A factored regression approach to modeling latent variable interactions and nonlinear effects.

Area of Science:

• Behavioral Sciences
• Psychology
• Quantitative Psychology

Background:

• Interaction effects are crucial for understanding human behavior in psychology.
• Existing methods for estimating latent variable interactions can be complex and limited.

Purpose of the Study:

• Introduce a flexible factored regression framework for estimating latent variable interactions and nonlinear effects.
• Provide a user-friendly approach for modeling complex data structures, diverse data types, and missing data.
• Offer graphical diagnostics for effective interaction probing.

Main Methods:

• Developed a factored regression framework.
• Conducted Monte Carlo simulations to compare with latent moderated structural equations and product indicator methods.
• Implemented the framework in Blimp software with practical examples.

Main Results:

• Factored regression performs comparably to, or better than, traditional maximum likelihood methods.
• The framework effectively handles complex data structures, diverse data types, and missing data.
• Graphical diagnostics aid in probing interactions.

Conclusions:

• Factored regression offers a flexible and effective approach for estimating latent interactions in behavioral research.
• The Blimp software implementation makes this advanced technique more accessible.
• This framework enhances the analysis of complex relationships in psychological data.