As computer-based tests become more common, there is a growing wealth of metadata related to examinees' response processes, which include solution strategies, concentration, and operating speed. One common type of metadata is item response time. While response times have been used extensively to improve estimates of achievement, little work considers whether these metadata may provide useful information on social-emotional constructs. This study uses an analytic example to explore whether metadata might help illuminate such constructs. Specifically, analyses examine whether the amount of time students spend on test items (after accounting for item difficulty and estimates of true achievement), and difficult items in particular, tell us anything about the student's academic motivation and self-efficacy. While results do not indicate a strong relationship between mean item durations and these constructs in general, the amount of time students spend on very difficult items is highly correlated with motivation and self-efficacy. The implications of these findings for using response process metadata to gain information on social-emotional constructs are discussed.