디지털 게임기반학습에서 학습자 행동 시퀀스 유형에 따른 중도이탈 생존 분석 = Survival Analysis of Dropout in Digital Game-based Learning By Learner Behavior Sequence Types

다국어 초록 (Multilingual Abstract)

This study aimed to classify learner behavioral sequence types in Digital Game-Based Learning and to investigate the patterns of dropout among these types. To achieve this, log data from 25,961 participants who engaged with the educational game ‘Wake: Tales from the Aqualab’ were analyzed. Sequence distances were computed using optimal matching based on in-game behavior types, followed by K-means clustering. Subsequently, N-gram analysis identified behavioral characteristics of each cluster, and Kaplan– Meier survival analysis with log-rank tests was conducted to examine differences among clusters. The results showed that the participating learners were classified into three types: ‘Trial-and-Error Explorer’(90.1%), ‘Immersive Strategist’(2.5%), and ‘Exploratory Strategist’(7.4%). The Trial-and-Error Explorer type was characterized by inefficiently repeating in-game actions. The Immersive Strategist type showed a prominent tendency to focus on efficient game paths centered around core activities. The Exploratory Strategist type demonstrated a pattern of focusing on exploration and basic in-game activities. The survival analysis and log-rank test revealed a significant difference in the survival curves among the three clusters. The median survival time was longest for the Immersive Strategists (46 completed missions), followed by the Exploratory Strategists (21 missions) and the Trial-and-Error Explorers (1 mission). This indicates a clear difference in gameplay persistence across learner behavior sequence types. The findings suggest the need for dropout-prevention strategies, such as early-stage adaptation support and type-specific intervention strategies.