From Theory to Evidence: The Comparative Method

From Theory to Evidence:
The Comparative Method
7 March, 2008
Is all social research comparative?
• We seek to describe, understand, explain
• Can this be done with a single case?
Research Design(s)
• The Case Study (one shot)
– examine phenomena at a single unit or event.
individual, state, society, culture, a conflict, a war.
– technically not a design because it does not
involve a comparison
• Comparative method
– examine variation in some phenomena as it occurs
across multiple units or events
Case Study
• Purpose is to describe a unit, not to test hypotheses
• inductive: start w/ observations, learn from them, and
generate theory or reach “understanding”
• Assume each case (nation) is unique
• many, many things may affect event/act (lots of
variables); all things have individual uniqueness
• interpretive approach: observer does rich observation,
think description of complex social relations, social
systems
• emphasize accuracy, nuance, uniqueness of case
Problems with single case study
•
•
•
•
•
•
•
No variation (no comparison)
non-generalizable
non-falsifiable (?)
non-causal
selection on the dependent variable
rare cases examined, not non events
non-cumulative
Comparative Method
• Observing and comparing carefully selected
cases on the basis of some stimulus being
absent or present
• Operates on the same logic as the
experimental design
• Limitation is that the ability to control the
political environment is so limited so causal
inferences are more difficult.
Approach
• goal is to make generalizations across many
cases
• Requires that concepts be simplified,
stretched
• Parsimony allows for generalization
• Take complex social systems and re-name
them “variables”
Problems with Big “N” comparative
method
• complex social phenomena hard to measure
• Culture
• A focus on the easily measured things can lead
researcher to emphasize the importance of trivial
forces
• Can lead to emphasis on questions that are trivial
• parsimony over accuracy
• Poor data
• “empirical, but disengaged”
Problems with small “N”
• too many variables, not enough cases, ie.
"degrees of freedom problem”
• Countries as the unit of analysis
• Often a large potential set of countries is
reduced by missing data
• Remaining cases are not representative
• Limited variety that imposes constraints on
rigor
Solutions
• Increase the number of cases as much as
possible
• Focus the comparative analysis on
“comparable cases”
• Focus on the “key” variables, reduce the
number of explanations
Example: Social Revolutions
• When, where, why? (T. Scotpol)
• Define: Violent (?) and/or rapid (?) change in
social order, internally generated
• Measurement issues
– How define it? civil war vs revolution
– political vs. social
Causal Factors
• working class consciousness (Marx), false
consciousness
• peasant solidarity
• weak state structure,
• illegitimate institutions
• fiscal crisis of state
• external military threat
• economic depression
• relative deprivation
How many cases?
•
•
•
•
•
Iran 1979
Russia 1918
France 1780s
China 1940s
US, Mexico, Latin America (political, but not
social revolutions)?
Problems
• too many variables, not enough cases, ie.
"degrees of freedom problem”
• How study where events did not happen?
• Places that had no revolutions tell us lots
about places that did have them
Comparable case analysis
• China 1910s (or Japan 1910s) to China 1940s
• Germany 1840s to France 1780s
• Iran 1979 to Iraq 1979
Voter turnout in cross national context
• Why is turnout higher in some places?
• Theory: rationalism and institutionalism
• institutions affect incentives for parties to
mobilize voters
• people respond to changes in context of
election
• “costs” of voting lower under some conditions
• “benefits” greater under some conditions
Causes
• Hypotheses:
•
PR vs. FPTP -> turnout
•
constituency level closeness -> turnout
• Other factors:
•
federalism
•
# of parties
•
frequency of elections
•
compulsory voting
•
weather
•
aggregate levels of wealth, education
How many cases
•
•
•
•
•
•
what’s fair to compare?
Established democracies
Fair elections
Just Europe, Asia, Africa, NA, etc.
w/ Africa, corruption = higher TO
w/o Africa, corruption = lower TO
Measurement issues
• PR -> many different forms (MMP, STV, closed
list) - binary?
• parties -> 2 party system, 3 party system, “multi”
• (what is UK? 2 party system, 3?)
• Can effect of PR be teased out from effect of # of
parties?
• Where is PR located (if cases largely European)?
– Scandinavia - a ‘consensual’ political culture
• Where is FPTP located?
– UK, US, Can, NZ (pre 96), Oz, Caribbean, India
How test for cultural effect?
• Hold culture constant: Change institutions in
one, two, N nations (NZ)
• Find variation in institutions where culture is
constant , ie. US south
Economic growth in comparative
perspective
• Why higher in some places?
• Theory: Social capital, institutions
• Hypotheses:
– Nations w/ more social capital -> growth
– Nations w/ certain institutions -> growth
autonomous central banks
Other factors
• colonial history
• social networks (trust -> trade)
• corporatism (left govt * union structure =
growth)
How many cases? what’s fair to
compare?
• All nations?
• OECD, just Europe, just Asia?
• NICs?
The problem of case selection
• Asia = Taiwan, China, Singapore, Japan, Malaysia,
Indonesia, Burma, Thailand, PNG, Laos,
Cambodia, Viet man, India
• we compare within Asia? Can we compare Asia
to Africa? to the Americas? to Europe?
• Europe = “old” vs. new” (post communist)
• The Norway issue: Hicks and Swank, Lange and
Garrett
– years of Left party control of government
– union structure / penetration / % of workforce
– ability to set top down agreements on wage hikes
Measurement issues
• Economic growth
• Confucian tradition
• Protestant work ethic = % Protestant?
Emphasis on Rationality and
institutionalism
• How do political & econ. institutions affect &
reflect behavior
• rationalism = individuals act consistent w/ their
preferences
• ‘utility’ maximizing behavior
• try to get more for less
• incentives & constraints (institutions resources)
• strategic interactions