An Experimental Investigation of Learning Mechanics in a Digital Geometry Game

• In Event: Designing and Implementing Educational Games: Exploring How to Increase the Effectiveness of Games for Learning

Abstract

With the growing interest in employing digital games and simulations for the purpose of teaching and learning, we are interested in studying features of well-designed learning games (Gee, 2003). Evidence Centered Design (ECD) is a useful framework for developing games for learning purposes since it requires designers to be specific about what learners need to do, say, or create in order to provide evidence of specific variables that are assessed (Mislevy et al., 2003). This is expressed in one of the key properties of a well-designed educational game, the learning mechanic, which we define as specialized game activities grounded in learning sciences (Plass et al., 2013). In this paper, we ask how two different learning mechanics, both grounded in middle school math education practice, impact learning outcomes in a geometry game.

Design & Measures
Middle-school students (N = 141) were randomly assigned to play one of two versions of a single player geometry puzzle game, Noobs vs. Leets, each featuring a different learning mechanic. In the Rule version, players were asked to identify which rule they would apply. In the Number version, players were asked to calculate the angle and select the correct number value. Participants were first administered a pretest of geometry knowledge (featuring conceptual items and applied items), followed by 25 minutes of independent gameplay, concluded by a geometry posttest and a situational interest survey.

Results & Discussion
Overall, analyses of covariance revealed that individuals in the rule condition performed better in the game, completing more levels. For students with low prior knowledge in conceptual understanding and arithmetic skills, the number condition resulted in greater situational interest than the rule condition.
Learning outcome results suggest that the effect of condition varied based on prior knowledge. For low prior knowledge learners, longer duration of game play was associated with better performance on post-test measures of conceptual geometry understanding (Figure 2). High prior knowledge learners also benefited from completing more levels; however, they performed better on the conceptual post-test in the rule condition than in the number condition.
Findings suggest that given sufficient prior knowledge, a conceptual approach to learning the rules of geometry leads to stronger conceptual learning outcomes. Conversely, a numbers-based mechanic results in increased geometry-specific arithmetic skills. These results highlight the importance of choosing a game mechanic that reflects the intended learning outcomes when designing games for learning.

Authors

• Jan L. Plass, New York University

• Bruce Douglas Homer, City University of New York

• Elizabeth Hayward, New York University

• Melissa L. Biles, New York University

• Jonathan Frye, New York University

• Tsu-Ting Huang, New York University