I often find handbooks in the social sciences to be particularly valuable resources for anyone interested in thinking about the big issues and challenges facing the social sciences at a particular point in time. Editors and contributors have generally made intelligent efforts to provide articles that give a good understanding of a theme, methodology, or issue in the given field of research, without diving into the full detail of a scholarly monograph on the minutiae. Earlier posts have highlighted a number of such handbooks (link, link, link).
An excellent example of such a volume is Hedstrom and Wittrock, Frontiers of Sociology (Annals of the International Institute of Sociology Vol. 11). This volume took its origin in the 2008 meeting of the IIS, supplemented by a number of additional pieces. It includes excellent contributions from highly impactful sociologists from Europe and North America.
The editors describe the motivation for the volume in these terms:
There may be a greater urgency today than for a very long time for sociology to examine its own intellectual and institutional frontiers relative to other disciplinary and scholarly programs but also relative to a rapidly changing institutional and academic landscape. In this sense, current sociology may be in a situation more analogous to that of the classics of sociology and of the IIS than has been the case for a large part of the twentieth century. (1)
A core theme in the volume is the idea of intersections between sociological research and adjacent disciplines.
The large topic areas include:
- The legacy and frontiers of sociology
- Sociology and the historical sciences
- Sociology and the cultural sciences
- Sociology and the cognitive sciences
- Sociology and the mathematical and statistical sciences
A handful of chapters are particularly relevant to themes that have come up frequently in Understanding Society. Here are a few representative ideas from these chapters.
Peter Hedstrom, “The analytical turn in sociology” [analytical sociology]
What I find particularly attractive in analytical philosophy is rather the general style of scholarship as well as a number of specific conceptual clarifications that are of considerable importance for sociological theory.
The style of scholarship one finds in analytical philosophy can be concisely characterized as follows:
- An emphasis on the importance of clarity: If it is not perfectly clear what someone is trying to say, confusion is likely to arise, and this will hamper our understanding.
- An emphasis on the importance of analysis in terms of breaking something down to its basic constituents in order to better understand it.
- An academic style of writing that does not shy away from abstraction and formalization when this is deemed necessary.
By adopting principles such as these, it is possible to devise a more analytically oriented sociology that seeks to explain complex social processes by carefully dissecting them, bringing into focus their most important constituent components, and then to construct appropriate models which help us to understand why we observe what we observe. Let me try to briefly indicate what I have in mind, starting with analysis in terms of dissection and abstraction. (332)
Philip Gorski, “Social ‘mechanisms’ and comparative-historical sociology: A critical realist proposal” [critical realism]
A second important principle of critical realism, which follows directly from the above, is ontological stratification. This principle is a familiar one for social scientists, who often speak of the various “levels” or “dimensions” of a particular “system” or “phenomenon.” But critical realism provides a basic principle for identifying these levels: namely, the principle of emergence. Consider the pin factory example once again. A modern manager or engineer could undoubtedly increase the output of the factory’s workforce simply by making various kinds of organizational or technical adjustments—to the sequencing of tasks, the spatial layout, the introduction of new tools or machines, and so on. From the perspective of critical realism, then, the factory qua organization or institution is an autonomous reality.
This is not to deny that the output of the pin factory could also be increased by changing the composition of the workforce—e.g., by hiring more skilled or dextrous or energetic workers. The principle of stratification should not be confused with the principle of holism. To say that the whole is greater than the sum of its parts is not to say that the composition of the parts is of no consequence. This brings us to a third important principle of critical realism: explanatory a-reducibility. A-reducibility is not the same as irreducibility. Irreducibility implies that one level or strata of reality cannot be explained in terms of another at all, that decomposing that strata into its constituent parts is useless; a-reducibility, on the other hand, merely implies that one level or strata cannot be fully explained in terms of another. So, to say that the output of the pin factory is a-reducible to the composition of its work force is not to say that the latter has no effect on the former, only that no individual-level property or power can fully explain the collective output of a factory, which is to say, that the factory has emergent powers and properties—that it is real. (148-149)
Richard Breen, “Formal theory in the social sciences” [agent-based models]
Game theory and social interaction models follow the rational choice model, but there is another approach to modelling the way in which context influences aggregate outcomes which has not developed from, nor necessarily makes use of, rational choice: this is agent based modelling. Agent based models (or ABMs) are simulations, usually computer based, of the interactions of agents. These simulations are used to generate aggregate properties of the agent population using very simple characterisations of the agents themselves. As Macy and Willer (2002: 146) put it in their review of the field: ‘ABMs explore the simplest set of behavioural assumptions required to generate a macro pattern of explanatory interest’. The roots of this approach are deep, but it is only within the past 15 years or thereabouts that agent based models have become popular. Nevertheless, one of the most influential pieces of agent based modelling predates the availability of powerful computers: this is Schelling’s (1971) model of residential segregation. The Schelling model consists of a set of agents, each of which is of one of two types—call them red and blue—arrayed on a network structure such as points on a circle or a lattice. These agents have a preference for a balance of neighbours of each type and in each round of the simulation those agents who are dissatisfied in this respect are allowed to move to any other available location. The strength of preference can be varied in the Schelling model, reflecting the extent to which agents want to be with a majority of their own type or are willing to be in a minority. It transpires that, even if agents’ preferences for neighbours of their own type are very weak, after several rounds of the simulation there is complete segregation, so that the network consists of areas that are completely red or completely blue but lacks any mixed areas. This outcome is a stylized version of residential segregation by race in the United States, but the model suggests that a strong desire for such segregation on the part of the agents is not necessary to generate segregation: it can arise from only weak preferences for residing with one’s own type. The Schelling model nicely illustrates the characteristic feature of agent based models: namely that a complex social outcome is generated as an emergent property of the interaction of agents who follow very simple rules of behaviour. A more recent and almost equally well known, but much more elaborate, example of an agent based model is the Sugarscape model of Epstein and Axtell (1996) in which a population of agents, again following simple rules, exploits the natural resource of a landscape and in so doing generates social outcomes, such as a distribution of wealth and cultural distinctions which, according to the authors, closely resemble their real world counterparts.
In obvious contrast to the agents of rational choice models, those of agent based models of the sort I have referred to follow simple rules of the kind ‘if X then do Y’—so, in the Schelling model, the rule is: if your preferences are not met move; if your preferences are met, stay where you are. To the extent that agents change their behaviour they are adaptive, rather than forward looking. (218-219)
Hans Joas, “The emergence of universalism: An affirmative genealogy” [new pragmatism]
The type of sociology I am representing here tries to understand—as I said—creative processes. The creative process that interests me most at the moment is the creation of new values. I prefer not to speak of the creation, but of the genesis of values, because it is one of the seemingly paradoxical features of our commitment to values that we do not experience them as being created by us but as captivating us, attracting and grasping us (“Ergriffensein”). There is a passive dimension in all creative processes that has always been described in terms like “inspiration”; but in the case of values it is clearly only from an observer’s standpoint that we can see values as created. The participants in these processes do not feel committed to an entity of their own making, but consider values as being discovered or rediscovered.
A sociological study of major innovations in the field of values is neither a philosophical attempt to offer a rational justification for these values nor a mere historical reconstruction of the contingencies of their emergence. Philosophical justifications do not need history. In the case of human rights, they mostly develop their argument out of the (alleged) character of reason or moral obligation as such, out of the conditions of a thought experiment or the fundamentals of an idealized rational discourse. The history of ideas is then mostly seen as the pre-history of the definitive solution that can be found in the work of Kant or Rawls or Habermas. —Historiography, on the other hand, certainly has some implicit elements of an evaluative character and of their justication and it can be a historiography of philosophical, political, or religious arguments concerning human rights and universal human dignity. But as historiography it seems to be nothing but an empirically tenable reconstruction of historical processes and not a contribution to the justification of values. In their division of labor philosophy and history support a strict distinction between questions of genesis and questions of validity. (18)
Jack Goldstone, “Sociology and political science: Learning and challenges” [internal organization of sociology as a discipline]
Both economics and political science have relatively simple structures to their academic fields. Economics divides itself, roughly, among micro- economics, macro-economics, trade or international economics, and economic history. Of course there are myriad specialties that cross these lines: labor economics, price theory, general equilibrium theory, development economics, finance, welfare economics, experimental economics, health economics, environmental economics, the economics of public goods, the economics of innovation/R&D, etc. etc. Nonetheless, economics gives itself great coherence as a field by having a simple fourfold division to guide its basic undergraduate and graduate foundation courses, and its staffing of departments.
Political science similarly has a four-fold division: American (or other home country) politics, comparative politics, international relations, and theory (actually the history of political thought). This can be even more simply conceptualized as the politics of my country, the politics of other countries, the relations among various countries, and the intellectual history of the field. Again, political science has myriad subfields that operate within and across these divisions: comparative democratization, legislative politics, the presidency, deterrence, political psychology, opinion research, public administration, etc. etc. But the basic four-fold division gives a certain coherence to structuring pedagogy and hiring.
Sociology, sadly, has no such internal structure. There is no reason for this, except for the history of the field as succumbing alternately to grand unifying visions (Marxism, Weberianism, Functionalism) and fractioning into a large number of different specialties. Sociology certainly could set itself up as having four basic fields: American (or the national society in which the department is located) society, comparative macro-sociology (the sociology of other societies and their relationships, including most of development and political sociology), micro-sociology (social psychology, small-groups, social identity, race/ethnicity/gender), and organizational or meso-sociology (professions, organizations, net- works, business, religion, education, medicine, etc.).
Of course sociology, like economics and political science, could retain its hundreds of sub-specialties that cut across these divisions. But requiring all students to take a course in each of these four areas, and specialize in one of them, would give much greater coherence and institutional structure to the field. Right now, sociology presents itself through one vast intro survey course and an unstructured mass of specialized courses that have little intellectual organization. (60-61)
Aage Sørensen, “Statistical models and mechanisms of social processes” [social mechanisms]
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.
We need to estimate relationship in order to figure out if our theoretical ideas have some support in evidence. For this purpose, we use statistics. Statistics is a branch of mathematics and might seem alien to sociologists for this reason. However, while sociologists are not very eager to formulate their theories as mathematical models, sociologists are very eager to learn statistics, and quite good at it. They therefore use statistical models to estimate the relationships that concern them in research. These statistical models are usually presented by statisticians as default models, to be used when a substantive model is lacking. The models are invariably additive, at least as a point of departure. They have the virtue of being parsimonious. The statistical models have the defect that they sometime are poor theories of the processes under investigation. (370)
I have argued that the integration of theory and evidence in sociology must take place by choosing theories that allow for such integration by producing ideas about what generates change in social processes. In order to be fully successful, this strategy should result in the formulation of mathematical models for change. These ideas are not novel. They form the justification for the creation of a specialty in sociology called mathematical sociology and the power of the ideas is illustrated with numerous examples in Coleman (1964) and related work. Nevertheless, sociologists have largely chosen a different strategy for going about testing their ideas with evidence. They adopt ad hoc statistical models as described in textbooks and they have become quite sophisticated at statistics. I have tried to show that this use of statistical models and the fascination with their elaboration has blinded sociologists to exploring fundamental properties of the processes they investigate, such as social mobility processes. In turn this has sometimes produced quite unreasonable substantive models for the processes being investigated, as in the case of research on schools effects.
There are several reasons for why statistical models came to dominate. I have argued elsewhere that the increase in computing power allowed sociologists to estimate what with less computing power had to be modeled (Sørensen 1998). So instead of formulating stochastic process models for the mechanisms of what we are interested in, we proceed directly to estimate how parameters of these processes depend on independent variables in hazard rate analysis. (396-397)
These are talented sociologists, searching for some new ways of approaching the task of understanding social change, social variation, and social stability. Further, like the larger universe of sociology today, there are important tensions among the approaches highlighted here — from historical and comparative research to causal mechanisms to statistical inference and causal modeling. And yet, in spite of the range of approaches offered in the volume, sociological methods and sociological theorizing would benefit from an even greater range of pluralism and innovation. Understanding the contemporary social world requires new appreciation of the complexity and heterogeneity of the social processes in which we live, and new tools for investigating and theorizing those processes. (Here is a post from 2008 that makes the case for theoretical and methodological pluralism in sociology — ironically, from the same year in which the Hedstrom and Wittrock volume appeared.)