Comparison of cognitive load on multi-question and very short-answer questions: an eye-tracking study.

• Grant number: 23/10189-5
• Support Opportunities: Scholarships in Brazil - Master
• Start date: August 01, 2023
• End date: April 30, 2024
• Field of knowledge: Health Sciences - Medicine - Medical Clinics
• Principal Investigator: Dario Cecilio Fernandes
• Grantee: Maria Gabriela Carneiro Queiroz
• Host Institution: Faculdade de Ciências Médicas (FCM). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• Associated research grant: 18/15642-1 - Investigating the impact of assessment strategies on medical students' learning, AP.JP

Abstract

Introduction: Short-answer questions (VSAQs) have been used to assess knowledge as they limit random guessing, promote application and integration of knowledge, and offer better information retention. Cognitive load assessment can be done through eye-tracking, providing accurate information about the cognitive processes involved. Objective: To compare the cognitive load, accuracy, and errors in response to multiple-choice questions (MCQs) and VSAQs. Hypotheses: Based on previous research, we do not anticipate significant differences in cognitive load between the two question types. Questions with higher complexity will induce a greater cognitive load compared to those with lower complexity. Additionally, different question models may exhibit similar cognitive loads, regardless of their complexity. Method: We will employ a randomized experimental study with a crossover design. A total of 40 questions (20 MCQs and 20 VSAQs) will be administered within a single session. The questions will be adapted from medical residency exams and the progress test, encompassing diverse subject areas and complexity levels. Participants will consist of 5th and 6th-year medical students, while students from other years and healthcare fields will be excluded. Data Analysis: The analysis will focus on the number of fixations and revisits in each area of interest using mixed linear models. We will conduct three distinct models: number of fixations, duration of fixations, and number of revisits. Covariates such as complexity level, accuracy and errors, word count, key information, and response time will be included. Post-hoc analyses will be performed, adjusting for pairwise comparisons using the Bonferroni method.