Print PageEngaging students in iterative construction and evaluation of systems models offers the potential to support teaching and learning about multi-scale complex biological problems. In a systems model, boxes contain major concepts linked by arrows describing relationships between concepts. In an introductory biology for science majors (n=180 students), we regularly asked students to construct systems models to illustrate their understanding about connections among major concepts in genetics (e.g., DNA, gene, allele, phenotype) and evolution (e.g. fitness). This approach was used in multiple assessments (e.g., homework, exams, in-class problems) and with multiple cases to emphasize the foundational nature of these principles throughout biology. We quantified the change in complexity and correctness of students’ models over a semester. Complexity was measured as both the ratio of relationships to concepts and using a web-complexity index (WCI) that accounts for the degree of branching within a model (i.e., multiple relationships linked to a single concept). Each relationship was then rated for correctness using a three point scale: (1) incorrect, (2) plausible but vague, or (3) correct. We compared students’ models in terms of correctness and complexity at two points in time: the mid-term and final examinations.
Results/Conclusions
Ten students were randomly chosen from those scoring above and below the median final course grade, for a total of 20 students. On average, students used the same number of relationships, but more concepts on the final than the mid-term (11 vs. 10.5). Spatial complexity was greater on the mid-term than the final as measured by both metrics (P<0.05). However, mean student correctness improved by 0.15 points (P=0.2) from the mid-term to the final exam. High- and low-performing students constructed similarly complex models on the mid-term exam, but higher performing students had significantly greater correctness (2.1 vs. 1.7, P<0.05). By the final exam, higher performing students used more concepts and relationships and continued to have significantly greater correctness (P<0.07). Students quickly learned model construction and included many relationships on the mid-term exam, even spurious relationships. By the final exam, students often used a more parsimonious model (correctness increased) even though complexity decreased. We speculate that high-performing students identify meaningful relationships earlier in the semester and are then able to correctly increase both the numbers of concepts and relationships. Building models of real world scenarios may prepare students to simplify and represent the interactive, simultaneous processes common in biologically complex systems.
