VanLehn Response

I think much of the work that needs to go into helping students construct a model of the spread of Zika is applicable to most, if not all, models.  A baseline of knowledge is essential to the successful construction and evaluation of models, after all.  In this vein, I think it is essential for students to be familiar and instructed in (at least briefly) the various kinds of modeling VanLehn discusses in his paper (374).  I venture that instruction in one kind of modeling (say, agent-based for Zika) could considerably streamline the classroom process, but it would do so in a self-limiting way.  If the goal of modeling is to engage students' critical thinking skills and teach them to think proactively, it should be made clear to them the flexibility modeling affords them, and the fact that modeling is as much a concept as it is a practice.  Certainly instruction may focus on a particular model for simplicity or ease of use, but students should be aware of alternatives.

In a similar vein, students should not only be aware of different types of models that exist, but also of the basic theory behind modeling; many of the problems VanLehn outlines as recurring challenges teachers face/shortcuts students take in the subject of modeling stems from a fundamental misconception of what exactly modeling is and why we do it and can lead to unmotivated student work.  Examples like inserting fudge factors, or as VanLehn put it: "models can be valid with respect to known observations but nonetheless make incorrect predictions about future observations".

Students should be instructed in the subject in question beforehand as well, in order to establish a common ground of knowledge and to familiarize students with some forms of content and the way a certain system functions.  It is unreasonable to expect, for example, a student to be able to account for the rate of the spread of Zika with respect to the resident mosquito population rates and reproduction rates; one cannot take into account factors they are not aware of or do not realize are relevant.

Practices that may be useful when specifically discussing Zika may be a focus on the decompositional model as outlined by VanLehn.  Particularly for a system with multiple variables, it may be overwhelming to students to attempt to construct a model from the start.  Instead, it may be more helpful to break down the spread of Zika into multiple components (depending on how complicated the desired model may be).  Synthesizing these components would almost certainly require considerable tweaking and bugfixing on the part of the student, but these activites in and of themselves are useful tools in understanding both how to functionally create a model and how variables interact when put together into a more complex, realistic system.  Bugfixing in itself could thus serve as a valuable introduction for students to both the scaffolding of the Zika problem and to higher order model building/functionality.

Meta-tutoring would likely be my favored method of providing feedback to the student.  Since the goal of modeling is to help students understand/think more about the natural world and how we can use these techniques to solve practical problems (or at least that is how I interpret it), meta-tutoring serves as a method of refining the student methodology and thought process.  Similarly, I think that use of natural language as outlined by VanLehn would be useful in testing how well students understand the actual workings of a system as opposed to being able to create a desired output.