[Observation and experiment: a causal perspective].

Area of Science:

• Scientific methodology
• Causal inference
• Research design

Background:

• Vague definitions of observation and experiment hinder understanding of research design and causal inference validity.
• Existing frameworks lack a unified approach to classifying research designs based on causality.

Purpose of the Study:

• To establish clear relationships among attribute, state, event, and phenomenon using causal thinking.
• To identify and differentiate the causal structures of human observation and manipulated experiments.
• To propose a unified framework for classifying research designs based on causality and measurement processes.

Main Methods:

• Borrowing the concept of phase transition from physics.
• Integrating causal thinking and causal diagrams to define relationships between concepts.
• Analyzing two strategies for classifying research designs: intervention and non-intervention methods.

Main Results:

• Established relationships among attribute, state, event, and phenomenon.
• Identified two distinct causal structures for acquiring phenomena: observation and manipulated experiments.
• Proposed intervention and non-intervention as core strategies for classifying research designs.

Conclusions:

• Observation and experiment should be viewed as unified concepts with a shared connotation.
• Accurate classification of research designs based on causality and measurement is crucial for scientific validity.
• The proposed framework offers a comprehensive and concise approach to understanding research design.