Causal Ecological Inferences

This note observes that the ecological inference problem is very closely related to the problem of estimating causal effects. Due to the common mathematical structure, instrumental variables techniques solve the ecological inference problem under essentially the same set of conditions under which they allow researchers to draw conclusions about causality. More generally, carefully addressing the issue of causality can be sufficient to generalize from aggregate data to the behavior of individuals. ∗Previous versions circulated under the title “Ecological Inference with Instrumental Variables.” I have benefitted from helpful conversations with Pablo Montagnes and Philipp Tillmann. I am also grateful to David Austen-Smith and Tim Feddersen for encouraging me to write up this note, and to Derek Neal for teaching me that regressions are always about the residual. Correspondence can be addressed to the author at MEDS Department, Kellogg School of Management, 2211 Campus Drive, Evanston, IL 60208, or by email: [email protected].