Main Content

Causal Inference and Statistical Workflows

“Correlation does not imply causation.” Students likely hear this sentence in every introductory lecture. However, instead of formal methods for justifying causal inferences outside of experiments, there is often a certain “taboo” against causal language in psychology. We research methods of causal inference that allow theoretical assumptions to be presented transparently and derive statistical procedures that enable causal conclusions conditional on certain assumptions. In particular, we focus on how results from different populations or cultures can be compared and how we can use theoretical models to improve statistical methods.

Image of a computational workflow for cultural evolution that combines abstract and real-world data
From Deffner et al., 2024, Proceedings of the National Academy of Sciences, under a Creative Commons Attribution 4.0 International License
Arrow diagrams to illustrate the transferability of causal effects to populations
From Deffner et al., 2022, Advances in Methods and Practices in Psychological Science, under a Creative Commons Attribution 4.0 International License

Publications

  • Deffner, D., Fedorova, N., Andrews, J. & McElreath, R. (2024). Bridging theory and data: A computational workflow for cultural evolution. Proceedings of the National Academy of Sciences, 121(48).
  • Sterner, P., Pargent, F., Deffner, D. & Goretzko, D. (2024). A Causal Framework for the Comparability of Latent Variables. Structural Equation Modeling, 1-12.
  • Deffner, D., Rohrer, J.M & McElreath, R. (2022). A causal framework for cross-cultural generalizability. Advances in Methods and Practices in Psychological Science, 5(3).