Related Experiment Video
Updated: Aug 14, 2026
Improving Student Outcomes with an Adaptable Molecular Cloning Course-Based Undergraduate Research Experience
Published on: November 15, 2024
The Impact of Changing Answers on Assessment Performance in a Doctoral-Level Anatomy Class
Mark Wilhelm1,2,3,4,5,6, Dominique Reynolds1,2,3,4,5,6, Craig Wassinger1,2,3,4,5,6
1Mark Wilhelm is the Associate Professor, Admissions Coordinator, Hybrid DPT Program at the Medical University of South Carolina, 151-A Rutledge Ave., Charleston, SC 29425 (wilhelmm@musc.edu). Please address all correspondence to Mark Wilhelm.
Introduction:
It is essential that students perform optimally on multiple-choice examinations such as the National Physical Therapy Examination, which is comprised entirely of multiple-choice questions. Students are frequently advised to avoid changing initial answers on multiple-choice examinations. This belief is reinforced by faculty and student misconceptions. This study investigated the frequency and impact of answer changes among physical therapy students completing anatomy examinations on a computer-based platform.
Review Of Literature:
Prior studies show that answer changing is generally beneficial. However, most studies were conducted decades ago, and only a few were found that were conducted in health care education. Almost all previously published studies used paper examinations and were limited to 3 main answer-changing patterns, whereas computer-based testing can more accurately identify patterns in how students change their answers.
Subjects:
Subjects include 174 first-year hybrid Doctor of Physical Therapy students.
Methods:
In this retrospective cohort study, participants completed 4 computer-based multiple-choice examinations. Exam snapshot data from ExamSoft were imported into MATLAB, which categorized items into 1 of 7 answer-changing patterns. Examination scores were calculated with and without answer changes. Analyses included descriptive statistics, a generalized linear mixed model, a linear mixed-effects model, Pearson correlations, and a partial correlation.
Results:
Across 30,133 answered items, 11.3% involved an answer change, with an average of 94.07% of students changing answers. The most common change was from incorrect to correct, representing 52.41% of changed items, whereas only 12.8% of changes were from correct to incorrect. Examination scores allowing answer changes were, on average, 5% higher than scores based on initial responses. Lower-performing students changed more answers, showing greater total benefit, while higher-performing students benefited more per question.
Discussion And Conclusion:
Students frequently changed answers on computer-based examinations, and answer changing generally improved performance. Although some students experienced detrimental changes, beneficial changes occurred far more often. Findings suggest that discouraging answer changing may disadvantage students, and educators should promote informed reconsideration of responses during multiple-choice testing.
