Discussion: View Thread

Dear Members of the ASA Statistical Consulting Section, 

On behalf of the AAAS Section on Statistics (U) and the AAAS Section on General Interest in Science and Engineering (Y), in collaboration with AAAS SciLine, we would like to invite you to a joint Zoom webinar on

Seeing Through Epidemiologic Fallacies: How Statistics Safeguards Scientific Communication in a Polarized Era

Friday, October 17 - 2:00 to 3:00 p.m. ET

Presenter: 
Jeffrey S. Morris
Perelman School of Medicine
University of Pennsylvania
The Wharton School

Moderator: 
William K. Hallman
Rutgers University

Welcome and closing: 
Tori Espensen
Media Training Manager
AAAS SciLine

The full abstract is reported after the signature line.

Please register here: https://aaas.zoom.us/webinar/register/WN_X2K9CTjSR3qAxqdxq_FRHQ

We look forward to your participation. 

Please feel free to  share this invitation with interested colleagues

Best,
Michele Guindani

Abstract

Observational data underpin many biomedical and public-health decisions, yet they are easy to misread, sometimes inadvertently, sometimes deliberately, especially in fast-moving, polarized environments during and after the pandemic. This talk uses concrete COVID-19 and vaccine-safety case studies to highlight foundational pitfalls: base-rate fallacy, Simpson’s paradox, post-hoc/time confounding, mismatched risk windows, differential follow-up, and biases driven by surveillance and health-care utilization.

Illustrative examples include:

1. Why a high share of hospitalized patients can be vaccinated even when vaccines remain highly effective.
2. Why higher crude death rates in some vaccinated cohorts do not imply vaccines cause deaths.
3. How policy shifts confound before/after claims (e.g., zero-COVID contexts such as Singapore), and how Hong Kong’s age-structured coverage can serve as a counterfactual lens to catch a glimpse of what might have occurred worldwide in 2021 if not for COVID-19 vaccines.
4. How misaligned case/control periods (e.g., a series of nine studies by RFK appointee David Geier) can manufacture spurious associations between vaccination and chronic disease.
5. How a pregnancy RCT’s “birth-defect” table was misread by ACIP when event timing was ignored.
6. Why apparent vaccine–cancer links can arise from screening patterns rather than biology.
7. What an unpublished “unvaccinated vs. vaccinated” cohort (“An Inconvenient Study”) reveals about non-comparability, truncated follow-up, and encounter-rate imbalances, despite being portrayed as a landmark study of vaccines and chronic disease risk in a recent congressional hearing.

I will outline a design-first, transparency-focused workflow for critical scientific evaluation, including careful confounder control, sensitivity analyses, and synthesis of the full literature rather than cherry-picked subsets, paired with plain-language strategies for communicating uncertainty and robustness to policymakers, media, and the public. I argue for greater engagement of statistical scientists and epidemiologists in high-stakes scientific communication.