In this article, Van Lehn outlines various practices, goals, and scaffolds that educators can use in the classroom in order to implement computational model construction. Van Lehn begins by reiterating the three different types of models that we already explored in our Zika projects--structural, functional, and behavioral. Since each of these models, as we already observed, serves to represent a different facet of model construction, it could be useful to have our students analyze the spread of Zika from each of these viewpoints before using NetLogo. (Van Lehn mentions that research has indicated that concept mapping should be done before constructing a computational model, so requiring students to think about and conceptualize the spread of Zika from a structural, functional, and behavioral lens could help them organize their thoughts and determine what aspects of Zika they may need to research prior to model construction).
Van Lehn also provides descriptions of various types of model-based thinking, which we could utilize in the classroom to help students consider the complexity and predictive power of model construction. He mentions that node diagrams and agent-based modeling can help students observe the causes and effects of manipulating single variables or behaviors, which is beneficial for overall model accuracy. Along these lines, students should develop a method for scoring their models (based on the number of accurate behaviors or conditions)--the higher the score, the more realistic the model becomes. We have discussed "curve-fitting" in modeling construction earlier in the course, and scoring a model based on certain criterion can help ensure that students' mechanisms for the spread of Zika are accurate (instead of simply matching current data). Van Lehn further posits that graphical and mathematical representations of agent-based models can allow students to make model-based predictions. Thus, before undergoing model construction, we as teachers should explicitly teach (or make sure that students already have) graphical and mathematical literacy. Lastly, the author indicates that research on modeling has shown that "meta-tutoring," with respect to modeling practices, can be beneficial for students. These tips can be divorced from direct content knowledge, as long as the students are able to comprehend and implement the various steps of effective model-making.
While these all seem to be good points, I did leave the article with a few unanswered questions. With regards to meta-tutoring, the author observed students using Betty's Brain program, which contained built-in meta-modeling construction. Assuming our students will be using NetLogo, which does not contain a built-in meta-tutoring component, what can we do to ensure that students think about the practices inherent in model construction and deployment? Also, as per usual, I wonder about the time commitments required in classrooms to effectively implement computational modeling. Van Lehn elaborated on the duration of NetLogo training in students who built models (48 hours) and students who simply explored existing models (2 hours). Given this much larger training period for students to create their own models, when is it appropriate to have students explore pre-made models, and when should they be required to build their own?
I like how you have incorporated concept mapping into the process of modeling because VanLehn makes the point that "a concept map is a model, but it is not an executable model;" (377) therefore, by having students complete this before they create their model they can see that these are two distinct entities and could be less likely to confuse the two. I also like the idea of having students develop a method of scoring their own models. That could work well into the NGSS standards incorporating 4,5, and 6: Analyzing and interpreting data, Using mathematics and computational thinking, and Constructing explanations (for science) and designing solutions (for engineering). Based on the score their model receives students can analyze how efficient/accurate their model was, creating the scoring system could involve manipulating mathematical equations, and finally based on the score their model receives students could be prompted to explain their successes and misconceptions.
ReplyDelete