2026 ASEE Annual Conference & Exposition

Understanding how test anxiety shapes students’ learning behaviors is critical for improving assessment design and supporting student well-being in engineering education. This study investigates how both trait anxiety (baseline, self-reported) and state anxiety (self-reported after each quiz) relate to study timing, study amount, and performance outcomes in a large undergraduate STEM course. The course employed repeated low-stakes quizzes over a semester, with varied testing environments (bring-your-own-device vs. in a computer-based testing facility). Our goal was to understand not only whether anxiety correlates with outcomes, but how it affects when and how much students study, and whether these effects carry over between quizzes.

We analyzed data from 248 students, integrating survey responses with detailed behavioral logs. We examined four categories of study behavior from practice quizzes:
* Percent of available questions attempted,
* Accuracy on attempted questions,
* Accuracy relative to the total question pool, and
* Attempt rate per practice quiz.

We also modeled study timing variables for exam preparation within 24 hours, within 48 hours, and greater than 48 hours before the exam, and included a binary variable for testing environment (whether students took the quiz in a computer-based testing facility or elsewhere). Our approach integrates repeated measures of self-reported anxiety with behavioral data, using ordinary least squares and linear mixed effects regression to examine how emotional responses influence study behavior both within and across quizzes.

Cross-sectional analyses revealed that higher state anxiety was associated with lower question-attempt rates and accuracy on practice quizzes. Trait anxiety similarly predicted reduced engagement, suggesting stable affective dispositions influence study behavior.

Lagged temporal analyses showed higher anxiety reported after one quiz significantly predicted earlier preparation and less cramming on studying for the next quiz. This suggests that anxiety may prompt students to adopt more proactive study habits. However, students in computer-based testing settings showed reduced early preparation rates, indicating that testing modality plays an additional role in the effect of test anxiety.

These findings demonstrate that students’ anxiety not only reflects their test experiences but also influences how they prepare for future assessments, particularly in terms of timing and effort. By capturing both immediate and delayed effects of anxiety across different testing contexts, this study provides evidence that emotional experiences can drive adaptive academic behavior, not just performance outcomes. Our results offer practical insights for instructors seeking to create testing environments that support sustained engagement and student well-being.