A Practical Approach to Teaching Type I Error in Hypothesis Testing Using False Positives in COVID-19 Testing
Academic Article
Overview
Overview
Abstract
Hypothesis testing is a fundamental topic taught in introductory statistics courses across disciplines. Despite its importance, students often struggle with misconceptions, particularly about its probabilistic nature and the implications of uncertainty. Type I errors (false positives) are typically emphasized in textbooks to avoid making false claims about the presence of an effect when there is none thus rejecting a true null hypothesis, the impact of Type I errors (false positives) can vary by context and may lead to significant economic consequences. To address this gap, we propose using COVID-19 diagnostic testing as a relatable example to illustrate the risks and consequences of Type I errors. False positives in testing can result in excessive absenteeism, anxiety, waste of personal protective equipment (PPE), and waste of human medical staff resources and facilities, delays in surgical procedures, hospital stays, etc. They can result in inappropriate treatments, introduce noise into crucial clinical data. Patients who falsely test positive might be less likely to avoid exposure to infected individuals, believing they have immunity, and possible forego vaccinations. Additionally, uninfected persons who receive false-positive results and are hospitalized are commonly moved into COVID-19 units where they have a higher risk of exposure to infected individuals. This is especially alarming for those who are elderly or ill, a risk of serious complications or even death. There are also the economic costs of lost wages for employees and lost productivity for companies. Possibly having far reaching consequences throughout the supply chain. This research approach fosters deeper engagement and understanding by connecting abstract statistical concepts to students’ real-world experiences. The effectiveness of this method was evaluated through pre-treatment and post-treatment surveys in introductory statistics courses.
