Case Studies,
Historical Analysis,
& Process Tracing
GLOA 720, Method Recap (1)
Any case study is a comparative analysis.
Even w/o a fancy case
to compare yours with.
Even w/o a fancy case
to compare yours with.
Because you
can always
compare to
yourself.
The job is not just to describe.
We ain't no reporters.
The job is not just to describe.
We ain't no reporters.
It is to analyze. In essence, we are doing:
The job is not just to describe.
We ain't no reporters.
It is to analyze. In essence, we are doing:
- within-case comparison: sub-units or over-time
- across-case: [apples vs. apples] or [apples vs. oranges]
- Or against a normal group
Things to care about: Structured and Focused comparison
Things to care about: Structured and Focused comparison
Focused
not all, but only select aspects of each case: consult the literature
discussions stay within the purview of the variables identified
Things to care about: Structured and Focused comparison
Focused
not all, but only select aspects of each case: consult the literature
discussions stay within the purview of the variables identified
Structured
Units at the same level (or category)
Equal treatment: all variables for each case.
Don't exclude the ones that don't fit the story. E.g., Skocpol 1979
An example of selection bias in case study:
Skocpol (1979) State and Social Revolution
- the idea: external military threat > elite split > chances for revolution
An example of selection bias in case study:
Skocpol (1979) State and Social Revolution
- the idea: external military threat > elite split > chances for revolution
Skocpol (1979), simplified
| revolution | no revolution | |
|---|---|---|
| elite split | Russia; France; China (20c) | |
| no elite split | Britain; Germany; China (19c) |
An example of selection bias in case study:
Skocpol (1979) State and Social Revolution
- the idea: external military threat > elite split > chances for revolution
Skocpol (1979), simplified
| revolution | no revolution | |
|---|---|---|
| elite split | Russia; France; China (20c) | |
| no elite split | Britain; Germany; China (19c) |
Geddes (1990)
Actual Implementation
Actual Implementation
Soaking and Poking: case immersion
- exploratory, but goes on detail
- literature review should be semi-established before this for 'focus'
Actual Implementation
Soaking and Poking: case immersion
- exploratory, but goes on detail
- literature review should be semi-established before this for 'focus'
Identifying the Variables
- Independent, Dependent, and Control Variables
- Operationalization: what would you mean by 'stronger'; 'increase' ?
- particularly important for qualitative research: 'structure'
Actual Implementation
Soaking and Poking: case immersion
- exploratory, but goes on detail
- literature review should be semi-established before this for 'focus'
Identifying the Variables
- Independent, Dependent, and Control Variables
- Operationalization: what would you mean by 'stronger'; 'increase' ?
- particularly important for qualitative research: 'structure'
explanation
- developing the narrative/aruement
- describe the relationship between the IV and DV
- with a structure, description = analysis
Actual Implementation
Soaking and Poking: case immersion
- exploratory, but goes on detail
- literature review should be semi-established before this for 'focus'
Identifying the Variables
- Independent, Dependent, and Control Variables
- Operationalization: what would you mean by 'stronger'; 'increase' ?
- particularly important for qualitative research: 'structure'
explanation
- developing the narrative/aruement
- describe the relationship between the IV and DV
- with a structure, description = analysis
Robustness (optional?)
- demonstrate the alternative explanations don't work
- could be just a good explanation of the control variables
Structured Designes (Mill's methods)
Structured Designes (Mill's methods)
| Most Similar | Independent | Control 1 | Control 2 | outcome |
|---|---|---|---|---|
| case A | ||||
| case B |
Structured Designes (Mill's methods)
| Most Similar | Independent | Control 1 | Control 2 | outcome |
|---|---|---|---|---|
| case A | ||||
| case B |
| Most Different | Independent | Control 1 | Control 2 | outcome |
|---|---|---|---|---|
| case A | ||||
| case B |
Structured Designes (Mill's methods)
| Most Similar | Independent | Control 1 | Control 2 | outcome |
|---|---|---|---|---|
| case A | ||||
| case B |
| Most Different | Independent | Control 1 | Control 2 | outcome |
|---|---|---|---|---|
| case A | ||||
| case B |
Structured and Focused?
An example (Pepinski 2009)
| Malaysia | Indonesia | |
|---|---|---|
| region | Southeast Asia | |
| economy | middle-income | |
| religion | Islam | |
| political coalition | native elites + native business |
native elites + Chinese business |
| post-1997 crisis | authoritarian resilience |
authoritarian breakdown |
But the actual writing
isn't this neat & simple.
But the actual writing
isn't this neat & simple.
Need to go step-by-step
using another example.
Give the general context
Give the general context
- basic background
Give the general context
- basic background
- some of the factors
of interest
Give the general context
- basic background
- some of the factors
of interest
- overview of the
empirical approach
Literature Review, of course
Literature Review, of course
Argument and
Variable Description
- explain clearly
what to expect if
the arugment holds
Now, finally, the empirical analysis: the three cases
Now, finally, the empirical analysis: the three cases
Now, finally, the empirical analysis: the three cases
Alternative Explanation
(robustness checek)
Alternative Explanation
(robustness checek)
- what are some plausible
explanations that can
screw up my finding?
- can my arguement couter them?
Comparative Historical Analysis
"By means of comparative analysis,
I want to preserve a sense of
historical particularity
as far as I can, while still
comparing different counts"
-- Reinhard Bendix (1976)
Comparative Historical Analysis: Case Study, but over time.
- Sequence matters.
- often, though not always, used for causal inference (hence the discussions about necessary and sufficient causes in Mahoney (2004))
Comparative Historical Analysis: Case Study, but over time.
- Sequence matters.
- often, though not always, used for causal inference (hence the discussions about necessary and sufficient causes in Mahoney (2004))
- methods
variants of Mill's methods, but with sequence; often descriptive and typological
process tracing: a complete anatomy of the intervening process
- sequence arguments: good at showing how; less of causal inference
others: formal theory and fuzzy sets: not very recommendable
Process Tracing
- Make sure all steps count.
An easy example: Ziblatt (2009), the second half
An easy example: Ziblatt (2009), the second half
An easy example: Ziblatt (2009), the second half
- Late 19C Germany; Hypothesis: high land inequality electoral fraud
An easy example: Ziblatt (2009), the second half
- Late 19C Germany; Hypothesis: high land inequality electoral fraud
- existence of strong landed elites
An easy example: Ziblatt (2009), the second half
- Late 19C Germany; Hypothesis: high land inequality electoral fraud
existence of strong landed elites
the adoption of universal male suffrage in Germany
An easy example: Ziblatt (2009), the second half
- Late 19C Germany; Hypothesis: high land inequality electoral fraud
existence of strong landed elites
the adoption of universal male suffrage in Germany
landed elites social power erodes
An easy example: Ziblatt (2009), the second half
- Late 19C Germany; Hypothesis: high land inequality electoral fraud
existence of strong landed elites
the adoption of universal male suffrage in Germany
landed elites social power erodes
landed elites pressure local officials (appointment; revenue)
An easy example: Ziblatt (2009), the second half
- Late 19C Germany; Hypothesis: high land inequality electoral fraud
existence of strong landed elites
the adoption of universal male suffrage in Germany
landed elites social power erodes
landed elites pressure local officials (appointment; revenue)
local officials remove voters from voter rolls
An easy example: Ziblatt (2009), the second half
- Late 19C Germany; Hypothesis: high land inequality electoral fraud
existence of strong landed elites
the adoption of universal male suffrage in Germany
landed elites social power erodes
landed elites pressure local officials (appointment; revenue)
local officials remove voters from voter rolls
systemic electoral fraud observed
An easy example: Ziblatt (2009), the second half
- Late 19C Germany; Hypothesis: high land inequality electoral fraud
existence of strong landed elites
the adoption of universal male suffrage in Germany
landed elites social power erodes
landed elites pressure local officials (appointment; revenue)
local officials remove voters from voter rolls
systemic electoral fraud observed