Modeling as a core component of structuring data

Abstract

We gave participants diagrams of traffic on two roads with information about eight attributes, including the type of each vehicle, its speed, direction and the width of the road. Their task was to record and organize the data to assist city planners in its analysis. Successfully encoding the information required the creation of a case, a physical record of one repetition of a repeatable observational process. We analyzed data sheets participants created including the methods they used to bind information together into cases. Overall, 79% of their data sheets successfully encoded the data. Even 62% of the middle school students were able to create a bound structure that could hold the critical information from the diagrams. A majority of these structures involved a hierarchy of cases rather than the "flat" case-by-attribute structure that virtually all statistical software require. Our sense is that participants strove to create a structure that modeled the real-world as closely as they could, constructing cases that corresponded to the different sorts of objects they perceived-vehicles with their characteristics nested within road segments with their characteristics.