Social: Theories and Models - Foundations for Designing User-Centered Systems - page 267

9.4.5 Computational Models of Social Behavior
As social models get more complex, it becomes necessary to use computers to
implement them and their predictions. Analytic models simulate the social pro-
cesses but without modeling the individual transactions in detail. Process models
model the information processing that is performed and they will typically also
model the interactions between agents.
9.4.5.1 Analytic Models
Analytic models provide a description of a social process. They may include
equations for how often people meet, or how often and what they communicate.
These equations are used to simulate how the social process unfolds.
As an example, a group of researchers modeled the transmission of a possible
influenza pandemic (Halloran et al. 2008 ). It can be viewed as a social model (as
well as a epidemiological model) because it primarily models social interactions
between people at schools, work, home, and neighborhoods. It uses equations to
model how often people interact, how often a disease would be transmitted, and
the impact of medical interventions to stop the transmission of the disease.
Importantly, these simulations explore how well public health interventions would
be followed and what would be the impact of these interventions on the spread of
the disease. The authors argue that the results are not accurate enough to be
predictions, but are useful for framing the discussion of how to handle pandemics,
and which interventions to consider. For example, closing schools is one that is
both easy to follow and greatly helps reduce the spread of disease. These simu-
lations do not cover the details, but they present a broad description of a large
social system. The Halloran et al. ( 2008 ) work modeled the population of greater
Chicago of 8.6 million people. This is one of the largest social simulation models
we know.
9.4.5.2 Process Models
As social processes are being better formulated, there has been an increase in
modeling them using information processing models. These models simulate
smaller groups, ranging from 2 to 100 people.
Current examples include work on optimizing the structures of groups to
increase performance (e.g., Levchuk et al. 2002 ), creating models for use as col-
leagues and adversaries (Jones et al. 1999 ; Tambe et al. 1995 ; Zachary et al. 2001 ),
and fundamental work on understanding how changes in group processes and
individual behavior lead to changes in group performance (e.g., Carley 1996 ). This
is an exciting emerging field, and further examples are available from conferences