Do the SBE Sciences Advance NSF’s Mission? Advancing Progress in Science with Innovative Theories, Methods, and Tools
In addition to contributing useful knowledge, the social, behavioral, and economic (SBE) sciences have produced a variety of theories, methods, and tools that are used to predict and explain behavior, identify problems, track them over time, and inform decision making. Social and behavioral research methods also produce rich datasets that are used to test hypotheses about behavior in addition to what can be learned through intuition and experience.
Some of these methods, such as polling or those used to determine the effects of interventions or policies, have become such a part of daily life that it is easy to forget their roots in fundamental SBE research. That research has addressed the complex problems that can complicate the generation of reliable survey data and has developed sophisticated methods for establishing cause-and-effect relationships. In particular, SBE theories and methods have led to a better understanding of conflict and cooperation, and to algorithms that are used for organ-donation matching (see Box 2), predicting international conflict, and modeling crowd behavior. Some of these methods have emerged and advanced with ever-increasing amounts of data and as SBE research addresses new, complex challenges, such as the spread of terrorism.
Groundbreaking Theories of Human Behavior
Although the concept of a theory can be abstract, it has an important role to play in explaining individual and social behavior. Three Nobel Prize-winning bodies of research exemplify some pioneering theories in SBE that have had wide-ranging practical applications.
Game theory explains how individuals reach agreements with one another through conflict or cooperation. It has been enormously influential in the SBE sciences: 11 game theorists have won the Nobel Prize in economics, many of whom received support from the National Science Foundation (NSF). It has been applied to labor markets, industrial organization, arms reduction negotiations, and the provision of public goods (see figure in Box 2).
For a different example, NSF-supported research found an important exception to a prominent social science theory, “the tragedy of the commons.” This long-held theory posits that individuals will compete to exploit a public resource, ruining the resource for everyone. One of the implications of this theory
was that only a government or governments could limit such individual competition. The new research showed how individuals around the world have cooperated to share resources and develop governance strategies for shared resources that are often better than top-down, government-driven solutions. The principles developed from this work are relevant to current debates about the use of a wide range of collective resources, including the Internet and knowledge in the public domain (e.g., Wikipedia).82
Even more recently, the theory of “nudging” describes an approach to policy design that accounts for systematic, irrational tendencies in people’s behaviors and decision making,83,84 building on Nobel Prize-winning work on the psychology of decision making that was funded by NSF.85 In essence, nudging involves small changes in how choices or options are presented. These near-costless interventions can have remarkable effects. Examples that have yielded demonstrable results include changing the default on organ donation or retirement saving decisions from opt in to opt out (so that no action is required to be in the donation or savings pool) and notifying consumers about their neighbors’ energy consumption.86 The individual and societal benefits of these interventions have been so large that both the U.K. and U.S. governments have established offices dedicated to implementing nudging approaches to a wide range of government programs, with demonstrated policy results.87
Understanding How People and Their Circumstances Change Over Time
Studies of education, the labor force, and aging that follow people over long periods of time (longitudinal studies) provide important information about the factors that lead to more or less positive life outcomes. Longitudinal research studies changes in behavior over time and can, in some cases, provide understanding of the long-range outcomes of an intervention. One example is the Health and Retirement Study (HRS) of people aged 50 and older—the premier source of information on the nation’s aging population. This large body of data from multiple sciences can help address a wide range of important questions about aging, such as how work, exercise, income, and other factors in middle age, affect circumstances in old age. Many countries around the world have modeled their own surveys after the HRS to understand their own aging populations. This type of research is made possible by access to large datasets—including data from federal statistical agencies and state administrative data systems—and helps to provide a more complete understanding of people and their well-being over time.
Simulations, Modeling, and Forecasting
Models and simulations that apply theories and principles of behavior can be used to develop and test policy ideas and interventions quickly, inexpensively, and safely. One such model of crowd behavior and suicide bombers yielded the surprising finding that remote sensing of suicide bombers and sounding alarms to notify crowds of their presence could actually expose more people to the blast and shrapnel and increase the number of casualties.88 Models that combine approaches from
statistics and demography to more accurately forecast life expectancy and mortality are now widely used around the world by national statistical agencies, public-sector pension agencies, the United Nations, and private-sector providers of life insurance and annuities.89,90 They are also used by the new longevity swap industry, which helps institutions manage the risks of unknown future costs, such as those for pension plans.91
Other examples of NSF-funded models that incorporate SBE research include algorithms that help prevent terrorist attacks92,93,94 and support sequential decision making to maximize the detection of illicit and hazardous cargo at U.S. ports. A model of pedestrian movement and crowd behavior in dense urban environments95,96 reveals the rapid exchange of nonverbal information in crowds and shows how the actions of a single individual can shape the dynamics of an entire crowd. Finally, forecasting methods developed by SBE research can now be applied to anonymized and aggregated datasets (“big data”) generated by search queries—user browsing logs and social media posts—to predict a wide range of collective human behaviors, such as consumer demand, unemployment claims, and mortgage default rates.97,98,99
New Methods of Collecting and Analyzing Data
More and more use is being made of data collected from surveys on the Web and smartphones. However, data collected in this manner do not satisfy a key requirement of standard survey methodology, which is that every member of a given population has to be equally likely to be surveyed. Thus, the data from these surveys are in some sense biased and nonrepresentative. NSF-sponsored research has begun to develop new models that use sophisticated statistical techniques for converting these inherently biased samples into unbiased estimates.100,101,102 These new methods could dramatically increase the scale, scope, and frequency of obtaining information from survey data by using real-time measures that draw on millions of responses to measure, for example, consumer or business activity, worker productivity, community well-being, or disease caseloads. With more development, these applications have the potential to inform all federal agencies that collect data, including the U.S. Census Bureau, the Bureau of Labor Statistics, the National Center for Health Statistics, the U.S. Office of Management and Budget, and the Centers for Disease Control and Prevention, as well as advance the work of industry and business.
As another example, respondent-driven sampling (also known as network sampling) is a relatively new method that allows researchers to collect important information about “hidden” or hard-to-reach groups, such as those at the greatest risk of infection from HIV/AIDS.103 This approach relies on members of those groups to recruit each other for the survey. Because data collected in this way are not representative of the total population of infected people, the method also includes statistical procedures for making the data more representative. NSF has supported the further development of respondent-driven sampling,104,105 and its use has been supported by the U.S. government through the U.S. President’s Emergency Plan for AIDS Relief.106
