In order to improve the feedback an intelligent tutoring system provides, the grading engine needs to do more than simply indicate whether a student gives a correct answer or not. Good feedback must provide actionable information with diagnostic value. This means the grading system must be able to determine what knowledge gap or misconception may have caused the student to answer a question incorrectly. This research evaluated the quality of a rules-based grading engine in an automated online homework system by comparing grading engine scores with manually graded scores. The research sought to improve the grading engine by assessing student understanding using knowledge component research. Comparing both the current student scores and the new student scores with the manually graded scores led us to believe the grading engine rules were improved. By better aligning grading engine rules with requisite knowledge components and making revisions to task instructions the quality of the feedback provided would likely be enhanced.

This experimental study examined the effects of an instructor’s face threat mitigation tactics on student self-efficacy for learning and perceived emotional support from the instructor in a written feedback setting. Participants (N = 401) were randomly assigned to one of four feedback scenarios in which level of face threat mitigation and instructor age and status were manipulated. Student grade orientation and state feedback apprehension were measured prior to being exposed to the feedback scenario. Results indicate that high face threat mitigation is positively associated with student self-efficacy for learning and perceived emotional support from the instructor. Results also revealed that state feedback apprehension predicts self-efficacy for learning and perceived emotional support from the instructor. Grade orientation predicted self-efficacy for learning but did not significantly predict perceived emotional support from the instructor providing feedback. Finally, scenarios manipulated for instructor age and status did not significantly differ in self-efficacy for learning or perceived emotional support from the instructor. Implications regarding theory, the measurement of feedback apprehension, and student-instructor communication are discussed.