Two metamodels of causal effects

Area of Science:

• Causal inference
• Quantitative methodology
• Statistical modeling

Background:

• Current methods for defining and measuring quantitative causal effects lack comprehensive frameworks.
• The role of contrafactual concepts in causal inference requires clarification.
• Generalization and statistical adjustment in causal analysis can be improved.

Purpose of the Study:

• To propose two novel metamodels, Model S and Model V, for the definition, measurement, and generalization of quantitative causal effects.
• To define and differentiate between effect change, total change, and remainder change.
• To elucidate the role of contrafactual concepts in causal inference and propose a non-contrafactual approach.

Main Methods:

• Development of two distinct metamodels: Model S (effect as part change in score) and Model V (effect as part change in variance).
• Definition of total change and remainder change in relation to effect change.
• Analysis of the integration and interpretation of contrafactual concepts within the proposed metamodels.

Main Results:

• Model S and Model V provide a structured approach to defining and measuring causal effects.
• Remainder change is attributed to factors other than the cause, while total change encompasses both effect and remainder changes.
• Contrafactual definitions are deemed inadequate; a non-contrafactual interpretation of statistical adjustment is advocated.

Conclusions:

• The proposed metamodels offer a robust framework for quantitative causal effect analysis.
• A non-contrafactual approach to causal inference and statistical adjustment is more appropriate and sufficient.
• These metamodels enhance the clarity and generalizability of causal effect research.