Partial identification for discrete data with nonignorable missing outcomes

Daniel Daly-Grafstein, Paul Gustafson

Nonignorable missing outcomes are common in real world datasets and often require strong parametric assumptions to achieve identification. These assumptions can be implausible or u…