Beyond reporting statistical significance: Identifying informative effect sizes to improve scientific communication

Area of Science:

• Psychology
• Scientific Communication
• Statistics

Background:

• Transparent research communication is vital for scientific understanding.
• Current reporting practices (p-values, Bayes Factors) may not optimally inform the public.
• Effect sizes offer a potentially more accessible way to communicate findings.

Purpose of the Study:

• To compare the subjective informativeness of various effect sizes and statistical measures.
• To determine which effect size is perceived as most informative by a general audience.
• To provide guidance on selecting effect sizes for public scientific communication.

Main Methods:

• Three studies were conducted with 652 participants.
• Participants rated the informativeness of five effect sizes, Bayes Factor, and significance statements.
• Subjective informativeness ratings were analyzed and compared.

Main Results:

• Cohen's U3 was rated as the most informative effect size across studies.
• A significant majority found Cohen's U3 more informative than Cohen's d.
• Educational background did not influence the perception of informativeness.

Conclusions:

• Cohen's U3 is recommended for general scientific communication to the public.
• The choice of effect size should align with the specific research finding being highlighted (e.g., differences vs. similarities).
• Emphasizing effect sizes can enhance public understanding of research.