Approximated adjusted fractional Bayes factors: A general method for testing informative hypotheses

Area of Science:

• Psychological Sciences
• Bayesian Statistics
• Statistical Modeling

Background:

• Informative hypotheses are crucial for capturing researcher theories in psychology.
• Bayesian analysis commonly employs default Bayes factors, like the fractional Bayes factor, for hypothesis evaluation.
• Existing methods for fractional Bayes factors require careful parameter specification.

Purpose of the Study:

• To approximate and adjust the fractional Bayes factor for evaluating informative hypotheses in general statistical models.
• To provide guidance on selecting the fraction parameter in the fractional Bayes factor.
• To introduce a software package simplifying the application of the adjusted fractional Bayes factor.

Main Methods:

• Approximation and adjustment of the fractional Bayes factor.
• Discussion of fraction parameter choices and a proposed setting scheme.
• Development of a software package for computing the approximated adjusted fractional Bayes factor.

Main Results:

• The paper presents an adjusted fractional Bayes factor applicable to general statistical models.
• A scheme for setting the fraction parameter is discussed, offering practical guidance.
• A software package is available to facilitate the computation and application of this method.

Conclusions:

• The approximated adjusted fractional Bayes factor enables straightforward evaluation of informative hypotheses in psychological research.
• The developed software simplifies complex Bayesian analyses for researchers.
• This approach enhances the practical utility of Bayesian methods in psychological sciences.