If the gear fits, spin it! Embodied education and in-game assessments

Two embodied gears games were created. Better learners should use fewer gear switches to reflect their knowledge. Twenty-three 7th graders, playing as dyads, used gestures to manipulate virtual gears. The Kinect sensor tracked arm-spinning movements and switched gear diameters. Knowledge tests were administered. Statistically significant knowledge gains were seen. For Game 1 (gear spun one direction), switching significantly predicted only pretest knowledge. For Game 2 (gear spun two directions) switching was also negatively correlated with both tests. For game 2, those who used fewer switches during gameplay understood the construct better scoring higher on both tests. Dyadic analyses revealed the winner used significantly fewer switches. In-process data can provide a window onto knowledge as it is being encoded. However, games should stay within the learner's ZPD, because if the game is too easy (Game 1), meaningful data may be difficult to gather. The use of in ludo data from games with high sensitivity may attenuate the need for repetitive traditional, post-intervention tests.