You are cordially invited to a webinar co-sponsored by CHANCE magazine and R Govys. Feel free to share with friends and colleagues.
Date: March 19, 3 PM (ET)
Registration (free): https://bit.ly/ChanceMarch. Click or tap if you trust this link." data-olk-copy-source="MessageBody">https://bit.ly/ChanceMarch
Title: Outliers & Alcohol: Revisiting Wilkinson's 2001 Graphic with Updated State Mortality Data, R Graphics, Shrinkage, and Structural Equation Modeling
Author: Eugene Gallagher
Abstract
Leland Wilkinson (2001) created a graphic of 1986 state liver-disease mortality rates versus alcohol consumption using marginal boxplots. After teaching boxplots with this graphic for decades, I updated it using 2023 data and modern R tools (ggplot2 and ggExtra), which were inspired by Wilkinson's Grammar of Graphics. I also analyzed alcohol-related automobile fatality rates. States with small populations often appear as outliers; this effect was evaluated using Fay–Herriot shrinkage. Finally, a structural equation model implemented in the R package lavaan assessed the relative importance of variables affecting liver disease and alcohol-related fatalities.
Bio
Eugene Gallagher earned his M.Sc. and Ph.D. in Oceanography at the University of Washington, where he designed controlled field experiments to test benthic succession theory. At the Marine Biological Laboratory in Woods Hole, working with Dr. Judy Grassle, he studied the ecological effects of benthic pollution. Gallagher helped found UMass Boston's doctoral program in Environmental Science and taught oceanography and statistics for more than 40 years. In retirement, he continues to study pollution impacts from Boston Harbor to the deep sea using R-based multivariate and Bayesian models.
------------------------------
Wendy Martinez
Pronouns: she, they
Senior Mathematical Statistician for Data Science
US Census Bureau
------------------------------