Themed session: Machine Learning & Spatial Techniques for Causal Inference

The last three years have seen a new emergence in the use of spatial information to conduct causal inferential analyses. These techniques - frequently seeking to create "treated" and "control" locations analogous to clinical trials - have now been used to examine topics including the effectiveness of programs designed to reduce deforestation, improve healthcare in impoverished communities, and mediate impacts of infrastructure on biodiversity. Most recently, machine learning techniques have begun to be applied to enable understanding of what geographic contextual factors drive the effectiveness - or lack thereof - of interventions. This session invites papers around the topic of using spatial data for causal inference, including:

  • Challenges to the use of spatially-explicit data for causal inference (including autocorrelation and spillover effects).
  • Machine Learning methods for identified studies.
  • Simulation and other approaches to uncertainty and Error Propagation for causal inference.
  • ANN, CNN, DID, or other novel approaches to causal inference being applied in spatial contexts.
  • Case studies with causal inferential foci.

How to apply

Please apply as normal, but select "Machine Learning & Spatial Techniques for Causal Inference" under the topics on the submission page. Although we can't guarentee all accepted submissions a place in the session of their choice, we will do our best to group papers thematically.

Further information

For further information, please contact the session chair: Dan Miller Runfola.