Agent-based modelling for social research- Block 2: Data Quality, Model Analysis and Psychological Experiments
Presenter(s): Jakub Bijak, Sarah Nurse, Jason Hilton, Toby Prike
Data Quality Assessment
In the video, Dr Sarah Nurse and Prof. Jakub Bijak use an example of international migration to discuss some of the more general issues with social data. It will focus on the limitations of measurement, but also present some ways in which we can overcome, or at least acknowledge these limitations. Finally, providing practical advice on how the imperfect data can be used in modelling.
Model Analysis and Uncertainty Quantification
In the video, Dr Jason Hilton and Prof. Jakub Bijak introduce the basic concepts related to the design of experiments used to help understand the behaviour of complex computational models, such as agent-based models. It looks at the relations between inputs and outputs of models, types of experimental design, and methods of analysing the results of simulation experiments. This video looks closely at the statistical emulators – or metamodels – and methods for uncertainty and sensitivity analysis.
In this video, Dr Toby Prike talks about increasing the psychological realism of agent-based models. In other words, making the agents within the models behave more like real humans, sharing examples of his research findings that can potentially be applied to agent-based models to make them more psychologically realistic. Please note- This video does not include any slides.
About the author
Jakub Bijak is Professor of Social Demography and currently a Joint Head of Department of Social Statistics and Demography at the University of Southampton, UK. At present, he leads an ERC funded project on Bayesian Agent-Based Population Studies and a H2020 project QuantMig: Quantifying Migration Scenarios for Better Policy. His email is firstname.lastname@example.org.
Sarah Nurse is a Research Fellow in the Department of Social Statistics and Demography at the University of Southampton. Her main research interests include migration, with a particular focus on forced migration and asylum, evaluation of migration data quality, and the demography of conflict and violence. Her email address is email@example.com.
Jason Hilton is Lecturer in Social Statistics and Data Science in the Department of Social Statistics and Demography at the University of Southampton. His research focuses on probabilistic projection of demographic processes, and on the management of uncertainty in agent-based demographic models. His email address is firstname.lastname@example.org.
Toby Prike is a Research Fellow in the Department of Social Statistics and Demography at the University of Southampton.
He holds a PhD in Psychology from Flinders University, Australia. His research looks at non-evidence based beliefs, probabilistic
reasoning, cognitive biases, and decision making under uncertainty. His email address is email@example.com.
- Published on: 27th April, 2021
- Event hosted by: University of Southampton
- Keywords: Agent-based modelling | Computational social science | Model analysis and experiments | Model documentation | Modelling and simulation | Modelling languages |
- To cite this resource:
Jakub Bijak, Sarah Nurse, Jason Hilton, Toby Prike. (2022) Agent-based modelling for social research- Block 2: Data Quality, Model Analysis and Psychological Experiments. National Centre for Research Methods online learning resource. Available at https://www.ncrm.ac.uk/resources/online/all/?main&id=20767 [accessed: May 24, 2022]