@ARTICLE{10.3389/feduc.2020.572546, AUTHOR={Cloude, Elizabeth B. and Dever, Daryn A. and Wiedbusch, Megan D. and Azevedo, Roger}, TITLE={Quantifying Scientific Thinking Using Multichannel Data With Crystal Island: Implications for Individualized Game-Learning Analytics}, JOURNAL={Frontiers in Education}, VOLUME={5}, YEAR={2020}, URL={https://www.frontiersin.org/articles/10.3389/feduc.2020.572546}, DOI={10.3389/feduc.2020.572546}, ISSN={2504-284X}, ABSTRACT={Quantifying scientific thinking using multichannel data to individualize game-based learning remains a significant challenge for researchers and educators. Not only do empirical studies find that learners do not possess sufficient scientific-thinking skills to deal with the demands of the twenty-first century, but there is little agreement in how researchers should accurately and dynamically capture scientific thinking with game-based learning environments (GBLEs). Traditionally, in-game actions, collected through log files, are used to define if, when, and for how long learners think scientifically about solving complex problems with GBLEs. But can in-game actions distinguish between learners who are thinking scientifically while solving problems vs. those who are not? We argue that collecting multiple channels of data identifies if, when, and for how long learners think scientifically during game-based learning compared to only in-game actions. In this study, we examined relationships between 68 undergraduates' pre-test scores (i.e., prior knowledge), degree of agency, eye movements, and in-game actions related to scientific-thinking actions during game-based learning, and performance outcomes after learning about microbiology with Crystal Island. Results showed significant predictive relationships between eye movements, prior knowledge, degree of agency, and in-game actions related to scientific thinking, suggesting that combining these data channels has the potential to capture when learners engage in scientific thinking and its relation to performance with GBLEs. Our findings provide implications for using multichannel data, e.g., eye-gaze and in-game actions, to capture scientific thinking and inform game-learning analytics to guide instructional decision making and enhance our understanding of scientific thinking within GBLEs. We discuss GBLEs designed to guide individualized and adaptive game-analytics using learners' multichannel data to optimize scientific thinking and performance.} }