Page 146 - Getting the Picture Modeling and Simulation in Secondary Computer Science Education

144
Chapter 6
and scope of their assignments: within a CS course, and not within a course on a particular (scientific, engineering, etc.) discipline. Finally, some of these techniques are rather advanced and belong into the repertoire of a professional modeler, rather than a student attending secondary education.
Another type of validation technique students did not report using was the engagement of external experts in any phase of their modeling process. They did not consult domain experts (Wilensky & Rand, 2015), performed no Delphi tests to seek consensus of experts on problematic outcomes (Carley, 1996), nor carried out structural walkthroughs of their models with peer groups (Sargent, 2013) — all of which they arguably could have done. None of them did a Turing test either, where they would have asked knowledgeable individuals whether it was possible to distinguish between the outcomes of the model and those of a real system (Sargent, 2013). We could have expected students to call upon experts. In one of our previous studies (Grgurina et al., 2016), we reported about our student talking to a medicine student to learn more about a particular disease; so it would not have surprised us if the students in this study had consulted their peers, teachers, or other people either to learn more about the phenomenon under scrutiny, or to seek feedback on any aspects of their models.
When we set side by side the findings from this study with the outcomes of the study we performed in 2016 (Grgurina et al., 2016), we noticed a few things.
Then, the students were expected to decide themselves what to model and some of them had difficulties coming up with a suitable problem. This time, the students choose problems from a list rather than coming up with their own problems. Consequently, all of the students this time had a clear idea of the purpose of the model they were developing and using.
Concerning the research the students did (or neglected to do), they reported similar sources of their knowledge — except this time, no-one reported consulting with an expert.
When it is time to state the assumptions the model is built upon, i.e. it is time to engage in abstracting, in this study we not only observed students having difficulty with this aspect of modeling — as we did in the 2016 study — but have also identified three distinct errors: oversimplification, omissions and circular reasoning.
Regarding the construction of the models and subsequent testing, in both of our studies, the students reported developing their models in small steps, continually testing, adjusting and expanding their models. This time we focused