Automating Guidance for Students' Chemistry Drawings

Generative educational assessments such as essays or drawings allow students to express their ideas. They provide more insight into student knowledge than most multiplechoice items. Formative guidance on generative items can help students engage deeply with material by encouraging students to effectively revise their work. Generative items promote scientific inquiry by eliciting a variety of responses and allowing for multiple correct answers, but they can be difficult to automatically evaluate. We explore how to design and deliver automated formative guidance on generative items requiring precollege students to draw the arrangement of atoms before and after a chemical reaction. The automated guidance is based on a rubric that captures increasing complexity in student ideas. Findings suggest that the automated guidance is as effective at promoting learning as teacher-generated guidance, measured both by immediate improvement on the revised item and preto post-test improvement on a near-transfer item. Immediate and delayed delivery of automated guidance are equally effective for promoting learning. These studies demonstrate that embedding automated guidance for chemistry drawings in online curricula can help students refine their understanding. Providing automated guidance can also reduce the time teachers spend evaluating student work, creating more time for facilitating inquiry or attending to the needs of individual students.