The assessment of learning outcomes of computer modeling in secondary science education

The computer modeling of dynamic systems is a topic that aligns well with the current trend in secondary science education to actively involve students in their knowledge construction, give room for inquiry, and offer realistic tasks. In this dissertation the specific learning outcomes that can be expected from computer modeling are investigated, as well as how they can be measured. The reasoning processes of modeling are explored and systematized as what is called the ‘ACE framework’ that describes modeling knowledge in three dimensions: 'type of reasoning', 'complexity', and 'domain-specificity'. The dimension of ‘types of reasoning’ includes applying (A), creating (C), and evaluating (E). The dimension of complexity distinguishes between reasoning with simple and complex situations. The dimension of domain-specificity describes the extent to which reasoning is dependent on the domain and distinguishes between domain-specific and domain-general. The framework was used to develop the 'ACE test' for the domain of the energy of the Earth. In three studies the ACE test is validated and used to compare different modes of instruction: modeling-based, simulation-based and an expository mode of instruction. A general pattern that arose from the comparative studies is the difference in performance between the groups with respect to simple and complex items and processes. Overall, the students in the modeling-based mode of instruction performed better on the complex items. The modeling processes are progressively more demanding with regard to the types of reasoning involved, from the straightforward reproducing of conceptual knowledge to the higher-order processes of applying, creating, and evaluating. The findings indicate a benefit for the learners in the expository and simulation-based instruction on the simplest process of reproducing conceptual knowledge. The modelers benefited progressively more than the learners in the simulation-based instruction on the complex items and the more demanding processes of applying and creating. However, no differences were found for the process of evaluating between learners in the modeling and the simulation-based mode of instruction.