Data-driven game design: The case of difficulty in educational games

Abstract

There is increasing interest in using data to design digital games that serve the purposes of learning and assessment. One game element, difficulty, could benefit vastly from applying data-driven methods as it affects both players’ overall enjoyment and efficiency of learning and qualities of assessment. However, how difficulty is being defined varies across the learning, assessment, and game perspectives, yet little is known about how educational difficulty can be balanced in educational games for each of the potentially conflicting goals. In this paper, we first review varying definitions of difficulty and then we discuss how we came up with a difficulty metric and used it to refine our game-based assessment Shadowspect. The design guidelines, metrics and lessons learned will be useful for designers of learning games and educators interested in balancing difficulty before they implement these tools in the classroom.