Sensor-based methods such as eye tracking are promising for user modeling in adaptive educational systems, as they enable continuous, fine-grained assessment of learners’ learning processes. However, implementing gaze-based user models requires selecting from a large set of available gaze measures, posing a challenge for adaptive-system designers. To inform this design process, we conduct an exploratory analysis of the associations between gaze measures, learner engagement, and learning outcomes in a video-based learning context. An analysis of data collected from 18 learners indicates that fixation dispersion and fixation rate on active areas of interest are indicators of engagement, while pupil diameter is an indicator of learning outcomes. To support further research on gaze-based user modeling and adaptation, we introduce EnGazement, a dataset comprising over 1.3 million processed eye-tracking data points, enriched with subjective and objective measures of engagement and learning.