Search papers, labs, and topics across Lattice.
The paper introduces ExAtlas, a framework that maps social science experiments by linking, reconciling conflicts, and bridging gaps between studies based on treatment and outcome similarity. ExAtlas composes effects from prior studies to predict target study outcomes, identifying consistent evidence, conflicts requiring moderator explanations, or gaps necessitating new experiments. The framework achieves 98.6% accuracy in recovering effect direction on locally supported held-out targets and generates plausible bridge experiments and useful conflict explanations.
Social science experiments contain far more latent structure than we realize, and ExAtlas reveals it by automatically linking studies, reconciling conflicts, and suggesting new experiments.
Social and behavioral science runs thousands of experiments each year, yet their findings rarely accumulate into a coherent map of what is known, what conflicts, and what remains missing. We introduce ExAtlas, a framework for turning an archive of experiments into an atlas: a structured map in which studies link, conflict, or leave bridgeable gaps. Given a target study, ExAtlas searches for prior studies that are locally close in treatment and outcome space and asks whether their observed effects can be composed to predict the target effect. This yields three cases. If the composition succeeds and agrees with the observed result, ExAtlas links the target to consistent prior evidence. If composition succeeds but disagrees, ExAtlas reconciles the conflict and proposes candidate moderators or higher-level theories that could explain it. If composition fails, ExAtlas proposes bridge experiments to close the gap. We provide an error bound for composition under local smoothness of the treatment-effect surface. On held-out targets certified as locally supported, ExAtlas recovers effect direction in 98.6% of cases. Human evaluations further suggest that its proposed bridge experiments are plausible and exhibit connectedness, and that its conflict explanations are useful for theory generation. These results suggest that the archive of social experiments contains more latent structure than current practice extracts -- and that making this structure explicit can guide both future theory and future experimentation.
