The Bayesian New Statistics: Hypothesis testing, estimation, meta-analysis, and power analysis from a Bayesian...

Area of Science:

• Psychology
• Statistics
• Data Analysis

Background:

• A conceptual divide exists in data analysis between hypothesis testing and estimation with quantified uncertainty.
• The New Statistics movement advocates for a shift from hypothesis testing to estimation in psychology.
• A further distinction lies between frequentist and Bayesian statistical methodologies.

Purpose of the Study:

• To elucidate how Bayesian methods can more effectively achieve the objectives of the New Statistics than frequentist approaches.
• To compare frequentist and Bayesian perspectives on hypothesis testing and estimation.

Main Methods:

• Review of frequentist and Bayesian techniques for hypothesis testing.
• Examination of frequentist confidence intervals and Bayesian credible intervals for estimation.
• Description of Bayesian applications in meta-analysis, randomized controlled trials, and power analysis.

Main Results:

• Bayesian methods provide a more coherent framework for estimation with quantified uncertainty.
• Bayesian approaches offer advantages in interpreting results and integrating prior information.
• The article demonstrates the practical utility of Bayesian methods across various statistical applications.

Conclusions:

• Bayesian methods are better suited to the principles of the New Statistics, emphasizing estimation and uncertainty.
• The adoption of Bayesian methods can enhance the rigor and interpretability of data analysis in psychology and beyond.
• Bayesian approaches offer a unified and powerful framework for modern statistical practice.