Online Program Home
My Program

Abstract Details

Activity Number: 534 - Contributed Poster Presentations: Section on Statistics in Epidemiology
Type: Contributed
Date/Time: Wednesday, August 1, 2018 : 10:30 AM to 12:20 PM
Sponsor: Section on Statistics in Epidemiology
Abstract #330930
Title: Ggdag and Confoundr: R Packages for Causal Inference with DAGs and Data
Author(s): Malcolm Barrett*
Companies:
Keywords: DAGs; confounding; causal inference; R; tidy data; data visualization
Abstract:

One of the primary issues in observational studies is the minimization of bias in the estimator of effect. Since early contributions by researchers like Pearl and Greenland, practices for analyzing directed acyclic graphs (DAGs) and detecting confounders have received much attention. ggdag and confoundr are R packages for visualizing and analyzing DAGs and detecting and visualizing confounders, respectively. The ggdag package makes it easy to produce visualizations based on algorithms in the dagitty package using tidy data principles common in the tidyverse ecosystem of R packages, including detecting adjustment sets, colliders, and pathways of bias that arise from adjustment, as well as functionality for presenting and analyzing common structures of bias. The confoundr package works with many different models to apply both classic methods, like change-in-estimate techniques, and more modern approaches, such as modified lasso regression (e.g. partially applying shrinkage to potential confounders), to detect confounders. Additionally, confoundr includes functionality for visualizing confounding and working with estimates with tidy data principles in mind.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2018 program