A causal research pipeline and tutorial for psychologists and social scientists

Area of Science:

• Psychology and Social Science Research Methodology
• Causal Inference and Machine Learning

Background:

• Causality is crucial for scientific understanding but often neglected in psychology and social sciences.
• A gap exists between causal theories and standard research practices, leading to under-specification.
• Growing recommendations advocate for adopting causal approaches in research.

Purpose of the Study:

• To reformulate research approaches in psychology to harmonize causal theories with the research pipeline.
• To present a novel process for developing, validating, and specifying theories using causal discovery and machine learning.
• To enable the estimation of causal effects from data using semi-parametric machine learning methods.

Main Methods:

• Incorporating causal discovery and machine learning for theory development and formal specification.
• Reducing complex theoretical models to fundamental submodels relevant to specific hypotheses.
• Employing semi-parametric machine learning for estimating causal effects when quantities are estimable.

Main Results:

• A new research pipeline is presented that facilitates causal inquiry.
• The process encourages transparent, unambiguous mathematical representation of theories.
• It bridges the theory-to-model gap, reducing under-specification issues.

Conclusions:

• The proposed pipeline yields causally meaningful and reproducible results.
• It supports scientific inquiry compatible with testing causal theories.
• Demonstrated through didactic examples with real-world data.