Causal Interpretations of Correlational Evidence Regarding Violence

Area of Science:

• Criminology
• Psychology
• Sociology

Background:

• Inferring causation from correlation can lead to flawed explanations of violent behavior.
• Ineffective or harmful interventions and policies may result from erroneous causal inferences.
• Understanding how researchers interpret evidence is crucial for advancing violence research.

Purpose of the Study:

• To explore the inferences violence researchers draw from correlational evidence.
• To assess whether researchers acknowledge limitations of non-experimental designs when inferring causation.
• To identify common errors in causal inference among authors of violence research.

Main Methods:

• Authors of violence research articles were surveyed about their causal inferences.
• Studies cited by participants were analyzed for research design (e.g., randomized experiments, non-experiments).
• Inferences drawn by researchers were coded for acknowledgment of alternative interpretations and causal claims.

Main Results:

• Most participants rarely claimed causal inference from non-experimental studies.
• A significant majority failed to acknowledge plausible alternative interpretations (e.g., reverse causality, third variables).
• Most participants incorrectly inferred causation from non-experimental study results.

Conclusions:

• Even experienced researchers may incorrectly infer causation from correlation in violence studies.
• There is a need for improved training in causal inference and research methodology among violence researchers.
• Addressing these inferential errors is critical for developing effective violence prevention strategies.