Combining independent p values: extensions of the Stouffer and binomial methods

Area of Science:

• Statistics
• Biostatistics
• Psychometrics

Background:

• Traditional methods for combining p values, such as the Stouffer z method, face challenges.
• These challenges include vulnerability to criticisms of individual studies, difficulty addressing the 'file drawer problem' (publication bias), and vague conclusions.

Purpose of the Study:

• To introduce novel analytical approaches that supplement existing methods for combining p values.
• To reduce or eliminate the identified problems associated with traditional meta-analytic procedures.

Main Methods:

• The study proposes supplementing standard combined probability tests with a variety of new analyses.
• These new analyses are designed to specifically address the limitations of existing methods.

Main Results:

• The proposed analyses offer a robust way to tackle the 'file drawer problem' without restrictive assumptions about unpublished studies.
• These methods enhance the conclusions obtainable from traditional meta-analysis.

Conclusions:

• The novel analytical strategies presented provide a more comprehensive approach to meta-analysis.
• These methods yield richer insights compared to widely used meta-analytic procedures, particularly in addressing publication bias.