Bayesian Multivariate Mixed-Effects Location Scale Modeling of Longitudinal Relations Among Affective Traits, States,...

Area of Science:

• Psychology
• Statistical Modeling

Background:

• Intensive longitudinal studies and experience sampling methods are increasingly used in psychology.
• Existing methods lack the capacity to fully characterize the interplay between personality traits and states.

Purpose of the Study:

• Introduce a novel Bayesian multivariate mixed-effects location scale model (M-MELSM).
• Enable simultaneous modeling of personality traits (location) and states (scale) within multivariate data.
• Facilitate the characterization of relationships between traits and states.

Main Methods:

• Developed a fully Bayesian multivariate mixed-effects location scale model (M-MELSM).
• The model estimates correlations between location and scale random effects across and within outcomes.
• Applied the M-MELSM to daily ratings of affect and physical activity from 194 individuals over 100 days.

Main Results:

• The M-MELSM successfully models multivariate personality data, capturing trait-state interplay.
• Visualizations demonstrated the richness of information provided by the M-MELSM.
• Bayesian hypothesis testing was performed on correlations between random effects.

Conclusions:

• The M-MELSM offers a powerful tool for analyzing complex within-person processes in personality research.
• The Bayesian approach provides essential information for psychological hypothesis testing.
• The model has potential for extensions and applications in personality science.