Transcript Case Study Research

Developing & Testing Theory
with
Comparative Case Studies
Mauro F. Guillén
(In collaboration with Anuja Gupta)
Comparative Case Studies
• N larger than one:
– Going from one case to two cases makes a huge
difference.
• It can be one case, but analyzed longitudinally.
• Long tradition of comparison in the social
sciences, though not so much in management.
• Distinction between qualitative and quantitative is
overdrawn, oversimplified, and, to a very large
extent, fallacious.
Misconceptions about Comparative Case Studies
• They are good only for theory building, not for
testing theory.
• They cannot help establish cause-effect
relationships (internal validity).
• They are qualitative in nature.
• They do not comply with standard methodological
requirements.
• They have limited generalizability (external
validity).
• They are good for teaching but not for research.
• They are harder to get published.
Our Mission in Life
• Show that comparative case studies can be:
– Useful for theory testing as well as theory
building.
– Every bit as rigorous as other types of research.
• Show that every research project can benefit
from adopting a “comparative” approach or
design.
• Help management scholars learn from
sociologists, political scientists, and
anthropologists.
Goals of Comparisons
• Theory building (à la Eisenhardt).
• Apply a general model to explain empirical
instances.
• Use concepts to develop a meaningful empirical
interpretation (interpretive method).
• Analyze causal regularities.
• Identify continuities and discontinuities.
• Validation.
• Nota bene: Importance attached to understanding
historical processes and to establishing causeeffect relationships as they unfold over time.
Theda Skocpol, ed., Vision and Method in Historical Sociology
(New York: Cambridge University Press, 1984).
Types of Comparisons
• Structure of the comparison:
– Longitudinal comparison of a single case study.
– Matched comparisons (including natural experiments).
– Variation-finding comparisons (larger N, but not matched).
• Levels of analysis:
–
–
–
–
–
–
–
–
–
–
Individuals (OB literature).
Decision situations (Vaughn, Allison).
Teams or groups (Perlow).
Organizational subunit or process (Zbaracki).
Organizations (Kanter, Lawrence & Lorsch, Kogut & Zander).
Clusters or networks (Powell et al.).
Industries (Westney, Benner & Tushman).
Regions (Saxenian).
Countries (Chandler, Useem, Cole, Porter, Murmann).
Global systems (Wallerstein, Gereffi).
Case Selection
• Random assignment? Difficult.
• Theory driven:
–
–
–
–
–
Matched case comparisons and natural experiments.
Most-significant case(s) design.
Most-similar cases that diverge.
Variation finding.
Standardized case comparisons (Walton 1973):
• Based on a stratified set of relevant independent variables (e.g.
technology, size, culture, etc.).
• Collection of a broad range of data through systematic
procedures.
• Strike a balance between in-depth analysis and systematic
comparison.
• Convenience:
– Access.
– Language proficiency.
Three Classic Designs
Table 1: Bendix, Geertz and Dore Compare
Bendix 1956
Timing:
Inception
of industry
Bureaucratized
industry
Geertz 1963
Organization:
Purely
traditional
Firm-type
Dore 1973
Technology:
Social position of entrepreneurs and managers:
Autonomous class
Subordinate to government control
England
Russia
United States
East Germany
Individualist
Social structure:
Group-based (seka)
Javanese bazaar (pasar)
Balinese cooperatives
Javanese stores
Balinese business concerns
Market-oriented
Social structure:
Organization-oriented
Small-batch
English Electric’s Bradford plant
Hitachi’s Furusato plant
Mass
production
English Electric’s Liverpool
plant
Hitachi’s Taga plant
Mauro F. Guillén,
The Limits of Convergence
(Princeton University Press,
2004).
Issues
•
•
•
•
Internal validity (cause-effect).
External validity (generalization).
Better than single case studies.
Comparison enables greater breadth without
losing too much depth.
• Excellent for theory development.
• Also good for testing theory (see Ragin).
Basic Analytical Idea
• J. S. Mill’s (1843) methods of experimental
inquiry:
– Method of difference:
A+B+C = Y.
A+B = not Y.
Therefore: A and B are necessary but not sufficient, while C is
necessary but not sufficient.
– Method of agreement:
A+B = Y.
A+C+D = Y.
B+C = not Y.
Therefore: A is necessary but not sufficient.
Fuzzy-Set Social Science
• Treat cases as configurations of different
values for variables of interest.
• Define an outcome of interest.
• Use Boolean algorithms to arrive at
generalizations about cause-effect
relationships (i.e. configurations that bring
about the outcome): necessary & sufficient
conditions.
• Calculate how confident you can be about the
generalizations.
Charles C. Ragin, Fuzzy-Set Social Science (University of Chicago Press, 2000).
An Example of a Fuzzy-Set
Case #
1
2
3
4
5
…
N
X1
0
0
0
0
1
X2
1
1
1
1
0
X3
1
1
1
0
0
Y
1
1
0
0
0
Probabilistic Criteria I
• When N>30, use a z test:
z = [(P – p) – 1/2N] / [p(1 – p)/N]1/2
N is the # of cases displaying the combination.
P is the observed proportion of the outcome.
p is the benchmark proportion:
.50 or “more often than not sufficient”
.65 or “usually sufficient”
.80 or “almost always sufficient”
Probabilistic Criteria II
• When N<30, use a binomial probability test:
b = [N!/(N-r)!] pr (1 – p)N-r
N is the # of cases displaying the combination.
r is the # of cases displaying the outcome.
p is the benchmark proportion:
.50 or “more often than not sufficient”
.65 or “usually sufficient”
.80 or “almost always sufficient”
Combining Methods
• Move beyond the qualitative/quantitative distinction.
• Comparative case studies can be every bit as rigorous as
any other kind of research.
• Use a comparative design as the overall structure of your
study.
• Within each case, collect large N data, conduct
experiments, gather archival data, interview people in
depth, organize focused groups, conduct an ethnography,
etc.
• Draw conclusions from the overall comparative scheme,
examining if:
– Outcome & process hold across cases.
– Outcome but not process is the same.
– Different outcomes.
Some Basic References
•
•
•
•
•
•
•
•
•
•
•
Burawoy, Michael. 1998. “The Extended Case Method.” Sociological Theory 16(1):4-33.
Dibble, Vernon K. 1963. “Four types of inference from documents to events.” History
and Theory 3(2):203-221.
Eisenhardt, K. M. 1989. “Building Theories from Case Study Research.” Academy of
Management Review 14(4):532-550.
Mill, John Stuart. [1881] 1950. “Of the Four Methods of Experimental Inquiry.” In
Philosophy of Scientific Method. Edited, with an introduction by Ernest Nagel. New
York: Hafner, 211-233.
Ragin, Charles C. 2000. Fuzzy-Set Social Science. Chicago: University of Chicago Press.
Skocpol, Theda ed. 1984. Vision and Method in Historical Sociology. New York:
Cambridge University Press.
Smelser, Neil J. 1976. Comparative Methods in the Social Sciences. Englewood Cliffs,
New Jersey: Prentice-Hall.
Stinchcombe, Arthur L. [1968] 1987. Constructing Social Theories. Chicago: The
University of Chicago Press.
Tilly, Charles. 1984. Big Structures, Large Processes, Huge Comparisons. New York:
Russel Sage Foundation.
Walton, John. 1973. "Standardized Case Comparison: Observations on Method in
Comparative Sociology." In Michael Armer and Allen D. Grimshaw, eds., Comparative
Social Research: Methodological Problems and Strategies. New York: John Wiley &
Sons, pp. 173-191.
Yin, R.K. 1994. Case Study Research. Thousand Oaks, CA: Sage.
Multiplicity of forms
One
Share of all
instances
Single
Multiple
Individualizing
Encompassing
Contrast specific instances
of a given phenomenon as a
means of grasping the
peculiarities of each case.
[Bendix; Dore; Westney;
Saxenian]
Different instances in a system
are compared so as to explain
their characteristics as a
function of their varying
relationships to the system.
[Wallerstein; Gereffi]
Universalizing Variation finding
All
Seeks to establish that
every instance of a
phenomenon follows
essentially the same rule.
[Rostow; Chandler; Kerr et
al.]
Aims at establishing a
principle of variation in the
character or intensity of a
phenomenon by examining
systematic differences among
instances.
[Moore; Gerschenkron; Porter]
Charles Tilly, Big Structures, Large Processes, Huge Comparisons
(New York: Russell Sage Foundation, 1984), p. 81-83.
Models of Management
_____________________________________________________________________
Dominant Pattern of Politics:
Liberal Democracy
Corporatism
ModernistTechnocratic
United States
Germany
Characteristic
Mode of
Thought:
TraditionalGreat Britain
Spain
Humanist
_____________________________________________________________________
Mauro F. Guillén, Models of Management (University of Chicago Press, 1994).
The Limits of Convergence
Table 8.1: Development Paths and Organizational Change in Argentina, South Korea, and Spain
Level of
Inward
Flows:
High
(Pragmatic)
Level of Outward Flows:
High
(Modernizing)
Low
(Populist)
SPAIN
Pragmatic-Modernizing:
HIGH: Allow Imports & Inward investment
HIGH: Export-led growth & Outward investment
ARGENTINA
Pragmatic-Populist:
HIGH: Allow Imports & Inward investment
LOW: Import substitution & Local investment
Organizational forms:
Foreign MNEs; SMEs; Large oligopolistic firms
Organizational forms:
Foreign MNEs; Business Groups; some SMEs
Organized labor’s ideologies:
Pro-globalization, MNEs = partners
Organized labor’s ideologies:
Anti-globalization, MNEs = necessary evils
4 1
3 2
Low
(Nationalist)
SOUTH KOREA
Nationalist-Modernizing:
LOW: Protectionism & Local ownership
HIGH: Export-led growth & Outward investment
INITIAL CONDITIONS, CIRCA 1950
Nationalist-Populist:
LOW: Protectionism & Local ownership
LOW: Import substitution & Local investment
Organizational forms:
Business groups; State-owned enterprises
Organizational forms:
State-owned enterprises; Large oligopolistic firms
Organized labor’s ideologies:
Pro-globalization, MNEs = arm’s length
collaborators
Organized labor’s ideologies:
Anti-globalization, MNEs = villains
Note: MNEs stands for multinational enterprises; SMEs stands for small and medium enterprises.
The Rise of Modernist Architecture
Table 7.1: A Cross-National Comparison of the Rise of Machine-Age Modernist Architecture
Variable:
Industrialization
Sociopolitical
upheaval
Class dynamics
(rise of the workerconsumer & mass
consumption)
New sponsors
(industrial firms &
the state)
Professionalization
of architecture
linked to
engineering
Strength of the
scientific
management
movement
Outcome:
Modernist
architecture
Britain
Pioneer of the
Industrial
Revolution,
though not of
mass
production.
No: Victorian
and
Edwardian
stability.
Early, which
thwarted
modernism in
favor of
kitsch.
Did not play a
role.
No: Weak
link; belated
adoption of
Beaux-Arts.
Weak until
1940s.
Late and
timid: 1940s.
France
Early
industrial
growth.
Germany
Late, but
successful
industrialization.
Italy
Late
industrialization in the
North.
Russia
Relative
laggard,
except in
certain
locations.
Spain
Laggard.
Mexico
Laggard.
Brazil
Laggard.
Argentina
Laggard.
U.S.A.
Most advanced
country in mass
production.
No:
Relative
stability
since the
1870
Commune.
Relatively
early: 1920s.
Yes:
Defeat,
revolution
(1918), and
inflation.
Yes:
Fascist
Revolution of
1922.
Yes:
Bolshevik
Revolution of
1917.
Yes:
Social
revolution,
1936-39.
Yes:
Vargas’s
Estado
Novo of
1930.
No:
Haphazard
transition to
mass politics,
1916-46.
No:
Constitutional
stability;
neutralized labor
unrest.
Late: 1930s
and 50s.
Late: 1950s.
Late: 1950s.
Late:
1960s.
Yes:
Revolution of
1910-17, and
socially
progressive
state.
Late: 1950s.
Late:
1950s.
Late: 1950s.
Relatively early,
which thwarted
modernism in
favor of kitsch.
Played a role,
but belatedly.
Strong role.
Strong role.
Strong role.
Weak role until
1930s.
Yes: Strong
and longstanding link
to
engineering.
Strong.
Yes: Strong
and longstanding link
to
engineering.
Strong.
Yes: Linked
to
engineering.
Strong role of Strong role
the state, not
of the state.
of firms.
Yes: Initially based on
Beaux-Arts, but influenced
by engineering since late
1920s or early 1930s.
Weak role.
No: Clash
between
Beaux-Arts
& technical
traditions.
Intermediate.
Weak role,
except late
1930s.
No: Strong
Beaux-Arts
influence.
No: Strong
Beaux-Arts
influence.
Yes initially, but
Beaux-Arts
influence,
1910s-20s.
Strong.
Weak until
1940s.
Intermediate.
Strong.
Weak.
Strong: pioneer.
Early
beginnings,
but late
consolidation.
Early pioneer;
heyday in the
1920s.
Early pioneer;
heyday in the
1930s.
Early pioneer;
heyday in the
1920s.
Late and
short-lived:
1933-39.
Relatively
early for its
region: late
1920s.
Relatively
early for its
region:
1930s.
Late, timid,
and shortlived: late
1930s.
Early pioneer,
but late heyday
in the 1930s.
Mauro F. Guillén, The Taylorized Beauty of the Mechanical (Princeton University Press, forthcoming).