Graphs do not lead people to infer causation from correlation

Area of Science:

• Cognitive Psychology
• Science Communication
• Data Visualization

Background:

• Media articles frequently report scientific findings, requiring readers to critically assess evidence and implications.
• Previous research suggests graphs enhance the persuasiveness of scientific data, potentially by evoking a sense of "science."

Purpose of the Study:

• To investigate whether the inclusion of graphs in online news articles leads readers to erroneously infer causation from correlational data.
• To examine the persuasive impact of graphs in scientific communication.

Main Methods:

• Two experiments were conducted using realistic online news articles.
• Participants evaluated research presented in text-only articles or articles accompanied by line or bar graphs.
• Participants were asked to assess the research and apply findings to a hypothetical scenario.

Main Results:

• No evidence was found that the presence of graphs influenced participants' interpretation of correlational data as causal.
• A direct replication attempt of a prior study (Tal & Wansink, 2016) claiming graphs are persuasive was unsuccessful.

Conclusions:

• The mere presence of graphs in media reports does not appear to increase the likelihood of readers incorrectly inferring causation from correlational data.
• Findings challenge the assumption that graphs inherently enhance the persuasive power of scientific information in media contexts.