Authoring tools for investigate-and-decide learning environments

The subject of this thesis is INDIE, a software tool for designing and implementing multimedia educational environments called Goal-Based Scenarios (GBSes). INDIE's products are simulations centered around tasks where students take the role of an expert who has to make a diagnosis, such as a geologist charged with determining if a volcano will erupt. The goals of the INDIE tool are that it be usable in many domains, be able to produce finished products as well as storyboards, allow incremental development, be able to develop a domain model separate from the interface and vice-versa, efficient for authors, accessible to non-programmers, and effective. A model-driven GBS development tool was first implemented (INDIE 1.0), but our design made it so authors had to develop complete domain models at every stage of development, making rapid prototyping of ideas difficult. INDIE 2.0 used an interface-to-model, incremental approach that let authors develop at any level of design, from storyboards to deliverable applications. Tool features that made INDIE 2.0 successful include an event/response architecture that implements interface actions, pre-specified interaction models for frequently-needed interface elements, and a simple but flexible rule-based engine to critique candidate student diagnoses. We analyzed the results of five major teams of authors that used INDIE to build applications in several domains. We found that authors (who were not trained as programmers) were able to build applications that showed nearly a ten-fold increase in complexity over those built in INDIE 1.0 without any increase in development time.