data & models paired

Are folks tempted in the quantitative sciences to present too much theory sans observation, reflected e.g. implicitly in this recent article? This counterpoint on poorly designed high school tests suggests a need for more robust content at the high school level as well.

One fix may be to introduce inverse (algorithm-selection) challenges in parallel with the usual forward (algorithmic-reasoning) challenges. In other words, give students data to measure and interpret while they are also learning the models used to predict what's going to happen, so they at least know there's a science of choosing (as well as of using) models in the subject at hand.

This would help close the loop on concreteness, and call into play complementary skills that deserve attention during the crucial period in which students are deciding their major. Our NanoScience Practicals course is example of a cross-disciplinary offering designed to explicitly confront students each week with new data, as well as with topic-specific concepts & models.

This is also consistent with recent requests in the biological sciences for the development of better principle-based as distinct from informal reasoning skills. The importance of principle-based tools for selecting models is perhaps even more obvious in life-science than in physical science areas.