Comparative-Historical Analysis and Knowledge Accumulation in the Study of Revolutions

Jack A. Goldstone
University of California, Davis
Forthcoming in Comparative Historical Analysis, edited by Dietrich Reuschemeyer and James
Mahoney (Cambridge University Press).

A TALE OF TWO METHODS

It is striking that those works on the subject of revolutions that have had lasting influence have
been almost exclusively built around comparative case studies. From Crane Brinton’s (1938)
Anatomy of Revolutions to Barrington Moore Jr.’s (1966) Origins of Dictatorship and Democracy
to Theda Skocpol’s (1979) States and Social Revolutions, this pattern holds true. Even books with
widely divergent theoretical perspectives, such as Samuel Huntington’s (1968) Political Order in
Changing Societies, Chalmers Johnson’s (1966) Revolutionary Change, and Ted Robert Gurr’s
(1970) Why Men Rebel build their arguments around series of brief analytic case accounts, rather
than efforts at statistical analysis of large data sets.
This is not because such data sets do not exist. Although it is sometimes argued that
revolutions—particularly great social revolutions, such as those of France in 1789 or Russia in
1917—are too few for formal data analysis, this seems to exaggerate the truth. Jeff Goodwin
(2001) lists no fewer than 18 “major social revolutions” (presumably the most restrictive category
of revolutions) from France in 1789 to the collapse of communism in Eastern Europe in 1989.
John Foran (1997a), using only a slightly less restrictive definition, lists 31 revolutions occuring
merely in the Third World since 1900. If we extend the definition to cover not only revolutions,
but also revolutionary movements, the Encyclopedia of Political Revolutions (Goldstone 1998c)
lists over 150 cases around the world since the 16th century. Studies using not merely cases of
revolution, but counting all years during which particular countries experience mass political
violence, further push their “n” into the hundreds or thousands.
Nor has there been any lack of effort to undertake the analysis of revolutions using large-N
statistical analysis. Gurr (1968) and Feierabend, Feierabend, and Nesvold (1969) sought to
develop statistical analysis of the causes of revolutions; Hibbs (1973) used sophisticated recursive
time-series models to probe the sources of mass political violence. Muller (1985), Weede (1987),
and Midlarsky (1982) used statistical analysis to examine the relationship between inequality and
revolutions. Nonetheless, these efforts, far from providing a more solid basis for inference and
knowledge cumulation, as some critics of small-N analyses have suggested (Goldthorpe 1997,
Lieberson 1991), have themselves failed to provide any cumulative insights. As Mark Lichbach
(1989) points out, studies of inequality and revolution have revealed conflicting results. In most
statistical studies of revolution, results vary according to the time period and the model
specification, so that no consensus emerged. As a result, most of these works are nowadays cited
far less frequently than case-study analyses, and the research tradition of large-N statistical studies
of revolution has borne modest fruit.
This essay attempts to account for this outcome, showing how comparative and historical
analysis of revolutions has led to steady progress and knowledge cumulation. In the following
sections, I first consider the logic and methods of comparative historical analysis, and point out its
advantages for the study of complex events like revolutions. Next, I offer a brief history of the
scholarship on revolutions, showing how case-study analyses of revolution have in fact
successively built upon prior analyses in rigorous and cumulative fashion, extending the range and
 2

detail of theory at the same time that they have deepened empirical knowledge. Finally, I close
with a brief compilation of the insights gained from comparative historical analysis of revolution.

THE GOALS AND INFERENCE METHODS OF COMPARATIVE HISTORICAL ANALYSIS

Large N-statistical inferences based on normal frequency distributions are designed to answer the
following problem: Given a large universe of comparable cases, how can we infer the
characteristics of those cases—including correlations among hypothesized causes and
effects—from a smaller sample? Starting with this problem, concerns about whether the sample is
biased, whether it is large enough to distinguish causes from random effects, and whether it is large
enough to discern probabilistic relationships, are of great importance (Keohane, King and Verba
1994).
However, this is not the problem addressed by comparative historical analysis (CHA).
Analysts using CHA generally face a finite set of cases, chosen against a backdrop of theoretical
interests, and aim to determine the causal sequences and patterns producing outcomes of interest in
those specific cases. Generalization is certainly a goal, but that generalization is sought by piecing
together finite sets of cases, not by sampling and inference to a larger universe.
To offer a homely example, let us say that explorers are surveying a large territory. If they
took the “large-N statistical” approach to developing knowledge of that territory, they would have
to sample enough locations to provide reliable inferences regarding the territory as a whole. If the
territory is fairly homogenous (or if its main characteristics are distributed in a statistically normal
distribution across the territory), such sampling would produce a fairly quick, accurate, and reliable
method of determining that territory’s main characteristics. If, however, the territory has
substantial local variations, and the explorers are as interested in those variations as in any general
characteristics, sampling will be useless. Moreover, as is commonly the case, if the territory has
six or seven distinctive zones, then sampling may just produce confusing or inconclusive results,
leading observers to imagine a fictitious “average” character that actually obtains nowhwere.
By contrast, if the explorers spread out, and each surveys and seeks to understand the
character of a different zone, they can put together a map of the entire territory with far greater
accuracy than an overall sample would provide. Of course, it takes time, and none of the explorers
by themselves would be able to make reliable inferences that extended beyond the zone that they
had studied; but together, assembling their distinct maps, they cumulate genuine knowledge of the
territory. Indeed, comparing notes, they may find deep regularities, or relationships among or
across different zones, that no statistical averages for the entire territory would reveal (Tilly 1984).
We noted above that large-N studies of revolutions have, to date, not been terribly fruitful.
This strongly suggests that the “territory” we seek to understand in this matter is far from
homogenous. Large-N studies that seek to understand the causes of revolution (and
non-revolution) by ignoring the differences between the Soviet Union and Burundi, or between
Cuba and Cambodia, and treating all of these as equivalent “cases,” seem bound to founder.
Assuming that certain variables have the same effects in all countries, or that specific outcomes are
always produced by the same causes, may simply be a wrong starting point.
Charles Ragin (1997) has argued that social phenomena are generally characterized by
“causal complexity,” in which the same outcome is often produced by varied and different
combinations and levels of causes, so that a single set of necessary and sufficient conditions for
 3

specific outcomes does not exist. CHA, by starting from the premise that specific cases first need
to be thoroughly understood before advancing to seek general patterns, never assumes an identity
or equivalence or representativeness of specific cases; instead the extent of similarity or difference
between cases is considered a crucial part of the investigation. CHA does not start out by assuming
the existence of, or seeking, universal causal or other patterns; rather it assumes that the degree of
generality of any particular causal mechanism or pattern is variable and part of what the
investigation needs to determine.
In reply, critics of case study methods would say (and Goldthorpe [1997] makes this
argument): how can such explorers even hope to map one such zone or case, without a large
number of observations? Changing one large territory into several small ones may reduce the
problem of generalizing somewhat, but it does not change the problem of making reliable
inferences from limited observations. In particular, establishing causal relationships even for one
case still requires a large number of observations. True enough, but once the problem is changed
from making inferences about causal relationships in a large universe from a limited sample, to
tracing the causal pattern or patterns in a small number of cases, new methods may be employed.
The most important of these used by CHA are 1) Bayesian analysis; 2) process-tracing; and 3)
congruence testing. Each of these methods allows practitioners of CHA to move toward their goal
of establishing causal relationships in a small and explicitly delimited number of cases, as the path
to cumulating knowledge.
Bayesian Analysis. While social scientists frequently and explicitly use statistical analysis
based on normal distributions, the use of other statistical methods—such as non-parametric and
Bayesian statistics—is far more rare (although see Western [2000] for a nicely explicit application
of Bayesian statistics in CHA). However, I would argue that CHA implicitly uses Bayesian
analysis quite extensively, and it is this approach that makes sense of how even single case studies
can be of immense value and influence (McKeown 1999).
“Normal” or “frequentist” statistical analysis presumes a world in which the only prior
knowledge one has is a null hypothesis, based on ignorance of causal relationships, and uses data
to identify causal relationships that depart from the null hypothesis. In order to establish
departures from the null, one needs enough cases to demonstrate that cause A is associated with
effect B to a degree that is highly unlikely starting from the assumption of no effect. One or a small
number of such associations provides no sure basis for such inferences.
By contrast, CHA rarely starts from ignorance of causal relationships. Quite the contrary;
most CHA squarely targets strong prior beliefs about causal relationships. Indeed, what makes
case analyses compelling is generally not that they pluck unknown causal relationships from
behind a veil of ignorance, but that they are specifically designed to challenge prior beliefs.
A Bayesian approach asks different questions of data than frequentist approaches. The
latter approach asks: what data do I need to convince myself that cause A is really connected to
effect B, and not merely randomly associated with it? Bayesian analysis asks: given that I have
strong prior beliefs about the relationship of A and B, how much would a particular bit of new data
shift that belief? If the prior belief is so strong as to be nearly deterministic, then simply one
opposing case may cast considerable doubt upon the prior belief.
Let us ponder how the two approaches would view recent trends in scholarship regarding
the French Revolution. Allow that scholars (such as Cobban 1964, Furet 1981, Taylor 1972) have
 4

uncovered persuasive evidence that the main elite actors in the French revolution were lawyers,
notaries, military officers, government officials, and recently-ennobled aristocrats pursuing the
ideal of a career open to talents against aristocratic privilege in the state and the army, and
opposing the monarchy mainly over its desire to raise taxes. However, merchants and
bankers—and for that matter policies that assisted the development of private property—played
little role. Proponents of a large-N statistical approach based on frequentist methods would say:
So what? This proves nothing about “revolutions” in general; how would we have any idea that the
same actors would be important in any other revolutions? This case, therefore, is of merely
historical interest.
Now consider what this case would mean for a scholar who held a strong prior Marxist
belief that revolutions revolved around class conflict, and that the key actors and policies in
revolutions against agrarian monarchies should be capitalists and their supporters interested in
advancing the gains to be made from private property. Given this prior belief, the findings of
scholars noted above would be strikingly important. It might still be that the Marxist view held in
other cases, but finding that it did not hold in one of the historically most important revolutions
(that is, a revolution in one of the largest, most influential, and most-imitated states of its day and
a frequent exemplar for Marxist theories) would certainly shake one’s faith in the value of the prior
belief.
CHA thus gains its impact not despite being statistically “faulty” by the standards of
frequentist analysis, but because that kind of analysis simply does not apply to the goals and
methods of most comparative historical studies. What CHA generally undertakes is the study of
a single, or small number, of cases, with the goal of identifying causal relationships in those cases,
in order to test (and shake) strong prior beliefs.
Demonstrating the presence of specific causal relationships or patterns within a single case,
or small number of cases, allows CHA to challenge prior beliefs in several ways:
(1) Existence/Absence: showing that causal relationships or patterns are important where they were
previously not seen or expected, or are absent where they were previously claimed to be seen or
expected. This can also be thought of as the “most likely/least likely” mode of critical case analysis
(Przeworski and Teune 1970, Collier 1993). In these analyses, CHA identifies and carefully
examines a case where existing theory claims some pattern or causal relationship is most likely to
be found and shows that in fact the claims of that theory cannot be substantiated. Or conversely,
CHA identifies and carefully examines a case where theory claims some pattern or causal
relationship is least likely to be found and shows that in fact, contra theory, that case exhibits that
pattern.
(2) Convergence/Divergence: showing that cases that theory claims are different in
significant ways in fact converge in ways not previously seen or expected, or conversely that cases
that theory claims are essentially similar in fact diverge in ways not previously seen or expected
(Collier 1993).
(3) Discrimination/Reconciliation: showing that where existing theories are in conflict,
certain cases (or elements of those cases) either discriminate between the two theories; or reconcile
them by showing how certain aspects of the cases are consistent with one theory while other
aspects are consistent with the other.
Many of the most striking and powerful CHA works manage to combine several of these
 5

approaches. Theda Skocpol’s (1979) work not only showed an unexpected convergence of causes
for liberal and communist social revolutions; it also showed systematic divergences between her
cases of successful social revolutions and other kinds of revolutionary events (e.g. the English
Revolution of 1640, the Japanese Meiji Restoration, and the German Revolutions of 1848). Critics
of Skocpol have misunderstood her work greatly by casting it as a study of “revolutions” in general,
as if it was arguing simply from the association of a social revolutionary outcome in a handful of
cases with a certain set of variables (or interaction among those variables) that those variables are
generally causal to that outcome across a large universe of cases. That would be an unwarrented
inference, and Skocpol (1994) explicitly disavowed it.
What astute critics of Skocpol’s work have pointed out (Mahoney 1999, Little 1998) is that
her accomplishment was to demonstrate the existence of certain patterns of events in historically
important but disparate cases of revolution—a pattern unseen and unexpected by prior theories.
Moreover, she also demonstrated that this pattern of events was absent (or differed in various
ways) in other kinds of revolutions. By so doing, she greatly shifted our prior beliefs about what
were the important causal elements associated with various particular revolutions.
It is not necessary to assume a deterministic world, or to think of this in strict Popperian
terms of “refuting” existing theories. It is sufficient to assume a Bayesian perspective that
essentially asks: what would you bet that you would find certain conditions in a given case before
a CHA of that case, and would you make the same bet after a CHA of that case? If the answer is
no, then the CHA has changed the prior belief and made a contribution to our knowledge.
Process-tracing and Congruence Testing. We still need to ask, however, how does CHA
persuade scholars to change their prior beliefs? What method of argument or inference provides
reliable statements regarding causal relationships in one particular case, or cases? There are two
main approaches used in CHA: process-tracing and congruence testing (Bennett and George 1997,
George and Bennett 1997).
Process-tracing consists of analyzing a case into a sequence (or several concatenating
sequences) of events, and showing how those events are plausibly linked given the interests and
situations faced by groups or individual actors. It does not assume that actions always bring their
intended consquences, only that actions are understandable in terms of the knowledge, intent, and
circumstances that prevailed at the time decisions were made. Process-tracing involves making
deductions about how events are linked over time, drawing on general principles of economics,
sociology, psychology, and political science regarding human behavior.
For example, in making my argument (Goldstone 1991) that population growth in agrarian
bureaucracies led to revolution, I did not proceed by showing that in a large sample of such states
there was a statistically significant relationship between population growth and revolution—that
would not be possible given the focus on only four main cases (England, France, the Ottoman and
Chinese Empires). Rather, I sought to trace out and document the links in the causal chain
connecting population growth to revolutionary conflict.
I began by noting that when demand exceeds supply, prices generally rise; thus population
growth in agrarian economies with technically limited growth capability is likely to produce
inflation.i Drawing on the economic history literature, I documented this association between
population movements and price movements over a century or more in each of the four cases. I
then pointed out that inflation will wreak havoc with state finances if states have fixed nominal
 6

incomes, forcing states to either raise taxes or increase their debts. While many factors influence
state finances, including wars and trade, I was able to show that in periods of rising prices, these
agrarian bureaucracies were more likely to depend on unsustainable expedients such as selling off
assets, imposing forced loans, currency devaluations, defaults, and/or efforts to impose new or
unconventional taxes. These expedients not only generally failed to fully restore states’ fiscal
health; they also imposed burdens on elites or popular groups, leading to potential conflicts.
In regard to elites, I observed that population growth will mean greater competition for elite
places, at the same time that inflation provides enrichment for some elites and groups while
eroding the incomes of others; this was documented with evidence on higher rates of social
mobility—both upward and downward—among elites during periods of sustained population
growth. Thus the state’s demands for greater revenue will be addressed to elites who are
themselves in turmoil and divided by heightened up-and-down social mobility. At the same time,
urban workers will face falling real wages as rising numbers of workers limit wages, and peasants
will be forced to pay higher rents or go landless, as increased numbers compete over a fixed supply
of land. It was also possible to document patterns of falling real wages and rising rents and
landlessless among the peasantry during the periods of population increase.
The combined results of these trends was that as rulers increased demands on elites and
popular groups, intra-elite competition tended to divide the elites into competing factions,
paralyzing the state and often leading some elite factions to seek popular support against the
regime. This situation in turn produced opportunties for distressed urban and rural groups to
mobilize in protest. In such circumstances, a critical event such as a state bankruptcy, military
defeat, or famine—which a healthy state would readily withstand—triggered escalating state-elite,
intra-elite, and elite-popular conflicts that combined to create a revolution.
This comparative historical analysis is certainly not made without quantitative analysis or
large amounts of statistical data. In fact I used hundreds of statistical and historical observations
regarding prices, wages, rents, population, the fate of families, declarations of support or conflict
by leading actors, and sequences of events to demonstrate that in England, France, Turkey, and
China from 1500 to 1750, those decades over which population mounted were marked by rising
prices, increasing state debts and fiscal difficulties, repeated conflicts over taxes, heightened elite
mobility, falling real wages, and growing political tensions. In contrast, those decades in this
period in which population stagnated or declined were marked by stable or falling prices, rising
state real revenues, sharply reduced elite mobility, rising real wages, and political settlements.
However, the data is used primarily to establish the validity of trends and relationships within
complex cases that are explored in considerable detail, which are then carefully compared for
similarity or differences, and not to establish the statistical significance of such trends or
relationships across large numbers of relatively unexplored cases.
It is this process-tracing, in which many hundreds of observations are marshalled to support
deductive claims regarding linkages in a causal chain, on which CHA rests, not simply causal
inference from associations among macro-conditions. Where generalization across cases is gained,
it is not in universal generalization to all agrarian bureaucracies at all times (which is neither
claimed nor demonstrated), but rather simply in showing that for a finite set of cases, where
convergence was not previously expected, a common pattern of causation of state upheavals in fact
can be found.
 7

Congruence-testing provides the basis for claims regarding “common patterns.”

Skocpol demonstrated by careful exploration of her cases that the roles of state crisis, elite revolt,
and popular mobilization formed nearly the same, or “congruent,” patterns in the French, Russian,
and Chinese revolutions. In doing so, she overturned the idea that the causal structure of the latter
two communist revolutions was different from that of the liberal French revolution. I showed that
in Britain, France, the Ottoman Empire and China, prior to periods of massive rebellion and state
overthrow, there is substantial evidence for extremely similar or congruent, patterns of sustained
population growth, price inflation, state fiscal distress, heightened elite mobility, falling wages,
and peasant landlessness. The congruence between the patterns found in the Asian empires with
the patterns established for the western monarchies provides a strong alteration in prior beliefs that
the western revolutions and Asian rebellions were different phenomena brought on by different
causes, and a powerful challenge to the idea that the western revolutions resulted from social
conflicts found only with the rise of western capitalism.
The goal of congruence testing is therefore not to establish universal generalizations across
a broad (and perhaps unconnected) range of cases. Rather, congruence testing uses the fruits of
process tracing to challenge and improve our understanding of how particular cases of interest are
related or different. If our goals are to understand how democratic and communist revolutions
were similar or different; to determine if western revolutions were truly distinct from Asian
rebellions; or to examine why military autocracies emerged in some Latin American or European
countries and not others, then congruence testing among selected cases can bring us closer to our
goals. Such comparative case studies provide solid building blocks for theories of social change
and development, allowing us to work toward larger generalizations gradually by incorporating
additional cases and patterns.
Both process-tracing and congruence testing are generally carried out through what Robin
Stryker (1996) has described as “strategic narrative.” Strategic narrative differs from a
straightforward narrative of historical events by being structured to focus attention on how patterns
of events relate to prior theoretical beliefs about social phenomena. Strategic narrative selects its
elements in response to a clearly articulated theoretical backdrop, and focuses on empirical
anomalies relative to such theory. Sometimes such narratives are designed to demonstrate the
strict causal connections between elements in a sequence of events (Abbott 1990, 1992; Griffin
1992); sometimes such narratives are designed to demonstrate the impact of contingency or
path-dependence in producing divergent or unexpected outcomes (Sewell 1996, Mahoney 2000b,
Goldstone 1999, Haydu 1998). Either way, the narrative marshalls evidence that particular
sequences or patterns unfolded in a particular way, for particular reasons, and relates that data to
our confidence in existing theory.
Neither process-tracing nor congruence-testing provide a means of generalizing to a
universe of cases beyond the finite samples. Instead, their goal is to test, challenge, and shift prior
beliefs about the cases under examination. Of course, if the prior beliefs are very strong and
general, and the results of process-tracing and congruence-testing shatter those beliefs, they may
be extensively modified or changed, to the point where the shift affects people’s expectations
about other cases as well.
The overall development of CHA in any particular field thus presents a research program
(Lakatos 1983) with the following features. At any one time, a finite number of cases deemed to
 8

be of interest to a research community are under investigation. Using process-tracing, scholars

seek to uncover casual sequences that produced the results or cases of interest. Using
congruence-testing, scholars make claims about the number of cases that “fit” a particular causal
sequence or pattern (or “model.”) This then sets up strong prior beliefs about which cases are
accurately explained by particular models.
The research program can then make progess in several ways. Further congruence-testing
of additional cases can extend the number of cases that are seen as fitting a particular model.
Where the fit is poor, new process-tracing in those additional cases may reveal new patterns,
leading to revisions of the old model or the need for new models. Moreover, new process-tracing
in previously-analyzed cases may challenge the validity of received models, leading to their
revision, modification, or abandonment. Throughout, the process is one of operating on finite
numbers of cases chosen to investigate how their analysis shifts prior beliefs (Bayesian inference),
not one of sampling cases in order to establish patterns in a general population (standard or
‘normal’ statistical inference). This method remains respectful of the historical context and details
of each case; advances generalizations slowly and circumspectly in stages,
and does not presuppose a deterministic view of the world. However, it constantly challenges
prevailing wisdom by confronting it with empirical details derived from further process-tracing of
old or new cases, and testing the congruency of those cases against prevailing expectations.
CHA is thus capable—in the best scientific sense—of extending and cumulating knowledge, while
rejecting theories that are inconsistent with empirical evidence.
For the specific field of revolutions, I will show in the following sections that this research
program has yielded steady and cumulative progress. A common indictment of theory
development in sociology is that it is often a matter of shifting fads—modernization, Marxism,
feminism, post-Modernism—with little cumulation or greater understanding of the empirical
world. Nonetheless, I believe that theories of revolution have made remarkable progress in
understanding the causes, processes, and outcomes of revolutions. Naturally, this progress
involves controversies, the falsification and abandonment of some theories, and the supersession
and incorporation of others. What is important, however, is that these shifts in theory have been
driven by the confrontation of theory with empirical events. This confrontation has led to
successive modification of theories, the introduction of new concepts, and the development of new
approaches. The result is that theories of revolution today address more cases, with more fidelity
to historical events, and with a greater grasp of the various elements that comprise revolutions,
than ever before.

DEFINITIONS OF REVOLUTION
We might first ask—what are revolutions? Again, a large-N statistical approach would require a
universal definition, so that the sample could be consistently coded to reflect “revolution” as a
dependent variable (to study causes) or as an independent variable (to study outcomes). However,
since CHA does not proceed by sampling and seeking significant associations between variables,
but instead by rigorously examining a finite number of cases, revolution can be roughly or even
implicitly defined through the initial selection of cases. Then, as the research program develops,
additional cases can be examined, and the initial definition modified or expanded according to
which features of the cases prove to be of lasting interest. In fact, the definition of revolution has
 9

developed over time in response to new empirical information, and to changes in our
understanding of the nature and possibilities of revolutionary events.
Throughout most of history, sudden changes in government were treated by historians and
political analysts as cyclical phenomena (hence the term “revolution.”) These changes might
involve cycling through different kinds of regimes (democracy, oligarchy, tyranny) as discussed by
Plato and Aristotle. Or these changes might involve an alternation of opposing factions or parties
in power, as occurred in the Renaissance Italian city-states and the seventeenth century British civil
wars. Yet by the Enlightenment of the eighteenth century, the notion of continuous improvement
had come into wide use and writers such as Thomas Paine viewed certain changes in
government—replacing traditional authorities with new institutions devised by reason—as part of
the march of progress. From the time of the American Revolution of 1776, and even more the
French Revolution of 1789, “revolution” came to be associated with the replacement of obsolete
governments with newer, more rational regimes. Revolution was henceforth defined as a
progressive and irreversible change in the institutions and values that provided the basis of
political authority.
The French Revolution of 1789, and even more the Russian Revolution of 1917, also
provided the image of popular uprisings and violence against authorities as key aspects of
revolution. From 1789 to 1979, the prevailing definitions of revolution thus incorporated three
elements: a dramatic and progressive change in a society’s values and institutions, mass action, and
violence (Huntington 1968, Skocpol 1979).
However, from the 1970s through the 1990s, it became increasingly clear that these
definitions were inadequate to the full range of events being studied by scholars of revolutions.
The revolution that overthrew the Shah of Iran in 1979 featured general strikes, and massive street
protests in Tehran, Tabriz, Qom, and other major cities, but little or no popular violence against
authorities. Such violence as occurred came mainly from the Shah’s forces firing on protestors, or,
after the Shah had fallen, in the attacks on Bahai, Jews, and those Iranians whose behavior
appeared to violate the proscriptions of increasingly strict Islamic authorities (Arjomand 1988).
The “people power” revolution that overthrew Marcos in the Philippines in 1986, and the
anti-communist revolutions in the Soviet Union and Eastern Europe, showed considerable mass
protest but startlingly little violence, to the point where some observers wondered if the latter
events should be considered revolutions at all (Garton Ash 1989).
These recent cases prompted scholars to take a second look at the role of violence in
historical cases. It was soon recognized that the pattern seen in Iran was not exceptional, and that
in general far more violence is committed by new revolutionary regimes after seizing power— in
order to consolidate their control of society, or in civil wars between revolutionaries and
counter-revolutionaries—than is committed by revolutionary actors against authorities before the
latter are overthrown. Thus in the French Revolution, it was the Terror and the actions to crush the
counter-revolutionary movements in the Vendée that produced the greatest number of deaths; in
the Russian Revolution it was the Civil Wars and Stalinist purges, not the workers’ insurrections
or peasant uprisings, that produced the greatest death toll. Revolutions are therefore no longer
defined in terms of violence against authorities. Instead, the role of violence in revolutions is now
seen as complex and contingent, with old regimes, revolutionary oppositions, and new
revolutionary regimes all capable of violence or non-violence to varying degrees at different times.
 10

In addition, many episodes that scholars study as being revolutionary have not led to the
permanent transformation of institutions and values. Efforts to remake society often fall short, and
even revolutions that take power sometimes are superseded by restorations of the old order.
After the fall of Napoleon, many of the key institutional changes of the French Revolution were
overturned. The monarchy was restored, as were the prestige of the nobility and the Catholic
Church. Studies of local wealth holding have shown that, despite the Revolution’s vocal attack on
the aristocracy, the families that held leading positions throughout France before the Revolution
were, in the main, the same families that held leading positions in the 1840s (Forster 1980). The
English Revolution of 1640 ended the monarchy, the monopoly of the Church of England, and the
House of Lords; yet all were restored in 1660 and kept their power for generations. The European
Revolutions of 1848 were largely defeated after brief but inspiring bursts of energy. And most of
the guerrilla movements in Latin America in this century, such as the Shining Path movement in
Peru and the Farabundo Martí National Liberation Front in El Salvador, though clearly episodes of
revolutionary conflict, failed to transform their societies. Thus success in changing values and
institutions, like violence, is now seen as a matter of degree and a contingent outcome in
revolutionary events, rather than part of the definition.
A more recent definition of revolution guided by these insights focuses on efforts to change
values and institutions and on mass action, but without insisting on revolutionary success or
violence. In this view, revolutions are “an effort to transform the political institutions and the
justifications for political authority in a society, accompanied by ... mass mobilization and
non-institutionalized actions that undermine existing authorities” (Goldstone 2001, p. 142). This
definition includes both successful revolutions and unsuccessful yet clearly revolutionary
challenges to state authorities. It excludes both coups and civil wars that change only rulers, not
institutions or the structure of political authority (such as most African and Latin American
military coups). It also excludes reform movements that achieve changes without mass
mobilization or attacks on existing authorities (such as the Prussian Reform Movement of the early
19th century, or the American “New Deal” under Roosevelt). It suggests that the key defining
element of revolution is always the attack on authorities’ right to rule, and mass involvement in
that attack; but that the level of success and violence involved are contingent factors to be
explained, not assumed as part of the definition.
In short, the very definition of revolutions—a key part of any theory—has changed over
time in response to careful case studies of how revolutionary events have unfolded.
The comparative and historical study of revolutions extends from classical times to the
present. Accounts of sudden changes in government and regime collapse were collected and
analyzed by Plato, Aristotle, Thucydides, Tacitus, Montesquieu, Machiavelli, de Tocqueville,
Marx, and Trotsky, among many others. However, it is convenient to focus on the cumulation of
knowledge in this century, beginning with the “natural history” school.

THE NATURAL HISTORY OF REVOLUTIONS

Scholars working in this perspective aimed to study revolutions the way biologists studied natural
history (that is, the history of life), namely by gathering the specimens, detailing their major parts
and processes, and seeking common patterns.
These scholars—notably Edwards (1927), Pettee (1938), and Brinton (1938)—compared
 11

sequences of events in the English, American, French, and Russian Revolutions, and traced
distinct patterns that reappeared across these cases. They laid out a scheme of events that they
believed all major revolutions followed, ranging from causes to outcomes. In brief, they noted that
prior to a revolution, the bulk of the "intellectuals"—journalists, poets, playwrights, essayists,
teachers, members of the clergy, lawyers, and trained members of the bureaucracy—cease to
support the regime, write condemnations, and demand major reforms. Just prior to the fall of the
old regime, the state attempts to meet this criticism by undertaking major reforms. However, the
actual fall of the regime begins with an acute political crisis brought on by the government's
inability to deal with some economic, military, or political problem rather than by the action of a
revolutionary opposition.
After the old regime falls, even where revolutionaries had united solidly against the old
regime, their internal conflicts soon cause problems. While the first group to seize the reins of state
are usually moderate reformers, who often employ organizational forms left over from the old
regime, more radical centers of mass mobilization spring up with new forms of organization, (such
as the Jacobin clubs in France, or Leninist vanguard parties). These radical groups challenge the
moderates, and eventually gather popular support and drive them from power. The great changes
in the organization and ideology of a society in revolutions occur not when the old regime first falls,
but when the radical, alternative, mass-mobilizing organizations succeed in supplanting the
moderates.
However, the disorder brought by the revolution and the implementation of radical control
usually results in forced imposition of order by coercive rule, often accompanied by a period of
state-imposed terror. The struggles between radicals and moderates and between defenders of the
revolution and external enemies frequently allow military leaders to take commanding, even
absolute, leadership of the new revolutionary regimes (for example, Cromwell, Washington, or
Napoleon). Eventually, the radical phase of the revolution gives way to a phase of pragmatism and
moderate pursuit of progress within the new status quo, but with an enlarged and more centralized
state.
The natural history school provided a clear and fairly comprehensive picture of the
preconditions, dynamics, and outcomes of the revolutionary process. Not only did this schema fit
rather well to the cases for which it was developed; it also fit remarkably well to cases that arose
decades later, such as the Islamic Revolution in Iran in 1979.
However, despite uncovering persistent and recurring patterns in the course of major
evolutions, this approach failed to present a convincing reason why revolutions should occur at
certain times and places but not others. The defection of the intellectuals, and the initiating
political crisis, simply appear as historical contingencies. The fall of the old regime then follows
as a matter of course.
Some political scientists and sociologists, however, noted a further temporal pattern in the
course of revolutions. They attributed the spread of revolutions through history to the process they
called “modernization.” In particular, they saw popular discontent with “traditional” regimes
bringing conscious efforts to modernize societies by means of revolution.

MODERNIZATION AND REVOLUTIONS

From the 1950s through the 1970s, there was a growing belief that most important features of
 12

long-term social change could be explained by focusing on a universal process of transition

described as “modernization” (Parsons 1966, Almond 1973). In this view, most societies
throughout history have been “traditional”—vesting political authority on the basis of custom and
heredity; having economies in which most activity was subsistence-producing, non-market, and
pre-industrial; and with social hierarchies that stressed conformity to traditional values and
privilege. In contrast, “modern” societies have political authorities (whether democratic regimes
or dictatorships) that are deliberately and recently established to provide efficient and bureaucratic
rule, economies that are market-oriented and industrialized, and social hierarchies based on
universal citizenship and achievement. Although this transition might sometimes be smooth,
modernization theorists of revolution argued that often the transition was jarring or uneven, with
traditional authorities trying to hang on to power in societies that were already partially
modernized; the imbalance then would produce a revolutionary situation and attacks on those
authorities.
Several different theories of revolution were derived from the basic modernization
framework. Chalmers Johnson (1966) suggested that in many developing countries, trade, travel,
and communications tended to break down traditional values, while economies and political
systems changed far more slowly, leading to a dysfunctional breach between modernizing values
and traditional social and political organization. Samuel Huntington (1968) focused more tightly
on political demands. He noted that developing countries often provided increased education and
participation of their people in the market economy while still maintaining a closed, traditional,
political system. Unless political participation was also expanded, he believed this imbalance
would lead to explosive demands for changing the structure of political authority. Ted Robert Gurr
(1970) claimed that even if economies were growing and political systems were developing,
imbalances could still arise. If people expected still greater change to occur than they actually
experienced, they would feel “relative deprivation.” This feeling of frustration could lead to
demands for still more rapid and extensive changes, fueling a revolutionary transformation of
society.
All these theories were variants of a model in which modernizing changes in traditional
societies disrupted established patterns and expectations regarding social and political life.
This may sound similar to Marx’s assertion that economic changes lead to revolutionary class
conflicts. However, Johnson, Huntington, and Gurr all eschewed class analysis, and argued that
economic change had no primacy—changes in values, education, and expectations were part of the
social disruption, not merely economic change. These theorists thought more in terms of a conflict
between governments and society at large, spurred by unbalanced social change, rather than
conflicts among classes, and kept their distance from Marxist theories.
In addition, because the modernization theory suggested there were universal processes of
transition that applied to all societies, it seemed an excellent candidate for large-N, global
statistical studies that could demonstrate a significant association between the strains of transition
and political violence. Several such studies were undertaken (Gurr 1968; Feierabend, Feierabend,
and Nesvold 1969, Hibbs 1973). However, they bore mixed results. It proved nearly impossible
to measure the independent variable in a way that would discriminate among competing versions
of modernization theory—how do you determine if political violence is associated with a change
in people’s values, or with a change in their desire for political participation, or with discontent
 13

with economic growth relative to their expectations? To get at these issues on a global basis would
have required comprehensive opinion polling in every country in the world at a time just prior to
their outbreaks of revolution. The large-N studies therefore used easier-to-measure proxies for
progess toward modernization, such as energy use, industrial output, and consumption. While
these studies did show that countries that were “less” modern had higher levels of political
violence than those countries that were “more” modern, there were three major problems in
interpreting these results.
First, it was never completely clear what were the causal mechanisms that linked changes
in material conditions and revolution. On the one hand, was it those individuals who were feeling
the most “deprived” or most left behind that actually created revolution? Or, on the other, was it
those individuals who had become most modernized in their values and thus were most discontent
with still partially traditional institutions? Or perhaps it had nothing to do with values or
discontent at all, but was a matter of elite competition for the fruits of growing economic output,
or loss of regime leverage over elites as the economic structure of society changed. For the most
part, modernization theories paid little attention to which elites, or which popular groups, were
most involved in conflict, and thus could not answer these questions.
Second, it was not clear that the dependent variable—“political violence”—was the same
as revolution. Many countries had high and persistent levels of political violence, but without their
governments succumbing to revolution (e.g. Colombia, South Africa). Moreover,
some revolutions, such as those in Iran in 1979 and the Philippines in 1986, came about mainly
through mass defections from the regime and urban street demonstrations, with relatively brief and
mild intervals of political violence producing major changes in government. Furthermore, in some
major revolutions, such as the Mexican Revolution of 1911 and the Russian Revolution of 1917,
all the major political violence occurred in the civil wars of 1911-1920 and 1917-21, after the old
regime had already been overturned. In these cases, revolutions apparently caused the political
violence, rather than the reverse. Huntington (1968) noted that some revolutions involved long
periods of guerilla warfare in the periphery of society before the fall of the old regime, while in
other revolutions the central regime experienced a sudden collapse. Clearly, a focus on the
correlates of political violence were telling us very little that was useful about the causes, patterns,
and dynamics of major revolutions.
Third, a surprising but consistent finding was that the best predictor of political violence
was almost always the presence of prior political violence. In other words, some
countries—regardless of their level of modernization—were simply more prone to political
violence than others. This finding called into question the whole premise that all countries could
be treated similarly, with differing levels of political violence varying mainly according to levels
of modernization. Instead, it increasingly appeared, even in the work of modernization theorists
of revolutions, that different countries were different in important ways; and that revolutions
themselves were different in how they unfolded, their levels of violence, and which elites and
groups were involved.
These points emerged most clearly in the work of two comparative-historical scholars
whose work—carefully grounded in CHA of specific countries and conflicts—criticized the
modernization approach.
Charles Tilly (1973, 1978) went directly after Huntington and Gurr, arguing that popular
 14

discontent could never by itself launch a revolution. In this he elaborated on the insights of Leon
Trotsky, who had organized and led the Red Army for Lenin’s Bolsheviks in the Russian
Revolution, and who wrote a brilliant, theoretically astute chronicle of the Revolution (Trotsky
1959). Trotsky observed that popular misery and frustration never lead to revolution by
themselves; if they did, he wrote, the masses would be in revolt at all times and everywhere. As
Lenin (1973) insisted, it required a revolutionary organization to channel popular discontent into
political grievances and to mobilize the masses for revolutionary action. Trotsky further observed
that in the course of a revolution, a critical period of contention occurs between revolutionary
forces and the government, during which the government is still in command but is losing popular
support, while the revolutionary forces are gaining support and resources but not yet in power. The
resolution of this period of “dual power” is the key to the revolution; only if the revolutionary
forces are able to build their strength and defeat the forces of the government can a revolution
triumph.
Tilly (1978) constructed from these observations a more formal theory of revolutionary
mobilization and contention. He argued that only when organized contenders emerge, capable of
mobilizing popular support against the government and defeating the government for control of
society, can revolution occur. Modernization might lead to the emergence of such contenders, but
it would not automatically provide for their success. That required action by the contenders to
assemble revolutionary forces, to acquire political, military, economic, and other resources to
confront the government, and then to defeat it. Only where modernization put this particular
sequence in motion would revolution follow.
A further challenge to modernization theories of revolution came from the work of
Barrington Moore, Jr. (1966). Moore argued that while modernization might be the fate of all
nations, there were in fact several distinct paths to modernization, with very different economic
and political outcomes. Moore highlighted a democratic path, founded on liberal revolutions such
as those that occurred in England and France; a communist path, founded on Leninist revolutions
such as those that occurred in Russia and China; and a fascist path, founded on nationalist
revolutions such as those that occurred in Facist Italy, Nazi Germany, and Meiji Japan.
Which path a particular nation took depended, as Marx had insisted, on their pattern of
class relations. Yet while prevailing Marxist theory had simplified those classes into landlords,
peasants, workers, and capitalists and postulated a similar historical pattern of class conflict in all
societies, Moore argued that patterns of class power and conflict could vary. Tracing out divergent
patterns among several cases, Moore showed that the emerging capitalist elite might unite with the
older landlord elite in order to suppress the masses and create facism; in other cases it might unite
with the masses to overturn the landlords and create democratic capitalism; in still others it might
remain so weak that it is overwhelmed by a popular revolution that creates socialism. Which path
a country followed depended not only on initial conditions, which were similar in all his cases circa
1500; it also depended on the precise sequence of events which followed. Where agriculture
turned early to capitalist organization and dependence on free wage labor, while industrialization
came later, then that labor force could be readily recruited to industry, creating a common interest
among wage laborers and industrial employers in opposing political control of the state by
landlords. This path led to bourgeois revolutions against aristocratic states, as in England, France,
and the United States (where Moore sees the 1860s Civil War as a revolution by northern
 15

industrialists to overturn political control of the national government by southern planter

landlords). Where agriculture continued to rely on coerced labor until industrialization began, then
industrialists would need to ally with landlords in order to gain access to a labor force, leading to
a conservative alliance that would prevent revolutionary mobilization from below, and instead
produce fascist regimes, as in Germany, Italy, and Japan.
Finally, where industry emerged very late, weak, and highly dependent on the state, while
landlords continued to control the state and a coerced rural labor force, revolutionary parties could
emerge as the main opponent of the landlords, leading to communist revolutions that would both
overturn the landlord regime and seize control of industry, as in Russia and China.
Moore’s model of divergent, path-dependent evolution leading to different kinds of
revolutions marked several major advances over both natural history and earlier modernization
theories. First, Moore incorporated more cases, indeed some cases which many scholars had not
treated as revolutionary situations (Nazi Germany, Fascist Italy, the United States during the Civil
War). Second, his demonstration that class conflicts and alliances were varied and contingent
overturned a fixed Marxist view of uniform social and political development, replacing it with a far
richer and more flexible view. Third, his demonstration that different kinds of revolutions had to
be distinguished not only according to their causes, but also by their different outcomes, laid out
multiple trajectories of revolutionary conflict and change, and offered far greater accuracy than
theories that pointed only to a modernization or strengthening of state power as the general
outcome of revolution.
By the late 1970s, theories that simply equated risks of revolution with transitions to
modernization were in retreat. Tilly’s mobilization theory gained acceptance as a key element in
any explanation of revolutions and other kinds of social protest. Moore had laid down a critical
challenge to modernization theory by demonstrating that modernizing revolutions were not all of
one sort, but had clearly distinct outcomes reflecting different kinds of transition to modernity.
However, additional process-tracing of further historical cases, and of newly emerging
revolutionary conflicts in the Third World, revealed that they had surprisingly little consistency
with regard to modernization; indeed perhaps no relationship to modernization at all. The Latin
American Revolutions of Independence, from 1808 to 1825, occurred in countries with few
measurable signs of economic or value modernization. More recently, the separatist revolts in the
Congo (Zaire) in the 1960s, the Sudanese Civil Wars, and the Cambodian Revolution occurred in
states where modernization seemed to have hardly begun. In contrast, the revolutions to found
communism in Cuba and in Eastern Europe after World War II occurred in countries that,
compared to most of the world and certainly compared to England in the 1640s or France in 1789,
were already quite well modernized. The transition to modernization per se thus seemed to be a
weak reed on which to rest any analysis of why revolutions occurred in certain societies at
particular times.
Modernization theories of revolution thus were coming under fatal fire. There was a clear
need to somehow improve upon modernization theory as a causal force in creating revolutionary
situations, while also incorporating Tilly’s theory of mobilization and Moore’s insights regarding
the variety of revolutionary outcomes. This challenge was met by the next major step forward in
theories of revolution, the social-structural theory of Theda Skocpol.
 16

SOCIAL STRUCTURAL THEORY—AND ITS CRITICS

In 1979, Theda Skocpol published States and Social Revolutions, which was immediately
recognized as a landmark in the theory of revolutions. Although it was preceded by several other
books which took a somewhat similar approach (Wolf 1970, Paige 1975, Eisenstadt 1978), it was
considerably more successful than any other works in resolving the dilemmas in revolutionary
theory at the end of the 1970s.
Skocpol did confine her range of coverage, in order to improve on the accuracy of her
theory and cover a number of distinct elements of the revolutionary process. She chose to focus on
the great social revolutions—namely the French, Russian, and Chinese Revolutions. She then
compared these cases with other cases of less extensive or less successful revolutionary change:
the English Revolution, the German Revolution of 1848, the Prussian Reform Movement, and the
Japanese Meiji Restoration. Skocpol also limited her methodological and theoretical range. She
worked almost entirely from comparing historical narratives, and focused on revolutions to the
exclusion of any other kind of social protest or collective action. Nonetheless, within these bounds,
she was able to make powerful advances in exploring new (or heretofore unappreciated) elements
of the revolutionary process, and creating a theory with far greater correspondence to the facts of
these great revolutions than was offered by any modernization-based theories.
Skocpol’s first step, following the lead of Wallerstein (1974), was to take modernization
out of the individual country, and move it to the world-system as a whole. Skocpol argued that it
was not the impact of modernizing change within a country that disturbed political stability; rather
it was international military and economic competition between states at different levels of
modernization that created destabilizing pressures. A regime that faced military pressure or
economic competition from a neighboring state at a more advanced level of development would
have no choice but to seek to increase its own resources by restructuring its financial, military, and
economic system. In her cases, eighteenth century France faced pressure on its trade and colonies
from more advanced Britain; Tsarist Russia was overwhelmed by German military might, and
China had to deal with a long stream of imperialist incursions from European powers and Japan.
Skocpol pointed out that if a state in this situation encountered resistance to reform from powerful
elites in its own country, this conflict could trigger a political crisis and precipitate a revolution.
Oddly, international pressures had remained largely absent from prior theories of revolution.
Adding this new element made revolutionary theory better correspond to observed conditions
facing states, and at the same time helped resolve the conundrums presented by modernization
theories that focused only on internal change.
Skocpol’s second innovation was to highlight the conflicts between state rulers and a
country’s political and economic elites. Prior revolutionary theories had more or less taken for
granted that state rulers and national elites would be natural allies. Even in Moore’s nuanced
accounts of class conflict, the question was which combination of elites would take state power,
not how rulers and elites might themselves come into conflict. Elite theories of politics, such as
those of Pareto and Mosca, spoke of the rotation of power among different elites, and conflict
between them, but not of an autonomous ruler facing off against elite groups. In pointing out that
state rulers have their own agendas and resources, and that the need for rulers to raise more
resources could come into conflict with elite claims to those resources, Skocpol presented a
scarcely-appreciated source of revolutionary dynamics. Harking back to Tocqueville, this aspect
 17

of her theory of revolutionary processes corresponded far better with the observed conflicts
between the French monarchy and the elite Parlements that led to the calling of the Estates-General
in 1789, or with the observed conflicts between the Chinese Imperial rulers and the warlords who
took over China after 1911, than did any modernization theories.
If conflict between a state ruler needing to pursue reforms and elites who resisted them
could precipitate a crisis, Skocpol nonetheless realized that this would not suffice for a revolution.
Following Tilly’s mobilization theory, Skocpol argued that an organizational framework was
needed that would allow popular groups to take advantage of conflict and crisis at the political
center. This could be a Leninist party, as was the case with the Communist Party in China.
However, Skocpol also pointed out that where peasants had their own village organizations and
leaders, who were accustomed to taking decisions for the village over matters such as taxation or
the distribution of land, this village structure could form the basis for local rural uprisings. In
countries like eighteenth century France and twentieth century Russia, strong local peasant villages
could therefore act when the political center was paralyzed by conflict, and launch rural uprisings
that would undermine the political and social order.
Skocpol thus presented her theory as laying out three conditions in a country’s social and
political structure that were necessary and sufficient for a great social revolution: international
pressure from a more advanced state or states; economic or political elites who had the power to
resist state-led reforms and create a political crisis; and organizations (either village or party) that
were capable of mobilizing peasants for popular uprisings against local authorities. Countries
lacking one of these conditions might have less extensive or unsuccessful revolutionary events, but
they could not undergo great social revolutions.
Skocpol did not rest with an analysis of the causes of revolution. She continued to consider
the variation in outcomes. She argued that social structure also constrained outcomes. Where the
sources of wealth in the economy were widely dispersed (such as land and small shops and
artisanal establishments), no revolutionary regime could seize control of the economy. In such
conditions, revolutions would produce new regimes that protected private property (as in England
in the 1640s and France in 1789). However, where industrialization had produced great
concentrations of economic power in the shape of large factories, electrical power grids, and
railroads, a revolutionary regime could (and most likely would) seize those resources in its drive
for power, creating a communist regime with the government commanding the economy. Here,
differences in the level of modernization drive differences in revolutionary outcomes,
respecting Moore’s insights while incorporating modernization theory into an account of
revolutionary outcomes, rather than causes.
Others authors also pointed out that revolutionary outcomes were constrained by
international economic structures. Eckstein (1982) showed for a number of Latin American
revolutions that despite variations in economic policy, revolutions did nothing to change the basic
position of poor countries dependent on raw material exports in the world economy.
Skocpol’s three-factor theory of revolutionary causes and her account of outcomes
appeared as a simple and elegant solution to the problems of prior theory. Even though she
covered only a handful of cases, she successfully illuminated new elements—international
pressures and state/elite conflicts—that appeared to have wide applicability to other cases.
Her treatment of modernization in terms of international competition and constraints on outcomes
 18

seemed far more accurate than the claims of prior modernization theories of revolution.
Skocpol’s argument that she had specified a set of necessary and sufficient conditions for
social revolutions did not rest simply on showing that those causes were present in her major cases.
Indeed, as critics of Skocpol have shown (Nichols 1986), her cases in fact are not identical with
regard to causation. Where France and China faced moderate levels of international pressure and
strong resistance to change from domestic elites, Russia faced extremely high levels of pressure in
World War I, strong enough to disable the government even with very weak resistance from
domestic elites. Skocpol granted this; her argument was that she had identified common
conditions that, in various combinations, produced the major revolutions of modern history, both
liberal and communist (Skocpol 1994). Through process-tracing and congruence testing, she had
demonstrated that these conditions better explained the details of these major cases than did prior
theories.
Skocpol’s social-structural approach became the dominant theory of revolutions for the last
two decades of the twentieth century. Yet almost as soon as it was published, the theory was
challenged by new empirical events. Revolutions in Iran and Nicaragua in 1979, in the Philippines
in 1986, and in the communist countries of Eastern Europe and the Soviet Union raised new
questions about the causes, processes, and outcomes of revolutions.
The Iranian revolution soon generated a debate on the applicability of Skocpol’s theory
(Skocpol 1982). The Iranian state under the Shah, far from being pressured by competition from
more advanced states, was the most powerful and advanced state in the Persian Gulf region, and
had the strong and committed support of the world’s leading superpower, the United States.
Yet the Shah’s government nonetheless provoked the opposition of almost all sectors of the Iranian
elites. While there were economic and political reasons for this opposition, a major role was
clearly played by ideological conflict, specifically between the Shah’s advancing of
Westernization and elites defending traditional Islamic practices and laws. Skocpol’s theory
allowed no role for such ideological roots of state crisis, stating that revolutions are not made, but
come as a result of external military or economic pressures. In Iran, the revolution did appear to
be made by the opposition, and the state crisis seemed to revolve around the escalation of
ideologically-motivated conflicts (Kurzman 1996, Rasler 1996, Parsa 2000).
Skocpol’s theory of revolutionary mobilization also began to seem deficient in light of new
events. The Iranian revolution had no peasant revolts, whether party-led or based in rural village
organizations. The Philippine Revolution also had no peasant revolts—or rather the revolutionary
party that organized rural rebellion played little or no role in driving the dictator Marcos from
power. In both Iran and the Philippines it was overwhelmingly urban protestors who provided the
popular base for the revolution (Gugler 1982). Shifting attention to urban crowds made it clear that
Skocpol had perhaps underestimated the importance of urban protest in her cases of France and
Russia, where the critical events in Paris and St. Petersberg were downplayed relative to peasant
revolts. Urban revolt was even more undeniably the story of popular protests in the overthrow of
communism. Here, neither vanguard parties nor any other formal organizations seemed to play a
major role in popular mobilization. Virtually spontaneous protests in Leipzig, Prague, and
Moscow, erupting from little more than a public call for protests, grew until the communist
regimes lost the heart to suppress them and collapsed (Opp, Gern, and Voss 1995; Oberschall
1994a, Urban et al 1997).
 19

Finally, Skocpol’s theory of structural constraints on revolutionary outcomes seemed

incapable of dealing with the emergent and varied patterns of revolutionary change. John Markoff
(1996) has established that for one of Skocpol’s central cases, the French Revolution, the key
outcome of abolishing feudalism was not the result of structural oppositions between the
monarchy’s interests and that of the revolutionary elite; it was the product of an emerging
interaction between bursts of peasant protest and efforts by the revolutionary elite to respond and
contain it. This led to more thorough-going attacks on elite privileges than either the peasants or
elites had demanded or foreseen in July of 1789.
The notion of structural constraints on outcomes was also at odds with the wide variety of
outcomes of revolutions in the 1970s, 1980s, and 1990s. In Iran, revolutionaries erected an Islamic
republic. In Nicaragua, socialist revolutionaries built a state that respected and preserved private
property, and where the vanguard Sandinista party surrendered to democratic elections. In the
Philippines, revolution also led to democracy, as it did in Eastern Europe and the Baltic states. Yet
in Russia and most of the other post-Soviet states, democracy was shaky or absent; and even
though all these states had a high degree of concentration of modern industries, most attempted to
create free-market economies and undo state control. It appeared that the ideals and programs of
the revolutionary regimes, rather than inescapable structural constraints, were dictating the pattern
of revolutionary outcomes (Selbin 1993, Foran and Goodwin 1993).
Despite these difficulties, scholars did not quickly abandon Skocpol’s social-structural
perspective. The key insights regarding the importance of international influences and state-elites
conflicts seemed sound, even if the details of recent events departed somewhat from the pattern of
Skocpol’s historical cases. The treatment of popular protest, it was felt, could be extended to
incorporate urban protests. Thus scholars worked to extend Skocpol’s perspective to cover a
variety of additional cases, including Iran and Nicaragua (Farhi 1990), anti-colonial revolutions
(Skocpol and Goodwin 1989), and guerrilla wars in Latin America (Wickham-Crowley 1992).
Nonetheless, the deficiencies in the social-structural approach grew more apparent,
especially with the collapse of communist regimes. In these cases, especially in the central case of
the Soviet Union, it was not clear that one could even speak of conflict between state rulers and
elites, since there were no independent economic or political elites outside the structure of the
communist party-state. Rather, it was cleavages within the party, setting reformist against
conservative factions of the military and political elites, that led to the fall of the Soviet regime.
New comparative studies of revolution therefore appeared that departed in various ways
from Skocpol’s specific structural theory, while still incorporating many of the elements of her
approach. I offered analyses of revolutions and large-scale revolts in Europe, the Middle East, and
Asia from 1500 to 1800 (Goldstone 1991), in which I argued that international pressures would
only create state crises if states were already weakened by fiscal strain. Moreover, I pointed out
that merely having a conflict between states and elites need not produce a revolution; strong states
could overcome elite opposition, while elites that were firmly united against the state could simply
insist on a change in rulers or reforms. It was only when elites themselves were severely divided
over how to meet those pressures that a paralyzing revolutionary crisis would occur.
Also, while Skocpol had argued that the masses always had reasons to revolt, in these cases
both urban and rural protests clearly had roots in declining economic opportunities. In the cities,
rising unemployment and falling real wages made workers open to revolutionary mobilization; in
 20

rural villages shortage of land and rising land rents similarly disposed peasants to protest action.
Moreover, a formal organization to orchestrate popular action, whether it be a vanguard party or a
village communal organization, was not always necessary. What mattered was that popular groups
saw the state as vulnerable, and were prompted by elites calling for change and claiming that it was
necessary. Popular groups could then call on a wide array of existing social networks—villages,
neighborhoods, workplaces, occupational groups, religious congregations—to mobilize for protest
action. Work by Glenn (1999), Gould (1995), Osa (1997), Parsa (2000), and Pfaff (1996) made it
clear that no one form of formal organization was needed for protest mobilization; rather a wide
variety of formal and informal networks could fulfil that role.
Perhaps the most significant aspect of this comparison of major European revolutions and
Asian rebellions was its demonstration of a different temporal pattern of state upheavals than
would have been predicted by Skocpol’s theory, that international pressures from more advanced
states produced the critical pressures leading to revolutions. Process-tracing and congruence
testing demonstrated that revolutions and major rebellions across Europe and Asia from 1500 to
1800, in states with highly varied levels of capitalist development and across times of both high
and low international conflicts, did in fact follow the pattern of population growth and decline
across those centuries. From 1500 to 1650 population increased, and a wave of revolutionary
upheavals grew from the late 1500s through the 1660s in Europe, the Ottoman Empire, and China.
From 1650 to 1730, population growth receded, and revolutions were rare, despite nearly
continuous and widespread international conflicts. But from 1730 to 1850 population growth
resumed, and revolutionary conflicts again spread all across Europe, the Middle East, and China
from the late eighteenth century through the 1840s and 1850s. Showing how population growth
could affect the key mechanisms of state crisis, elite conflicts, and popular unrest allowed an
explanation of this wave pattern across three centuries. In addition, it extended the coverage of
revolutionary theory to new cases of state breakdown, such as regional revolts in the Ottoman
Empire and the collapse of the Ming Dynasty.
The structural basis of Skocpol’s theory of revolutionary outcomes, as noted above, also
needed revision in light of these cases. The great revolutions in Europe and the major revolts in
Asia seemed to have very similar economic and political origins; they all also led to substantial
institutional changes. Yet in Europe, these changes were often accompanied by a major shift in
values and a condemnation of the old regime as obsolete and in need of replacement; in Asia, these
changes were usually accompanied by little or no shift in values, and a condemnation of the old
regime as flawed for its failure to live up to traditonal values and ideals. I therefore argued that
social-structural conditions alone could not account for the different character of revolutionary
outcomes; rather, the revolutionary breakdown of the former regime offers a uniquely fluid
situation in which the new revolutionary leadership has choices as to how to present itself, and how
to rebuild the social and political order. The ideology of the new revolutionary regime can
therefore be decisive in determining the outcome and trajectory of the post-revolutionary state.
These revisions to Skocpol’s approach thus continued the progress of the theory of
revolutions. The range of historical cases covered now ranged from the absolutist states of Europe
to the great empires of the Middle East and China. The elements of the revolutionary process were
now expanded to include international pressures, fiscal strain, intra-elite conflict, a wide range of
popular protest and mobilization, underlying population pressure on resources, and coordination
 21

between opposition elites and popular protest to produce revolutionary situations, as well as the
pivotal role of revolutionary ideologies in guiding outcomes. Correspondence to empirical events
continued to improve, as various details of state actions, elite conflicts, and popular action could
be grounded in revolutionary theory. The ability of that theory to provide an explanation for the
wave pattern of revolutionary crises all across Eurasia in the late pre-industrial period was
particularly striking.
Skocpol’s emphasis on international factors also opened the way for a major area of
research, including works by Walt (1996), Halliday (1999), Armstrong (1993) and Snyder (1999)
showing how the origins, development, and outcomes of modern revolutions were affected by
international forces, and how international relations among states were affected by revolutions.
These scholars showed that international wars and shifts in international alignments were highly
likely to follow revolutions, since revolutionary regimes tended to distance themselves not only
from old regime domestic policies, but from their international policies and alliances as well.
Yet deficiencies in the theory of revolutions still remained. It was highly problematic to
apply Skocpol’s or my approach, both focused on the dynamics of agrarian bureaucracies, to the
revolutions in Eastern Europe and the Soviet Union, where the limits of traditional agrarian
economies no longer applied. While I had called for a critical role of ideology in guiding
revolutionary outcomes, critics of structural theory further insisted that ideology must be given a
far greater role in the origins of revolutions, in shaping the opposition and guiding revolutionary
mobilization (Sewell 1985, Foran 1992, Emirbayer and Goodwin 1994, Selbin 1993, Katz 1997,
Mahoney and Snyder 1999). In particular, they noted that structural accounts gave far too little role
to leadership and conscious decision-making by both rulers and revolutionaries. It may have been
true, as Skocpol argued, that Lenin’s precise designs for communism in Russia could not be
implemented. Yet it is hardly credible to go from that observation to theorizing as if Lenin’s
decisions and actions were wholly irrelevant to the development and outcome of the Russian
Revolution. Several scholars therefore undertook important studies of modern revolutions that
demonstrated the key role of ideology and leadership at multiple stages in revolutionary processes
(Foran 1997a, Katz 1997, Goodwin 1997, Parsa 2000, Selbin 1993).
Finally, an entirely separate line of theorizing regarding revolutions was developing among
political scientists using game theory (also called “rational choice analysis”). Instead of studying
empirical cases in detail, rational choice analysts created formal mathematical structures whose
logic was held to embody the choices faced by actors in revolutionary situations. Oddly, the initial
formulation of these formal models predicted that revolutions should not happen at all. Olson
(1965) and Tullock (1971) showed that for any individual who contemplated joining in a
revolutionary action, the risks and efforts of protest that he or she faced were normally quite
considerable, while their contribution to the success of that protest was usually small. In fact, for
most individuals their contribution was so modest that the outcome for society—whether or not the
revolution succeeded—would likely be the same whether they joined or not. Thus for any
individual, the logical incentives were strongly in favor of not joining a revolutionary protest, as
they faced substantial risks but were unlikely to change the outcome.
This paradox of collective action was held to show a major deficiency in sociological
theories of revolution based on analysis of large-scale, national revolutionary trajectories. Without
showing how the paradox of collective action was overcome, rational-choice theorists accused
 22

sociological theorists of offering theories that lacked solid foundations in individual behavior
(Kiser and Hechter 1991).
Over the course of the twentieth century, comparative-historical studies of revolutions had
made enormous strides in identifying the causal patterns behind the major pre-industrial
revolutions, in explaining their timing and their outcomes. Yet for all its triumphs, the
social-structural approach was under severe assault.

EXTENDING SOCIAL-STRUCTURAL APPROACHES: NEW PERSPECTIVES AND NEW

CASES
As has happened before when revolutionary theory based on CHA was confronted with new
empirical and theoretical challenges, scholars have delved into new cases of revolutions, and
re-examined old ones. While it is too soon to identify a new dominant perspective on revolutions,
the building blocks for that new perspective are already apparent. Older theorists such as
Goldstone, Tilly, Oberschall, Opp, and younger scholars such as Foran, Gould, Lichbach,
Goodwin, Wickham-Crowley, and Parsa have used additional case analyses to test prior theories
and lay out new principles for understanding a still broader range of revolutionary events.
Let us begin with the challenges posed by rational choice theory. Mark Lichbach (1995)
has led the way in showing how the collective action problem can be overcome. He has identified
over a dozen different mechanisms by which opposition groups can motivate people to join
revolutionary movements, consistent with rational action. Whether through contracts, incentives,
prior commitments, or exercising pre-existing authority over individuals, opposition movements
have a variety of tools available to promote mobilization. The key to these approaches is to realize
that individuals do not make isolated choices whether or not to join revolutionary protests, as the
early rational-choice theorists suggested. Rather, individuals come to social interaction already
bound up in a web of social connections and obligations. It is through building on these extant
linkages and commitments that revolutionary mobilization occurs (Opp and Roehl 1990,
Goldstone 1994b, Gould 1995).
However, formal models alone have not been persuasive to scholars—it required testing
those models in the context of detailed case studies, using process-tracing to demonstrate
congruence with the predictions of rational-choice models, to convince analysts of revolution of
the value of rational-choice approaches (Kiser 1996). These case-studies have particularly focused
on Eastern Europe (Opp 1989; Opp, Gern, and Voss 1995; Oberschall 1994b, Kuran 1991, 1995),
and shown that mass mobilization was explicable as a “rational” response to a perceived increased
opportunity to protest against and overturn the communist regimes. However, as Opp (1999) has
argued, this approach involved developing a “thick” rationality, rich in case-study insights of the
determinants of people’s decision-making in those particular cases, to supplement the “thin”
rationality of abstract formal models. Indeed, many leading proponents of applying rational choice
models to understanding political change now argue in favor of a synthetic approach, using
“analytic narratives” that combine insights from both rational-choice logic and close analysis of
specific historical cases (Bates et al. 1998).
Another area in which case-studies of revolutionary process are blending with work in
adjacent fields is efforts to link revolutionary theory with theories of social movements. As
McAdam, Tarrow and Tilly (1997) and I (Goldstone 1998a) have pointed out, case-study analyses
 23

of major social movements (e.g. McAdam 1982) and of revolutions were in fact converging on
similar sets of conditions and processes for protest and revolutions. Social movement theory had
argued that protest movements arose when three conditions were conjoined: (i) Political
opportunities, which are brought by state weakness or intra-elite conflicts; (ii) active mobilization
networks, drawing on pre-existing social linkages or organizations; and (iii) a cognitive framework
showing the need for and effectiveness of protest, created by leaders who “frame” the need for
protest in a persuasive ideology (Tarrow 1998). The first two of these conditions were, in effect,
already part of the theory of revolutions. The last condition, ideology, clearly needed to be brought
into revolutionary theory.
The role of ideology, however, is extremely complex, and still being explored. Selbin
(1983), in a comparative set of case studies of modern Latin American revolutions (Bolivia, Cuba,
Nicaragua, and Grenada), has argued that ideology is linked to leadership, and that effective
leadership is needed to both inspire revolutionaries through persuasive ideology, and to build and
maintain revolutionary institutions. Using the approach of demonstrating divergence among cases
that appear similar in light of past theories, Selbin showed that only in Cuba were the revolutionary
transformations both institutionalized and consolidated. In each of the other cases, either
revolutionary gains were reversed, or the revolutionary regime itself failed to become
institutionalized. Selbin showed that structural theories are unable to account for this divergence,
and that the effectiveness (or lack of effectiveness) of leadership in both sustaining a vision
focused on revolutionary goals, and building institutions to implement them, is crucial to
determining revolutionary outcomes. Foran and Goodwin have further shown, in a comparison of
revolutionary outcomes in Nicaragua and Iran, that the emergence of democractic characteristics
and economic changes are strongly affected by “which set of organized political leaders within the
revolutionary coalition is able to consoldiate its hold on state power after the overthrow of the old
regime” and the “ideological visions of the new revolutionary leadership” (1993: 210-211).
A number of scholars have also used comparative case studies to illustrate the role of
ideology in the origins of revolutions. Foran (1997a) and Wickham-Crowley (1992) have argued
that almost all societies maintain a stock of ideologies of protest or rebellion from past conflicts
with authorities, and that revolutionary leaders need to tap into these memories of conflict
experiences in their own societies to construct persuasive and attractive ideologies of rebellion.
Wickham-Crowley, (1991), Foran (1997) and Parsa (2000) have argued that it is ideologies that
provide the glue for alliances between elites and popular groups. Calhoun (1994) has further
pointed out that ideologies carry notions of identity and group commitment, which help sustain and
invigorate the social networks that mobilize people for revolutionary action. It is clear that
ideologies play so many crucial roles in revolutionary processes that the task of integrating
ideology into revolutionary theory is just beginning.
The largest challenge to revolutionary theory is to accurately encompass the major
revolutions of the late twentieth century, including anti-colonial, nationalist, and anti-dictatorial
revolutions such as those in Cuba, Iran, the Philippines, and the former communist regimes. Here
too, theories of revolution have made significant progress, with efforts by many scholars to grapple
with different sets of cases.
Foran (1997), Wickham-Crowley (1992), Goodwin (2001), Goodwin and Skocpol (1989),
and Goldstone et al (1991) have all shown that modern revolutions have common features that they
 24

share with the revolutions in agrarian monarchies and empires. Most of those features were
already laid out in the social-structural approach. International pressures, state crises, intra-elite
conflicts, and popular mobilization remain key components in the development of revolutionary
situations even in modern regimes. However, these broad factors need to be modified to gain
accurate correspondence with the events observed in modern revolutions, and a role needs to be
added for ideology as well. Although these scholars overlap to a good degree in identifying the
factors important to modern revolutions, they do not agree on precisely which conditions should
comprise the core of revolutionary theories.
International pressures, for example, are not sufficiently described in terms of international
military or economic competition. Increasingly, direct foreign intervention and support—or
withdrawal of that support—are common. In Cuba, Iran, Nicaragua, and the Philippines, the
pattern of initial U.S. support for dictators, followed by reduction or withdrawal of that support,
contributed to the outbreak of revolutionary conflicts (Halliday 1999, Snyder 1999).
State crisis is also not merely a problem of states running into elite resistance to reforms.
In modern states, the relationships between rulers and elites are quite varied. Elites may oppose
rulers because of failures of patronage, problems in economic development, exclusion from
political power, or lack of progress on nationalist goals. Goodwin (1994, 2001), Goldstone
(1994b), Wickham-Crowley (1992), and Snyder (1998), all building on the key work of Dix (1983,
1984), have argued that one type of modern dictatorship—the personalist, or neopatrimonial
dictatorship, in which a single individual has amassed great power and gains elite loyalty mainly
by dispensing favors and positions—is especially vulnerable to social revolutions. This kind of
regime can appear quite strong when the dictator is able to keep elites factionalized and dependent
on his favor. Yet when economic or international political pressures weaken the dictator’s ability
to control the elites, there are no other traditional or institutional supports for his regime. Such
regimes can thus fall with remarkable swiftness when they lose elite support. In addition, because
the dictator has no established organizations or institutions to enlist and maintain popular support,
beyond the repression of the police and military, elites can fairly easily encourage popular
opposition if the repressive institutions appear weakened or unwilling to defend the regime. This
pattern of neopatrimonial regime collapse can be identified in the revolutions in Cuba (1959),
Nicaragua (1979), and Iran (1979), and was expected—and in fact occurred—in the Philippines
(1986), Indonesia (1998), and Zaire (1998). In contrast, dictatorships linked to and strongly
supported by a particular social group, be it the military, landed elites, industrialists, or a political
party organization, are more stable.
The study of modern revolutions has also shown that if intra-elite conflicts are necessary to
create conditions favorable to revolution, a strong revolutionary movement generally requires
cross-class coalitions linking elite and popular groups (Parsa 2000; Foran 1992, 1997a; Goodwin
2001). This is particularly true where revolutionary movements take the form of guerrilla war
against the government. For such movements to succeed, the rural guerrillas must find allies
among elites and urban groups; otherwise they remain isolated in rural rebellion
(Wickham-Crowley 1991, 1992).
Scholars have demonstrated that even the collapse of communist regimes can be brought
within the scope of revolutionary theory. In an analysis of the collapse of the U.S.S.R., I have used
process-tracing to demonstrate the role of causal patterns and sequences familiar from prior
 25

revolutions (Goldstone 1998b). International competitive military and economic pressure from the
West brought a faction of Soviet elites led by Gorbachev to seek major reforms in the communist
regime (Glasnost and Perestroika). However, by the time of those reforms, a sharp contraction in
Soviet economic growth had already weakened the state’s ability to provide continued
employment improvement and higher living standards for its population. In addition, the
stagnation of professional employment produced a large number of educated elites (technicals)
with college degrees who were stuck in blue-collar jobs and excluded from political or economic
advancement. A fiscally-strapped regime thus faced elites who were divided and distressed over
state policy and problems of social mobility.
Economic decline, along with grievous pollution and inefficient health care, had also
produced a marked decline in life expectancy since the mid-1970s. Health and safety concerns
were particularly strong in those areas with the most polluting and dangerous factories and mines.
Urban groups and miners thus became a particular source of popular mobilization against the
communist regime.
The technical elites, urban workers, and miners saw Gorbachev’s reforms as an opportunity
not merely to reform, but to throw out communist control over political and economic life. They
supported the reformist faction, led by Yelstin in Russia and by other nationalist leaders in the
Baltic, Caucasus, and Central Asian Soviet Republics, and pressed for the end of communist
control. The key elements in this process—international pressure, state weakness, intra-elite
conflicts, cross-class coalitions among elite reformers and popular groups facing depressed living
conditions, and ideologies of rebellion (democracy and nationalism) that united regime opponents
and inspired rebellion—are precisely those found in other major revolutions. As a consequence, I
suggested that as in most other major revolutions, struggles for power would likely continue, and
in the absence of leaders strongly committed to democracy—an absence felt in most former
communist regimes except for a few in Eastern Europe and the Baltics—democracy was not likely
to develop, whatever the hopes of Western aid contributors and well-wishers.
Goodwin (2001), in a comparative case study of the anti-communist revolutions in Eastern
Europe, further pointed out that differences in the way those revolutions unfolded reflected
differences in state structure. Romania’s revolution against the neopatrimonial Ceaucescu regime
was most like the sudden and violent overthrow of other personalist dictators, while the limited
resistance put up by the communist regimes in East Germany and Czechoslovakia meant that these
events unfolded more like social protest movements.

A NEW SYNTHESIS: FOCUSING ON CAUSES OF STABILITY

While these efforts have shown how revolutionary theory can be modified and applied to explain
modern revolutions, they have also created their own problem in the complexity that resulted from
the extension of theory to cover more cases. In the course of these efforts, the earlier goal of
identifying a short and consistent list of causes that lead to revolution seems to have been lost. For
example, while state weakness or crisis is frequently identified as a condition of revolution, what
are the causes that produce that condition? Scholars have enumerated such causes as military
defeat or competition, economic downturns, excessive state debt or fiscal strain, long-term
demographic pressures, uneven or unsuccessful economic development, strains on colonial
powers’ resources or domestic support, rulers’ departures from religious or nationalist principles,
 26

foreign interventions, shifts in state resources due to changing export or import prices, ethnic
conflicts, erratic or excessive repression by rulers, shifts in international standards for human rights,
harvest failures or other natural disasters (including earthquakes, flood, and droughts), with no two
revolutions having precisely the same causes. Similarly, intra-elite conflicts may arise from ethnic
divisions, political exclusion, state actions that favor a particular group, religious divisions,
conflicts over resources or economic development, or shifts in social mobility that produce new
elite aspirants or that undermine the continuity of existing elite families.
In short, it has become more difficult to look behind the “conditions”of revolution—such
as state crisis or elite conflicts—for concrete historical causes. The latter are simply too diverse,
and vary too greatly across cases, to serve as the foundation for general theory. Moreover, there is
a realization that the unfolding of revolutions has emergent properties; once a revolutionary
situation has begun, the actions and decisions of rulers and revolutionaries influence further events.
Certain mechanisms persist across revolutionary situations (McAdam, Tarrow, and Tilly
2001)—popular mobilization, leadership, ideological framing, building cross-class coalitions,
confronting authorities with mass actions—but they may develop in different ways in different
cases. Thus simply listing conditions for revolution does not tell us how revolutions develop,
because these conditions combine and concatenate in different ways.
For example, in Russia intra-elite conflict was not important until after WWI produced a
collapse of the Tsar’s military; Lenin and Trotsky’s leadership and mobilization of workers then
proved critical in the ensuing intra-elite struggles for control of the post-Tsarist state. In Iran, by
contrast, the military was never weakened by international conflicts. Rather, intra-elite conflicts
precipitated protests over the Shah’s religious policies and alliance with the United States; these
protests then led to more constrained U.S. support and a decision by the Shah’s military to
withdraw from defending the Shah against mass opposition. In other cases, revolutions develop
through guerrilla wars; in still others through relatively peaceful strikes and protests. A list of
causes of revolution is thus too static to capture the reality of the revolutionary process, in which
causal factors enter in different combinations and temporal patterns, interacting with the decisions
of rulers and revolutionaries to produce varied revolutionary trajectories. The efforts of scholars to
achieve greater accuracy in understanding the causes of revolutions, in an increasing number of
cases, thus appears to have succeeded at the expense of parsimony and generality (Przeworski and
Teune 1970: 20-23).
I have tentatively suggested (Goldstone 2001) a way out of this impasse by turning the
problem of revolutionary causation around. Theories of revolution have generally proceeded as if
states have a huge variety of ways to operate normally, but only a small and particular set of
conditions is sufficiently disruptive to produce revolutions. However, over a century of empirical
events and investigations now suggests that may not be so. Instead, it appears that a huge variety
of events are involved, in various combinations, in producing different varieties of revolutionary
situations. Thus, instead of treating stability as unproblematic, and piling up a set of conditions
that lead to revolution, I have argued that we may be better served by treating stability as the
problem, and seeing revolutions as the result of the loss or undermining of the conditions that
maintain stability. A small number of conditions may in fact be critical to sustaining state stability,
while many different combinations and patterns of causation can undermine these conditions and
lead to state breakdown. The different ways this happens will generate different trajectories
 27

leading to various processes and outcomes of protest and/or revolution.

The conditions for stability seem fairly easily enumerated, by abstracting from and
inverting the much longer list of revolutionary conditions. They are: (1) rulers (or ruling
organizations) that are widely perceived as both effective and just; (2) elites that are unified and
loyal to the regime; and (3) popular groups that are able to secure their customary standard of living
in a reliable manner. States that meet these conditions are reliably stable; the more that states fall
short of meeting these conditions the more likely they are to experience conflict, protest, and
revolution.
Departures from the conditions for stability then create situations in which revolutionary
leaders can begin to mobilize oppositions, and rulers to plan their response. The decisions and
actions of rulers and revolutionaries, the particular elite cleavages and groups that are mobilized,
the specific ideologies used to generate support, and international reactions then generate
the variety in the pattern of protests and revolutions that we observe—some involving guerrilla
warfare, some involving the sudden collapse of a state, some occurring in neopatrimonial
dictatorships, some in one-party states, some in democracies.
By focusing on the loss or undermining of conditions of stability, revolutionary theory can
allow much greater scope for dynamic processes such as leadership, mobilization, and
decision-making by rulers and revolutionary opponents, and for the interplay of these processes to
vary over time. In addition, specifying conditions for stability allows for the enormous range of
specific concrete factors shown to be important in revolutions—such as military or economic
failures, failures to uphold religious or nationalist principles, or excessive or inadequate repression
or response to disasters, to name a few—to play a role without having to specify the same set of
causal factors being involved in all revolutions. In this way, those specific factors specified by
Skocpol—military pressure that undermines state effectiveness, state reforms that undermine elite
unity and support for the state, and village communal organizations facing more onerous taxes and
rents—become simply one particular set of conditions undermining state stability. The
demographic pressures I had noted in late pre-industrial societies, and the ways that personalist
dictatorships may decline in effectiveness or forfeit elite loyalty, similarly become special cases
subsumed under the broader theory of regime stability. The ineffectiveness of the Wiemar regime
that led to the Nazi revolution and of the Chinese imperial court; the elite court/country divisions
in 17th century England and the reformist/conservative blocs in the Communist Party of the Soviet
Union in the late 20th century; the impact of demographic pressures and of commercialized
agriculture on peasants across time and space, all can be brought into a common framework of
factors that undermine the conditions of state stability. When all three of those conditions of
stability are simultaneously lost, by whatever means, a revolutionary situation is the result.
Harking back to Weber’s view of revolutions as requiring the institutionalization of new
values, the process and outcomes of revolutions can then be conceptualized as an ongoing effort to
recreate the conditions of stability. Revolutionary leaders use ideological struggles and
mobilization to secure elite support for the revolution, while building a stronger state is essential to
gain effectiveness for the new revolutionary regime. Securing a stable material existence for the
population is often the most difficult condition to meet, and thus it is normal to expect recurrent
popular protests as part of the revolutionary process. Indeed, the difficulty of rebuilding stable
conditions implies that revolutionary struggles will often last years, or even decades, and that the
 28

fall of the old regime will generally not, in itself, bring a society to popular prosperity, elite loyalty,
and effective and just governance (Stinchcombe 1999).
It is too early to tell if a theory based on conditions of stability will prove able to simplify
theories of revolution, while yet comprehending a wide range of cases and aspects of the process
of revolutions. However, it appears promising, as is the range of excellent work done on ideology,
on revolutions in neopatrimonial states, on guerilla wars, and on the collapse of communist
regimes. Revolutionary theory thus appears well on its way to meeting current challenges.
By focusing on international pressures, state weakness, intra-elite conflicts, cross-class
coalitions, personalist regimes, ideologies of rebellion, mobilization networks, popular living
standards, and revolutionary leadership, comparative historical analyses of revolution have moved
toward a common explanatory framework covering events ranging from the French Revolution to
the collapse of the U.S.S.R. This framework has its roots in the insights of the fathers of
sociological theory, but has continually developed over time in response to new events, expanding
to incorporate new cases and additional details of the revolutionary process.
Lest we end on too triumphal a note, we should be clear that important further challenges
for revolutionary studies still remain. Though receiving increasing attention, the role of women
and gender issues in revolution has yet to be adequately analyzed (Moghadam 1997, Tétreault
1994, Wasserstrom 1994). Although rational choice theory has offered some guidance, no
mathematical formalization of revolutionary processes is widely accepted. And although there are
promising starts (Goodwin 1997, Scott 1990), macro-level theories of revolution still need to be
harmonized with social-psychological theories of individual behavior. As in any fruitful field of
research, further frontiers yet await.

KNOWLEDGE CUMULATION IN THE STUDY OF REVOLUTIONS

In this section, I wish to briefly summarize some of the key findings presented by scholars using
CHA regarding revolutions. I believe it will be evident that these findings could not have been
uncovered by any other methods than careful process-tracing and congruence-testing of discrete
historical cases.

Conjunctural Causation
CHA of revolutions has shown that there is no single set of necessary and sufficient causes that lay
behind all revolutions. Rather, revolutions are the result of a conjuncture of distinct causal streams,
and the type and development of a revolution depends on the precise combination and components
in those causal streams. The key to understanding collapses of authority of a magnitude that can
lead to revolution is that they involve the breakdown of authority at several levels in society: state
command over elites fails; elite unity or cooperation is sundered; popular acceptance of political
authority and/or certain elite privileges is replaced by active rejection.
An enormous range of events and conditions can contribute to such conjunctures. Across
varied cases, scholars have identified as frequently important all of the following: international
military pressures on state fiscal health and effectiveness; demographic pressures on agrarian
economies and traditional status recruitment; elite alienation stemming from political exclusion by
personalist dictatorships; elite and popular alienation due to striking repression or cultural and
religious deviation by the regime; economic downturns or natural disasters that not only bring great
 29

popular distress but are perceived as caused or exacerbated by the regime; the development of
cross-class or multi-class coalitions of opposition; the spread of compelling ideologies of
opposition rooted in nationalist, religious, or other widespread and popular cultural idioms; the
emergence of leadership that is both inspiring and pragmatically effective; and mobilization
networks, whether formal or informal, that draw on personal ties and group affliations to forge
oppositional identities.
The complexity of the list above is inseparable from the complexity of the events we
collectively label “revolutions.” However, this does not mean they are so complex as to resist
explanation or even fairly straightforward analysis. By abstracting from concrete causes, such as
those enumerated in the preceding paragraphs, and focusing on more general conditions that
indicate the collapse of state stability, we can arrive at the following list of five key conditions that
must arise for a revolutionary episode to begin:ii
(1) A crisis of state authority, in which the state is widely perceived by elites and popular
groups as both ineffective and unjust;
(2) A crisis of elite relationships, in which elites become divided, alienated, and polarized
into factions that disagree over how to reconstitute state authority;
(3) A crisis of popular welfare, in which urban and/or rural groups find it difficult to
maintain their customary standard of living through accustomed means;
(4) The emergence of an elite/popular coalition to attack the authority of the state;
(5) An ideology of opposition that binds elite and popular groups in their attack on the
authorities, justifies that attack, and suggests alternative bases for authority.
While the presence of any one or several of these conditions can lead to a political crisis,
such as a peasant rebellion, urban uprising, elite revolt or coup, or civil war, it seems to require all
of these conditions for a truly revolutionary situation to arise.

Variety in Processes and Outcomes

One of the most critical insights of CHA analyses of revolutions is the need to clearly separate the
study of origins, processes, and outcomes of revolutions. The causal origins of revolutions—that
is, the factors that undermine the stability of regimes—interact with prior conditions and with
actions by states, opposition groups, and international actors to produce contingent and emergent
revolutionary processes of coalition formation, revolutionary tactics and confrontations, and
factional struggles. These processes of revolutionary conflict can last for decades before a stable
outcome is achieved. That outcome, in turn, is highly contingent on the particular leaders,
coalitions, ideologies, and experiences that emerge as dominant from the revolutionary process as
it unfolds.
We have already mentioned that some revolutions develop slowly out of visibly active
revolutionary movements that gradually gain strength, while other revolutions erupt suddenly and
unexpectedly from the implosion of the old regime. Each pattern reflects a different combination
of causal components. Where states retain considerable elite support and fiscal strength, but act
unjustly and repressively, revolutionary movements often build on the periphery, awaiting a shift
in state strength or external alliances to provide an opportunity for successful rebellion. Led by
marginal or dissident elites, they need to battle to win over dominant elites and extend their popular
support. This was the pattern in the Cuban, Nicaraguan, and Chinese Communist revolutions.
 30

Where states are weakening in their material resources, are seen as increasingly ineffective, and
elite defections are widespread, a sudden collapse in state authority may occur, with elites
mobilizing popular groups first for reform and then the overthrow of the regime. This was the
pattern in the English, French, Mexican, Russian, and anti-Soviet revolutions.
Both such patterns can in turn produce a variety of outcomes. Here the ideology of
rebellion, the constellation of external and internal struggles facing the revolution, and the
character of the revolutionary leadership are critical. Revolutionary leaders who believe in
democracy as a chief goal, trumping other aims, appear to be essential to create democratic
outcomes. If they have that goal—such as Mandela in South Africa, the Sandinistas in Nicaragua,
or Aquino in the Philippines—then even counter-revolutionary threats, severe ethnic conflicts, and
dire economic conditions do not prevent the construction of democratic institutions. In contrast,
leaders who are willing to subordinate building democratic institutions to other goals—whether
economic, religious, or military—tend to produce party-states or dictatorships, even facing far
lesser threats, as in Khomeini’s post-revolutionary regime in Iran.
Other economic and political aspects of revolutionary outcomes are also strongly related to
the ideology of the revolutionary leadership, and how that intersects with their experience and
conditions they face. Every revolutionary leadership faces the problem of how to restore stability,
which in turn requires effective fiscal and military action by the new regime, gaining the loyalty of
elites, and reducing popular grievances. How the leadership frames the solution to these problems
often determines the outcome of the regime. Often, heightened centralization and
bureaucratization of the new regime compared to the old is necessary for effective mobilization of
money and manpower. However, even more important is how the leadership views the faults of
the old regime. If the old regime was seen as failed because it departed from past virtues, which
now need to be restored, the revolution may have a highly conservative cast, even as it creates new
institutions of governance, as for example in Meiji Japan. In contrast, if the old regime was seen
as failed because it held tight to old, obsolete, invalid ideals and principles, the revolution will take
a radical stance, even if in many respects it simply extends and modifies institutions of the old
regime, as in the French and Russian Revolutions.
Lastly, it must be noted that many of the processes and outcomes of revolutions are
emergent, unknowable until they are produced in reaction to the very events of the revolutionary
process itself. Many revolutions begin as essentially reformist, only to become radicalized under
the failure of moderates to secure stability, or in the heat of battle with counter-revolutionary
foreign and domestic foes. For example, Madero’s reformist revolution in Mexico against Diaz
only led to massive peasant rebellions and civil war because of the imposition of the
counter-revolutionary Huerta regime. Similarly, in the wake of the moderate New Economic
Program in Russia’s revolution and Lenin’s death, there arose a power struggle between those who
wished a more radical return to communism, and those who sought to continue on a more graduate
path to socialism. The victory of the former under Stalin put the Russian revolution on a more
distinctly communist and authoritarian path.

CONCLUSION
Revolutions differ profoundly in their concrete causes, their patterns of unfolding, and their
outcomes. However, the conditions for a revolutionary situation to arise can be simply described
 31

in terms of the conditions that denote a collapse of state stability. Nonetheless, these conditions do
not determine the development and outcome of a given revolution; those are emergent properties
and depend on events and struggles within the unfolding revolution, on the actions of particular
leaders, international actors, and popular groups.
Process-tracing can reveal those causal mechanisms and how they operate in varied
contexts, while congruence-testing can demonstrate the similarities and differences among
revolutionary events. These methods of CHA have continually confronted prior beliefs about
revolutions with new empirical findings, challenging and expanding our views. This process is
certain to continue, as it seems unlikely that we have exhausted all the varieties of causation or
outcomes that revolutions could produce. Especially with that prospect, CHA seems an essential
method to continue the cumulative progress in our knowledge about revolutions and revolutionary
movements.

BIBLIOGRAPHY
Abbott, Andrew. 1990. “Conceptions of Time and Events in Social Science Methods: Causal
and Narrative Approaches.” Historical Methods 23:140-50.
-------. 1992. “From Causes to Events: Notes on Narrative Positivism.” Sociological Methods
and Research 20:428-55.
Almond, Gabriel A., ed. 1973. Crisis, Choice, and Change: Historical Studies of Political
Development. Boston: Little, Brown.
Arjomand, Said A. 1988. The Turban for the Crown: The Islamic Revolution in Iran. New York:
Oxford University Press.
Armstrong, David. 1993. Revolution and World Order: The Revolutionary State in International
Society. Oxford: Oxford University Press.
Bates, Robert H., Avner Greif, Margaret Levi, Jean-Laurent Rosenthal, and Barry R. Weingast.
Analytic Narratives. Princeton: Princeton University Press.
Bennett, Andrew and Alexander L. George. 1997. “Process Tracing in Case Study Research.”
Paper presented at the MacArthur Foundation Workshop on Case Study Methods, Belfer
Center for Science and International Affairs (BCSIA), Harvard University, October
17-19.
Berger, Joseph and Morris Zelditch Jr., eds. 1993. Theoretical Research Programs: Studies in
the Growth of Theory. Stanford: Stanford University Press.
Brinton, Crane. 1938. The Anatomy of Revolution. New York: Norton.
Calhoun, Craig. 1994. Neither Gods nor Emperors: Students and the Struggle for Democracy in
China. Berkeley: University of California Press.
Cobban, Alfred. 1964. The Social Interpretation of the French Revolution. Cambridge:
Cambridge University Press.
Collier, David. 1993. “The Comparative Method.” In Political science: The State of the
Discipline II, ed. Ada Finifter, pp. 105-19. Washington, DC: American Political
Science Association.
Collier, David and James Mahoney. 1996. “Insights and Pitfalls: Selection Bias in Qualitative
Research.” World Politics 49:56-91.
Dix, Robert. 1983. “The Varieties of Revolution.” Comparative Politics 15:281-95.
 32

-------. 1984. “Why Revolutions Succeed and Fail.” Polity 16:423-46.

Eckstein, Susan. 1982. “The Impact of Revolution on Social Welfare in Latin America.” Theory
and Society. 11:43-94.
Edwards, Lyford P. 1927. The Natural History of Revolutions. Chicago: University of
Chicago Press.
Eisenstadt, S.N. 1978. Revolution and the Transformation of Societies. New York: Free Press.
Emirbayer, Mustafa and Jeff Goodwin. 1994. “Network Analysis, Culture and the Problem of
Agency.” American Journal of Sociology 99:1411-54.
Farhi, Farideh. 1990. States and Urban-based Revolution: Iran and Nicaragua. Urbana and
Chicago: University of Illinois Press.
Feierabend, Ivo K., Rosalind L. Feierabend, and Betty A. Nesvold. 1969. “Social change and
Political Violence: Cross-national Patterns. In Violence in America: Historical and
Comparative Perspectives, ed. Hugh D. Graham and Ted R. Gurr, pp. 606-68. New
York: Praeger.
Foran, John. 1992. “A Theory of Third World Social Revolutions: Iran, Nicaragua, and El
Salvador Compared.” Critical Sociology 19:3-27.
-------. 1997a. “The Comparative-historical Sociology of Third World Social Revolutions: Why
a Few Succeed, Why Most Fail.” Pp. 227-67 in Foran 1997b.
-------, ed. 1997b. Theorizing Revolutions. London: Routledge.
Foran, John and Jeff Goodwin. 1993. “Revolutionary Outcomes in Iran and Nicaragua:
Coalition Fragmentation, War, and the Limits of Social Transformation.” Theory and
Society 22:209-47.
Forster, Robert. 1980. “The French Revolution and the New Elite, 1800-1850.” Pp. 182-207
in The American and European Revolutions, 1776-1850, ed. J. Pelenski. Iowa City:
University of Iowa Press.
Furet, François. 1981. Interpreting the French Revolution. Cambridge: Cambridge University
Press.
Garton Ash, Timothy. 1989. “Refolution in Hungary and Poland.” New York Review of Books
36 (August 17): 9-15.
George, Alexander, with Andrew Bennett. 1997. “The Role of the Congruence Method for Case
Study Research.” Paper presented at the 38th Annual Convention of the International
Studies Association in Toronto, March 18-22.
Glenn, John K. 1999. “Competing Challengers and Contested Outcomes to State Breakdown: the
Velvet Revolution in Czechoslovakia.” Social Forces 78:187-212.
Goldstone, Jack A. 1991. Revolution and Rebellion in the Early Modern World. Berkeley:
University of California Press.
-------. 1994a. “Is Revolution Individually Rational?” Rationality and Society 6:139-66.
-------. 1994b. “Revolution in Modern Dictatorships.” Pp. 70-77 in Goldstone 1994c.
-------, ed. 1994c. Revolutions: Theoretical, Comparative, and Historical Studies. Fort Worth,
TX: Harcourt Brace. 2nd ed.
-------. 1998a. “Social Movements or Revolutions? On the Evolution and Outcomes of Collective
Action.” Pp. 125-45 in Democracy and Contention, ed. Marco Guigni, Doug McAdam,
and Charles Tilly. B-oulder, CO: Rowman & Littlefield.
 33

-------. 1998b. “The Soviet Union: Revolution and Transformation.” Pp. 95-123 in Elites, Crises,
and the Origins of Regimes, ed. Mattei Dogan and John Higley. Boulder, CO: Rowman &
Littlefield.
-------, ed. 1998c. The Encyclopedia of Political Revolutions. Washington, D.C.: Congressional
Quarterly Press.
-------. 1999. “Initial Conditions, General Laws, Path Dependence, and Explanation in Historical
Sociology.” American Journal of Sociology 104:829-845.
-------. 2001. “Toward a Fourth Generation of Revolutionary Theory.” Annual Review of
Political Science 4: 139-87.
Goldthorpe, John H. 1997. “Current Issues in Comparative Macro-Sociology: A Debate on
Methodological Issues.” Comparative Social Research 16:1-26.
Goodwin, Jeff. 1994. “Old regimes and Revolutions in the Second and Third Worlds: A
Comparative Perspective.” Social Science History 18:575-604.
-------. 1997. “The Libidinal Constitution of a High-risk Social Movement: affectual ties and
solidarity in the Huk rebelion, 1946 to 1954. American Sociological Review 62: 53-70.
-------. 2001. No Other Way Out: States and Revolutionary Movements, 1945-1991.
Cambridge: Cambridge University Press.
Goodwin, Jeff and Theda Skocpol. 1989. “Explaining Revolutions in the Contemporary Third
World.” Politics and Society 17:489-507.
Gould RV. 1995. Insurgent Identities: Class, Community, and Protest in Paris from 1848 to the
Commune. Chicago: University of Chicago Press.
Griffin, Larry J. 1992. “Temporality, Events, and Explanation in Historical Sociology: An
Introduction.” Sociological Methods and Research 20:403-27.
Gugler, Josef. 1982. “The Urban Character of Contemporary Revolutions.” Studies in
Comparative International Development 17: 60-73.
Gurr, Ted Robert. 1968. “A Causal Model of Civil Strife: A Comparative Analysis Using New
Indices.” American Political Science Review 62:1104-24.
-------. 1970. Why Men Rebel. Princeton, NJ: Princeton University Press.
-------, ed. 1980. Handbook of Political Conflict. New York: Free Press.
Halliday, Fred. 1999. Revolution and World Politics. London: Macmillan.
Haydu, Jeffrey. 1998. “Making Use of the Past: Time Periods as Cases to Compare and as
Sequences of Problem Solving”. American Journal of Sociology 104: 339-71.
Hibbs, Douglas A. 1973. Mass Political Violence: A Cross-national Causal Analysis.
New York, Wiley.
Huntington, Samuel P. 1968. Political Order in Changing Societies. New Haven, CT: Yale
University Press.
Johnson, Chalmers. 1966. Revolutionary Change. Boston: Little, Brown.
Katz, Mark. 1997. Revolutions and Revolutionary Waves. New York: St. Martin’ss Press.
King, Gary, Robert O. Keohane, and Sidney Verba. 1994. Designing Social Inquiry: Scientific
Inference in Qualitative Research. Cambridge: Cambridge University Press.
Kiser, Edgar. 1996. "The Revival of Narrative in Historical Sociology: What Rational Choice
Theory and Contribute." Politics and Society 24:249-271.
Kiser, Edgar, and Michael Hechter. 1991. “The Role of General Theory in Comparative-
 34

Historical Sociology.” American Journal of Sociology 97:1-30.

Kuran, Timur. 1991. “Now out of Never: The Element of Surprise in the Eastern European
Revolution of 1989.” World Politics 44:7-48.
-------. 1995. Private Truths, Public Lies: The Social Consequences of Preference
Falsification. Cambridge, MA: Harvard University Press
Kuhn, Thomas S. 1996. The Structure of Scientific Revolutions. 3rd ed. Chicago, IL: University
of Chicago Press.
Kurzman, Charles. 1996. “Structural Opportunity and Perceived Opportunity in Social Movement
Theory.” American Sociological Review 61:153-70.
Lakatos, Imre. 1983. The Methodology of Scientific Research Programmes, edited by John
Worrall and Gregory Currie. Cambridge: Cambridge University Press.
Lenin, Vladimir Ilich. [1973]. What is to be Done? Burning Questions of our Movement.
Translation of the 1st ed. of 1902. Peking: Foreign Languages Press.
Lichbach, Mark I. 1989. “An Evaluation of ‘Does Economic Inequality Breed Political
Conflict?’ Studies. World Politics 41:431-70.
-------. 1995. The Rebels’ Dilemma. Ann Arbor: University of Michigan Press.
Lieberson, Stanley. 1991. “Small N’s and Big Conclusions: An Examination of the Reasoning
in Comparative Studies Based on a Small Number of Cases.” Social Forces 70:307-20.
Little, Daniel. 1998. Microfoundations, Method, and Causation: On the Philosophy of the Social
Sciences. New Brunswick, NJ: Transaction Publishers.
McAdam, Doug. 1982. Political Process and the Development of Black Insurgency, 1930-1970.
Chicago : University of Chicago Press, 1982. viii, 304
McAdam, Doug, John D. McCarthy, and Mayer N. Zald, eds. 1996. Comparative Perspectives
on Social Movements. Cambridge: Cambridge University Press.
McAdam, Doug, Sidney Tarrow and Charles Tilly. 1997. “Toward a Comparative Perspective
on Social movements and Revolution.” Pp. 142-73 in Comparative Politics: Rationality,
Culture, and Structure. ed. Mark Lichbach and Alan Zuckerman. Cambridge: Cambridge
University Press.
-------. Forthcoming. Dynamics of Contention. Cambridge: Cambridge University Press.
McKeown, Timothy J. 1999. “Case Studies and the Statistical Worldview.” International
Organization 53:161-91.
Mahoney, James. 1999. “Nominal, Ordinal, and Narrative Appraisal in Macrocausal Analysis.”
American Journal of Sociology 104:1154-96.
-------. 2000a. “Strategies of Causal Inference in Small-N Analysis.” Sociological Methods and
Research 28:387-424.
-------. 2000b. “Path Dependence in Historical Sociology.” Theory and Society 29:507- 48.
Mahoney, James and Richard Snyder. 1999. “Rethinking Agency and Structure in the Study of
Regime change.” Studies in Comparative International Development 34:3-32.
Markoff, John. 1996. The Abolition of Feudalism. University Park, PA: Pennsylvania State
University Press.
Midlarsky, Manus I. 1982. “Scarcity and Inequality: Prologue to the Onset of Mass
Revolution.” Journal of Conflict Resolution 26:3-38.
Moghadam, Val M. 1997. “Gender and revolutions.” Pp. 137-67 in Foran 1997c.
 35

Moore, Barrington Jr. 1966. Social Origins of Dictatorship and Democracy. Boston: Beacon
Press.
Muller, Edward N. 1985. “Income Inequality, Regime Repressiveness, and Political Violence.”
American Sociological Review 50:47-61.
Nichols, Elizabeth. 1986. “Skocpol on Revolution: Comparative Analysis vs. Historical
Conjuncture.” Comparative Social Research 9:163-86.
Oberschall, Anthony. 1994a. “Protest Demonstrations and the End of Communist Regimes in
1989.” Research in Social Movements, Conflict, and Change 17:1-24.
-------. 1994b. “Rational Choice in Collective Protests.” Rationality and Society
6:79-100.
Olson M Jr. 1965. The Logic of Collective Action: Public Goods and the Theory of Groups.
Cambridge, MA: Harvard University Press.
Opp, Karl-Dieter. 1989. The Rationality of Political Protest. Boulder, CO: Westview.
-------. 1999. "Contending Conceptions of the Theory of Rational Action." Journal of
Theoretical Politics 11:171-202.
Opp Karl-Dieter, Peter Voss, and Christianne Gern. 1995. Origins of a Spontaneous Revolution:
East Germany, 1989. Ann Arbor: University of Michigan Press.
Opp Karl-Dieter and Wolfgang Roehl. 1990. “Repression, Micromobilization, and Political
Protest.” Social Forces 69:521-47.
Osa, Maryjane. 1997. “Creating Solidarity: the Religious Foundations of the Polish Social
Movement.” East European Politics and Society 11:339-65.
Paige, Jeffrey M. 1975. Agrarian Revolution. New York: Free Press.
Parker, David, ed. 2000. Revolutions and Revolutionary Tradition in the West, 1560-1991.
London: Routledge.
Parsa, Misagh. 2000. States, Ideologies, and Social Revolutions: A Comparative Analysis of
Iran, Nicaragua, and the Philippines. Cambridge: Cambridge University Press.
Parsons, Talcott. 1966. Societies: Evolutionary and Comparative Perspectives. Englewood
Cliffs, NJ: Prentice-Hall.
Pettee, George S. 1938. The Process of Revolution. New York: Harper and Row.
Pfaff, Steven. 1996. “Collective Identity and Informal Groups in Revolutionary Mobilization:
East Germany in 1989.” Social Forces 75:91-110.
Przeworski, Adam and Henry Teune. 1970. The Logic of Comparative Social Inquiry. New
York: Wiley-Interscience.
Ragin, Charles C. 1997. “Turning the Tables: How Case-Oriented Research Challenges
Variable-Oriented Research.” Comparative Social Research 16:27-42.
Rasler, Karen. 1996. “Concessions, Repression, and Political Protest in the Iranian Revolution.”
American Sociological Review 61:132-52.
Scott, James C. 1990. Domination and the Arts of Resistance: Hidden Transcripts. New
Haven, CT: Yale University Press.
Selbin, Eric. 1993. Modern Latin American Revolutions. Boulder, CO: Westview Press.
Sewell, William Jr. 1985. “Ideologies and Social Revolutions: Reflections on the French case.”
Journal of Modern History 57:57-85.
-------. 1996. “Three Temporalities: Toward an Eventful Sociology.” In The Historic Turn in
 36

the Human Sciences, ed. Terence J. McDonald, pp. 245-280. Ann Arbor: University of
Michigan Press.
Skocpol, Theda. 1979. States and Social Revolutions. Cambridge: Cambridge University Press
-------. 1982. “Rentier State and Shi’a Islam in the Iranian Revolution.” Theory and Society 11:
265-303, with responses by Nikki Keddie, Walter Goldfrank, and Eqbal Ahmed.
-------. 1994. Social Revolutions in the Modern World. Cambridge: Cambridge University Press.
Snyder, Richard. 1998. “Paths Out of Sultanistic Regimes: Combining Structural and Voluntarist
Perspectives.” Pp. 49-81 in Sultanistic Regimes, ed. H.E. Chehabi and J.J. Linz.
Baltimore, MD: Johns Hopkins University Press
Snyder, Robert S. 1999. “The U.S. and Third World Revolutionary States: Understanding the
Breakdown in Relations.” International Studies Quarterly 43:265-90.
Stinchcombe, Arthur L. 1999. “Ending Revolutions and Building New Governments.” Annual
Review of Political Science 2:49-73.
Stryker, Robin. 1996. “Beyond History Versus Theory: Strategic Narrative and Sociological
Explanation.” Sociological Methods and Research 24:304-352.
Tarrow, Sidney. 1998. Power in Movement: Social Movements and Contentious Politics.
Cambridge: Cambridge University Press. 2nd ed.
Taylor, George V. 1972. “Noncapitalist Wealth and the Origins of the French Revolution.”
American Historical Review 67: 956-77.
Tétreault M.A., ed. 1994. Women and Revolution in Africa, Asia, and the New World. Columbia:
University of South Carolina Press.
Tilly C. 1973. “Does Modernization Breed Revolution?” Comparative Politics 5: 425-47.
-------. 1978. From Mobilization to Revolution. Reading, MA: Addison-Wesley.
-------. 1984. Big Structures, Large Processes, Huge Comparisons. New York: Russell Sage
Foundation.
Trotsky, Leon. 1959. The Russian Revolution. Selected and ed. By F. W. Dupee, trans. Max
Eastman. New York: Doubleday/Anchor Books.
Tullock G. 1971. “The Paradox of Revolution.” Public Choice 1:89--99
Urban, M., V. Igrunov, and S. Mitrokhin. 1997. The Rebirth of Politics in Russia. Cambridge:
Cambridge University Press.
Wallerstein, Immanuel. 1974. The Modern World System. Vol. 1. New York: Academic Press.
Walt, Stephen M. 1996. Revolution and War. Ithaca, NY: Cornell University Press.
Wasserstrom, Jeffrey N. 1994. “Gender and revolution in Europe and Asia.” Journal of Women’s
History 5:170-83, 6:109-20.
Weede, Eric. 1987. “Some new Evidence on Correlates of Political Violence: Income Inequality,
Regime Repressiveness, and Economic Development.” European Sociological Review
3:97-108.
Western, Bruce. 2000. “Bayesian Thinking About Macrosociology.” Estudio/Working Paper
2000/152. Madrid: Juan March Institute.
Wickham-Crowley, Timothy. 1991. Exploring Revolution: Essays on Latin American Insurgency
and Revolutionary Theory. Armonk, NY: Sharpe.
-------. 1992. Guerrillas and Revolution in Latin America. Princeton, NJ: Princeton University
Press.
 37

Wolf, Eric. 1970. Peasant Wars of the Twentieth Century. New York: Harper.

i. Of course overall price inflation, as opposed to relative price shifts, requires an increase in either
the velocity or the supply of money. In a separate paper, (Goldstone 1984) I showed that if
population increase is accompanied by urban growth and craft specialization, the velocity of
money will greatly increase. At the same time, states and merchants facing rising nominal prices
generally responded by devaluing coinage and extending public and private credit, thus increasing
liquidity. The increased demand fueled by population growth was therefore accompanied by
increases in the velocity and supply of money.

ii. This list is compiled, with considerable abstraction, primarily from the works of Skocpol (1979),
Goldstone (1991), Foran (1997a), Goodwin (2001), and Wickham-Crowley (1992).