Key premises of analytical sociology

Image: residential segregation by race, NYC 2010

In Dissecting the Social Peter Hedström describes the analytical sociology approach in these terms: 

Although the term analytical sociology is not commonly used, the type of sociology designated by the term has an important history that can be traced back to the works of late nineteenth- and early twentieth-century sociologists such as Max Weber and Alexis de Tocqueville, and to prominent mid-twentieth-century sociologists such as the early Talcott Parsons and Robert K. Merton. Among contemporary social scientists, four in particular have profoundly influenced the analytical approach. They are Jon Elster, Raymond Boudon, Thomas Schelling and James Coleman. (Dissecting the Social, kl 113) 

And here is how Hedström and Bearman describe the approach in their introduction to The Oxford Handbook of Analytical Sociology: 

Analytical sociology is concerned first and foremost with explaining important social facts such as network structures, patterns of residential segregation, typical beliefs, cultural tastes, common ways of acting, and so forth. It explains such facts not merely by relating them to other social facts -- an exercise that does not provide an explanation -- but by detailing in clear and precise ways the mechanisms through which the social facts under consideration are brought about. In short, analytical sociology is a strategy for understanding the social world. (Hedström and Bearman, eds. 2009 : 3-4) 

Peter Demeulenaere makes several important points to further specify AS in his extensive introduction to Analytical Sociology and Social Mechanisms. He holds that AS is not just another new paradigm for sociology. Instead, it is a reconstruction of what valid explanations on sociology must look like, once we properly understand the logic of the social world. He believes that much existing sociology conforms to this set of standards -- but not all. And the non-conformers are evidently judged non-explanatory. For example, he writes, “Analytical sociology should not therefore be seen as a manifesto for one particular way of doing sociology as compared with others, but as an effort to clarify (“analytically”) theoretical and epistemological principles which underlie any satisfactory way of doing sociology (and, in fact, any social science)” (Demeulenaere, ed. kl 121). So this sets a claim of a very high level of authority over the whole field, implying that other decisions about explanation, ontology, and method are less than fully scientific. 

Analytical sociology rests on three central ideas. 

First, there is the idea that social outcomes need to be explained on the basis of the actions of individuals. Hedstrom, Demeulenaere, and their colleagues refer to this position as methodological individualism. It is often illustrated by reference to "Coleman's Boat" in James Coleman, Foundations of Social Theory (Coleman, 1990, 8) describing the relationship that ought to exist between macro and micro social phenomena (link). The boat diagram indicates the relationship between macro-factors (Protestant religious doctrine, capitalism) and the micro factors that underlie their causal relation (values, economic behavior). Here are a few of Hedström's formulations of this ontological position: 

In sociological inquiries, however, the core entity always tends to be the actors in the social system being analyzed, and the core activity tends to be the actions of these actors. (Dissecting, kl 106) 

To be explanatory a theory must specify the set of causal mechanisms that are likely to have brought about the change, and this requires one to demonstrate how macro states at one point in time influence individuals' actions, and how these actions bring about new macro states at a later point in time. (Dissecting, kl 143) 

In other words: according to analytical sociologists, a good explanation of a given social outcome is a demonstration of how this outcome is the aggregate result of structured individual actions. In particular, an explanation should not make reference to meso or macro level factors. 

In his introduction to Analytical Sociology and Social Mechanisms Demeulenaere provides an analysis of the doctrine of methodological individualism and its current status. He believes that criticisms of MI have usually rested on a small number of misunderstandings which he attempts to resolve. For example, MI is not "atomistic", "egoistic", "non-social", or exclusively tied to rational choice theory. He prefers a refinement that he describes as structural individualism, but essentially he argues that MI is a universal requirement on social science. Demeulenaere specifically disputes the idea that MI implies a separation between society and non-social individuals. That said, Demeulenaere fully endorses the idea that AS depends upon and presupposes MI: “Does analytical sociology differ significantly from the initial project of MI? I do not really think so. But by introducing the notion of analytical sociology we are able to make a fresh start and avoid the various misunderstandings now commonly attached to MI” (Demeulenaere ed. kl 318). 

A theory based on the individual needs to have a theory of the actor. Hedström and others in the AS field are drawn to a broad version of rational-choice theory -- what Hedström calls the "Desire-Belief-Opportunity theory". This is a variant of rational choice theory, because the actor's choice is interpreted along these lines: given the desires the actor possesses, given the beliefs he/she has about the environment of choice, and given the opportunities he/she confronts, action A is a sensible way of satisfying the desires. (It is worth pointing out that it is possible to be microfoundationalist about macro outcomes while not assuming that individual actions are driven by rational calculations. Microfoundationalism is distinct from the assumption of individual rationality.) 

Second is the idea that social actors are socially situated; the values, perceptions, emotions, and modes of reasoning of the actor are influenced by social institutions, and their current behavior is constrained and incentivized by existing institutions. (This position has a lot in common with the methodological localism; link.) Practitioners of analytical sociology are not atomistic about social behavior, at least in the way that economists tend to be; they want to leave room conceptually for the observation that social structures and norms influence individual behavior and that individuals are not unadorned utility maximizers. In the Hedström-Bearman introduction to the Handbook they refer to their position as “structural individualism”: 

Structural individualism is a methodological doctrine according to which social facts should be explained as the intended or unintended outcomes of individuals’ actions. Structural individualism differs from traditional methodological individualism in attributing substantial explanatory importance to the social structures in which individuals are embedded. (Hedström and Bearman, 2009, 4). 

Demeulenaere explicates the term by referring to Homans’ distinction between individualistic sociology and structural sociology; the latter “is concerned with the effects these structures, once created and maintained, have on the behaviour of individuals or categories of individuals” (Demeulenaere, Analytical Sociology and Social Mechanisms, 2011, introduction, quoting Homans, 1984). So “structural individualism” seems to amount to this: the behavior and motivations of individuals are influenced by the social arrangements in which they find themselves. 

This is a direction of thought that is not well developed within analytical sociology, but would repay further research. There is no reason why a methodological-individualist approach should not take seriously the causal dynamics of identity formation and the formation of the individual's cognitive, practical, and emotional frameworks. These are relevant to behavior, and they are plainly driven by concrete social processes and institutions. 

Third, and most distinctive, is the idea that social explanations need to be grounded in hypotheses about the concrete social causal mechanisms that constitute the causal connection between one event and another. Mechanisms rather than regularities or necessary/sufficient conditions provide the fundamental grounding of causal relations and need to be at the center of causal research. This approach has several intellectual foundations, but one is the tradition of critical realism and some of the ideas developed by Roy Bhaskar (link). 

Here is Hedström's statement of the position:

The position taken here, rather, is that mechanism-based explanations are the most appropriate type of explanations for the social sciences. The core idea behind the mechanism approach is that we explain a social phenomenon by referring to a constellation of entities and activities, typically actors and their actions, that are linked to one another in such a way that they regularly bring about the type of phenomenon we seek to explain. (Dissecting, kl 65) 

A social mechanism, as defined here, is a constellation of entities and activities that are linked to one another in such a way that they regularly bring about a particular type of outcome. (kl 181) 

Demeulenaere also emphasizes that AS depends closely on the methodology of social causal mechanisms. The "analytical" part of the phrase involves identifying separate things, and the social mechanisms idea says how these things are related. Causal mechanisms are expected to be the components of the linkages between events or processes hypothesized to bear a causal relation to each other. And, more specifically to the AS approach, the mechanisms are supposed to occur at the level of the actors--not at the meso or macro levels. So this means that AS would not countenance a meso-level mechanism like this: "the organizational form of the supervision structure at the Bhopal chemical plant caused a high rate of maintenance lapses that caused the accidental release of chemicals." The organizational form is a meso-level factor, and it would appear that AS would require that its causal properties be unpacked onto individual actors' behavior. (I, on the other hand, will argue below that this is a perfectly legitimate social mechanism because we can readily supply its microfoundations at the behavioral level. So this suggests that we can legitimately refer to meso-level mechanisms as long as we are mindful of the microfoundations requirement. And this corresponds as well to the tangible fact that institutions have causal force with respect to individuals. Here is an earlier discussion; link.) 

In addition to these three orienting frameworks for analytical sociology, there is a fourth characteristic that should be mentioned. This is the idea that the tools of computer-based simulation of the aggregate consequences of individual behavior can be a very powerful tool for sociological research and explanation. So the tools of agent-based modeling and other simulations of complex systems have a very natural place within the armoire of analytical sociology.  These techniques offer tractable methods for aggregating the effects of lower-level features of social life onto higher-level outcomes. If we represent actors as possessing characteristics of action X, Y, Z, and we represent their relations as U, V, W -- how do these actors in social settings aggregate to mid- and higher-level social patterns? This is the key methodological challenge that sociologists like Gianluca Manzo have explored (Agent-based Models and Causal Inference), and it produces very interesting results. 

This brief summary of the central doctrines of AS provides one reason why AS theorists are so concerned to have adequate and tractable models of the actor -- often rational actor models. Thomas Schelling's work provides a particularly key example for the AS research community; in field after field he demonstrates how micro motives aggregate onto macro outcomes (Schelling, 1978, 1984). And Elster's work is also key, in that he provides some theoretical machinery for analyzing the actor at a "thicker" level -- imperfect rationality, self-deception, emotion, commitment, and impulse (Elster, Ulysses and the Sirens). 

In short, analytical sociology is a compact, clear approach to the problem of understanding social outcomes. It lays the ground for the productive body of research questions associated with the "aggregation dynamics" research program. There is active, innovative research being done within this framework of ideas, especially in Germany, Sweden, and Great Britain. And its clarity permits, in turn, the formulation of rather specific critiques from researchers in other sociological traditions who reject one or another of the key components. However, the framework of analytical sociology should not be mistaken for a general approach to all sociological research and explanation. It is well suited to some problems, and less so to others.

(Here is an earlier post summarizing Peter Demeulenaere's account of analytical sociology; link.)

The functionality of artifacts

We think of artifacts as being "functional" in a specific sense: their characteristics are well designed and adjusted for their "intended" use. Sometimes this is because of the explicit design process through which they were created, and sometimes it is the result of a long period of small adjustments by artisan-producers and users who recognize a potential improvement in shape, material, or internal workings that would lead to superior performance. Jon Elster described these processes in his groundbreaking 1983 book, Explaining Technical Change: A Case Study in the Philosophy of Science.

Here is how I described the gradual process of refinement of technical practice with respect to artisanal winegrowing in a 2009 post (link):

First, consider the social reality of a practice like wine-making. Pre-modern artisanal wine makers possess an ensemble of techniques through which they grow grapes and transform them into wine. These ensembles are complex and developed; different wine "traditions" handle the tasks of cultivation and fermentation differently, and the results are different as well (fine and ordinary burgundies, the sweet gewurztraminers of Alsace versus Germany). The novice artisan doesn't reinvent the art of winemaking; instead, he/she learns the techniques and traditions of the elders. But at the same time, the artisan wine maker may also introduce innovations into his/her practice -- a wrinkle in the cultivation techniques, a different timing in the fermentation process, the introduction of a novel ingredient into the mix.

Over time the art of grape cultivation and wine fermentation improves.

But in a way this expectation of "artifact functionality" is too simple and direct. In the development of a technology or technical practice there are multiple actors who are in a position to influence the development of the outcome, and they often have divergent interests. These differences of interests may lead to substantial differences in performance for the technology or technique. Technologies reflect social interests, and this is as evident in the history of technology as it is in the current world of high tech. In the winemaking case, for example, landlords may have interests that favor dense planting, whereas the wine maker may favor more sparse planting because of the superior taste this pattern creates in the grape. More generally, the owner's interest in sales and profitability exerts a pressure on the characteristics of the product that run contrary to the interest of the artisan-producer who gives primacy to the quality of the product, and both may have interests that are somewhat inconsistent with the broader social good.

Imagine the situation that would result if a grain harvesting machine were continually redesigned by the profit-seeking landowner and the agricultural workers. Innovations that are favorable to enhancing profits may be harmful for safety and welfare of agricultural workers, and vice versa. So we might imagine a see-saw of technological development, as the landowner and the worker gains more influence over the development of the technology.

As an undergraduate at the University of Illinois in the late 1960s I heard the radical political scientist Michael Parenti tell just such a story about his father's struggle to maintain artisanal quality in the Italian bread he baked in New York City in the 1950s. Here is an online version of the story (link). Michael Parenti's story begins like this:

Years ago, my father drove a delivery truck for the Italian bakery owned by his uncle Torino. When Zi Torino returned to Italy in 1956, my father took over the entire business. The bread he made was the same bread that had been made in Gravina, Italy, for generations. After a whole day standing, it was fresh as ever, the crust having grown hard and crisp while the inside remained soft, solid, and moist. People used to say that our bread was a meal in itself.... 

Pressure from low-cost commercial bread companies forced his father into more and more cost-saving adulteration of the bread. And the story ends badly ...

But no matter what he did, things became more difficult. Some of our old family customers complained about the change in the quality of the bread and began to drop their accounts. And a couple of the big stores decided it was more profitable to carry the commercial brands. 

Not long after, my father disbanded the bakery and went to work driving a cab for one of the big taxi fleets in New York City. In all the years that followed, he never mentioned the bread business again.

Parenti's message to activist students in the 1960s was stark: this is the logic of capitalism at work.

Of course market pressures do not always lead to the eventual ruin of the products we buy; there is also an economic incentive created by consumers who favor higher performance and more features that leads businesses to improve their products. So the dynamic that ruined Michael Parenti's father's bread is only one direction that market competition can take. The crucial point is this: there is nothing in the development of technology and technique that guarantees outcomes that are more and more favorable for the public.

Mechanisms, singular and general

Let's think again about the semantics of causal ascriptions. Suppose that we want to know what  caused a building crane to collapse during a windstorm. We might arrive at an account something like this:

• An unusually heavy gust of wind at 3:20 pm, in the presence of this crane's specific material and structural properties, with the occurrence of the operator's effort to adjust the crane's extension at 3:21 pm, brought about cascading failures of structural elements of the crane, leading to collapse at 3:25 pm.

The process described here proceeds from the "gust of wind striking the crane" through an account of the material and structural properties of the device, incorporating the untimely effort by the operator to readjust the device's extension, leading to a cascade from small failures to a large failure. And we can identify the features of causal necessity that were operative at the several links of the chain.

Notice that there are few causal regularities or necessary and constant conjunctions in this account. Wind does not usually bring about the collapse of cranes; if the operator's intervention had occurred a few minutes earlier or later, perhaps the failure would not have occurred; and small failures do not always lead to large failures. Nonetheless, in the circumstances described here there is causal necessity extending from the antecedent situation at 3:15 pm to the full catastrophic collapse at 3:25 pm.

Does this narrative identify a causal mechanism? Are we better off describing this as a sequences of cause-effect sequences, none of which represents a causal mechanism per se? Or, on the contrary, can we look at the whole sequence as a single causal mechanism -- though one that is never to be repeated? Does a causal mechanism need to be a recurring and robust chain of events, or can it be a highly unique and contingent chain?

Most mechanisms theorists insist on a degree of repeatability in the sequences that they describe as "mechanisms". A causal mechanism is the triggering pathway through which one event leads to the production of another event in a range of circumstances in an environment. Fundamentally a causal mechanism is a "molecule" of causal process which can recur in a range of different social settings.

For example:

• X typically brings about O.

Whenever this sequence of events occurs, in the appropriate timing, the outcome O is produced. This ensemble of events {X, O} is a single mechanism.

And here is the crucial point: to call this a mechanism requires that this sequence recurs in multiple instances across a range of background conditions.

This suggests an answer to the question about the collapsing crane: the sequence from gust to operator error to crane collapse is not a mechanism, but is rather a unique causal sequence. Each part of the sequence has a causal explanation available; each conveys a form of causal necessity in the circumstances. But the aggregation of these cause-effect connections falls short of constituting a causal mechanism because the circumstances in which it works are all but unique. A satisfactory causal explanation of the internal cause-effect pairs will refer to real repeatable mechanisms -- for example, "twisting a steel frame leads to a loss of support strength". But the concatenation does not add up to another, more complex, mechanism.

Contrast this with "stuck valve" accidents in nuclear power reactors. Valves control the flow of cooling fluids around the critical fuel. If the fuel is deprived of coolant it rapidly overheats and melts. A "stuck valve-loss of fluid-critical overheating" sequence is a recognized mechanism of nuclear meltdown, and has been observed in a range of nuclear-plant crises. It is therefore appropriate to describe this sequence as a genuine causal mechanism in the creation of a nuclear plant failure.

(Stuart Glennan takes up a similar question in "Singular and General Causal Relations: A Mechanist Perspective"; link.)

New thinking about causal mechanisms

Anyone interested in the topic of causal mechanisms will be interested in the appearance of Stuart Glennan and Phyllis Illari's The Routledge Handbook of Mechanisms and Mechanical Philosophy. Both Glennan and Illari have been significant contributors to the past fifteen years of discussion about the role of mechanisms in scientific explanation, and the Handbook is a highly interesting contribution to the state of the debate.

The book provides discussion of the role of mechanisms thinking in a wide range of scientific disciplines, from physics to biology to social science to engineering and cognitive science. It consists of four large sections: "Historical perspectives on mechanisms", "The nature of mechanisms", "Mechanisms and the philosophy of science", and "Disciplinary perspectives on mechanisms." Each section consists of contributions by talented experts on genuinely interesting topics.

A good introduction to the general topic of mechanisms is the introduction to the volume by Glennan and Illari, and more especially their article, "Varieties of mechanisms." They directly confront one of the large issues in the field, the wide dispersion of definitions and applications of the idea of a causal mechanism. They correctly observe that the concept of mechanism is used fairly differently in various areas of science and philosophy, but they argue that there is a common core of elements that underlie most or all of these usages. The variety that exists is the result of differences in the nature of the phenomena across different areas of scientific investigation, and differences in methodology in use in various sciences. They provide a rather general definition of a mechanism:

A mechanism for a phenomenon consists of entities (or parts) whose activities and interactions are organized so as to be responsible for the phenomenon. (92)

They then attempt to provide a basis for classifying different kinds of mechanisms according to several different criteria. The dimensions of variation they identify include the kind of phenomenon produced, the kind of entities and activities constituting the mechanism, the way in which entities and activities are organized, and the etiology of the mechanism.

Also interesting is Petri Ylikoski's contribution, "Social mechanisms." Ylikoski structures his exposition of the theory of social mechanisms around the Coleman boat diagram (link). To provide a mechanism for a social phenomenon is to provide an account at the level of the actors of how a macro-level event or entity causally brings about another macro-level event or entity. Ylikoski insists that this is a matter of explanatory adequacy rather than reductive analysis, and is therefore not ontologically reductionist. But it does fundamentally imply that social mechanisms occur at the level of interactions among actors. In prior posts I have argued against this presupposition (link). I argue that it is perfectly intelligible to suggest that there are meso-level causal mechanisms. Ylikoski also underlines the affinity that exists between social mechanisms and agent-based modeling: a good ABM demonstrates the process through which a set of conditions at the micro-level aggregate to a certain kind of macro-level outcome. See this earlier post for a small amount of doubt about the adequacy of ABM models to perform this kind of social aggregation for realistic social scenarios; link. (Several of these points are developed in my New Directions in the Philosophy of Social Science.)

Povich and Craver address the topic of the relationship that exists between mechanisms, levels, emergence, and reduction in their contribution, "Mechanistic levels, reduction, and emergence". This is a key question within the philosophy of social science. And the idea of  mechanism seems to have great relevance to the idea of various levels of phenomena. At the level of the organization we see, perhaps, chronic inefficiency in the use of certain kinds of resources. In searching for the mechanisms that cause this inefficiency we may choose to drop down a level and examine the incentives and constraints that guide the behavior of individuals within the organization. And we arrive at a theory of the individual-level mechanism that produces the meso-level outcome. This is a mechanism that falls along strut 3 of Coleman's boat; it is an aggregative mechanism. But not all social mechanisms have this nature. If we want to know why rebellious segments of an agrarian society locate themselves in remote, mountainous areas, it is enough to know a few meso-level facts about the functioning of traditional military forces and the meso-level fact that mountainous terrain gives a tactical advantage to rebel commanders. This appears to be a meso-level mechanism from start to finish.

A particularly intriguing and original contribution is Abrahamsen, Sheredos, and Bechtel's "Explaining visually using mechanism diagrams." We tend to think of scientific explanations as mathematical demonstrations or text-based derivations of outcomes. Abrahamsen, Sheredos, and Bechtel point out that visual diagrams play a crucial role in the presentation of many scientific results; and these diagrams are not merely heuristic or illustrative. A visual presentation serves to designate how the hypothesized mechanism works: what its parts are, how the parts influence each other, and how the functioning of the mechanism over time produces the outcome in question. The authors make an admirable attempt to provide a philosophy-of-science analysis of the components and logic of a visual diagram as an expository device for presenting a causal mechanism or process. They highlight the logical problems of representing entities, spatial location, and temporal duration within a diagram in a way that permits the viewer to gain an accurate understanding of the hypothesized mechanism or process. And they note that it is a conceptually simple step to introduce computational modeling into the graphical representation described here, so the processes in question can step through their interactions on-screen.

Taken together, the essays collected here constitute a valuable contribution to the literature on mechanisms and explanation. The handbook also gives the reader a concrete experience of how deeply varied the mechanisms literature is, leading to very interesting questions about cross-disciplinary communication. It appears to be genuinely challenging to formulate an abstract analysis of the idea of a causal mechanism that will mean approximately the same thing to researchers trained within significantly different research traditions. Unlike many handbooks, this collection warrants reading cover to cover. Researchers who believe that the mechanisms approach provides a valid way of understanding the metaphysics of causal inquiry and explanation will find every article stimulating and helpful.

(Here are a couple of prior posts on the challenge of providing a classification scheme for social mechanisms; link, link.)

Causal diagrams and causal mechanisms

There is a long history of the use of directed causal diagrams to represent hypotheses about causation. Can the mathematics and graphical systems created for statistical causal modeling be adapted to represent and evaluate hypotheses about causal mechanisms and outcomes?

In the causal modeling literature the structure of a causal hypothesis is something like this: variable T increases/ decreases the probability of the occurrence of outcome E. This is the causal relevance criterion described by Wesley Salmon in Scientific Explanation and the Causal Structure of the World. It is a fundamentally statistical understanding of causality.

Here is a classic causal path model by Blau and Duncan indicating the relationships among a number of causal factors in bringing about an outcome of interest -- "respondent's first job".

This construction aims at joining a qualitative hypothesis about the causal relations among a set of factors with a quantitative measurements of the correlations and conditional probabilities that support these causal relations. The whole construction often takes its origin in a multivariate regression model.

Aage Sørensen describes the underlying methodological premise of quantitative causal research in these terms in his contribution to Frontiers of Sociology (Annals of the International Institute of Sociology Vol. 11):

Understanding the association between observed variables is what most of us believe research is about. However, we rarely worry about the functional form of the relationship. The main reason is that we rarely worry about how we get from our ideas about how change is brought about, or the mechanisms of social processes, to empirical observation. In other words, sociologists rarely model mechanisms explicitly. In the few cases where they do model mechanisms, they are labeled mathematical sociologists, not a very large or important specialty in sociology. (370)

My question here is whether this scheme of representation of causal relationships and the graphical schemes that have developed around it are useful for the analytics of causal mechanisms.

The background metaphysics assumed in the causal modeling literature is Humean and "causal-factor" based; such-and-so factor increases the probability of occurrence of an outcome or an intermediate variable, the simultaneous occurrence of A and B increase the probability of the outcome, etc. Quoting Peter Hedstrom on causal modeling:

In the words of Lazarsfeld (1955: 124-5), "If we have a relationship between x and y; and if for any antecedent test factor the partial relationships between x and y do not disappear, then the original relationship should be called a causal one." (Dissecting the Social: On the Principles of Analytical Sociology)

The current iteration of causal modeling is a directed acyclic graph (DAG). Felix Elwert provides an accessible introduction to directed acyclic graphs in his contribution to Handbook of Causal Analysis for Social Research (link). Here is a short description provided by Elwert:

DAGs are visual representations of qualitative causal assumptions: They encode researchers’ expert knowledge and beliefs about how the world works. Simple rules then map these causal assumptions onto statements about probability distributions: They reveal the structure of associations and independencies that could be observed if the data were generated according to the causal assumptions encoded in the DAG. This translation between causal assumptions and observable associations underlies the two primary uses for DAGs. First, DAGs can be used to prove or disprove the identification of causal effects, that is, the possibility of computing causal effects from observable data. Since identification is always conditional on the validity of the assumed causal model, it is fortunate that the second main use of DAGs is to present those assumptions explicitly and reveal their testable implications, if any. (246)

A DAG can be interpreted as a non-parametric structural equation model, according to Elwert. (Non-parametric here means simply that we do not assume that the data are distributed normally.) Elwert credits the development of the logic of DAGs to Judea Pearl and Peter Spirtes, along with other researchers within the causal modeling community.

Johannes Textor and a team of researchers have implemented DAGitty, a platform for creating and using DAGs in appropriate fields, including especially epidemiology (link). A crucial feature of DAGitty is that it is not solely a graphical program for drawing graphs of possible causal relationships; rather, it embodies an underlying logic which generates expected statistical relationships among variables given the stipulated relationships on the graph. Here is a screenshot from the platform:

The question to consider here is whether there is a relationship between the methodology of causal mechanisms and the causal theory reflected in these causal diagrams. 

It is apparent that the underlying ontological assumptions associated with the two approaches are quite different. Causal mechanisms theory is generally associated with a realist approach to the social world, and generally rejects the Humean theory of causation. The causal diagram approach, by contrast, is premised on the Humean and statistical approach to causation.  A causal mechanisms hypothesis is not fundamentally evaluated in terms of the statistical relationships among a set of variables; whereas a standard causal model is wholly intertwined with the mathematics of conditional correlation.

Consider a few examples. Here is a complex graphical representation of a process understood in terms of causal mechanisms from McGinnes and Elandy, "Unintended Behavioural Consequences of Publishing Performance Data: Is More Always Better?" (link):

Plainly this model is impossible to evaluate statistically by attempting to measure each of the variables; instead, the researchers proceed by validating the individual mechanisms identified here as well as the direction of influence they have on other intermediate outcomes. The outcome of interest is "quality of learning" at the center of the graph; and the diagram attempts to represent the complex structure of causal influences that exist among several dozen mechanisms or causal factors.

Here is another example of a causal mechanisms path diagram, this time representing the causal system involved in drought and mental health by Vins, Bell, Saha, and Hess (link).

Here too the model is not offered as a statistical representation of covariance among variables; rather, it is a hypothetical sketch of the factors which play in mechanisms leading from drought to depression and anxiety in a population. And the assessment of the model should not take the form of a statistical evaluation (a non-parametric structural equation model), but rather a piecemeal verification of the validity of the specific mechanisms cited. (John Gerring argues that this is a major weakness in causal mechanisms theory, however, in "Causal Mechanisms? Yes, But ..." (link).)

It seems, therefore, that the superficial similarity between a causal model graph (a DAG) and a causal mechanisms diagram is only skin-deep. Fundamentally the two approaches make very different assumptions about both ontology (what a causal relationship is) and epistemology (how we should empirically evaluate a causal claim). So it seems unlikely that it will be fruitful for causal-mechanisms theorists to attempt to adapt methods like DAGs to represent the causal claims they want to advance and evaluate.

Microfoundations and mechanisms

The topics of microfoundations and causal mechanisms have come up frequently in this work. The microfoundations thesis maintains that social attributions and explanations based on macro-level entities and structures depend upon pathways at the level of the individual actors through which the entities and processes are maintained. The causal mechanisms thesis maintains that the best way of understanding causal assertions linking A to B is to identify the concrete causal mechanisms through which the powers of A bring about the properties of B.

Is there a relation between these two bodies of philosophical theory about the social world? There is, in a fairly obvious way. When we ask for the microfoundations of a hypothesized social process, we are really asking about the lower-level social mechanisms that bring the process about.

For example: What is it about an extended population that creates the observed features of the spread of rumor or panic? Or in other words, what are the social mechanisms through which socially interacting actors spread rumors or contribute to a broader occurrence of panic and fear? When we provide an account of the ways in which individuals communicate with each other and then demonstrate how messages diffuse through the given network structure, we have identified one of the social mechanisms of the social process in question.

Asking for the microfoundations of X is asking for an answer to two related questions: What is X (at the micro level)? And how does X work (also at the micro level)? The latter question can be paraphrased as: what are the sub-level mechanisms through which the X-level processes work? The first question is not so clearly a question about mechanisms; it is rather a question about composition. What is it about the substrate that gives rise to (constitutes) the observed macro-level properties of X? But in their book In Search of Mechanisms: Discoveries across the Life Sciences Craver and Darden argue that mechanisms play both roles. Mechanisms can be invoked to account for both process and structure (link). Here is their diagram illustrating the role that mechanisms can play with respect to higher-level structures and processes:

So here is a preliminary answer to the question of whether microfoundations and mechanisms are related. In the most immediate sense, we might say that the search for microfoundations is a search for a group of lower-level social mechanisms, to account for both the constitution and the causal dynamics of the higher-level structure. Searching for microfoundations involves learning more about the substrate of a given level of structure and process, and the causal mechanisms that occur at that lower level. Microfoundations is the question and mechanisms is the answer.

This response is not fully satisfactory, however, for several reasons.

First, there is an implication in this analysis that mechanisms live at the substrate level -- in the case of the social world, at the level of individual social actors. This is clearly assumed in the analytical sociology literature (Hedstrom, Dissecting the Social: On the Principles of Analytical Sociology). But this is an unnecessary and narrow stipulation about causal mechanisms. It is plausible to maintain that there are causal mechanisms at a range of levels (link); for example, at the cognitive level, the motivational level, the organizational level, or the system level (link).

Second, we might also observe that various social mechanisms themselves possess microfoundations. There are processes in the causal substrate that constitute the causal necessity of a specified mechanism. A spark in the presence of methane and oxygen brings about an explosion. This is a mechanism of combustion. The substrate is the chemical composition of methane and oxygen and the chemical processes that occur when an electrical spark is introduced into the environment. So the question of "level" is a relative one. A given set of objects and causal processes has its own substrate at a lower level, and simultaneously may serve as the substrate for objects and processes at higher levels.

We might also consider the idea that the two concepts have a different grammar. They play different roles in the language of science. The microfoundations conceptual scheme immediately invokes the idea of level and substrate. It brings along with it an ontological principle (the higher level is constituted by the properties of the substrate), and a partial methodological principle (the generative strategy of showing how higher-level processes come about as a consequence of the workings of the substrate). The mechanisms conceptual scheme does not inherently presuppose higher-level and lower-level structures; instead, a mechanism is something like a unit of causation, and it may be found at any level from molecular biology to organizational change.

(In an earlier post I considered a similar question, the relation between powers and mechanisms. There I argued that these two concepts are symmetrical: mechanisms lead us to powers, and powers lead us to mechanisms.)

Social mechanisms and ABM methods

One particularly appealing aspect of agent-based models is the role they can play in demonstrating the inner workings of a major class of social mechanisms, the group we might refer to as mechanisms of aggregation. An ABM is designed to work out how a field of actors of a certain description, in specified kinds of interaction, lead through time to a certain kind of aggregate effect. This class of mechanisms corresponds to the upward strut of Coleman's boat. This is certainly a causal story; it is a generative answer to the question, how does it work?

However, anyone who thinks carefully about causation will realize that there are causal sequences that occur only once. Consider this scenario: X occurs, conditions Ci take place in a chronological sequence, and Y is the result. So X caused Y through the causal steps instigated by Ci. We wouldn't want to say the complex of interactions and causal links associated with the progress of the system through Ci as a mechanism linking X to Y; rather, this ensemble is the particular (in this case unique) causal pathway from X to Y. But when we think about mechanisms, we generally have in mind the idea of "recurring causal linkages", not simply a true story about how X caused Y in these particular circumstances. In other words, for a causal story to represent a mechanism, it needs to be a story that can be found to hold in an indefinite number of cases. Mechanisms are recurring complexes of causal sequences.

An agent-based model serves to demonstrate how a set of actors give rise to a certain aggregate outcome. This is plainly a species of causal argument. But it is possible to apply ABM methods to circumstances that are unique and singular. This kind of ABM model lacks an important feature generally included in the definition of a mechanism-- the idea of recurrence across a number of cases. So we might single out for special attention those ABMs that identify and analyze processes that recur across multiple social settings. Here we might refer, for example, to the "Schelling mechanism" of residential segregation. There are certainly other unrelated mechanisms associated with urban segregation -- mortgage lending practices or real estate steering practices, for example. But the Schelling mechanism is one contributing factor in a range of empirical and historical cases. And it is a factor that works through the local preferences of individual actors.

So this seems to answer one important question: in what ways can ABM simulations be said to describe social mechanisms? They do so when (i) they describe an aggregative process through which a given meso-level outcome arises, and (ii) the sequence they describe can be said to recur in multiple instances of social process.

A question that naturally arises here is whether there are social mechanisms that fall outside this group. Are there social mechanisms that could not be represented by an ABM model? Or would we want to say that mechanisms are necessarily aggregative, so all mechanisms should be amenable to representation by an ABM?

This is a complicated question. One possible response seems easily refuted: there are mechanisms that work from meso level (organizations) to macro level (rise of fascism) that do not invoke the features of individual actors. Therefore there are mechanisms that do not conform strictly to the requirements of methodological individualism. However, there is nothing in the ABM methodology that requires that the actors should be biological individuals. Certainly it is possible to design an ABM representing the results of competition among firms with different behavioral characteristics. This example still involves an aggregative construction, a generation of the macro behavior on the basis of careful specification of the behavioral characteristics of the units.

Another possible candidate for mechanisms not amenable to ABM analysis might include the use of network analysis to incorporate knowledge-diffusion characteristics into analysis of civil unrest and other kinds of social change. It is sometimes argued that there are structural features of networks that are independent of actor characteristics and choices. But given that ABM theorists often incorporate aspects of network theory into their formal representations of a social process, it is hard to maintain that facts about networks cannot be incorporated into ABM methods.

Another candidate is what Chuck Tilly and pragmatist sociologists (Gross, Abbott, Joas) refer to as the "relational characteristics" of a social situation. Abbott puts the point this way: often a social outcome isn't the result of an ensemble of individuals making discrete choices, but rather is a dance of interaction in which each individual's moves both inform and self-inform later stages of the interaction. This line of thought seems effective as a rebuttal to methodological individualism, or perhaps even analytical sociology, but I don't think it demonstrates a limitation of the applicability of ABM modeling. ABM methods are agnostic about the nature of the actors and their interactions. So it is fully possible for an ABM theorist to attempt to produce a representation of the iterative process just described; or to begin the analysis with an abstraction of the resultant behavioral characteristics found in the group.

I've argued here that it is legitimate to postulate meso-to-meso causal mechanisms. Meso-level things can have causal powers that allow them to play a role in causal stories about social outcomes. I continue to believe that is so. But considerations brought forward here make me think that even in cases where a theorist singles out a meso-meso causal mechanism, he or she is still offering some variety of disaggregative analysis of the item to be explained. It seems that providing a mechanism is always a process of delving below the level of the explananda to uncover the underlying processes and causal powers that bring it about.

So the considerations raised here seem to lead to a strong conclusion -- that all social mechanisms can be represented within the framework of an ABM (stipulating that ABM methods are agnostic about the kinds of agents they postulate). Agent-based models are to social processes as molecular biology is to the workings of the cell.

In fact, we might say that ABM methods simply provide a syntax for constructing social explanations: to explain a phenomenon, identify some of the constituents of the phenomenon, arrive at specifications of the properties of those constituents, and demonstrate how the behavior of the constituents aggregates to the phenomenon in question.

(It needs to be recognized that identifying agent-based social mechanisms isn't the sole use of ABM models, of course. Other uses include prediction of the future behavior of a complex system, "what if" experimentation, and data-informed explanations of complex social outcomes. But these methods certainly constitute a particularly clear and rigorous way of specifying the mechanism that underlies some kinds of social processes.)

Varieties of social methodology

What are the frameworks that generally come to mind in discussions of methodology in the social sciences? Several families of methodological frameworks are indicated in the diagram above. These are deliberately presented as a wheel, with no sense of priority among them.

(A) Quantitative methodology -- what Andrew Abbott refers to as the variables paradigm. This is the approach that analyzes the social world as a set of individuals, groups, and properties, and simply sorts through to find correlations, associations, and possible causal relationships using a range of statistical tools. This is an inductivist approach. In this approach, the role of the social sciences is to accurately observe the facts and to build up systems of regularities among them. This seems like an assumption-free framework, but there is an underlying ontology here no less than the other frameworks mentioned here -- the idea that the the social world is governed by some system of underlying laws or regularities.

(B) Interpretive methods. Clifford Geertz recommends an approach to social research within a generally interpretive worldview. He maintains that the most important feature of the social world is the fact of meaning. He urges us to consider as most important the meanings individuals and groups attach to behaviors and performances. So the research task is to reconstruct those meanings by observing and interacting with the social actors in a particular setting. The observer should observe the patterns of action and interaction he/she finds and carefully investigate the patterns of meaning the participants weave around their worlds.

A related framework is ethnomethodology, the approach taken by qualitative sociologists like Goffman and Garfinkel. This is the idea that one important function of the social sciences is to figure out the underlying grammar of the assumptions and rules that individuals are following as they interact with each other. The ontological assumption here is that individuals are basic in the social world, and individuals are complex. On this approach the methodology is to observe ordinary behavior and try to discover the underlying rules and expectations that indicate something like a grammar or normative frame that drives or generates interpersonal behavior.

(C) A family of approaches we might call realist methodology. These approaches begin with the premise that the social world consists of certain kinds of entities, forces, and processes, and then guides the researcher to attempt to discover the characteristics of those structures. This is a process of hypothesis formation and theory development and of testing out theories -- large or small -- of things like class, charisma, or bureaucratic state apparatus.

(D) A number of methods of analysis developed in the comparative social sciences, including causal methods associated with comparative historical sociology. That includes the methods of paired comparisons, Mill's methods, and methods of similarity and difference.  The researcher attempts to work out which factors are necessary or sufficient, enhancing or inhibiting. We can call this comparative methodology.

(E) Causal mechanisms methodology. This framework is a variety of realist methodology. On this approach we work on the assumption that there are social causes and that causes take the form of concrete causal mechanisms. The task of research is to gain enough empirical detail about selected cases to be able to piece together assumptions about the mechanisms at work. Ideas associated with the notion of process tracing have a natural fit here.

(F) Methods emphasizing techniques of formal modeling. This methodology is especially prominent in political science and economics. Here the goal of the research is to arrive at elegant, simple mathematical models of the phenomena. On this approach the evaluation of the model is not so much empirical but rather mathematical and formal. This approach is commonly faulted exactly because it is not sufficiently responsive to empirical constraints do standards. Its empirical relevance is not so clear. If we believe the social sciences are empirical then a formal model that is a valued for its abstract elegance is unsatisfactory.   It needs to contribute to an understanding of real empirical phenomena. At the least this means that we should be able to tie the model to some real behavioral characteristics. In the best case -- for example, with ABMs or CGE models-- we should be able to begin to reproduce important features of real empirical cases by calibrating the model to empirical circumstances.

(G) There are two other aspects of methodology that need to be called out. One has to do with the methods of data collection which are recommended, which differ substantially from domain to domain. The other is methods of empirical evaluation of the theories we advance. Social sciences differ substantially in both these ways -- what kind of data is needed, how it should be collected (e.g. survey methodology), and how we should validate the results. These are often discipline-specific and substantially more concrete than prescriptions in the philosophy of science. The ways in which we should evaluate a social science construction also varies significantly by national research tradition. Gabriel Abend points out that schemes of evaluation vary substantially across the sociological traditions of mexico and the United States in terms of the standards in play about what constitutes rigor,empirical argument and theoretical argument."

Each of these is a methodology in the loose sense I favor.  It is a guide for the researcher, indicating what kinds of factors he or she should be looking for by postulating a social ontology; an indication of what an explanatory account ought to look like, and an indication of what counts as warrant for such explanations.

We might observe that if we favor a pluralist social ontology, according to which there are properties individuals, relationships, structures, networks, meanings, and values, then the method we use to acquire knowledge about these things should be pluralistic as well. Our methods should allow us to pose research questions about all these kinds of things.

Heuristics for a mechanisms-based methodology

Let’s imagine that I’m a young sociologist or political scientist who has gotten interested in the social-mechanisms debates, and I’d like to frame my next research project around a set of heuristics that are suggested by the mechanisms approach. What might some of those heuristics look like? What is a "mechanisms-based methodology" for sociological research? And how would my research play out in concrete terms? Here are a few heuristics we might consider.

1. Identify one or more clear cases of the phenomenon I’m interested in understanding
2. Gain enough empirical detail about the cases to permit close examination of possible causal linkages
3. Acquaint myself with a broad range of social mechanisms from a range of the social sciences (political science, economics, anthropology, public choice theory, critical race studies, women’s studies, …)
4. Attempt to segment the phenomena into manageable components that may admit of separate study and analysis
5. Use the method of process-tracing to attempt to establish what appear to be causal linkages among the phenomena
6. Use my imagination and puzzle-solving ability to attempt to fit one or more of the available mechanisms into the phenomena I observe
7. Engage in quasi-experimental reasoning to probe the resulting analysis: if mechanism M is involved, what other effects would we expect to be present as well? Do the empirical realities of the case fit these hypothetical expectations?

These heuristics represent in a rough-and-ready way the idea that there are some well understood social processes in the world that have been explored in a lot of empirical and theoretical detail. The social sciences collectively provide a very rich toolbox of mechanisms that researchers have investigated and validated. We know how these mechanisms work, and we can observe them in a range of settings. This is a realist observation: the social world is not infinitely variable, and there is a substrate of activity, action, and interaction whose workings give rise to a number of well understood mechanisms. Here I would include free rider problems, contagion, provocation, escalation, coercion, and log-rolling as a very miscellaneous set of exemplars. So if we choose to pursue a mechanisms-based methodology, we are basically following a very basic intuition of realism by asking the question, "how does this social phenomenon work in the settings in which we find it?".

So how might a research project unfold if we adopt heuristics like these? Here is a striking example of a mechanisms approach within new-institutionalist research, Jean Ensminger's account of bridewealth in the cattle-herding culture of Kenya (Making a Market: The Institutional Transformation of an African Society). First, some background. The cattle-herding economic regime of the Orma pastoralists of Kenya underwent substantial changes in the 1970s and 1980s. Commons grazing practices began to give way to restricted pasturage; wage labor among herders came to replace familial and patron-client relations; and a whole series of changes in the property system surrounding the cattle economy transpired as well. This is an excellent example for empirical study from a new-institutionalist perspective. What explained the particular configuration of norms and institutions of the earlier period? And what social pressures led to the transition towards a more impersonal relationship between owners and herders? These are questions about social causation at multiple levels.

Ensminger examines these questions from the perspective of the new institutionalism. Building on the theoretical frameworks of Douglass North and others, she undertakes to provide an analysis of the workings of traditional Orma cattle-management practices and an explanation of the process of change and dissolution that these practices underwent in the decades following 1960. The book puts forward a combination of close ethnographic detail and sophisticated use of theoretical ideas to explain complex local phenomena.

How does the new institutionalism approach help to explain the features of the traditional Orma cattle regime identified by Ensminger’s study? The key institutions in the earlier period are the terms of employment of cattle herders in mobile cattle camps. The traditional employment practice takes the pattern of an embroidered patron-client relation. The cattle owner provides a basic wage contract to the herder (food, clothing, and one head of cattle per year). The good herder is treated paternally, with additional “gifts” at the end of the season (additional clothing, an additional animal, and payment of the herder’s bridewealth after years of service). The relation between patron and client is multi-stranded, enduring, and paternal.

Ensminger understands this traditional practice as a solution to an obvious problem associated with mobile cattle camps, which is fundamentally a principal-agent problem. Supervision costs are very high, since the owner does not travel with the camp. The owner must depend on the herder to use his skill and diligence in a variety of difficult circumstances—rescuing stranded cattle, searching out lost animals, and maintaining control of the herd during harsh conditions. There are obvious short-term incentives and opportunities for the herder to cheat the employer—e.g. allowing stranded animals to perish, giving up on searches for lost animals, or even selling animals during times of distress. The patron-client relation is one possible solution to this principal-agent problem. An embedded patron-client relation gives the herder a long-term incentive to provide high-quality labor, for the quality of work can be assessed at the end of the season by assessment of the health and size of the herd. The patron has an incentive to cheat the client—e.g. by refusing to pay the herder’s bridewealth after years of service. But here the patron’s interest in reputation comes into play: a cattle owner with a reputation for cheating his clients will find it difficult to recruit high-quality herders.

This account serves to explain the evolution and persistence of the patron-client relation in cattle-camps on the basis of transaction costs (costs of supervision). Arrangements will be selected that serve to minimize transaction costs. In the circumstances of traditional cattle-rearing among the Orma the transaction costs of a straight wage-labor system are substantially greater than those associated with a patron-client system. Therefore the patron-client system is selected.

This analysis identifies mechanisms at two levels. First, the patron-client relation is the mechanism through which the endemic principal-agent problem facing cattle owners is solved. The normal workings of this relation give both patron and client a set of incentives that leads to a stable labor relation. The higher-level mechanism is somewhat less explicit, but is needed for the explanation to fully satisfy us. This is the mechanism through which the new social relationship (patron-client interdependency) is introduced and sustained. It may be the result of conscious institutional design or it may be a random variation in social space that is emulated when owners and herders notice the advantages it brings. Towards the end of the account we are led to inquire about another higher-level mechanism, the processes through which the traditional arrangement is eroded and replaced by short-term labor contracts.

This framework also illustrates the seventh heuristic above, the use of counterfactual reasoning. This account would suggest that if transaction costs change substantially (through improved transportation, for example, or through the creation of fixed grazing areas), that the terms of employment would change as well (in the direction of less costly pure wage-labor contracts). And in fact this is what Ensminger finds among the Orma. When villages begin to establish “restricted grazing areas” in the environs of the village, it is feasible for cattle owners to directly supervise the management of their herds; and in these circumstances Ensminger finds an increase in pure wage labor contracts.

What are the scientific achievements of this account? There are several. First, it takes a complicated and detailed case of collective behavior and it makes sense of the case. It illuminates the factors that influence choices by the various participants. Second, it provides insight into how these social transactions work (the mechanisms that are embodied in the story). Third, it begins to answer -- or at least to pose in a compelling way -- the question of the driving forces in institutional change. This too is a causal mechanism question; it is a question that focuses our attention on the concrete social processes that push one set of social behaviors and norms in the direction of another set of behaviors and norms. Finally, it is an empirically grounded account that gives us a basis for a degree of rational confidence in the findings. The case has the features that we should expect it to have if the mechanisms and processes in fact worked as they are described to do.

A final achievement of this account is very helpful in the context of our efforts to arrive at explanations of features of the social world. This is the fact that the account is logically independent of an effort to arrive at strong generalizations about behavior everywhere. The account that Ensminger provides is contextualized and specific, and it does not depend on the assumption that similar social problems will be solved in the same way in other contexts. There is no underlying assumption that this interesting set of institutional facts should be derivable from a general theory of behavior and institutions. Instead, the explanation is carefully crafted to identify the specific (and perhaps unique) features of the historical setting in which the phenomenon is observed.

(Here is a nice short article by David Collier on the logic of process-tracing; link. And here is an interesting piece by Aussems, Boomsma, and Snijders on the use of quasi-experimental methods in the social sciences; link.)

Realism and methodology

Methodology has to do with the strategies and heuristics through which we attempt to understand a complicated empirical reality (link). Our methodological assumptions guide us in the ways in which we attempt to collect data, the kinds of data we collect, the explanatory hypotheses we bring forward for that range of empirical findings, and the ways we seek to validate our findings. Methodology is to the philosophy of social science as historiography is to the philosophy of history.

Realism is also a set of assumptions that we bring to empirical investigation. But in this case the assumptions are about ontology -- how the world works, in the most general ways. Realism asserts that there are real underlying causes, structures, processes, and entities that give rise to the observations we make of the world, natural and social. And it postulates that it is scientifically appropriate to form theories and hypotheses about these underlying causes in order to arrive at explanations of what we observe.

This description of realism is couched in terms of a distinction between what is observable and what is unobservable but nonetheless real -- the "observation-theoretic" distinction. But of course the dividing line between the two categories shifts over time. What was once hypothetical becomes observable. Extra-solar planetary bodies, bosons, and viruses were once unobservable; they are now observable using various forms of scientific instrumentation and measurement. So the distinction is not fundamental; this was an essential part of the argument against positivist philosophy of science. And we might say the same about many social entities and structures as well. We understand "ideology" much better today than when Marx theorized about this idea in the mid-19th century, and using a variety of social research methods (public opinion surveys, World Values Survey, ethnographic observation, structured interviews) we can identify and track shifts in the ideology of a group over time. We can observe and track ideologies in a population. (We may now use a different vocabulary -- mentality, belief framework, political values.)

There are several realist methodologies that are possible in the social sciences. The methodology of paired comparisons is a common part of research strategies in the historical social sciences. This is often referred to as "small-N research." (Here is a description of the method as practiced by Sid Tarrow; link, link.) The method of paired comparisons is also based on realism and derives from causal ideas; but it is not specifically derived from the idea of causal mechanisms.  Rather, it derives from the simpler notion that causal factors function as something like necessary and/or sufficient conditions for outcomes. So if we can find cases that differ in outcome and embody only a small number of potential contributing causal factors, we can use Mill's methods (or more general truth-table methods) to sort out the causal roles played by the factors. (Here is a discussion of some of these concepts; link.) These ideas contribute to methodology at two levels: they give the investigator a specific idea about how to lay out his/her research ("seek out relevantly similar cases with different outcomes"), and they embody a method of inference from findings to conclusions about causal relations (the truth-table method). These methods allow the researcher to arrive at statements about which factors play a role in the production of other factors. (This is a logically similar role to the use of multiple regression in quantitative studies.)

Another possible realist approach to methodology is causal mechanisms theory (CM). It rests on the idea that events and outcomes are caused by specific happenings and powers, and it proposes that a good approach to a scientific explanation of an outcome or pattern is to discover the real mechanisms that typically bring it about. It also brings forward an old idea about causation -- no action at a distance. So if we want to maintain that class privilege causes ideological commitment, we need to be able to tell an empirically grounded story about how the first kind of thing conveys its influence to changes in the second kind of thing. (This is essentially the call for microfoundations; link.) Causal mechanisms theory is more basic than either paired comparisons or statistical causal modeling, in that it provides a further explanation for findings produced by either of these other methods. Once we have a conception of the mechanisms involved in a given social process, we are in a position to interpret a statistical finding as well as a finding about the necessary and/or sufficient conditions provided by a list of antecedent conditions for an outcome.

It is an interesting question to consider whether realism in ontology leads to important differences in methodology. In particular, does the idea that things happen as the result of an ensemble of real causal mechanisms that can be separately understood lead to important new ideas about methodology and inquiry?

Craver and Darden argue in In Search of Mechanisms: Discoveries across the Life Sciences that mechanisms theory does in fact contribute substantially to contemporary research in biology, at a full range of levels (link). They maintain that the key goal for much research in contemporary biology is to discover the mechanisms that produce an outcome, and that a central component of this methodology is the effort to explain a given phenomenon by trying to fit one or more known mechanisms to the observed process. So working with a toolbox of known mechanisms and "problem-solving" to account for the new phenomenon is an important heuristic in biology. This approach is both ontological and methodological; it presupposes that there are real underlying mechanisms, and it recommends to the researcher that he/she be well acquainted with the current inventory of known mechanisms that may be applied to new settings.

I think there is a strong counterpart to this idea in a lot of sociological research as well. There are well understood social mechanisms that sociologists, political scientists, and other researchers have documented -- easy riders, prisoners dilemmas, conditional altruism -- and the researcher often can systematically explore whether one or more of the known mechanisms is contributing to the complex social outcomes he or she is concerned with. A good example is found in Howard Kimeldorf's Reds or Rackets?: The Making of Radical and Conservative Unions on the Waterfront. Kimeldorf compares two detailed case histories and strives to identify the concrete social mechanisms that led to different outcomes in the two cases. The mechanisms are familiar from other sociological research; Kimeldorf's work serves to show how specific mechanisms were in play in the cases he considers.

This kind of work can be described as problem-solving heuristics based on application of a known inventory of mechanisms. It could also be described as a "normal science" process where small theories of known processes are redeployed to explain novel outcomes. As Kuhn maintains, normal science is incremental but creative and necessary in the progress of science.

A somewhat more open-ended kind of inquiry is aimed at discovery of novel mechanisms. McAdam, Tarrow and Tilly sometimes engage in this second kind of discovery in Dynamics of Contention -- for example, the mechanism of social disintegration (kl 3050). Another good example of discovery of mechanisms is Akerlof's exposition of the "market for lemons" (link), where he lays out the behavioral consequences of market behavior with asymmetric knowledge between buyer and seller.

So we might say that mechanisms theory gives rise to two different kinds of research methodology -- application of the known inventory to novel cases and search for novel mechanisms (based on theory or empirical research).

Causal-mechanisms theory also suggests a different approach to data gathering and a different mode of reasoning from both quantitative and comparative methods. The approach is the case-studies method: identify a small set of cases and gain enough knowledge about how they played out to be in a position to form hypotheses about the specific causal linkages that occurred (mechanisms).

This approach is less interested in finding high-level generalizations and more concerned about the discovery of the real inner workings of various phenomena. Causal mechanisms methodology can be applied to single cases (the Russian Revolution, the occurrence of the Great Leap Forward famine), without the claim to offering a general causal account of famines or revolutions. So causal mechanisms method (and ontology) pushes downward the focus of research, from the macro level to the more granular level.

The inference and validation component associated with CM looks like a combination of piecemeal verification (link) and formal modeling (link). The case-studies approach permits the researcher to probe the available evidence to validate specific hypotheses about the mechanisms that were present in the historical case. The researcher is also able to try to create a simulation of the social situation under study, confirm as much of the causal internal connectedness as possible from study of the case, and examine whether the model conforms in important respects to the observed outcomes. Agent-based models represent one such set of modeling techniques; but there are others.

So the methodological ideas associated with CM theory differ from both small-N and large-N research. The search for causal mechanisms is largely agnostic about high-level regularities -- either of things like revolutions or things like metals. It is an approach that encourages a more specific focus on this case or that small handful of cases, rather than a focus on finding general causal properties of high-level entities. And it is more open to and tolerant of the possibility of a degree of contingency and variation within a domain of phenomena. To postulate that civil disorders are affected by a group of well-understood social mechanisms does not imply that there are strong regularities across all civil disorders, or that these mechanisms work in exactly the same way in all circumstances. So the features of contingency and context dependence play an organic role within CM methodology and fit badly in paired-comparisons research and statistical modeling approaches.

So it seems that the ontology of causal-mechanisms theory does in fact provide a set of heuristics and procedures for undertaking social research. CM does have implications for social-science methodology.

Entropic social mechanisms

Many of the examples of mechanisms that we turn to in the social sciences are purposive, agental, and designed. But there is a fundamental feature of the natural world that seems to have relevance to the social world as well that is distinctly non-purposive -- the workings of entropy. The second law of thermodynamics holds that the overall entropy (disorder) of the universe increases, and it requires an input of energy to maintain local structure against disorder. The discovery of Brownian motion was the impetus to this fundamental insight into the natural world: random, stochastic forces constantly interact with all levels of physical systems, leading to unpredictable disturbances and gradual decay of orderly structures (link).

This basic fact about the natural world seems applicable to the social world as well. In place of the heat-induced motions of particles in a solution we have the fact of multitudes of individuals choosing to act in a variety of ways, impinging on the social structures and rules that surround them. These "bumps" lead to local changes, and sometimes these changes accumulate to a process of drift in the structures upon which they impinge. A small group of racists begin demonstrating their beliefs in a small Kansas town, and somehow they manage to disrupt the prior racial harmony. This is an example of path dependency. And it is an example of how small random events can have large outcomes.

So are there features of social process that we might refer to as entropic mechanisms?

It would seem that there are. Take the idea of "the fog of war." The basic idea here is that generals like to think of the conduct of war as a purposive, intelligent marshaling of forces to secure clear goals against the adversary. But those who highlight the fog of war emphasize two fundamental facts: it is difficult to collect information during war, and it is difficult to mount coherent focused action in these circumstances. Warfare is a complex activity involving hundreds of leaders, thousands of combatants, scores of unforeseen circumstances, and a practical inability to gather accurate information rapidly enough to control one's forces effectively. The fog of war impedes control in both directions. It makes intelligence gathering difficult, but it also makes the direction of force and tactics difficult as well. By the time French generals in the Franco-Prussian War realized they needed to concentrate forces in Sedan, the disorder in the rail system made it impossible to do so (link).

Or take another basic idea of thermodynamics, the fact of friction. Friction is the interaction between an object and its environment that causes it to lose energy, momentum, and direction. The hockey puck on ice follows the course predicted by classical mechanics from stick to goal. But the same puck when slapped on asphalt or grass pursues a dramatically different course. It slows rapidly to a stop.

Friction can be thought of as a countervailing force. But more generally, it is an expression of the world's stickiness in response to change. Systems rarely perform exactly as pure theory would predict (classical mechanics or rational choice theory). And this is true in the social world as well. Take a large agency like the Veterans Administration. Top executives may declare that long waiting lists for seriously ill veterans are no longer acceptable, and they may put in place a set of institutional reforms designed to reduce the average wait. Six months later we may examine the system as a whole and find that some hospitals have quickly implemented the reforms; others have attempted to do so but have failed; and yet others have not taken any action. How can we explain this mix of outcomes? The facts of friction and delay in the system are key factors. Transmission of commands and reforms through an institutional system is always a partial affair, and an unavoidable interference with intention that is a combination of organizational rigidity, resistance, and imperfect communication is the result.

Or take the decline of a religious or ideological movement as a third example. Maintaining a high level of passionate commitment to the movement's ideas and values takes the expenditure of organizational resources. Individual followers have a range of other motivations that compete with their ideological fervor. And this is particularly true when there is a cost associated with activism. So we should expect a gradual decay of activist mobilization unless there is a powerful countervailing force -- effective grassroots mobilization efforts that keeps the faithful fired up.

Each of these seem to be recognizable social tendencies or processes that have a lot in common with entropy in physical systems. Stochastic events, friction, and loss of focused energy are all familiar in the social world. And these factors have a distinct flavor of thermodynamics.

(I've really posed two questions here: is there such a thing as social entropy? And are some features of entropy reasonably classified as mechanisms? It is possible that the examples I've mentioned here do in fact succeed in identifying entropic features of the social world but do not identify entropic mechanisms.)