You are on page 1of 16

Sociological Theory http://stx.sagepub.

com/

Processual Comparative Sociology: Building on the Approach of Charles Tilly


Chares Demetriou Sociological Theory 2012 30: 51 DOI: 10.1177/0735275112437172 The online version of this article can be found at: http://stx.sagepub.com/content/30/1/51

Published by:
http://www.sagepublications.com

On behalf of:

American Sociological Association

Additional services and information for Sociological Theory can be found at: Email Alerts: http://stx.sagepub.com/cgi/alerts Subscriptions: http://stx.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav

>> Version of Record - Apr 3, 2012 What is This?

Downloaded from stx.sagepub.com by guest on September 17, 2013

437172
7172DemetriouSociological Theory 30(1) 2012

STXXXX10.1177/073527511243

Processual Comparative Sociology: Building on the Approach of Charles Tilly


Chares Demetriou1

Sociological Theory 30(1) 5165 American Sociological Association 2012 DOI: 10.1177/0735275112437172 http://st.sagepub.com

Abstract
Charles Tillys work on process analysis offers a methodological approach to comparative-historical sociology that can be considered paradigmatic. Y et the approach has been widely criticized for lack of rigor. This paper maintains that the problem lies in insufficient clarification of the approachs central concept: mechanism. Once scrutinized, the concept reveals a tension between its connotation and its denotation. This can be addressed in two ways: either by maintaining what the concept connotes according to Tilly but limiting what it denotes (thus limiting the paradigms scope conditions) or by limiting what it connotes and maintaining what it was intended by Tilly to denote (thus maintaining wide scope conditions). Elucidating the possibilities of processual comparison is particularly important for comparative-historical sociology because the subfield rests upon processual presuppositions.

Keywords
Charles Tilly; processes; mechanisms; comparison.

The concept of process is as old as sociology itself yet for the most part it has remained in the background. Although sociologists often refer or allude to social processes as they explain phenomena such as social change, there have been few metadiscussions of processes. The work of Charles Tilly is an exception. Not only does Tillys work reconstruct and compare wideranging historical processes, but it also treats processes more abstractly. One of the key features of processes for Tilly is the quality of emergence resultant from other processes. Thus, to study processes is to study their relationships with constituent subprocesses. Tilly calls these subprocesses mechanisms, and it is under this label that Tilly builds his metatheory, holding mechanisms to be the central units of comparative analysis and explanation. His varied processual analytic narratives pivot on the relations mechanisms have with each other as well as with other elements of the social worldincluding social structure but also conjuncture and temporality. Although Tillys writing has generated a voluminous secondary literature, most of this has been concerned with his substantive theoriesof revolutions, urbanization, state formation, inequality, democratization, contentious politics. Only a small portion of commentaries address methodological issues. Most that do take the form of reviews of his collaboration with Doug McAdam and Sidney Tarrow (McAdam,
1

Queens University, Belfast, United Kingdom

Corresponding Author: Chares Demetriou, School of Politics, International Studies and Philosophy, Queens University, Belfast, 21 University Square, Belfast, BT7 1PA, UK Email: chares.demetriou@gmail.com

Downloaded from stx.sagepub.com by guest on September 17, 2013

52

Sociological Theory 30(1)

Tarrow, and Tilly 2001). These reviews have tended to be critical of the approach and hold that Tilly and his collaborators fall short of their goal of introducing conceptual and analytical rigor vis--vis mechanisms and comparison. Pamela Olivers (2003) review goes to the heart of these objections: The mechanism (I would call it a process) that comes up most often [in the volume] is called brokerage, a term borrowed from network theory [...] to refer to people who connect previously unconnected groups, and the book stresses its importance in virtually every episode. But the nature of this brokerage varies tremendously from case to case. For the Mau Mau, it is a semi-planned recruitment strategy that is pretty similar to a diffusion process: particular groups are recruited and converted, and then sent out as recruiters of others. Taxi-drivers are brokers here because they meet a lot of people and can thus provide communication bridges to new groups. But in other situations, e.g., the development of the Italian state, the brokers are more in line with the usual meaning of the term, people who help parties with partially conflicting interests find a mutually agreeable bargain. Mediating a bargain over interests between extant groups and providing a communication link between previously unconnected individuals are not the same process, and simply calling them both brokerage does little to advance understandings. (Pp. 12122) Oliver raises a fundamental challenge to Tillys process-mechanism approach. As she illustrates the lack of clarity with regard to the definitions and referents of the mechanisms named by Tilly and his coauthors, she helps us see what the problem is at the root: namely, the concept of mechanism in the abstract. There is indeed a lack of clarity in Tillys work about the phenomena indicated by the concept. Olivers demand, in effect, is to clarify what the concept of mechanism denotes and what it connotesor, in Giovanni Sartoris (1970) classic terms, to clarify its extension and intension at the high end of the ladder of abstraction. Yet a challenged prospective paradigm may mean a failure in the making or a success in infancy. The quandary needs to be addressed because the stakes are not trivial. From Tillys point of view, they are epistemological and rather stark. Mechanistic explanation is for him particularly fitting for historical and other qualitative research because it avoids both the pretentions of positivism and the paralysis of epistemological skepticism (Tilly 2002, 2008b). Here one may object, of course, that the two options rejected by Tilly are merely straw men. Still, Tilly has a point. Taken as a method, his approach is arguably one of the few developed challengers to the approach to qualitative comparison following the positivist, variableoriented paradigm of quantitative analysis (e.g., King, Keohane, and Verba 1994). To the extent that it has cohesion and efficacy, it can plausibly be considered a paradigmatic alternative. Tillys perspective is, in general terms, realist. For him to explain scientifically is to describe operative processes (Manicas 2006; Tarrow 2008). But realism lacks internal coherence, and commentators see affinities between Tillys work and different branches of realism. For example, George Steinmetz (2010) views him, mutatis mutandis, as a close relative of critical realism, while Neil Gross (2010) views him as a not-so-close relative of pragmatism, itself arguably a school with an affinity to realism. When it comes to the philosophy of mechanisms, Tilly does not elaborate on the link between mechanisms existing on the real, ontological level and mechanisms devised on the explanatory, epistemological level. The relation between these two levels is a vexing issue, and Tillys assumption that his explanations describe real mechanisms does not address it adequately (Demetriou 2009). However, I argue that Tillys brand of realism is closer to Mario Bunges (1997, 2004) philosophy of science than anything else. The fit is not perfect, especially since Tilly does not accept Bunges emphasis on the material foundations of mechanisms, but Tillys general epistemology, methodology (e.g., valuing the discovery of knowledge through conjecture), and definition of mechanism remain essentially Bunge-esque. In this essay I reevaluate the paradigmatic prospects of Tillys approach in light of my interpretation of him as a Bungean realist. I do so through a twofold analysis. First, I develop an exegesis of Tillys conceptualization of mechanism, emphasizing in particular Tillys ambiguity regarding the concepts connotation, that is, its possessed characteristics. It is as a result of this ambiguity, I maintain, that some of the

Downloaded from stx.sagepub.com by guest on September 17, 2013

Demetriou

53

most serious criticisms of the paradigm have been raised, though the criticisms leave the ambiguity unarticulated. Second, I develop two basic alternative ways in which the paradigm can proceed in order to remedy its shortcomings. The one alternative upholds what Tillys concept of mechanism formally connotes, but narrows what it denotes (the phenomena it indicates), a move that implies a narrowing of scope conditions. The other alternative limits what the concept connotes in order to maintain what Tilly intends it to denote, a move that allows wide scope conditions. As will become clear in the course of this analysis, Tillys inadvertent leaning is toward the second alternative. In the concluding section I revisit the question of the epistemological and methodological stakes for comparative-historical sociology.

TILLY AND HIS CRITICS


At the conceptual level, Tillys approach pivots on the definition of mechanism and on the relation between mechanism and process. These two theoretical facets relate to the conceptual specification of the constituent elements of a mechanism, a task that Tilly fails to complete adequately. I will argue that these two theoretical facets, as they are presented by Tilly, are in tension with each other and that this tension provides a key hurdle for the development of the paradigm. Let us start with the definition. Throughout different publications, Tilly defines a mechanism as a delimited class of events and occurrences that alter connections among social units (McAdam et al. 2001:24; Tilly 2001:25-26; Tilly and Tarrow 2007:29). Elsewhere, he paraphrases it thus: A mechanism is similar events that produce essentially the same immediate effects across a wide range of circumstances (Tilly 2003:20).1 In both versions of the definition, a mechanism explicitly includes an outcome and corresponding antecedent events, while implicitly including an initial condition. Schematically, the mechanism can be presented as X Y, where X is the initial condition, Y the outcome, and the arrow the series of events.2 What is the relation between a mechanism and a process? According to Tilly, it is a constitutive relation: Mechanisms constitute a process. However, a mechanism can, in its turn, be broken down to constituent submechanisms, while, conversely, a concatenation among processes may constitute a new, larger process. What is called a mechanism and what a process is therefore arbitrary and dependant on the context of the analysis; what in one context is a mechanism, in another context is a process. In other words, the labels mechanism and process only make sense in relation to each other (McAdam et al. 2001:27). This means that the scheme X Y also describes a process, where X is the initial condition, Y the outcome, and the arrow a mechanism concatenation. The obvious question arising at this juncture is about the compatibility between a mechanism concatenation making up a process and a series of events making up a mechanism. Are they the same thing? Tilly never discusses this, but one can presume that a mechanism concatenation and a series of events are, at an ontological level, the same sort of thing. Similar to a mechanism concatenation, after all, a series of events may have an initial condition, an effect, and constituent events. (Not all series of events have distinct outcomes, of course, but then neither do all mechanism concatenations.) The difference between a mechanism concatenation and a series of events, one may presume further, is an epistemological one. That is, a mechanism concatenation has a more precise abstract form and analytical specificity than a series of events. If this understanding sufficiently addresses the question of the relation between the two concepts, Tillys conceptual scheme becomes smooth: The connection among processes, mechanisms, and events follows a perfect symmetrya logic of regress, of parts and wholes. But this is empirically untenable. As a package X Y, a mechanism will not work across all empirical contexts. Indeed, some of the most severe criticisms leveled against Tillys approach, and most particularly against the version of this approach developed by McAdam et al. (2001), stem from the expectation that a mechanism is supposed to manifest itself in a standard form and work in a standard way across episodes. Thus, in reviewing the mechanisms and processes named by McAdam et al. (2001), Earl (2008) and Staggenborg (2008) follow Oliver (2003) and argue that the links between broad mechanisms and

Downloaded from stx.sagepub.com by guest on September 17, 2013

54

Sociological Theory 30(1)

constituent mechanisms are generally weak, while Rucht (2003) and Koopmans (2003) argue that the same concatenation of mechanisms is presented in certain instances to explain different outcomes. Rucht (2003) argues further that there are instances in McAdam et al. (2001) in which the same outcome is said to be explained by different concatenations of mechanisms. Norkus (2005), finally, examining Tillys work more comprehensively, claims that the series of events that amount to mechanisms and processes are not always specified by Tilly, leaving the mechanistic explanation incomplete and thus unable to demonstrate productive continuity (more on this term below). These are sound and penetrating criticisms. What is particularly interesting, however, is that Tillyalone or with his collaboratorsnot only fails to live up to the definition he starts with but is apparently not interested in doing so. That is, the expectation that X Y would appear as a package from one episode of contention to another is not one that Tilly holds of his approach. In pursuing comparisons across different episodes of contentious politics, Tilly is not interested in presenting the mechanisms and processes that he singles out in a way such that the outcome (Y) is matched by a cross-episode constant mode (the arrow in the diagram) and a constant initial condition (X). This is abundantly clear in his discussion of the various processes he names. Consider as an example references Tilly and Tarrow (2007) make to scale shift, a process they define as the increase or decrease in the number of actors and/or geographical range of coordinated claim making (p. 217). They present it as a complicated process that may be variously constituted through such events as strikes, suicide bombings, and court decisions (p. 95). These references alone should point to varied cross-episode constitutions of the process, but the authors go on to acknowledge such variety as they produce an illustrative figure demonstrating, in their words, alternative routes to upward scale shift. Thus, as they explain, possible routes may start from local action and feature a successive concatenation of the mechanisms of brokerage, attribution of similarity, and emulation, and thereby produce coordinated action on a larger scale, or they may start from local action and feature a successive concatenation of attribution of similarity, relational diffusion, and emulation, and thereby produce coordinated action on a larger scale (p. 96). The cross-episode constitution of the process is clearly varied, then, and the same holds for every process identified by McAdam et al. (2001). Each is theorized to operate in different episodes of contention through different mechanism concatenations. Given that the choice between the labels process and mechanism is arbitrary in the face of a concrete set of social patterns, one can readily extend the above assessment of processes to cover mechanisms as well. It is not empirically tenable, therefore, to hold that a named mechanism features a constant mode of constituent events across episodes. However, while Tilly is clear about the constitution of processes by corresponding mechanisms, he muddles the point when he refers to named mechanisms and their constitution. Consider the mechanism diffusion, defined as the transfer in the same or similar shape of forms and claims of contention across space or across sectors and ideological divides (McAdam et al. 2001:68). In the course of their substantive analysis, McAdam et al. (2001) identify varied modes of this mechanisms constitution. Thus, in the context of the secessionist mobilization within the USSR in the 1980s, diffusion is said to have resulted in part from deliberate proselytizing by the Balts, but it was also transmitted by media reports and by the elite. (p. 252), and in the context of the genocide in Rwanda, diffusion of killing is said to have taken place through a combination of interpersonal networks and mass media (p. 339). It becomes clear from these references that mass media and interpersonal networks, as well as potentially other means, can variously work to produce diffusion. No cross-episode constant mode can therefore be said to exist regarding this mechanism. Ultimately, this point could be repeated about all other mechanisms identified by Tilly in his various writings. While no standard mode characterizes the constitution of either a process or a mechanism, Tillys substantive discussions do not paint a symmetrical picture. He identifies constitutive patterns when he refers to named processes because he links them to concatenations among named mechanisms, but when he refers to named mechanisms he does not specify their constitutive patterns with the same precision; he elaborates substantively on the operation of named mechanisms, to be sure, but he neither abstracts nor names the form of such operation. Ultimately, he ignores the theoretical issue, and only implicitly does he suggest that, in

Downloaded from stx.sagepub.com by guest on September 17, 2013

Demetriou

55

thinking about it, there should be conceptual gravity toward the outcome. Still, it is important that this implication about the weighting of the outcome is present. It can be inferred not only from references such as the ones above but also by considering the logic followed by Tillys definition of named mechanisms. For these mechanisms are essentially defined by the outcome that is produced rather than by the combination of outcome and mode (not to mention the initial condition). To refer to the examples already discussed, it is by virtue of the outcome scale shift that the mechanism scale shift is conceptualized, and it is by virtue of the outcome diffusion that the mechanism diffusion is conceptualized. Nowhere in the formal definition of these mechanisms, and in any other mechanism identified by Tilly, do we see a reference to the mode through which the outcome is generated. Clearly, Tillys distancing of himself from a theoretical discussion of mechanism constitution has not saved his approach from criticisms. Again, these criticisms boil down to the question of the mechanisms mode. They can be summarized here in two interrelated observations: first, that different modes may well generate the same outcome; second, that different outcomes may well be generated by the same mode. Examples of the first condition (often referred to as multiple realizability) were given earlier; examples of the second condition include a single series of verbal exchanges between two political actors that generate both polarization (defined as increasing ideological distance between political actors or coalitions; Tilly and Tarrow 2007:217) and boundary activation (defined as the increase in the salience of the usthem distinction separating two political actors; Tilly and Tarrow 2007:215). What does all this imply for the paradigmatic status of Tillys approach? I argue that this discussion clarifies two pathways for the paradigm: either (1) to pursue the empirical search for the package X Y, something that would imply it narrowing of scope conditions and differentiation among them, or (2) to render the concept of mechanism a practical synonym of outcome and distinguish it from the concept process, something that would allow sweeping scope conditions.

AMBITIOUS INTENSION AND MODEST EXTENSION


In this section I discuss the first of these two possible pathways. I will maintain that a mechanism can repeatedly instantiate itself as a package X Y only within well-structured social configurations that give hypostasis to both the activity and the entities constituting the mechanism. This means that mechanisms so conceptualized can become part of a research program only when scope conditions are specified according to terms pertaining to this sort of social configuration. The idea that a mechanism has a constitutive material dimension is well established among philosophers of mechanism, even among ones representing relatively divergent perspectives. It is telling that Bunge and Roy Bhaskar, two philosophers who established their own distinct schools in what is broadly considered to be the realist tradition, both hold the interplay of material entity and activity to be constitutive of mechanism. They differ from each other, among other ways, in how they conceptualize mechanism. Bunge (1997, 2004), much like Tilly, conceives mechanisms as events generating an outcome, while Bhaskar (1979) conceives them as generative qualities that may be realized or remain potential. Bunge proposes a two-way relation between system and mechanism in which a system, conceived to be comprised by bonds among material parts, enables the operation of mechanisms that, in their turn, help maintain the system. Bhaskar, too, sees material structure as being constitutive of mechanism, for he holds that it is the inner and environmental structure of a thing that creates the generative quality that the mechanism stands for (see Demetriou 2009 for a more sustained comparison between the two philosophers). Explanations through mechanisms have gained acceptance particularly in biology, and it is here perhaps that the generative role of the entity-activity interplay can most clearly be seen. Thus, echoing Bunge and Bhaskars general orientation, Machamer, Darden, and Craver (2000) point out that various entities in biological organisms possess capacities for activity that, when realized, produce a concatenation of changes and ultimately a distinct phenomenonfor example, protein synthesis. They add, the organization of these entities and activities determines the ways in which they produce the phenomenon. Entities

Downloaded from stx.sagepub.com by guest on September 17, 2013

56

Sociological Theory 30(1)

often must be appropriately located, structured, and oriented, and the activities in which they engage must have a temporal order, rate, and duration (Machamer et al. 2000:3). In this sort of structured material context, they add, mechanisms recur in regular ways from beginning to end. And any intelligible description of mechanisms, they therefore conclude, must demonstrate productive continuity, showing without gaps the stages through which a mechanism operates from beginning to end. For biology, then, explanations through mechanisms involve an account of the generative role of parts in the context of structured wholes. In Cravers (2006) words, such explanations explain the behavior of the mechanism as a whole in terms of the organized activities of and interactions among its components (p. 369; italics in the original). Accordingly, it is possible to have precise expectations of mechanistic explanations that relate distinctly to the outcome, the components, and the organization of the components. Regarding the outcome, Craver maintains that one can analyze it in terms of the rate of rise, peak magnitude, refractory period, side effects, and so on, as well as in terms of its inhibiting conditions, nonstandard conditions, and other background conditions. Regarding the components, he maintains that one can analyze their stability as clusters, the robustness of their interlinkages, and even their transposability into other sets of components and activities. Regarding the organization of the components, finally, he maintains that their various spatial and temporal dimensions can be analyzed, thus allowing for evaluations of density, distribution, sequence, and so on (Craver 2006:368-73). If the mechanistic explanatory program is such a straightforward affair in biology, however, can it be the same in sociology? After all, one may object to any suggested parallels between biology and sociology: biologys subject matter is far more structured than is the social world. While this fact is indisputable, some parallels may hold, for it is also true that the social world does feature bounded units with well-defined parts. Examples of this are many and include formal organizations and institutions. It is not a forgone conclusion that such bounded social units repeatedly produce distinct types of outcomes through the same mode of interaction among their components, but this is something that happens often enough, as exemplified by bureaucratic procedureselection processes, university admission processes, religious rituals, and so forth. Attributing paramount significance to the entity-activity bound, rather than to activity alone, has repercussions for comparison. If the entity-activity interplay is constitutive of mechanism, mechanistic comparisons across social configurations cannot be presumed to be appropriate on a wide scope. They can be appropriate among similar configurations but not among diverse ones. This is true of biology as well as of sociology. In biology, for example, polarization as a mechanism in the context of the cell body is not analogous to polarization in the context of an organ, for example, the heart. This is so because the two respective sets of components (cell body and organ) and their operations are not analogous, even though both produce an outcomepolarization of somethingthat may be deemed analogous in some respect. Likewise, polarization in the social worldfor example, increased ideological distance between actors, as per Tillycannot entail equivalent operations when taking place in nonequivalent contexts, say, for example, between the context of voluntary associations that allow exit and the context of organizations that strictly sanction dissent. (As the previous section showed, Tilly does not discusmuch less delineate equivalent structural contexts in the sense expounded here, so one cannot demonstrate equivalent operations of a mechanism such as polarization through his substantive analyses.) If a mechanistic comparative program becomes possible through the narrowing and differentiation of its scope, then the analysis needs to treat similar social configurations. The contrast of similarity rather than the comparison of dissimilarity is, strictly speaking, the modus operandi of the program. In searching for recurrent mechanisms, the existence of some variation in mechanism operation must be taken to be the rule rather than the exception because even with regard to well-bounded social units the mechanism cannot be expected to work in exactly the same way from instantiation to instantiation. This is so, for one, because identified mechanisms always imply the ceteris paribus clause. That is, any given mechanistic explanationthe matching of regular effects to corresponding regular modes and initial conditionsis in fact an idealization of a multitude of actual generative operations, most of which are either

Downloaded from stx.sagepub.com by guest on September 17, 2013

Demetriou

57

ignored or not understood by the analyst (this is so despite the premises of the realist approach; see in particular Demetriou 2009; also, Craver 2006). Moreover, variation in the operation of any given mechanism across instantiations should be expected because in the social world, in contrast to the biological world, the coherence and enclosure of social configurations relating to mechanisms are due more to organizational factors than to material ones; the components of social units, after all, include humans who are also components of other social units. This means that chains of interaction constituting a mechanism in the social world may be ramified well beyond the mechanisms bounds. Such extension of interaction means, in turn, that even subtle changes in the social configuration may generate a different mechanism. Thus, for example, in the scheme Z |X Y, elements X and Y may be bounded structurally so that X Y is a mechanism, but a change in social configuration may shift the bound (the vertical line in the scheme) so that Z Y now becomes the mechanism (with Y, perhaps, slightly altered). To see how plausible such a change is, one only needs to remember that any formal organization is affected not only by its members and internal operations but also by external stakeholders and external operations. External influences may affect the organizations primary functions but may also, if sustained in time, play a part in the evolutionary reproduction of the organization itself. If one were to treat the primary functions of the organization as mechanisms, therefore, one would readily conclude that changes in the organizational form would redefine these mechanisms. Take as an example the Paris Opera in its early years, as described by Johnson (2007). The Opera was formally a royal academy but, unlike other such academies (e.g., the Academy of Science, the Academy of Painting and Sculpture, etc.), it gave public performances and charged admission. It was therefore an organization with a hybrid form, combining the form of a commercial troupe with characteristics of the other royal academies. The commercial facet of the Opera predominated, and it is elements related to it (e.g., entertainment market forces) that primarily determined the pattern of stage productions. However, elements relating to the organizational form of the royal academies were also present. Thus, for example, royal premises could and did provide venues for performances, royal subsidies could and did contribute to stage productions, and so on. In this context one may consider stage production to be a mechanism (complicated for sure, since it includes anything from repertoire, casting, and rehearsals to accounting and advertising) where any particular stage production would represent the mechanisms instantiation. It can be readily seen that the operation of the mechanism depended on the organizational fluctuations of the Opera with regard to its commercial and royal facets. Thus, for example, the royal input in stage productions could vary from time to time, as when, plausibly, the cost of production in a given year was too high to be balanced off by commercial revenues. This means, then, that the bounds of the configuration that the Opera was could shift and that the mechanism to which this configuration gave hypostasisthat is, stage productionscould shift along the way, for the constitutive elements and dynamics going into a production changed depending on whether market forces or subsidies predominated. This example helps us see where the focus of the mechanistic paradigm in this path of development is. It is on specific mechanisms that are generated in equivalent social configurations and are compared by way of contrasting their instantiationsand by this I mean that it is possible to compare, say, productions among different Opera organizations as well as ones by the same organization. Knowledge may therefore advance through the analysis of the mechanisms outcome, components, and components interactions. This may include an analysis of the specific dimensions discussed by Craver or may follow other similar analytical strategiesclearly, organizational sociology has pursued similar analyses, albeit without focusing on the concept of mechanism systematically. But once the social configuration is deemed well structured, it is possible to make analyses and comparisons with regard also to submechanismsfor example, the casting of actors within the Operafor these would relate to the elements changing regularly along the path of the mechanisms productive continuity. Also, analysis can be made with regard to the functions of the mechanism within its environmentfor example, the role of Opera productions in legitimating the kingbut this is a concomitant line of analysis. Lastly, given that the social world features variation even among well-bounded units, it is important to stress that this effort will require important analytical calls as

Downloaded from stx.sagepub.com by guest on September 17, 2013

58

Sociological Theory 30(1)

to which outcomes can be considered regular and which secondary, which mechanism modes can be considered regular, possible, or exceptional, and so on. In sum, the path of research seeking to analyze mechanisms as X Y packages must pay attention to the social configurations that give hypostasis to such packages and allow them to instantiate themselves repeatedly. This will require narrowing the scope conditions of comparison, where only commensurate social configurations are compared and contrasted. For this, the yardstick must be social configurations generative capacities. What this means in practical terms may vary (e.g., see Mingers 2011), but it is worth noting that the general orientation in sociology sometimes termed American structuralism may not be this paths best guide. For a social configuration with generative capacities is not the same as the form of the content, which Simmel advocated with regard to sociation and which American structuralism largely adopts. This sort of abstraction does not tell one much about whether or not what is identified as structure gives hypostasis to a mechanism. It may do so in some instances but not in others, and so it cannot be the yardstick in the analysis of mechanisms.

MODEST INTENSION AND AMBITIOUS EXTENSION


In this section I turn to a second, alternative pathway for Tillys paradigm. If the emphasis is placed on the mechanisms outcome, rather than on the combination of mode and outcome, then the possibility opens up for a paradigmatic direction in which the connections of the outcome to its environment gain analytical primacy. This means that the way the outcome combines with other outcomes can be conceptualized to be the mode of a process. In this approach, it is crucial to draw a distinction between a mechanism and a process. As seen, the notion that a mechanism operates as a package X Y in diverse contexts and in different aggregations of events (or chains of interaction) is empirically untenable. This limitation is taken for granted in biology: The cell, the organ, the organism, and the ecology are not considered equivalent contextseven though they are linked to each otherand are not thought to foster the same mechanisms. But if we do not want to break down the social world into entities in the fashion of biology, what remains equivalent about it for purposes of analysis through mechanisms? How can a wide range of phenomena in the social world, in other words, be subjected to mechanistic explanation? The key to the answer is to acknowledge that a mechanistic explanation in this direction, unlike the direction of the first paradigmatic pathway, cannot provide a comprehensive account of all generative forces at work relative to a phenomenon. This not only accounts for a mechanisms multiple realizability but suggests that factors beyond an identified set of mechanisms contribute to the precipitation of the outcome of a process. It is possible to conceptualize a mechanism through its outcome. In this move the mode is bracketed and the Y of the X Y diagram is essentially called a mechanism. As a result, the identification of the mechanism-cum-outcome in a wide range of contexts becomes possible. As Tillys work effectively showed, one can identify such named mechanisms-cum-outcomes as polarization, certification, brokerage, diffusion, and so on, across a wide range of social circumstances. But whereas Tilly is criticized for not providing sufficient abstractions of the constitutive elements of the mechanisms he names, an explicitly outcome-driven conceptualization of mechanism can face no such objections. The conditions and forces generating such outcomes are simply not at issue vis--vis the explanation. It is useful to give a distinct label to what it has been referred to above as mechanism-cum-outcome. Let us call it m-outcome. This label aims at showing two things: first, that the outcome recurs as such in a variety of contexts, and it is therefore not a one-time outcome; second, that a commitment is maintained on the realist position that the social world is comprised of interactive generative forces at various levels of reality and by corresponding emergent social outcomes (Bhaskar 1986; Manicas 2006). The latter point therefore recognizes that though the operations generating an m-outcome are mechanisms, their superfluity is not subject to idealization by the sociologist, either because this is not well understood or for reasons of economy vis--vis the explanatory narrative. In this approach, then, an m-outcome stands for an emergent outcome the generating forces of which are bracketed through an epistemological move.

Downloaded from stx.sagepub.com by guest on September 17, 2013

Demetriou

59

This conceptualization of mechanism is helpful for research purposes first because it disciplines empirical work by requiring the identification and labeling of distinct outcomesmuch like the ones Tilly identifies and labels. Furthermore, it is the first step to a comparative program. In particular, comparisons using this notion of mechanism pivot on the outcomes relation to its environment. Whereas the first paradigmatic pathway focuses on the mechanisms internal dynamics, the second paradigmatic pathway focuses on m-outcomes extended dynamics. An m-outcomes connections to its broader context vary. They may include linkages to structural conditions, such as organizational contexts, as well as more fluid linkages. Nothing prevents research within this paradigm from studying such connections comprehensively. Yet a focus on the linkages an m-outcome has with other m-outcomes is at the center of a line of research especially appropriate given mechanistic-realist presuppositions. That is, given the task of explanation, an m-outcomes function in the explanation stems from its concatenation with other m-outcomes such that the concatenation explains a distinct emergent outcome, which is the explanandum in the context of a given stream of interaction. The concatenation of m-outcomes plus the explanandum can be thought to constitute a process. This resembles the logic of Tillys approach. However, the crucial difference between Tillys take and the present proposal rests on explicitly drawing the distinction between m-outcome and process. This is a distinction on the epistemological level (rather than on the ontological level) holding that the process is subject to an analysis of the featured relation between outcome and mode, but that the m-outcome is not subject to such an analysis. Compared in this approach are processes that are deemed similar by virtue of similar explananda. These phenomena are comparable because of their similar end points rather than the patterns that lead to those end points, which in all likelihood are not similar across time and space. The patterns then become subject to analysis by way of process tracing in which m-outcomes are key accentuation points in the process. Since this approach is similar to Tillys, adjusting and correcting for conceptual lacunae and unexplored presuppositions, a reference to Tillys substantive analyses can offer an illustration. In his book Democracy, Tilly (2007) argues that the complicated process of democratization is constituted by three subprocesses: the integration of trust networks into public politics, the insulation of public politics from categorical inequality, and the reduction of autonomous power clusters. Each of these subprocesses is comprised across episodes by an array of possible mechanisms (m-outcomes, under the new label), some of which Tilly discusses on the theoretical level and others of which he analyzes substantively with reference to specific historical contexts. Regarding the insulation of public politics from categorical inequality, for example, he analyzes it within the context of democratization in South Africa. He argues that in the last two decades of Apartheid rule two mechanisms (m-outcomes) generated through the antiapartheid struggle were particularly important, a mechanism (m-outcome) producing sustained popular resistance against the direct inscription of racial categories into politics and a mechanism (m-outcome) forging powerful coalitions across racial and ethnic categories. Beyond this substantive analysis, Tilly (2007) suggests that the mechanisms (m-outcomes) constituent of insulation of the public politics from categorical inequality include the dissolution of state controls that support current relations among social categories, the equalization of assets and/or other well-being across categories within the population at large, and the reduction or state containment of private controlled armed forces (p. 106-32). What this example from Tillys work illustrates is, first, that similar end points of processes (e.g., an explanandum such as the insulation of public politics from categorical inequality) render these processes comparable across episodes and, second, that such processes are variously constituted. Herein lies a major presupposition of this approach to comparisons (left unacknowledged by Tilly): An identifiable set of m-outcomes cannot be considered to exhaust all the constitutive elements of a process. It is partly because of this that a process cannot be reduced to the same pattern of constituent mechanism, that is, something that would have singlehandedly explained the process across episodes. The South Africa example shows also that it is the epistemological context that determines whether a phenomenon is treated as an m-outcome, a process, or bothnotice how insulation of public politics from categorical inequality is both an explanandum, explained by a variety of m-outcomes, and an explanans of

Downloaded from stx.sagepub.com by guest on September 17, 2013

60

Sociological Theory 30(1)

democratization. This does not negate the distinction between process and m-outcomes exactly because this distinction exists only on the epistemological level. In this line of comparison, therefore, not unlike the original intent of Tilly, what can be labeled an m-outcome (constitutive of a process) can in another epistemological context be labeled a process (an explanandum) to be explained through different constituent m-outcomes. In short, whereas Tilly was criticized for not demonstrating productive continuity (an expectation better suited for the first pathway of research), an outcome-driven conceptualization of mechanism may face no such objections since the mechanisms constitution, or, better, the abstraction of the mechanisms constitution, is bracketed. The concatenation among m-outcomes matters, of course, and it can be criticized if it should appear to rest on a dubious empirical analysis. Nevertheless, the acknowledgment that identified concatenations do not provide a full explanation of a processbecause they do not exhaust the processs determinantsmust qualify the efficacy claim of this paradigmatic pathway and thereby lower the expectations one should have from it.

CONCLUSION: PROCESS AND COMPARATIVE-HISTORICAL SOCIOLOGY


Sociologists in general and historical sociologists in particularly consider happenings or events as they unfold in time to be the raw elements of their subject matter, which then they typically reconstruct in some sort of structure-action narrative. But while many of them also see processes in their prefigured subject matter and accordingly discuss processes in their analytical narratives, few are ready to go into details about the abstract constituents of processes. Tilly, alone and with his two main collaborators, stands out for promoting the idea that processes can indeed be delineated abstractly. His work takes processes to be comprised of mechanisms, and it is around mechanisms that his analytical narratives weave other, more widely accepted abstract elements of processes, such as sequence, contingency, and conjuncture. The epistemological-methodological dimension of Tillys effort, however, has failed to gain the favor of most established historical and political sociologists, with a number of them taking issue in particular with Tillys lack of conceptual clarity and empirical rigor in relation to it. Doug McAdam and Sidney Tarrow, eventually recognized the problem. In an introductory article of a special issue of the journal Mobilization that commemorated the 10th anniversary of the publication of Dynamics of Contention (DOC), they write: [I]n DOC we defined mechanisms as delimited sorts of events that change relations among specified sets of elements in identical or closely similar ways over a variety of situations. We now think our use of the term events was confusing because it may have led readers to assume we were talking about occurrences rather than the mechanisms that drive them. (McAdam and Tarrow 2011:4) In the present essay I turned to the same concern, digging deeper than McAdam and Tarrow to examine the connotation (intention) of the concept mechanism. I maintained that the lack of rigor one may find in Tilly is precisely the result of insufficient clarification of this key concept. Once clarified, the concept reveals a tension with regard to its constituent elements, a tension that, as I have shown, can be addressed in two ways: by limiting the denotation of the concept and, thus, the scope conditions associated with its usage, or by limiting its connotation, which would allow for wide scope conditions. Rather than suggesting the abandonment of explanations through what Tilly understands as mechanisms, my argument creates space for the progress of such explanations by clarifying their potential as well as limitations. The extension of Tillys approach as it is described in the second pathway allows research in comparative-historical sociology to focus on those elements that are germane to processes. This approach carries some promise to move the metatheory of comparative-historical sociology forward. Such metatheorizing does not have to develop into the central task of comparative-historical sociology, nor would it prove

Downloaded from stx.sagepub.com by guest on September 17, 2013

Demetriou

61

to be unproblematic. But if what makes comparative-historical sociology less ad hoc than historiography is an attention to methodology and conceptual systematization, then the effort to promote the epistemology of the process-mechanism approach is justified. Here I would briefly suggest two areas where our understanding of processes, and by extension comparative-historical sociology, needs improvement. The first area concerns the question of how would one would go about classifying different kinds of processes in an abstract way. The field is open for creative replies, but it is worth visiting familiar ground as a start. Though generated in philosophy, the distinction between productive processes and transformative ones resonates strongly with sociological theory (Rescher 1996). In fact, this is a distinction that also takes us back to the question of m-outcomes combining to generate an outcome (productive processes) and mechanisms link to well-structured configurations (transformative processes). How would historical-comparative sociology view this distinction? A hint can be found in Tilly. Although in his work we mostly find discussion of productive processes since he focuses on the extended effects of mechanismstransformative processes are also featured. Consider his analysis of parliamentarization in Great Britain (Tilly 1997). If the British Parliament became a well-structured configuration between the 1780s and the 1830s, it became so through mechanisms that gave it hypostasis as an effective political broker. While these power-wielding mechanisms were many and varied, Tilly focuses particularly on the claims average people brought to the more powerful, that is, on bottom-up mechanisms. As Parliament was locked in a struggle with the Crown over power, Tilly suggests that bottom-up mechanisms contributed to the shift of power toward Parliament, since the people recognized the already increasing power of the parliamentarians and augmented it by seeking them as brokers. Hence the institutionalization of the British Parliament was a transformative process. Other circular processes, whereby an activity is repeated within certain margins of variation, become in time transformative processes, transforming precisely the configurations that give them hypostasis in the first placehence they are actually spiral processes rather than circular. In addition to the political process via parliament, one may note the labor process, the judicial process, and so on. One point to draw from this, therefore, is that the genealogy of institutions falls within what might be called the historical sociology of transformative processes. A second, and more important point, however, suggests that such processes cannot unfold in history merely as internal processes. That is, the circular operations characterizing them take place in conjunction with other processes, and this is a point Tillys just cited work also makes clear: Important forces behind parliamentarization pertained to the struggle between Parliament and Crown, and that is a set of dynamics outside the process defining the institutionthat is, the spirals of brokerage between constituencies and the state. So, historical sociology ought to view the distinction between transformative and productive processes as a heuristic for research rather than a steadfast distinction of actual reality, since pure transformative processes probably do not exist anywhere. Other ways to classify processes abstractly may prove much more useful than this on the level of reality and thus epistemology. This also points to the better fit of comparative-historical sociology with the second pathway rather than the first. Crucial here is the notion of temporality/time. Time requires an ontology against which to be defined, being hardly conceivable without this linkage (Abbott 2001a). More precisely, what we can know is not time in and of itself but times mark on processes, be those processes physical (e.g., aging) or social (e.g., the performance of an activity; Adam 1990). Temporality, accordingly, is a rather vague and general notion that aims at clarifying times effects on processes (and vice versa, but this is a matter that need not concern us here). Hence temporality-related concepts such as timing and tempo relate directly to this timeprocess bind. Consider tempo first. This can be thought of as the rhythm of a process, an effect that can be generated within short or long spaces of time. Relating to this, timing may be thought of as the actors adjustment to the changing tempos of processes (though a nonsubjective definition of timing is also possible). How may these two notions relate to historical-comparative sociology? Going back to Tilly and to the same work on parliamentarization can offer, once again, some clarifications. As the importance of the British Parliament was growing, and peoples reliance on parliamentarians was increasing, people abandoned forms of protest

Downloaded from stx.sagepub.com by guest on September 17, 2013

62

Sociological Theory 30(1)

promising quick returns. Thus their new, emerging mode of political engagement had two characteristics. As Tilly (1997) writes, (1) they could almost never, even in principle, accomplish their proclaimed objectives in the short run; (2) they depended for effectiveness on extensive anticipation, cumulation, and coordination of efforts by multiple actors, often substantially removed from one another in time and space (p. 235). In a review essay of Tillys parliamentarization argument (and other chapters by Tilly), Arthur Stinchcombe (1997) chooses to underline the changing time horizon of the actors. He couches his point mostly (though not exclusively) in subjective and intersubjective terms: People take into account the time into the future, their knowledge of the future becomes more complex, and they have a longer sense of future anticipation. But while this reading of Tilly is not wrong, it places the emphasis on peoples changing timing and not on the changing tempo of the processes underpinning their timing. What Tilly describes, however, is not merely a changing subjectivity and intersubjectivity of time, but a changing temporalitythat is, a change in the rhythm of the mechanisms of political demands, political brokerage, and the political process more generally. Thus he made some strides in an area that remains in need of much improvement the comparative-historical sociology aiming to treat temporality through the complicated time-process bind rather than just through the phenomenology of time. What can be derived from this is that processual investigations carry a promise to enrich our understanding of temporality by placing old subject-object questions under fresh ontologies and epistemologies. Beyond temporality and abstract descriptions of processes, more widely recognizable elements of process are also important. These are elements that inform processual ideas of causality, such as conjuncture, contingency, and sequencing. In concluding this discussion, therefore, it must be stressed that clarifying a sensible way forward along the road charted by Tilly is important not so much because this road opens up new horizons for comparative-historical sociology but because it safeguards some of comparative-historical sociologys valued epistemological commitments. This is the case as this road presents a realist paradigmatic alternative to the approach to qualitative comparison emulating the methodological presuppositions of the quantitative method of covariance. The paradigm centered on the analysis of covariance, called methodological positivism by George Steinmetz, may not be currently as prominent in historical-comparative sociology as in previous decades (Calhoun 1996; Steinmetz 2005); nevertheless, it is still the leading paradigm in general American sociology in general and therefore retains a certain appeal for comparative-historicists. The paradigm is characterized by a tendency to uphold empiricist ontological presuppositions, to assume time independence from social-theoretical concepts, to search for constant conjunction of events, and to privilege quantitative methods (Steinmetz 2005). However, while its application in mainstream sociology has produced a wealth of useful research, it is gravely taxed in the face of subject matters featuring event sequences that, like historical processes, are context dependent and do not betray repeated regularity, since here explanations through claims of constant conjunction become elusive. Likewise, this paradigm is not particularly useful if the analysis does not seek to reconstruct linear temporality or is unwilling to eschew complicated sequencing, conjuncture, and contingency (Abbott 2001b; Ragin 1987; Steinmetz 2005). In other words, the paradigm of covariance is ill suited for the analysis of processes, which means that it is ill suited for comparative-historical sociology as well as those less macroscopic areas of study that all the same aim at reconstructing emergencemost social movements scholars, for example, would think that their subject matter qualifies one way or the other. Yet those who study processes are often allured by this paradigms language. It is telling that McAdam and Tarrow slip into the mainstream discourse as they comment on Dynamics of Contention. Continuing the passage quoted above, they write, But the [social] ontological thrust of our program should have been clear: we were interested not in the strength of the correlations between variables but in how variables are linked to one another through causal mechanisms (McAdam and Tarrow 2011:4). The authors, then, wish to distance themselves from the hard science that purports to measure the strength of covariance but paradoxically fail to distance themselves from the label variable. This is so even though the book McAdam and Tarrow coauthored

Downloaded from stx.sagepub.com by guest on September 17, 2013

Demetriou

63

with Tilly keeps a safe distance from the labels of the covariance paradigm. Not only is the word variable nowhere to be seen in it, but the authors insist on comparing episodes rather than cases. More important still, their book avoids the paradigms logic, spirit, and tropes: It rejects John Stuart Mills Method of Agreement and Method of Difference when it comes to large comparisons and rejects the usefulness of searching for constant conjunction when it comes to less macroscopic comparisons, including those that may involve some microfoundations of behavior. Instead, in covering the range from macro to micro, it places its emphasis on dynamic patterns and emergence, contingency and conjuncture, and the sequencing and timing of eventsthat is, on the standard stuff of processes. The point is that it is precisely the terminology and logic of variable analysis that the process-mechanism realist approach aspires to overcome. In this approach, analysis through mechanisms does not connect variables nor unpack them but rather replaces thinking in terms of variables with thinking in terms of more fluid and complicated causality. And while it aspires to rigor, it is rigor not at the level of a model, where variables connect to each other computationally or logically but not necessarily realistically, but at the level of describing actual sequences of events. This last point is worth emphasizing because a developing discourse on mechanisms muddles it, drawing unwarranted connections among ontology, epistemology, and methodology. The discussion of mechanism has expanded in recent decades, and Tillys is but one of the available treatments. One only needs to read through the various existing definitions of what a mechanism is in order to hear the emergent polyphony (Gerring 2010; Gross 2009; Mahoney 2003). At the same time, many of the perspectives offered are by and large aligned with traditional epistemological and methodological positions. Being realist, Tillys approach differs most distinctly from analytical approaches to mechanisms. These do not belong to a coherent group. Some approaches, for example, combine the analysis of mechanisms with rational choice presuppositions (e.g., Kiser and Hechter 1991) or with other presuppositions of individualist behavior (e.g., Boudon 1998), while other approaches try to combine variable analysis with process tracing (e.g., George and Bennett 2005). In general, however, the analytic approach to mechanisms is a self-adopted label intended to show affinity with the effort to construct models of mechanisms, typically agent based, that are held to be more rigorous the more they are quantified (Hedstrom and Swedberg 1998; Hedstrom and Ylikoski 2010; Manzo 2010). So, whereas this approach does not necessarily adopt classical empiricism and in principle may remain open to a realist ontology, it does adopt positivism. Thus it takes on the standard perspective of quantitative sociology where attention is placed on the computational efficacy of a model rather than on access to real generative social dynamics through qualitative research and unquantifiable conceptualizations. However useful this approach may be for other areas of sociological research, it is ultimately ill suited for comparative-historical sociology. Considered from a point of view respectful of emergence through messiness, processual epistemology and methodology must remain modest. And indeed, the approach led by Tilly is modest, even though at times the author sets the expectation too high by referring to the work of robust mechanisms. His critics missed this modesty, expecting him to have reconstructed mechanism concatenations displaying productive continuity. But as this paper argued, this is an expectation tied to a pathway to research that Tilly did not take, one that would focus on the generative capacities of well-structured configurations. By contrast, the pathway to research started by down Tillythe one searching for mechanisms in wideranging social contextsexpects changing mechanism concatenations to be the norm.

Acknowledgments
I conducted research for this paper while holding a postdoctoral fellowship at Masaryk University, Brno, in 20092010. An early draft of the paper was presented at the Seminar on Political Mobilization, Institute for Comparative Political Research, Masaryk University, on April 28, 2010. I thank the participants of the seminar, and especially Ondrej Csar, for their comments. The paper benefited also from comments by two anonymous reviewers.

Downloaded from stx.sagepub.com by guest on September 17, 2013

64

Sociological Theory 30(1)

NOTES
1. McAdam, Tarrow, and Tilly (2001) primarily and McAdam, Tarrow and Tilly (2008) secondarily remain the two texts in which Tilly and his coauthors writes the most on the conceptualization of mechanisms and on related metatheory; even Tilly and Tarrow (2007), meant to be a text clarifying McAdam et al. (2001), operates on a less abstract level and is therefore less probing on epistemological and ontological questions. Accordingly, the present discussion relies heavily on McAdam et al. (2001). What subsequent writings do is elaborate substantively on specific mechanisms and processes, and my analysis consults them as it examines the empirical application of the concept of mechanism; for example, it refers to Tilly (2007), though not to Tilly (2008a), the authors last published book, since that book deliberately avoids the terminology developed in McAdam et al. (2001) (Tilly 2008a:xv). 2. Several schematizations of mechanism following nonrealist epistemological traditions consider only the arrow to be the mechanism. In such approaches, X and Y are held to be variables and the arrow/mechanism to be the specification that unpacks covariance (see, e.g., Gerring 2010). It must be underlined that the schematization implied by Tillys definition is distinct from such nonrealist schematizations.

REFERENCES
Abbott, Andrew. 2001a. Temporality and Process in Social Life. Pp. 20939 in Time Matters: On Theory and Method. Chicago: University of Chicago Press. Abbott, Andrew. 2001b. Transcending General Linear Reality. Pp. 3763 in Time Matters: On Theory and Method. Chicago: University of Chicago Press. Adam, Barbara. 1990. Time and Social Theory. Philadelphia: Temple University Press. Bhaskar, Roy. 1979. The Possibility of Naturalism. Sussex, England: Harvester. Bhaskar, Roy. 1986. Scientific Realism and Human Emancipation. London, England: Verso. Boudon, Raymond. 1998. Social Mechanisms without Black Boxes. Pp. 172203 in Social Mechanisms: An Analytical Approach to Social Theory, edited by P. Hedstrom and R. Swedberg. Cambridge, England: Cambridge University Press. Bunge, Mario. 1997. Mechanism and Explanation. Philosophy of the Social Sciences 27:41065. Bunge, Mario. 2004. How Does It Work? Philosophy of the Social Sciences 34:182210. Calhoun, Craig. 1996. The Rise and Domestication of Historical Sociology. Pp. 30538 in The Historic Turn in the Human Sciences, edited by T. J. McDonald. Ann Arbor: University of Michigan Press. Craver, Carl. 2006. When Mechanistic Models Explain. Syntheses 153:35576. Demetriou, Chares. 2009. The Realist Approach to Explanatory Mechanisms in Social Sciences: More Than a Heuristic? Philosophy of the Social Sciences 39:44062. Earl, Jennifer. 2008. An Admirable Call to Improve, But Not Fundamentally Change, Our Collective Methodological Practices. Qualitative Sociology 31:35559. George, Alexander L. and Andrew Bennett. 2005. Case Study and Theory Development in the Social Sciences. Cambridge, MA: MIT Press. Gerring, John. 2010. Causal Mechanisms: Yes, But . . . Comparative Political Studies 43:1499526. Gross, Neil. 2009. A Pragmatist Theory of Social Mechanisms. American Sociological Review 74:35879. Gross, Neil. 2010. Charles Tilly and American Pragmatism. American Sociologist 41:33757. Johnson, Victoria. 2007. What Is Organizational Imprinting? Cultural Entrepreneurship in the Founding of the Paris Opera. American Journal of Sociology 113:97127. Hedstrom, Peter and Richard Swedberg, eds. 1998. Social Mechanisms: An Analytical Approach to Social Theory. Cambridge, England: Cambridge University Press. Hedstrom, Peter and Petri Ylikoski. 2010. Causal Mechanisms in the Social Sciences. Annual Review of Sociology 36:4967. King, Gary, Robert O. Keohane, and Sidney Verba. 1994. Designing Social Inquiry: Scientific Inference in Qualitative Research. Princeton, NJ: Princeton University Press. Kiser, Edgar and Michael Hechter. 1991. The Role of General Theory in Comparative-Historical Sociology. American Journal of Sociology 97:130. Koopmans, Ruud. 2003. A Failed RevolutionBut a Worthy Cause. Mobilization 8:11619. Norkus, Zenonas. 2005. Mechanisms as Miracle Makers? The Rise and Inconsistencies of the Mechanism Approach in Social Science and History. History and Theory 44:34872. Machamer, Peter, Lindley Darden, and Carl Craver. 2000. Thinking about Mechanisms. Philosophy of Science 67:125. Mahoney, James. 2003. Tentative Answers to Questions about Causal Mechanisms. Paper presented at the Annual Meeting of the American Political Science Association, August 28, Philadelphia, PA.

Downloaded from stx.sagepub.com by guest on September 17, 2013

Demetriou

65

Manicas, Peter. 2006. A Realist Philosophy of Social Science: Explanation and Understanding. Cambridge, England: Cambridge University Press. Manzo, Gianluca. 2010. Analytical Sociology and Its Critics. European Journal of Sociology 51:12970. McAdam, Doug and Sidney Tarrow. 2011. Introduction: Dynamics of Contention Ten Years On. Mobilization 16:110. McAdam, Doug, Sidney Tarrow, and Charles Tilly. 2001. Dynamics of Contention. Cambridge, England: Cambridge University Press. McAdam, Doug, Sidney Tarrow, and Charles Tilly. 2008. Methods for Measuring Mechanisms of Contention. Qualitative Sociology 31:30731. Mingers, John. 2011. Explanatory Mechanisms: The Contribution of Critical Realism and Systems Thinking/ Cybernetics. Working Paper No. 241, Kent Business School, University of Kent, Canterbury, England. Oliver, Pamela. 2003. Mechanisms of Contention. Mobilization 8:12229. Ragin, Charles. 1987. The Comparative Method: Moving beyond Qualitative and Quantitative Strategies. Berkeley: University of California Press. Rescher, Nicholas. 1996. Process Metaphysics: An Introduction to Process Philosophy. Albany, NY: State University of New York Press. Rucht, Dieter. 2003. Overcoming the Classical Model. Mobilization 8:11216. Sartori, Giovanni. 1970. Concept Misformation in Comparative Politics. American Political Science Review 64:103353. Staggenborg, Suzzane. 2008. Seeing Mechanisms in Action. Qualitative Sociology 31:34144. Steinmetz, George. 2005. The Epistemological Unconscious of U.S. Sociology and the Transition to Post-Fordism: The Case of Historical Sociology. Pp. 10958 in Remaking Modernity: Politics, History, and Sociology, edited by J. Adams, E. S. Clemens, and A. Shola Orloff. Durham, NC: Duke University Press. Steinmetz, George. 2010. Charles Tilly, German Historicism, and the Critical Realist Philosophy of Science. American Sociologist 41:31236. Stinchcombe, Arthur. 1997. Tilly on the Past as a Sequence of Futures. Pp. 387409 in Roads from Past to Future by Charles Tilly. Lanham, MD: Rowman & Littlefield. Tarrow, Sidney. 2008. Charles Tilly and the Practice of Contentious Studies. Social Movement Studies 7:22546. Tilly, Charles. 1997. Parliamentarization of Popular Contention in Great Britain, 17581834. Pp. 21744 in Roads from Past to Future. Lanham, MD: Rowman & Littlefield. Tilly, Charles. 2001. Mechanisms in Political Processes. Annual Review of Political Science 4:2141. Tilly, Charles. 2002. Stories, Identities, and Political Change. Lanham, MD: Rowman & Littlefield. Tilly, Charles. 2003. The Politics of Collective Violence. Cambridge, England: Cambridge University Press. Tilly, Charles. 2007. Democracy. Cambridge, England: Cambridge University Press. Tilly, Charles. 2008a. Contentious Performances. Cambridge, England: Cambridge University Press. Tilly, Charles. 2008b. Explaining Social Processes. Boulder, CO: Paradigm Press. Tilly, Charles and Sidney Tarrow. 2007. Contentious Politics. Boulder, CO: Paradigm Press.

Bio
Chares Demetriou is a Marie Curie Fellow at the School of Politics, International Studies, and Philosophy, Queens University, Belfast. His interest in processual theory is expansive, covering not only sociological theory but also methodology, the philosophy of science, and metaphysics. Demetriou also works on substantive research, which includes two monograph projects: one on the historical sociology of Ireland and Cyprus in connection to the issue of political violence legitimization, and a second on the comparative study of processes of political radicalization.

Downloaded from stx.sagepub.com by guest on September 17, 2013

You might also like