Confidence plays a central role in metacognitive regulation, yet most educational research measures it through explicit self-report. Larry Michaelsen and early adopters of Team-Based Learning (TBL) asserted that having students spread points across the options in the multiple-choice questions during the individual Readiness Assessment Test (iRAT) is an implicit measure of confidence.  Very few, if any, studies examine whether this point-spreading technique actually correlates with student confidence. This study examines whether confidence is reflected in students’ response behavior during assessment. Using data from 380 students in an undergraduate TBL courses, we required students to distribute points across multiple answer options and to report their confidence for each question. As hypothesized, we found that ordered logistic regression results indicate that point-spreading entropy is strongly associated with stated confidence: a one-unit increase in entropy corresponds to an approximately 22 percentage-point decrease in the probability of reporting higher confidence, even after controlling for course grade, gender, age, and accumulated academic credits. There is more to the story, however. Exploratory cluster analysis reveals discernible student subgroups, based on their point-spreading behavior and their academic performance. The groups identified include students who express elevated behavioral uncertainty despite average performance and students who exhibit high confidence and decisiveness despite weaker academic outcomes. The existence of clusters suggest that groups of students across observable characteristics might perform different in point dispersion instruments. Taken together, these findings suggest that students’ point-spreading behavior during retrieval contains information related to their reported confidence. In instructional contexts such as Team-Based Learning, point-spreading tasks may have heterogeneous effects in different subgroups of the student population.