”Every move you make in every game you play”: can game logs determine player’s engagement?

Context: Metrics and instruments available to measure player’s engagement usually involve the application of questionnaires or interviews. Problem: Despite their importance and validity, these approaches can affect the player experience by interrupting the interaction process. In addition, the literature reports a growing interest in the development of games capable of real-time self-adapting based on the player state. This type of game requires strategies and tools to allow an automatic and non-noticeable evaluation of the user’s experience, which is a challenging and open research field. Solution: Game analytics, which embraces techniques that can automatically collect and analyze data, can contribute in this context by providing data that can be interpreted in order to establish eventual relation with players’ experience. IS Theory: Flow theory. Methods: This study investigates whether there is a correlation between interaction logs with the self-assessed engagement. An experimental evaluation was performed with 37 volunteers interacting with two games. The correlations were analyzed considering each log individually and their classification in three categories: neutral, positive and negative. Results: The results of the experimental evaluation reveal that the correlation coefficients found are considered weak (about 0.44). On the other hand, it is interesting that the categorization of logs between neutral, positive and negative presented the main significant results in both games. Contributions and Impact in the IS area: The experiment showed weak correlation between game logs and engagement in the most cases. However, they suggest that the categorization of logs in neutral, positive and negative logs can be an interesting approach.