Causal diagrams and change variables.

Area of Science:

• Causal inference
• Statistical modeling
• Methodology

Background:

• The assumption that 'true change' between two time points has causal implications is common but lacks formal theoretical basis.
• Existing research often relies on intuitive understanding rather than rigorous justification for change variable causality.

Purpose of the Study:

• To formally investigate the causal properties of a 'change variable' using causal directed acyclic graphs.
• To critically evaluate the theoretical underpinnings of assuming causality for variables representing change over time.

Main Methods:

• Utilized causal directed acyclic graphs (DAGs) to represent and analyze competing causal structures.
• Examined the role and nature of a 'change variable' within formal causal frameworks.

Main Results:

• A 'change variable,' representing the difference between two time points, is identified as a derived variable.
• The study demonstrates that 'true change' does not possess causal efficacy and is not a variable of direct causal interest.

Conclusions:

• 'True change' is not a fundamental variable in the physical world.
• The practice of modeling change between two time points is causally justified only in specific, limited circumstances.