Students are asked an open-ended question relating to a topic of interest. Experts on this topic are then given two student responses, and they judge which of the two displays a better conceptual understanding of the concept without using a specified scoring rubric. This process is repeated several times, and a rank/scale is created for the responses from “worst” to “best.” Using statistical modeling, a z score parameter and standard error is created for the responses. Studies have found that comparative judgement displays internal consistency, reliability, and validity in varying (mathematical) contexts.