In Covid times, politicians have to make difficult decisions on measures to prevent the disease's further spread. They make choices between complete lockdown, wearing face-masks, or even an evening clock. It is not always clear which of these measures will have the best effect on the number of infections. In fact, their trade-off is even more complicated when we consider that they can enforce these measures for an entire country or only for the most affected regions.
Simulation models can help politicians to decide on which intervention methods are most effective. However, this requires models that do simulate not only disease spread but also human behavior. When a government enforces a lockdown, the number of contacts between individuals will decrease, leading to less subsequent infections. This will only happen when citizens stick to the rules. If they do not feel the risk of becoming infected, they may ignore the implemented measures making them less effective. There are many examples, like illegal parties or visits with many family members during the Christmas break.
Protection Motivation Theory (PMT) is a framework developed by Rogers in 1975 and further adjusted in 1983 (Rogers, 1983). It divides the decision-making process into two steps: risk assessment and coping appraisal. In the first step, the level of risk is assessed. In case there is a risk, the second step will decide on the action to take. Risk assessment is conducted in many agent-based models, e.g. to evaluate the risk of flooding (Huang, Parker, Filatova, & Sun, 2013), to determine the risk for disease (Abdulkareem, Augustijn, Filatova, Musial, & Mustafa, 2020; Abdulkareem, Augustijn, Mustafa, & Filatova, 2018) etc. When an agent experiences risk, this will influence the behavior of this agent. The agent will try to make decisions that reduce the risk level or prevent risk completely. However, in most of the current disease models, this risk perception is not included.
We are currently working on a Covid model in which politicians perceive risk at a country level to decide which measures to apply. However, we think that risk should also be perceived at the local level (municipalities). The main reason for this is that locally, the number of infections might be higher, and hospitals' pressure may be beyond the capacity.
In this MSc topic, three steps will be taken:
- Adjusting the PMT model to include municipal risk perception
- Extend the current agent-based model in such a way that control measures can be taken at both the local (municipality) and national level
- Evaluate model output comparing a national versus a local strategy

This topic includes programming in Netlogo and basic programming skills are an advantage. The topic may also include working in R as the currently implemented risk model uses a Bayesian Network in R to steer agent behavior. Data is important to calibrate models. Pre-processing of data may be part of this topic.