Adjusting For Non-Ignorable Selection Bias In Surveys
Brady West and Rebecca Andridge
January 29, 2026 – 9am-1pm EST, 6050 ISR-Thompson/Zoom
Please join as we kick off the 2026 PDHP workshop series with Adjusting For Non-Ignorable Selection Bias in Surveys, presented by Rebecca Andridge (The Ohio State University’s College of Public Health) and Brady West (University Of Michigan’s Institute for Social Research). This workshop introduces new indices for measuring the non-ignorable selection bias in survey estimates (means, proportions, and regression coefficients — called SMUB, MUBP, and MUBNS, respectively), including techniques for adjusting population inferences based on these estimates. The presenters will provide concrete descriptions of model assumptions and data requirements necessary for implementing these indices. Attendees will receive hands-on practice with computation and interpretation of these indices, using real survey datasets and annotated R code, as well as an audience Q&A session about the use of these indices in practice.
Topics include:
- Introduction of the SMUB, MUBP, and MUBNS indices for measuring non-ignorable selection bias in survey estimates
- Model assumptions and data requirements for implementing these indices
- Hands-on practice with computation and interpretation of these indices using R
- Q&A about the use of these indices in practice
Video:
Workshop video is available here
Slides:
Slides are available for download hereThis linked document/file has unknown accessibility or has not been evaluated
Lab Materials:
Download the lab materials hereThis linked document/file has unknown accessibility or has not been evaluated
Recommended Software:
The workshop will provide examples in R, as configured below:
––R Studio (strongly recommended)
––R packages can be installed using the following code
# install R packages
install.packages(c("boot", "coda", "magrittr", "MASS", "MCMCpack", "mnormt", "msm", "nlme", "mvtnorm", "tidyverse", "survey", "knitr"))