| Research on science instruction indicates that the traditional science classroom is not always effective in improving students' scientific understanding. Physics courses, in particular, do not promote the ability to apply scientific principles for many reasons, based on their focus on procedural problem-solving and lab exercises. In this dissertation, I propose the Designing-to-Learn Architecture (DTLA), a design-centered goal-based scenario (GBS) architecture, theoretically grounded in the literature on design-centered learning environments, goal-based scenarios, intelligent tutoring systems and simulations. The DTLA offers an alternative approach to addressing the issues encountered in the traditional science classroom. The architecture consists of an artifact with associated design goals; components with component options; a simulation; a reference database; and guided tutorials. I describe the design of Goin' Up?, the prototype DTL application, serving as the basis for evaluating the effectiveness of the DTLA. I present results of interview and testing protocols from the formative evaluation of Goin' Up?, suggesting that learning outcomes, though not statistically significant, could be improved through DTLA enhancements informed by usage patterns in software sessions. I conclude with an analysis of the results and suggestions for improvements to the DTLA, including additional components to address reflection, provide support for novice designers, and offer tutorial guidance on the analysis of the artifact. |
