[Online] Lessons from COVID-19: Non-random Missing Data and Its Consequences

## How to join the webinar:

NOTE: You can join via your browser (no app download required). Use Chrome or Firefox.
Pre-register for the webinar:
https://www.bigmarker.com/neo4j/Data-Umbrella-Webinar

--------------------------------
Video Recording
--------------------------------
This event will be recorded and placed on our YouTube. We usually have it up within 24 hours of the event. Subscribe to our YouTube and set your notifications:
https://www.youtube.com/c/DataUmbrella/

--------------------------------
Time
--------------------------------
16:00 UTC
9 am PT / 12 pm ET / 7 pm EAT / 9:30 pm IST

## Speaker

Mitzi Morris

## Talk Level

Intermediate

## Pre-reqs

Familiarity with statistical modeling, ideally survey statistics and Bayesian inference.

## Prep Work

1. Modeling rates of disease with missing categorical data
2. Multiple Imputation of Missing Data

## Resources

https://github.com/rtrangucci/epi-missing-data

## Event

A fundamental challenge for survey and observational datasets is that not all records in the dataset are complete; key pieces of information may be missing.

In this talk I work through the models and methods from the paper
MODELING RACIAL/ETHNIC DIFFERENCES IN COVID-19 INCIDENCE WITH COVARIATES SUBJECT TO NON-RANDOM MISSINGNESS

They write:
In emergency situations, such as a surging pandemic, it is easy to see how the disease process itself may induce non-random missingness of covariates. For example, during a period of rapidly increasing caseloads, such as the Delta and Omicron surges of the COVID-19 pandemic, the overwhelming number of cases is likely to limit the ability of case investigators to collect data that are as detailed as those collected during lower-incidence periods. These differences may also be more pronounced when comparing wealthier and poorer jurisdictions with differential resources for case-finding and intervention.

Using the Stan language and CmdStanR interface, together with a simulated dataset of Covid-19 cases and population demographics, where age, gender, race/ethnicity, and neighborhood have varying degrees of missingness, we will demonstrate how different approaches produce different estimates of Covid-19 prevalence among key demographics.

This event is being co-promoted with R-Ladies NYC. R-Ladies NYC is part of a world-wide organization to promote gender diversity in the R community. We aspire to encourage and support women and gender minorities interested in learning and sharing their experiences in R programming by hosting a variety of events including talks, workshops, book clubs, data dives, and socials. (https://www.rladiesnyc.org/)

## CONNECTING WITH US

We invite you to follow us on our social networking sites to keep up to date on the latest news we will be sharing.

• Meetup: https://www.meetup.com/data-umbrella/
• Job Board: https://jobs.dataumbrella.org/
• Newsletter: https://dataumbrella.substack.com/
• Twitter: https://twitter.com/DataUmbrella
• YouTube: https://www.youtube.com/c/dataumbrella
• LinkedIn: https://www.linkedin.com/company/dataumbrella/?
• Instagram: https://www.instagram.com/data.umbrella/