In his book, Charles Tilly argues that a relatively small set of causal mechanisms underlies the entire range of collective violence. On its own, this argument is an important and provocative challenge to existing explanations of political violence. It suggests that riots, rebellions, civil wars, revolutions, and other categories of violent events that are often treated as separate phenomena by political scientists are, in fact, built from common and identifiable causal forces. While this argument is substantively important, I believe that the book is equally important from a methodological viewpoint. Building on recent research (e.g., see Doug McAdam, Sidney Tarrow, and Charles Tilly, Dynamics of Contention, 2001), Tilly rejects a standard variable-based approach and instead adopts a mechanism-based method to explain variation in collective violence. Students interested in violence will benefit from reading this book, but so will political scientists interested in the application of a methodology that seeks to highlight causal specificity. After a brief review, I will return to this methodological issue.

In The Politics of Collective Violence, Tilly first develops a typology of collective violence determined by two dimensions—salience of short-run damage (that is, the extent to which relationships are dominated by violent rather than nonviolent interaction) and the extent of coordination among violent actors. The result is seven types of violent episodes: violent rituals, coordinated destruction, opportunism, brawls, broken negotiations, scattered attacks, and individual aggression. Each type is typical of a certain space on the salience-coordination field of variation. For example, violent rituals are high on salience and coordination; brawls are high on salience but low on coordination; broken negotiations are low on salience but high on coordination. Tilly's goal is to explain why different times and places experience variation in these types of violent episodes. Moreover, the work attempts to explain how one type of violent episode evolves or creates the conditions for violence of a different type. If we can understand these types of processes, then we will be in a better position to address such puzzling questions as why collective violence often occurs in waves, why social interactions so easily shift from nonviolence to violence, and how regime types influence different levels and forms of collective violence.

Tilly searches for the causes of this variation in combinations and sequences of mechanisms, rather than in laws or models. As he defines, “Mechanisms are causes on the small scale: similar events that produce essentially the same immediate effects across a wide range of circumstances” (p. 20). For example, “brokerage” is the joining of two or more previously less-connected social sites by a third actor. “Certification” involves the validation of actors by external authorities. In total, Tilly employs perhaps 12 to 15 such mechanisms in the book. It is important to note that the same mechanisms show up in different forms of violence. It is the specific combination and sequence, as well as the setting, that determines the causal effect on collective violence. Changes in type of violence are explained through changes in the set of operative mechanisms or their sequence. For example, coordinated destruction, involving high salience and high coordination, results from mechanisms of brokerage, network-based escalation, and activation of political identity. However, if brokerage fails to operate, the level of coordination among social sites will drop, and the violence may mutate from coordinated destruction to opportunism. Such an explanation helps explain how politically oriented violence can descend into violence marked by pillage and rape.

Chapters 4 through 9 each illustrate one type of violence. Each chapter lists a set of causal mechanisms and processes and theoretically describes how they create a certain type of collective violence. Tilly then uses a diverse set of cases, ranging both across centuries and across the globe, to illustrate the argument.

Some of the substantive findings of the book are less than surprising. For instance, the conclusion that high-capacity democratic regimes are far less prone to violence than low-capacity nondemocratic regimes will not come as a shock to most students of conflict. However, Tilly's goal in this work is not so much to uncover new relationships among broad factors as it is to develop a much higher level of causal specificity for explanations of political violence. In his view, a concentration on mechanisms provides this causal specificity, while retaining a measure of parsimony. Tilly's goal here is to develop a new way of thinking about collective violence that goes beyond correlation and beyond present compartmentalized debates.

There are costs to a mechanism-based approach. If too many mechanisms are cited, the explanation lapses into description simply dressed up with a new terminology. Mechanism-based explanations are difficult to falsify. Some will argue that good variable approaches based on coherent theory will already contain, at least implicitly, an idea of causal mechanism (for an extended discussion, see Arthur Stinchcombe, “The Conditions of Fruitfulness of Theorizing About Mechanisms in Social Science,” Philosophy of Social Science 21 [1991]: 367–88).

Given these potential problems, the book raises the question of why Tilly feels compelled, especially after so many years as a leader in the field, to explicitly adopt a method based on mechanism and process. Perhaps his answer is most clearly stated in a previous article. After describing the standard political science approach as a combination of propensity and covering law explanations trying to capture general uniformities in human political behavior, he writes: “Despite more than a century of strenuous effort, political scientists have securely identified no such uniformities. But they have recurrently identified widely operating causal mechanisms and processes. Rather than continuing to search for propensity-governing laws, it would therefore make sense to switch whole-heartedly toward the specification of mechanisms and processes” (“Mechanisms and Political Processes,” Annual Review of Political Science 4 [2001]: 25).

Whether the majority of political scientists would agree and make such a switch is highly unlikely. However, the numbers of those already practicing a mechanism-based approach is growing. One problem is that a plethora of definitions and usages of mechanism and process has created confusion. My own work on collective violence, for example, is also mechanism based. However, the mechanisms underlying my explanations are all at the individual level and primarily cognitive. While advancing our substantive knowledge of collective violence, The Politics of Collective Violence also provides an exemplar of one type of mechanism approach that serves to clarify and advance this methodology. As Tilly hopes, the book should open doors on several issues.