Predicting student achievement in learning management systems by log data analysis

Previous attempts to understand the activity in learning management systems have failed to provide log data analysis methods that significantly predict student achievement. The number and frequency of keystrokes and mouse clicks have little to say about cognitive activities. On the other hand, self-reporting may include a sufficiently accurate insight into cognitive activities, but this insight is blurred by learners' distorted self-perception during intensive cognitive activities. This study proposes a linear model that includes previous knowledge and log file-extracted online activity as predictors of student achievement. The model displayed a good fit with data collected in three different cases (CFI up to .98, RMSEA down to 0.028) and it explained R2 = approx. 0.50 of the variance in learning outcome. In conclusion, the relationship between log data and cognitive activities is discussed, and design recommendations for learning management systems are drawn.