An Introduction to Handling Missing Data in R 

Missing data is a common complication when doing applied statistics and data science. In this talk, I will review the various types of missingness that can arise, and survey the most common ways of accounting properly for missing data when conducting statistical analyses. I will talk about some tools for summarizing missingness, managing it, and analyzing data in its presence. The emphasis will be on basic ideas and practical tools for handling missing data in R. I will discuss the taxonomy of missingness (missing at random, missing completely at random, not missing at random), strategies for analysis when missingness is a problem (complete case analysis, available case analysis, imputation, multiple imputation, and full information maximum likelihood estimation), and I will provide some practical recommendations. Examples with R code will be sprinkled through the presentation. Handling missing data properly can be very challenging, but this talk will give you the tools and basic ideas to get started on your own and to know when to seek help.