Online talk by Prof. Jeffrey S. Morris (University of Pennsylvania)
The novel virus SARS-CoV-2 has produced a global pandemic, forcing doctors and policymakers to “fly blind,” trying to deal with a virus and disease they knew virtually nothing about. Sorting through the information in real time has been a daunting process—processing data, media reports, commentaries, and research articles. In the USA this is exacerbated by an ideologically divided society that has difficulty with mutual trust, or even agreement on common facts. The skills underlying statistical data scientists are central to this knowledge discovery process, filtering out biases, aggregating disparate data sources together, dealing with measurement error and missing data, identifying key insights while quantifying the uncertainty in these insights, and then communicating the results in an accessible balanced way. As a result, we have had a central role to play in society to bring our perspective and expertise to bear on the pandemic to help ensure knowledge is efficiently discovered and put into practice. I have authored a website and blog
covid-datascience.com that represents my own person efforts to disseminate information I have found reliable and insightful regarding the pandemic, accounting for subtle scientific and data analytical issues and uncertainties about our current knowledge, and seeking to filter out political and other subjective biases.
Using experiences with the
covid-datascience.com blog as a backdrop, I will highlight how statistical and data scientific issues have been central in understanding the emerging knowledge in the pandemic. I will discuss various broad issues I have seen impede the knowledge discovery process, including subjective bias causing individuals to ignore some information and magnify others, viral misinformation spread on social media platforms, danger of rushed and inadequately reviewed scientific studies, conflating of political concerns and scientific messaging, and incomplete and messaging from scientific leaders to the broader community. I will discuss these concepts in various specific contexts, vaccine safety and efficacy, durability of immune protection and risk of reinfections or breakthrough infections, and the emergence of variants of concern and how this affects the pandemic moving forward. I will point out several of the key epidemiological biases and data fallacies that have been evident in the dynamic nature of the pandemic that have led to much confusion and fed misinformation and misunderstandings. I will focus in particular on Simpson’s paradox, that has manifest in various ways throughout the pandemic especially in relation to the structured vaccine rollout first focusing on older and more vulnerable populations and then later vaccinating the younger and healthier cohorts. Throughout, I will mention the political and sociological dynamics that have shaped our ability (and inability) to reach consensus on emerging knowledge and put it to work for effect (or ineffective) pandemic management.
Jeffrey S. Morris (University of Pennsylvania)