Are Current Educational Games Suitable for Data-Driven Education?

Gaming as a practice could be described as a constant and active learning process. Even when playing a game not specifically designed for educational purposes, the player has to learn new skills, new game mechanics, and new patterns to beat the game. It is the perfect combination of skill mastery and challenge. For a game to be enjoyable, a good game needs to be equally challenging but also manageable. Otherwise, it would be either too boring or too hard to play. This is often described as the state of flow. If this is applied well, together with well-defined learning objectives, highly effective educational games can be created.

When it comes to digital games, it is more than common for the developers to use various data-gathering techniques and analytics to better understand the behavior of the players to make their gaming experience as good as possible. On top of that, the interactive and flexible nature of digital games as a medium makes data gathering rather accessible. Considering this, a few questions come to mind. Is it possible to analyze and track how people learn through playing educational games? If so, can it be used to create better, more personalized, and more effective learning experiences? And are educational games good for data-driven education?

Nowadays, most educational game design is still more cognitive-based than data-driven. In other words, it is more rooted in theory than in actual learner behavior research. That is because the current approaches to research in this field are usually either overly simplifying or costly. They frequently omit the complexity of the game by simplifying the user behavior to what suits them, they focus more on knowledge tracing instead of focusing on how people learn, or they focus on motivation and engagement tracking (Lee et al., 2014). In contrast, the data-driven approach is based on end-user interaction. It can capture learning over time (Zook & Riedel, 2012), analyze where the learners make mistakes, and how they interact with the system (how and what they explore, which pieces of the puzzle are they solving first). Based on that they are later able to predict where the learner might struggle, and provide them with more tasks to perfect the desired skill or even offer hints and instructions. According to Hooshyar et al., the best approach is the combination of the two since the ”data-driven, and theoretical approaches can mutually reinforce each other, by continually testing and validating theories while also generating further hypotheses for study” (Hooshyar et al., 2016).

The potential is there, and it is grand. By tracking the inputs, immediate, and highly personalized responses can be provided, making learning even more effective (Lee et al., 2014). It also provides the creators and educators with a better understanding of learners’ behavior and effective learner modeling practices, and it is also less expensive. If handled correctly, a data-driven approach to educational games could be the answer to the demand for adaptability and personalization in education, making learning more effective, engaging, and efficient.

Unfortunately, according to Kadel et al. (2019), the current serious game design (another term for educational games) is rigid. They recommend the implementation of AI and machine learning to make serious games more dynamic and able to handle various scenarios (Kadel et al., 2019). Serrano-Laguna et al. (2018) also recommend the implementation of non-disruptive in-game tracking, meaning that the gameplay is not stopped to ask the player to give direct feedback to the developer, or fill out a feedback form, as it is often done now. Kadel et al. (2019) also recommend eye-tracking.

In conclusion, even though the data-driven approach to educational game design has immense potential to make the learning experience more personalized and effective, no complex methodology has been proposed and proven efficient, making educational games not quite suitable for data-driven education as of now.

References:

1. Hooshyar, D., Lee, C., & Lim, H. (2016). A survey on data-driven approaches in educational games. 2nd International Conference on Science in Information Technology (ICSITech), 291–295.
2. Lee, S. J., Liu, Y.-E., & Popvić, Z. (2014). Learning Individual Behavior in an Educational Game: A Data-Driven Approach. EDM.
3. Kadel, R., Krishna, P., & Gurung, M. P. (2019). A Review on Educational Games Design, Development, and Effectiveness Measurement. 2019 IEEE International Conference on Engineering, Technology, and Education (TALE), 1–7.
4. Serrano-Laguna, Á., Manero, B., Freire, M. et al. (2018) A methodology for assessing the effectiveness of serious games and for inferring player learning outcomes. Multimed Tools Appl 77, 2849–2871.
5. Zook, A. E., & Riedel, M. O. (2012). A temporal data-driven player model for dynamic difficulty adjustment. AIIDE’12: Proceedings of the Eighth AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment, 93–98.