Page 142 - Getting the Picture Modeling and Simulation in Secondary Computer Science Education

140
Chapter 6
144 cheeses are what, that you could see in the chart.” Interestingly, student S11, who modeled bank counter queues, made extensive use of monitors in the model to observe its behavior and outcomes, but does not explicitly refer to them when asked about validating the model.
Interpreting Outcomes
In order to assess the validity of a model, the observed outcomes can be interpreted, either subjectively or objectively — for example using statistical tests.
When possible, the model’s outcomes can be compared to the behavior of the real system.
Three groups reported comparing the events occurring in their model to those of a real system as they perceive it, thus employing event validity technique and checking for consistency with previously verified theories. However, they made no use of historical data. When asked what they think would happen in reality if the farmer did not intervene, student S1 replied, “I think the affected plants would take over. And here, the new, healthy plats are added all the time” and later went on to add, “affected plants die and in reality, a new plant would not be planted there. So it is, wait first for everything to be harvested and only then will the new plants be added.” When the model results indicated that use of fertilizer made no difference, student S2 commented, “right, that is not realistic, so we knew through our common knowledge [...] that the outcomes were not good.” Yet, this group found their model sufficiently valid, as described in the section on plausibility and accuracy. Group G2 compared the behavior of their model with their own experiences of a fire drill at school and concluded that the model was satisfactory. Student S11 commented, “In reality people enter the building and join the queue, the shortest queue.”
Reasoning about the Model
When the data from the real system are not available, validity of a model can be derived through reasoning. Here, one asserts that if the assumptions and the implementation of a model are correct, then it follows that the model is valid. This is what group G5 did and student S9 said, “so then we looked if this and this happened one after another, does that happen one after another in reality too? If that’s right, then it should be right in the model too, because, say, we had no real information if it was really right, so no real information from cheese producers that you could fill in, like, are the outcomes the same.”