University of Insubria, Varese
Davide Secchi was invited to deliver a one-hour seminar 'How many times should my simulation run?' at the University of Insubria, Varese (Italy), Friday, September 30th, 2016. After introducing 'agent-based models' (ABM), the talk covered needs and problems related to the lack of a method to determine how many times a simulation should be performed. In a multi-paper cooperation with Raffaello Seri (U. of Insubria), they indicate that statistical power analysis can solve the problem and estimate the number of runs for a simulation, given the configurations of parameters are known (as they usually are). In the talk, Davide shows that most ABM studies are underpowered, meaning that each configuration of parameters is not performed enough times to make sure results are not random occurrences. The talk ended with an explanation of the benefits to use power calculations and the floor was then open to questions.