Democracy → Statistics Section

 Question: What distinguishes democracies from nondemocracies?

o Best answered with cross-sectional statistical analysis.
 Przeworski, Alvarez, Cheibub & Limongi (2000):
o Built a dataset of 141 countries, annually from 1950–1990.
o Coded regimes as democracies (dichotomous variable).
o Used also Coppedge-Reinicke scale → got similar results.
 Their findings (linked to Lipset 1959):
o Strong relationship between economic development and democracy.
o But → modernization is not the cause of democracy.
o Instead, democracy survives more successfully once economic growth is attained.
o Poor democracies are more likely to fall into dictatorship.
o Dictatorships can survive for years even in wealthy countries.
o Many dictatorships fell in poor countries.
 Probability of survival:
o If per capita income > $7000 → zero probability of democracy collapse.
o If per capita income < $1000 → 12% chance of collapse.
o Argentina collapsed at $6055 (highest in database).
o 3 of 4 collapses at >$4000 were in Argentina, 1 in Uruguay.
o Per capita income is most powerful predictor.
o Correctly predicted 77.5% of 4126 annual observations with income alone.
 Additional findings:
o Democracies survive more successfully under economic growth.
o Dictatorships fail equally under both growth and decline.
 Other scholars (Linz & Stepan):
o Emphasize institutions, not just GDP.
o High GDP favorable, but not enough.
o Five necessary conditions for democracy survival:
1. Vibrant civil society
2. Autonomous political society
3. Rule of law
4. Usable state
5. Institutions for an economic society
o Seven independent variables that help predict consolidation:

 State–nation relationship
 Type of prior regime
 Leadership base of prior regime
 Pattern of transition
 Legitimacy of institutional actors
 Environment of constitution drafting
 (others implied under nominal categories)
 Political culture arguments:
o Lipset (1994): legitimacy needs supportive culture.
 o Diamond (1999): regime legitimacy, corruption, behavior of representatives
matter.
 People’s support for democracy depends less on economics, more on
institutional behavior.
 Legitimation requires:
 All actors see democracy as appropriate.
 All competitors see democracy as “the only game in town.”
 Political variables affect survey support for democracy more than
economic ones.
o Inglehart (1990): half of variance in democratic persistence explained by political
culture.
 Challenges to political culture argument:
o Muller & Seligson (1994):
 Used Inglehart’s data + Latin America.
 Found interpersonal trust is a result of democracy, not cause.
 Only variable with independent effect: population favoring moderate
reform.
 Other factors examined by Przeworski et al:
o Duration of democracy (not significant once economic controls added).
o Cultural factors (religion) not significant.
o Ethnic fractionalization adds little predictive power.
o Education adds predictive power, even controlling for income.
o Parliamentary vs. presidential systems:
 Suggestive but not conclusive.
 28% of parliamentary regimes died vs. 54% of presidential regimes.
 But not confident in robustness (poor parliamentary regimes poorer than
poor presidential).
 Limits of Przeworski et al:
o Political system variables not well specified (parliamentary/presidential too
crude).
o Need better measures:
 Cox (1997): electoral systems differ in coordination failures.
 Tsebelis (1995): “veto points” better capture minority protection.
 Niou & Ordeshook: integrated vs. bargaining systems.
o Did not test:
 Protection of property rights
 Rule of law
 Corruption
 Civil society strength/density
 Military institutional power
 Opportunities for new research:
o Link new institutional variables to democratic consolidation.
o Treisman: use comparative corruption data.
o Collect systematic civil society data (Putnam 1993, Schmitter 1997).
o Add survey data on legitimacy and culture to Przeworski dataset.
  Muller & Seligson suggest “favoring moderate reform” is promising
variable.
 Final limitation of cross-sectional data:
o Hard to interpret causally.
o Cannot narrate progression from democracy to dictatorship in terms of abstract
variables like per capita income.
o Need theory + case narratives (Linz & Stepan).

Narratives (Democracy)

 Key questions:
o What pushes some countries at certain times into democracy?
o How do new democracies survive crises?
o These questions → need historical/narrative analysis (sensitive to change over
time).
 Limits of statistical approaches:
o Przeworski et al’s dataset allows some diachronic analysis.
o Linz & Stepan are attentive to sequencing.
o But both focus less on who is acting and their social class positions.
 Return to Barrington Moore tradition:
o Studies on historical role of social classes in democracy’s rise/fall.
o Classic: Moore (1966) → bourgeoisie = source of modern democracy.
 Bourgeois strength = democracy’s survival in interwar period.
 Weak bourgeois + landed classes alliance = fascism.
 Weak bourgeois + mobilized peasants by leftists = communism.
o Many studies since then have refined/challenged Moore.

Luebbert (1991):

 Focus: survival of democracies during interwar depression.

 Like Moore: middle classes = key to democratic strength.
 BUT → rejects “marriage of iron and rye” (bourgeoisie + landed elite) as cause of
fascism.
 Findings:
o Rural support for fascism didn’t require landed elites.
o In Germany, Spain, Italy → fascism strong where family peasants predominated.
o Only in southern Italy did landed elites matter, but they initially sided with
liberals.
 Fascism gained elite support after Mussolini took power.
 Democracies survived in Britain, France, Switzerland, Belgium, Netherlands because:
o Middle classes not divided by religion, language, region, or urban–rural splits.
o United middle class wasn’t afraid of workers → slowly integrated them.
 o Workers got dignity and inclusion, even if not maximum material gains.
 Where middle classes divided by preindustrial cleavages:
o They feared alliance with workers.
o Workers built strong unions instead.
o After WWI: united workers + divided liberals = strong worker demands.
o Long-term peace required subordination of markets to politics.
 Alliances:
o Social democracy = urban workers + middle peasants/family farms (e.g., Norway,
Sweden, Denmark, Czechoslovakia).
o Fascism = middle classes + family peasants, triggered when socialists tried to
mobilize agrarian proletariat (Germany, Italy, Spain).
o Irony: workers fighting for rural poor often drove peasants toward
middle-class/fascist alliances.

Rueschemeyer, Stephens & Stephens (1992):

 Large historical study (Western Europe, Latin America, Caribbean).

 Challenge to Moore:
o Bourgeoisie not consistent driver of democracy.
o Middle classes → ambivalent once included.
 Instead: working class plays decisive role.
o Unlike peasants, can organize politically.
o Helps create balance of power (no single group hegemonic).
 Sequence: capitalist development → changes in class structure → strengthens working +
middle classes, weakens landed elite → conflicts → eventually democracy.
 Initial assumption: working class = consistently pro-democracy.
o But refined:
 Only true if party system protects upper-class interests.
 Otherwise, upper classes resist democracy.

Collier (1999): Paths Toward Democracy:

 Examines 19th + 20th century cases, Western Europe + Latin America.

 Identifies 7 paths to democracy.
 Labor plays a role in 4 of them.
 Provides evidence of labor’s role across history.
 Problem:
o Only studies successful democracies.
o Doesn’t code labor mobilization for unsuccessful attempts.
o Risk of selection bias:
 Maybe stronger labor mobilization reduces chances of democracy (right
resists more strongly).
o Needs fuller dataset to confirm or refute.
General Assessment of Narratives vs. Statistics:

 Moore-tradition narratives = rich, detailed histories of democracy and reversals.

 But findings often conflict with statistical correlations.
 Rueschemeyer et al:
o Many different regional sequences can lead to democracy.
o Development-democracy correlation may be “fortuitous.”
o Underlying common factor = balance of power between classes & between civil
society vs. state.
o But balance can be reached in many ways.
 Tension:
o Luebbert → narrowly focused, historically circumscribed.
o Rueschemeyer/Collier → broad, many paths, harder to generalize.
 Problem:
o Either findings limited to a small set of cases, or too many possibilities → weak
predictive power.
o Need better-specified variables (that can apply beyond original cases).
 Until then: narrative approaches won’t fully challenge statistical findings.
 Both statistical and narrative work have created new problems for theory (third element
of tripartite method).

Theory (Democracy)

 Recent democratic theory has focused on microdynamics of stability:

o Przeworski (1991): Why do actors out of power not rebel against democratic
rulers?
o Weingast (1997): Why do democratic rulers not confiscate property/voting rights
from enemies to stay in power?
 Przeworski → stability against revolution from below.
 Weingast → stability against revolution from above.
 Gap between theory and empirical findings:

1. Economic wealth and democracy:

 Empirical data: per capita income strongly linked to democracy.
 Theoretical models → missing this link.
 Should include parameter for income showing instability when income is
low.
2. Institutional forms:
 Empirical: parliamentary more stable than presidential.
 Theoretical gap: Why? What exact features explain it (e.g., veto points)?
 Theorists should also address how institutions are chosen (Geddes 1996 on
Latin America/Eastern Europe).
 3. Class mobilization:
 In societies with workers, middle class, peasants, landed elites → under
what conditions does worker mobilization bring democracy?
 Luebbert’s idea (bourgeoisie + workers compromise) matches Przeworski
& Wallerstein (1982).
 But this can’t be the only democratic equilibrium → other Moore-tradition
patterns should be modeled.
 Defining democracy:

o Problem: Definitions differ across contexts.

 Shaffer (1998): Senegalese “democracy” = different meaning than in
West.
 Collier & Levitsky (1997): Even political scientists lack a universal
standard.
 Threat: without a single standard, large-N statistical comparisons are
shaky.
 Alternative approaches to defining democracy:
o Bollen (1993): Defines “liberal democracy,” uses structural equations +
confirmatory factor analysis to fix measurement biases.
o O’Donnell (1997): Real democracy requires informal institutions that block
favoritism/particularism.
o Coppedge & Reinicke (1990): Scaling technique to capture democracy’s depth.
 Przeworski et al’s minimalist definition:
o Democracy = only two conditions:

1. Contestation: Opposition has real chance of winning.

2. Autonomy: Winners actually rule.
o They argue for minimalism because:

 Adding too many “good things” (e.g., equality, accountability, welfare)

makes it impossible to isolate relationships.
 Example: Dahl’s polyarchy definition mixes multiple features, making
analysis difficult.
 Benefits of disaggregation:
o Lets us study how variables interact separately (e.g., contestation + education, or
contestation + responsiveness).
o Allowed Przeworski et al to find surprising results:
 Dictatorships can grow fast or slow.
 On average, democracies and dictatorships → no significant difference in
economic performance (after controlling conditions).
 Implications:
o For data collection: Better, less biased data needed (not just statistical
corrections).
o For theory: Models must explain how factors like elections, median voter
influence, information, and equality fit together.
 o Right now, theory assumes they cohere but hasn’t shown under what conditions
they do.
 Conclusion:
o Past decade = strong empirical progress (stats + cases).
o If these empirical findings re-stimulate democratic theory → big achievement.

Order — Statistics

 After WWII, comparative politics largely ignored Hobbesian disorder; focus turned to
who gets what, when, how.
 Earlier work on revolutions (Skocpol → Goldstone) treated revolutions as too few for
statistical methods; they used historical-comparative narrative.
 Late 1980s events (collapse of second-world states; collapses in third-world states)
brought “order” back into focus.
 Two global trends became clear: inter-state war probability decreased; civil-war
probability increased.
 Civil wars rose partly because many became interminable, while interstate wars more
often ended in negotiated settlements.
 Re-specifying the dependent variable as a state’s ability to withstand collapse/insurgency
produced enough cases for statistical work.
 New datasets (Minorities at Risk; State Failure) let scholars statistically sort polities by
rebellion risk (Gurr 2000; Collier & Hoeffler 2000; Fearon & Laitin 1999).
 Theoretical work began identifying causal mechanisms behind statistical findings, often
building on Horowitz (1985) on ethnic conflict.
 Case-based narrative work (e.g., Kalyvas, Varshney) uses theory and statistics to clarify
mechanisms.

Order — Narratives

 Case studies analyze two related problems: collapse of state authority; eruption of ethnic
violence/civil war.
 Causes of Soviet collapse debated among specialists: national consciousness (Suny),
nationalist mobilization (Beissinger), Leninist institutional sclerosis (Roeder), loss of
will/intelligentsia (Hough), agents’ loss of confidence and asset-grabbing (Solnick).
 Lohmann’s work on informational cascades parallels narratives of rapid regime
breakdown (East Germany).
 These narratives suggest political order is an equilibrium of coordinated expectations;
even institutionalized polities can unravel quickly.
 African state collapse literature shifted toward “shadow state” concepts (Bayart,
Mamdani, Reno): informal networks (officials, chiefs, thugs, foreign patrons, traders)
exercise real authority; formal state is often a façade.
 Rebel resources include foreign patrons (e.g., Qaddafi), aid/NGO flows, smuggled
commodities; colonially thin state structures made collapse easier (Ellis on Liberia).
  Ethnic-violence narratives (Brass, Kalyvas): “riot professionals” and alliances between
urban ideologues and local actors can escalate local incidents into large communal
violence; ethnic and ideological civil wars can show similar dynamics.
 Bunce compares USSR/Yugoslavia/Czechoslovakia: military interests and whether
dominant groups had pre-existing institutions matter for violence level in breakup.
 Varshney: civic engagement patterns (cross-community associations) help prevent
communal violence in some Indian cities.

Order — Theory

 IR theorists applied security-dilemma logic to post-Soviet violence (Posen; Walter &

Snyder), explaining violence via overestimation of others’ weakness and windows of
opportunity.
 Fearon (1998) challenged the security-dilemma mechanism and proposed commitment-
problem logic: ruling factions cannot credibly commit to protect losing minorities after
independence, so minorities rebel early.
 Kuran (1998) proposed cascade/tipping dynamics for rapid rise in group solidarity;
Snyder (1993) points to state weakening as a cascade trigger.
 Collier & Hoeffler (2000) modeled rebellion as akin to organized crime: success depends
on access to primary-product rents, recruitable looters, and weak state forces.
 Fearon & Laitin (1999) modeled insurgency as young men choosing between the
legitimate economy and rebellion while the state allocates counter-insurgency resources;
their model predicts country-level GNP, primary resources, and geographic concentration
(esp. in mountains) are strong rebellion predictors.
 These models (Collier & Hoeffler; Fearon & Laitin) suggest economic and opportunity
factors predict rebellion better than grievance measures.
 Despite theoretical advances, gaps remain:
o State Failure findings (e.g., link to low trade openness) are not well theorized.
o Statistical non-findings for grievances/discrimination → theory hasn’t explained
this null result.
o Narrative work reveals important actors and fragmentation (e.g., “riot
professionals”), but theory hasn’t specified implications of internal divisions
among insurgents or conflicts between armies and presidents.
 Closing line: greater attention to disorder details in narratives will strengthen future
theoretical models.

Forms of Capitalism

 The study of the political economy of advanced industrial states grew after the 1970s oil
shocks and is an energetic research program.
 This research program draws on historical institutionalism (influenced by Gerschenkron
1962).
 Katzenstein (1978) treated the dependent variable as the political strategies OECD states
used to adjust to the end of Bretton Woods and the oil shocks.
  Different countries, because of their historical trajectories, developed distinct institutions;
policymakers were constrained by those institutions in choosing strategies.
 Institutional capacities and interests of finance ministries, central banks, commercial
banks, and labor unions limit and enable policy responses to shocks.
 Historical institutionalists saw a continuum of capitalist types: from states strong relative
to society (mercantilist strategies) to societies strong relative to state (liberal strategies).
 The field did not assign a single numeric value per country/year for “forms of
capitalism”; instead, different studies put varied outcomes (growth, stability, wage
equality, redistribution, who bears adjustment costs, political strategy) on the left side of
their equations.
 By the 1990s there was an explicit sense that such outcomes cluster into coherent
“packages” (Esping-Andersen’s “worlds” of capitalism), so Laitin highlights the
dependent variable for this research community as these different “worlds” or forms.
 Much research specified relations within each form (e.g., how party electoral power
affects wage equality), but the central question (Gourevitch 1978) is how distinct political
economies (the dependent variable) adjust to common international challenges
(independent variables).
 Debates in the field focus on causal factors explaining distinct political economies:
sectoral balances, timing of industrialization, social partnership levels among classes,
land/labor/capital ratios.
 Historical institutionalists produced narrative-rich case studies and comparisons (Zysman
1977; Katzenstein 1985a; Gourevitch 1986); even Rogowski (1989), theoretical, used
historical narratives.
 Statistical research (Europe and U.S.) began to show stability to the institutional patterns
described by institutionalists, with many policy outcomes conditioned by type of
capitalism.
 Globalization and rapid technological change continued to affect political economies;
research retained “forms of capitalism” as the dependent variable while developing along
tripartite-method lines (statistical, theoretical, narrative).
 Econometric analyses of OECD data became stronger, but cross-sectional work forces
tighter variable specification than vague “three forms of capitalism,” and the statistical
tradition still needed more from narratives.
 Theorists criticized historical institutionalists for not formalizing discovered patterns;
unanswered questions included: why institutional convergence toward more “efficient”
patterns did not occur, and what maintains institutional patterns over time.
 Historical institutionalists emphasized political–economic interaction but did not convert
that into testable models; theorizing forms of capitalism as equilibria pushed the field to
new questions and promised future statistical tests highlighting endogeneity of politics
and economics.
 The review will cover (in sequence) cross-sectional statistical research, theoretical
developments, and continuing narrative work in historical institutionalism, aiming at
better integration of the three approaches.

Statistics subsection (from the text)

  “Forms of capitalism” is a vaguely specified, nominal dependent variable — suited to
historians’ coherent narratives rather than high R-squares.
 Rogowski (1993) tried to refashion institutionalist literature for econometric testing and
treated comparative economic growth as a consequential dependent variable.
 A decade earlier the problem looked different: Japan’s continued growth and continental
Europe’s respectable growth contrasted with U.S./U.K. declines; theories (ideas,
institutions, interests) were offered.
 A decade later growth patterns reversed (Japan’s long recession), complicating causal
claims — the rapidly changing world economy makes it hard to place countries stably on
any dimension.
 Meanwhile, new OECD data on many dimensions became available; some variables (like
inflation) were standardized by governments; others (corporatism, central bank
independence, union concentration) required scholarly construction.
 These data enabled statistical tests of many theoretical speculations.
 Research moved from political responses to oil shocks to variables like the inflation–
unemployment tradeoff, trade openness, and wage equality.
 Statistical work repeatedly showed: social-democratic (corporatist) models are stable and
associated with larger government sectors, greater equality, public investment in task-
specific skills, growth advantages in certain sectors, and comparative advantage in the
face of shocks.
 A third form (Christian Democracy) presents a package of high equality and low taxes,
sacrificing some growth (Swenson 1989; Iversen & Wren 1998).
 The major debate: globalization’s impact on different forms of capitalism.
o Consensus: globalization’s impact is strong but unclear.
o Rodrik (1998): globalization pressures increase demand for welfare spending as
insurance against shocks, at cost of losing mobile capital.
o Scharpf (1991) and Lambert (2000): globalization undermines the welfare state
even in corporatist governments.
o Garrett (1998): democratic corporatism can cushion globalization and be a good
response, attracting mobile capital.
o Iversen & Cusack (2000): challenge the strong-globalization view, arguing
technological change is a stronger force than globalization.

If globalization is strong, expect convergence of OECD structures and strategies; if technological

change dominates, expect continued variation in growth, welfare-state growth, and wage equality
depending on countries’ technological profiles.

Theory

 Hall & Soskice (forthcoming) specify historical institutions as equilibria, linking

historical institutionalist and statistical work.
 Since the 1960s, political economy has sought to explain cross-national patterns of
economic policy and performance, stressing institutions’ importance.
 Endogenous growth theory suggests national growth rates depend on institutional
structures of national economies.
 Hall/Soskice focus on comparative institutions: institutions shape technological progress
and growth; understanding them as equilibria explains growth.
 Their approach is based on the “new economics of organization,” with firms as the main
unit adjusting to shocks.
 Firms reduce risk by commitments to workers and other firms in five spheres:
(a) wage bargaining and working conditions,
(b) securing skilled labor,
(c) obtaining finance,
(d) coordinating with firms (e.g., standards),
(e) ensuring employees act as firm agents.
 Strategies for these problems depend on national institutions.
 National political economies are the main unit of analysis; most significant variations
occur at the national level; regulatory regimes are nation-state domains.
 Dependent variable dimension:
o Liberal Market Economies (LMEs): coordination via market mechanisms,
arms-length relations, formal contracting.
o Organized Market Economies (OMEs): non-market coordination, incomplete
contracts, shared private info, collaborative inter-firm relations.
 Equilibrium predictions:
o LME firms invest in “switchable” assets usable for other purposes.
o OME firms invest in specific assets relying on others’ cooperation.
 Components are complementary: high returns in one part increase returns in others (e.g.,
long-term jobs + finance tolerant of short-term losses).
 Complementarity explains clustering of solutions to coordination problems.
 Predictions under globalization:
o Economists’ assumption of universal liberalization is wrong (profitability ≠ just
low labor costs).
o Hypothesis 1: OME firms may move to LMEs for radical innovations; LME firms
may move to OMEs for quality control and skilled labor.
o Hypothesis 2: Labor-capital conflict under globalization: low in OMEs (joint
support for regimes, strong unions); high in LMEs (push for deregulation,
weakening unions).
 Hall/Soskice account for stability of social democracy and correlations between
government spending, welfare, and union density as packages.
 Endogenizing politics and economics requires new statistical testing (Alt & Alesina
1996).
 Their approach aligns with historical institutionalists’ equilibrium view of capitalism.
 Explains why globalization does not produce full convergence; allows for national
variation.
 Provides alternative to price theory, treating institutions as stable equilibria rather than
efficiency constraints.
 But formal theory diverges from narrative findings in historical institutionalism.
Narratives

 Historical institutionalists produce empirically dense narratives aligned with

statistical/formal dependent variables but with limited impact on modelers.
 Katzenstein (1985a, 1985b):
o Explained small European states’ political stability despite economic flexibility
through corporatist governance.
o Identified liberal and social sub-types of democratic corporatism.
o Smallness → vulnerability → ideology of social partnership sustains institutions.
o Sectoral analyses: Austria (social corporatist), Switzerland (liberal corporatist).
o Ideology variable (hard to generalize) should have influenced future
theory/statistics but rarely did.
 1990s focus on institutional breakdown:
o Dore & Berger (1996): narrative studies of Japan/Europe showed national
political economies retained institutional integrity under globalization.
o Thelen & Kume (1999): similar findings for labor policy in Germany/Japan.
o These studies haven’t forced globalization-convergence modelers to revise their
theories.
 Pierson (1994):
o Narratives on welfare state retrenchment under Thatcher/Reagan.
o Prediction: LMEs under conservatives → big retrenchment.
o Finding: grand goals, limited success.
o Institutional structures (e.g., veto points) didn’t explain outcomes; Thatcher less
successful than Reagan.
o Key variable: “policy feedback” — features of programs themselves made
dismantling difficult.
o Feedback not statistically tested; difficult since each program has multiple
feedback dimensions.
o Suggested building middle-range theories; also proposed:
1. Replicate study across LME/OME divide to test policy feedback.
2. Develop coding of program criteria for exploratory cross-national
statistical testing.
o Pierson (1996): compared Sweden/Germany with UK/US; found no OME
advantage in sustaining welfare state.
 Narrative uncovered plausible new variable (“policy feedback”) needing attention in
theory/statistics.
 Herrigel (1996):
o Historical study of German industrialization shows no unified national economy;
two intersecting institutional frameworks existed.
o Implication: cross-sectional studies using OECD country-level data risk error.
o For variables like Katzenstein’s (social networks), data must be disaggregated to
lowest administrative level.
o Requires major effort to gather such regional-level data.
  Narratives highlight challenges and opportunities for statistical/formal models in
comparative political economy.
 The research program remains vibrant but needs to reconcile inconsistent findings across
the tripartite methodology.

Conclusion

 The review is not comprehensive; it illustrates the tripartite methodology (statistical,

formal, narrative) using three dependent variables.
 Mentions two additional dependent variables with active research:
o Selection of political institutions (electoral laws, presidential vs. parliamentary
systems, central banks).
 Inspired by Riker’s 1980 point: if institutions shape outcomes, their choice
must be endogenous.
 Propelled by post-Soviet constitutions, new democracies, and EU
institutionalization.
o Formation/re-formation of political identities.
 Rooted in Lipset & Rokkan (1967) on cleavages.
 Shift from class to ethnic identities has motivated new research.
 Focus: relationship between mobilized identities and party formation.
 Thesis implication: researchers using all three methodologies (statistics, formal models,
narratives) are interdependent, since they all address variance on dependent variables.

Most robust findings of the past decade:

 Democracy: High GDP per capita helps explain consolidation of democracy, but not its
initiation.
 Order:
o Models using group culture or minority grievances fail to explain ethnic civil
wars.
o Better explained by country-level variables (GDP per capita, population size,
terrain, economic growth).
 Forms of Capitalism:
o Each form is a robust equilibrium.
o Less affected by globalization/electronics revolution than predicted.
o Suggests social democracy is unlikely to disappear soon.

Future of comparative politics research:

 Not reasonable to demand all comparativists master all three methodologies.

 Losing any one skill set would damage the discipline.
 Fear 1: formal theorists driving fieldworkers out (like a Chomsky-style revolution).
 Fear 2: fieldworkers isolating themselves from formal/statistical findings.
 Ideal model: physics.
o Division of labor between theorists and experimentalists.
o Despite disrespect, neither side ignores the other’s findings.
 In comparative politics:
o Interdependence across the tripartite divide, with grudging toleration, is essential
for progress.
o Narrative, formal, and statistical scholars all face challenges in reacting to one
another.
o Evidence shows a common focus is emerging:
 Consequential dependent variables.
 Joint attempts to explain variance across polities.
o Those working in this division of labor are remaking the comparative method.