Error Reasoning in Complex Systems : Training and Application Error for Decision Models
Figure 2 . The user interface of the agent-based model in NetLogo . The dark blue cells represent the classrooms . The circles represent the student agents . The different colors represent the agents ’ “ true ” class level ( cyan agents ’ true class level is 1 , green agents true class level is 2 , and yellow agents true class level is 3 ).
The “ school ” has 36 classrooms with a capacity of 25 students each . We set the total number of students at the school to be 720 ( 80 % of total school capacity ). Each classroom is provided a class level and students are given a “ true ” class level TrueL . Students who are placed in a classroom that matches their true class level are said to have a class mismatch of 0 ( TrueL == AssignedL ). Students placed in a classroom that does not match their true class level are said to be mismatched by | TrueL − AssignedL |. At model initialization , students are placed in classrooms based on a predetermined percentage of students to be mismatched by 1 and by 2 , φ 1 and φ 2
, respectively . In the event that no classroom space is available that meets a student ’ s mismatch criteria , the student will be randomly placed in an available space . A specified percentage of students , θ are then randomly selected to be “ disruptors ”. Table 1 outlines the model ’ s input parameters .
At each tick of the simulation , students can be influenced to become disruptors . A student ’ s decision to become a disruptor is a function of their mismatch and vision threshold . A student with a mismatch of 1 or 2 will “ look ” at its neighbors within its vision . If another student within this vision is a disruptor , the student will be “ influenced ” to also become a disruptor . The greater the student ’ s mismatch , the larger the vision threshold . The model out-
173