Quantifying virtual education preferences in medical sciences during crises: a discrete choice analysis.

introductionThe COVID-19 pandemic and other global emergencies accelerated the shift to virtual learning in higher education, ensuring instructional continuity but also revealing significant challenges. Disparities in technology access, instructional design, and student engagement highlighted critical gaps in preparedness. Understanding the factors that shape students’ preferences for virtual education is essential for building resilient, equitable, and adaptable systems.

methodsA discrete choice experiment (DCE) was conducted to identify determinants of student preferences for virtual learning. Six attributes were examined: technology infrastructure, support structure, e-learning method, student characteristics, teacher characteristics, and evaluation. Attribute levels were derived from the literature review and qualitative analysis. Data were collected via an online questionnaire, including demographics and 15 choice scenarios designed using orthogonality and D-efficiency principles. Sampling was stratified by faculty and field of study, with non-random recruitment continuing until 300 completed responses were obtained. Data were analyzed using descriptive statistics and nominal logistic regression in SAS JMP.

resultsParticipants had a mean age of 22.3 years (SD = 3.38); most were female (66.3%), single (93.1%), and enrolled in bachelor’s (48.2%) or M.D. programs (47.2%). Overall, digital, economic, and educational preparedness was adequate. Regression analysis indicated that the e-learning method and technology infrastructure were the most influential factors, followed by evaluation methods and teacher characteristics. Blended learning significantly increased selection likelihood compared to synchronous learning (OR = 1.85), while asynchronous learning showed a smaller positive effect (OR = 1.21). Enhanced tools and technologies increased preferences relative to basic platforms (OR = 1.40). Communication competency and virtual exam quality had lower odds compared to time-management ability (OR = 0.80) and exam type (OR = 0.88). Demographic factors were generally non-significant, but prior virtual learning experience, academic/technical requirements, and economic status had meaningful effects.

conclusionBlended learning models, robust technology infrastructure, and context-sensitive strategies are critical for effective and resilient virtual education. Faculty development in digital pedagogy, infrastructure investment, and adaptive course design aligned with learner profiles is essential for improving student experiences and outcomes. Future studies should explore longitudinal effects and broader student populations to validate these findings.