
his
pp. categories
the
to
the are
turn,
this
evidence
between
by
proper
sociolo
achieve
theory,
certain
that
compo
behav
itself);
Loss
119-22.
rather
family
justify
loss THEORY
level,
death
inter
Once
Con
cate
joint
then.
itself
rela
that
will
and
and
is
be
it.
is
of
is
a
nections
abstractions
Watson, sized
aim
zation tively
ceptual
areas.
established
will
under
literally
relevance, signed
focus
long
harder
disbelief
data
‘ng tion
“square
on
in
theory
ally
ing,
quacy
most categories,
lished
selection
this,
emergent
that
Generating
27.
This
While finding
at
them.
not
By
with
at
See being
for
run then’
to few
on
as
the
Similarities
study,
Merely achieving relevant
Op.
for
position
of
as
level.
to
by
to
between
contrast,
to
he
categories”
apply
possible.
the Hie
of
the
the categories
Cit.,
indicators
borrowed
they
mawj
forcing,
major
find,
the
The
indicators
the
ignore
another
conceptualizations,
fullest
Also,
after
of
Theory
developed
contaminated
because
both
in
case
emergence
verification
pp.
selecting
tentative
categories
situation
and
purpose,
to
order
e
may
on
result
feer
much
Ice
uniformities
the
history
data
305-09. believe colleagues
emergent
when
and The
categories
the
possible
and
the the
dr
theory
is
for
fit,
categories
the
not
and
to
analytic
literature
buttressed
in is
eonvergences
dn.’ersiñj
synthesis
and of
and
diversity
richness, best
data
generating
on
under
emergent
relationship
assure
of they we
major
and
that
of
number,
conceptual
in
be
that
by
tends
this
categories
generality
checked
lower
fitted
categories
getting
theory try
and
or for
relevant,
must
concepts
properties. and
our
core
There
study
the
problem
that
effort
is
properties
in
of provides
of
An
to
to
a
laymen
by
categories
to and
more
sariations
and
emergent
forcing
generation
category theory
conceptual
theory
be
hinder
aims
the
fit
of
agreement
effective
for
the
with
a
between
and
are
we
is
solves
and
respecifled.
long
cnr’fronted
categories
highem and
more
a
not
usually
emergence
not
relevance.
data,
difficult
from
a
at
can
hypothetical
readily
category
does
the
and
the
is
meaning
of
number
as
are justifleatory
establishing
is
generation. that
the
categories,
suited
strategy
on
sure have
As
“round
of
literature
level
the the
rich;
les seldom
generation
among
put
fact
prove
not
problems
by
the
they
apparent
has
since
theory
el
has
start, 27
to
In
of
Also
two.
much
from
a
of
on
Reisman
has
to
since
conceptual
are
exactly
same
premium
been
categories
arouse
is,
short.
data”
a are
reasons
emerged. colleagues
generali
to
different
the
they
svpthe
problem.
explana
impor
but
the
can
Forcing
continu
should a
at
another
trouble
Work emerg
be
of
in
of
con
rela
con
estab
area
first,
md
our into
ade
data
new
de
be
the
are
the
fit,
the
for
on 37 38 THE DISCOVERY OF GROUNDED THEORY Generaizrig Theory 39 taut consequences both for the sociologict and for sociology. enable one to gras the reference in terms of one’s own experi As the sociologist uses standard sociological concepts, he soon ence. To make concepts both analytic and sensitizing helps discovers that they usually become very differently defined, the reader to see and hear visidly the people in the area under dimensioned, specified. or typed. Typical boundaries of the study, especially if it is a substantive area. This perception. in standai d concept become broken Furthermore, the boundaries turn, helps the reader to grasp the theory developed for the of the established battery of sociological concepts are also area, To formulate concepts of this nature, bringing together broken, As he discovers new categories, the sociologist realizes the best of tw’o possible worlds, takes considerable study of how few kmds of behavior can be coped with by many of our one’s data and requires considerable data collection of incidents concepts, and recognizes the need to develop more concepts by bearing on a category. If, when a category is but scarcely estab straying out of traditional research areas into the multitude of us icd. the sociologist turns to collecting data for another poten substantive unknowns of social life that never have been touched ml ategory, slighting the newly established one, the latter is —to gie only a few examples, building subcontracting, auction- l’l—elyto lack development both in sensitizing and in some ccring, mortgaging, or the producing of plays by amateur theater ef its analytic aspects. A balance must be struck hetw’een the grouns. two lines of effort in accordance w’ith the theoretical saturation As one thinks about the broad spectrum of social life, one of categories (a strategy we shall discuss in Chapter III). realizes that sociologists (with the focused aid of foundations Hypotheses’. The comparison of differences and similarities have reall worked in only a small corner of it when posing among groups not only generates categories, hut also rather the larger questions of deviance, social problems, formal organ speedily generates generalized relations among them. It must izatons, education, mental health, community government, un he emphasized that these hypotheses have at first the status of dercleveloped countries, and so forth, One also realizes that a suggested, not tested, relations among categories and their prop great many more formal theories of sociology have yet to be rties, though they are verified as much as possible in the course generated about such additional areas as loneliness, brutality, o research resistance, debating, bidding systems, transportation, mail-order Whether the sociologist, as he jointly collects and analyzes distribution, corporate collusion, financial systems, diplomacy, qualitative data, starts out in a confused state of noting almost and world interdependence through business systems. One verything he Sees because it all seems significant, or whether strategy for bringing the generation of theory to greater impor F starts out with a more defined purpose, his work quickly tance is to work in non-traditional areas where there is little or leads to the generation of hypotheses. When he begins to no technical literature, Finding non-traditional areas is also a hypothesize with the explicit purpose of generating theory, the strategy for escaping the shackles of existing theory and con researcher is no longer a passive receiver of impressions but is temporary emphasis. The sociologist who does so can easily draw n naturally into actively generating and verifying his find himself not merely generating a new theory but also open h potheses through comparison of groups. Characteristically, in ing a new area for sociological inquiry—virtually initiating a this kind of joint data collection and analysis, multiple hypothe new portion of sociology. ‘Whether he studies less or more tra ses are pursued simultaneously. Some are pursued over long ditional areas, hossever, the first requirement for breaking the i eriods of time bei.ause their generation and verification are bounds of established sociology is to generate theory from data. linked with developing social events. Meanw’hile, new hypotheses The type of concept that should be generated has two, joint, a e continually sought essential features, First, the concepts should be analytic—suffi Generating hypotheses requires evidence enough only to ciently generalized to designate characteristics of concrete enti ties, not the entities themselves. They should also he sensitizing 28 On sensitizing concepts see Flerbert Blumer, “What is Wrong with Social Theory,” American Sociological Review, 19 (February 1964). —yield a “meaningful” picture, abetted by apt illustrations that to 3-lu. 40 ISlE DI5C0ERY OF GROUNDED THEORY Generating Tleory establish a suggestion—not an excessive piling up of evidence bon the fullest range of conceptual levels: anyone o uses the to establish a proof. and the consequent hindering of the gen integrated theory can start at a more general level and, focus eration of new hypotheses. In field work, however, general rela ing upon a specific area within the theory, ‘aork down to data. tions are often discovered in vivo; that is, the field worker still guided by hypotheses for limited, specific situations. For 1vliteral sees them occur This aspect of the “real life” character those who rise the theory, these less information-packed hypothe of field work deserves emphasis, for it is an important dividend ses may be as important as the more general theoretical ones; in generating theory. (We shall say more about this point when for instance, a sociologist studying the awareness of dying discussing the credibility of analyses of qualitative field data in p tients on a surgical ward, or nurses trying to apply aware Chapter IX) sess theory to family relations as observed on an emergency In the beginning one’s hypotheses may seem unrelated, but a d (although not on all wards) as categories and properties emerge develop in abstraction, and I must be emphasized that integration of the theory is become related, their accumulating interrelations form an inte be t when it emerges, like the concepts. The theory should grated central theoretical f”arne’vork the core of the emerging icver just he put together, nor sh )uld a formal-theory model 1f/icon The core becomes a theoretical guide to the further be applied to it until one is sure it will fit, and will riot force .collection and anal si of data. Field storkers have remarked the data. Possible use of a formal model of integration can he upon the rapid crystallization of that framework, as svell as the determined oni after a substantive model has sufficiently iapid emergence of categories ‘When the main emphasis is emerged. The truly emergent integrating framework, which on crifying theory, there2 is no provision for discovering novelty, encompasses the fullest possible diversity of categories and and potentially illuminating’ perspectives, that do emerge and properties, becomes an open-ended scheme, hardly subject to might change the theory, actually arc suppressed. In verification, be-ing redesigned. It is open-ended because, as new categories one feels too juickly that he has the theory and now must pi operties are generated and related, there seems alw iys to check it out.” When generation of theory is the aim, however, s place for them in the scheme, For substantive theory, the one is constantly alert to emergent perspectives that will change st is very likely to discover an integrating scheme within and help develop his theory. These perspectives can easily occur i b to, since the data and the interrelations of his theory lie yen on the final day of study or when the manuscript is re USC together i viewed in page proof. so the published word is not the final fosvever the comparative analysis of diverse kinds of sub one, but only a pause in the never-ending process of generating antive groups though aimed at generating “grounded” fosmal theory, W’hen erifieation is the main aim, publication of the ti’eor can takc the researcher far from from emergent substan study tends to give readers the impression that this is the last Integrations. Then existing formal models of process and w’ord ure and analysis become useful guides to integi ating 11w Integration. IntegratIon of the theorv—sshieh takes place at i’s-s of a formal theory—provided that integration is not the many les els of generality that emerge—does rc, ‘i on the theory. not necessitate \lodels of integration for substantis €‘ a distinction between “working” (or “ordinary”) and theoretical lu-Cl v that are c’orived from the data are hot necessanli aprite hypotheses. Our emphasis on integration takes into considera 5b1 to o’her u st-intise are, s em transfer si on be at pted s th gr ‘it autien i d ni after tiy’r is -Ovci 3 29 Our colleague, Leonard Schatzman, has called this the “momentum ° a seig n t gnat r first effect ‘i he emergence of categories and theoretical perspective gains such momsntum that a researcher must usually retire from the field after the C e a r i egr o bsta sv i or h first few days to appraise the data and establish an order for what is hap pening. He stops being drowned flood x by the of data and startsto plan his For cl rs dtr the Ste ate theme of ‘ ‘ and theoretical sampling. surss ‘s Con osu CliapF-r 13 CIa cr arid Sin is Awar iSf 30 Zcttcrbcrg,op. oft, p. 21, and paasimr, Jing.
by
two phasize a stantive culating
loss—includes
th nurses’
that are distribution this social
are cation cussed peop]e
Hypotheses Pioperties umerenc rizing
CategorY
42
stat-is
would
“Temporal
useful
Elements
experts.
Pavir
3:3.
32
lie
generated
beginning
r’es
kinds
same
“work”
passage—legitimatine.
provide loss
.
The
Claser is
of
in
attitudes
theories
too
following
is
social
of
e
beginning
referred
The
Chapter
Aspects
ill
integrative
of
heed sehe
of
wa
of
Our
of
Category’
strongly—whereby
of
and
corres
a
Theory
dyir,g
predict
theory
from
concepts
move
the
descriptions
useful
we
lose
n
services
substantive
to
Strauss,
that
and
of
hav
to
chart
I\
efici
04
these
Dying
j-atiei
data,
cud
for
‘,Q(
beginning
from
that
in
deal
and
iehavior
schemes,
mt
integrated
t
lal
“rue
patients
nurses
care.
rationales The and Calculating
away on
teristics Social
loss
Chapter
(I)
as
integrating
provides
strioti of
an
‘rrs’lati
as bid
announcing
we
is
explain—and
substantis
s—i
loss
basis
of
affected
of
integration,
ith
The
the
apparent
a
higher
Social insure
Substantive
his
to (2)
a have
loss
son-Scheduled
a
develop
ndei
re
dying the
of
tories
substantive as
the
better
of to
socia 1
a
InC
proc’ess—srhich
to
death
We
iis
lU
of
thec’rv
The
patient
Loss
social
examples
the
learned
the
on
explain
social
is
study
discussed:
that
DISCOVF,RY
dying
s
a
patient,
by
e
and
charac
eli
he
canrot
and
his
Type
loss
ould
social
em
more
loss
to formal
of
and
“real”
loss
however,
of
the
do
major
as
of
substantive
uoordrnating
Dying
value
formal
will
dying of
status
gence
of
the
not
apply
to
Status
and
social
of
say
Theory
is
OF
The
services less istics
ences
basis
apparent Calculating
value
Social
dying
theory
aloe
be
categories
strictures
orld,
Patients,”
of
impact
elements
sound
as
GBOUCDED
passage
formal
we
whether
levels
delay would
and
Passage,”
higher
people
to
explicitly
a
of
of
value
in
of
salsie
Formal
the
These
non-scheduled
cannot
patients
and
from
learned
other
a
receiving
person
character
about
the
‘sviridy”
he
pcrson
of
in
O
of
passage—
theories
the
provide
social
of
on
forrna 1
of
served
of
of
experi
or
general, IHEORY
op. social
experts
ipp 1 i
theo
people
rules
Cit.
sub
social
em
cal
and
on
dis
the
the ot
the
the
os
the
cit.
t
and
and
generation
Chapter
operation, tion
leing are
such
tendency’ should
mflt of
other
investigation three
important
Generating
ished
ie
oupled
data
Ir
discussion
slighted
analysis,
analysis,
hinders
a
chapters
operations
coded
concluding
rules
studies
blur
definite
is
III
sith
is
Theory
the
aspect
the
of
and
to
or or
we
and
as
which
generation
to
the
is
of
theory
of
idea do
underlying
tocus
ignored.
plain
ill
intertwine
data
its
discuss
be
our
description
this
notion
a of
all
focus
may
end,
fresh
imply
done
on
generating
routine—thus
book
three
are
chapter.
To
This
one
the
on
of
of
be
To
collected.
analytic
operation
continually,
together
pursue
considerable
simultaneously’.
Joint
theory
the
theory.
operation
disregarded
relations and
be
definite
sic
relations
theory
collection,
sure,
verification,
uish
this
idea
as
stifling
as
F
The
separation
or
proc
between
from
at
in
much
to
vital
that
coding;
emerges
because a
example.
between
any
generation
at
hut
nphasizc
time
coding.
ss,
the
pervades
tactic
as
that
there
investigation
requires
in
beginning
data
that
of
in
possible.
that
many
of
joint
moment
and
each
if
Chapter
further,
is
one’
the
pre-estab
of
collection
data
jolts
typically
this
analysis
that
coding
theory,
(if
opera
others
highl
of
They
and
the
the
are
not
the
V,
an all
in
43
I ii
covers
of
gories
eepts
hold
nor
mission that will
work
features
framework.
The
to
1cm
marks
eial
emerging
emerges
cisions
and
and
gcnerating
these
The
students
Theoretical
are
there
study initial
on
sociological
where
area
analyzes
ahoet
of
of
that
or
for
they
his
procedures sociologist
concepts of
local”
his
how
This
will
theory,
decisions
theoretical
the
research,
some
For
theory
such
to
the
in
likely
tbeorw
he
his
policemen
structure
find
process
medical
sampling
example.
concepts.
problem
perspectie
anticipated
to
doctors,
as
whether
may
data
to
whereby
These
them,
are
his
Of
His
how
become
collection
begin
of
course,
not and
problem—this
school
and
categories
itself,
he
act
is
nurses,
designating
concepts
substantive
in
confidence
data
based
the
and
decides
the
knows
processes
“local”
part
toward 45
order
the
he
that
of
not
collection
Theoretical
process
analyst
on
and
research
on
of
data
does
are
give
to
its
before
concepts
a
turns
the
what
a
Negroes
aides,
a
or
men
problem
general
in
develop
preconceived
are
not
situation.
more
few
him
core
formal,
jointly
of
the
vith
is
based
them
data
studying
know
handle
and
data
principal
a
controlled
likely
situations
explanatory
may
or
subject
beginning a
must
his
wards
to collects
The
into
what
only
the partial
klso,
Sampling
collection
remain
theory
collect
to
prospective
theoretical
a
emerge—
relevancy
initial
on
doctors’).
or
be
or
happens
hospital
and
he
that
frame
by
a
codes
cate-
prob
gross
foot
con
as
dis
un-
gen
next
ad
de
the
he
for it 46 TIlE DISCOVERY OF GROUNDED TBEORY Theoretical Sampling 4T used in the situations releant to his problem--doctors may, for ceived theory is likel) to be readily dropped or forgotten be the problem, be called therapists—and he discovers many more cause it now seems irrelevant to the data.’ structural and processional “bear concepts than he could hae Beyond the decisions concerning initial collection of data, anticipated before his research further collection cannot be planned in advance of the emerging Thc.sociologist should also be sufficiently theoretically ,sen-si theory (as is done so carefully in research designed for verifi e ‘a h—can cc’—“pual’-e re-I f’rriulatc a theory as it cation and cleec’ription) The emerging theory points to the next emerges from the data. Once started, theoretical sensitivity is steps—the sociologist does not know them until he is guided forever in continual development. It is developed as over many by emerging gaps in his theory and by research questions sug years the sociologist thinks in theoretical terms about what he geted by previous answers. knows, and as he queries many different theories on such ques The basic question2 in theoretical sampling in either sub tions as ‘“What does the theory do? How is it conceived? What stantive or formal theory) is: what groups or subgroups does one is its general position? \Vhat kinds of models does it use?” turn to next in data collection? And for what theoretical pur Theoretic-al seiisitixitv of a sociologist has two other character pose? In short, how does the sociologist select multiple compari istics First, it involves hic personal and temperamental bent son groups? The possibilities of multiple comparisons are in Second, it involves thc- sociologist’s ability to have theoretical finite, and so groups must be chosen according to theoretical insight into his area of research combined uith an ability to ri teria, make something of his insights (see Chapter XI). In actuality, many sociologists escape this problem of select These sources of developing theoretical sensitivity con ing groups by studying only one group during a given research, tinually build tip in the sociologist an armamentarium of cate uith some slight effort at delineating subgroups, and with occa gories and sional references hypotheses on substantive and formal levels. This (usually in footnotes 1 to comparative findings theory that exists within a sociologist can be used in generat on another group, typically followed by a brief description of ing his specific theory if, after study of the data, the fit and rele differences, but not by a theoretical analysis. In other studies, vance to the data are emergent. A discovered, grounded theory, particularly survey research, comparisons are usually, and quite then, will tend to combine mostly concepts and hypotheses that arbitrarily, based on only one different substantive group (such have emerged from the data with some existing ones that are as natural scientists compared with social scientists, or scientists clearly useful, We have put most emphasis on the emergent vith engineers); or the comparisons are based on several sub concepts—those coming from the data. Still, whether the theo groups within the substantive group. And in “comparative retical elements are emergent or already exist with fit and tudies” of more than two groups, the sociologist usually tries relevance that emerges, the strategies of comparative analysis to compare as many as he can of the groups for which he can presented in this and the next two chapters apply. Potential theoretical sensitivity is lost when the sociologist 1. For an excellent discussion of this phenomenon see James Coleman, commits himself exclusively to one specific preconceived theory “Research Chronicle: The Adolescent Society,” in Philip Hammond (Ed,). Soczologist.s at Work (New York: Basic Books, (e.g.. formal organization) for then he becomes doctrinaire and 1964), pp. 198-204. 2. For example, in our study of the patient’c awareness of dying related can no longer “see around” either his pet theory or any other. t- medical staff-patient interaction, after we had saturated the various texts con in which this occisrrecl, we realized that we should collect data on He becomes insensitive, or even defensive, toward the kinds of additional situations where patient awareness is discounted So we looked questions that cast doubt on his theory; he is preoccupied with oseb for this at 1 staff-patient interaction on an emergency ward. See Bar testing, modifying and seeing everything from this one angle. ney C. Glaser and Anseim L. Strauss,Auareness of Dying (Chicago: Aldine Publishing For this person, theory will seldom truly emerge from data. Co, 1965), Chapter 7. 3. The reader may consider aggregates or single people as the In the few instances where lents equiva theory does emerge, the precon of groups, with respect to the strategies of comparative analysis. 48 THE DISCOVERY 0 GROUNDED THEORY Iheoretica Sampbrg 49 obtain data within the limits of his own time and money and r matter F desinn, ‘ poor t r d If F ar es his task to his degree of of mee these access to those groups. The resulting set groups unanticipated contingencies, readers may judge that is then justified by citing common4 factors and relevant differ his facts have been contasni iated by his personal violation ences, stating that this constitutes all the available data any o the preconceived impersonal rules. Thus he is controlled by how Further comparison are left to future researchers ii s mpersonai rule aid Fas to cortrol over th relevancy of Although these methods of choosing groups yield worth hr data, even as he sees it go rtrnv while research, they do not employ the criteria for theoretical sampling that we shall discuss in this chapter. Our criteria are those of theoretical purpose and relevance—not of structural cir Selecting Comparison Groups cumstance, Though constrained by the same structural circum” stances of research, In this sect o i we do not base research on them. Ihe we focus in tsso questions: which groups are criteria may appear flexible (too much so for validity, one dec ed, why and how? critic has said), but the reader must remember that our main purpose is to generate theory, not to establish verifications with Which Groups? the “facts,” We trust that these criteria will also appear to create a more systematic, relevant, impersonal control over data Ti e basic cr tenon g verning the s°l°ctiop ‘omparison oups for collection than do the preplanned, routinized, arbitrary criteria discovering theory is their theoretical releeance for Ii thering based on the existing structural limits of everyday group the deselopment of emerging categories. The re earcher boundaries. The latter criteria are used in studies designed to chooses any groups that will help generate to the ulIest extent, get the facts and test hypotheses. One reason for emphasizing as many properti’s of the categories as possible, s id that this difference in control is immediately apparent. The criteria viii help relate categories to each oth r and to their of properties. s theoretical sampling are designed to be applied in the on Thus, as e said in Chapter 11 group comparisons a e conceptual; going joint collection and analysis of data associated with the they sre nade by compar’ng diverse or similar idence generation of theory. Therefore, they are continually tailored to iidicating the same coreptual categories and proper bes, iot fit the data and are applied judiciously at the right point and by comparing the e idence fir ts o n ak Compara e analysis moment in the analysis. The analyst can continually adjust his kes ful’ ad r ag c the ‘afrchangeability’ indicators control of data collection to ensure the data’s relevance to nd de c op i pr eeds a broad rarge of icceptable the impersonal criteria of his emerging theory. ndicators for catogc ries and proj erties C By i cc contrast, data collected according to a preplanned rou gio Ip na4 be chos ii for i ig comparisoi onl tine are more likely to force the analyst into r in be r o deli i irrelevant direc re ‘ eel epla r d c o’ groups tions and harmful pitfalls. He may discover unanticipated rat or co ry e Ic con al o mo go e s F e a e tingencies in his respondents, in the library and in the field, xc I but is unable to adjust his collection procedures or even redesign n i y lii not pe n it we to 11 s his whole project. In accordance with conventional practice, I ph Ce ‘‘its hat oud’ (ut rI 4 ‘3 s ti dy’ Har r ond I pa the researcher is admonished to stick to his prescribed research trio s [or h develop d a e Y ‘, h S C 0 ore I pr ‘nr 4 For examples see Coleman, op. cit., and The Adolescent Society d itocj ov ed (New York: Free Press of Glencoe, 1961) Morris Janowitz, The Military c H S lv ‘‘1 F n erpla of oc I in the Political Development I of New Nation.s (Chicago: University of Chi itS u , ii So ‘ol l.a ol VII cago Press, 1964), or Seymour Martin Lipset and Reinhard Bendix, Social .9 0 ) Mobility in Industrial Society (Berkeley: University of California Press, 1959). 50 THE Dli OVEHY OF GROF,NDED THE SIlT Theoretical Sampling in comparative st idks macic for accurate descriptions and veri To be sure these rules of comparability are important when fication) In research carried out for discovering theory, the accurate evidence is the goal but they hinder the generation of sociologist cann( t cite the number and types of groups from theory’, in which “non comparability” of groups is irrelevant. uhich he collected data nnf ii the research is completed In an They prevent the use of a much wider range of groups for de evtrer e ca h may then find that the development of a h veloping properties of categones Such a range, necessary for major cateorv may have been based on comparisons of differ the categories’ fullest possible development, is achieved by ent sets of groups. For example. one could write a substantive comparing any groups, irrespective of differences or similarities, theory about scientists’ authority in organizations, and compare as long as the data apply to a similar category’ or property. Fur very different kinds of organizations to develop properties asso thermore, these two rules divert the analyst’s attention away dated with the diverse categories that might emerge: authority from the important sets of fundamental differences and simi orer clients, administration, research facilities, or relations with larities. which, upon analysis, become important qualifying con outside organizations and communities: the degree or type of ditions under which categories and properties vary’.These differ aliation in the organization; and so forth. Or the sociologist ences should be made a vital part of the analysis, but rules of may wish to write a formal theory about profescional authority comparability tend to make the analyst inattentive to conditions organizations; then the sets of comparison groups for each that vary findings by’ allowing him to assume constants and to category are likely to be much mon diverse than those used disqualify basic differences, thus nullifying their effort before in developing a substantive theory about scientists, since now the analysis. the field of possible comparison is far greater. It is theoretically important to note to what degree the Our logic of ongoing inclusion of groups must be differenti properties of categories are varied by diverse conditions. For ated from the logic used in comparative analyses that are example. properties of the effect of awareness contexts on the focused mainly on accurate evidence for description and veri interaction between the nurse and the dying patient within a fication, That logic, one of preplanned inclusion and exclusion, hospital can usefully be developed by making comparisons with warns the analyst away from comparing “non-comparable” the same situation in the home, in nursing homes, in ambu groups. To be included in the planned set, a group must have lances. and on the street after accidents. The similarities and “enough features in common” with the other groups. To be differences in these conditions can be used to explain the simi excluded, it must show a “fundamental difference” from the lar and diverse properties of interaction between nurse and others. These two niles represent an attempt to “hold constant” patient. strategic facts, or to disqualify groups where the facts either The principal point to keep clear is the purpose of the re cannot actually be held constant or would introduce more un search, so that rules of evidence will not hinder discovery of wanted differences. Thus in comparing variables (conceptual theory. However, these goals are usually not kept clear (a con ard factual) one hopes that, because of this set of “purified dition we are trying to correct) and so typically a sociologist groups,” spurious factors now will not influence the findings starts by applying these rules for selecting a purified set of and relationships and render them inaccurate, This effort of puri groups to achieve accurate evidence. He then becomes caught fication is made for a result impossible to achieve, since one up in the delights of generating theory, and so compares every’- never really knows what has and has not been held constant, thing comparable; but next he finds his theory development severely limited by lack of enough theoretically relevant data, 7. For exariple see Janowitz, op cit., Preface and Chapter 1; and Ed because he has used a preplanned set of groups for collecting scard A Shils, On the Comparative Study of New States’ in Clifford Geertz (Ed.), Old Societies and Ncw States (New York: Free Press of his information (see Chapter VI) In allowing freedom for Glencoe, 1963), pp 5, 9. comparing any groups, the criterion of theoretical relevance 52 THE DISCOVERY OF GROUNDED THEORY Theoretical Sampling 53 used for each comparison in systematically generating theory or local-cosmopolitan. However. onlv a handful of survey re controls data collection without hindering it. Control by this searchers have used their skill to create multiple comparison criterion assures that ample data will be collected and that the subgroups for discovering theory. This would be a very worth data collection makes 5 sense (otherwise collection is a waste of while endeavor (see Chapter VIII on quantitative data). t r e However aoplyi ig theoretical control over choice of TI’e tactic of creating groups is equally applicable for soci comparison groups is more difficult than simply collecting data ologists who work with qualitative data, When using only inter from a preplanned set of groups, since choice requires continu views, for instance, a researcher surely can study comparison 5ou thought, analysis and search. groups composed of respondents chosen in accordance with his The sociologist must also be clear on the basic types of emergent analytic framework. And historical documents, or other groups he wishes to compare in order to control their effect on library materials, lend themselves wonderfully to the compara generality of both scope of population arid conceptual level of tue method. Their use is perhaps even more efficient, since the his theory. The simplest comparisons are, of course, made among researcher is saved much time and trouble in his search for different groups of exactly the same sul’stantive type, for in comparison groups which arc, after all, already concentrated stance, federal bookkeeping departments. These comparisons ‘n the library (see Chapter VII). As in field work, the re lead to a substantive theors that is applicable to this one type searcher who uses library material can always select additional of group. Somewhat more general substantive theory is achieved comparison groups after his analytic framework is well de by comparing different types of groups; for example, different veloped. in order to give himself additional confidence in its kinds of federal departments in one federal agency. The scope credibility. He will also-—like the field worker who sometimes of thc theory is further increased by comparing different types tumbles upon comparison groups and then makes proper use of groups within different larger groups (different departments of them—occasionally profit from happy accidents that may ir different agencies). Generality is further increased by mak occur when he is browsing along library shelves, And, again ing these latter comparisons for dificre it regions of a nation like the researcher who carefully chooses natural groups, the or, to go further, different nations. The scope of a substantive sociologist who creates groups should do so carefully according theorii can be carefulit, increased and controlled by such con to the scales of generality that he desires to achieve, scions choices of groups. The sociologist may also find it con As the sociologist shifts the degree of conceptual generality venient to think of subgroups within larger groups. and of for which he aims, from discovering substantive to discovering jute nal and external groups, as he broadens his range of corn formal theory, he must keep in mind the class of the groups p son and attcrrpts t keep tractable Fis substantise theory’s he selects. For substantive theory, he can select, as the same substantive class, groups regardless of where he finds them. He he sociologi t deve o up substa tive or formal theor can may, thus, compare the “emergency ward” to all kinds of medi also isefuIiv crc’iitc groups provided he keep cn mind that cal wards in all kinds of hospitals, both in the United States they are an artifact 01 ins resec’rch design. and so does not and abroad. But he may also conceive of the emergency ward start assuming in his anal si that thc have properties pnswcsed as a subclass of a larger class of organizations, all designed to by a natucal group. Scuse’. researchers are ad’pt at creating render immediate assistance in the event of accidents or break gr U s md s ati tieafly o iding th ii relevarce (as by Ia toi’ 8. In fact, in b’ickstage discussions about which comparative groups to seil-ig r c i na e t u.s e crc they re, ‘reate and choose in survey analysis, the answer frequently is “Where the reaksin the distribution are convenient and save cases, and among these grou n t cat a .nir fu I Iei mc s cw n cc choose the ones that give the ‘best findings,.’” Selvin, however, has devel they have bcci created, exam e teacher igh. mcdi cm. oped a systematic method of subgroup comparison in survey research that The Effects and lw on “apprehension”. or option, middle. acid lower cias: prevents the opportunistic use of “the best finding” criteria. See of Leadership (Glencoe, Iii.: Free Press, 1960). 54 THE DISCOVERY OF GROUNDED ThEORY Theoretscal Sampling downs. For example, fire, crime, the automobile, and even plumb non-comparable comparison with the apparently comparable ing problems have all given rise to emergency organizations that ones. The non-comparable type of group comparison can greatly are on 24-hour alert. In taking this approach to choosing dissimi aid him in transcending substantive descriptions of time and lar, substantive comparative groups, the analyst must be clear place as he tries to achieve a general, formal theory.’ about his purpose. He may use groups of the more general class 1 to illuminate his substantive theory of, say, emergency wards. He Why Select Groups may wish to begin generating a formal theory of emergency or ganizations. He may desire a mixture of both: for instance, This concern with the selection of groups for comparison bringing out his substantive theory about emergency wards raises the question: Why does the researcher’s comparison of within a context of some formal categories about emergency groups make the content of the data more theoretically relevant organizations. than when he merely selects and compares data? The answer is On the other hand, when the sociologist’s purpose is to dis threefold. Comparison groups provide, as just noted, control 9cover formal theory, he will definitely select dissimilar, sub over the two scales of generality: first, conceptual level, and stantive groups from the larger class, while increasing his second, population scope. Third, comparison groups also pro theory s scope. And he will also find himself comparing groups vide simultaneous maximization or minimization of both the that seem to be non-comparable on the substantive level, but differences and the similarities of data that bear on the cate that on the formal level are conceptually comparable. Non- gories being studied, This control over similarities and differ comparable on the substantive level here implies a stronger ences ia vital for discovering categories, and for developing degree of apparent difference than does dissimilar. For example, and relating their theoretical properties, all necessary for the while fire departments and emergency wards are substantially further development of an emergent theory. By maximizing or dissimilar, their conceptual comparability is still readily appar minimizing differences among comparative groups, the sociolo ent. Since the basis of comparison between substantively non- gist can control the theoretical relevance of his data collection. comparable groups is not readily apparent, it must be explained Comparing as many’ differences and similarities in data as on a higher conceptual level. possible (as mentioned in Chapter II) tends to force the Thus one could start developing a formal theory of social analyst to generate categories, their properties and their inter isolation by comparing four apparently unconnected mono relations as he tries to understand his data (see Chapter V graphs: Blue Collar Marriage, The Taxi-Dance Hall, The also). Ghetto and The Hobo (Komarovsky, Cressey, Wirth, Ander Minimizing differences among comparison groups increases son) ,10 All deal with facets of “social isolation,” according to the possibility that the researcher will collect much similar data their authors. For another example, Goffman has compared on a given category while he spots important differences not apparently non-comparable groups when generating his formal caught in earlier data collection. Similarities in data that bear on theory of stigma. Thus, anyone who wishes to discover formal a category help verify its existenc by verifying the data be theory should be aware of the usefulness of comparisons made hind it. on high level conceptual categories among the seemingly non- The basic properties of a category also are brought out by comparable; he should actively seek this kind of comparison; similarities, and by a few important differences found when do it with flexibility; and be able to interchange the apparently minimizing group differences. It is helpful to establish these 9. Cf. Shils, op. cit., p. 17. properties before differences among groups are maximized, For 10. Respectively, Mirra Kornarovsky (New York: Random House, 1962); Paul Cressey (Chicago. University of Chicago Press, 1932); Louis Wirth 11. This statement is made in implicit opposition merely to “writing’ (Chicago: University of Chicago Press, 1962 edition); and Nels Anderson one’s theory in a general formal manner, on the basis of sheer conjecture (Chicago: Universih f Chicago Press, 1961 edition). or on the basis of one group, as is typical of journal articles. E 56 THE DISCOVERY OF GROUNDED THEORY Theeretical Sampling example. the basic property ot calculating the social loss of hassc work is accomplished howex er. he should turn to maxi— dying patients is their age, as was discovered by observation on ivuzing differences among co nparison grnups. in aeordanec geriatric and nursery wards, It was important to establish this wit the kind of t icon>’he v ishes to develop substantis e or for property before going on to establish other properties of social al and with the requirements of his emergent theory. When loss by studying dying on other kinds of nards.’ maximizing differences among comparative groups (thereby \ linimizing differences among comparison2 groups also helps niaxin ‘zing differences in data) he possesses a more powerful establish a definite set of conditions under which a category racan for stimulating the generation of theoretical properties exists, either to a particular degree or as a tvpe_which in turn once his basic framework has emerged.’ Maximizing brings out establishes a probability for theoretical prediction. For example, the 1estv:c possibie cox erage on ranges. continua, degrees. types. “open awareness contexts” about dying—where the patient and o idor nitics, variations, Cauws conditions, consequences. proba the staff are aware that he is dying—arc expectable whenever bihti s of relationships, strategies, process, structural mechan patients are held “captive’ in a government hospital (whether ‘aris and so forth all ec ‘ssary for elaboration of the theory national, state, or county). ‘Captive” patients may be convicts, As th sociologist maximizes differences by changing the eterans, or research patients.’ scope of his research—for example, by going to different organi The other approach, maximizing differences among compari zations, regions, cities or nations—he discovers more startling son groups, ii’’reases the probability that the researcher siii differences in data. His attempts to understand how these differ collect different and varied data bearing on a category, while ences fit in are likely to have important effects on both his yet finding strategic similarities among the groups. The similari research operations and the generality of scope of his theory. ties that occur, through many diverse kinds of groups, pro These differences from other organizations, regions, or nations vides, of course, the most genesal uniformities of scope within v ill make him wonder where he could have found the same dif his theory. As the analyst tries to understand the multitude of ferences at original research sites. And how can he continue differences, he tends to develop the properties of categories his theoretically focused research along this line when he speedily and densely and, in the end, to integrate them into a turns to home base) theory that possesses different levels of conceptual generality. At the same time the scope of his theory is broadened, not thereby delimiting the theory’s scope. The sociologist does not qualified. For example, one of us once noted that in Malavan merely look for negative cases bearing on a category (as do hospitals families work in caring for dying patients. This obser others who generate theory); he searches for maximum differ vation was interesting because tip to this point we had consid ences among comparative groups in order to compare them on ered the family member, in the United States, as either being the basis of as many relevant diversities and similarities in the treated as another patient (sedated, given rest) or just ignored data as he can find, as a nuisance. Review log our American data, though, w’e dis When beginning his generation of a substantive theory the covered that the family is used in several ways for the care sociologist establishes the basic categories and their properties of dying patients. We had failed to focus on this not-so by minimizing differences in comparative groups)’ Once this ohserx 0abl occurrence, Thus, we discos ered a cross-national uniformit —not a difference—by noting abroad what we had 12. See Barney C’ Glaser and Anselrn L. Strauss, “The Social Loss of missed in America, We then proceeded to study it at our home Dying Patic r,t, .tmerican Journal of Nursing, ol 64. \o 6 (June, 1964 li 13 See Clasi r ‘mci Strauss 4warene.ss of Dying op. cit., Chapter 6. base, where we had more time for the inquiry. We had similar 14 Good sul,stantivc thcors c in result fioin tlsc studs of one group, if expeiiences when comparing hospitals in various regions of the the analyst carelulls’ sorts data into comparative subgroups For example, United States with those closer to home, in San Francisco. see Evans-Pritchard, Witchcraft, Oracles and Magic Amanc’ the Azande Oxford, England: Clar”ndon Previ, 1937 .and our diseu’:sion cf this hock in Chapter \‘I. 15. Shils, op. czt., p. 25.
remember
must to
data,
events active
group, groups
do
Differences
select
Minimized
58
maximizing
Maximized CHART
toii,
(sew
in
Part
Groups
see
How
16,
Chart
not
Leonard
York:
not
them
See
is
sampler
1
is—or
where
with for
To
give
of
the
CONSEQUENCES
Merton’s
1
IN
an
that
Basic
Select
theoretically
the
presents
groups
Broom,
systematically
COMPARISON
or
problem
him
ethnogapher
is
the
he
of
Books
without
sociologist’s
not—likely
Groups
discussion
theoretical
next
basic
is scope uniformities gory, Lying
conditions
Maximum
Spotting
data
data, for
of
ditions
(3)
and
in
an
the
categorY.
prediction.
1959,
of
Establishing
oc
generating
leads
properties;
usefulness
theoretical
Leonard
(2)
relevant
active GRoUPS
he
a
basic
can
how
MiNIMIZING
fundamental Similar
of
trying
preplanned
calculate
Generating
similarity
for
must
to:
to
p
be
strategic
decision
of
These
relevance,
to
sampler
xxvi.
consequences
take
ro
used
(1)
S.
a
THE
greatest
data of
and
set
Data
go
to
continually
degree
theory.
Cottrell
question
con
Veri
GENERA’IING
cate
DISCOVERY
in
research
place.’ 6
of
AND
get
about
where
collection.
on
about
of
research
he
Category
the MAXIMIZING
(Eds.l,
theoretically
gories
theory.
property
Dense data Maximum vary
differences
category
(3)
(2) Spotting
choosing
sites
will
must
a
of
If
OF
fullest
which
analyze
Delimiting
THEORY
Integrating
quickly
given
ongoing minimizing
GROUNDED
in
and
developing
design. First,
take
Sociology
Diverse
of
fundamental
be
Robert
and
diversity
properties;
under
DIFFERENCES
data
categories;
groups
particular
forces:
prepared
order
the
him.
hypotheses
he
relevant
scope
events of
K
As
THEORY
on
Today must
which
of
data
Mer
cate
and
in
He
an
(1)
of
to
of
a
conceptions
comparative
these
to
the
objective
of
another
ilso
sshich, r
Di,ing
ssill to
Theoretical
comparative
son
widely
comparisons
America
course
which weeks
times
ceptual
expectedness ceptual
scheduling
And
the
17.
and
were
I dying
great
sc
observe
problem
quick,
expectancy
Dalton awareness
comatose)
tl’en lows:
manipulate
wished
general The
‘r
The
is
‘Once
happen.
above
Visits
rapid,
the looking
typically
we
her
in
research
at
in
needed but
describes
to
of
contrasting
tended
I
s
so
and
following
structure
emphasis
in
analysis
scheme—which
on
1 iscusiori
wished
can connection returned
a
turn,
country,
further
comments continuous
then
Sampling
I
the
asks
where
analysis
to
theoretically
types
neurosu
(and
of
observed
Methods
a
and
is
of
I
and
at
be
the
service
theoretical
checking
requires
wished
increased
types
to
staff
to events
directs
himself
memo
some
of
investigated
questions,
so
so
first
including how
the
look
of
various
be
rate
is
not
and
countries.
to
of
memo
industrial
of and
gial
first
looked
in
service
upon
services
of
on
using slow.
observation,
patient’s
where
different
with
to
the
next
its
a
of
Merton by at
the
Men
which
relevant
or
him
often
search an
service
the
gap
looked
x
specific
dying—as
look
conditions
included
service
medical
cry
type
had
words
the
each
from
extending
matters So
Intensive
to
active
at
Who
cross-national
That
staff
to
to
is at
organizations, next
after
of
on
greatly
look
might
at
“next
been
types
I
for
shows
identified,
an
of
good
other seek
ways. was
with
at
ctrategies.
patients
looked
Manage.”
developing
where services our
expectations
or
hospital.
is,
in
services
empirical thcoietically
the
emergency
search
a
hypotheses
not
at
well
question where
missed
of
Care a”tions
directed
the or
premature
order
advantage.”
particular
its
by
the
types
The dying
how
initial research
services,
at
might
next
patients
as
distinctive
was
it
major
choosing
first
that
in
(Ibid..
Unit.
comparisons
were
in
probability
death
for
The
leads
in to
by
materials
comparisons,
groups
theory
technique”
of
two
in
Melville
Hammond, great
at in
the
by
of
check not
service.
envisioned.
a
minimized
data
unit
about hospitals.
other
a
we
baby
a
Then
scheduled
relevant
order
for
groups
pp.
almost
developing
teiniinality
\Vc or
was
verc
initial
a
situation
be,
cancer
by
different
and
rationale,”
also
three
general
are
of
xxiii-xxiv.)
through
upon Awareness
occurs
Dalton,
have
major
unexpected
service
I
awareness,
linking
of
and
While
to
frequently
compari
to
made dying
as
wished
observed
period. 17
on.
selected:
for
fruitful
guide
or
service
So
question,
see
a
Some
here
patient
detailed
where
items
as
where
unit
cit.
and four
matter
con
were
which con
study:
as
“Pre
them
our
was
“The
and
we
in
fol
what
See
to
his
the
is
of
59
H
11
II, ii
s
o
a
should
tcglc
Degree
60
tingle
g
unexpected; where
would
patients slow, and services by serve hospital
dominantly structural observe gr regions
studied will Acian
s
and of
of
America.
theore
sards)
ry
elati he
ount
p1mm
atier
epends
tuations
Vhcr
hosp
comparison,
ups
‘ 1 uc’sfoiu
the
The
to
usual
this
ones
be
mm
‘ely
and
ts overseas
of
y, countries
‘OUl
tend
two
right
conceptual
of
tal
in
where
isi
selection
tend
s’ho
may
therefo
hospitals
which
at or
means
I
choose
which
rapid.
I
where
wards
American
the
the
heoretical
oosine
l
conditions
service
expected
are
to how
locales
hospitals,
satura
will
wards
within
to
are have
co
countries.
United
he
will
all
as
rn°rig
different
will
that
man’
e,
be
I
where
he
I
degrcr
they
observe,
of
of
man’,
we
for
would
framework
four
I
groups
To
since
hope
where
be
where differing
the
of
on.
dies
hospitals
by
conditions
high
I
maximize hail
have
States:
Sampling
hospitals
do we
Chinese
‘,vhi
to
guided,
will
low
conditions
same
predominant
staff
patients
But
ss
‘s
a
to
of
attempt
and see,
patients
begin
as not:
id
where
seen
fhmr ft
c
hase
at
have
social
onditions
observe
sanmpl1n,
n
and
degree
services;
well
namely, under
city.
ar
developed
I
one.
‘nsues
as
wards
of
the
thm some and
TH
know
in
d
consict
of
also
with, to
others
to
t 1 utjO noted
alm’.
possible,
in
services
d
tend
as
are
our
cities,
high
I
visit what DXSCOVERY
i
on
hospitals
the
should
mode
a
shall
iaximize
where
by
wards
of re
t
where
for
and,
own
maximally
asise: mmd
where
cat
to where
of
and
various
social
as
hospitals
to
are curient
the
even
circumstances
instance,
social
such
know
at
only
also
of
we 11
hospitals,
in
as
date.
relevance,
where
the
ea’ires
three
d
me
which
dying
specific different
i-io’
th dying
OF
dying
for
‘pth
m
in
hospitals
.wlue.
[four
choose
that
one
terminal
of
value
they
as terminal
GRO1’NOED
colleut Asia,
I
k
different
Malaya,
in
mixed
are
dying
that
rids
‘,vill
those
of
many where
tends
I
large
tends
is
very
contrasting
important]
is
are would
conditions
chc’ussions
ft
relatively
wards
different
the
of
t md
the
than
want
in
In
e
patient
(or
data
is
dying,
ethnic
wards
study,
to
ward.
dying
o many
groups
to
there
there
retic
same some from
THEOCY
each
type
pre
svill
ob
stra
be
be
on
or
to
in
on
categories,
not
gap ing
When
during
possible confident
analysis
to can
sible,
that
sample
gory
tanees ologist
data.) are
ing must
groups.
and
Thus rrturning
usually
group
with
collected
groups
will
annot
vhose Theoretical
hen
look
One
Even
Theoretical
As
completely
in remains
distributed
develop
no
to
sample
a
is
just
generating
state
to
saturation
he
his
analysis
study
may
we
over
multiplicits
at
must
for
additional rules
tate
for
the
reaches
of
range
is
stop (This
during
that
continually
to
theory,
from
the to
Sampling
and
data.
have
the
engaged
groups
be
each
category’s
and make
during
hut
involves
properties
at
them,
sampling
continually of Saturation
end.
a
of
of number
throughout
situation
collected
attempt
filled,
the
a
category
research
theory
occurs,
theoretical
over
evidence
said,
(See
to
data
especially
data.)
theoietical
data
certain
that single
Since
the
of
outset
while
go
in
verification
is
again,
In
Chapter
the
on
theoretical
are
the situations
of
of
he
stretch
entire
collecting
to
on
the
dealing
When
contrasts
for
data
focused
that
is
trying
judge
the
group—although
groups
the
governing
various
of
would sociologist
simultaneously
being
saturate different
in
to
saturation
analyst
saturated,
the
only
point.
his
category.
his
saturation
for
study;
category.
new
V
diversity
one
researcher
how
with
or
to
research
found
for find
within
major
on that
various
saturation. one
categories,
with
data
dew
category
will
reach these
groups
groups
The
theory,
he
a
the
trying
many
a
it
svill
He category—the
by
discussion
ription
w’hereby
categories,
that
from
usually
bard
multiplicity
As
of
is
can
each;
criterion
collection
categories
new
how
seeking
saturation
goes
joint
becomes
based
be
data
pertinent
data
he
for
is
however
groups
to
only
of
and
Saturation
to
older
sampled.
saturated,
many
categories
in
sees
while
out
discover
find
data
collection
count
the
as the
on
from
he,
which
of
count
ness
is
for
far
empirically
of
the
are
similar
of
he groups.
that
sociologist
of
saturation
researcher
sociologist
to
almost,
he
groups
therefore,
on
absorbed
the
his
all
a
judging
as
reliable
accord
groups.
a groups.
widest
usually
should
means
people up
maxi
theory
noth
other
some
given
also, cate
these
soci
pos
way
and
in
the
he
if
or 61 62 THE DISCOVERY OF GROUNDED THEORY Theoretical Sampling 63 mizcs differences in h s groups in order to maximize the sane that we should look for (as re ‘earLhers and reader of rewarch) ties of data bearing on a category, and thereby develops as is very different, many diverse properties of the category as possible. The criteria The adequate theoretical sample is judged on the basis of for determining saturation, then, are a combination of the how widely and diversely the analyst chose his groups for satu pirical limits of the data, the integration and density of the rating categories according to the type of theory he wished fhec, d the aaIyt’s theoretica sensiti”ity. to develop The adequate ctatictic’al sample, on th other hand, Saturation can never be1 attained by studying one incident is judged on the basis of techniques of random and stratified in one group. What is gained by studying one group is at most sampling used in relation to the social structure of a group the discovery of some basic categories and a few of their prop or groups sampled. The inadequate theoretical sample is easily erties. From the study of similar groups (or subgroups within spotted. since the theory associated with it is usually thin and the first group), a few more categories and their properties are not well integrated. and has too many obvious unexplained yielded. But this is only the beginning of a theory. Then the exceptions. The inadequate statistical sample is often more diffi sociologist should try to saturate his categories by maximizing cult to spoh usually it must be pointed out by specialists in differences among groups. In the process, he generates his methodology, since other researchers tend to accept technical theory. For example, from studying one incident in one group sophistication uncritically. ‘si’ rnlghf diseoser that an important property of Pursing stu The researcher who generates theory need not combine dents’ perspectives about course work is their assessment of the sandom sampling with theoretical sampling when setting forth differential importance of certain kinds of course work to the relationships among categories and properties. These relation faculty; but this discovery tells us almost nothing. To find ships are suggested as hypotheses pertinent to direction of re out such properties as when and how an assessment is made and lationship, not tested as descriptions of both direction and shared, who is aware of given assessments, and with what conse magnitude. Conventional theorizing claims generality of scope; (luences for the students, the faculty, the school, and the pa that is, one assumes that if the relationship holds for one group tients whom the students nurse, dozens and dozens of situations under certain conditions, it will probably hold for other groups in many diverse groups must be observed and analyzed under the same conditions.19 This assumption of persistence is comparatively. I subject only to being disproven—not proven—when other soci ologists question its credibility. Only a reversal or disappearance of the relationship will be considered by sociologists as an im Theoreticaland StatisticalSampling portant discovery, not the rediscovery of the same relationship It is important to contrast theoretical sampling based on in another group; since once discovered, the relationship is as the saturation of categories with statistical (random) sampling. sumed to persist. Persistence helps to generalize scope but is rheir differences should he kept clearly in mind for both de usually considered uninteresting, since it requires no modifica signing research and judging its credibility. Theoretical sam tion of the theory. pling is done in order to discover categories and their properties, Furthermore, once discovered the relationship is assumed to and to suggest the interrelationships into a theory. Statistical persist in direction no matter how biased the previous sample sampling is done to obtain accurate evidence on distributions of data was, or the next sample is. Only if the hypothesis is dis of people among categories to be used in descriptions or veri proven do biases in the sample come under question. For gen fications. Thus, in each type of research the “adequate sample” erating theory’ these biases are treated as conditions changing the relationship, which should be woven into the analysis as 18. Fred Davis and irginia Olesen, “Problems and Issues in Collegiate Nursing Education” in Fred Davis (Ed.>. The Nursing Profession (New 19. See discussion on this in Hans L. Zetterberg, On Theory and York: John Wiley and Sons, 1966). pp. 138-75. Verication in Sociology (Totowa, N.J.: Bedminster Press, 1963), pp. 52-56. 61 TifF DISCO’SERY OF GROUNDF.O I SEORY fheretica1 So op rg such Thus, random sampling js not necessary for theoretical 15i the -r f0 bcrr’i’op. ‘u must eflflnfl(’ cause the sampling, either to discoser the’ relationship or check out its rules of aceurato cv:denee require the fulit 5 doserage to achieve existence in other groups, However, when the sociologist the snort acc’usate C000t. If the researchet wishes to cliserge wishes also to describe°2 the magnitude of relationship within a design arpartc”i group, random ramp!ing. or a highly ‘stematic 1observatson orocedure done over a specified time is necessary. Oiit\ ante’ or labenioeslv integrat his ness approach into the For example. after we discovered the positive relationship be ieseaj’c’hdesig to allow a new prepl’snned attack on tInt total tween the attention that nurses gave dying patients and the , blem. He must not deniate from this ness dcsign cither; nurses’ perceptions of a patient’s social loss, we continually c stually it leads hint back into the ‘a ne “bind, ‘ 21 found this relationship throughout our research and were quick to note conditions altering its direction, But we could never state the precise magnitude of this relationship on, sax, cancer Slice of Data waids, ,ince our sampling snas theoretical In theoreti al sompling no one kind ot data on a category Another important difference between theoretical and statis is’ ‘ technique fer data coilecti,,,r is necessarili appropriate. sampling tical is that the sociologist must learn when to stop Different kinds of data give the analyst different views or using the former. Learning skill takes this time. analysis and vantage’ points from which t understand a r’ategory and to flexibility, since making the theoretically sensitive judgment develop its propertIes; these different view’s we have called about saturaton is never precise. The researcher’s judgment slices of data. W’hile the’ sociologist max’ use one technique becomes confidently clear only toward the close of his joint of data collection primarily, theoretical sampling for saturation collection and analysis, when considerable saturation of cate of a category allows a multi-faceted investigation, in which gories in many groups to the limits of his data has occurred, there are no limits to the techniques of data collection, the way so that his theory is approaching stable integration and dense they are used, or the ts pes of data 22acquired. One reason for development of properties. this openness of inquiry is that svhen obtaining data on differ By contrast, in statistical sampling the sociologist must con cnt groups, the sociologist svorks tinder the diverse structural inue with data collection no matter how much saturation he conditions of each group: schedules, restricted areas svork tern pcrceives. his Jo case, the notion of saturation is irrelevant pos, the different perspectives of people in different positions. to the study. Even though he becomes aware of what his find and the availability of documents of different kinds, Clearly, to ings will be. and knows he is collecting the same thing over and 20. We has’,’ takei, a position in direct opposition to TJclv, who says: 21. Fnr example. 5Uiy says, “The coding operation proved to be very i “Any ecarch of any type whatsoever which seeks to make generalizations tedious ‘dr,g w’ork’ in the worst sense of the terms, I . , was nosy attempt— y. nd tie material studied ins oives problems of sampling.... [The mu to resist, rather than encourage flights of imagination. I had to accept i (“‘ar’fl”r] is implicitly identifying a larger population, of which his eases the fact that there 55 crc caps in the data about ss’hich I could do nothing” nurncr t he i ii flrcsi’ntativc sample. and contending that certain rela— op. cit., pp. 178-79 0 To as oid this binrl. many sociologists lure data 5tionshms sire rvr’d in his sample could not have occurred there by chance. collectors and coders in prr’planned research for description and erification. it is 0mph ‘set tine that one cia ‘ixold ‘implirg problems by proceeding Then, however, d’scovr’ric’. are made too latr to cifeet changcs in data in words i stead at numbers or bs avoiding tht use of statistical techniques, nilection. Scc thc tug—of—warwagc d between Riesman and Watson on this though it is uiitisrtiuiately tine that by avoiding such methods onc can bind: Riesman continualls wanted to break out and Watson wanted to often ket p u I ipli epor blr’ms from becoming explicit ‘ Udy’s gross ate Ir’untain tight control, David Rie, man and Jeanne Watson, “The Sociabil gone ii position could a odiflerl to c osapatibility with ours. x’, e believe, ity Project: Chroniek if Frustration and Achievement,” in Hammond, if he thought rather tC in. of diverse purposes of sese’ircls and the degree op cit., pp. 269-84 to v hich cad purp.’ rcquir( s a rd itionship to be dc ‘crihed in terms of its 22. For examples of niriltifaceterl insestigitions, see in Hammond, op. various properties.5 cristroce, direction, magnitude, n’it’ire. and conditions, cat.: the research chronicles of Renee Fox, c “An American Sociologist in the etc In any event, a sv lint s later he then admits that “one cannot really Land of Belgian Resesreh’; Dalton; and Seymour NI. Lipset “The Biog solse them” (problems of representativeness). Ldy, op. dC, PP. 169-170. naph of a Research l’roiect: Union Democracy,”
strained
tion obtain
guess
useful
than
secret
tise treme
democracy composition that
pp. when
survey
testing the
one
technique
understanding
diverse
he
slices
on
ferent
conditions ologist,
between collection). on
to
variety
evidence
be succced
coLlecting
66
26.
24,
25.
23.
Most
question ethical generate
Among
desired? methodological
107-119.
The
categories
is
remembered
bewildering
field
conditions
he
only
Only
its
Dalton,
but
of
Thus,
Compare
Lipset
example, the
of
personnel
ways
techniques
is
providing
compared
result
the
by
he
often,
data
problems
composition
coming
data
was
the
for
best
that
one
highly
arise: under
The
data
the
any
under
must
International
said
This
properties
such
of
op.
of
is
generated.
verifications,
various
the
kind—and
is,
best
answer to
many
than
should
discussion
of
however,
Dalton
which usually
to
an
permit
particular
from
he
cit., knowing
records
if
this
of
be
obtain? of
flexibility makes
beneficial,
that
structural
terms
motivation
the
which
differences
wished
he
can
Riesman
executive
we
course,
pp.
flexible
any
slices
is
new
method
from
group
slices
has be
meaningles.
as
See
technical,
had
category
Typographers’
with
this
obtain
its
66
when
about
wished
the
conditions
The
the so one
slice
to
about
he
in based
they
and used.
Seymour
of
However,
and
use
a
names, 26 to
differences.
in
test
ethics
would
because
that
conditions
comparative
of
research
to
tries
whether
theory
sociologist’s variety
data
answer
to
hierarchy, mode
of
the
and
bribe
67
his
Watson,
not
in
data,
group
his were
THE
a
to
keep
data on
under
he
of
of
Often
information
give
to
methods
that
category
M.
doctrinaire.
some on
theory
is
evaluate
Union,
a Dalton,
DISCOVERY
the
a
of
survey
of
could
to
understand
it
is,
for
Thereby
the in
group
Lipset
very
23
standard him
op.
collectd, 24
the
secretary
may
yields slices
as
the
the
knowing
diverse
of
researcher’s
terms
rather
manner. 25
of
generating
cit.,
objective,
best
strategy
analysis
What
course,
data who
at
op.
exciting
formed
and
which
researcher
find
union
in
be
virtually
more
pp.
of
them OF
data
his
desired,
more
cit.,
Hammond,
are
collected,
problems
of
in
actually
data
in conditions,
than
is out
GROUNDED
260-69.
the
allows
democracy
(technique
task.
theory
on
the
theory
pp.
both
better his
will
But
the
available order
of
For
to
as
information
the
the
theory
is
differences
theoretical
that
force
59-62,
collection
trying
means
provided
both
the
different
kinds
The
different
forced
happened
be
accurate
it
on
or
informa
was
an
op.
ethnic
to
which of
THEORY
would
worse
union
must
con
soci
by
does
with
him
see
dif
not
this
cit.,
ex
the
not
are
for
to
to
to
of
a
they
situation
Some may
hounding
talking
same
ent
may relevancies,
parisons
else
veloping
comparisons
of
data
comparison.” must
mformation important
or of
a
on
on tantively
nsent
operating.
the
en ber
erates
(done onc
of
be
sary
and
Theoretical
data
national
As
pain
reading,
Another
data
illness
a
the
documents
observed,
sociologist
has
be ,
tend
is
at
offer
on
social-structural
subject,
be
relevant
sociologists
from
he
everyone
properties,
for
to
dying
what
life
can
continually
in
he
is
theoretically
other
he
core
to
relativism
to
and
will
different
in
trivial
as
Possibly
Sampling
hospitals.
unbearable
the
styles
slice
is
be any
learned and
market-research
to
start
are
claim
times,
themselves;
Through
and
“the
categories.
talked
(patient,
category.
pain
able
groups
observes
on
soon
meat-packing
trusted
a
knows,
information
the
especially
of
see
researcher
they
of
and
data
to
the
facts”
that
his
people.
changing
by
He
to
of
data
about
professionals know
with
stories
information.
these relevant,
develop
same
that
most
his
to
vary
facts—no
obtain,
theory
a
if
can
should
nurse,
or
only
For
different
nurse
The
that only
the
those
very
own
talks
overheard,
the
offer
what
useful
circumstances
Furthermore,
help,
that
subject
any
of
example,
who
their that
sun’ev
as experience
industry).
researcher
will
should
the
doctor,
realize
Similarly,
comparing
it
different
or
category
who experiences,
others,
with
the
information
slice
useful
data
type
if
and
For
people patient
by
is
category
happens
receive
in
method
can
it
group
wish
studying
in
about
starting be
examining,
can
that
is
example,
interviewed
chaplain
one
yields
sensitize
the
of
comparisons.
a
be
The
accurate.
information
was
can a
and
used
the
may
in
to
situation
as
yield
considerable
their
awareness
no
down-to-earth
can
changes,
can
sociologist
quite
which
his
general
meat
considerably
verify
different
presenting
ask
“lived.” useful
make
data
matter
information
research
is
the
yield
gain
no
give
way;
or
the
differences
himself
the
himself
for
contradictory.
Since
they
matter
managemem
family
need
consumption
where
or
or
same
quick
information
knowledge,
useful
and
very
the
this
what
“anecdotal
about
Anecdotal
This
even context
describe,
positions can surveyed
and develop
indicate
an
such
“accu not
differ
where
to article
group,
useful
neces
when
whom
some
itself mem
com
slices
kind
gain
data
sub
gen
nfl
the
de
its
be
a
is 67 DISCOVERY OF GROUNDED THEORY Theoretical Sampling 69 rate evidence, Other meth ds ti hcy might use only yield done different slices of data are seen as tests of each other, biased or impressionistic data, and so can be discounted 27 not as different modes of knowing that must be explained and Ifeing this argument, they take onh one slice or mode of know integrated theoretically. The result is that, without comparative ing as giving the “facts.” Since they do not seek other modes, analysis, even men who generate theory tend to use and fall they remain untroubled For example. iii em’ noted study into the rhetoric of verification.29 They miss out on the rich of adolescents in high schools. Only the adolescents were sur disersitv of modes of knowing about their categories. And they veyed; and in a study of workers n a factory. only workers were fail to tell their readers of their other data, since they believe, observed and 25interviesxed, quite ssrongly, that it disproves their theory, when it would But when different slices of data are submitted to com base actually enriched it immensely. parative analysis, the result is not unbouncling relativism, In stead, it is a proportioned view of the evidence, since, during Depth of TheoreticalSampling comparison, biases of particular people and methods tend to reconcile themselves as thu analyst discovers the underlying The depth of theoretical sampling refurs to the amount of causes of variation This con inual correction of data by com data collected on a group and on a category. In studies of parat’ve analysis gives thu sociologist confidence in the data verification and description it is typical°3to collect as much data upon which he is basing his theory, at the same time forcing as possible on thu “whole” group Theoretical sampling, though, him to generate the properties of his categories. The continual does not require the fullest possible coverage on the whole correction of data also makes the sociologist realize clearly an group except at the very beginning of research, when the important point: when used elsewhere, theory generated from main categories are emerging—and these tend to emerge very just one kind of data never fits, or works as well, as theory 31fast. Theoretical sampling requires only collecting data on generated from diverse slices of data on the same category. categories, for the generation of properties and hypotheses. The theory based on diverse data has taken into consideration Even this kind of selective collection of data, hosvever, tends more aspects of the substantive or formal area, and therefore to result in much excess data, from which new and related can cope with more diversity in conditions and exceptions to categories emerge. For example, after a full day in the field, hypotheses. when the field svorker is tired and jammed with dozens of inch If the sociologist has two slices of data (such as field and dents to report in his field notes, he need only dictate data about survey data), but does not engage in comparative analysis, he his categories. Going through his categories also helps him to will generate his theory from one mode of collection and ignore remember data he may have forgotten during his full day. the other completely when it disproves his theory—although he With these categories firmly in mind, directing his attention, the may selectively use confirmatory pieces of the other data as field worker can focus on remembering the details of his day’s supporting evidence Thus, when no comparative analysis is obsersations with the confidence that the notes will be implicitly
27 For example, “The significance of the quantitative case study, then, 29. These same sociologists tend to be debunkers who try to dig up is (1) that it stimulates the kind of theoretical insights that can be derived something out of their own reading to disprove the theory presented by only from quantitative analysis as well as the kind that results from close their colleague. They do not understand they are merely offering a new observation of an empirical situation, and (2) that it provides more severe slice of data that under comparative analysis would enrich his theory by checks on these insights than an impressionistic study and thus somewhat providing or modifying properties and categories increases the probably validity of conclusions,” Peter Blau, “rhe Research 30. 5ee the instructive discussion on “depth” by Udy, op cit, pp. Process in the Study of the Dynamics of Bureaucracy,” in Hammond, op. 164-65. cit,p 20 31. For examples on the quick emergence of relevant categories see, 28. Coleman, op. cit., and see, for the study of workers, Donald Roy, Blanche (,eer, “First Days in the Field,” in Hammond, op. cit; and Blau, “Efficiency and the Fix: Informal Intergroup Relations in a Piecework Ma op. cit., pp 33-34. Blau discovered the significance of the “consultation” chine Shop,” American Journal of Sociology, 60 11954), pp 255-266. p, ttern with’n s sveek after starting his field research, 70 THE DISCOVERY OF GROUYOED THEORY Theoretical Sampling 71 guided by his categories. Any additional information he decides has been collected on a previous one. The sociologist should to note afterwards is “gravy” for theoretical consideration, not continue to saturate all categories until it is clear which are a required chore for the fullest coverage. Theoretical sampling, core categories. If he does not, he risks ending up with a vast thereforc can save much time in note-taking. array of loosely integrated categories, none deeply developed. It is not too difficult to compare as many as forty groups This results in a thin, unvalenced theory. Since stable integra On the basis of a defined set of categories and hypotheses tion of the tneory requires dense property development of at (not on the basis of the “whole” group). and when groups within least some core categories, it then becomes difficult to say groups are compared (e.g.. different and similar wards within which of the array are the core categories; that is, those most different types of hospitals). These groups can be studied one relevant for prediction and explanation. at a time, or a number can be studied simultaneously. They can also be studied in quick succession, to check out major hypotheses before too much theory is built around them, With Temporal Aspects of Theoretical Sampling out theoretical sampling, the field worker, or the writer of a survey questionnaire, collects as much data as he can and hopes When generating theory through joint theoretical collection, that this full coverage will “catch enough” that later will prove coding and analysis of data, the temporal aspects of the re relevant. Probably, though, it will prove too thin a basis for search arc different from those characteristic of research where a developed 32theory. Theoretical sampling reduces the mass of separate periods of work are designated for each aspect of the data that otherwise would be collected on any single group. research. In the latter case, only brief or minor efforts, if any, Indeed, without theoretical sampling for categories one could are directed toward coding and analysis while data are collected. not sample multiple groups; he would he too bogged down Research aimed at discovering theory, however, requires that all trying to Cover just one. three procedures go on simultaneously td the fullest extent The depth to which a category should be sampled is another possible; for this, as we have said, is the underlying operation matter. The general idea is that the sociologist should sample when generating theory. Indeed, it is impossible to engage in a category until confident of its saturation, but there are quali theoretical sampling without coding and analyzing at the same fications. All categories are obviously not equally relevant, and time so the depth of inquiry into each one should not be the 33same. Theoretical sampling can be done with previously collected Core theoretical categories, those with the most explanatory research data, as in secondary analysis, but this effort requires power, should be saturated as completely as possible. Efforts a large mass of data to draw on in order to develop a theory to saturate less relevant categories should not be made at the )f some density of categories and properties. The sociologist cost of resources necessary for saturating the core categories. cngages in theoretical sampling of the previously collected data, As his theory develops and becomes integrated, the sociologist sshich amounts to collecting data from collected data. Also, learns which categories require the most and least complete he is bound to think of ways to make quick, brief data-collec saturation, and which ones can be dropped. Thus, the theory tion forays into other groups, to find additional relevant com generates its own selectivity for its direction and depth of parative data. Therefore, in the end, theoretical sampling and development. data collection for discovering theory become simultaneous, In actual practice, even the saturation of core categories whether the sociologist uses collected data or collects his own can be a problem In field work especially, the tendency always data, or both. How much time and money are available is im is to begin collecting data for another category before enough portant in deciding to what degree the data to be sampled will have been collected previously by the researcher or anyone else 32 For example see Riesman and V,atson, cit., op. p 295. who compiles data. 33. See Sbus, op. cit., p. 17. 72 THE DISCOVERY OF GROUNDED THEORY Iheor(4i(1l Sampling 73 All studies require respites from data collection for the relief pickIng up loose ends. To pace the alternating tempo of these and health of their personnel. Generating theory by joint collec three operations. the sociologist coon learns that anahsis can tion coding and analysis requires such respites for additional, be usefully accomplished at various times: immediately after obvious reasons. The sociologist must engage continually in leaving the field; during the evening between successive days some systematic. coding (u-ually just jotting c’itegories and of data collection and during two or thre d or weekl properties on the margins of his field notes or other recorded resp tes from data collection However, the sstematic formu data) and analytic memo writing (see Chapter V). He must be latio i of the core structure of his theory r ia take considerable looking for emeigent categories, reformulating them as their ti though it need not. In either event, the ociologast should properties emerge, selectively pruning his list of categories he v y flexible about timing his work, lie should not be afraid while adding to the list as the core of his theory emerges, along to take. literaih, months off his data collection, if necessary with deeloping his hypotheses and integrating his theory—in and if possible. to thmk through his emergent theory before order to guide his theoretical sampling at each step of the way returning to tlìe field. if he does not take respites for reflection and analysis, he can The continual irwermeshing of data collectLn and analysu not avoid collecting a large mass of data of dubious theoretical beai c directly on how the data collection is brought to a close, relevance, k rsearcher can iways try to collect more iota for checking Most generating of theory should be done in uninterrupted hynotheses or fo g e atirig new properties ategories a quiet, away from the field or the machine room. This is true h potheses When vriting is done in or near the field, the especially during earlier stages of the project, when more time temptatior to go back is especially strong. These final searches is needed for careful formulation. At later stages, the sociologist toi data tend to be for either specific confirmation (the re will find that analysis can proceed more easily during moments ‘earcher mo’.ing now with considerable sureness and speed or of data collection. When his categories are firmer in integration elaboration (the researcher wishing to round out his work by and development, he usually can spot what he is doing in exploring some area that was previously untouched or even . theoretical terms w’hile collecting data. At this time, he may unconsidered ) Ti;ev can be strongly tempting if personal observe in a few minutes all that he needs to know about a elations formed in the field are atisfving or if exciting new group with reference to a given theoretical point. However, cvents are developirg here. Howevrr collection of additioral actually generating theory at the moment of collecting data is da a can be a va te of time fo ategerd’ airead s-itrr ed never easy; usually it takes reflection afterward to discover what o f catEgories t of core value o the theor Sometime, one has actually found. In addition, if one has colleagues on th s t nden y to ait n ti’ fe d just ‘ ‘we something the same project, they all must have respites from data collec new si d hopper t often it not— n I he stud tion to discuss what they are doing and should do next. Such p”o]o ed ulinacessati’ , This tenderic mao lse related to tc discussion is difficult or impossible in the field because they are esearohei ‘s anxiety to ‘know t’ver,’tbing,” whsch i no necec either scattered in different places or cannot talk freely in other for theorotical wturatinij. people’s presence. 34 \, Strou I Ocr’117’100. f B. Buoii’r, D fI.’1 o.. e”3 ‘1 — The sociologist eventually learns to pace the alternating r hiot” !Jc. j’ lIla I,utlt’O’lns ‘sec, 1.,-Yu Froc’ 31111 ‘‘1(001 tempo of his collecting, coding and analyzing in order to get 193i,, (1 flJ 000 .1110 e ktr B. C’ F Flughc 3 S S i nc z each task done in o ‘3hiu,,c’ ‘ of C o Prt cc 0 2 appropriate measure, in accordance with the r stage of his research and theory development. At the beginning, S in F1gF1 is ti (1 T ill t 11(1 1 (11 1 there is more collection than coding and analysis; the balance 10 1 C -isol n I selve k ard 1’. ‘ 1 ‘10110, then gradually C month r F’too, i [ie”l em changes until near the end when the research ‘a pc. ii has -izais ‘ 0 ‘shin, involves ‘rk ““tfl i It lyE’ lllt,1, p mostly analvsis with brief collection and coding for ‘C ., 30 ,,0 11 0 It. because 74
sociologist theory
1cm, involve but project? amount
and theory
time what not substantive groups. geographical of
how In generality’ countries)
necessary can ter least retical of numbers in amount
they on researcher ever, sampling surely
The
types Because
field situations
Detailed it
a
state
the
For VII). talk
description,
in
his
many
degree
for
depend
research
is
after
which
tempo
long
of while analysis presenting it will
This
Specifying
temporal
of and project
example, of necessary
difficult
the
and beforehand
is
time
for and
occur From
who will the
large or time
th
can
affect
describing periods largely
is he
survey to
breakdowns
data may
relatively completion
on positions areas of
formal,
to
different
a proposal necessary
be sociologist
be
and will will
which necessary
intends
sometimes
(what maximize the question these
units to the
open-endedness
when
the proposals
to
come
collection, the sampled. observed
contingent
research, answer of
where take study
writing
research
achieve directions
how
that
detailed
kinds
of
(such gestation descriptions, caches the require
shifts,
easy
and
designed
quickly
of
for to
people that
of But
who
for
differences he
each many
kinds
his
describe for for he in In
how
can
of data for generate
the the as
and
rough
wishes is
respites) THE lunches of
he review
timing
on one
wishes preplanned
library
project, the
studies research groups
one,
will difficult number
material at
to
those groups continue
desired This
of often
can
of timing
collection DISCOVERY
for realizing
the
group,
some be
emerging
theoretical he estimates
groups he
to
structural for boards
study
state of
discovering
or tempo
difficulty
to interviewed
research,
devoted theory
do will are
focused
allowing
generate,
can
cannot
he to his of grants,
developing
scope
to
staff of points
generate
hours
for
situations
O
also the schedules.
that
know be
cities,
indicate and
will
project.
to estimate
research certain want
of
theory
GROUNDED
meetings)? can
years.
used
anticipate
and type
sampling.
be say
difficult raises
on conditions to
how and the
for of the the
ample study
and
beforehand,
theory, regions, answered—
be theory
or
verification studied, observation,
generating hoss
conceptual (see
the properties.
emerging
of final sociologist
for the at
takes,
kinds does
surveyed
the
give given
a (number
THEORY
theory, and
time
others
whole
to
prob routine
Chap much
range
How’ theo can
What
since time
the the
and
give the
that
and
the
to of
of
at
Theoretical are meet necessary?
tion r
his from action with during Colleagues affect (as ‘icualiiation
groups into
is sometimes viewed depth, could people
when rapport a minutes data hears current order
to since ing
tematic that terviews tions which
few periods
establishing
groups the
Finally,
time obtain
In
are
a suicides)? polio)? documents,
the
excludes before timing
of
also
all or sampling in
to proposal.
categories, field the even
best can to
respondents
or
is for
activities; there,
interviewing
time. or Sampling
people
get observes
situations slices allow
What usually it briefly
of
difficult. observed.
often require sociologist
also
half another building
hours,
studies. ho
The
at before being of this
it
waiting
Often
observation.
rapport such
all, How have
how
him of quickly,
many help
Since kinds a interviewing,
not
to
researcher
for
exploratory
the
and
consist
recognizing data this clay
shooed days
Particularly
he
arc be
occurring
to the time-consuming theoretical
had a as
To some
long necessary.
long
of
in
may
for
sociologist
house
comparative writes the of
he study
amount
without
with
sampled,
a allowed training are
researcher
judging of without establish
experiences
sample encapsulated
unscheduled
he does
may
of of spend
off
core
At
the
might
relevant.
in
them the
his
these
period,
will
open-ended
the the and
over
his
of
in obtain it weck, order
In sampling the
theory
to
temporal periods,
anticipated explanations,
may weeks of
people
or
rapport
take find
time beginning field
premises
need
sits purpose. at
later
observing
talk structual
groups
people
respondents,
time
to
in aspect
There
will. he to
or obtain
back it his
to situations
studies periods too, would
similar
with or
get
study
to stages
worthwhile
might
who times
usually
seasons,
quickly (such follow
data Theoretical contingencies.
conversations even
quickly he
fill
and for though
is He
the
timing of contingencies
of
no
his
at
subcontractors),
begin little, are or
talks
on research
out
of of of
the interfering clandestinely
data
listens may gain
months
imposed the
kind as or
the
data
to
requires to
one is.
the
data the
job
to
the for
interviewing establishing research, of
recovering
with,
be
if to occur
same
to
course of obtain
a collection
be
sampling
in research of research, group
year
while periods
any,
explore data
theory. clearer collec
and/or allowed
appear
course, getting
during a
limita inter
data
over
read that
time,
with
(as
few
to
sys
of his 75
the on
in
in
in UIE DSCO EHY OF GROUNDE iHORY Theoretical Sampling 77 respondents tell their stories. Later. ;v”en interviews and ob research is also e’cciting work but, as we have detailed, it lacks servations are directed by the emerging theory, he can ask the more extensive commitment to discovery of theory displayed i direct questions bearing on his categories. These can be an by research utilizing theoretical sampling. I swered sufficiently One final and important and fairly quickh. Thus, the time for any point: since each researcher is likely I[ One interview giows shorter as the n’mher of interviews in to encounter special conditions in his research, he will inevitably cseases because the researcher now onestions many peonle. in acid to the discussion of theoretical sampling as outlined in this different positions ar d different grout s about the same topics. chapter. We would scarcely wish to limit this type of compara Aithor gh the time taker by most interviews decreases as the t s e anal>sis to what we can say about it, from either our own ti eory develops, the sociologist still cannot st’ te how long all ‘esearch or our knowledge of others’ research. We have merely h’s interviews will take because a new category might emerge pesed up the topic The motto should be: the more studies are at any time: this emergence will call for lengthy open-ended based on theoretical sampling, the more effective should future conversations and prolonged obsers ations within some groups ieoretical sampling and comparative analyses become—pro lsc theoretical sampling aimed at fobs dig an incident or ssdecl researchc a write about their strategies and techniques. observing over a eriod of time requires sequential interviews, I with no clear notion of when the sequence will be terminated.
:;‘si Conclusion
I:’I Theoretical sampling, then, by providing constant direction to research, gives the sociologist momentum, purpose and confi dence in his enterprise He develops strong confidence in his categories since they have emerged from the data and are con stantly being selectively reformulated by 36them. The categories, therefore, will fit the data, be understood both to sociologists and to laymen who are knowledgeable in the area, and make the theory usable for theoretical advance as well as for practi cal application. The sociologist will find that theoretical sam pling, as an active, purposeful, searching way of collecting data, is exciting, invigorating and vital. This point is especially im portant w’hen one considers the boring, dull, and stultifying effects on creativity of the methods involving separate and rou tine data collection, coding and analysis which are used fre quently in descriptive and verificatory studies, Conventional field
36. Theoretical sampling would have avoided the dilemma facing Wat son and Riesman (op. cit.) in their study of sociability. Watson feared the muof he” detafled, prco”iceh ed code when starting to collect data, since Riesman lacked confidence in it and wanted to change it completely, If they had undertaken an active theoretical search for categories that worked and fit, then tile preconceived code could base been selectively reformu lated with the approval and confidence of both researchers. particular
to
ever necessary, to generated strategic formal gives become
latter
properties
some ‘t
ure application from failure, the parative parative and
using
al
is
that the
Since
As
education,
1,
relevance
in
political
may
difficult a
this an
For
not
way
colleague: development
science. we
theory.
and
specific
analysis failure,
almost
initial have link
substantive
term,
example,
only
substantive
to in
and directly
remarked still stimulated
behavior.
many
of
start
to
in
mental
The important
in
these We
provides
that others
area.
‘comparative
direction
automatically
“Thanks
find
choosing the
one the
areas
notion
from
believe
wouid
Others
references
of
A a
sent
fonnulation
theory
formal
health)
author in
area
by
grounded of
a
theory
very
a
data,
general
of
facilitate
in Chapter
life,” grounded theory
phoned
a
stimulus possible
that
f’tilure”
(work,
comparative
much
of developing
substantive
theory
provided
is
a
it
Other
at
it
79
this
and
although
grounded
implications
springboard
to
is
might
formal
for such
generating
From
arid
modes to
II,
book
give
juvenile
to
most
from
formal
research
colleagues
your
religion,
immediate
a
a
substantive
generation failure their
he
received “good”
relevant
theory.
theory
conceptual
article
a
desirable,
Formal
of
formal
Substantive
in
substantive
a taken
theory,’
ideas
references, delinquency,
integration.
and
formal
marriage,
wrote or
would
research
on material
idea,
the that
Often
about stepping Iv
categories relevance,
theory
to
comparative
of
letters
theory theory
following
and
seem
level,
apply
was
Theory but
Though
grounded
comparative social
one, for
the
on
Indeed,
usually
can
detailing
to it
of not medi
a stone
how
only
is
sub-
The
and one also
class, and
corn have cons
note
fail to
be
not
in
a Sub8tUfltitL to 80 THE DISCOVERY OF GROUNDED THEORY From For’nal Theory 81 stantive and formal theories are formulated by different authors. A quick perusal of any sociological journal ill demonstrate Sometimes in formal theory the substantive theory is implicit, that almost all sociologists believe this is the way to write formal having been developed preiously by the author or another wory! For example, Selvin and Hagstron recently hare pub writer lished an article entitled ‘Two Dimensions of Cohesiveness in S Groups’ In this chapter we sFall orly begin the discussion of the rail but this artic does not if r thc brouncled processes by which a substantive theory is advanced to a formal rmai theory its title implies, only a grounded substantive theory one. We should emphasize that, since our experience and knowl (about college women) written up a notch, At the close of the edge arc least extensive in this area, most of our discussion will daper. some comparatirc speculation is offered about broader be concerned w’ith general rules, positions, and examples of implications; there is no comparative research or analysis to initial efforts at generating formal theory. More specific pro establish formal theory. cedures await the time when enough sociologists will have gen Such rewriting techniques applied to a substantive theory erated grounded formal theory that their procedures can be roduce only an adequate start toward formal theory, not an codihed, Although we Ia k many specific examples, we feel iëqnate formnl theory its lf Pir bahly thc rescarchers are, certain of our general position on the ways that formal theory is tr ical, responding to the substantive stimulation with should he generated. Near the end of the chapter, we shall dis ome general implications. All they have done is to raise the cuss the closely related questions: “Why go on to formal theory?” conceptual level of their work mechanically’; they have not and “What are its uses?” raised it through comparative understanding. They have done nothing to broaden the scope of their theory on the formal level y comparative lnvestigation o different substantive areas, Generating Formal Theory I G have rot escaped the tii se ar d place of their substantive oscarch though their formal riting of the theory rsay lead eaders into thinking One-AreaFormal Theory so. A classic example if this type of theory riting is Merton and Kitt’s theory of reference group behavior There are at least two “rewriting” techniques for advancing We can only wonder what such theories might have looked like a substantive to a formal theory that is grounded in only one sub f their authors had done the comparative analyses implied by stantive area, The sociologist can simply omit substantive words, heir writing, phrases or adjectives: Another instead of saying “temporal aspects of danger of the mist hug tech s iue as used on a dying as a nonscheduled status passage” he would say “temporal ingle su t i vc area i that f tie rcadcr, it terds to dis aspects of nonscheduled status ocia h passage” He can also rewrite a da fr r the brmal theory Wise he theory s vcry substantive theory up a notch: instead of writing about how met, I comes icr to see hot it cnme from the da of doctors and nurses give stg,dx, i medical attention to dying patients cc the to na! thecin r ‘s rende ‘s the’ data ithont according to the patient’s social value, he can talk of how pro sn!’stantisy theory interrenini fessional services are distributed according to the social value of .lsn, toe tormal theory annot fit or work very is cli when eiients.2 By applying these rewriting techniques to a substantive littl : tTO (uth :onc mb’tantn’e area. and usually only one theory, the sociologist of can change the focus of attention from ae the Ca ‘, hc Sr se it canr ot reall:y F develop I suffi substantive to formal concerns. He writes a to one-area formal tl into s all h ntir e c and s ial flea theory on the basis of a i h i I substantive theory; he does not gen v ‘ n t d s s ii s ti ea vi ci erate the formal theory directly from the data, r ‘Moe o C 2. See Barney G. Glaser and Anselm L. Strauss, “The Social Loss of 4 Bob u ‘h ‘ nI TI.eori, ‘intl S’n-iol ‘uciun’ Dying Patientc” American Jnnrnal Non York of ursir,g, Vol. 64, No, 6 (June, 1964). ‘ Pr , ,.c,., . t,o !‘J’. -O
are As
thus few
too
begin of required.
when collaborators.
of
parative
in for “Awareness
wall tiple comparatix
subsumed lating in
or 507-14. tix
for it
Press,
goneralizod other
dictions
becomes, it new.
82
will
e
is
formal
sociologists.
order Quota
.5.
6
Two
Multi-Area
we
its
When
busy
obtaining
not generating
theory,
good
put too
of
For
29
groups
1959),
C.
the
theories
to
a
be
For
more
have
formal
confined
theory
examples
Wright
(1964),
example,
Achievement,”
analysis
to
working
and sparsely
generate
theory,
the Yet
applied.
division
in
advancing
synthesizers
in
further
Contexts
e
p.
generate
which
by
analysis
discussed
Formal
of
the
abstract
actuality,
wall
the
65
more
It
explanations.
There through
that
core
Mills,
theory
p.
comparative
one
and
see
it
discussion
process.
need
to
is
task
on
from
dcx
All
676.
of
formal
and
becomes provided
together. 6
data
Theory
ts
passim.
Donald
the
substantive categories
situations
of
their
The
a
formal
are
labor eloped
that
American
level
can
in
from
who Social
not
the
our substantive
and
groups
treated
same
from
never
Sociological
‘Ihe
a
sec
theory
own
happens
Roy.
be
within
own comparathe
be
harder
works.
recent
wish
the
Thus
and
an
theory.
to
analysis
Barney
Interaction,”
These
logic
done
for
many
in
is
and
Sociological
“Work
data! relegated
enough
use
efficient
THE
as
area,
work
research
still
through
the
which
to
generating
sociology
to
the
Ihe
article,
is
theory
a
used
C.
their
by
DISCOVERY
in Imagination
While
worthy
kinds
search
generate
the
that Satisfaction
wider also
substantive
will
Claser
making
one
bricks
rexsriting
one
guide
method,
American
in
people
most
properties
to
bricks
Review,
comparative it
suggest
“awareness
to
will
of
the
discovering
sociologist
area
and
out
will
will
people
the
range
OF
a
either
substantive
and
the
powerful
to
(New trustworthy
provide
process
and
formal
Anselm
GROUNDED
the
are
of
be
have
techniques
distant formal
since
selecting
Sociological
latter
18
theory
there
how
a
Social
of modified
are
and
substantive
bricks
dying,
York:
(1953),
multitude
contexts”
built
or
L
analysis by
research
one,
substan
one
of
because
a
method
usually
are
formu
Reward
theoiy
future
Strauss,
ThEORY
a
to
areas,
guide
Grove
com
itself
mul
and pre
few
but
can
the
too
the
are
Re
pp.
be by
respective
is
is
are (‘it,
of physiasl
some
instance. tic
quic
trern
archs
also
“wise’
ens
present
\egro resented
of
interactants.
which
the preliminary
nations,
exist
Negroes cars,
our
substantive
quently.
contexts,
as
From
two,
Cr?
n
the
ison
as
Comparisons
insignia
an pp.
Quick
kly
Au
of
nailni
Glae
of
dying
vary
found
substantive
primary are
.
hustling
are,
tional
Substantive
three
pr
persons
people
are
669-679.
staff inter
they
camps
leads
And
asso during
(sk
and
by
upon
marks
as
:n?Ty
if
sex.
we
ii
for
hierarchically
and
iesc
scrutiny
ger
:‘tuation
members’
thr
we
areas.
can
analy
sit
ctar
or
whites,
uniforms
iated
Different
the
of
to
arc
The
instance,
asici
c
in
who
S
who
the orally
the
iat
more)
irjci
context,c.
the
jntcractars
ra
course
of
wish to
of
Jo
he mutual
theory
pool
‘s
automatically
sis
Formal
seconciars Here
an,
ethers
each
hypotheses.
s,
identifying interaction
sk1] know
of
development
‘a
stat
do usefisliv
.
spying
i
to
halls
Az
and
or
five
c’ert
,w,nbors
suggests
these
Different
clowning
not
the
n
differences
category
of
is
suspicion
all
(same
develop
Theory
his
the
as
of
Situations
rca
s.
how
clothing
co
may
)
awareness
stakes
comparative
as
kinds
i
seceindai’y.
compared.
Ihe
situations
igns—those
The
e
secret,
is
agility
N
signs
may
signs
bchaxioral led
a
Suppose
this
(on
of
and
or
-ev
numbc-’rc
g
of
ratios
at
for
ass
usual
a
of
positions
of
of
mteractantr
to
different
betwee
var”
st
properties
might
forn-wi
circuses,
cci of
‘
prismerc
one
and
uhre
diverse
of
of
and
the
social analyzing
‘
contexts
i
of
status
practice
c the
The
irs status.
categories
nd
bidding
a
mark,
that
(one
millions
of
so
interaction
tnat
insiders
z
be
-i
America.
ih
S
s 1 ecch theory
of
s
card
interaction)
groups
awa
substantive
o
s
level
signs
buying
done
eli ‘1
of
one
and
Sal
i.
genItalia
patient
interactants
nay
strongly
tj
c-an
uc
Some
(in
one
data
For
war
carried
Interaction”
eness
a
shark),
the
of
and
ti
the
of or
of
of
or
focuses,
h
starting
ir
be
dying
can
Negro,
our
skin
and
any white
from
n
may
passing
signs
gestrre) tI’ indicators
them
the
awareness
and
formula-
outsider.s
tcrnis
thought
involved
suggest
contexts
Chinese
he
groups
out
re
gisen
Conse
earlier
selling
color
some
othc
bier
many
sass
may
doz
r
and
with
five
are
and
for
op.
of
h
of
of
r
83
I
ii w 84 THE DISCOVERY OF GROUNDED THEORY From Substantwe to Formal Theory 85 status especially when found in conjunction with primary nno ence to byst ir den or e cci to police arid later to a court signs—would be, for “Negro,” “kinky hair” and perhaps “south f law, Sometimes identifying signs are “rectified” falsely! The ern-style speech”; and, for sex, clothing, hair style, and gesture. ew signs are believed and accepted, even though the original rhe visibility of such signs depends on learned ability to recog indications were really true, In “mutual pretense” situations, the nize them, for instance, many people have never learned to dying patic it in some sense rectifies the notion that he is hoinci,xuah and utlier uuuld sw,t know an Amen lying By anting much alise; given the ambiguity of can Indian if they saw one most signs, other people act up to his false rectification, until Understandably, some interactants may not even recognize the signs are either so unambiguous that the game is hard to the signs of their own status; for instance, the dying person play, or until he drops the pretense and admits his real situa may be kept unaware of his own position (closed awareness tion 10 A subjective and subtle variation occurs when an inter actant’s status is rejected context). Signs can be manipulated, both crudely and subtly. and he himself begins to doubt who I For instance, they ma simply be removed from vision, as when he is, as in Nazi Germany when gentiles with faint Jewish stigmata are concealed. They can be disguised, as when kinky Imeage came to doubt their true identities because their claims hair is straightened or, as John Griffin did when passing for to be non-Jewish were denied, Negro. skin color is changed temporarily with chemicals. Signs Such comparisons of diverse groups in terms of identifying can also be suppressed. as when an interactant8 chooses not to “signs” quickly lead to both useful properties and hypotheses indicate that he is really an American spy, or when a Japanese- about this facet of a formal theory of awareness context. Just as “1 American visiting Japan speaks Japanese at a department store in the development of substantive theory, the hypotheses will be so as not to be recognized as a “rich American.” All these tactics, concerned with such matters as tactics and counter-tactics, as of course, are aimed toward minimizing potential recognition by well as with their structural conditions, their consequences, and other mteractants. so on. But it is important to understand that this kind of inquiry Counter-tactic’s consist of eliciting important “give-away” can be furthered immensely by systematic analysis, not only of signs, to avoid having to wait for signs and hoping to recognize a single category but of combinations of categories: signs and them. Some counter-tactics for recognizing persons who are stakes, for instance; or signs, stakes, ratios of insiders-outsiders. suppressing their identity depend on “passing” as a member of and numbers of group representatives present at the interaction. their group (all FBI man posing as a Comrnnnist, or on getting This kind of analysis becomes increasingly richer, because it information from others within the group. Persons of similar leads the researcher to ask “Where can I find another compari status may use conventjonal signs to further recognition; the son group that differs in one more specified respect?” When he deliberate use of these signs will vary, depending on whether finds that group, its examination leads him to further generation outsiders are present or absent, and whether they are “w’ise” and qualification of this theory. By such means, exceedingly complex sympathetic to insiders ) or not.i Usually there are places and sseli-grounded formal theory can be developed. It where the gathered insiders can forgo their efforts to disguise is precisely by such means that a substantive theory of aware or suppress iclentilsing signs. But they max’need (as with drug ness contexts can be extended upward in conceptual generality addicts counter tactics to avoid betrayal even in such secluded and outu ard in scope. In doing so, many more useful types of places. awareness contexts would be generated and related to inter It is worth emphasizing that identifying sigm sometimes actants’ behavior. need to be rectified—as uhen a customer in a store is mistaken Status passages Our second example is the initial generation for a salesman, or a man mistaken for a thief must prove his of a formal theory of status passages, prompted by our substan
8. John H. Griffin Black Like Me (New York, Signet Books, 1962). 10. See Glaser and Strauss, Awareness of Dying (Chicago: Aldine Pub 9. See Erring ColEman Stigma (New York’ Prentice Hall, 1963). lishingCompany, 1965), Footnote 4, p. 279. 86 THE DISCOVERYOF GROUNDEDTHEORY From Substantive to Formal Theory 87 tive theory on the status passage involved in dying)’ We have in hospitals can he located b all thesc structural dimensions in written about the “nonscheduled status passage” of dying; sev the following svav: the status passage is nonscheduled, nonpre eral other dimensions (properties of status passage also arose scribed. undesirable and, after a point, inevitable. The passage from our study, of is a passage is One these whether or not status sometimes regulated but sometimes not; and sometimes rela follows an institutionally prescribed sequence of transitional tivciv unambiguous but except for its end status ) sometimes statuses. For instance, many ethnographic descriptions of grow not. ing up and aging, and many descriptions of organizational The next step is to study different types of status passage careers, delineate prescribed passages. (Such may or in passages order to begin generating a formal theory. Various conihina may not be precisely scheduled.) tions of the above dimensions provide ways of typing different “Transitional status” is a concept denoting time in terms of status passages as n eli as some of the conditions under which the social structure. It is a social systems tactic for keeping a the passage is managed. Differences between two sets of these person in passage between two statuses for a period of time, He conditions will, therefore, tend to explain why two types of is put in a transitional status, or sequence of them, that deter tatus passagc s are managed differently. mines the period of time that will be in a status passage. he For example, in the United States the engagement status pas TF us the transitional status of “initiate” will, in a particular sage (between the statuses of being single and married) is case, carry with it the given amount of time it will take to usually institutionally nonscheduled like dying, though unlike make a non-member a member---a civilian is made a soldier by dying it is desirable to the parties involved, Because it is a spending a given number of weeks as a basic trainee; an ado status they have chosen, the status occupants themselves deter lescent spends a number of years ‘in training” to be an adult. mine when they are in passage, what the transitional statuses A third dimension of status passage is the degree to which will be, and for how long a 1] period they will he in each one, In II) it is regulated; that is, to what degree there are institutionalized contrast, couples involved in personally undesirable or forced operations for getting an occupant in and out of beginning, engagements, such as sometimes found in Europe and Japan, transitional, and end statuses and for keeping others informed especially among the upper class, do not control their own of the passage. Rites of passages are instances of such regulated transition. operations. It is notable in of dying patients that our studies A status passage that contrasts with both the engagement the nonscheduled status passage involved both fairly regulated and dying is the defendant status passage, which links the and fairly unregulated temporal elements. One regulated aspect statuses of citizen and prisoner. It is scheduled and undesirable. is that at certain points in the passage the doctor must announce Commitment to a state mental hospital can be regarded as an the death to a family member. But less regulated is the typical instance of the defendant passage. In contrast to dying, while problem: when (if ever) does the physician tell the patient that the legalized legitimator of the passage is a judge, the unofficial he is dving The regulated and unregulated elements of the legitimator can he, in fact, a lawyer, a general practitioner, a nonscheduled status passage together generate one structural psychiatrist, the family, or the “defendant” himself. Thus, any condition leading to differential definitions among parties to the one who would be an unofficial legitimator must develop tactics passage. to make both his claim as such ‘stick” and his definition of the Further dimensions of status passages include to what degree defendant’s sanity status accepted by the court. Comparative the passage is considered undesirable: whether or not it is analysis of the characteristic tactics in this situation with those inevitable; and how clear are the relevant transitional statuses used during engagement or dying passages can be useful for and the beginnmg and end statuses of the passage itself. Dying developing a formal theory. Also, 11. Glaserand Strauss, ‘Temporal Aspects of Dying as a Non-Scheduled useful comparisons between the recover’ and dying Status Passag,” American status Journal of Sociology (July, 1965), pp. 48-59. passage are provided by a study of the polio patients who From Substantive to 88 niE DiSCOVERY OF GROUNDED THEORY Formal Theory 89 reflections recover from their acute ‘ittacks of polio hut who suffer varying about various kinds of institutional settings (e.g., degrecs of muscular impairment 12 This particular kind of bureaucratic, small establishments) and situations (courtship, recovery passage is non-institutionally scheduled or prescribed demotion). undesirable and, after a point, inevitable. One difference with An examination of his paper quickly shows that, in fact. Goffman begins dying ‘s that the end ta us—ref crc the passage will lead—is by pointing to comparison group that he does not frequuitlv unclear The doctor is uncertain about the degree to later build systematically into his comparative analysis. He which the patient sill regain use of the affected muscles. As a uses these initial comparisons to set his own point of view result, the doctor as legitimator is often very chary with infor squarely before the reader (quite like Cressey in I he Taxi- mation to family and patient, both in the hospital and after Dance Hall),14 Thus, “losing a role” may occur through promo tion, discharge (even though after a time he may form a clear idea abdication or demotion, Likewise, demotion may or may not of where the patient will end up). ‘Ihis lack of clear announce involve reflection of the person’s capacities. Each of these ments about the end status stimulates the patient and family to comparisons, in fact, can be built into the emerging theory to engage in vigorous searches Ioi ues which might define just give it much more scope and depth. Even if demotion alone how much better the patient can he expected to get. is focused on, Goffman has offered useful cues for extending In Davis’s account of the polio recovery there is very little his analysis Thus, what happens when demoters and demoted information or analysis concerning the coordination of people’s both agree he has been demoted, as over against when they behavior that is obtained by defining statuses correctly. The define him as demoted but he does not? What about the reverse reason is easy to find: while our study was focused upon medi situation? What about when demoters (and bystanders) do not cal personnel in the hospital, his study—especially in later phases agree among themselves:- And when they’ are differentially ‘4 of the passage to “getting better”—focused largely upon the above or below him in status? And when there are variable family outside the hospital. The medical personnel would not dimensions of “awareness context”—whether “open,” “closed” or -s be so concerned with coordinating a passage outside their organ “suspicion”_concerning agreement or disagreement? Also, what S izational jurisdiction. about the distance that he is demoted? And when more than The above examples are taken from our research; however, one person is demoted simultaneously? Other cues for theoretical ‘4 as we noted earlier, anyone can begin generating formal theory sampling are offered in passing by Goffman. He remarks that directly from published theory. For instance, he might system criminal gangs sometimes can afford not to cool out the client, atically extend Erving Goffman’s “On Cooling the Mark Out.” but department stores necessarily must be concerned. The In this useful paper, Goffman focused on the type of status implications of that important point—including when each demotion that reflects on the incapacity of the demoted person. party can or cannot afford to cool out—are not followed through. “Cooling out” means demoting him while simultaneously’ taking We are told also, through a passing remark, that agents who measures to minimize those of his reactions that would be most cool out may themselves react (as with guilt) to their actions. destructive to the institutional setting n here the demotion But what different kinds of agents. under what conditions, occurs. Goffman’s theory of “cooling out” encompasses such react similarly or differently? Also, if we scrutinize what we are matters as when this process occurs, what typical tactics are offered in the way of tactics for cooling out or situations where used in cooling out, and what happens when the demoted per it typically occurs, then we find lists of tactics and situations son refuses to be cooled out. The theory is built on Coffman’s that are related in the analysis only rather loosely to different types of organizations or situations. Systematic comparison of Bobbs-Merrill, 12. Fred Davis, Passage Through Crisis (Ind’anapolis’ organizationseither through field research or, 1963). quite feasibly, 13. Psychkitry, 15 (1952), pp. 451-63, 14 See our commentary on this common practice in Chapter III. through
will 90
theory theory else to straetly all, away”
he comparisons what
another the will own of scendence
earlier, on
possible have can periences. making to
Glaser, motion tional
‘d
and
Davis’ (1965),
120-32. Interaction,”
firsthand This
has
grounded
ithin
Direct
Formal s
generate quickly many
what
15.
16.
theory.
occur
data.
emerge
Douglas
formal
comparisons
usually
suggests
through
suggested because Scientists
For
paper
“Stable kind secondary For Industrial since than
theory, in
Cf.
to
pp.
any
Formulation
substantive comparisons This
to research of
Industrial extension
instance, formulate
our
theory begin
research.’
American As
are of
Moore,
714-24
in
a theory, theory.
on stimulation ColEman’s,
the
other substantixe
come Carecr
comments that theoretical
(Indianapolis: formal strategy
of scrutin
further the
these
“Deviance Oigaiiizations,”
suggested
in
analyst,
to
analysis
his
are
formulated
passing “Demotion,”
one of
and sociologist’s an sociologists’
of
N.
from Management.”
but
densify
Sociological That
of
formal
Coffman’s
areas
analyses initial
Formal theory
The important
suggested
may comparisons—directed
Martin Comparative the not
on
and
scholarship.
and
and
allows
Disavowal,”
just formulations
a
sampling
of his
theory
directly but
only
Bobbs-Merrill,
begin
strategy
author
is
theory substantive illustrative
the
guidance,
focus; may
Theory Journal published
paper
by
and as THE
Social
hard
work, has directly
Revietv,
feed
mind speedy
theories.
permits
emergent
if generating strategy
comparative
American
by
max
DISCO’,ERY be
A.
Failures,”
not
in by
directly. he
has
second, is
Problems, Business,
He along
to consists
Social
into
“Awareness
one’s
Strauss,
from
tied
his
29 followed
to even were
find, from incorporation from he extension
substantive
may
the
1964);
theory. forgotten
(1964),
these
begin in
analysis;
the
Problems, fonnal
formal
much Sociological
developed
Ch’iptr
his
OF
own
The if efficient
29
what thinking generating
as of
9 begin
“Patterns
comparative
one
by unacknowledged,
GROUNDED
analysis,
lines, development
Fred
(1962),
reading,
(1956
up;
Contexts asking, we
with
However,
core pp.
reflections
theory
that more of
theory.’5
area; comparisons
10 or
immediately
and
third, research—
9
have
see Goldner,
669-79.
Coffrnan’s h generation
categories about in and
someone
Review,
of (1961),
theory— less
pp. “thrown i’p
Barney
from first closely THEORY
and formal
without fourth, Organiza
life
though
Mobility
noted what
101—10; 213-20.
tran
data
it ab.’
Social
his
“De
on of Fred
ex
of
pp. is
30 G.
Rerolution of ization,’° variables one
1961;, Social he thing appears sue this cud ‘rus-dkal borrowing and thus patory a cult substitute models missing.
From that The bforc right” but significance? dangers directly 1i-onies data do
irs
S.d
Pt
way
ideas,
ph
18.
17.
and, For This
abandons The
111
0.
-essiun
the
1,
problem
of a no
task gix
or the
procedures Substantive File
See, For
Sos’
pp.
C and
set
is
to
of-
“useless” succession”
Merton’s he Sch,
its lug one
of
example, categories sociologist
approach above.
in
This
formal
The
Since arise
ii
and
(Chirago iple ‘In
13s’rnard the when
362-75, thinking of
for
for’
dubious
the the and
\,w-
is
do his
ad
up
will
exariple
I
properties may
If
s
generating through,
tactic research
data sociologist
his
xanip
ways 0
memos neat, they out
the
when
4nn the to person-sets
\ork:
to
not,
they
categories
can really-
“anomie’
Loseoson,
and
are
slip
fit
the
us—hecausi of-ha Formal theory
takes
search
fit
or Unix about must
borrowers
of Lookng
has
kr’or,
w explain of
clear, the
to
is’hri
Davrd bear
N
a “person-set.” and the ti into essentially
w from
authors
i’I
he errn
logico know
Cl
tht
formal
for
also
categories considerable
also
a Theory
s must
the
his on for Viiev’snd
“Buss J.
Ntieh.
guidance will
cc
ti
real work? adaptations, logico-dedectix’e
on are
ran
grounded it
Cath,xitz,
his
‘4
Sr1ir around been
e
predict
functional
what data
th theory, in be theory deductive that
of
P
rhetoric an
world
i relevant v make Chicago hase of comparisons their
ory
rcraOc re!evancc
C
wary
1Z-iu Y
theory
Are the
abstract
used
is that
the
to
are
tha
Sons.
k
Bk or
Seoial
Student has of
pertinent
atte many
nay
a The
certain
formal
and
do .‘sity
discipline
same Suc
they
of
Prms,
IL
u. of
theories would useless
model sounds with
in
to data
formal a
requisites
not
so
1964).
npts
with
as
using another . Exche,ni
sound
ession,” sociologist
Change
substantive
interpret order data. Microfilms. category
fledgling
often
clearly
formal
been an of Faculty umeba
1939), Parsons’s as
can
theory’ through
“nice
help
through for
it.
to even substantive
explanation
theorists.
the
those
on
In because
“nice
demonstrated not
he in
in lack data.
formal
integrating
of
him
Chapters
and theories
ani indicated Relatkius Amitai
other
rhetoric
antici1,atorv
the
if
like aud a anvthmg
in
substantive
to faced
Inc.,
an
patterned pilot
empirical very they and suggested
the theory
“neat’
In
generate
favor Indutrle1
Power
run
V’inst
“antici
theory,
organ several
of
words,
Etzioni
1960)
short.
areas,
neat” rhet
1,
with diffi “feel
tests
as
in
any and
out
but
hut
II
rn
by
of it
of ir
is 91
a a referents,
question is
mal and easier eration 92
data neat” case
sound nates conceptual categories, turns merely
tion in reasons stantive more tive about securities
rewards emanate On knows sociologists difference
havior and covery
relevant
Another working
Most
First
20.
21.
theory to
that
of
under
areas
of
Chapter
speculations,
out,
the about
to
(New
broader Formal so
See
Upon
Cf.
grounded
an
his of
being
for
of
level.
of
of
sociologists ask
the
from
it
Again
logical.
theory;
to
Smeiser, orderly, Fred
formal well, of
and
from
a logico-deductive
between
data with
danger
mature framework,
all,
this
York:
the
will learn them
III:
logically the
and
borrowing
logical such
Theory?
In
colleagues
so
Red
generalities,
a Scientist’s
avoidance. and
data high
the
inside data
“Filling
formal help Free
researcher answer
op.
have
theory addition
the over The
very which
“postal
to
expertise
and
specialization
result
a
unquestionably
feels
cit..
is
with Press
level that
worked-out
b.
result
formulate
dilettante
data the no
Anseim and
the early
not theory; is
the Career,”
a
Chapter one
with
to
and clerk”
increasingly my are of
disciplining
is never
abstractness
abstractions,
to
years.
tendency theory:
we
Boxes.” out.
tends
the Glencoe,
are
used in
no
is
feels
the formal
the
of
L. the
grounded THE
a a
they
believe Social
11:
approach
simple
a
then done This
whe.
my Strauss,
and
set
specialized
The
growing
dictum
must
inherently
to
are
wider
to
grounded DISCOVERY
See “Some
How
tend 1963).
stay
theory
of
know to
theory? Problems
generate
conviction a effect internal forcibly
comfortable by
abetted
means directly also
he
and
categories.
there
slip
ordering
“The
true
fonnal
do
public.2’
principally
Empty
to
that
to asking
correct
love
one
purports Smelser,
in
from OF
I
avoid
on
sorting formal
field,
greater
Impact
know
This
“pro.” are
(1965),
that
feeling
satisfactions
there GROUNDED
ordered
Theoretical how by or generating
properties
theory.
of
the
tends of
the
because
several
Furthermore,
two the
The rewards
as
one’s logic
question
Collective
the
this
theory,
of
out
at a
to the
The
difficulties
following pp. is
true
he
confident
formula Rapid
mass
the substan
handle,
relative
to
by
a
theory facts.2° THEORY
further
theory
297-311.
domi
“nice, learns
Boxes,”
other
latter
great gen
they
keep
sub for
and
that
and but
the
Dis
of
is Be
theory
change of much
moved
substantive
in data
logico-deductive to
their
vant have a
ology’s
to rests
theory.
theory, league
that “Also
applicability colleague
much substantie Thus, as structures,
of seem
ic as
and working
claim
sociologists
From
substantive
the
grounded
the
The generate
itc Other
too
Another
too
formal
from
own pass
theory,
on
future
neither comprehend
and
cipposod logico-deductiye I
real
Substant,ve
real. from data,
although
future,
their
“findings.”
unbounded,
but
depersonalization had
abstract,
suppose
theories
research.
colleagues as
more
many
its
wrote
world.
theory
wished
theory They
Many
for
area
the nor formal them.
just reason
sociological
from minds say,
theory
time
two
and
hut
depersonalizatitig
formal
sociologists abstractly,
theories,
real time I substantive
to
too of
published of
for
us, see
Depersonalization
links
am
so
great
researching sociologists powers
for
to Formal This
from
nor
for
theory
and
substantive if
interest
to
have
divorced
is world.
use go for
conceptual
apropos
sufficiently
they
only
ones. theories help
avoiding
with formal
inclination
the amounts
task
no
theory
the
research, for
of
Theory
in
told
because
of
because
further
know a
focus
for those one parsimony
areas, formal
Those
data:
We
consultation
it
is, time
explanation remarkably
from
resist
theory__there
more
his
us
is
are areas
to
which
the
of
area.
offended
are
level
effect. of
a
truly sociologists
rather
want
many
that
and they
own
in and
to
colleagues this
thinking
theory
course, people
formal
they data is generation
and
widely,
confident
(period!) generalizing
generate
they
It
minimized
and
of
may
the
difficult theory
do
Formal place no
area
substantive
distrust on
before
or
is
feel
its
by
plausible
and
is
theory
not part and
important scope
have
future
almost
not
will lean
lectures,
and
grounded
terms
most
the
who
of
who
of
they
is
that
actually
prediction. of
and theory
everyday
a
to of
they
really
not
no
based specialization:
airy
the
heavily
certainly specific
and
it
of can
formal
theory
need
it.” of
apparent relate
do
too
exclusively
many
must
so substantive to
areas
theory,
always
the
separation
can sociology
assertions but
for
help minimal
is This
not a
proceed
limiting.
far
formal
trust
on
a
formal
viewed
theory
them
theory
on life
from
amass
social
rele safely and will soci
who
see
the from One
re
in
col
nor
be
in
a its
to
a
93 a may from formal
musicians
94
ways. theory,
derived area, legion. Uses
tries materials Robert
of cation chances conditions tive show contrast, qualification.
alizes
the
substantive effort
post-hoc sometimes data prepare
substantive
Chicago
Insofar
research
A
A Probabh
22. 23.
be
process,
analysis,
of
with the
some
second and
First,
collection. third a theory
to
of
Robert H. currently
Blauner
Formal
found
from
new
f
Press,
lectures. or
our
give comparative
and enterprise
the
an
Becker, explain
respect as
small
(see
the is
area development
approach
he
theory.
not theory;
about
the important approah use the except prominent
central
typically It
Blauner, marijuana
1964).
his
in
Theory
formal
may
systematically
Chapter
Outsiders
portion the uses
that
The
he
most Becker’s sociologist of
variations
materials
to
A the
in
comparative variations insofar
tends
set product.
guiding
their
Alienation
the good
is
comparative analysis
sociologist
confined research
theories
widespread
is
social
body
users,2’
theorists.
out of to
of
(New
VI).
sociologist
to
Otrtsiders;
degree
to
example
one greater
in
new apply as
does
to
study
theory,
control
of
handle THE
an
and scrutinized York:
the and
of
This
to direct
after
or
verify,
theory
analysis
theoretical
established
concern also
Such
DISCOVERY
his
careful
several
of
old
more Freedom explanations
meaning. use
analysis with
of
Free
already
checking approach he
and
the
“alienation.”
it
substantive portions,
grounded
does
theory
in
verification of
based
Press
in carefully
formal
comparti7P
data creation generates
variation
himself
a formal
a
formal
OF (Chicago:
several
this conventionally, number
given
writing, knows
He general
of
seems
on GROUNDED
it
is
tends
became
Glencoe,
at
theories,
formal
under does
collected;
to
the
theories
23
with
generates of
theory, theories
least,
substantive
and
studies
alternative
and
well,
University This to
order
theory.
deviance
of
compara
as
to
reearc1i
this
ThEORY theory
rec1uire
formal diverse 1962). qualifi
part
indus
gener
when
of block often
in
type
how-
to
as
are
on
and
but
his
In
an a
to
of
a
a
of
(Ed.),
tation Rushing’s
can formal illustrated further formal I-Iarvard,
hunches, w’th tive begins
ever,
From
instrumental his reducing power concepts. from very ment tivelv on fully
conventional pressed fessional ing His central Like
I because then tions on can, three who
24.
read
present
area
fleldnotes
Using its
two
Rushing
Designed Substantive problem professionals and
Sociologists See
is
institutionalized,
in
indeed, are again
institutionalized Thibaut, on
but other theory
separate
institutionalized
theory
strategies), with experiences social
the
this:
thesis
a
distinct Blau’s notions, by
under The
the
individuals
interested
Mertonians
roles biased strategies),
in
asking
Among
qualitative
Psychiatric
book.
for
commentators
when
the
spent versus
theory the
psychological is a
role
and
Psychiatric engage
abound.
primarily
“hen
under
that and its
readings.
in
Homans, to at how
consideration,21
loose
general levels:
himself
following
in tow
interviews,
theory
Format concepts, description,
Work
the
influence, the
a
Udy’s the
in expressive
is
the
since
to
who
a
number
initiating in
relative ard analysis,
Role
psychiatric
but disappointing university mental
a
key
analyze conceptual structures from
However,
hospital
for
theoretical
formal
(New
a He
“modern
what
too lack
Merton
particular
approach Theory
work
Pro
the very
is
graduation.
on
theory sociologists
aspects
concepts—derived,
can
review
cost
static: and
Rushing
of
fess’ion.r:
“in
Columbia),
deprivation, and
activities. hospital of
York:
psychiatric
is
abetted
this
researcher’s
there).
useful
months
specific
read
theory
disappointing
(teaching) and This clear-cut
(and process
rather hospitals,
Staff.
hypotheses
therefore
mental
the finally
for
and
process,
(the
its
framework
Basic
in
professor (by
the
discusses
Parsons—.are
examples,
exercise
characteristic cost and
Power, For
forte setting. framework
some by
descriptive. observing
Examples
than
impact is
reread
book
researches
Books, of
Strauss)
hospital”
hospitals,
reference frequent
consensus
secondarily
this
inducing,
this
ungrounded needs
as
with
is
psychiatric
graduate institutionalization.”
regard.
those
to
first about
Conflict
in
by
So
chapter
I
well analysis,
book
in
of it
(Parsonians
1964).
particular
of
illuminate
gather,
the supplementary
and
for
giving
Philip
he
the for
is
of of
that power
is A
quotes
he
formal
This
group,
preventing,
can
a about as its difficulties
description
the
often
takes
interview
sociologist for its
not
William
this
establish
by
training
and
was
hospital.
sociologist
he
descrip are
Hammond
be
mainly himself with
are theory,
is
people
substan
chap-
focus
from
“yet”
finds
(and
rela
as
use
Ada pro read how and
linked
not
im ideas,
from
well
a
his
A.
p
of
in
it
95 96 TNF DiSCOVERY OF GROUNDED Tuvonv From Substantive to Formal Theory 97 of various auxiliary development. ter the social positions, plights and strategies He has not especially caught development in recreational descriptions, his personnel in the hospital: notably, social workers, which are relatively static, or in his theorizing, psychiatric nurses. The dis which is essentially workers, clinical psychologists, and non-processual. While the book is very useful a systematic and step by step presentation for its descriptive cussion turns around materials and detailed quotes, I believe it is of hypotheses, with qualitative evidence bearing upon them. Two also useful as an object lession about a type of prevalent re quotes from the concluding ch sptcr will conscy the kinds of earch style in the use of logical formal theory. hypotheses which he presents: “the typology of power strate 2 implementing cost-inducing, structural cost-reducing, and gies rhe several uses of formal theory discussed in this quotation cost-preventing. . . We hypothesized that this maintaining arc enterprises quite different from is related to the institutionalization process: the chaiac the generation of grounded typology formal theory, tcr of the particular power strategy—its function for the actor— accomplished through systematic study of mul tiple depends upon the degree to which social relationships have been comparison groups and substantive theories. Perhaps the closest institutionalized” (page 241) relative to such formulation is the kind of essay writing descriptive material offered throughout the book is The or established many years ago by Georg Simmel, dered by the theoretical requirements of each chapter. Anyone and nurtured by such contemporaries as who has observed psychiatric hospitals closely—including state Erving Goffman and David Riesman, in which the hospitals where the winds of current doctrine happen to blow essayist—with or without systematic data before him —develops softly—will i ecognize many features either explicitly dis a series of general propositions of relatively high or his descriptions abstraction, ussed by Rushing implicitly touched upon by Such writing can be criticized as being, at best, and by his interviewees’ remarks. The book teems with illustra full of insights and, at worst, as pure speculation. (Some “in tions of the ambiguity associated with auxiliary personnel’s tasks, sights” may later be of them and the “tested” by more rigorously minded soci conflict among these personnel and between ologists.) psychiatrists, of strategies for getting work done and professional From our viewpoint, such writing is exceedingly valuable, interests accomplished. but as theory it lacks both integration of well-defined Nevertheless my response to the book is that it is not suc concepts and sufficiently credible grounding in careful compara cessful in portraying—through joined description and analysis— tive research, a hospital that is very much “in process.” I lay the blame on an The more prestigious unwillingness to abandon conventional role theory for something style of logico-deductive, systematic “grand theorizing” bolder, something more suited to, as Rushing aptly regards it, is, in the hands of its most brilliant practi tioners, the non-institutionalized hospital. Rushing’s assumption is that more than merely esthetically satisfying: it also gives these hospitals are moving toward institutionalization—which is impetus to considerable useful, precise verification of hypotheses. probably incorrect, and if so still raises questions as to the most But it provides no directive—any more than it did a century fruitful ways of studying their institutionalization. Careful as is ago when Comte Rushing’s development and Spencer were its spokesmen—to closing of social psychological theory, it suffers that from the all too customary effort to fit combined bits of logical embarrassingly noticeable gap between highly abstract theory and formal theory to a substantive area, Not much, I suspect, is the multitude of miniscule substantive studies so really added to the formal theories other than indicating how characteristic of current sociology. It should be evident that portions of them can be applied in this particular substantive 23 we put greater faith in grounded formal theory to close that area. If I am incorrect in that assertion, then at least the book gap, for it readily fails to indicate how those formal theories (bearing on power, fits “what’s going on” in everyday situations, Possibly influence, cost, reference groups, relative deprivation) were the main benefit yielded by grand theories is their use modified, qualified or extended. of abstract models (mathematical, process, system, functional, As for the relationships among professionals in the hospital: interdependences, equilibrium, etc.). The integration of formal immersed as I have been in similar hospital settings, I miss in his 25. The review was account a quality of ongoing development of relationships. He published in Social Forces (Chapel Hill: The Unisersity of North Carolina Press, 1964), portrays very well the conflict and tension among personnel and 26. The gap was already embarrassingly noticeable touches occasionally upon outcome of conflict and tension, but Herbert in 1940 when Blumer commented on it, See “The Problem of the Concept in there is conveyed hardly any sense Social Psychology” of institutional or professional American Journal of Sociology, 45 (1940), pp. 709-19. 98 THE DISCOVERY OF GROUNDED THEORY From Substantive to Formal Theory 99 theory often requires more guidance from such explicit models organizations or governments. Most consultants are well known than substantive theory does, because of a greater level of for their research and everyday experience in a particular area, abstraction. However, as we stated in Chapter II, the integration and perhaps for a portion of their substantive theory if they of a formal theory can begin very usefully with the emerging have generated some The transferability of formal theories to ‘ntegration schcme that vas used for the sub lantive theory that diverse substantive areas is seldom done in sociological consulta ctually stimulated the formal theory’s generation tion because most formal theories are ungrounded, and there Because grounded formal theory fits and works, we see its fore not trusted by either sociologists or laymen when they face use in research and teaching as more trustworthy than logico “real life circumstances.” deductive theory, for the simple reason that the latter often Theoretical consultation is an area of sociological work that requires forcing of data into categories of dubious relevance to would be suitable for many sociologists, but cannot really be the data’s meaning. Grounded formal theory is also more trust opened up until there are many more grounded formal theories. worthy for sensitizing the researcher to the generation of new Then, for example, a general theorist, not only the well-known substantive theory and for helping him to formulate it. recearchers. could he called in for consultation about juvenile Grounded formal theory is thus also highly useful in predic delinquency because he is especially skilled at applying I tions and explanations when we are consulted about substantive grounded formal theory to substantive areas. Sociology cannot areas where we have no theory, and no time or inclination to reach this stage of development if we continue to plod on with I develop one. Explanations and predictions from logico-deduc grand logical theorizing and miniscule verifications. But this tive formal theory are used mainly where they will do no harm; stage can be reached through the generation of grounded sub that is, in the classroom, as “tacked-on” explanations of accom stantive and formal theories, Whether a substantive problem is plished research (as mentioned in Chapter I), and as hypotheses theoretical or practical, and whether extensive research is called (prediction) in the service of the perennial testing of parts of a for or not, general theorists skilled at applying grounded formal formal theory with the eternal hope that it can be modified to theories are needed as consultants for making cogent predictions fit reality. and explanations, and for helping decide the course of action Grounded formal theory is more trustworthy for consultations for research or practical action. because both laymen and sociologists can readily see how its predictions and explanations fit the realities of the situation. This is strategically important. While in research, predicting and explaining have few real risks (the researcher merely modifies the theory according to his findings), a layman does not trust a prediction of what will happen in his situation unless he can readily see how it applies Similarly, he will not accept a theo retical explanation unless he can readily see how it explains his situation, and gives him a sound basis for corrections and future predictions. Grounded formal theory, like substantive theory, earns the trust of laymen and sociologists alike. Both consultant and constiltee must have this trust in order to work together see Chapter X). As yet there is not much of this type of consultation in sociology. Seldom is such a general theorist (if you can find one’ called in for consultation by other sociologists, laymen, V
The Constant Comparative Method of Qualitative Analysis *
Currently, the general approaches to the analysis of quali tative data are these: 1. If the analyst wishes to convert qualitative data into crudely quantifiable fonn So that he can provisionally test a hypothesis, he codes the data first and then analyzes it. He makes an effort to code “all relevant data [that] can he brought to bear on a point,” and then systematically assembles. assesses and analyzes these data in a fashion that will “constitute proof for a given proposition.” l 2. If the analyst wishes only to generate theoretical ideas— new- categories and their properties. hypotheses and interrelated hvpotheses—l cannot be confined to 0th practice of coding first and then analyzing the data since, in generating theory. he e1is constantly redesigning and reintegrating his theoretical notions as he reviews his 2materiaL Analysis after the coding operation * 5’e wish to thank the editors of Social Problems for permission to publish thu papr .s Chapter V. See Barney C. Glaser, Social Problems, 12 1965), pp 436-45, 1. Howard S. Beehr and Blanche Geer, “The Analysis of Qualitative Field Data’ in Richard N Adams and Jack J. Preiss (Eds.), Human Organization Research (Homewood. Ill.: Dorsey Press, Inc., 1960), pp. 279-89. See also Howard S. Becker, “Problems of Inference and Proof in Participant Observation,’ American Sociological Review, (December, 1958), pp. 652-60; and Bern’ird Berelson, Cont nt Analysis (Glencoe, Ill,: Free Iress, 1952), Chapter III and p 16. 2. Constantly redesigning the analysis is a well-known normal tendency in qualitative sesearch (no matter what the approach to analysis) which eurs throughout the whole research experience from initial data collec 101
stimulated
l’ootnote
I Paul
Functions articulated.”
in
I
something
actually
ment
and
then.
may
approach
strategy. of
PP. for
n
ous 1 v
noted
retical lion
for Theory
method
sinab 1 tic
allowed
approach, The
and
approach
constant tatis
F,nglc’wooo
.ipset
memos
data
burdensome pose,
102
ailabie
publications
hich
ould
the
3.
an
Systematizing
390-92
cosniopolitans
F through
be
prox’isioi>al
h
c
purpose
he
Our
analysis
in
‘uid
for
purpose
collected
excellent
e
a
and hind
1
>u
indicated
developed
sampling
data—
1. not
will
Becker
wish
macic
On
:‘arsf”ld
presented
hid
of
of
by
procedures.
the
of ness
comparison
other
Neil
Op.
and
Chili,
the
However,
this
Social
comparative
these
coding
it
Qualitatis univ
to
the
the
a
to
of
work
of
in
task
he one
example
detailed
the
is J•
properties
cit and
explicit
purpose
the
method
the
or
by
the
suggest
the
connection
testing.
elarits
possihic
N.J
second
Snielser
as-on
Strncture
. unnecessarily
her’,
Only
Sec to
properties
whether
sociological
to
the
Ceer,
p390.
compiled tnat
a
of
development
As
first
this
style
constant e
.
generate
direct final
Allen
account
Prentiec-lIall,
of
While
Becker
the
Analysis
when
the
imum ir
second
a
with
approach;
tendency
combines,
op.
does
a
odirig
to
(lids.
aiiah
approach.
how
not
analysis
of
presenting
result,
This
of
tendency
third
(New
mity
wstls
quotation
H.
codify
explicit
for
cit..
fr>ternih
his a
which
qualitative
theory
(>1
comparative it
and
jot
more
sis
(IISC()5 B 1 rtii
.
is,
considerable
infiuentj:ds
theory
in
itself
collecting
goes
his
the
approach P. delay
Sociology: approach
theosetical
York.
may
and
of
but
iiiethods the
of
is
1Hi
Social
adhere
Ceer,
270,
is
1961).
sx
own
quantitative
by
the
coding
by
from
to course,
theory
on.
s
ering.
avs
development
used
and
in
often
writing.
far
have
analyst
SITS
Free and
stematic
more
be
Berelson.
using
constant
an
the
see
qualitative
in
and
Research,” the
data,
Robert
Paul
of
OVERY
It
the purposeli
“This
new to completely
used
body
to
method which
also
(inspecting
interfere
second
analytic because seems
Press
Robert
procedure
of
categories,
qualitative
practice
is
The
the
of be
systematically,
Progress
explicit
inerc’lv
F
the
the
niethods
hypotheses. 3
data
OS
comparative op.
of K.
jointly
part
suppressed
than
Berelson,
qualitative
of
analysis
Lazarsfeld,
tendency
approach
position
K. an
analysis
liv
such
GROUNSIRD
\Icrton—a
basic
in
of
of
cit.,
Glencoe,
of
it
procedure
with
of
or
Merton,
as
unnecessary,
of
joint
analysis
Seymour
incorporating
inspects
to
coding
the
the
is
of
with
p.
and
our
reports
have
position
data
an
on
a
designed
the
of
cited
that
has
of
the
125;
his
anlyses
in
Decade
second,
and
analytic
second
method
coding
“Some
previ report,
1957), writes
quali
THEORY
theo
Social
locals
state
been favor
This
with
first. than
been
and
and
pur
has
first
and
are
M.
the
his in
of
of
achieving
presentation.
his to
tractable method
after ferent constant
tween hypotheses quantitative enough
approach. Chapter
suggest, theoretical
coding.
approach
be tested, fIrst
theory. ing
the
and.
m partially.
rt’d
plant
Still that
,st
constant
Bather.
The
1 n’tntitatixe
esults;
more
A
the
The
If
used
findings
analyst,
n-ness
independently
Constant
signing
cicpendeiit
because
fourth
js
prosing
wishes
the
a
methods
same
levels
typically’
integrated,
it
is
B
first rigorous
theory.
eomparatis to the
fom
researcher
X).
comparative
so
some
of operationalized
is
skills
They
concerned
sampling,
and
couched
with
I.
general
contrast, yield who
a
time
both
Comparative
analysis
designed
constant
that
Of
inspection
of
of
approach
to
or
These
developing
synthesis
is
on
Partial
fhxiLiliC
and
theoretical
generality.
corn are discover
he
more
possesses
disproving
approaches
is
provisional
that
it
provisional
consistent,
the
e
with
in
approach
using
makes
at
coded
the
sensitivities
come
se.
method.
are)
can
the
with
to
a
comparatis
method
general
the
skills
testing
the
Method
also
form
and
the
data
the
allow,
at
theory)
c
be some
to
for
that
the
these
many
onstant
later
first
saturation, his
same
only
The
different
his
analyst
and
differs
(sometimes
memo-writing
guarantee
same
testing
collected
clear
provisionally
It
to
plausible.
ideas
of
testing
tests
of
first
is
discos-cries
approach
with
ad
or
hypotheses,
is
in
abilities,
m 1 tiahtatis
required
reason
hypotheses
enough
e
Qualitative
theory,
sensitivities
level
not
by
usually
the
all
comparative
enough
data
method
approach
the
using
suggested
in
and
discipline,
levels as
this
designed
are
in
of
are
scientific
of
that
another
for close
eicatisc
they
will
qualitative when
in
discovering to
must
methcd
quantitative
e
the
concerned
not
generality,
in
this
not by
to
Analysis
tested
of
analy
along
generating
is
attempt
two
generate,
this
synthesized
of
generating
achieve
be
coded
hypotheses
using
to
by
generality. (coding
extensise
arc
approach
necessary,
designed
keep
for
(as
the
method
enterprise
wax’
readily’, is
analysts
difference
the
his
getierotioll
does
in
with
in
some
hut
analyst,
methods is
the
extensively with
the
to
theory’
data,
data—and
while
the
from
the
“analytic hence
the research.
not
a
all
explain
enough
explicit usually
second theory.
cannot
at might,
theory’
if
to
theory
of
is
work a
to
same
same
thus
(see data
sup
onk
first
few
dif
our
left
be
the
the
the
aid
be
103
to
in
of of
249-331.
extent
cles,
parisons
theory information,
Free
constant
likely parative The
are
other
saturation tain 1947.
attempt
properties
dimensions,
unlike
of
these
hypotheses
method
formulation.
(but with of
generated
of
sisting
theory
a addiction,
Analytic
proving
a
induction
104
4.
Clearly
limited
the
services
manner
the
In
the
Method
Press
either
books,
See
not
to
properties.
pp.
negative
properties
of
contrast
supplement
phenomena.
analytic
phenomena of
data
of
is
required Alfred
be
comparative
method
is
ar
of
comparing
provisionally
12-14:
numbers
induction
of
causes
made
embezzlement).
number
the should
by
the
concerned
of
applied
and
according
different
Clencoe,
including
which
about
types,
restricted
integrated,
data—not
Sociology
R.
the
purposes
universality
cases,
to
Donald
induction
so
Lindesmith,
in
by
Since
max’
by
accounting
result analytic
reformulation
each
of
combines
forth.
Following
in of
forced
contrast
processes,
general
1953),
the
both
has
method,
and
clearly
from
observations,
cases
the
he
to
R.
hypotheses
testing)
to
(New
no
ith
consideration
constant
other,
of limited,
in
been
Cressey,
the
one
As
methods
causes,
same
what
others
proof
p.
by
In
Opiate
or
induction,
generating
both
the
to
which
an
York:
defined
a
problems
the
16
for
social
unlike
line
the
kind
constantly
the
etc.
analytic
as consequence,
concerned
many
study
integrated
data
and
constant
THE
is
Other
of
these
first
precise,
a
Addiction
comparative
are
second
as
ss
well
proof
Farrar
with
differ
involved,
of
do
specific
interviews,
In
ith
hypotheses
value
passirn;
rnscovpaiy
analytic
and
in
categories,
and
to of
conditions,
and
clearly
the
People’s
and
the
methods not
both
as
induction
all
analytic
any
of
in
all
and
approach, carefully
confionting
e
comparativc
universally
ideas
constant
(Bloomington.
the
second
of with
g.,
first
breadth
theory. plausibly
suggested
and
confirm
behavior
available
available
kind
Rinehart,
the
the
induction,
approaches,
defined
clients) Os
Money
method
first
documents,
the
approach.
Florian
and
for
are
generating
properties.
crsoenen
constant
induction,
constant
consequences,
of
requires
approaches
selected
and the
Further,
comparative
of
distribution the
generating
compared. qualitative redefinition
(New
suggesting
the
(e.g..
case.
data,
1934),
causes applicable
data,
to
Znaniecki,
purpose,
Some
theory Principia,
method. 4
is
second
current
theory
ascer it
SHF,ORY
more
com
these
com
York:
arti
only
cases
The
drug
tests
con
nor
and
but
pp.
and
no
or
of
in
is
patients.
comparisons
category —each
categories
data method
analyst stages
cessive
tive
anals
ddimiting gory,
I
to
tional
findings
ter of
erating qualitative
approaches.
Gciserating
The
he
qualitatve
We the
1
VIII).
Constant
method,
Constant
sis
emerge
Yes
No
Comparing
approaches
(2)
do
constant
stage
TABLE
starts
stage
of
shall
and
theory
within
of
Each
Theory
the
remain
of
generating
analysis.
integrating
Comparatice
of
social
All
each
after
until
by
I.
analysis
that (1)
analysis
describe
Comparative
theory,
relevant
nurses’
subgroups
Usi
in
comparative
four
coding
comparing
according
incidents
provides
in
hypotheses
loss” with
Combining
analyzing Analytic
analy’zing
Coding
a
the
fit
quantitative
OF
Table
time
operation
and
methods
theory
responses
an
and
as
coding
APPROACHES
analysis
in
response Method
of
categories
each
for
possible,
existing
induction
is
(4)
dying and
I
provides
four
data
data
Method
continuous
inspection
applicable
to
locates
Provisional
test,
transformed
(2)
for
method
is
Yes
incidents
incident
writing
of
with
their
provide
(1)
is
a
(1)
test,
simultaneously
to
stages
patients research,
along
involved
then
Qualitative
continuously
terminated.
TO
category.
(4)
the
as
and
the then
a
external
purposes.
QU-LITATIVE
for
can
scheme
the
Testing
in
categories
potential
applicable
development
the rise
to
different
their
emerged
his
into
also
theory.
the
Ethnographic
method Constant
Inspection
Then
(2)
each
Analysis
of
constant
groups
data
For
of
for
the
be
nurse’s
The
these
growing
properties,
throughout
Theory
deaths
ANALYSIS
category.
alternatives
one
emerge
Although
locating
used
example,
to
into (3)
quickly
next—earlier
comparative
general
for
No
approaches
(see
to
each
appraisal compara
compares
description
as hypotheses
for
of
its
process
Chap
many
or
from
their addi
cate
suc
‘Ihe
gen
idea
this
(3)
the
the
as
105
to ‘it
‘ai!tcmem,t
cc’
ot
h i that
Iov
sa it
lion
sta
SI t
actual t
I tLP
o
Ire ke
Occupation.
the of
il6ctor.
‘usband
I
il
t
a
1
106
f
Ut
orti,.
atm
aaras-d
lii/st
i is
i, P
at
social
ile
‘i
P
This
aai
the
pc
the
I
tit’
the ed
Codir it Ct
si
-i anus
1 c’aa
ilifi’
picu’auncc’d
to
and
S d
ii’cideit
1
onstaitt
nci P1’
5.
of
cml
co
d
t
tia
1
marital,
degiec
ainte
a
other
‘caved a
u
a
‘a,
(4
constant
tie 1
IIJ.
derrr “She
a.
Cm
tI
bc
aiau
lisT
do
ikng
that te
air
stir
s
tc’gorv
iS
los tand
‘
,lmai’m
mcmory
thc P
-
c-c P
-
la,a
tinT
e
I
Iv
th
t
or
vs
doni
oji
I (
- I
‘
eategoa
f
t ijd,
c-s cd
n’ine
onparaPvc
1lUi(
pnticrst
athont
it of
)
mpitals
a
ioi
laoforc
tatus.
muari
in
it o”ctic
of
Sock
0dmc
tI
me the
c
pa
c’verv
unit
ol corn
ails or
xi
di
tIcs
a
con
comparison
14
I
its
tcrrns
a
c mm
ti
ss
t)
as
‘thr
social
s bath
i
t
‘oaamnis
lull
P
to
1- tv
lPcr
comparisc
a
c-s
laer
a
is
m,
tie hat
s
it
C
damens
dire ist
I
also
care
education
snails
a
a
1 -d
I
establish pres
a
ded
ati
prop
iith
spcn
nag
a’
hte
aatier
arthr
dab
of lea
‘ii
i-Ic
Pitcnts,
-nd social ni
Iii
Tr
iicid
lns.
h
e
gory
niethorl:
‘if s”eial
-‘
apparent
tends
d
Baimi,
ions we
tF
a
in
Ii
ned, crtic
r
i/ic
.1.
widea
therc
‘1
or was
i
it.s
c
ts,
rat
ot
it, mid ‘as
a
scat c
n.
nt
i/u
atient
add
od1p.
at’
loss
I
)
THE
“W
a
Since
in,-.Cli-c oti
-
iv
p1’
P
Tm-s
to
It
roup a
I’I
in
‘fhia
dega
so
il1
o maJor
t sam
ts
Is
‘d
hat
c 1
a also
11
er
a
ting
s DISCOVERY
the S
riOi
th
aea
that
mci
which
il
airy
ncc ssould
young.’
on
lan
Pa cc 1,1
“
ale
.p
I
cc
properties
eo
obsura
c-I
will
c
1 I Pu)
-t
in
I
how
zed others
asic’. consequences
‘ategoly
coding
some
cc
F
of
‘P
nr
c-haute
tcgor’cs a
iicideuit
dations
ositia
of -‘itS
on
sticiat ‘a
luau
e
the
,id ame
a-as vs
s
Ii
i’ a
he
social
that
d anon
OF
hich
gc
typcs
un aims
“He
hr
cards
001
o
defining
P
nc-i’
x
to
children
social
whale
civ
P
I
as
ry t cry
GROUNDED
appal
c -us
au
i
lations
n-i
somc
(0cc
his
patse asni’x’attri
Fm a
under
lP,
loss
led
a
was f
-
the
refci
with
‘n
Tf
coioparisoii
or I I
on
a In” o
•.
soon
re
e
low
family
pre’
attributes
ent
s
I
c.
c
It
t spatiola
Ii
example.
us
C
continua
are
x
a
paticots
pond
its
incid
rule
to
mc
ic
ts
and
azuig
i sorir
-
an-aR ‘-
detree
(tv,x
,r
Stran
to iids
should
s;,f
wliicF to THEORY
Ci
social
starts
fro witi
tar
seer rein—
be
‘u
p
the
th
licr
r,
his
r Pt
6
to
nt
t
r
st a
i
s, a
r
gories,
takea
to are
depending clay’, tant ideas.
theory’, hours analyst’s
necessary thoughts. wall the family’
mates he comparative time, loss while
and tion
ments have discover
as
to substantive The
sariables (grounded maintains
consultation, the the
analyst
discuss
6.
determine
‘social
If As
After
no
in
explanations, 6
Constant
analyst
actual
of
compared.
as to
be (For
Thus
been
one
his
trying
on
the
a
This
scheduled
started
like
can
and
emergence
there
emphasi7e
ategories
a
will
or
continual
two
coding
theoretic’al
one musing
theoretical
personal In
is
to
dying
her
concepts
loss”
we example.
piocesses
on
in
abstracted
processes
of
consolation,
rule
“I
situation
Coin
help
will
working method
find
to
doing
kinds:
is
reflect
page
the
the
have
course the
crying
composure
was
in
At
paratce
concentrate
or
for
is
patient
notice
routine
conflicts
and
data,
For
over
alternate
process
bring
predesignecl
relevance
studies
that
or
of
to aensitia
‘calculation”
so,
this
designed
“composure”
constructed
those
afraid
a
and
notions
is:
notions
the
on vs
over
be
he
arid
example,
category
their
from
new
contraception,
the
ill
not
Method
for
stop
theoretical
point,
that
described
a
will
may
out
mood
carry
covering
of
of
that
ity tend
(a
in
their
behaviors
research
analyst
speculative)
of
joint
ways
categories.
the
properties
delinquency, vs
on
coding
of
change.
and
to
the
the
behaa
points affect
code
itli
into
losing
he of
the
research,
a
of
his perhaps
the
to
his
child,”)
language
by
of
tap
was
coding
Qualitatice
nurse’s
one
and
emphases
by
concepts
has
to
the
should
the
be
uon 5 ideration.
second
twenty
thinking
social
mr
material,
etc.
notions
the
team,
anti
study
the
I
relieve
which
explained
or that derived
missed,
my
an
constructed
analyst,
current
I
a
amount
emerge.
stage
justice.
conclusions.
analyst
more three
As
record in
The
perception
these
and initial
explanation
of
composure
loss
take
it
of
are
pages
rule
Analysis
his
abstracted
his
of
it
is
and,
to
the
the
analyst
saturation
analysis
from
the
teammates of
I;
or
since
usually add
“becoming,” labels
should
to
also
his
as
presence.
to
theory
a
freshness
the
its
will
of
and
four
formulation
research
in
conflict
memo
next
These
himself
be
much
be
at
nurses’
thinking.
the
most
a
points
of
this
a analyst
It
tend
may’
)
in
those
coded
become
good
times, explained,
half
the
when
there
be
the
develops,
from
is
how
incident.
constant
of
method on
time
use
factors
impor
Team logical
of
in
coded
spend
stigma,
to
social
(such
situa
state
hour,
same
cate
yoar
they
idea
that
will
can
per
the
she
the
the
the
his
He
for
be
107
the
of
as I’ S S I 108 THE DXSCOVERY OF GR0U,DED THEOflY The Constant Comparative Method of Qualitative Analysis 109 have run across in their own coding and data collection, and adult. since for a person of this age. education was considered crosscheck his points. They, too, begin to compare the analyst’s to be of most social worth, This example also shows that con- notions with their own ideas and knowledge of the data; this taut comparison causes thc accumulated knowledge pertaining : comparison generates additional theoretical ideas, With clearer to a property of the category to readily start to become inte ideas on the emerging theory systematically recorded, the ana rrated, that related in manr differcrit says, resulting in 1st then i a returns to the data for more coding and constant nified whole comparison. In addition, the diverse properties themselves start to become From the point of view of generating theory it is often useful 5 integrated. Thus, we soon found that the calculating and recal to write memos on, as well as code, the copy of one’s field culating of social loss by nurses was related to their develop notes. Memo writing on the field note provides an immediate ment of a social lOSS “stor” about the patient. When asked illustration for an idea. Also, since an incident can be coded bout a dying patient nurses would tell what amounted to a for several categories, this tactic forces the analyst to use an tory about him, The ingredients of this story consisted of a incident as an illustration only once, for the most important ontinual balancing out of social loss factors as the nurses among the many properties of diverse categories that it mdi earned more about the patient. Both the calculus of social loss c’atcs. He must look elsewhere in his notes for illustrations for and the social loss story were related to the nurse’s strategies his other properties and categories. This corrects the tendency for coping with the upsetting impact on her professional com to use the same illustration over and over for different properties. posure of, say, a dying patient with a high social loss (e.g., a The generation of theory requires that the analyst take I mother with tsso children), This example further shows that apart the story sithin his data. Therefore when he rearranges the category becomes integrated with other categories of analy his memos and field notes for writing up his theory, he suffi sis, the social loss of the dying patient is related to how nurses ciently “fractures” his story at the same time that he saves apt maintain professonal composure whilc attending his dying. illustrations for each idea (see Step 4). At just this point in his 7 Thus the theory develops, as different categories and their writing, breaking down and out of the story is necessary for properties tend to become integrated through constant clear integration of the theory. compari sons that force the analyst to make some related theoretical 2. Integrating categories and their properties. This process sense of each comparison. starts out in a small w’ay; memos and possible conferences are If the data are collected by theoretical sampling at the same short, But as the coding continues, the constant comparative time that they are analyzed (as ‘ae suggest should be done). units change from comparison of incident with incident to com then integration of the theory is more likely to emerge by itself. parison of incident with properties of the category that resulted By joint collection and analysis, the sociologist is tapping to the from initial comparisons of incidents. For example, in comparing fullest extent thc in vivo patterns of integration in the data incident with incident we discovered the property that nurses itself; questiors guide the collection of data to fill in gaps and constantly recalculate a patent’s social loss as the’ learn more to extend the theory—and this also is an integrative strategy. about him. From then on. each incident bearing on ‘calcula Emergence of integration schemes also occurs in analyses that tion” was compared with “accumulated knowledge on caleulat are separate from data collection, but more contrivance may be ing”—not with all other incidents involving calculation. Thus, necessary when the data run thin and no more can be collected. once ve found that age ssas the most important characteristic (Other aspects of integration ha e been discussed in Chapter in calculating soc al loss, we could discern how a patient’s age II,) affected the nurses’ recalculation of social loss as they found out 3. Delimiting the theory. As the theory develops, various more about his education, We found that education was most influential in calculations of the social loss of a middle-aged 7. See Clacer and Strauss “Asareness and the 1urse’s Composure,” in Chapkr 13 in A.,urLns.s.s of Dying (Chicago: Aidine Publishing Co.. 1965).
can cated higher
areas), fessionals for The into mainly
even tained damage, he portant—reduction.
eralizing, should example, covering he details
staff
become
discovered used erties, to
taking decided relevant dents delimiting
should
among lying
First,
Delimiting
Here
110
would
curb
Thus, Through
By
lived,
a
at
the
underlying
be
(not
by
uniformities
normal
why
the
is
of
this and
reduction
care
to
be
level
into
the
next
on
themselves,
fewer
generalized
major what
to properties,
or
an
the
that
a
now
how
with
be
theory
the
affect
forced
just
given.
point
the features
that
can be
occurs
the
analyst
elaborate
category
of
further
the
illustration
receive
better
concepts.
life.
nurses
care
our
in
and
be
could
nurses).
patient,
patients
outline
they
reduction
then
order
the
the
uniformity
be
we
solidifies,
theory
constant,
by
in socially
at
theory
fewer
of
starts
off
of
based integrating
rationales
reduction
as
distribution
could
mean
decide
to
a
the
formulate
otherwise
constant
to
two
our
all
of
of
the
dead.
a
given
service,
This
with
On
maintain
which
its
original
and
theory
to
as
patients
clarifying
interrelated
of
could levels:
on worthless;
theory
that
unendurable
in
constant
all
the
properties.
was
achieve
the
who
varying
delimits
terminology
For
his
the
the
some
of
which
comparisons
be
the
the
and elaborating
become
shows level
THE
of
analyst
be
of
degree
set
that
terminology
the
literature
example,
sense
professional
by
among
(not
how
considered
theory
analyst their
DISCOVERY
consequent
generalized the
what
comparative
two of
in
degrees
nurses
adding
theory
of
its
categories
all
the
categories
that
the
pain,
Later
just
spite
compares
formal
of
an
logic,
major
services
terminology
many
these
and
calibre
with
he
may
of social
integration
social
details
overwhelming
major
(some
dying
used, and
OF
detailed
or
of
of
we
other
would
modifications
composure
“loss
consequent
taking
requirements
theory,
a have
GROUNDED
rationales discover
waiting
reduction:
so
social
the
and—most
the
loss,
were method
values
to
smaller
or
of
the
when
modifications
of
ones)
comparisons
professional
that
rationales.”
clients;
have their
social
service
categories.
no
properties
next
strategies
would, and
out
loss.
also
of
it
of chance
talking
it
under clients
THEORY
by
set
begin
could
while
prop
brain
more
indi
task.
non-
gen
mci
text.
loss.
pro
per
dis
We
We
im
are
for
all
he
of
of
if
for
theoretical
to
the op. her
conditions
frequencies
back outweighed
spondence, the great
incidents
ease
negative referring
across
calculating
cable
only
coded
For times,
gories,
lvst focused.
of
coding, of
present
becomes
mass
His
tht.
scope
tions, 8 theory
The
code
9.
8.
cit., incidents
incident.
coding.
frequency categories
Another
social
The
example,
original
commitment
Constant
has
to
Merton.
If
of
adds
social
incident
of
in
a
he
and
pp.
is
while
the
developing
is
for
and
second
an
case
boundaries
under case),
qualitative (1)
to FIe
the
coded
that
reduced,
loss.
that
can
learns
considerably
128-34. bulk
sampling;
analyst’s
Furthermore,
a
social
compared.
her
loss,
age
85-year-old
factor,
on.
of analyzing
clearly
Comparative
list
category
applicability
can
keeping
parsimornj
where
to
also
after the
for
points
In
which
incidents,
cit.. the
age
level
for
to
count incidents
we
of
to
addition,
devote
loss,
See the
now’
generate collecting
purpose,
and
become
the
we
which
p.
calculating
case
categories
see
as
discovered
applicable
data,
one
age
to of
to
Becker for
theory,
thus
a
for
260
no
If
Merton
the
of allows
had
establish
coded
reduced
increasingly
close
a
quickly his
Method
can dying
op
of
was
establishing more
did
no,
delimiting
new
incidents
longer
of
new
the
for
still
besides
most
the
he
thcoreticathj
established variables
and
justify theory.
cit.,
since
and
the
not
correspondence
the
coded
has
him
data
aspect.
the analyst
theoretical
saves
amount
for
time
social
further
provisional
woman
that
to
of
when
Geer
important
whether
theory
pp.
made
did
appear
incident
coding
developing
frequency
Qualitative
this
time-consuming,
coding.
same
to
can
provisional
and
her
the
and
works
to
In
283-87,
time
for
we
loss,
If
cut
and
the becomes
the
of
smaller
delimits
the
become to
age
who
“wonderful
theory
turn,
yes,
saturated.
ideas.
a
nothing
then
need
proofs, to
category
data,
or
data
additional
down
new
is
However,
in a
data
factor
formulation,
As
better
and
constant
theory,
be
Analysis
as
wide
not
then
not
of
was
proofs
studying
See
his
compared.
the
the
to
method
set
the
according
the
is
then
are
committed
cross-tabulation
theory
the
more
the
the
to
coded,
careful
for
code
the
for range
a
Berelson
is
analyst
consideration.
theory
considered
After
a
of
point,
base
base
obtained
may
personality”
reduction
he the
list
also
original
if
comparison
calculating
next
number
incident
categories.
of ordering
select
we
incidents
most
and
his
counting;
and
also
of
of
theory. 9
line
counting
in
since
an
to
line
In
grows,
needs to
on
appli
came
corre
situa
count
data to
cate
data.
ana
code
feed
and
the
by
the
for (2)
the
list
(a
of
of 111
it.
is
in
it
a a
data,
his
large
mized tended toward
data
and
ments uses
library
coding
central
enough
theory. is
both it
then
after
If
he
concerning
theory.
to If
the whether
theory.
tinue
been start
sshether
pages.
emerge
a
(or
112
can
saturated; strategy
it
The
the
observed
go
time
Theoretical
Theoretical
notes,
is
saturation
has,
incident
is
additionally
from
field
since
comparison
to
for
missed
by be
back
based
of
unrecorded
to
by
to
The
universe
If
of
their If
data
first
after
code
and
the
then
or
a
this
added
collect
theoretical
the
the
the
studies, a beausc
the
time in
working few
categories.
answer
and
or
return
not
on
or,
delimitated
effort
if
place
new
will
and
coping
hundreds
theory,
for
theoretical
category
it
remembered
heard
field
saturation
saturation
of
the hundred
the
to
if
to
of try
is
to
more
it
the
category
probably
collected)
category
categories.
with
it
unnecessary
incident
theoretical
time
the
only
data
to
for
in
can
reduction
category
go
to
within
notes,
with
indicates
sampling. but
the
new
data
If
the
data.
does
next large
saturate
back
of
either
long
he
that
pages
on
the
and
memo
delimiting
did
of
can
another
pages
category
theory.
can
collection
incident
has
will
limits and
have
applies
data
memo
not
categories
analyst
studies
is
not
lists
and
the
of
data
then,
saturation
Thus,
be
help
a
of
be
to
minor.
Research becomes it
THE
become
saturate,
the
remember
new
for
of
record.
relevant
constant
little
coding,
ignored
of
go
re-code
forces
included
problem:
This
provided
and
have
to where
solve
coding,
theory DISCOVERY
of
if
has
is
generates
the
returning
possibly
back,
according
property
“no:’
the necessary,
an
However,
also
modifying
thus
incident
collected
What
theoretically
suggests
a
then
collected
the
or
resources
still
because
been
to
established
comparative
all
it
and
possible
directive
either
other
in
can
and
new
until
integrated
it The
his
emerges,
OF
analyst
useful
the
previously
a
another
does
a
is
of
the
to
be
to
memo
GROUNDED
coded,
the
alone new
categories.
analyst categories
incidents central
if
that
the
to
his
carefully
the
analyst
effect
the
the
the
delimitation
employed
universe
universe
he
it
are
for the
question
categories
own
remaining
to
saturated.
category,
category,
category,
what
becomes
and
category
do
require
directed
problem
field
into
may
method
further
field
econo
to
on
spend
should
to
THEORY
coded
needs
nos?
data,
con
that
will
ex
the
has
the
the
In
see
be
of
or
of
or
as
is
aehieement
provide
Tch
Institutions to
have
all
Properties pinpoint haps,
Buchr, writing couched
composure.”
use—then
work
ably thcory
“the
major
and
analysis,
behind likely of monstrous
incidents less
retical
and dents,
and
The
calculating
the
10. Using each
collected
When
memos
4.
likely
iques
Constant
a
thousands
money
been
accurate
further
forms
Writing
impact
coded
to
theory.
themes
On
each
Danuta
criteria
later
the
one’s
category,
illustrations’ 0
in
data
the
waste
the
and
(New
a
to
the
“pinpointing’
of
he
written
on
task sd
.
a
categories,
of
a
Comparative
data,
yield
social
data
of
Without
the
presented
analyzing
constant
analyst
theory,
form
can
Strategies
The
systematic
Ehrlich
calculating
behind
categories.
theon, 1 .
researcher (section
statement
which
are
of
York: of
to
time
a
Theory
which
when
if
an
publish
pages
complex fit
about
discussions
very
social
loss.”
that
done
Free
integrated
and
on
possesses
the
could
it
At
a
theoretical
comparative
ee
of
titles)
which
before
Method
in necessary
necessary
is
hypothesis
is
others
what
of
of
substantive
categories.
Melvin this
Press
available
I’ea,n
is
“the
social
at
Anselm
his
first
easily
loss
papers
be
theory
uotes
the
convinced
all,
stage
for
become
in
results
of
may
writing
coded
Fieldwork,”
of
patient’s
necessary going coded
product;
Shabshin,
loss
matters
can
his
on
Glencoe, accomplished
Qualitative
criteria,
our
Strauss,
for
embodying
to
or
resources
method
that
in
or
later
Thus,
memos
validate
for
theory,
become
a
the
paring
into
with
paper
books.
the
data,
the
about
gaps
that
multitude
corresponds
summarizing social
studied,
the
Psychiatric
1964),
prove
Leonard
to
process
delimiting
the
we
major
nurse’s
confidence.
Analysis
makes
provide
his
a
in
on
analyst
that
collate
down
of
it.
a
very
For
thousands
brought
series
same
since
suggested
loss
the
Chapter
to
social personnel,
One
analytic
themes
Schatzrnan, it
and
of
of
probable arbitrary
example,
be
Ideologies
an
theory,
is
professional
is
the
story,”
the
ways.
the
field
can
of
qualitative
closely
a
and,
irrelevant
loss
also
that
a
otherwise
2,
together
To
unix
memos,
content
reason
memos
memos
of
frame
return
“Logic,
of
point
could
more
were time,
theo
start
per
it
inci
and
and
erse
the
and
Rue
and
the
the
to
113 is 114 THE DISCOVERY OF GROUNDED THEORY The Constant Comparative Method of Qualitative Analysis 115 the data, since the constant comparisons force the analyst to taming to a conceptual area (such as stigma, deviance, lower consider much diversity in the data. By diversity we mean that class, status congruency, organizational careers, or reference each incident is compared with other incidents, or with proper groups) ,12 To be sure, as we described in Chapter IV, the level ties of a category in terms of as many similarities and differ of generality of a substantive theory can be raised to a formal ences as possible This mode of comparing is ir cortrast to theory. (Our theory of dying patients’ social loss ouId be raised coding for crude proofs; such coding only establishes whether to the level of how professional people give service to clients an incident indicates the fes properties of the category that are according to their respective social value.) This move to formal being counted. theory requires additional analysis of one’s substantive theory. Thc constant comparison of incidents in this manner tends and the analyst should, as stated in the previous chapter, in to result in the creation of a “developmental” theory)’ Although clude material from other studies with the same fonnal theo this method can also be used to generate static theories, it retical import, however diverse their substantive content) The especially facilitates the generation of theories of process. se point is that the analyst should be aware of the3level of gen quence, and change pertaining to organizations, positions, and erality’ from which he starts in relation to the level at which social interaction. But whether the theory itself is static or he wishes to end. deselopmentaL its generation. by this method and by theoretical The constant comparative methcd can yield either discus sampling, is continually in process. In comparing incidents, the sional or propositional theory-. The analyst may wish to cover analyst learns to see flis categories in terms of both their many properties of a category in his discussion or to write internal development and their changing relations to other formal propositions about a category’. The former type of presen categories. For eVample. as the nurse learns more about the tation is often sufficiently useful at the exploratory’ stage of patient, her calculations of social loss change; and these recal theory development, and can easily be translated into proposi culations change her social loss stories, her loss rationales and tions by, the reader if he requires a formal hypothesis. For her care of the patient. example, two related categories of dying are the patient’s social This is an inductise method of theory development. To make loss and the amount of attention he receives from nurses. This theoretical sense of so much diversity in his data, the analyst is can easily he restated as a proposition: patients considered a forced to develop ideas on a level of generality higher in con high social loss, as compared with those considered a low social ccptual abstraction than the qualitative material being ana loss, will tend to receive more attention from nurses, lyzed. He is breed to bring out underlying uniformities and diversitics, and to use more abstract concepts to account for 12. For an example, see Barney G. Claser, Organizational Careers (Chi publishing 1967) differences in the data, ‘To master his data, he is forced to cago’.Aldine Co., engrge ir 13. “. , the development of any one of these coherent analytic per reduction of terminology. If the analyst starts with spectives is not likely to come fromthose who restrict their interest exclu aw data, h wil nd up initially with a a ibstantive theory’ a sively to one substantive ares’ From F ving Goffman. Stigma: Notes on the Management of Spilled Idertity (Englewood Cliffs,N.J.’ Prentice-Hail. th ory or thc substantive area on whicl- I c I as done rescarcli 1963), p. 147. See also Reinhard Bendix, “Concepts ard Generalizations (for example patient care or gang behavior) If he starts with in Comparative Sociological Studies,” American Sociological Review, 28 the finding drawn f orr many stu die yertaining to an abstract (1963’,, pP 332-39 sociological category hc will end up with a formal theory per I “c rail. h,’ mO- de”elopmc xta,, opposed to sta’ic’, theones havi been r ade by vilbe Moore ‘Predicting D’scontinuit es in Social Ch nge Americas ciologi at Reicw 29 (1964), p 322; Howard S &cker, Ou siders (N w Yor : Free Press of (,lencoe, 1962), pp. 22-25, and Barney G Glaser a id Anseim Strauss, Awareness Contexts and Social Inter ctior ‘ op C t