Until recently, discussion of causal effects in empirical research have been limited to the context of randomized control trials. The reason being that reliable causal inferences cannot be guaranteed in observational or non-randomized experimental research, where, absent a randomization procedure, the relation between “cause” and “effect” variables of interest is likely to be confounded.
However, avoiding discussion of causality in these types of research is based on a harmful conflating of the means of research with the ends: although randomized experiments are not always possible due to practical or ethical reasons, the core goal of much research remains the investigation of causal effects.
Interventionist causal inference gives us a framework by which to describe and investigate causal effects in observational and non-randomized experimental settings.
Typically, interventionist approaches take on one of two complimentary flavors:
- the potential outcomes approach, which views causal inference as a missing data problem
- the structural causal model approach, which typically makes use of graphical models also known as Bayesian networks or Directed Acyclic Graphs
Topics of research within the department span both approaches. Examples include:
- the use of statistical network models for exploratory causal investigation: https://github.com/ryanoisin/SEset
- Imputation methods for individual causal effects: https://www.iops.nl/students/current-iops-students/mingyang-cai/
- Email: email@example.comPhone: 030 253 5857Researcher
- Email: firstname.lastname@example.orgPhD Candidate
- Email: email@example.comProfessor
- Email: firstname.lastname@example.orgPhone: 030 253 4911Associate Professor
- Email: email@example.comProfessor
- Email: firstname.lastname@example.orgPhD Candidate
- Email: email@example.comAssistant Professor