Skvoretz & Fararo (2011) - Mathematical sociology

7/24/2019 Skvoretz & Fararo (2011) - Mathematical sociology

1/14

Sociopedia.isa 2011 The Author(s)

2011 ISA (Editorial Arrangement of Sociopedia.isa)

John Skvoretz and Thomas J Fararo, 2011, Mathematical sociology, Sociopedia.isa,DOI: 10.1177/2056846011102

1

Mathematical sociology weds mathematics and sociol-ogy to advance the scientific understanding of socialstructures and social processes. Within the broadarena of sociology, it stands in that corner defined bya generalizing orientation, by the belief that a scienceof social orders is possible, by a commitment to a log-ical derivation of empirical regularities from formallystated axioms or assumptions, and by a concern forthe integration and unification of sociological theory.Mathematics in this context is used to directly formu-

late theory and derive testable hypotheses.Mathematics is also widely used in data analysis in thesocial sciences but such use is not intended to be inthe service of theory formulation. Nevertheless, prac-titioners of quantitative data analysis and practitionersof mathematical sociology can overlap substantiallysince the requisite skill set, facility with mathematicalformulations and the training to reason logically frompremises, is common to both enterprises. This articlesurveys the field of mathematical sociology, its histo-ry, its presuppositions and current areas of develop-ment. It then outlines some of the basic challenges

faced by mathematical sociology, closing with a com-mentary on its future directions.As an intended scientific discipline, sociology is

relatively new and mathematical sociology evennewer. The canonical problems of the discipline arenot agreed upon and a variety of theoretical perspec-tives exist as competing paradigms. The array of such

general perspectives in sociology is what is usuallymeant by sociological theory. At the level ofmacrosocial analysis, one perspective (functionalism)treats values in terms of their social integrative signif-icance while another (conflict theory) treats values asweapons in struggles among groups. At the level ofmicrosocial analysis, one perspective (exchange theo-ry) treats social interaction in terms of interchange ofresources of any kind, while another perspective (sym-bolic interactionism) treats social interaction as com-

munication of social meanings. These perspectivesand others have arisen in pursuit of the goal to formu-late fundamental sociological problems what sociolo-gy should explain and how. In turn, the use ofmathematics in sociological theory has arisen in thecontext of attempting to find more effective means ofthe pursuit of explanatory goals, drawing upon butnot limited to existing theoretical perspectives.

History

Interest in mathematical sociology took off in thepost-Second World War period. The pre-war work ofRashevsky (1939a, 1939b, 1940a, 1940b, 1941,1942) was emblematic of early efforts in mathemati-cal sociology problems were approached in a grandtheory fashion with little attention to relevant data.The post-war period, the 1950s and 1960s, was a

Mathematical sociology

John Skvoretz University of South Florida, USA

Thomas J Fararo University of Pittsburgh, USA

abstract The history, presuppositions, current developments, challenges and prospects for mathemat-

ical sociology are described. Common research threads in the field from its earliest post-Second World

War days to the present are identified. A case for the importance of mathematical sociology to the future

of sociological theory is made.

keywords mathematical sociology model building process sociological theory structure


2/14

2

Skvoretz and Fararo Mathematical sociology

fruitful one for mathematical sociology in partbecause theorists began by paying attention to signif-icant empirical regularities, the precision of whichvirtually demanded formal analysis. Many of theseregularities derived from laboratory studies of small

group processes. Specifically, there were the studiesof participation in task groups by Bales and col-leagues (Bales, 1950; Bales et al., 1951), the studiesof the effect of the structure of a communicationnetwork on group performance and individual out-come by Bavelas and his colleagues (Bavelas, 1948,1950), Leavitt (1949, 1951), Smith (1950) andBavelas and Barrett (1951) and the studies of con-formity by Asch (1951, 1952, 1956).

The importance of small group research wasclearly recognized in the title of one of the earlybooks to organize mathematical thinking in sociolo-

gy, Types of Formalization in Small Group Research(Berger et al., 1962). Useful to this day is their cate-gorization of formal models by three types of pri-mary goal: explication, representation andtheoretical construct. In the first type mathematics isused in the explication, or rendering of precisemeaning, to one or more basic concepts (Berger etal., 1962: 7). Their illustration is the use of graphtheory by Cartwright and Harary (1956) to rendermore precise the idea of balance, the primary driverin Heiders theory of interpersonal relations (1944,1946, 1958). In the second type, the theoristattempts to represent in as precise and formally sim-

ple a manner as possible a recurrent but specificinstance of an observed social phenomenon (Bergeret al., 1962: 7). The paradigmatic illustration isCohens (1958, 1963) Markov chain model of theAsch conformity process. In the third type, the the-orist aims to provide a direct means of developing a

generalexplanatory theory which formally accountsfor a variety of observed processes (Berger et al.,1962: 67; italics in original). The exemplar they dis-cuss is the stimulus sampling learning model ofEstes and Burke (1955), a model proposed to explainoutcomes in one type of learning experiment (dis-

crimination learning) whose conceptual machinerywas sufficiently general that it was elaborated toapply to multi-person situations (Atkinson andSuppes, 1958).

Another common route to mathematical modelswas to recast verbally stated theoretical claims in for-mal terms. Simons (1952) formalization of Homans(1950) theory is the classic example. The conceptualfoundation of Homans theory is a social systemmodel in which there are two subsystems, externaland internal. Each subsystem is described by threevariables: activity, interaction and sentiment. In theexternal system, the variables are specified as task-related and consisting of interactions needed for the

task; in the internal system, the three variables char-acterize interpersonal relations such as friendships.The general idea is that the internal relationshipsemerge out of the social activity and interactionrequired for adapting to the groups environment but

then feed back to alter or reinforce that adaptation.For instance, a work groups strong internal ties mayenable it to resist some new practice initiated byhigher management (environment).

Homans made several claims about how in gener-al activity is related to sentiment, sentiment to inter-action, and so on from his analysis of five groups thathad been subjects of detailed field studies by socialscientists. Simon formalized these assertions by spec-ifying mathematically the functional relationshipsbetween the variables of activity, sentiment andinteraction and connecting them via differential

equations into a dynamic system characterizing thegroup. The system of equations was analyzed for theconditions under which equilibrium points existedand were stable. The hope was that such conditionscould then be interpreted in meaningful sociologicalterms that expanded the theoretical understanding ofthe establishment and maintenance of social groups.

Other foundational work appeared not in sociol-ogy journals but in the Bulletin of MathematicalBiophysics. In a series of papers, Anatol Rapoport andcolleagues set out and explored formal models forpotentially extremely large groups (Landau, 1952;Rapoport, 1951a, 1951b, 1953a, 1953b, 1953c,

1957, 1958, 1963; Rapoport and Horvath, 1961;Rapoport and Solomonoff, 1951). In particularRapoports random and biased net theory focused onmodeling social networks the structure of socialties of a particular type in a bounded population.The models used probabilistic formalisms first todescribe properties of a random net and then tomodel departures from randomness by biases thatmade certain connections more likely than chancewhen the relational context was favorable. Forinstance, the idea that a person, ego, is more likely tochose another as a friend if the other has also chosen

ego as a friend was formally expressed by a reciproc-ity bias. The basic problem concerning Rapoportwas the tracing problem: how biases and the ran-dom chance of connection affected reachability, theproportion of nodes in a population that could bereached from a randomly selected starter node. Theaverage proportions of newly reached nodes in eachgeneration, as links were traced out from an arbitrarystarting set, were termed the biased net structure sta-tistics (Fararo and Sunshine, 1964). Biases weredefined in the context of the tracing procedure.Rapoport derived formal expressions for the struc-ture statistics but only under some strong approxi-mation assumptions that later research showed


3/14

3


needed correction to improve fit to available data(Skvoretz, 1990).

By the 1960s and 1970s, a sufficient number ofexemplars of the use of mathematics in sociology hadaccumulated that there was a spate of textbooks

directed at introducing a reader to mathematicalsociology: Colemans Introduction to MathematicalSociology (1964), Doreians Mathematics and theStudy of Social Relations (1970), Fararos

Mathematical Sociology: An Introduction toFundamentals (1973) and Mathematical Sociology(1975) by Leik and Meeker, although the first textwas arguably Karlssons Social Mechanisms: Studies inSociological Theory(1958). Second, there were sever-al texts that either aimed for a wider audience ofsocial scientists, such as Mathematical Models in theSocial Sciences (1962) by Kemeny and Snell and

Introduction to Models in the Social Sciences(1975) byLave and March, or developed applications of a par-ticular mathematical formalism for a more narrowlydefined set of applications, such as BartholomewsStochastic Models for Social Processes(1967). Finally,mention should be made of Boudons work usingmathematics to explore aspects of mobility and edu-cational stratification systems in MathematicalStructures of Social Mobility (1973) and Education,Opportunity, and Social Inequality (1974), a studywhich also employed numerical simulation.

Of the mathematical textbooks, the ones byColeman and by Fararo were the most ambitious and

comprehensive yet they were quite different efforts.Colemans introduction grew out of immersion inlarge-scale survey analysis and a recognition thatmuch of the data in sociology came in the form ofproportions or percentages. Because of his earlytraining as an engineer, Coleman was well aware thatsystematic measurement was necessary to scientificadvance and that it was lacking in the social sciences witness the title of his Chapter 2, The problems ofquantitative measurement in sociology. Measuresbased on counting, however, bypassed some of theproblems facing social scientists trying to measure

cherished concepts like solidarity or the division oflabor. In Colemans version of mathematical sociolo-gy, mathematical models of social phenomena weredriven by this recognition that counting things inclasses provided the systematic measurement basisnecessary for the science of sociology. Hence his fun-damental type of model was a continuous time, dis-crete state stochastic process in which personsrandom movement from one state to another wasdetermined probabilistically by endogenous andexogenous forces representing social effects likeinfluence. Colemans introduction did not aim toacquaint the reader to the variety of mathematicsthat could be used to formalize sociological theories.

Fararo was equally aware of the importance ofsystematic measurement to scientific advance yet thematerial he presented was much less driven by themeasurement basis of counting and more oriented toa comprehensive survey of useful mathematical tech-

niques in social science. Topics covered included sto-chastic process models, specifically Markov chains assource models for the formation of images of strati-fication, the formation of expectation states and therepresentation of social mobility; abstract algebra assource models for social relational structures like kin-ship; and the theory of games as proposed by VonNeumann and Morgenstern (1947) as source modelsfor interactive choice situations or, more narrowly,rule-governed interactive choice situations, thoughtof as the core generator of sociological puzzles andproblems. He also provided extensive background

introduction to key mathematical concepts likemappings, sets, vectors, matrices, Markov chains andabstract groups. Fararos introduction was squarelyaimed at acquainting the reader with the variety ofmathematics that could be used to formalize socio-logical theory and model sociological processes andstructures.

The appearance of textbooks was one of the indi-cators marking the start of the institutionalization ofmathematical sociology as a field within the disci-pline. Other indicators included the establishment ofjournals and professional associations and the found-ing of graduate programs. On the first point, the first

issue of The Journal of Mathematical Sociologywaspublished in 1971. Later in the decade, came thejournal Social Networks, founded in 1978, followedby the annual publication Advances in GroupProcesses, founded in 1984, the journal Rationalityand Society, founded in 1989, and then much later in1998 the online Journal of Artificial Societies andSocial Simulation and in 2000 the electronic journalof the International Network for Social NetworkAnalysis,Journal of Social Structure. The formationof professional associations lagged about a decadebehind these developments. Early on the scene was

the Japanese Association for Mathematical Sociologyfounded in 1986. But it was not until 1997 that aSection on Mathematical Sociology of the AmericanSociological Association was established. In 2000,the section collaborated with the Japanese associa-tion to organize the first joint JapaneseAmericanConference on Mathematical Sociology (held inHawaii). The activity of these groups is helping tochange the earlier situation of the field, described byFararo (1997) as one of lacking common identityand solidarity.

As early as 1970 the University of Pittsburgh hada mathematical sociology track in its PhD programlargely due to the efforts of Tom Fararo, Pat Doreian,


4/14

4


and somewhat later Norm Hummon. The most suc-cessful program of the day was associated withHarrison White at Harvard. Although White neverauthored a textbook in mathematical sociology, heprovided key monograph-length exemplars of the

creative use of mathematical concepts to modelsocial structure and social process. Trained initiallywith a PhD in physics before taking a PhD in soci-ology, Whites first monograph in mathematical soci-ology, An Anatomy of Kinship (1963), tackled astructure problem using abstract algebra to analyzekinship structures as compositions of elementarykinship relations and argued that different kinds ofkinship structure emerge from placing differentcompositions of relations in equivalence to oneanother. A second monograph in 1970, Chains of Opportunity, tackled the process problem of social

mobility in a well-defined career system and stoodthe prevailing conceptualization on its head. Thetypical approach viewed the flow of persons througha set of positions as the modeling problem, whereasWhite viewed it as the flow of a vacancy through thatset of positions. He showed that the vacancy flowcount could be modeled as a Markov process whilecreating by implication a flow of persons throughpositions that were non-Markovian and far more dif-ficult to model. The program at Harvard underWhites intellectual leadership produced an accom-plished generation of mathematically inclined schol-ars: Phil Bonacich, Ron Breiger, Ivan Chase, Bonnie

Erikson, Mark Granovetter, Joel Levine and BarryWellman, among others.

Finally, mention should be made of two long-standing research programs that have been quite sup-portive of formal models in sociology, even if theywould find it awkward to be claimed as research pro-grams in mathematical sociology. The first and mostprolific is the expectation states program based atStanford University and founded by Joseph Berger,Bernard P Cohen and Morris Zelditch Jr (Correlland Ridgeway, 2003). The number of outstandingscholars trained by this program and associated with

it is truly remarkable. The programs consistent focushas been on the development of power and prestigeorders in task groups and the myriad ways in whichexogenous status considerations and endogenous sta-tus processes impact an actors position in the powerand prestige order. The second program is that ofDavid Heise and colleagues on affect control theory,a theory designed to model action and emotionbased on the affect profiles of situational identitiesand activities (Robinson et al., 2006). Rooted in theclassical tradition of symbolic interactionism, thisresearch program uses control theory and its associ-ated mathematical machinery of matrix algebra tounderstand how the meanings of separate elements

comprising a situation (as measured by the semanticdifferential) combine to reinforce those fundamentalsentiments. These two and a number of other theo-retical research programs that employ mathematicalmodel building are included in the edited volume

New Directions in Contemporary Sociological Theory(Berger and Zelditch, 2002).

Overview and presuppositions

At the heart of the scientific research enterprise is thecoordination of theory with relevant data. Scientistssystematically collect information in order to testtheories sets of ideas about how the natural worldworks. Social groups and their workings are part ofthe natural world and the subject matter of the social

sciences, sociology in particular. Verbal formulationsof typical sociological theories are useful up to apoint. However, precise understanding requires(among other things) that hypotheses be stated asprecisely as possible and hence the utility of mathe-matics to sociology (and to all other sciences as well).Mathematics is particularly valuable to the social sci-ences where theories stated in ordinary language caneasily import layers of hidden meaning and culturalcontent that skew testing and analysis. This claim is,of course, not widely shared by theorists who eschewformal argument.

This claim should also be distinguished from the

point that mathematics brings value to social sciencevia statistical techniques for data analysis. The differ-ence between mathematical models and statisticalmodels has been addressed in different ways byCollins (1988: Appendix A), Skvoretz (1998) andSrenson (2009). Collins suggests that behind anystatistical analysis is a substantive theory in whichoperative causal mechanisms produce random distri-butions and that explicitly building on such a theo-ry should be a priority for social science. Srensonargues that statistical models are often used becausesociologists do not have good substantive models

based in theory and so default to the statisticians rec-ommendations. Finally, Skvoretz demonstrates howconsequential the choice between a default statisticalmodel and a theoretically derived model can be forassessing whether a theory is generally disconfirmedon one hand or clearly supported on the other. Allthree underscore the point that mathematical sociol-ogys value lies on the theoretical side of the scientif-ic enterprise rather than on its methodological side.

Fararo (1973) provides a useful overview of thepresuppositions and steps involved in any researchproject in mathematical sociology. FollowingToulmin (1953), Fararo holds that theoreticalscience searches for novel and systematic ways of


5/14

5


representing phenomena. The representations giveexact form to the phenomenon of interest. So, forinstance, the conformity process as studied by Asch(1951) is a process by which a person makes asequence of choices, each of which either conforms

to a groups judgment or not. The representation ofthis process by a Markov chain, as proposed byCohen (1963), gives exact form to this process there are a number of states (four) that a person mayoccupy, each state corresponds to a particularresponse with probability 1, and between one choiceand the next, a person may move from one state toanother, the probabilities of movement dependenton the exact parameters of the Markov process andinterpretable in terms of the referent phenomenon ofconformity. Fararo notes that one advantage of for-mal representation is that it makes certain questions

meaningful (how does the time to absorption intothe state of permanent conformity depend onparameters?) and rules out others (does failure toconform to the group judgment indicate exceptionalindependence of character and leadership poten-tial?). Another advantage is that the formal represen-tation, if found to be generally descriptivelyadequate, allows the theorist to raise the deep ques-tion of why it should be so, the question of explana-tory adequacy.

A generic method of representation provides, inFararos (1973: Sect. 1.3) terms, a framework and heembeds model building, the key step in any research

project in mathematical sociology, in the context offramework construction. The research project beginsimplicitly or explicitly with the framework in a cur-rent state and that state, along with interest in someempirical phenomena, leads to a scientific problem.For example, the general framework which stipulatesthat influence flows through a group represented as asocial network of significant connections among per-sons, along with the interest in the process of adop-tion of some new technology, leads to the scientificquestion of how adoption is driven by contagionfrom connections and/or by homophily or attribute

similarity among persons (Aral et al., 2009). Modelbuilding then attempts to solve the scientific prob-lem. The activity of building models has three steps:(a) model setup, (b) model analysis and (c) modelapplication. A valuable research project may stopwith the first two of these steps: Simons formaliza-tion of Homans Human Group is an example.

Model application itself, as Fararo specifies, hasfour substeps: (a) identification of abstractions, (b)estimation of parameters, (c) calculation of predic-tions and (d) evaluation of goodness of fit. Skvoretz(1981) provides an exemplar for the model buildingstage: the research question is the distribution of par-ticipation in task focused groups of any size and the

model is set up as probabilistic process in which thestrength of a persons inclination to participate rela-tive to the strengths of all others in the group deter-mines the probability of participation and strength isdetermined by the persons comparative status in the

group. Analysis derives relationships between the sta-tus composition of the group and the chances thatpersons occupying particular statuses in the groupparticipate. The application phase uses data fromadministrative conferences in a psychiatric hospital the operative status dimension is identified, parame-ters calibrating status effects on participation are esti-mated, predicted distributions are derived and,finally, comparisons of predicted to observed distri-butions assess fit.

Fararo (1973: 6) makes the important observa-tion that feedback loops connect the steps in the

research process as the results of a model evaluationcan feed back to the first two steps of the modelbuilding and that in turn can feed back to reformu-late the scientific question. Over time such reformu-lations of the scientific question can change the stateof the framework suggesting different or morerefined methods of representation that evoke a newcycle of model building efforts. Again work onaccounting for status effects in the distribution ofparticipation in task groups can serve as an example:limitations in the data addressed in Skvoretz (1981)led to new data collection efforts to evaluate themodel (Smith-Lovin et al., 1986) and the evaluation

led, in turn, to more sophisticated models (Skvoretz,1988) but also to concern with emergent status dis-tinctions from behavioral cues and their integrationwith exogenous diffuse status effects on participa-tion. The scientific question is reformulated and theframework shifts to a discrete state stochastic processin which emergent and activated status orders (andthus states of the stochastic process) are representedby an evolving network of precedence relations(Skvoretz and Fararo, 1996).

It is in the model building phase that the fullmenu of options offered by modern mathematics

comes into play. Following Fararo (2001) there are,broadly speaking, two types of models: process mod-els and structure models. Process models aim to cap-ture the trajectory of a system as it moves over timethrough a state space. Key elements of a processmodel are the conceptualization of the time domainas discrete or continuous, the definition of the statesas discrete or continuous, the conceptualization ofthe states as probability distributions or determinis-tic positions, the specification of the change-of-staterules and the conceptualization of parameters usedby these rules as discrete or continuous. Simonsprocess model, for example, falls in the category ofcontinuous time, continuous and deterministic state,


6/14

6


with continuous parameters in change-of-state rulesembodied by differential equations. Key questionsfor the analysis of such models are the existence andstability of equilibrium states and their dependenceon the relative values of key parameters. These equi-

librium states may be specific positions or states orprobability distributions over a set of states or posi-tions.

These basics of a dynamic process model are use-ful not only to elucidate specific content areas butalso as a metaphor for the aims of general theoreticalsociology. Fararo (1989: 109) uses these basic ideasto define the four fundamental problems of theoret-ical sociology: (1) the existence and forms of socialstructures (are there equilibrium states of socialprocesses?); (2) the stability of social structures (doesthe system return to an equilibrium state if small

shocks move it out of that state?); (3) the compara-tive statics of social structure (how do the equilibri-um states of social processes depend on parametersof the processes?); and (4) social change conceptual-ized as movement from one equilibrium state toanother (what higher level process can producemovement from one such state to another via para-metric change?). The last problem provides a formalviewpoint on the classic micromacro problem insociological analysis. It adds the intriguing idea thatthe connection is made through a feedback loopfrom the state space of the micro process to itsparameter space which constitutes the state space of

the macro process.Structure models, on the other hand, focus on a

set of abstract objects, held to represent some socialphenomenon of interest, and study ways of charac-terizing important properties of these abstractobjects. The notion of balance in signed graphs isone of the clearest examples of a structural model.The type of abstract object is a collection of nodeswith pairwise edges between them that have eitherpositive or negative valence, meaning that they haveweights that can be interpreted as positive or nega-tive constants (usually 1 or 1). Interpretively, the

collection could be a group of persons and the edgesconnecting pairs who like or dislike each other. Amuch cited result of the analysis of this structuralmodel is Cartwright and Hararys (1956) proof ofthe structure theorem: that the nodes in every bal-anced signed graph may be partitioned into two sub-sets (one of which may be empty) such that positiveedges join nodes in the same subset and negativelines join nodes in different subsets.

Signed graphs are part of a larger family of struc-ture models in which structure is represented by anetwork. Three other families of structure models areones in which structure is represented by distribu-tions, by rule systems (grammars) and by games

(Fararo, 2001). Distribution models focus on prop-erties of the (static) distribution of persons into posi-tional categories. The classic use of these models isfound in the work of Blau (1977), whose concernwas how properties of these distributions affected

intergroup relations. Rule system models find use inrepresenting institutions as systems of social roles(Fararo and Skvoretz, 1984, 1986). Game-theoreticmodels have been widely used to represent interac-tive choice situations such as public goods problemsand emergence of cooperation in prisoners dilem-mas and are much more prevalent in the work ofeconomists than sociologists (Macy and Skvoretz,1998). Game-theoretic concepts also play a role inthe rational choice framework associated with thelater work of James Coleman (1990) and others. InColemans framework, the structure model is defined

by actors initial endowments of (a) interests overevents or resources and (b) control over those eventsor resources. Under a utility maximization assump-tion, control over events of lesser interest is tradedfor control over events of greater interest until acompetitive equilibrium is reached. Coleman usesthis model to analyze power, trust, norms and con-stitutions.

This categorization into process models vs struc-ture models should not be pushed too far. Fararo(2001) includes a category of models combiningprocess and structure. Indeed, there are researchprojects in mathematical sociology that have ele-

ments of both: for instance, the models developed inSkvoretz and Fararo (1996) for participation in taskgroups are process models in which the statesthrough which a task group moves are represented bya structure model, namely, networks of directed tiesof precedence connecting persons. Also the rationalchoice framework can incorporate process elements(Coleman, 1990: 899931; Fararo and Skvoretz,1993). Nevertheless the distinction is useful forthose just learning about mathematical sociologyand wanting to learn more about its signatureachievements.

Perhaps the most recent development in mathe-matical sociology that represents the biggest depar-ture from its classical foundations is the use ofagent-based models to investigate social phenomena.This modeling framework developed outside ofmathematical sociology, although one of its earliestsuccesses, Schellings segregation model (1971), waspublished in the first issue of the Journal of

Mathematical Sociology. The use of this framework isso distinctive that the research area has its own name,computational sociology, a term first widely used byHummon and Fararo (1995).

Agent-based models focus on systems consistingof multiple agents and the emergence of system


7/14

7


regularities from local interactions between agents.Agents have internal states and behavioral rules andthe rules may be fixed or changeable through experi-ence and interaction. Agents are boundedly rational;they have only limited information processing and

computational capacity. Agents interact in an envi-ronment that provides resources for their actions.Typically, agents and/or the rules they use thrive ordie based upon their success in obtaining resources.

The setup of an agent-based model requires thatsimulation be used to analyze its consequences. Insuch a model, there are typically many agents andprobabilistic considerations figure in the determina-tion of who interacts with whom and in the determi-nation of the changes of agent state. Mathematicalanalysis of such a system for equilibrium solutions isnot feasible. The only way to explore logical conse-

quences is through simulations. In this field, thedesign of the agents and the rules under which theyinteract are the assumptions of the formal theory andsimulation plays the role of deduction from that the-ory. In general, the aim is to derive regularities at theaggregate level from the interaction of agents follow-ing relatively simple rules at the micro level. Suchregularities are emergent relative to the lower-levelrules of interaction and agent state change and thus,in principle, not predictable from these rules.Therefore, simulation is used to detect such emer-gent regularities.

An agent-based research project has three phases:

model setup, model implementation and executionand inductive analysis of model output. In the firstphase, decisions are made about how agents mayinteract and what rules govern their changes of state.In the next phase, the system of agents is encoded ina computer program and then various runs of theprogram made. In the last phase, the output is thenanalyzed for regularities that can be reasonablyattributed to the underlying assumptions about thebehavior constraints on agents encoded in the pro-gram. Care must be taken so that substantive mean-ing is not attributed to regularities that are artifacts

of implementation. The modeling exercise is con-vincing when the assumptions about behavior areclear and intuitively reasonable or based clearly onexisting theory, the program implementation istransparent, a full range of initial conditions and val-ues of basic parameters is explored, clear regularitiesemerge and variation in these regularities can beinterpreted in terms of the models original assump-tions. There remain difficult issues in the coordina-tion of such models with observational data.Occasionally, the simple derivation of results by sim-ulation is treated as an empirical test but this is amistake at best, such derivations can demonstrateplausibility, but full-scale testing of the model

requires the usual steps of parameter estimation andderivation of specific empirical consequences. Theproblem of empirically testing agent-based modelshas been addressed recently in a special issue ofEcology and Societyedited and introduced by Janssen

and Ostrom (2006). They suggest four approachesincluding laboratory experiments, role-playinggames, case studies and derivation of stylized facts.

It is clear that agent-based models straddle thedistinction between process and structure models.They also formally address a fundamental problemin theoretical sociology, the macromicro problem:namely, how is it that the micro-behavior of actorsaggregates to macro-regularities of the system theycompose as Raub et al. (2011: 2) state: establish-ing micromacro links to explain social macro-levelphenomena as a result of the behavior of individual

actors is a core aim of model building in sociology.Both observations, along with plentiful computingpower, are reasons for the accelerated interest and useof the agent-based modeling strategy in mathemati-cal sociology. There is, possibly, an even deeper rea-son: agent-based models comprise the only way toachieve a scientific understanding of situations whenan agent interacts with a moderate number of othersover structured networks of connections, situationswhere according to Miller and Page (2007: 221) ourtraditional analytic tools break down. In their viewthe traditional analytic tools of mathematical socialscience produce tractable models when either very

few (usually two) or very many (an infinite number)of agents are postulated.

Agent-based models often make excellent use ofgame-theoretic formulations of situations of interac-tion. Game theory, with its working hypothesis ofutility maximization, had its natural first home ineconomics. Sociologists have had much troubleaccepting the utility maximization hypothesis and somuch of rational choice theory, founded in an eco-nomic perspective on social life, has been criticized,occasionally unfairly, for this reason. However, thenotion of bounded rationality, as developed by

Simon in his Models of Man (1957), is far moreacceptable to sociologists. The problem, however, isthat while there is one clear route to optimization,there are many ways to be boundedly rational.Hence developing formal models that rest on bound-edly rational agents is a challenge for agent-basedmodels. In these models interactions are modeled asrepeated plays of games in the formal sense and themodel must be explicit about the ways in whichagent rationality is bounded. For instance, the agentsin Skvoretz and Fararos early (1995) model of theevolution of reciprocity recall only the last twoencounters and formulate a plan of action based onthe outcomes in these last two encounters. The


8/14

8


action plan specifies which alternative to select in thenext encounter and the choice when coupled withthe selection of the agent encountered determinespayoffs to both parties according to the game matrix.Effectively in this more sociological approach, sys-

tems of agents solve the game through evolution ofstrategies. Economists, too, have analyzed boundedrationality versions of game-theoretic situations,with a particularly advanced formal treatment foundin Young (1998).

Current areas of development

To capture recent trends in mathematical sociology,an examination of issues of the premier journal inthe field is useful, in particular issues of The Journal

of Mathematical Sociology(JMS) from 2008 to 2011.Several trends stand out: the importance of smallgroup experiments in providing data regularities forformal explanation, the continuing interest in bal-ance as a master concept, the permeability of theboundary between pure mathematical sociology andproblems in quantitative methodology, the use ofsimulations to derive results from models, the peren-nial interest in the micromacro theme and theproblem of emergence, and the fertility of problemsgrowing out of research on networks of social rela-tions.

Before these trends are inspected in detail, how-

ever, it should be pointed out that mainstream soci-ological journals, like American Journal of Sociologyand American Sociological Review, have been andcontinue to be hospitable outlets for mathematicalsociology articles. Early on these journals publishedthe influential works of Granovetter (1978), whoanalyzed the emergence of collective action usingthreshold models and Srenson (1977), whoemployed a vacancy chain model as one aspect of atheoretical analysis of status attainment. These jour-nals also published articles that advanced the foun-dational work of Rapoport on biased net models,

namely Skvoretz (1983) and Fararo and Skvoretz(1987). There are many recent examples of mathe-matical sociology, too numerous to mention, appear-ing in these journals. This sections focus on the

Journal of Mathematical Sociologyis meant to give thereader a more comprehensive survey of the diversityof work in the field than could be gained from a sur-vey of mathematically oriented work in mainstreamsociology journals. However, that such work hasappeared and continues to appear in these outletsoffers evidence that mathematical sociology has arecognized place at the table of contemporary sociol-ogy.

The first trend, the use of data from experiments

in small groups to drive formal modeling, is exempli-fied by Buskens and Van de Rijt (2008), Dogan andVan Assen (2009), Willer and Emanuelson (2008),Gchter and Thoni (2011) and Aksoy and Weesie(2009). The first three use data from experiments on

networks in which exchanges can be made betweenconnected actors and the aim to model the observedregularities in differential earnings by positions, onaverage and over time. The other two articles explorechoice behavior in game-like environments designedto replicate public goods and asymmetric investmentsituations.

A second recent trend in contributions appearinginJMSis the idea of balance continuing to motivateresearch. Deng and Abell (2010) and Abell andLudwig (2009) use analysis and simulation to inves-tigate how a local rule for sign change toward bal-

ance affects the long-term structure of the networkof positive and negative ties. Van de Rijt (2011)investigates networks in jammed states in whichmany triads are unbalanced in the traditional senseyet no change to the valence of any one relation canproduce a net reduction in the number of imbal-anced triads. Montgomery (2009) investigates bal-ance phenomena when an additional consideration,the awareness of the evaluations of the others, isallowed to vary. Finally, Mrvar and Doreian (2009)reconceptualize the original set of balance ideas as, innetwork terms, a two mode phenomenon and theyexplore how blockmodeling techniques from net-

work analysis can be used to characterize balancedstructures as two mode blockmodels.

Several articles illustrate a third trend, namely thepermeability of the boundary between pure mathe-matical sociology and problems in quantitativemethodology, a field that concentrates primarily onstatistical models for data analysis. Snijders (2010),Kejzar et al. (2008) and Ziberna (2008) advancequantitative methods in network analysis. Threeother articles address issues in general structuralequation modeling and demography: the identifica-tion problem in structural equation modeling

(Bollen and Bauldry, 2010), modeling time seriesdata (Singer, 2010), demographic mobility indices(Bevaud, 2008) and the two sex problem (Mic etal., 2008).

Simulation studies, the fourth trend, abound.Two of the previously mentioned articles use simula-tion to illustrate results (Abell and Ludwig, 2009;Deng and Abell, 2010). Fossett (2011) continues theresearch program of Schelling in using agent-basedmodels to study residential segregation. Flache andMacy (2011) explore opinion polarization on differ-ent network typologies with agent-based methods.Helbing et al. (2011) examine the evolution of coop-eration in the prisoners dilemma game in a spatial


9/14

9


and a network environment. Fioretti (2010) usessimulation to derive consequences of a stochasticprocess model for a vacancy chain representation ofresource flows. Hevenstone (2009) studies the use oflabor market intermediaries (temp agencies) with an

agent-based model. Finally, Centola (2009) uses sim-ulation methods to compare complex and simplecontagion in different network topologies, ones withscale-free vs exponential degree distributions, subjectto disruption by random removal of vertices.

Articles using agent-based models illustrate thefifth trend, interest in the micromacro problem andemergent phenomena, because they are often set upto demonstrate how micro specifications of behaviorcan, when agents interact, aggregate to emergentphenomena like waves of cooperation. In addition,however, several articles provide analytical results for

such problems. Raub et al. (2011) and Opp (2011)discuss meta-theoretical issues with respect to themicromacro problem as framed by the Colemanboat image (Coleman, 1990). This image visualizesa macromacro proposition as explained by threelinked processes: a propositional link from macro tomicro, a horizontal link from micro to micro (theconceptual location for individual rational choices)and an up-link from micro to macro. The logic ofColemans boat provides an explanatory frame for acomparative statics macro-level proposition corre-sponding to an empirical generalization through adynamic system model at the micro level based on

rational choice theory and action.Other papers also illustrate this trend. Yamaguchi

(2011) assesses how population heterogeneity withrespect to cost-benefit comparisons affects collectiveoutcomes in two examples, crime and enforcementand gender role attitudes. Jasso (2010) offers aframework for studying the links between micro andmacro phenomena that proceeds from a simple basisof populations characterized by variables. Menicucciand Sacco (2009) and Bischi and Merlone (2009)derive limits on the global behavior of agent-basedmodels for prosocial behavior (preferences that are

sensitive to the payoffs of others) and for binarychoice games with externalities. Finally, Kitts (2008)also derives bounds for collective outcomes for a pre-viously investigated agent-based model of formaland informal control.

Finally, JMS continues to attract mathematicalwork relating to social network analysis. Grassi et al.(2010) prove some general results about the relation-ship between degree and eigenvector centrality fortrees as a class of networks. Friedkin (2010) usesmulti-level event history analysis on a classic networkdata set, the medical innovation data of Coleman etal. (1966), to demonstrate the importance of net-work position in adoption of tetracycline.

Agneessens and Roose (2008) show how exponentialrandom graph models can be used to analyze a two-mode network of persons (theater goers) connectedto events (theater performances).

The remaining papers published in these four

volumes are an eclectic collection. Three have statusas a theme: an evolutionary psychology model link-ing occupational status and fertility behavior(Hopcroft and Whitmeyer, 2010); inequity effectson behavior in some standard game theory designs(Tutic and Liebe, 2009); and the contingent emer-gence of a Matthew effect in status systems(Bothner et al., 2010). Three papers look at proper-ties of some special applications of stochastic models,diffusion of rumors (Molchanov and Whitmeyer,2010), turnover in law firms (Denrell and Shapira,2009) and success in air combat (Simkin and

Roychowdhury, 2008). Two papers by Wyburn andHayward (2008, 2010) apply standard equilibriumanalysis to a system of differential equations model-ing the interaction of linguistic majorities andminorities. The remaining three papers fail to fallneatly into any of the above clusters. Plos et al.(2010) use modal logic to analyze the concept oflegitimation; Hajeeh and Lairi (2009) apply an engi-neering decision analysis method to female selectionof marriage partners in Kuwait; and Mayer (2008)applies differential equations to study transforma-tions and deformations in the space of political posi-tions in the USA over time.

This brief and selective survey of recent work inmathematical sociology gives testimony to the vitali-ty and diversity of the field. Topics from the earliestdays, like balance and network models, continue tobe of contemporary interest. The tools of researchersremain many of the standard and well-known meth-ods of mathematics: differential equations, stochasticprocesses and game theory. Newer tools like agent-based models are prominently represented. Perennialsubstantive problems still drive research: diffusion,social influence, status origins and consequences,segregation, cooperation, collective action, power.

Challenges and prospects

Just as perennial as the substantive problems thatengage mathematical sociologists are the challengesthey face. Mathematical sociologists are or shouldthink of themselves as first and foremost theorists.Putting sociological theory into formal terms is themain task. This is challenging in a field not knownfor its hospitality to formal theory. There are signsof change: the recent prominence of analytical soci-ology as a subfield of theoretical sociology is encour-aging. Analytical sociologists are very much


10/14

10


interested in the mechanisms by which social andcollective phenomena are produced: analytical soci-ology explains by detailing the mechanisms throughwhich social facts are brought about and these mech-anisms invariably refer to individuals actions and the

relations that link actors to one another (Hedstrmand Bearman, 2009: 4). Much of mathematical soci-ology works to lay bare such mechanisms in waysthat make clear how the social facts at issue followfrom actions and relations.

A second perennial challenge is relevance. Amathematical model must abstract from everydayexperience it must represent social life in limitedterms. Natural scientists, especially physicists, havetaken this step with their subject matter the colorof the apple or its taste is irrelevant to its behavior ina gravitational field. No one would think to criticize

Newton for irrelevance because all apples have sometaste and therefore an apples taste must be includedin any explanation of its flight. In comparison, soci-ologists often abstract away from the personalities ofactors when analyzing a social system. Homans inthe Human Group specifically sets aside considera-tions of personality and Simons model makes thisquite clear. Yet there are those who would object tosuch an abstraction and argue that personality differ-ences must be included in social system accountsbecause they are there. This overlooks the analyticalcharacter of scientific theorizing. However, thingsare clearly not so settled in sociology. Hence, the ade-

quacy of a representation/abstraction for a particularset of questions can be a contentious issue and evenif judged adequate, the importance of the questionsmay then be challenged. Nevertheless, mathematicalsociologists must persevere, believing that, asDoreian (1970: 153) has expressed, the future ofsociology as a viable discipline will largely depend onthe use of mathematics in an informed and imagina-tive manner.

Despite these challenges, the long run prospectsfor work in mathematical sociology are robust. Theproblems that attract social scientists and the social

problems that assail contemporary societies are bothproblems that at their core have massive interde-pendency as their signature element. This interde-pendency among the actions of agents and variouslevels of social organization cannot be wished awaynor can it be assumed irrelevant to a systems trajec-tory and outcome. Coming to terms with such prob-lems can only be accomplished throughformalization and creative use of mathematical mod-eling and simulation. Heckathorn (2002) showsthat important applied results can emerge out ofbasic research that includes the construction ofmathematical models. He describes a theoreticalresearch program that concentrated on such basic

sociological problems as how collective action is gen-erated and how social norms emerge but then led toapplied research dealing with interventions to pre-vent the spread of AIDS. In turn, the results of thisapplied research were implemented in real-life situa-

tions. One of his most important conclusions is thatthe stark separation traditionally existing in sociolo-gy between theoretic and applied work need notexist (Heckathorn, 2002: 105). He might haveadded that the use of mathematical models is a keyfeature in the success of such efforts and that suchsuccess can help to advance the place of mathemati-cal sociology within the broader discipline of sociol-ogy.

Annotated further reading

Edling C (2002) Mathematics in sociology.AnnualReview of Sociology 28: 197220.Edlings review of trends in mathematical sociologyhighlights the convergence among process, structureand action in models of social phenomena. Heacknowledges that the rubrics of process, structureand action are common terms to characterize mathe-matical work but argues that recent trends suggest ablurring of boundaries. Of special interest is the factthat Edling conducted short interviews with leadingmathematical sociologists (Peter Abell, PhillipBonacich, Kathleen Carley, Patrick Doreian, ThomasFararo and Harrison White) and these interviews

inform Edlings observations on the past and futureof mathematical sociology.

Freeman LC (1984) Turning a profit from mathematics the case of social networks.Journal of MathematicalSociology 10: 343360.Freeman discusses the early history of mathematics insociology, both successes and conspicuous failures(the failures deriving from a too literal imitation ofthe use of mathematics in the physical sciences). Hethen describes the state of the art (at the time) insocial network analysis in which the alliance betweenmathematics and social science had been extremelyproductive. For Freeman, such work is appropriately

mathematical, problem driven rather than purelyemulative of what worked in other sciences.Heise DR (2000) Thinking sociologically with mathe-

matics. Sociological Theory 18: 498504.Heise recounts his development of affect control the-ory (ACT) from the seed of a sociological question ofhow people learn to perform appropriate role actionsand avoid deviant behaviors without learning anextensive catalog of dos and donts to a full fledgedtheory with extensive formulation in some very stan-dard mathematics. He points out specific examples ofthe advantages formalization brought to the sociolog-ical intuitions.


11/14

11


References

Abell P and Ludwig M (2009) Structural balance: Adynamic perspective. The Journal of MathematicalSociology33: 129155.

Agneessens F and Roose H (2008) Local structural prop-

erties and attribute characteristics in 2-mode net-works: p* models to map choices of theater events.The Journal of Mathematical Sociology32: 204237.

Aksoy O and Weesie, J (2009) Inequality and proceduraljustice in social dilemmas. The Journal ofMathematical Sociology33: 303322.

Aral S, Muchnik L and Sundararajan A (2009)Distinguishing influence-based contagion fromhomophily-driven diffusion in dynamic networks.PNAS106: 2154421549.

Asch SE (1951) Effects of group pressure upon the mod-ification and distortion of judgment. In: GuetzkowH (ed.) Groups, Leadership, and Men. Pittsburgh, PA:

Carnegie Press, 177190.Asch SE (1952) Social Psychology. Englewood Cliffs, NJ:Prentice-Hall.

Asch SE (1956) Studies of the independence and sub-mission to group pressure: I. A minority of oneagainst a unanimous majority. Psychological

Monographs70, No. 9.Atkinson RC and Suppes P (1958) An analysis of two-

person game situations in terms of statistical learningtheory.Journal of Experimental Psychology55:369378.

Bales RF (1950) Interaction Process Analysis: A Method forthe Study of Small Groups. Cambridge, MA: Addison-Wesley Press.

Bales RF, Strodtbeck FL, Mills TM and RoseboroughME (1951) Channels of communication in smallgroups.American Sociological Review16: 159184.

Bartholomew DJ (1967) Stochastic Models for SocialProcesses. New York: Wiley.

Bavaud F (2008) The endogenous analysis of flows, withapplications to migrations, social mobility and opin-ion shifts. The Journal of Mathematical Sociology32:239266.

Bavelas A (1948) A mathematical model for group struc-ture.Applied Anthropology7: 1630.

Bavelas A (1950) Communication patterns in task-ori-ented groups.Journal of the Acoustical Society of

America 22: 271282.Bavelas A and Barrett D (1951) An experimentalapproach to organizational communication. Personnel27: 366371.

Berger J and Zelditch M Jr (eds) (2002) New Directionsin Contemporary Sociological Theory. Lanham, MD:Rowman and Littlefield.

Berger J, Cohen BP, Snell JL and Zelditch M Jr (1962)Types of Formalization in Small-Group Research.Boston, MA: Houghton-Mifflin.

Bevaud F (2008) The endogenous analysis of flows, withapplications to migrations, social mobility and opin-ion shifts. The Journal of Mathematical Sociology32:239266.

Bischi G and Merlone U (2009) Global dynamics in

binary choice models with social influence. TheJournal of Mathematical Sociology33: 277302.

Blau PM (1977) Inequality and Heterogeneity: A PrimitiveTheory of Social Structure. New York: Free Press.

Bollen KA and Bauldry S (2010) A note on algebraicsolutions to identification. The Journal of

Mathematical Sociology34: 136145.Bothner MS, Haynes R, Lee W and Smith EB (2010)

When do Matthew effects occur? The Journal ofMathematical Sociology34: 80114.

Boudon R (1973)Mathematical Structures of SocialMobility. New York: Elsevier.

Boudon R (1974) Education, Opportunity, and SocialInequality. New York: John Wiley and Sons.

Buskens V and Van de Rijt A (2008) Sequential power-dependence theory. The Journal of MathematicalSociology32: 110128.

Cartwright D and Harary F (1956) Structural balance: Ageneralization of Heiders theory. Psychological Review

63: 277293.Centola D (2009) Failure in complex social networks.The Journal of Mathematical Sociology33: 6468.

Cohen BP (1958) A probability model for conformity.Sociometry21: 6981.

Cohen BP (1963) Conflict and Conformity: A ProbabilityModel and its Applications. Cambridge, MA: MITPress.

Coleman JS (1964) Introduction to MathematicalSociology. New York: The Free Press.

Coleman JS (1990) Foundations of Social Theory.Cambridge, MA: Belknap Press.

Coleman JS, Katz E and Menzel H (1966) MedicalInnovation: A Diffusion Study. Indianapolis, IN:

Bobbs-Merril.Collins R (1988) Theoretical Sociology. New York:

Harcourt Brace Jovanovich.Correll SJ and Ridgeway CL (2003) Expectation states

theory. In: Delamater J (ed.) Handbook of SocialPsychology. New York: Springer US, 2951.

Deng H and Abell P (2010) A study of local sign changeadjustment in balancing structures. The Journal of

Mathematical Sociology34: 253282.Denrell J and Shapira Z (2009) Performance sampling

and bimodal duration dependence. The Journal ofMathematical Sociology33: 3863.

Dogan G and Van Assen M (2009) Testing models of

pure exchange. The Journal of Mathematical Sociology33: 97128.Doreian P (1970)Mathematics and the Study of Social

Relations. New York: Schocken.Estes WK and Burke CJ (1955) Application of a statisti-

cal model to simple discrimination learning inhuman subjects.Journal of Experimental Psychology50: 8188.

Fararo TJ (1973)Mathematical Sociology: An Introductionto Fundamentals. New York: Wiley Interscience.

Fararo TJ (1989) The Meaning of General TheoreticalSociology: Tradition and Formalization. New York:Cambridge University Press.

Fararo TJ (1997) Reflections on mathematical sociology.

Sociological Forum 12: 73101.


12/14

12


Fararo TJ (2001) Theoretical sociology in the 20th cen-tury.Journal of Social Structure 2(2).Available at:www.cmu.edu/joss/content/articles/volume2/Fararo.html.

Fararo TJ and Skvoretz J (1984) Institutions as produc-tion systems.Journal of Mathematical Sociology10:

117182.Fararo TJ and Skvoretz J (1986) Action and institution,

network and function: The cybernetic concept ofsocial structure. Sociological Forum 1: 219250.

Fararo TJ and Skvoretz J (1987) Unification researchprograms: Integrating two structural theories.

American Journal of Sociology92: 11831209.Fararo TJ and Skvoretz J (1993). Methods and problems

of theoretical integration and the principle of adap-tively rational action. In: Berger J and Zelditch M Jr(eds) Theoretical Research Programs: Studies in TheoryGrowth. Stanford, CA: Stanford University Press,416450.

Fararo TJ and Sunshine M (1964)A Study of a BiasedFriendship Net. Syracuse, NY: Syracuse UniversityYouth Development Center and Syracuse UniversityPress.

Fioretti G (2010) A model of vacancy chains as a mecha-nism for resource allocation. The Journal of

Mathematical Sociology34: 5275.Flache A and Macy MW (2011) Small worlds and cul-

tural polarization. The Journal of MathematicalSociology35: 146176.

Fossett M (2011) Generative models of segregation:Investigating model-generated patterns of residentialsegregation by ethnicity and socioeconomic status.The Journal of Mathematical Sociology35: 114145.

Friedkin NE (2010) A multilevel event history model ofsocial diffusion:Medical Innovation revisited. The

Journal of Mathematical Sociology34: 146155.Gchter S and Thni C (2011) Micromotives,

microstructure, and macrobehavior: The case of vol-untary cooperation. The Journal of MathematicalSociology35: 2665.

Granovetter M (1978) Threshold models of collectivebehavior.American Journal of Sociology83:14201443.

Grassi R, Stefani S and Torriero A (2010) Extremalproperties of graphs and eigencentrality in trees witha given degree sequence. The Journal of Mathematical

Sociology34: 115135.Hajeeh M and Lairi S (2009) Marriage partner selectionin Kuwait: An analytical hierarchy process approach.The Journal of Mathematical Sociology33: 222240.

Heckathorn DD (2002) Development of a theory of col-lective action: From the emergence of norms toAIDS prevention and the analysis of social structure.In: Berger J and Zelditch M Jr (eds) New Directionsin Sociological Theory. Lanham, MD: Rowman andLittlefield, 79108.

Hedstrm P and Bearman P (2009) What is analyticalsociology all about? An introductory essay. In:Hedstrm P and Bearman P (eds) The OxfordHandbook of Analytical Sociology. New York: Oxford

University Press, 324.

Heider F (1944) Social perception and phenomenalcausality. Psychological Review51: 358374.

Heider F (1946) Attitudes and cognitive organization.Journal of Psychology21: 107112.

Heider F 1(958) The Psychology of Interpersonal Relations.New York: Wiley.

Helbing D, Yu W and Rauhut H (2011) Self-organiza-tion and emergence in social systems: Modeling thecoevolution of social environments and cooperativebehavior. The Journal of Mathematical Sociology35:177208.

Hevenstone D (2009) Employment intermediaries: Amodel of firm incentives. The Journal of

Mathematical Sociology33: 137.Homans GC (1950) The Human Group. New York:

Harper and Row.Hopcroft RL and Whitmeyer JM (2010) A choice model

of occupational status and fertility. The Journal ofMathematical Sociology34: 283300.

Hummon NP and Fararo TJ (1995) The emergence ofcomputational sociology.Journal of MathematicalSociology 20: 7987.

Janssen MA and Ostrom E (2006) Empircally based,agent-based models. Ecology and Society11(2): 37.Available at:www.ecologyandsociety.org/vol11/iss2/art37/.

Jasso G (2010) Linking individuals and societies. TheJournal of Mathematical Sociology34: 151.

Karlsson G (1958) Social Mechanisms: Studies inSociological Theory. Glencoe, IL: The Free Press.

Kejzar N, Nikoloski Z and Batagelj V (2008)Probabilistic inductive classes of graphs. The Journalof Mathematical Sociology32: 85109.

Kemeny JG and Snell JL (1962)Mathematical Models inthe Social Sciences. Waltham, MA: Ginn.

Kitts JA (2008) Dynamics and stability of collectiveaction norms. The Journal of Mathematical Sociology32: 142163.

Landau HG (1952) On some problems of random nets.Bulletin of Mathematical Biophysics14: 203212.

Lave CA and March JG (1975) Introduction to Models inthe Social Sciences. New York: Harper and Row.

Leavitt HJ (1949) Some effects of certain communica-tion patterns on group performance. Unpublisheddoctoral dissertation, Massachusetts Institute ofTechnology.

Leavitt HJ (1951) Some effects of communication pat-terns on group performance.Journal of AbnormalPsychology46: 3850.

Leik RK and Meeker BF (1975)Mathematical Sociology.Englewood Cliffs, NJ: Prentice-Hall.

Macy MW and Skvoretz J (1998) The evolution of trustand cooperation between strangers: A computationalmodel.American Sociological Review63: 638660.

Mayer T (2008) The transformation of American politi-cal space 19822002. The Journal of MathematicalSociology32: 156.

Menicucci D and Sacco PL (2009) Evolutionary selec-tion of socially sensitive preferences in randommatching environments. The Journal of Mathematical

Sociology33: 241276.


13/14

13


Mic JC, Caselles A, Soler D, Sanz T and Martnez E(2008) A side-by-side single sex age-structuredhuman population dynamic model: Exact solutionand model validation. The Journal of MathematicalSociology32: 285321.

Miller JH and Page SE (2007) Complex Adaptive Systems:

An Introduction to Computational Models of SocialLife. Princeton, NJ: Princeton University Press.

Molchanov S and Whitmeyer JM (2010) Two Markovmodels of the spread of rumors. The Journal of

Mathematical Sociology34: 157166.Montgomery JD (2009) Balance theory with incomplete

awareness. The Journal of Mathematical Sociology33:6996.

Mrvar A and Doreian P (2009) Partitioning signed two-mode networks. The Journal of MathematicalSociology33: 196221.

Opp K (2011) Modeling micromacro relationships:Problems and solutions. The Journal of Mathematical

Sociology35: 209234.Plos L, Hannan MT and Hsu G (2010) Modalities insociological arguments. The Journal of MathematicalSociology34: 201238.

Rapoport A (1951a) Nets with distance bias. Bulletin ofMathematical Biophysics 13: 8591.

Rapoport A (1951b) The probability distribution of dis-tinct hits on closely packed targets. Bulletin of

Mathematical Biophysics 13: 133137.Rapoport A (1953a) Spread of information through a

population with socio-structural bias: I. Assumptionof transitivity. Bulletin of Mathematical Biophysics15:523533.

Rapoport A (1953b) Spread of information through a

population with socio-structural bias: II. Variousmodels with partial transitivity. Bulletin of

Mathematical Biophysics15: 535546.Rapoport A (1953c) Spread of information through a

population with socio-structural bias: III. Suggestedexperimental procedures. Bulletin of MathematicalBiophysics16: 7581.

Rapoport A (1957) A contribution to the theory of ran-dom and biased nets. Bulletin of MathematicalBiophysics19: 257271.

Rapoport A (1958) Nets with reciprocity bias. Bulletin ofMathematical Biophysics 20: 191201.

Rapoport A (1963) Mathematical models of social inter-

action. In: Luce RD, Bush RR and Galanter E (eds)Handbook of Mathematical Psychology, Volume 2. NewYork: Wiley, 493579.

Rapoport A and Horvath WJ (1961) A study of a largesociogram. Behavioral Science6: 279291.

Rapoport A and Solomonoff R (1951) Connectivity ofrandom nets. Bulletin of Mathematical Biophysics 13:107117.

Rashevshky N (1939a) Studies in mathematical theoryof human relations. Psychometrika4: 221239.

Rashevshky N (1939b) Studies in mathematical theoryof human relations, II. Psychometrika 4: 283209.

Rashevshky N (1940a) Studies in mathematical theoryof human relations, III. Psychometrika5: 203210.

Rashevshky N (1940b) Studies in mathematical theory

of human relations, IV. Psychometrika5: 299303.Rashevshky N (1941) Note on the mathematical theory

of interaction of social classes. Psychometrika6:4347.

Rashevshky N (1942) Contributions to the mathemati-cal theory of human relations, V. Psychometrika7:

117134.Raub W, Buskens V and Van Assen MALM (2011)

Micromacro links and microfoundations in sociolo-gy. The Journal of Mathematical Sociology35: 125.

Robinson DT, Smith-Lovin L and Wisecup AK (2006)Affect control theory. In: Stets JE and Turner JH(eds) Handbook of the Sociology of Emotions. NewYork: Springer US, 179202.

Schelling T (1971) Dynamic models of segregation. TheJournal of Mathematical Sociology1: 143186.

Simkin MV and Roychowdhury VP (2008) Theory ofaces: High score by skill or luck? The Journal of

Mathematical Sociology32: 129141.

Simon HA (1952) A formal theory of interaction insmall groups.American Sociological Review17:202212.

Simon HA (1957)Models of Man. New York: JohnWiley.

Singer H (2010) SEM modeling with singular momentmatrices part I: ML-estimation of time series. The

Journal of Mathematical Sociology34: 301320.Skvoretz J (1981) Extending expectation states theory:

Comparative status models of participation in n per-son groups. Social Forces59: 752770.

Skvoretz J (1983) Salience, heterogeneity and consolida-tion of parameters: Civilizing Blaus primitive theory.

American Sociological Review48: 360375.

Skvoretz J (1988) Models of participation in status-dif-ferentiated groups. Social Psychology Quarterly51:4357.

Skvoretz J (1990) Biased net theory: Approximations,simulations, and observations. Social Networks12:217238.

Skvoretz J (1998) Theoretical models: Sociologys miss-ing links. In: Sica A (ed.) What is Social Theory? ThePhilosophical Debates. Oxford: Blackwell, 238252.

Skvoretz J and Fararo TJ (1995) The evolution of sys-tems of social interaction. Current Perspectives inSocial Theory15: 275299.

Skvoretz J and Fararo TJ (1996) Status and participation

in task groups: A dynamic network model.AmericanJournal of Sociology101: 13661414.Smith SL (1950) Communication pattern and the adapt-

ability of task-oriented groups: An experimentalstudy. Group Networks Laboratory, ResearchLaboratory of Electronics, Massachusetts Institute ofTechnology, Cambridge, MA.

Smith-Lovin L, Skvoretz J and Hudson CG (1986)Status and participation in six-person groups: A testof Skvoretzs comparative status model. Social Forces64: 9921005.

Snijders TAB (2010) Conditional marginalization forexponential random graph models. The Journal of

Mathematical Sociology34: 239252.

Srenson AB (1977) The structure of inequality and the


14/14

14


process of attainment.American Sociological Review42: 965978.

Srenson AB (2009) Statistical models and mechanismsof social processes. In: Hedstrm P and Wittrock B(eds) Frontiers of Sociology. Boston, MA: Brill,369399.

Toulmin S (1953) The Philosophy of Science. London:Hutchinson.

Tutic A and Liebe U (2009) A theory of status-mediatedinequity aversion. The Journal of MathematicalSociology33: 157195.

Van de Rijt A (2011) The micromacro link for the the-ory of structural balance. The Journal of MathematicalSociology35: 94113.

Von Neumann J and Morgenstern O (1947) Theory ofGames and Economic Behavior. Princeton, NJ:Princeton University Press.

White HC (1963)An Anatomy of Kinship. EnglewoodCliffs, NJ: Prentice-Hall.

White HC (1970) Chains of Opportunity. Cambridge,MA: Harvard University Press.Willer D and Emanuelson P (2008) Testing ten theories.

The Journal of Mathematical Sociology32: 165203.

Wyburn J and Hayward J (2008) The future of bilin-gualism: An application of the Baggs and Freedmanmodel. The Journal of Mathematical Sociology32:267284.

Wyburn J and Hayward J (2010) A model of language-group interaction and evolution including language

acquisition planning. The Journal of MathematicalSociology34: 167200.

Yamaguchi K (2011) Population heterogeneity andbetween-group substitutability and complementarityof social actions. The Journal of MathematicalSociology35: 6693.

Young HP (1998) Individual Strategy and SocialStructure: An Evolutionary Theory of SocialInstitutions. Princeton, NJ: Princeton UniversityPress.

Ziberna A (2008) Direct and indirect approaches toblockmodeling of valued networks in terms of regularequivalence. The Journal of Mathematical Sociology

32: 5784.

rsum Dans cet article, il dcrit lhistoire, les prsuppositions, les dveloppements en cours, les

preuves y les perspectives de la sociologie mathmatique. Les tches de recherche courantes dans cet

domaine depuis la dernire guerre jusquau prsent sont identifies. Il donne des arguments en faveur delimportance de la sociologie mathmatique pour la future de la thorie sociologique.

mots-cls construction des modles procd sociologie mathmatique structure thoriesociologique

resumen En este artculo, se describe la historia, las presuposiciones, el desarrollo en curso, los desafos

y las perspectivas para la sociologa matemtica. Los hilos de investigacin comunes en este campo

despus de la segunda guerra mundial hasta el presente son identificados. Se dan argumentos a favor de

la importancia de la sociologa matemtica para el futuro de la teora sociolgica.

palabras clave construccin de modelos estructura proceso sociologa matemtica teorasociolgica

John Skvoretz is Professor of Sociology at the University of South Florida, CarolinaDistinguished Professor Emeritus at the University of South Carolina and President of theInternational Network for Social Network Analysis. Research interests include biased net theo-ry models for social network analysis and e-state structural models for power and prestige orders

in focused interaction settings. [email: [email protected]]

Thomas J Fararo is Emeritus Distinguished Service Professor of Sociology at the Universityof Pittsburgh. His articles and books deal with theoretical and mathematical sociology andincludeMathematical Sociology(1973).

Skvoretz & Fararo (2011) - Mathematical sociology

Documents

Transcript of Skvoretz & Fararo (2011) - Mathematical sociology