Transcript Yes

Content and Context:
Textual Analysis for Qualitative Research
Klaus Weber
Northwestern University
PDW “The Power of Richness”
Academy of Management
Hawaii, 2005
Overview
• What’s in a Text? A Semiotic Framework
• Text Analysis in Qualitative Research Designs
• What to do with it? Generic Types of Textual Analysis
• Example from Analyzing Corporate Discourse
• Basic Practical Decision Points
• Software Support Options
• Conclusion
Klaus Weber, Textual Analysis for Qualitative Research, Academy of Management Meeting 2005
2
What’s in a Text? A Semiotic Perspective
• Text = words arranged in order,
but of interest is more often the
meanings or actions that the
words represent
Conceptual image
of innovation
Sense /
Concept
• The semiotic problem: words,
concepts, and referents do not
correspond one-to-one
Sign
Unit
Sign
vehicle
Word in a Text
e.g., “innovation”
Referent
Innovation activity
in organization
• Hence: What is the researcher’s
object in the analysis? Cognitive
structures, factual information,
communication strategies or
something else?
Solutions to the semiotic problem:
–Interpretation of words through categorization (e.g., codes) and connection (e.g., grammar)
–Meaning can be inferred based on information within the analyzed text, e.g. based on the
proximity and grammatical position of words, or based on contextual information and
communication context, e.g. who created the text, for whom, when, why
Klaus Weber, Textual Analysis for Qualitative Research, Academy of Management Meeting 2005
3
Text Analysis In Qualitative Research Designs
Data Generation &
Collection
Data Storage &
Organization
Categorizing and
Connecting
Uses of Formal Text Analysis
• Text analysis can be used as a tool within a
larger research process, it is not an end of
itself (at least for non-linguists)
• Formal text analysis helps to :
–structure the analysis process (discipline)
–simplify the richness and complexity of data
–present patterns to oneself and to others
Coding and
Summarizing
Presentation and
Simplification
• Note: textual data may be generated through
many processes, e.g., interviews; recordings
from observation; archival documents, pictures
and video; responses to open ended surveys
Klaus Weber, Textual Analysis for Qualitative Research, Academy of Management Meeting 2005
4
Generic Types of Structured Textual Analysis
Content analysis
• Small text units in isolation, e.g. categories (yields e.g., category
schemes, frequencies, trends)
Semantic analysis
• Relationship between content units, e.g. associations and grammar
(yields e.g., scripts, networks of associated concepts, causal maps)
Narrative analysis
• Structure of larger text units, e.g. elements, turns, plots in a story (yields
more complex stories and rhetorical practices and beliefs)
Discourse analysis
• Several texts, e.g. broad regimes of interpretation (yields broad
ideologies, institutional myths and political contradictions)
Klaus Weber, Textual Analysis for Qualitative Research, Academy of Management Meeting 2005
5
Illustration: Corporate Cultural Repertoires
- Categorizing and Connecting Phrase
Phrase A1
…
Phrase An
Repertoire Element
Script
Regulation
Issues
Phrase B1
…
Phrase Bn
Technology change
Phrase C1
…
Phrase Cn
Uncertainty
Phrase D1
…
Phrase Dn
Attractive
Phrase E1
…
Phrase En
Form alliance
Phrase F1
…
Phrase Fn
Restructure
Text Unit
Connection
Interpretations
Action Response
Concept (+ Referent)
“Technological change…
… provides attractive
opportunities…
… which we capture
through new alliances.”
Grammar
Klaus Weber, Textual Analysis for Qualitative Research, Academy of Management Meeting 2005
6
Illustration: Corporate Cultural Repertoires
- Coding and Summarizing* Annual Report, Squibb Co., 1985
Codes assigned
The development of truly innovative, cost effective health care products is the major
mission of Squibb.
[means: (58) product development] [style: (71) efficient]
[domain: (10) product market]
[evaluation: (24) neutral importance] [aspiration: (40) vision]
In pursuit of this mission, we are committed to
superior excellence in science, product quality,
customer service and management - indeed, in
every aspect of our business.
[aspiration: (40) vision] [style: (69) commitment]
[aspiration: (36) comparative] [domain: (14) technology]
We believe that the evolution of Squibb over
these past few years, including the ever
improving productivity of our research and
development efforts, has positioned our
company to take advantage of the
opportunities provided by the present
economic environment.
[domain: (17) production] [actors: (08) customers]
[trend: (29) continuity] [means: (60) improve organization]
[means: (58) product development] [resources: (43) positioning]
[evaluation: (27) opportunity] [domain: (18) macroeconomics]
* Coding performed with custom dictionary in TextQuest software
Klaus Weber, Textual Analysis for Qualitative Research, Academy of Management Meeting 2005
7
Illustration: National Differences in Action Repertoires
- Presentation and Simplification Means of action: 1980
Means of action: 2001
restructure
optimize org
select people
restructure
invest
optimize org
expand geo
enter alliance
downsize
build capacity
develop people
develop product
sell product
acquire
select people
invest
expand geo
enter alliance
downsize
build capacity
develop people
develop product
sell product
strength. position
acquire
strength. position
Germany average
U.S. average
Klaus Weber, Textual Analysis for Qualitative Research, Academy of Management Meeting 2005
8
Some Basic Practical Decision Points
Conceptual level of analysis
• Words, concepts, actions?
Category schemes
• Inductive / emergent, custom standard, generic?
Sampling
• Code everything or selectively, and if less then what?
Coding unit
• Words, sentences, responses, documents, etc.?
Computer support
• Manual coding, computer-assisted, automated coding, data mining?
Klaus Weber, Textual Analysis for Qualitative Research, Academy of Management Meeting 2005
9
Software Support Options: Packages
Functionality
Functionality
Functionality
Functionality
Functionality
Type of
Software
Examples
of popular
software
Storage,
retrieval
Developing
and linking
categories
Automated
content
coding
Mapping,
display of
coded data
Quantification,
statistics
Theory
Building
Support
ATLAS.ti,
Ethonograph,
Kwalitan,
MaxQDA,
NUD*IST/NVivo
Yes
Yes
Some
Some
Little
(best for
smaller
volumes)
(main focus)
(best for
smaller
volumes)
(mostly basic)
(export to other
software)
Coding
Support
Diction,
TextQuest,
VbPro
Yes
Little
Yes
Little
Little
Mapping
AutoMap,
DecisionExplor
er
Some
TextAnalyst,
SAS plug-in,
WordStat,
TAKMI
Yes
Text
Mining
(main focus,
efficient for
high volume)
Little
Little
Yes
Yes
(especially for
large volumes)
Klaus Weber, Textual Analysis for Qualitative Research, Academy of Management Meeting 2005
(export of other
software)
Yes
Some
(main focus)
(e.g. concept
centralities)
Some
Some
(e.g. built in
algorithms)
10
Software Support – A Word of Caution
Key advantage:
• The ability to document choices and easily re-code data as category
schemes emerge
But, never underestimate the effort to prepare the data!
• Most programs only read digital files with proper punctuation
• Can use OCR and voice recognition software, but quality is often
poor, especially when documents or recordings are old
And, software tends to promise more than what you hope for
• Structuring data and standard analyses work fine, but that’s often not
what qualitative researchers are looking for
– "Both packages offer a variety of features that effectively help researchers run associations and
present results. However, in extracting themes from unstructured data, both packages were only
marginally helpful. The researcher still needs to read the data and make all the difficult
decisions.“ The American Statistician 2005: 59(1): 89-103 in a comparison of SAS Text Miner and
WordStat packages
Klaus Weber, Textual Analysis for Qualitative Research, Academy of Management Meeting 2005
11
Conclusion
Structured text analysis is a useful addition to qualitative
researchers’ toolkit
• Draws in many insights and techniques that have been developed
and used in other social science disciplines and humanities
• Leaves the choices about how structured the analysis process
becomes and how much the richness of data is reduced
But text analysis techniques cannot substitute for good
research questions and designs
• Does textual analysis actually help answer the research question?
• How should the data be interpreted, e.g. representing cognitive
structures, factual reports, institutional conventions, etc.?
• Think of the approach before the software support!
• Use contextual information to interpret the text!
Klaus Weber, Textual Analysis for Qualitative Research, Academy of Management Meeting 2005
12