Beruflich Dokumente
Kultur Dokumente
Barney G. Glaser
Stable URL:
http://links.jstor.org/sici?sici=0037-7791%28196521%2912%3A4%3C436%3ATCCMOQ%3E2.0.CO%3B2-T
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained
prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in
the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/journals/ucal.html.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academic
journals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers,
and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community take
advantage of advances in technology. For more information regarding JSTOR, please contact support@jstor.org.
http://www.jstor.org
Fri Jul 27 14:35:27 2007
ters. Even here it may be attenuated increased attention to residence and
in areas with strong religious and status in investigating illegitimacy. It
family controls. may well be that the generalizations
Restriction of this analysis to a par- based largely on studies of urban popu-
ticular metropolitan area precludes lations do not apply uniformly either
generalization of the findings to the to all segments of the urban social
country as a whole. The findings point structure or to the suburbs.
to the desirability, however, of giving
OF QUALITATIVE ANALYSIS *
BARNEY G. GLASER
U n i v e r ~ i t yof California
Research into social problems, prob- topics, stigma, and legality, and because
lems of deviation, of control and of people in these situations are usually
crisis, and the like-the general sub- adept at covering the facts when nec-
ject matter to which Social Problems essary, often the only way a researcher
is devoted-is still mainly feasible can obtain any data, or data that is
through methods which yield qualita- accurate, is some combination of ob-
tive data. Because these areas raise serving what is going on, talking in
problems of secrecy, sensitivity, taboo rather loose, sharing, fashion with the
people in the situation, and reading
* This paper developed out of problems some form of document that they have
of analysis arising during the study of
terminal care in hospitals; particularly the written. These methods best allow the
interaction of staff and dying patients. researcher either to gain the trust of
T h e study is sponsored by the National the people in the situation or, if nec-
Institutes of Health, Grant GN9077. An- essary, to accomplish clandestine re-
selm Strauss, Fred Davis, and Stewart
Perry have been strong sources of encour- search. In view of this distinctive rele-
agement in the preparation of this paper. vauce of qualitative data collection
I am particularly indebted to the extensive and analysis for many areas of social
editorial work of Robert K. Merton. Sub- problems, the constant comparative
stantive papers from this study are: Anselm
Strauss, Barney G . Glaser, and Jeanne method of qualitative analysis will in
Quint, "The Non-Accountability of Ter- particular, I trust, increase the battery
minal Care," Hospitals, 36 (Jan. 16, 1 9 6 4 ) , of alternative approaches useful to re-
pp. 73-87; Barney G . Glaser and Anselm searchers in these areas.
Strauss, "The Social Loss of Dying Pa-
tients," American Journal of Nursing, 64 My other purpose in presenting the
(June, 1964) pp. 119-121; Barney G. constant comparative method may be
Glaser and Anselm Strauss, "Awareness stated by a direct quotation from
Contexts and Social Interaction," American Robert K. Merton-a statement he
Sociological Review, 29 (Oct. 1 9 6 4 ) , pp.
669-678; Barney G. Glaser and Anselm made in connection with his own
Strauss, "Temporal Aspects of Non-Sched- qualitative analysis of locals and cos-
uled Status Passage," (to be published in mopolitans as community influentials:
the American Journal of Sociology) ; and
a forthcoming book, Barney G. Glaser and This part of our report, then, is a bid
Anselm Strauss, Awareness of Dying: A to the sociological fraternity for the
Study of Social Interaction, Chicago: Aldine practice of incorporating in publications
Press. a detailed account of the ways in which
Constatzt Compavative Method of Analysis 437
qualitative analyses actually developed. his material.3 Not only would analysis
Only when a considerable body of such after a coding operation unnecessarily
reports are available will it be possible delay and interfere with his purpose,
to codify methods of qualitative analysis
with something of the clarity with but explicit coding itself often seems
which quantitative methods have been an unnecessary, burdensome task. As a
articulated.1 result, the analyst merely inspects his
data for new properties of his theo-
retical categories and writes memos on
these properties.
Two general current approaches to In this paper, I wish to suggest a
the analysis of qualitative data are third approach to the analysis of quali-
as follows: (1) If the analyst wishes tative data, combining, by an analytic
to convert qualitative data into crudely procedure of constant comparison, the
quantifiable form in order to test pro- explicit coding procedure of the first
visionally an hypothesis, he codes the approach and the style of theory devel-
data first and then analyzes it. An opment of the second. The purpose of
effort is made to code "all relevant data the constant comparative method of
[that] can be brought to bear on a joint coding and analysis is to generate
point," and then the assemblage, as- theory more systematically than al-
sessment, and analysis of this data is lowed by the second approach by using
accomplished systematically in a fashion the explicit coding and analytic proce-
that will "constitute proof for a given dures. At the same time, it does not
proposition."2 forestall the development of theory by
( 2 ) If the analyst wishes only to adhering completely to the first ap-
generate theoretical ideas-new con- proach which is designed for provi-
cepts and their properties, hypotheses sional testing, not discovering, of
and interrelated hypotheses-the anal- hypotheses.
ysis cannot usefully be confined to the Systematizing the second approach
practice of coding first and then ana- by this method does not supplant the
lyzing the data, since the analyst, in skills and sensitivities required in in-
direct pursuit of his purpose, is con- spection. Rather the constant compara-
stantly redesigning and reintegrating
tive method is designed to aid analysts
his theoretical notions as he reviews with these abilities in generating a
1 O p . cii., p. 390. This is, of course, theory which is integrated, consistent,
also the basic position of Paul F. Lazars- plausible, close to the data, and in a
feld. See Allen H. Barton and Paul F. form which is clear enough to be
Lazarsfeld, "Some Functions of Qualitative
Analysis in Social Research," in Seymour M. readily, if only partially, operation-
Lipset and Neil J. Smelser (eds.), So-
ciology: T h e Progress of a Decade, Engle- 3 Constantly redesigning the analysis is
wood, N.J.: Prentice-Hall, 1961. It is the a well known normal tendency in qualita-
position that has stimulated the work of tive research (no matter what the approach
Becker and Geer, and Berelson cited in to analysis) which occurs throughout the
footnote 2. whole research experience from initial data
2 Howard S. Becker and Blanche Geer, collection through coding to final analysis
"The Analysis of Qualitative Field Data" and writing. It has been noted in Becker
in Human Organization Research, edited and Geer, op. cit., 270, Berelson, op. cit.,
bv Richard N . Adams and Tack J. Preiss, 125; and for an excellent example of how
~ o m e w o o d :Dorsey Press, inc., i960, pp. it goes on, see Robert K. Merton, Social
279-289. See also Howard S. Becker, Theory and Social Structure, New York:
"Problems of Inference and Proof in Par- Free Press, 1957, pp. 390-392. However,
ticipant Observation," American Sociolog- this tendency may have to be suppressed
ical Review, Dec., 1958, pp. 652-660, and in favor of the purpose of the first ap-
Bernard Berelson, Content Analysis, Glen- proach, but in the second approach and
coe: Free Press, 1952, Chapter 111, and the approach to be presented here, it is
page 16. used purposefully as an analytic strategy.
alized for testing in quantitative re- combines the first and second ap-
search. Depending as it still does on proaches in a manner different from
the skills and sensitivities of the ana- the constant comparative m e t h ~ d . ~
lyst, the constant comparative method Analytic induction is concerned with
is not designed (as methods of quanti- generating and proving an integrated,
tative analysis are) to guarantee that limited, precise, universally applicable
two analysts working independently theory of causes accounting for a spe-
with the same data will achieve the cific phenomenon, e.g., drug addiction
same results; it is designed to allow, or embezzlement. Thus, in line with
with discipline, for some of the vague- the first approach, it tests a limited
ness and flexibility which aid the crea- number of hypotheses with all availa-
tive generation of theory. ble data, which are numbers of clearly
If the person applying the first defined and carefully selected cases of
approach wishes to discover some or the phenomena. In line with the
all of the hypotheses to be tested, his second approach, the theory is gene-
discoveries are typically made by using rated by the reformulation of hypoth-
the second approach of inspection and eses and redefinition of the phenomena
memo-writing along with explicit forced by constantly confronting the
coding. In contrast, the approach pre- theory with negative cases.
sented here cannot be used for provi- In contrast to analytic induction, the
sional testing as well as discovering constant comparative method is con-
theory, since the collected data, as will cerned with generating and plausibly
be seen in the foregoing description, suggesting (not provisionally testing)
are not coded extensively enough to many properties and hypotheses about
yield provisional tests, as they are in a general phenomenon, e.g., the distri-
the first approach. The data are coded bution of services according to the
only enough to generate, hence, to sug- social value of clients. Some of these
gest, theory. Partial testing of the properties may be causes; but unlike
theory, when necessary, is left to more analytic induction others are conditions,
rigorous, usually quantitative, ap- consequences, dimensions, types, proc-
proaches which come later in the sci- esses, etc., and, like analytic induction,
entific enterprise. they should result in an integrated
The first approach differs in another theory. Further, no attempt is made to
way from that presented here. The first ascertain either the universality or the
approach is usually concerned with a proof of suggested causes or other
few hypotheses at the same level of properties. Since no proof is involved,
generality, while the constant compara- the constant comparative method, in
tive method is concerned with many contrast to analytic induction, does not,
hypotheses synthesized at different as will be seen, require consideration
levels of generality. The reason for this of all available data, nor is the data re-
difference is that the first approach stricted to one kind of clearly defined
must keep the theory tractable for case. The constant comparative method
provisional testing in the same pre- may be applied for the same study to
sentation. Of course, the analyst using any kind of qualitative information, in-
the first approach might, after either cluding observations, interviews, docu-
proving or disproving his hypotheses, ments, articles, books, and so forth.
attempt to explain his findings with As a consequence, the constant com-
some more general ideas suggested by
his data, thus achieving some synthe- 4 See Alfred R. Lindesmith, Opiate
Addiction, Bloominaton: Principia, 1947,
sis at different levels of generality. pp. 12-14, and Donald R. Cressey, Other
Another approach to qualitative People's Money, New York: Free Press,
analysis is "analytic induction," which 1953, p. 16 et passim.
Constant Comparative Method of Analysis
TABLE I
USE OF APPROACHES
TO QUALITATIVE
ANALYSIS
Provisional Testing of Theory
Yes No
( 2 ) Inspection for hypotheses ( 2 ) Inspection for hypotheses
along with ( 1 ) coding for
Generating test, then analyzing data
Theory Yes ( 3 ) Constant Comparative
( 4 ) Analytic Induction Method
No (1) Coding for test, then ana- Ethnographic Description
lyzing data
ops as different categories and their we were also discovering that our
properties tend to become integrated theory could be generalized to one
through constant comparisons which which concerns the care of all, not
force the analyst to make some related just dying, patients by all staff, not
theoretical sense of each comparison. just nurses. Even more generally, it
3. Delimiting t h e theory. As the could be a theory of how social values
theory develops, various delimiting of professionals will affect the dis-
features of the constant comparative tribution of their services to clients:
method set in to curb what could for example, how they decide who
otherwise become an overwhelming among many waiting clients should
task. This delimiting occurs at two next receive a service and what calibre
levels: (1) the theory and (2) the of the service to give him. Thus, with
original list of categories proposed for reduction of terminology and conse-
coding. First, the theory solidifies in quent generalizing which are forced by
the sense that major modifications be- constant comparisons-some of which
come fewer and fewer as one compares can now be based on incidents found
the next incidents of a category to in the literature of other professional
properties of it. Later modifications are areas-the analyst starts to achieve two
mainly on the order of logical clarity; foremost requirements of theory: (1)
paring off non-relevant properties ; in- parsimony of variables and formula-
tegrating elaborating details of proper- tion and (2) scope in the applicability
ties into the major outline of interre- of the theory to a wide range of situa-
lated categories; and most important, tions,7 while keeping a close corre-
reduction. By reduction I mean that spondence of the theory to data.
a higher level, smaller set of concepts, Second, delimiting the theory results
based on discovering underlying uni- in a delimiting of the original list of
formities in the original set of cate- proposed categories for coding. As
gories or their properties, might occur the theory grows, reduces, and increas-
to the analyst by which to write the ingly works better in ordering a mass
theory, hence, delimiting its termino- of qualitative data, the analyst becomes
logy and text. An illustration showing committed to it. This commitment now
both integration of more details into allows him to delimit the original list
the theory and some consequent reduc- of categories for coding according to
tion is the following. W e decided to the boundaries of his theory. In turn,
elaborate the theory by adding detailed his consideration, coding, and ana-
strategies which the nurses used to lyzing of incidents become more select
maintain their professional composure and focused. H e can devote more time
while taking care of patients with to the constant comparison of incidents
varying degrees of social loss. W e dis- clearly applicable to a smaller set of
covered that the rationales which they categories.
used among themselves could all be Another factor, which then further
considered "loss rationales." The un- delimits the list of categories for cod-
derlying uniformity was that all ra- ing, is that categories become theo-
tionales indicated why the patient, reticdlly satzlrated. After one has coded
given his degree of social loss, would, incidents for the same category a num-
if he lived, now be socially worthless; ber of times, it becomes a quick opera-
in spite of the social loss, he would be tion to see whether or not the next
better off dead. (For example, he applicable incident points to a new
would have brain damage, be in con- aspect of the category. If yes, then
stant, unendurable pain, or have no the incident is coded and compared.
chance for a normal life.)
By further reduction of terminology 7 Merton, op. rit., p. 260
If no, the incident is not coded, since saturate it, if the category is central
it only adds bulk to the coded data to the theory.
and nothing to the theory.8 For exam- Theoretical saturation helps solve
ple, once we had established age as another problem concerning categories.
the base line for calculating social If the analyst has also collected the
loss, it was no longer necessary to data, then he will be remembering
code incidents referring to age in from time to time other incidents he
calculating social loss. However, if observed or heard that were not re-
we came across a case where age did corded. What does he do? If the
not appear to be the baseline (a nega- unrecorded incident applies to an es-
tive case), it was coded and then tablished category, it can, after com-
compared. In the case of an 85-year- parison, either be neglected as a
old, dying woman who was considered saturated point or, if it is a new
a great social loss, we discovered her property of the category, it can be
"wonderful personality" outweighed added into the next memo and thus
her age as the most important factor integrated into the theory. If the re-
in calculating- her social loss. membered incident generates a new
The fact that categories become category, both incident and category
theoretically saturated can be employed can be included in a memo bearing
as a strategy in coping with another on their place in the theory. This may
problem: new categories will emerge be enough data if the category is
after hundreds of pages of coding. The minor. However, if the category be-
question is whether or not to go back comes a central part of the theory,
and re-code all previously coded pages. the memo becomes a directive either
The answer for large studies is "no," for returning to the notes for more
not until starting to code for the new coding, or for returning to the field or
category at the page when it occurs, and library for more data or for future
waiting for a few hundred pages of research.
coding, or when the remaining data The universe of data used in the
have been coded to see whether or not constant comparative method is based
the new category has become theo- on the reduction of the theory and the
retically saturated. If yes, then it is not delimination and saturation of cate-
necessary to go back because theoretical gories. Thus, the collected universe of
saturation suggests that what has been data is theoretically delimited and, if
missed will in all probability have necessary, carefully extended by a re-
little modifying effect on theory. If turn to data collection according to
the category does not saturate, then theoretical requirements. This theo-
it is necessary to go back and try to retical delimiting of the universe econ-
omizes research resources, since it
If the purpose of the analyst, besides forces the analyst to spend his time
developing theory, is also to count inci- and effort on data relevant only to his
dents for a category to establish provisional
proofs, then he must code the incident. categories. For large field studies with
Furthermore, Professor Merton has made long lists of possibly useful categories
the additional point in correspondence that and thousands of pages of notes em-
counting for establishing provisional proofs bodying thousands of incidents, each
may also feed back to the development of
theory, since frequency and cross-tabulation of which could be coded a multitude
of frequencies can also generate new theo- of ways, theoretical criteria are of
retical ideas. See Berelson on conditions great necessity in paring down an
under which one can justify time consum- otherwise monstrous task to the re-
ing, careful counting, op. cit., pp. 128-134. sources of the people and the pro-
See Becker and Geer for a new method of
counting frequency of incidents, op. cit., ject's allotted time and money. With-
pp. 283-287. out these criteria the delimiting of a
Constant Cornpdrative Method of Analysis 443
sis of one's substantive theory, and the ample, two related properties of a
analyst should include material from dying patient are his social loss and
other studies with the same formal the amount of attention he receives
theoretical import, however diverse the from nurses. This can easily be re-
substantive content.lG The analyst stated as a proposition: patients con-
should be aware of the level of gene- sidered a high social loss compared
rality at which he starts in relation to to those considered a low social loss
the level at which he wishes to end up. will tend to receive more attention
The constant comparative method from nurses.
can yield either property or proposi-
tional theory. The analyst may wish 15 ". . . the development of any one
to proliferate many properties of a of these coherent analytic perspectives is
category or he may wish to write prop- not likely to come from those who restrict
their interest exclusively to one substantive
ositions about a category. Property area," Erving Goffman, Stigma: Notes on
theory is often sufficient at the ex- the Management of Spoiled Identity, Engle-
ploratory stage of theory development wood Cliffs, N.J.: Prentice-Hall, Inc., 1963,
and can easily be translated into propo- p. 147. See also Reinhard Bendix, "Con-
cepts and Generalizations in Comparative
sitions if the work of the reader Sociological Studies," American Sociological
requires a formal hypothesis. For ex- Review, August, 1963, pp. 532-539.
HYMAN RODMAN
Merrill-Palmer Institz~te
Many things have been much said form important functions for individ-
about family sociology: it deals with uals and society, and as a consequence
sensitive issues and therefore the objec- family sociology is an important area
tivity of researchers and the acceptabil- of research and has produced some of
ity of research by the public have come the most important studies in the social
slowly; everybody is an expert in sciences.
family sociology, having lived most of The above issues have been dealt
his life in families, and therefore much with many times, and I shall therefore
of family sociology is trivial and com- not elaborate. Nor shall I go into a
monsensical; many groups have a general review of family research,
vested interest in "the family" and because the family area has perhaps
therefore family sociology is either had more than its fair share of such
subserviently conservative or insolently reviews and critiques. Textbooks, how-
radical; family sociology boasts of ever, have seldom been looked at crit-
many diverse studies but few binding ically, except in the course of reviews
theories; it has low status, unless sprin- about a single text at a time. I shall
kled with terms like "kinship" or therefore, in a critical vein, concentrate
"comparative" or "stmctural-func- upon family sociology textbooks, to the
tional" ; it is popular with students relative neglect of readers, and of texts
(either because of intrinsic interest or that are primarily practical (family
easy grading) and therefore suspect. life) or cross-cultural (anthropolog-
Despite this rather gloomy picture, ical) in orientation. I have made no
families of one form or another are attempt to rank the texts in order of
universally found, and universally per- their excellence; all of those that I