Beruflich Dokumente
Kultur Dokumente
(1971), Warwick and Lininger (1975), and Williamson and colleagues 49. See Hiller (1979).
(1982:315-319) For an additional discussion of access issues, see Armer 50 See Ember (1977), Harris (1976), and Headland and associates (1990),
(1973:59) and Foim (1973)..
51. For additional discussions of equivalence, see Anderson (1973),
43. See Haikness, van de Vijver, and Johnson (2003) and Smith
Armer (1973), Frey (1970), Holt and Turner (1970), Przeworski
(2003,2004),
and Teune (1970,1973), and Warwick and Osherson (1973)
44. Coupei and de Leeuw (2003) discuss nonresponse, and Johnson
52. SeeHazelrigg (197.3).
and van de Vijver (2003) describe social desirability issues
53. See Novick (1988:612-622) for an extensive discussion of the
45. For more on back translation, see Anderson (1973), Gtimshaw
David Abraham case
(1973), and Hymes (1970)
54. For a discussion of archived data, see Odette and Mautner
46. See Harkness, van de Vijver, and Johnson (2003)
(2004) and Richardson and Godfrey (2003)
47. See Frey (1970), Gtimshaw (197.3), and McDaniel (1978).
SeeMikkelsen (1995|
48 See also Bradshaw and Wallace (1996)
Much of the best work in sociology has been carried out using qualitative methods
without statistical tests. This has been true of research areas ranging from
organization and community .studies to microstudies of face to face interaction and
macrostudies of the world system Nor should such work be regarded as weak or
initial exploratory approaches to those topics..
Randall Collins, Statistics versus Words, p 340
S'
f Qualitative data are in the form of text, written y. words,
phrases, or symbols describing or representing people,
actions, and events in social life.
: Qualitative researchers rarely use statistical analysis
Their data analysis can be systematic and logically
rigorous, although in a different way from
quantitative or statistical analysis
In the past, few qualitative researchers explained how they analyzed data. In fact, a common
criticism of qualitative research was that data analysis was not made explicit or open to inspection
Qualitative data analysis has moved to a more ex-
plicit and systematic step-by-step approach 1 Nevertheless, no single qualitative data analysis approach is widely accepted.
This chapter is divided into four parts . We first
examine similarities and differences between qualitative and quantitative data analysis. Next, we look
at how researchers use coding and concept/theoiy
building in the process of analyzing qualitative data.
Third, we review some of the major analytic strategies researchers deploy and ways they think about
linking qualitative data with theory. We also look at
what researchers do not see, or how they use the
457
the covariation of one variable with another, usuevidence showing that a theory, generalization,
or interpretation is plausible.
ally across many cases . . , The quantitative reThe fourth difference is the degree of
searcher typically has only broad familiarity with
abstracts ' tion or distance fr om the details of social
the cases. (Ragin; 1994a: 107)
L
Fourth, in both qualitative and quantitative
forms of data analysis, researchers strive to avoid
errors, false conclusions, and misleading inferences
Researchers are also alert for possible fallacies or
illusions. They sort through various explanations,
discussions, and descriptions, and evaluate merits of
rivals, seeking the more authentic, valid, true, or
worthy among them.
Differences
First, quantitative researchers choose from a specialized, standardized set of data analysis techniques
Hypothesis testing and statistical methods are similar
across diffrent social research projects or across the
natural and social sciences. Quantitative analysis is
highly developed and builds on applied mathematics
By contrast, qualitative data analysis is less
standardized. The wide variety m qualitative research is
matched by the many approaches to data analysis.
Qualitative research is often inductive. Researchers
rarely know the specifics of data analysis when they
begin a project. Schatzman and Strauss (1973:108)
remarked, Qualitative analysts do not often enjoy the
operational advantages of their quantitative cousins in
being able to predict their own analytic processes;^
consequently, they cannot refine and order their raw
data by operations built initially into the design of
research.
A second difference is that quantitative researchers do not begin data analysis until they have 7
collected all of the data and condensed them into
numbers. They then manipulate the numbers in
order to see patterns or relationships. Qualitative
researchers look for patterns or relationships,
early in a research project, while they are still
collecting data . The results of ear ly data
analysis guide subsequent data collection Thus,
analysis is less a distinct final stage of research
than a dimension of resear ch that stretches acr
oss all stages
Another difference is the relation to social
theory. Quantitative researchers manipulate
numbers that r epresent empirical facts in order
to test an abstract hypothesis with variable
constructs By contrast, qualitative researchers
create new concepts and theory by blending
together empirical evidence and abstract
concepts Instead of testing a hypothesis, a
qualitative analyst may,illustrate or color in
life. In all
data analysis, a resear cher places raw data into
cat- : egories that he or she manipulates in order
to iden
tify patterns In quantitative analysis, this
process is clothed in statistics, hypotheses, and
variables. Quantitative researcher s assume that social
life can be measured by using numbers, then
manipulate II the numbers with statistics to reveal
features of so- * rial life.
Qualitative analysis does not draw on a
large,
; well-established body of formal knowledge from
mathematics and statistics. The data ate relatively J
imprecise, diffuse, and context-based, and can have
more than one meaning. This is not seen as a
disadvantage
Words an not only more fundamental
intellectually, one may also say that they are necessarily
superior |f ;. to mathematics in the social str ucture of
the discipline For words are a mode of expression with
greater open-endedness, more capacity for con- necting
various realms of argument and experience, and more
capacity for reaching intellectual audiences (Collins,
1984:353)
Qualitative analysis can eliminate an explanation by showing that a wide array of evidence contradicts it The data might support more than one
explanation, but all explanations will not be consistent with it In addition to eliminating less plausible explanations, qualitative data analysis helps to
verify a sequence of events or the steps of a process
This temporal ordering is the basis of finding associations among variables, and it supports causal
arguments
Some qualitative researchers are almost entirely descriptive and avoid theoretical analysis In
general, it is best to make theories and concepts explicit. Without an analytic interpretation or theory pr
ovided by the r esear cher , the r eader s of
qualitative r esear ch may use their own everyday,
taken-for- granted ideas. Their commonsense
framework is likely to contain implicit assumptions,
biases, eth- nocentrism, and ill-defined concepts
from dominant cultural values.2
CODING AND CONCEPT
Qualitative researchers often use general ideas,
themes, or concepts as tools for making generalizations. Qualitative analysis often has nonvariable
concepts or simple nominal-level variables.
Conceptualization
Quantitative researchers conceptualize variables and
refine concepts as part of the process of measuring
variables. By contrast, qualitative researchers form new
concepts or refine concepts that are grounded in the
data. Concept formation is an integr al part of data
analysis and begins during data collection Thus,
conceptualization is one way that a qualitative r esear
cher organizes and makes sense of data
A qualitative researcher analyzes data by organizing it into categor ies on the basis of themes,
concepts, or similar features. He or she develops new
concepts, formulates conceptual definitions, and
examines the relationships among concepts. Eventually,
he or she links concepts to each other in terms of a
sequence, as oppositional sets (X is the opposite of Y),
or as sets of similar categories that he or she
interweaves into theoretical statements Researchers
form concepts as they read through and ask critical
questions of data (eg,, field notes, historical documents,
secondary sources, etc.) The questions can come from
the abstract vocabulary of a discipline such as
sociologyfor example: Is this a case of class conflict?
Was role conflict present in that situation? Is this a
social movement? Questions can also be logicalf or
example: What was the sequence of events? How does
15.1 mmsrnMMrnMMmmmd
"A good thematic code is one that captures the qualitative richness
of the phenomenon It is usable in the analysis, the interpretation,
and the presentation of research" (Boyatzis, 1998:31). To code data
into themes, a researcher first needs to learn how "to see" or
recognize themes in the data Seeing themes rests on four abilities:
(1) recognizing patterns in the data, (2) thinking in terms of systems
and concepts, (3) having tacit knowledge or in-depth background
knowledge (e g , it helps to know Greek myths to understand
Shakespeare's plays), and (4) possessing relevant information (e g ,
one needs to know a lot about rock musicians and music to code
Illustration
"
oncewhen 1 got marriedand look what happened to me'. By God, the undertaker will have to
put the next one on 1 ordered a beer, then asked
him, Why did you get married?" He replied,
What the hell you goin to do? You just cant go on
shacking up with girls all your life1 did plenty of
that when 1 was single witha smile and wink. He
paused to order another beet and light a c igarette,
then continued, A man, sooner or later, likes to
have a home of his own, and some kids, and to have
that you have to get married.Theres no way out of it
they got you hooked " I said, Helen [his wife]
seems like a nice person " He returned, Oh, hell,
shes not a bad kid, but shes a goddamn woman
and they get under my skin. They piss me off. If you
go to a party, just when you start having fun, the
wife says 'lets go home." (Adaptedfrom LeMasters, 1975:36-37)
;BOX 15-2
Coding qualitative data, whether it is in the form of observational
field notes, video or audio recordings, open-ended interviews, or
detailed historical documents, is a challenge despite attempts by
Strauss (1987) and others to systematize and simplify the process,
making it appear as a fixed three-step sequence with open, axial, and
selective coding. Some researchers rely on text-coding software
programs (see discussion later in this chapter) that force them to
create codes, but the software is just one tool in a larger coding
process .
Weston and associates (2001) described their coding process
in detail, Weston worked as part of a six-person research team and
noted that team collaboration helped to make coding processes more
explicit. The ideal associated with grounded theory that a researcher
begins with a completely open mind and without prior expectations
is just that, an ideal. In reality, a person's academic training, awareness of concepts and theoretical assumptions, and expectations from
the audience who will read the research report shape data coding. In
Weston's study, the process began with one researcher on the team
creating a coding system that had four codes based on a first reading
of open-ended interview transcript data The system had a definition
for each coded idea and rules with examples for converting raw data
into codes. Others on the research team then used the system to code
selections of raw data Based on experiences with this preliminary
system, they revised the coding system and added subtypes of the
original codes. The process was repeated several times with the
team members individually coding raw data, meeting together to
discuss coding, and revising the coding system. After months of
coding and meetings, the initial four codes became three master concepts with two of the three containing two types and each type
having four to seven more refined codes. This yielded 34 coding
distinctions. Over the next two years, the research team applied the
system to hundreds of pages of raw data . Team members con tinued
the process of reflecting on codes, meeting to discuss coding, and
refining the system Eventually their coding system had four tiers
three master concepts, seven types under the master concepts, two
subtypes within three of the seven types, and several refined codes
within each of the subtypes In total, they created 58 codes
Over the next two years, as they continued to examine the data and
present findings to the scientific community, the team kept on
refining and adjusting the coding system They were following a
strategy of successive approximation (see later in this chapter). A
few new codes emerged and system's structure shifted a little, but
four years into the project, after hundreds of hours of meetings and
repeated passes through the raw data, the coding system stabilized
As you see, a coding system can be more than a way to code raw
data. It offers a system of analysis that provides a structured
interpretation By the way, Weston's research topic was improving
university teaching, and the team's data were from detailed openended interviews with six professors gathered during one semester
During axial coding, ask about causes and consequences, conditions and interactions, strategies and
pr ocesses, and look for categories or concepts that
cluster together You should ask questions such as: Can
I divide existing concepts into subdimensions or
subcategories? Can I combine several closely related
concepts into one more general one? Can I organize
categories into a sequence (i.e. A, then B, then C), or
by their physical location (i .e, where they occur), or
their relationship to a major topic of interest? For
example, a field researcher studying working-class life
divides the general issue of marr iage into subparts
(e .g., engagement, weddings) He or she marks all
Knights and other political parties, including temperance gr oups and the Prohibition party. The array of
concepts and themes that are r elated to temperance in
axial coding helps him or her discover how the
temperance issue facilitated or inhibited alliances.
During selective coding, major themes or con-
1--------------------------------------1
'..... .... ' 1
. . . . _ 1 ' ________________________1 ___________________ 1
11
'i
J Analytic ] i
1 | Other | i
1
i Memos i---------------------------- i Files i-------------------------------- 1
Filed by Concept
or Theme
Filed by Purpose
11
8,
9.
fO
ANALYTIC STRATEGIES
Techniques of coding, memo writing, and looking for
outcroppings are generic approaches to the
terpreted in terms of a social theory or the setting in
which it occurred. This allows the qualitative researcher to move from the description of a historical
event or social setting to a more general interpretation,
A source of confusion is that data take multiple forms
in various stages of qualitative research For example,
field research data include raw sense data that a resear
cher experiences, recorded data in field notes, and
selected or pr ocessed data that appear in a final report
Other ________________
Sources
Artifacts
Documents, Maps
Observations of Others
FIGURE 1 5 , 3 Data in Field Research. (Data 1 = Raw sense data, experiences of researcher;
Data 2 = Recorded data, physical record of experiences; Data 3 = Selected, processed data in a final report) Source: Adapted
from Ellen (1984a:2t4
)
Contrast Contexts. Researchers who adopt a organize qualitative data, An analogy is a statement
strongly interpretive approach may use ideal types to that two objects, processes, or events ate similar to
interpret data in a way that is sensitive to the context each other. Resear chers use them to comand cultural meanings of member s. They do not test
municate ideas and to facilitate logical comparhypotheses or create a generalizable theory, but use the
isons. Analogies transmit information about patideal type to bring out the specifics of each case and to
terns in data by referring to something that is
8
emphasize the impact of the unique context, Resear
already known or an experience familiar to the r
chers making contrasts between contexts often choose
eader. Analogies can describe relationships buried
cases with dramatic contrasts or distinctive features.
deep within many details and are a shorthand
For example, in Work and Authority in Industry,
method for seeing patterns in a maze of specific
Reinhard Bendix (1956) compared management
events. They make it easier to compare social
relations in very different contexts, Czarist Russia and
processes across different cases or settings . 9 For
industrializing England
example, a researcher says that a room went silent
When comparing contexts, some researchers do
after person X spoke and a chill like a cold gust of
not use the ideal type to illustrate a theory in different
air spread through it This does not mean that the
cases or to discover r egularities. Instead, they acroom temper atur e dr opped or that a breeze was
centuate the specific and the unique By contrast, other
felt, but it succinctly expresses a rapid change in
r esearcher s use ideal types to show how unique
emotional tone, Likewise, a researcher reports that
features shape the operation of general processes As
gender relations in society Y were such that
Skocpol and Somers (1980; 178) explained:
women were viewed like pr operty and treated
Above all, contrasts are drawn between or among
like slaves This does not mean that the legal and
individual cases. Usually such contrasts are desocial relations between genders were identical to
veloped with the aid of references to broad themes
those of slave owner and slave. It implies that an
or orienting questions or ideal type concepts.
ideal type of a slave-and-master relationship would
Themes and questions may serve as frameworks for
show major similarities to the evidence on relations
pointing out differences among cases Ideal types
between men and women if applied to society Y.
may be used as sensitized devicesbenchmarks
Ideal type analogies used to analyze qualitative
against which to establish the particular features of data oper ate as a heuristic device (i .e ., a device that
each case
helps one learn or see) . Analogies are especially
valuable when researcher s attempt to make sense of
Thus, one use of the ideal type is to show how
or explain data by referring to a deep structure or an
specific circumstances, cultural meanings, and the per
underlying mechanism.10 Ideal types do not provide a
spectives of specific individuals are central for
definitive test of an explanation.. Rather, they guide
understanding a social setting or process The ideal
the conceptual reconstruction of the mass of details
type becomes a foil against which unique contextual
into a systematic format.
featur es can be more easily seen
Analogies. Ideal types are also used as analogies to
Successive Approximation
This method involves repeated iterations or1
cycling through steps, moving towar d a final
analysis, Over time, or after sever al iter ations, a
resear cher moves from vague ideas and concrete
details in the data toward a comprehensive analysis
with generalizations, This is similar to coding
discussed earlier:.
A researcher begins with research questions
and a framework of assumptions and concepts. He
or she then pr obes into the data, asking questions
of the evidence to see how well the concepts fit the
evidence and reveal features of the data He or she
also creates new concepts by abstracting from the
evidence and adjusts concepts to fit the evidence
better. The resear cher then collects additional
evidence to address unresolved issues that appeared
in the first stage, and repeats the process. At each
stage, the evidence and the theory shape each
other;. This is called successive approximation
because the modified concepts and the model
approximate the full evidence and are modified
over and over to become successively more
accurate.
Each pass through the evidence is provisional
or incomplete. The concepts ate abstract, but they are
rooted in the concrete evidence and reflect the context.
As the analysis moves toward generalizations that are
subject to conditions and contingencies, the researcher
refines generalizations and linkages to reflect the
evidence better:,11
Spradley defined the basic unit in a cultural setting as a cultural domain, an or ganizing idea or concept. His system is built on analyzing domains.
Domains ate later combined into taxonomies and
broader themes to provide an overall interpretation of
a cultur al scene or social setting. Cultur al domains
have thr ee parts: a cover term, included terms, and a
semantic relationship. The cover term is simply the
domains name. Included terms are the subtypes or
parts of the domain A semantic relationship tells how
the included terms fit logically within the domain. For
example, in the domain of a witness in a judicial
setting, the cover tferm is witness . Two sub- types
or included terms are defense witness and expert
witness, The semantic relationship is is a kind of .
Thus, an expert witness and a defense witness are
kinds of witnesses. Other semantic relationships are
listed in Chart 15.1
Spradleys system was developed by analyzing
the argot of members in ethnogr aphic field research,
)
CHART 15 ,1 Forms of Relationships in Cultural Domains
RELATIONSHIP EXAMPLE OF USE
A bus is a kind of motor vehicle [kinds of vehicles]
A tire is a port of a car [parts of cars]
Cheating is a way to get high grades in school [ways students get high grades] ,
is a kind of
A train is used for tr ansporting goods [ways to transport goods]. High unemployment is a reason for public
is a part of As
a place in is a
way to
is used for is
a reason for
Analytic domains contain terms from the researcher and social theory. They are most
helpful when the meanings in a setting are tacit, implicit, or unrecognized by participants
The researcher infers meaningful categories and identifies patterns from observations and
is a stage of
artifacts, then assigns terms to them Domains are constructed from data notes. You r ead
is a result of/is a your notes, looking for common semantic relationships (eg.., is a kind of place, is a kind
of person, is a kind of feeling, etc.) in or der to find them You proceed by identifying a list
of cover terms. In the examples, a witness in a judicial setting or an attitude toward a
cause of is a
childs death are cover terms. Once you have a list of cover terms, you next or ganize the
place for
information from the notes as included terms. Prepare a worksheet for each domain relationship, The worksheet contains the cover term,
the list of included terms, and the semantic relationship. An example worksheet is
is a
shown in Box 15.4,
characteristic
Next, you locate your examples of the domain relationship from your notes. The
of
analysis pr oceeds until all relevant domains have been identified . You then organize
the domains by comparing their differences and similarities. Finally, reorganize domains into typologies or taxonomies and reexamine the domains to create new,
broader domains that include other domains as included terms.
Spradleys domain analysis formalizes six steps common to other forms of
qualitative data analysis. A researcher (1) rereads data notes full of details,
(2) mentally repackages details into organizing ideas,
(3)
constructs new ideas from notes on the subjective meanings or from the researchers organizing
ideas,
(4) looks for relationships among ideas and puts them into sets on the basis of logical similarity, (5) orga nizes them into larger gr oups by comparing and contrasting the sets of ideas, and (6) reorganizes and links
the groups together with broader integrating themes. The process builds up fr om specifics in the notes to an
overall set of logical relationships.14
Analytic Comparison
The British philosopher and theorist John Stuart Mill (1806-1873) developed a logic of comparison that is
still in wide use today. His method of agreement and method of difference form the basis for analytic
comparison.15 Researchers can use the ideal type, successive approximation, the illustr ative method, and
domain analysis to examine qualitative data from a single case or from multiple cases; however; for analytic
comparison multiple cases are required. Analytic comparison uses aquasi-experimental approach that
combines deductive with inductive theorizing. Basically, a researcher identifies many factors for a
Analytic domain A rultui il domain developed by i lesearcher using categories or terms he or she devi I oped to undo
stand a sou si etting
AnalyticcomparisonQualinmedatimnsisinvvhicharesearcher,risesthemethodofagreementaridthemethodofdifferencetodiscovercasualfactorsthat c tomeamohgasetof
cases.
15.4
'BOX
i
1.
Semantic relationship: Strict indnsinn
2.
Form: X (is a kind of) Y
3.
Example: An oak (is a kind of) tree
INCLUDED TERMS
laundromat hotel lobby
SEMANTIC
RELATIONSHIP
is a kind of
bridcie
boy
car alley
public toilet steam qrate
Structural questions: Would vou call an allev a flop?
s?
SEMANTIC
RELATIONSHIP
INCLUDED TERMS
trusty ranger
bull cook mopper
COVER
TERM
COVER
TERM
is a kind of
------
iail inmate
CAUSAL FACTOR
OUTCOME
CASE
State Breakdown Peasant Revolt
Revolution?
France
Yes
Russia 1917
China
England
Russia 1905
Germany
Prussia
Japan
Yes
15 5 Example of Method of
Agreement and Difference:
BOX
Yes
Yes
Yes
Yes
No
No
Yes
Yes
No
Yes
No
No
No
No
No
Yes
Yes
No
No
No
No
No