This document discusses two methods for assessing student engagement and affect in educational games: quantitative field observations and automated detectors. Quantitative field observations involve experts observing students' engagement and affect through peripheral vision during gameplay and noting behaviors like collaboration, boredom, and frustration. Automated detectors use models to assess engagement and affect in real-time from student behavior within software, without sensors. The author's lab has used these methods to compare engagement between games and intelligent tutoring systems, and to study how social behavior influences affect dynamics in games. Both methods provide ecologically valid assessments of student disengagement and experiences of specific emotions during learning.