|
Abstract:
|
David Freedman studied causal inference for nearly three decades. He scrutinized methods used by political scientists, economists, sociologists, and epidemiologists, asking, "Do these methods yield rigorous causal inferences from non-experimental data?" The iterated answer: causation was built into the methods' assumptions, not inferred from the data. A new volume, "Statistical Models and Causal Inference: A Dialogue with the Social Sciences," collects twenty of David's papers. These cover: foundational issues, statistical inference in social science and epidemiology, and the (mis)behavior of regression when the usual regression assumptions are false. I will give highlights of perhaps eight of those papers.
|
