������������������
made recognizing that in reality , consumers may not be able to directly pinpoint the cause of foodborne disease , or may not realize that they have been affected by a mild foodborne illness . Further research to determine the exact factors that would cause consumers to return to a restaurant where they had hygiene concerns will be needed to help refine this assumption in the model . ������������� : Consumers in the second or alternative scenario will only go to restaurants with a low re-inspection priority rating for 15 time steps after healing . This is consistent with hysteria and indifference swings ( Beck , 1992 ) and the observed tendency of consumers to gradually resume prior consumption patterns following an outbreak ( Bocker & Hanf , 2000 ). ����������� ����������� : One assumption used in the model is that consumers and inspectors are unable to tell if a store is contaminated prior to arrival . This assumption is driven by the notion of asymmetric information ( Akerlof , 1970 ). Elements of a restaurant that are relevant to safety , such as the cleanliness of food preparation areas , are generally hidden from consumers , providing a further element of asymmetric information ( Filion & Powell , 2009 ). However , the simulation allows for this fundamental asymmetry to be gradually eliminated , depending on agent type and scenario .
The model ’ s key results and outputs include the number of sick consumers , the number of sick and at-risk consumers , the number of consumers that never get sick , the number of inspected and contaminated restaurants , and the number of restaurants for each re-inspection priority level . Most of the results seem to arise because of the rules and assumptions of the model . However , the reduction in variability seen in the second scenario is likely an emergent result that requires further discussion .
Consumers adapt their behavior by updating the list of bad restaurants and avoiding these locations in the future , even if an inspector has inspected the restaurant . This captures the consumer objective of avoiding sickness . In the alternative scenario , consumers further pursue this objective by only visiting lowrisk restaurants following illness , a behavior motivated by risk aversion and imposed by the model ’ s rules . In the current model , neither restaurants nor inspectors demonstrate adaptive or learning behavior . Given that some literature has shown that the introduction of grade score cards in Los Angeles County restaurants led to increases in inspection scores ( Jin & Leslie , 2003 ), some form of adaptive restaurant behavior should be introduced in future versions of this model .
Inspectors have different sensing abilities than consumers in the baseline scenario . Inspectors are able to observe the re-inspection priority level of a restaurant and search for a restaurant to inspect based on this criterion . Consumers are only able to use this information in the alternative scenario . Importantly , neither consumers nor inspectors can tell if a restaurant is contaminated prior to arriving at it . Also , consumers cannot sense whether a restaurant has recently been inspected or whether consumers near them are sick .
Consumers and inspectors do not interact in the model . Consumers interact with restaurants by visiting them , but they do not interact with other consumers who may also be present at that location at that time step . Inspectors interact with restaurants by inspecting them and changing their re-inspection priority scores and contamination variable .
146