Mining big data to help make informed decisions for designing effective digital educational games.

More research is needed on how to best use analytics to support educational decisions and design effective learning environments. This study was to explore and mine the data captured by a digital educational game designed for middle school science to understand learners' behavioral patterns in using the game, and to use evidence-based findings to inform the design and implementation of the game's next iteration. The data was from 3,963 sixth-grade students from 31 teachers across nine U.S. states who used the game. The dataset consisted of 3,063,267 lines of log data. The findings showed significant positive relationships between students' performance scores and their tool usage in terms of both frequency and duration of tool use. This was further confirmed by the relationships between different levels of students' performance and tool usage: the more frequently and longer that students used the tools, the higher their performance scores, and there were significant differences between students with high performance scores and those with low performance scores. Comparing three different cases over the individual days of game use provided further evidence that in all three comparisons, the students who used more essential tools and for longer time had higher performance scores. Implications are discussed.