������������������
spection priority score , and whether they are compliant . Patches in the computational environment that are not defined as restaurants represent empty space . As in most NetLogo models , the computational environment is a torus that measures 33x33 grid squares with a central origin point , comprised of 1089 patches . Realizations and interactions of the model progess for 75 time steps . The temporal scale of a time step and spatial scale of a patch are not specified .
������������������������������� : The following processes take place once per time step in the following order .
( 1 ) �������������������� : The likelihood of restaurants becoming contaminated depends on whether or not they comply with regulations . Restaurants that are compliant have a 0.5 % chance of becoming contaminated each time step , but restaurants that are not compliant have a 3 % chance of becoming contaminated each time step . Restaurants that become contaminated change their contaminated variable from 0 to 1 . This is shown visually in the model by changing the color of the restaurant to red . ( 2 ) ������� : Consumers who are not sick select a restaurant within their operating range that does not belong to their current list of “ bad restaurants ” ( places where they previously got sick ) and move there to eat . If the consumer lands on a contaminated restaurant they have a chance of becoming sick , which also varies depending on whether they belong to an at-risk group . If the consumer becomes sick , they update their list of “ bad restaurants ” and remain sick for a specified number of time steps , depending on whether the agent is part of an at-risk group . In the second scenario , the consumer also behaves in a risk averse manner for 15 time steps after healing , meaning that they will only go to low re-inspection priority restaurants . If there are no suitable restaurants within the consumer ’ s range , the consumer simply wanders to look for restaurants in future time steps . ( 3 ) ���� : The inspector agent prioritizes restaurants based on three levels of re-inspection priority . Each time step it selects a restaurant to inspect . It first chooses high priority restaurants within its range , then moderate , then low . If the chosen restaurant is contaminated , the inspector fixes this by changing the restaurant ’ s contaminated variable from 1 to 0 , and subsequently raises the re-inspection priority of that restaurant . If there is no contamination upon inspection , the inspector lowers the restaurant ’ s re-inspection priority ( if possible ). ( 4 ) ���� : Consumers that have been sick for three time steps heal and then re-circulate . If they belong to an at-risk group , it takes five time steps to heal .
������� �������� : Since this model represents a stylized restaurant inspection system , much of its design has been informed by the food safety literature . The following basic principles are incorporated into the design of the current model .
��������� ������� ����� : Producers , distributors , and suppliers that make up the global supply chain are not explicitly observed in the model . Since consumers only directly interact with the supply chain at the retail level and the model focuses on restaurant inspections , only restaurants are included . Grover and Dausch ( 2000 , as cited by Knight et al ., 2009 ) estimate that a foodborne illness outbreak could cost food service outlets $ 100,000 and up to a 30 % loss in sales due to decreased consumer trust . Even though a restaurant
144