Snyder Scaling Down
Snyder Scaling Down
Snyder Scaling Down
zation of what were once only nominally federal systems in cases such as Rus-
sia and Mexico shifted the distribution of power between national and
subnational governments in ways that greatly strengthened the latter. More-
over, democratization was associated with both the "federalization" of for-
merly unitary systems (e.g., Spain) and the breakup of federal systems (e.g.,
Czechoslovakia, Yugoslavia, Soviet Union). An extreme form of decentraliza-
tion is seen in cases where a change of political regime resulted in the collapse
of the central state and the emergence of anarchical warlord politics (e.g., Congo/
Zaire, Liberia, Sierra Leone, and Somalia).
In response to major centrifugal trends such as these, comparativists
have focused increasing attention on subnational units of analysis. Thus
many recent studies of topics such as economic policy reform, democra-
tization, and ethnic conflict rely on comparisons across subnational po-
litical units. 1 Despite the proliferation of works that employ subnational
comparisons, insufficient attention has been devoted to the methodologi-
cal issues that arise in the comparative analysis of subnational units. 2 To
help fill this gap, this article explores how the subnational comparative
method can expand and strengthen the methodological repertoire available
to comparative researchers.
I argue that the subnational comparative method offer three key advantages.
First, it can help better manage some of the characteristic limitations of a small-
N research design. The use of subnational units is an especially effective strat-
egy for increasing the number of observations and thus mitigating the problem
of "many variables, small N. ''3 Moreover, a focus on subnational units can
make it easier to construct controlled comparisons that increase the probabil-
ity of obtaining valid causal inferences in small-N research.
Second, subnational comparisons strengthen the capacity of comparativists
to accurately code their cases. Comparativists too often rely on national-level
means and aggregate data when studying countries with high degrees of inter-
nal heterogeneity. This tendency to unreflectively gravitate toward national-
level data and national units of analysis--a tendency that Stein Rokkan (1970)
some time ago called "whole-nation bias"--has contributed to a miscoding of
cases that can distort causal inferences and skew efforts at theory building. 4 A
greater sensitivity to within-nation variation and complexity can help
comparativists avoid these pitfalls.
Third, a focus on comparing subnational units better equips us to handle the
spatially uneven nature of major processes of political and economic transfor-
mation. Processes such as democratization and economic reform often have
varied effects across the territorially-defined subunits of a political system. 5
Because the subnational comparative method makes it easier to see this within-
nation variation, it contributes to a more adequate description of complex pro-
cesses of change. In addition to providing a salutary increase in our ability to
accurately describe complex processes, a focus on subnational units has im-
portant implications for how we theorize such processes. Disaggregating coun-
tries along territorial lines makes it possible to explore the dynamic linkages
among the distinct regions and levels of a political system. Analyzing these
Snyder 95
The subnational comparative method has two key strengths with regard to
research design: (1) it can serve as a powerful tool for increasing the number
of observations, and (2) it makes it easier to construct controlled comparisons.
All comparative research faces the challenge of coding cases: the analyst
must employ implicit or explicit indicators for categorizing the cases included
in the study. Subnational comparative analysis can help mitigate two forms of
bias that affect how national cases are coded. The first involves the inappropri-
ate use of national-level averages in studying countries with a high degree of
98 Studies in Comparative International Development / Spring 2001
internal heterogeneity. The second kind of bias results when the attributes of a
well-studied or especially salient subunit are improperly used to characterize
the national case as a whole.
a new theoretical agenda that focuses on the linkages between distinct modes
of industrialization within a single national unit.
In sum, mean-spirited analysis and the misuse of the best known place pose
important problems of measurement and description: mean-spirited analysis
inappropriately employs highly aggregated data to code cases, whereas the
misuse of the best known place inappropriately employs disaggregated data to
make inferences about national units. As the earlier examples suggest, in het-
erogeneous societies the distortions associated with these two problems can
lead to the invalid coding of national cases.
that kept the populist regime in power. 29 From this perspective, the standard inter-
pretation of Latin American populism as a labor-based phenomenon was insuffi-
ciently attuned to the linkages between metropolitan and peripheral regions.
Consequently, populism's metropolitan face was seen as its only face.
What is wrong with this understanding of populism? First, such a perspec-
tive potentially distorts the forces that kept populist governments in power: a
focus limited to the metropolitan side of populism can obscure the fact that
labor alone was electorally insufficient to sustain populist policies. Moreover,
ignoring the provincial dimension of populist governments miscasts the rela-
tionship between these governments and the old, oligarchic order that preceded
them. The lopsided view of populism as an urban, labor-based phenomenon
either pits populist governments against peripheral, traditional elites or, more
simplistically, relegates such elites to a marginal role in national politics. By
contrast, a balanced understanding of populism as a composite of metropoli-
tan and provincial coalitions provides a very different, less adversarial per-
spective on how populist governments interacted with rural elites: It highlights
how populist governments accommodated and even depended on provincial
oligarchs in order to "link populism to the traditional order, give it coherence
as a national electoral force, and extend its reach throughout the national terri-
tory" (Gibson 1997: 341). 30
Focusing on the dynamic linkages among the levels and regions of a politi-
cal system provides a new way of looking at the relationship between contrast-
ing political phenomena observed at the "center" (i.e., the national level) and
at the "periphery" (i.e., the subnational level). In the study of political and
economic liberalization in contemporary Latin America, a regionally differen-
tiated framework highlights how illiberal political and policy regimes that per-
sist, and even proliferate, in the peripheral areas of many countries may be
undercurrents that are causally integral components of the process of liberal-
ization seen at the center. 31 By contrast, studies that focus only on politics at
the national level either do not "see" these illiberal peripheries, or, alterna-
tively, inappropriately dismiss them as isolated backwaters, disconnected from
and left behind by the processes of political transformation observed at the
center. Such a center-centered perspective, which has tended to dominate re-
cent research on political and economic liberalization in developing countries,
can lead us to misunderstand these fundamental processes as far more uniform
than they in fact are.
The Spatial Unevenness of Democracy. Guillermo O'Donnell's pioneering
work on the shortcomings of contemporary democratic regimes in Latin America
highlights how a focus on the interaction among the territorially defined com-
ponents of a political system helps us better understand the process of de-
m o c r a t i z a t i o n . O ' D o n n e l l a r g u e s that p o l i t i c a l l i b e r a l i z a t i o n and
democratization are territorially uneven processes. Consequently, within
a particular country, the rules and norms that define the national regime
and state cannot be assumed to apply equally throughout the country. For
102 Studies in Comparative International Development / Spring 2001
the agenda of neoliberal reform set at the national level by the president. In
fact, Salinas did not merely tolerate such subnational deviations from his
neoliberal line, he selectively promoted reregulation by vigorously supporting
populist governors in certain states. Thus liberalizing economic reforms at the
national level and the proliferation of illiberal policy enclaves at the subnational
level were complementary and closely articulated components of a single process:
the latter helped secure the political stability necessary for the former to proceed.
Gibson and Calvo (2000) also employ a subnational comparative method to
explain the territorially uneven nature of economic reforms in Argentina. They
show how electoral rules heavily overrepresented sparsely populated rural ar-
eas in the national legislature. This situation gave national politicians strong
incentives to shield rural areas from neoliberal policy reforms in order to se-
cure the legislative majorities n e e d e d to i m p l e m e n t these reforms in
underrepresented, urban areas. 34 Thus the persistent illiberal policy regimes in
the Argentine periphery were not stagnant backwaters left behind by the pro-
cess of economic reform in urban areas. Rather, they were causally integral
components of the overarching economic reforms.
As these examples suggest, understanding economic transformation in con-
temporary Latin America requires that we explore the linkages between liberal
and illiberal areas in the context of a single, unified system. The subnational
comparative method is a key tool for achieving this goal.
Conclusion
Notes
* I appreciate helpful commentson this material from Nancy Bermeo, Dexter Boniface, David
Collier, John Gerring, Edward Gibson, Robert Kaufman,Juan Linz, James Mahoney, Kelly
McMann, Gerardo Munck, Peter Nardulli, David Samuels, Judith Tendler, and two anony-
mous reviewers. I also benefitedgreatlyfrom the insightfulcommentson an earlier draft pro-
vided by the participantsin the conferenceon "Regimesand PoliticalChangein LatinAmerica;'
held at the Universityof Illinois at Urbana-Champaignin August 1999.
1. See, for example, Kohli (1987, 1990); Fishman (1993); Fox (1993, 1996); Putnam (1993);
Locke(1995); Hagopian(1996); Lockeand Jacoby(1997); Rubin (1997); Stoner-Weiss(1997);
Tendler (1997); Samuels (1998); Snyder (1999a, 2001); Firmin-Sellers (2000); Gibson and
Calvo (2000); Heller (2000); Montero(2000); Remmerand Wibbels (2000); Varshney(2001);
and Jones Luong(2002). Althoughsubnationalunitscommandincreasingattention,subnational
political analysis is not an entirely new phenomenon in comparative politics: a number of
earlier works focusedon subnationalcases (for example,Tilly 1964; Linz and de Miguel 1966;
Kesselman and Rosenthal 1974; and Tarrow 1967, 1976).
2. Theseissues do receive limited attentionin Lijphart(1971); King, Keohane,and Verba (1994);
Snyder 105
and Peters (1998). Linz and de Miguel (1966) is one of the few works that focuses primarily on
methodological issues involved in comparing subnational units.
3. The classic discussion of this problem is Lijphart (197t). See also Collier (1993)-
4. The problem of whole-nation bias is also discussed in Lijphart (1975:166-9).
5. See, for example, O'Donnell (1993); Fox (1994); Kurtz (1999); Snyder (1999a, 2001); Gibson
and Calvo (2000); and Remrner and Wibbels (2000).
6. Collier (1993:112) also discusses the value of within-case, "internal comparisons" for increas-
ing the number of observations.
7. In this context, King et al. discuss Atul Kohli's (1987) research on social policy in India. To
test hypotheses about the effects of state-level political regimes on poverty policy, Kohli uses
additional observations from the local, panchayat level (Kohli also looks at other countries to
evaluate the observable implications of his hypotheses). Recent work on divided government
at the state level in the United States provides another good illustration of how moving to a
lower level of aggregation can help solve the small-N problem by increasing the number of
observations. See, for example, Alt and Lowry (1998); and Lowry, Alt, and Ferree (1998).
8. Lieberson (1985: 108) adopts a quite different view of this issue, concluding that evidence at a
lower level of aggregation "is irrelevant for determining the validity of a proposition about
processes operating on the higher level" and that "no useful understanding of the higher-level
structure can be obtained from lower-level analyses."
9. "Galton's problem" is so named because of the objection raised by Francis Galton at the meet-
ing of the Royal Anthropological Institute in 1889 to a pioneering paper by Edward B. Tylor
introducing the cross-cultural survey method. Galton pointed out that because traits often spread
by diffusion (for example, by borrowing or migration), observations of such traits across cul-
tures were not necessarily independent instances (Naroll 1961: 15).
I0. As discussed below, dynamic interactions among the subunits of a political system should not
necessarily be seen as a troubling hindrance to making valid causal inferences. Rather, such
interactions can be a fertile source of theoretical innovation and progress.
ll. See Locke (1995) for an intriguing research design that combines the study of varied subnational
units with an analysis of multiple sectors. Locke offers a subnational study of the Italian auto-
mobile and textile industries.
12. Rustow (1968: 45) made a related point when he argued that "mere geographic proximity does
not necessarily furnish the best basis of comparison."
13. As O'DonneU himself observes (1973: 21), such a strategy overlooks the possibility that the
presence or absence of a large, underdeveloped periphery may have a crucial impact on na-
tional politics. Hence, even though the modem sectors in Argentina and Brazil are similar in
size, their effect on national politics may differ because of the different sizes of the agrarian
sectors in the two cases.
14. This between-nation comparison thus enables Lipset to employ a "most similar" systems de-
sign. On "most similar" and "most different" systems designs see Przeworski and Teune (1970).
15. Another commonly-employedstrategy of between-nation subnational comparison involves fo-
cusing on the same economic sector across distinct countries (see, for example, Evans 1995;
Bates 1997; Karl 1997; and Paige 1997). Because of cross-country differences on a wide
range of other variables, such an approach may be less effective than within-country compari-
sons for exploring how variation in political factors shapes economic outcomes.
16. In the context of the Saskatchewan-Alberta-Manitobajuxtaposition, Lipset (1950:217) makes
an intriguing cross-temporal comparison, noting that "during the period between the 1890's
and the early twentieth century, when economic and ecological conditions in Manitoba ap-
proximated those of Saskatchewan and Alberta today [ 1950], the Manitoba agrarian political
movement was strong"
17. This difficulty of achieving independent observations among spatially proximate units is dis-
cussed in the econometric literature on "spatial autocorrelation" (e.g., Anselin 1988). The spa-
tial autocorrelation problem is in many ways analogous to the "temporal autocorrelation"
problem in time-series analysis (Eagles 1995: 7; King 1997: 166).
18. The value of between-nation comparisons of non-contiguous units for increasing the indepen-
dence among cases is suggested by Waldo Tobler's "first law of geography" which posits that
106 Studies in Comparative International Development / Spring 2001
"everything is related to everything else, but near things are more related than distant ones"
(Eagles 1995: 7).
19. See, for example, Lipset (1959) and the huge literature inspired by this study.
20. Peters (1998: 44) makes a related point when he observes that whole countries are often coded
as outliers when only one part of the country has extreme values, whereas the rest of the
country actually has "normal" values with respect to the cross-national distribution.
21. See, for example, Nelson (1990) and C6rdoba (1994).
22. A similar reliance on easily available, national-level OECD data in much recent work on the
comparative political economy of advanced industrial countries has also undoubtedly obscured
important processes occurring below the national level.
23. Moving to the subnational level does not necessarily guarantee a correct coding of national
cases because scholars can miscode the subnational units themselves. For example, Gaines
(1999) shows that scholars miscoded province-level party systems in Canada as two-party
systems and thus incorrectly concluded that Canada conformed to Duverger's law at the pro-
vincial, if not at the national, level (Duverger's law states that plurality rule elections result in
two-party competition). By contrast, Gaines shows that province-level party systems in Canada
actually have more than two parties. Thus Canada stands as a genuine counterexample to
Duverger's law.
24. On the rather different problem of"the misuse of the little known case" see Kenworthy (1973).
Ruth Berins Collier and David Collier's inversion of Kenworthy's formulation--"the misuse
of the best known case"--has obviously influenced my formulation (Collier and Collier 1991:
14).
25. The problem of misusing the best know place might be characterized as the well-known eco-
logical fallacy "in reverse" in that disaggregated data from subnunits results in a biased infer-
ence about the aggregate level. The ecological fallacy refers to the biases that can result when
using aggregate data to make inferences about individual-level behavior (Robinson 1950). A
method for making unbiased inferences from aggregate data is presented in King (1997).
26. In its view of national political economies as complex composites of diverse subnational sys-
tems, Herrigel's work exemplifies what might be called the "MIT school" of political economy
associated with the work of scholars such as Suzanne Berger, Michael Piore, Charles Sabel,
and Judith Tendler. Richard Locke's research on Italy also exemplifies the MIT school. In
his important book on the Italian case, Locke (1995) argues for a "micro-politics" ap-
proach to political economy rooted in the assumption that "national political economies
are not coherent systems but rather incoherent composites of diverse subnational patterns
that coexist (often uneasily) within the same national territory." Locke (1995: 16-20) criticizes
tendencies by some scholars to extrapolate the economic successes and dynamism of the so-
called "Third Italy" based on small enterprises and flexible production to the national case as
a whole. This tendency provides a good example of what I mean by misusing a well-known
place.
27. In their critique of "dualist" perspectives on development that divided countries into a modern,
capitalist economy and a backward, pre-capitalist one, dependency theorists often expressed a
strong concern with describing and explaining the connections between contrasting regions of
a country. See, for example, Stavenhagen (1968) and Portes (1983).
28. The term "center-centered" is from Rubin (1997: 12).
29. See Tendler (1999: 105-19) for an intriguing extension of Gibson's argument that shows how
"two-pronged" coalition-building strategies that marry modern and traditional constituencies
have been implemented at the subnational level by contemporary Brazilian governors. Roberts
(1995) makes a related argument about "neopopulism" as a coalition-buildingtool in the con-
text of the implementation of free-market economic reforms in Peru during the 1990s. On
neopopulism, see also Weyland (1996, 1999).
30. Hagopian's important work on Brazil also shows that a keen sensitivity to variation at the
subnational level can highlight how transformational national regimes accommodate, rather
than marginalize, traditional elites. As in the literature on populism, work by political scien-
tists on Brazil's bureaucratic-authoritarian regime tended to focus exclusively on urban, met-
ropolitan areas. In this respect, Hagopian (1996: xiii) notes, "Many country studies that identify
themselves as national in scope are in fact analyses of the most advanced regions--Rio de
Snyder 107
Janeiro and especially S~o Paulo. As such, they have a collective blind spot and...have paid a
price for their myopia?' See also Migdal, Kohli, and Shue (1994) on the value of subnational
analysis in the study of state-society relations.
31. By "illiberal" 1 mean non-democratic in the political sphere and non-free-market in the eco-
nomic sphere.
32. On subuational authoritarian regimes in Mexico, see Fox (1994) and Snyder (1999b). On the
important, yet often overlooked, role of non-democratic rural elites in securing political and
economic liberalization at the national level in Chile, see Kurtz (1999).
33. We might want to consider how we should code such a case. Is a country characterized by the
coexistence of a national democratic regime alongside subnational authoritarian regimes a
full-fledged democracy?
34. Remmer and Wibbels (2000) also analyze the divergent effects of neoliberaI economic reforms
across Argentine provinces. They argue that the persistence of illiberal policy regimes at the
provincial level may jeopardize national economic performance and thus undermine the
sustainability of reforms at the national level.
35. On the related trade-off between generalizability and conceptual validity, see Sartori (1970;
1984); Collier and Mahon (1993); Collier and Levitsky (1997); Collier and Adcock (1999);
and Munck (1998).
References
Alt, James E. and Robert C. Lowry. 1994. "Divided Government and Budget Deficits: Evidence
from the States." American Political Science Review 88 (December): 811-28.
Anderson, Jeffrey J. 1992. The Territorial Imperative: Pluralism, Corporatism, and Economic Cri-
sis. New York: Cambridge University Press.
Anselin, Luc. 1988. Spatial Econometrics: Methods and Models. Boston: Kluwer Academic Pub-
lishers.
Bates, Robert H. 1997. Open-Economy Politics: The Political Economy of the World Coffee Trade.
Princeton: Princeton University Press.
Collier, David and Robert Adcock. 1999. "Democracy and Dichotomies: A Pragmatic Approach to
Choices about Concepts." Annual Review of Political Science 2: 537-65.
Collier, David and Steven Levitsky, 1997. "Democracy with Adjectives: Conceptual Innovation in
Comparative Research." World Politics 49 (April): 430-451.
Collier, David and James E. Mahon, Jr. 1993. "Conceptual 'Stretching' Revisited: Adapting Cat-
egories in Comparative Analysis?' American Political Science Review 87 (December): 845-
55.
Collier, Ruth Berins and David Collier. 1991. Shaping the Political Arena: Critical Junctures, the
Labor Movement, and Regime Dynamics in Latin America. Princeton: Princeton University
Press.
C6rdoba, Josr. 1994. "Mexico." In The Political Economy of Policy Reform, ed. J. Williamson.
Washington, D.C.: Institute for International Economics.
Di Tella, Torcuato S. 1965. "Populism and Reform in Latin America?' In Obstacles to Change in
Latin America, ed. C. V61iz. New York: Oxford University Press.
. 1968. "The Working Class in Politics?' In Latin America and the Caribbean - A
Handbook, ed. C. V61iz. New York: Praeger.
Eagles, Munroe. 1995. "Spatial and Contextual Models in Political Research: An Introduction." In
Spatial and Contextual Models in Political Research, ed. M. Eagles. London: Taylor and
Francis.
Evans, Peter. 1995. Embedded Autonomy: States and Industrial Transformation. Princeton: Princeton
University Press.
Firmin-Sellers, Kathryn. 2000. "'Institutions, Context, and Outcomes: Explaining French and Brit-
ish Rule in West Africa" Comparative Politics 32 (April): 253-72,
Fishman, Robert M. 1993. "Divergent Paths: Labor Politics in Barcelona and Madrid" In Politics,
Society, and Democracy: The Case of Spain, ed. R. Gunther. Boulder, CO: Westview Press.
Fox, Jonathan. 1993. The Politics of Food in Mexico: State Power and Social Mobilization. Ithaca:
Comell University Press.
108 Studies in Comparative International Development / Spring 2001
Lowry, Robert C., James E. Alt, and Karen E. Ferree. 1998. "Fiscal Policy Outcomes and Electoral
Accountability in American States?' American Political Science Review 92 (December): 759-74.
Migdal, Joel S., Atul Kohli, and Vivienne Shue, eds. 1994. State Power and Social Forces: Domina-
tion and Transformation in the Third World. New York: Cambridge University Press.
Miles, William. 1994. Hausaland Divided9 Ithaca: Cornell University Press9
Montero, Alfred E 2000. "Delegative Dilemmas and Horizontal Logics: Subnational Industrial Policy
in Brazil and Spain." Paper presented at the Third Meeting of International Working Group on
Subnational Economic Governance in Latin America and Southern Europe. San Juan Puerto
Rico, August 26-28.
Munck, Gerardo. 1998. "Canons of Research Design in Qualitative Analysis?' Studies in Compara-
tive International Development 33 (Fall): 18-45.
Naroll, Raoul. 1961. "Two Solutions to Galton's Problem." Philosophy of Science 28 (January): 15-39.
9 1966. "Scientific Comparative Politics and International Relations?' In Approaches
to Comparative and International Politics, ed. R. Farrell. Evanston, IL: Northwestern Univer-
sity Press.
Nelson, Joan M. 1990. "Introduction: The Politics of Economic Adjustment in Developing Na-
tions." In Economic Crisis and Policy Choice: The Politics of Adjustment in the Third World,
ed. J. Nelson. Princeton: Princeton University Press.
O'Donnell, Guillermo. 1973. Modernization and Bureaucratic-Authoritarianism: Studies in South
American Politics. Berkeley: Institute of International Studies, University of California, Ber-
keley.
9 1999. Counterpoints: Selected Essays on Authoritarianism and Democratization.
Notre Dame, IN: University of Notre Dame Press9
Paige, Jeffrey M. 1997. Coffee and Power: Revolution and the Rise of Democracy in Central America.
Cambridge: Harvard University Press.
Peters, B. Guy. 1998. Comparative Politics: Theory and Methods. New York: New York University
Press.
Portes, Alejandro. 1983. "The Informal Sector: Definition, Controversy, and Relation to National
Development?' Review 7 (Summer): 151-74.
Przeworski, Adam and Henry Teune. 1970. The Logic of Comparative Social Inquiry. New York:
John Wiley and Sons.
Putnam, Robert D., with Robert Leonardi and RaffaellaY. Nanetti. 1993. Making Democracy Work:
Civic Traditions in Modern Italy. Princeton: Princeton University Press.
Remmer, Karen L. and Erik Wibbels. 2000. "The Subnational Politics of Economic Adjustment:
Provincial Politics and Fiscal Performance in Argentina." Comparative Political Studies 33
(May): 419-51.
Roberts, Kenneth. 1995. "Neoliberalism and the Transformation of Populism in Latin America: The
Peruvian Case." World Politics 48 (October): 82-116.
Robinson, William S. 1950. "Ecological Correlations and the Behavior of Individuals?' American
Sociological Review 15 (June): 351-57.
Rokkan, Stein. 1970. Citizens, Elections, Parties: Approaches to the Comparative Study of the
Processes of Development. New York: David McKay Company.
Rubin, Jeffrey W. 1997. Decentering the Regime: Ethnici~, Radicalism, and Democracy in Juchit6n,
Mexico. Durham: Duke University Press9
Rustow, Dankwart A9 1968. "Modernization and Comparative Politics: Prospects in Research and
Theory." Comparative Politics 1 (October): 37-51.
Samuels, David J. 1998. "Careerism and its Consequences: Federalism, Elections, and Policy-
Making in Brazil," Doctoral Dissertation, Department of Political Science, University of Cali-
fornia, San Diego,
Samuels, David J. and Richard Snyder. 2001. "The Value of a Vote: Malapportionment in Compara-
tive Perspective." British Journal of Political Science 31, 4 (October).
Sartori, Giovanni. 1970. "Concept Misformation in Comparative Research." American Political
Science Review 64: 1033-539
Sartori, Giovanni, ed. 1984. Social Science Concepts: A Systematic Analysis9 Beverly Hills: Sage,
Snyder, Richard. 1999a, "After Neoliberalism: The Politics of Reregulation in Mexico." World Poli-
tics 51 (January): 173-204,
110 Studies in Comparative International Development / Spring 2001