Computational Sociology

Computational sociology is a specialty area in sociology that has gained momentum through advances in computer technology over the past few decades. In computational sociology, techniques such as mathematical modeling, computer simulation, and agent-based modeling are used to analyze social phenomena to better understand underlying factors and predict future behavior. As opposed to most conventional statistical methods, these tools do not assume that there is an underlying linear relationship between variables. In addition, the tools of computational sociology allow researchers to investigate the complex relationships often seen in real world behavior without the limitations posed by more conventional statistical methods, ethical considerations, or practicalities of real world experimental design. As the full potential of the tools of computational sociology are increasingly explored and realized, researchers and theorists will be better able to understand, explain, and predict social behavior.

Keywords Chaos Theory; Computer Simulation; Ethics; Experiment; Hypothesis; Model; Society; Statistics; Subject; Survey; Variable

Overview

After a long day in front of the computer doing research and writing, there is little that I enjoy more than sitting in front of the computer for even more hours playing computer games. Solitaire and Soduku are fine, of course, the former improving hand-eye coordination and the latter helping reasoning processes. However, what I really enjoy is a good role playing game or simulation. I have happily chased electronic dwarfs down labyrinthine corridors in dark caves for hours, searching for a golden plover's egg. I have built cities and refought ancient battles, putting in different parameters just to see their effects on societal stability or battle outcomes, all from the comfort of my desk chair. I play these games for fun, of course, and although they feed my curiosity, they have little to do with my own areas of scientific inquiry.

Social scientists use similar tools to explore the effects of different variables and parameters on society. This application of mathematical modeling and computer simulation to the analysis of real world social phenomena is called computational sociology.

Traditional statistical methods can be very useful for analyzing data and interpreting real world phenomena. However, like all tools, they have their limitations. Most conventional statistics that are used to analyze experimental data assume that there is a linear relationship between variables and that the effect on the dependent variable is proportional to the changes in the independent variable(s). However, life is infinitely complex, and it cannot be assumed that there is a linear relationship between real world variables. For example, if one were studying the relationship between certain kinds of stimuli and how these affected the anger response of a subject, it might be assumed that the more irritating stimuli to which the subject was exposed, the more s/he would exhibit symptoms of anger. However, although this linear assumption might be true for some people, many people are able to control their anger — and all its symptoms — until a certain threshold is reached, at which point they display their anger. This nonlinear hypothesis would be more difficult to test using conventional statistical methods and it can be difficult to articulate a set of equations that can be used to predict the characteristics of people's responses to anger stimuli (or other system).

Computer Simulations

The field of computational sociology came into being in the closing decades of the twentieth century. This field of endeavor, including the tools of computer simulation and mathematical modeling, has great potential for helping social scientists better understand and predict social processes. One of the most frequently used tools of computational sociology is computer simulation. This is a particular type of modeling. Computer simulations do not make assumptions about the linear relationship between variables. Computer simulations are representations of real world conditions using a computer program to develop a mathematical model that imitates the internal processes of a situation as well as the end results. The situations modeled in computer simulations tend to be complex in nature. In the social sciences, computer simulations are used to analyze social phenomena in order to better understand and predict behavior in society. They can be very helpful in understanding social behavior because it is often difficult or impossible to manipulate variables in real world situations in order to see what the result is.

Professional ethics for working with human subjects requires that the researcher first does no harm to the subjects either physically or psychologically. However, the manipulation of variables related to the questions that many sociologists are interested in often brings with it the potential for harming the subject. For example, although one could easily develop a controlled research paradigm to investigate factors relating to an individual's successful readjustment after the death of a spouse or the coping mechanisms of different kinds of adults to abuse as a child, it would be highly unethical (as well as illegal) to manipulate the variable of whether or not one's spouse dies or whether or not one experiences child abuse. Social scientists, therefore, often perform hypothetical experiments using computer simulations in which they manipulate the value of variables in a mathematical model of a real world situation in order to observe the effect on the results.

Chaos Theory

Computer simulation frequently uses chaos theory to examine how complex behavior can result from relatively simple activities and antecedents. Chaos theory is a variation of nonlinear systems theory used in modeling for the physical, life, and social sciences. Chaos theory attempts to explain such things as how small changes may result in unexpectedly large consequences or why multiple patterns of social behavior may share general tendencies yet not have the same result. Computer simulations — just like mathematical models — use variables related to a theoretical model of the real world.

Applications

Examples of Computer Simulations in Computational Sociology

Scientists manipulate the values of variables in order to examine the effects of these changes on the results. Gilbert and Troitzsch (2005) offer a simple example of the use of simulation in the social sciences. This example is based on the way that different people choose a marriage partner. One way to approach this problem is for one to continue to date and look for a mate until one is found that meets all of the individual's ideal characteristics. Other people, however, only participate in dating behavior until they find someone who is "good enough." To better understand individuals' mate-searching behavior, the individual interviewed them or asked them to complete a survey instrument. However, such an approach is unlikely to be successful because most people do not have a conscious strategy for dating, may not be able to articulate the strategy if they do have one, or be unwilling to share that strategy with a stranger.

Therefore, a social scientist might develop a computer simulation that includes plausible assumptions regarding how various individuals choose mates. This computer model could be used as a tool to aid in the development of the theory concerning the ways in which individuals choose their mates. Because a computer program is more precise than a written description, it allows easier and more controlled manipulation of variables so that the theory can't be refined. Once the theory has been formalized into a computer program, it can be run in order to observe how the simulation behaves. For example, the research in this study could develop a population of simulated suitors, each of which could be assigned a suitability score at random. This simulated individual looking for a mate (referred to as the "agent") could "date" the potential suitors in random order. After each simulated date, the agent would make a decision as to whether or not to break up or settle down with the simulated suitor. As in real life, each decision would be made without knowing the relative suitability of the other members of the population who had not yet been "dated." This simulation could be repeated numerous times with different random suitability scores and dating order each time. Although the output of any one run of the simulation would be of minimal use to the scientist, the average score or pattern over a large number of runs could help the scientists better understand dating and mating behavior.

There are a number of uses of computer modeling and simulation for use in sociology. First, these tools allow one to better understand various features of the social world. In particular, modeling and simulation allows social scientists to better understand behavior that is difficult to discover directly (e.g., dating strategies). Another use of simulation is for purposes of prediction. If the model can be developed that adequately and accurately simulates real world behavior, it can be used to predict behavior in the future given a set of parameters. This use of modeling and simulation is employed frequently in demographic research. For example, a computer model could be used to predict how the size and age structure of a population might change over a given period of time in the future. To do this, the model might incorporate data such as age-specific fertility and mortality rates. Another use of computer simulation and modeling in sociology is as a tool to help social scientists to discover information about the actions and consequences of human behavior in social settings and to formalize these into testable theories.

Stockard (2000) describes the use of a computer simulation in the conduct of a hypothetical experiment to investigate the effects of racial and ethnic segregation on the development of the underclass and the concentration of poverty within urban centers. Massey and Denton, who designed the simulation, constructed four imaginary cities, each of which had the same size population and which were each divided into 16 distinct neighborhoods. In order to reflect the racial and ethnic composition of the United States, they made the assumption that each city be comprised of 25 percent African Americans. It was further assumed for purposes of the simulation that the poverty rate among African Americans was 20 percent and among European Americans was 10 percent. The hypothetical cities were designed with different amounts of residential segregation based on race and ethnicity. Hypothetical City 1 was designed with no segregation (i.e., everyone lived in a neighborhood that comprised 25 percent African Americans). Hypothetical City 2 was designed to have low racial and ethnic segregation (i.e., 25 percent of the neighborhoods had no segregation and the other 15 neighborhoods were equally segregated). Hypothetical City 3 had a high level of segregation and hypothetical City 4 was completely segregated (i.e., all African Americans occupied 25 percent of the neighborhoods and all European Americans occupied the other 75 percent of the neighborhoods) (Stockard, 2000).

A computer simulation was then used to determine the characteristics of populations living in different neighborhoods in each of these hypothetical cities. In hypothetical City 1 (no segregation situation), the poverty rate was 12.5 percent. However, as racial-ethnic segregation increased in the other three cities, the simulated African American population in the city tended to live together in areas with a higher concentration of poverty whereas the European Americans tended to live together in areas with lower concentration of poverty. For example, in City 3, the average simulated African American family lived in a neighborhood with a 15 percent poverty rate whereas the average European-American family lived in a neighborhood with an average 11.7 percent poverty rate. In City 4 (total segregation), the simulated African American families lived in neighborhoods with twice the poverty rate of the simulated European-American families. Although this simulation did not take into account the factor of social stratification, it does demonstrate how economic down turns can increase the relative poverty in ghetto areas. The simulation also showed how social structure in cities (i.e., segregation by neighborhood) can increase the probability that African Americans will live in areas with high poverty levels than will European Americans (Stockard, 2000).

Agent-based Models

Computational sociology typically employs the tools of computer simulation in order to better understand social behavior in groups and to predict future behavior. Another approach to modeling that is increasingly being used in other social sciences is agent-based models. Sociology, however, has not yet fully embraced this approach to modeling despite the potential it has shown in other social and behavioral sciences (Macy & Willer, 2002). As opposed to the types of simulations discussed above, agent-based models can help researchers and theorists discover the causal mechanisms that underlie the associations uncovered by other statistical methods. In agent-based models of human social interaction, one models the interaction between members of a simulated population rather than the population as a whole or the individual members of the population.

As opposed to traditional approaches in sociology that look at social life as a hierarchical system of institutions and norms that affect behavior, the agent-based approach to modeling takes into account the fact that there is great complexity in social behavior and that the behavior of humans in groups can be nonlinear, path-dependent, and self-organizing. In addition, agent-based models of interaction can help theorists bridge the gap between the micro and macro levels. Agent-based models can help the researcher discover previously undiscovered theoretical ideas that often have broader application than those arising from other modeling techniques. Further, agent-based modeling is an experimental method that allows social scientists to test theories, not just explore them. Using agent-based modeling, one can apply careful, systematic scientific method that will advance the state of the art in sociology. Agent-based models also allow researchers to manipulate structural conditions in order to test not only microsociological theories but macrosociological theories as well (Macy & Willer, 2002).

Conclusion

One of the difficulties in hypothesis testing and theory building in sociology has always been the fact that real world data can be extremely complicated with many variables, whose relationships are not only complex but nonlinear. In such situations, even if one were able to conduct an experiment to examine the complex relationship between multiple variables, the results might not be replicable. Further, as opposed to the physical sciences, ethical considerations frequently limit the degree to which variables can be manipulated in the field. In addition, even when survey instruments can be developed to gather the data desired for hypothesis testing, subjects may be unable or unwilling to articulate them. Fortunately, computer technology has now advanced to the point where it is possible to relatively easily develop models of complex real world situations and repeatedly run simulations to better understand the aggregate affect of variables. In addition, the tool of agent-based modeling is beginning to come into its own and should become an important tool for use in sociological research and theory building. Computational sociology is an exciting field with great potential for advancing the state of knowledge about sociology and the actions of people in society.

Terms & Concepts

Chaos Theory: A variation of nonlinear systems theory used in modeling for the physical, life, and social sciences. Chaos theory attempts to explain such things as how small changes can result in unexpectedly large consequences or why multiple patterns of social behavior may share general tendencies yet not have the same result.

Computer Simulation: The representation of real world conditions using a computer program to develop a mathematical model that imitates the internal processes of a situation as well as the end results of these processes. The situations modeled in computer simulations tend to be complex in nature. In the social sciences, computer simulations are used to analyze social phenomena in order to better understand and predict behavior in society.

Demographic Data: Statistical information about a given subset of the human population such as persons living in a particular area, shopping at an area mall, or subscribing to a local newspaper. Demographic data might include such information as age, gender, or income distribution.

Ethics: In scientific research, a code of moral conduct regarding the treatment of research subjects that is subscribed to by the members of a professional community. Many professional groups had a specific written code of ethics that sets standards and principles for professional conduct and the treatment of research subjects.

Ethnicity: A social construct used to describe a relatively large group of people that shares a common and distinctive culture such as a common history, language, religion, norms, practices, and customs. Although members of an ethnic group may be biologically related, ethnicity is not the same as race.

Experiment: A situation under the control of a researcher in which an experimental condition (independent variable) is manipulated and the effect on the experimental subject (dependent variable) is measured. Most experiments are designed using the principles of the scientific method and are statistically analyzed to determined whether or not the results are statistically significant.

Hypothesis: An empirically-testable declaration that certain variables and their corresponding measure are related in a specific way proposed by a theory.

Model: A representation of a situation, system, or subsystem. Conceptual models are mental images that describe the situation or system. Mathematical or computer models are mathematical representations of the system or situation being studied.

Population: The entire group of subjects belonging to a certain category (e.g., all women between the ages of 18 and 27; all dry cleaning businesses; all college students).

Race: A social construct that is used to define a subgroup of the human population that has common physical characteristics, ancestry, or language. Racial groups are often neither objectively defined nor homogenous, and racial categories may differ from culture to culture.

Society: A distinct group of people who live within the same territory, share a common culture and way of life, and are relatively independent from people outside the group. Society includes systems of social interactions that govern both culture and social organization.

Statistics: A branch of mathematics that deals with the analysis and interpretation of data. Mathematical statistics provides the theoretical underpinnings for various applied statistical disciplines, including business statistics, in which data are analyzed to find answers to quantifiable questions. Applied statistics uses these techniques to solve real world problems.

Subject: A participant in a research study or experiment whose responses are observed, recorded, and analyzed.

Survey: (a) A data collection instrument used to acquire information on the opinions, attitudes, or reactions of people; (b) a research study in which members of a selected sample are asked questions concerning their opinions, attitudes, or reactions are gathered using a survey instrument or questionnaire for purposes of scientific analysis; typically the results of this analysis are used to extrapolate the findings from the sample to the underlying population; (c) to conduct a survey on a sample.

Variable: An object in a research study that can have more than one value. Independent variables are stimuli that are manipulated in order to determine their effect on the dependent variables (response). Extraneous variables are variables that affect the response but that are not related to the question under investigation in the study.

Bibliography

Boden, M. A. (1994). Artificial intelligence and social forecasting. Journal of Mathematical Sociology, 9 , 340-355. Retrieved 15 September 2008 from EBSCO Online Database Academic Search Premier http://search.ebscohost.com/login.aspx?direct=true&db=aph&AN=9669762&site=ehost-live

Cioffi-Revilla, C. (2010). A methodology for complex social simulations. Journal of Artificial Societies & Social Simulation, 13, 12. Retrieved October 25, 2013, from EBSCO Online Database Academic Search Complete http://search.ebscohost.com/login.aspx?direct=true&db=sih&AN=51671069

Gilbert, N., & Troitzsch, K. G. (2001). Simulation and social science. In Simulation for the Social Scientist (2nd ed.). Maidenhead, Berkshire, UK: Open University Press/McGraw-Hill Education, 1-14. Retrieved 17 September 2008 from http://cress.soc.surrey.ac.uk/s4ss/S4SS-sample-chapter.pdf

Hummon, N. P. & Fararo, T. J. (1995). The emergence of computational sociology. Journal of Mathematical Sociology, 20 (2/3), 79-87. Retrieved 15 September 2008 from EBSCO Online Database Academic Search Premier http://search.ebscohost.com/login.aspx?direct=true&db=aph&AN=15371955&site=ehost-live

Kroska, A., & Harkness, S. K. (2011). Coping with the stigma of mental illness: Empirically-grounded hypotheses from computer simulations. Social Forces, 89, 1315-1339. Retrieved October 25, 2013, from EBSCO Online Database Academic Search Complete http://search.ebscohost.com/login.aspx?direct=true&db=sih&AN=60914645

Macy, M. W., & Willer, R. (2002). From factors to actors: Computational sociology and agent-based modeling. Annual Review of Sociology, 28 , 143-166. Retrieved 15 September 2008 from EBSCO Online Database Academic Search Premier http://search.ebscohost.com/login.aspx?direct=true&db=aph&AN=6969516&site=ehost-live

Salgado, M., & Gilbert, N. (2013). Emergence and communication in computational sociology. Journal for the Theory of Social Behaviour, 43, 87-110. Retrieved October 25, 2013, from EBSCO Online Database Academic Search Complete http://search.ebscohost.com/login.aspx?direct=true&db=sih&AN=86026116

Stockard, J. (2000). Sociology: Discovering society (2nd ed.). Belmont, CA: Wadsworth/Thomson Learning.

Sutcliffe, A., & Wang, D. (2012). Computational modelling of trust and social relationships. Journal of Artificial Societies & Social Simulation, 15, 8. Retrieved October 25, 2013, from EBSCO Online Database Academic Search Complete http://search.ebscohost.com/login.aspx?direct=true&db=sih&AN=77870974

Suggested Reading

Brent, E., Thompson, A., & Mirielli, E. (1995). Disambiguating verbal comments in social interaction: A computer model of meaning. Journal of Mathematical Sociology, 20 (2/3), 109-125. Retrieved 15 September 2008 from EBSCO Online Database Academic Search Premier http://search.ebscohost.com/login.aspx?direct=true&db=aph&AN=15372019&site=ehost-live

Dukstra, W., Draisma, S., & Van der Zouwen, J. (1995). Simulative response behavior in sociological survey interviews. Journal of Mathematical Sociology, 20 (2/3), 127-144. . Retrieved 15 September 2008 from EBSCO Online Database Academic Search Premier http://search.ebscohost.com/login.aspx?direct=true&db=aph&AN=15372051&site=ehost-live

Hegselmann, R. (2012). Thomas C. Schelling and the computer: Some notes on Schelling's essay "On letting a computer help with the work". Journal of Artificial Societies & Social Simulation, 15, 13. Retrieved October 25, 2013, from EBSCO Online Database Academic Search Complete http://search.ebscohost.com/login.aspx?direct=true&db=sih&AN=89459335

Klüver, J., Schmidt, J., & Stoica, C. (2005). The emergence of social order by processes of typifying: A computational model. Journal of Mathematical Sociology, 29 , 155-176. Retrieved 15 September 2008 from EBSCO Online Database Academic Search Premier http://search.ebscohost.com/login.aspx?direct=true&db=aph&AN=16968162&site=ehost-live

Sallach, D. L. (1972). Systems analysis and sociological theory. Sociological Focus, 5 , 54-60. Retrieved 15 September 2008 from EBSCO Online Database SocINDEX with Full Text http://search.ebscohost.com/login.aspx?direct=true&db=sih&AN=14643719&site=ehost-live

Squazzoni, F. (2012). Agent-based computational sociology. Hoboken, NJ: Wiley & Sons.

Essay by Ruth A. Wienclaw, Ph.D.

Dr. Ruth A. Wienclaw holds a Doctorate in Industrial/Organizational Psychology with a specialization in Organization Development from the University of Memphis. She is the owner of a small business that works with organizations in both the public and private sectors, consulting on matters of strategic planning, training, and human/systems integration.