We develop new quasi-experimental tools to understand algorithmic discrimination and build non-discriminatory algorithms when the outcome of interest is only selectively observed. These tools are applied in the context of pretrial bail decisions, where conventional algorithmic predictions are...

Studies of racial discrimination often condition on endogenous measures of race or on earlier decisions that might themselves be affected by discrimination. We develop quasi-experimental tools for estimating the impact of racial misclassification on measures of unwarranted disparity, and for...

Black children in the U.S. are twice as likely as white children to spend time in foster care. Such racial disparities raise concerns of discrimination, but might also reflect differences in the underlying need for intervention. This paper estimates unwarranted disparities (UDs)--racial...

A recent econometric literature shows two distinct paths for identification with shift-share instruments, leveraging either many exogenous shifts or exogenous shares. We present the core logic of both paths and practical takeaways via simple checklists. A variety of empirical settings illustrate...

The ISCHEMIA Trial randomly assigned patients with ischemic heart disease to an invasive treatment strategy centered on revascularization with a control group assigned non-invasive medical therapy. As is common in such "strategy trials," many participants assigned to treatment remained untreated...