PIs: William Finzer, Timothy Erickson, Frieda Reichsman, Michelle Wilkerson-Jerde
This project refines and studies technology for ‘data science games’: essentially, a game is embedded in a data analysis environment, in which the game can only be ‘won’ by doing data modeling. Research will examine how students learn to analyze and model data in high school biology, chemistry, and physics; how the game can support this learning; and how such games can fit into high school science classrooms (as tested in a large urban school district).
This project uses design-based research methodology to understand how students engage with and learn about data science (specifically, concepts of center, spread, distribution, and inference) and to identify social, technological, and pedagogical supports to allow classroom use, including various types of data representations in the data games (flat, hierarchical, tree, digraph, and relational). Semi-clinical interviews and direct observation of students using the games in controlled settings will lead to broader workshop- and classroom-based observations of individuals and dyads using think-aloud protocols. This data will be analyzed both using grounded theory and using diSessa’s knowledge analysis method and analysis of discourse using Hmelo-Silver et al.’s CORTDRA diagrams. In addition, teacher focus groups and classroom video will be used to help identify affordances related to classroom adoption. The design and development work will be driven by the empirical research, and will utilize Squire’s game-based learning design principles, starting with the existing CODAP (Common Online Data Analysis Platform) software. Four design iterations will take place, each culminating in user testing, initially with a summer workshop of students, then with two rounds of classroom testing in six high school classrooms, and finally in an open adoption phase in which 50 teachers will be recruited for data collection (but more teachers may adopt the software and curricula).