Supporting Students Modeling in a Digital Game: Adaptively Modulating Abstraction of Self-Explanation

• In Event: 63.025 - Supporting Modeling Epistemologies in the Science Classroom

Abstract

Objectives. The current study presents an approach that adapts the abstraction of self-explanation prompts based on a player’s performance to support players engaging in modeling in a digital game.

Theoretical Framing. Modeling, interpreting, translating, and manipulating across formal representations is central to scientific practice (Pickering, 1995; Lehrer & Schauble, 2006a, 2006b; Duschl et al., 2007), but modeling as an epistemological undertaking is challenging for students. Games as a medium have been demonstrated to be effective in supporting various aspects of science learning (e.g., Authors, 2013). This poster explores the design of adaptive self-explanation functionality to support deeper model-based thinking in games (Chi, Bassok, Lewis, Reimann, & Glaser 1989; Roy & Chi, 2005; Chi & VanLehn, 1991). Previous research suggested that self-explanation functionality can be effective in games for learning, but this previous research also illuminated the challenges in balancing and integrating the demands and abstraction of self-explanation functionality with the demands of the game, particularly for games that are themselves cognitively more complex (e.g., Authors, 2014; O’Neil et al., 2014; Johnson & Mayer, 2010). Accordingly, the current study presents an approach that adapts the abstraction of the self-explanation prompts based on a player’s performance in the modeling game.

Methods and Data. The current study was conducted with the 7th grade classrooms of two teachers in two different middle schools in the southeastern United States. A total of 210 students, including 124 from the first school and 86 from the second school, assented to participate in the study. Both schools are racially and culturally diverse. In terms of socio-economics, 43% of the students at the first school and 83% of the students at the second school are eligible for free or reduced lunch. Students were randomly assigned to one of three game conditions within each classroom (i.e., each classroom included students in all three conditions). In the navigation-only condition, players programmed their trajectories without any self-explanation prompts. In the navigation+abstract condition, these navigational challenges are paired with a self-explanation prompt that focuses on abstract connections between the navigational challenges and Newtonian relationships. In the navigation+adaptive condition, the navigational challenges are paired with self-explanation prompts that adaptively increase from low abstraction (in which the prompts focus concretely on navigational moves) to high abstraction (in which the prompts focus more abstractly on the navigational challenges in terms of overarching Newtonian relationships).

Results and Scholarly Significance. The results demonstrate that students in the navigation+adaptive condition (a) scored significantly higher on the post-test than students whose self-explanation prompts were not adaptively adjusted and were always abstract and (b) scored higher, but not significantly so, than students who did not receive self-explanation functionality. Analyses of game-play metrics suggest that trade-offs in terms of progress through the game may explain some aspects of these post-test comparisons. Analyses also demonstrate that both self-explanation conditions significantly outperformed the navigation-only comparison condition on a game-play metric that suggests deeper model-based thinking.

Authors

• Doug Clark, University of Calgary

• Satyugjit Singh Virk, Teachers College, Columbia University

• Jackie Barnes, Northeastern University