Joint Bayesian inference reveals model properties shared between multiple experimental conditions.

Area of Science:

• Statistics
• Psychophysics
• Data Analysis

Background:

• Statistical models offer interpretable descriptions of experimental data via parameters.
• Interpreting parameter shifts from experimental manipulations requires other model components to remain constant.
• Joint analysis of multiple conditions with invariance constraints can lead to complex models lacking standard procedures.

Purpose of the Study:

• To develop a robust method for joint analysis of multiple experimental conditions in statistical modeling.
• To enable separate analysis of conditions while implicitly accounting for all data.
• To provide a framework for validating the assumption of invariance across conditions.

Main Methods:

• Formulating a solution for joint analysis through repeated application of standard procedures.
• Introducing an additional assumption to facilitate separate condition analysis.
• Validating the supplementary assumption using simulation studies.
• Developing a method to assess the appropriateness of joint treatment.

Main Results:

• The proposed method allows for separate analysis of experimental conditions while implicitly incorporating all data.
• Simulations confirm the validity of the supplementary assumption.
• A natural check for the suitability of joint treatment is presented.
• The method is illustrated using the psychometric function but is broadly applicable.

Conclusions:

• The developed statistical modeling approach simplifies the joint analysis of multiple experimental conditions.
• This method enhances the interpretability of model parameters derived from complex experimental designs.
• The procedure is applicable to various models involving multiple experimental conditions beyond psychophysics.