[May cross-sectional studies provide causal inferences?]

Area of Science:

• Epidemiology
• Biostatistics
• Causal Inference

Background:

• Cross-sectional studies (CSS) are widely criticized for their inability to establish causal inference.
• Inherited flaws include issues with synchronicity, statistical association, and survivor bias in measured variables.

Purpose of the Study:

• To redefine real and measured cross-sections within a causal inference framework.
• To clarify the analytic role of CSS by addressing limitations in temporal ordering and measurement assumptions.

Main Methods:

• Utilized causal thinking and diagrams to differentiate theoretical and real cross-sections.
• Analyzed the non-synchronic nature of variable measurements in real-world CSS.
• Drew parallels with cumulative case-control and historical cohort studies regarding historical reconstruction.

Main Results:

• Established that real cross-sections, unlike theoretical ones, involve non-synchronic measurements.
• Demonstrated that measured variables act as agents for their historical counterparts.
• Identified a key precondition for causal inference: a causal relationship between measured and historical variables.

Conclusions:

• The temporal order of measurements in CSS is less critical than the causal link to historical events.
• Understanding the analytic role of CSS requires recognizing them as historical reconstructions.
• CSS can contribute to causal inference when preconditions regarding variable representation are met.