Scientific Reasoning Across Different Domains

This study seeks to establish which scientific reasoning skills are primarily domain-general and which appear to be domain-specific. The subjects, 12 university undergraduates, each participated in self-directed experimentation with three different content domains. The experimentation contexts were computer-based laboratories in d.c. circuits (Voltaville), microeconomics (Smithtown), and the refraction of light (Refract), Subjects spent three 1-1/2 hour sessions working with each laboratory and took pretests and posttests that assessed their learning. Specific patterns of strategies used in each laboratory depended primarily on the structural form of the discovery task and the nature of the domain. In a situation that required the discovery of correlational regularities, evidence-generation activities, like the heuristic of controlling variables, were primary. In contexts where the regularities were functional rules, evidence interpretation became important. When the rules were quantitative, mathematical and algebraic heuristics were important. Students appeared very sensitive to the task demands of each laboratory, and adjusted their strategies accordingly. Regardless of this adaptation to specific conditions, they learned more as they proceeded from domain to domain, indicating that they were becoming more effective in planning and carrying out experiments, and in formulating and testing hypotheses based on those experiments. The findings suggest that the most generally useful skills for direct instruction may be those for evaluating the kind of problem at hand and for selecting the most appropriate processes and strategies.