Policy and Complex Systems
Figure 5 . ABM environment for base run in the preliminary model .
Outcome One
We analyze the ratio of protesters
under several scenarios with different agent ’ s grievance level distribution and different threshold levels for the ratio of protesters to cops . We test randomly uniform , randomly normal , and randomly exponential distributions during the experimental phase . Figure 6 and Table 6 show the protester ratio among all civilians under each scenario . In most of the scenario runs , agents experience oscillatory behaviors due to interactions among agents ; however , the amplitude of the oscillations differs in some cases such as case i and c . The case with normally distributed grievance levels ( c ) has a noisier graph under the highest threshold compared with the case with exponentially distributed grievance levels ( i ). Exponentially distributed grievance levels can create more realistic results ( in terms of stability ), but it should be considered that the threshold level might also have an effect on these oscillatory behaviors . When the threshold is at its lowest value , which means cops move as soon as they see one protester , agents become more risk averse and are not as eager to protest . Therefore , for these cases ( b , e , h ), the ratio of protesters among civilians is lower than in other cases .
Table 6 . Different agent ’ s grievance level distribution and different threshold levels for the ratio of number of protesters over the number of cops
27