Virtual Levels and Role Models: N-Level Structural Equations Model of Reciprocal Ratings Data

Area of Science:

• Psychometrics
• Statistical Modeling
• Social Psychology

Background:

• Traditional Multilevel Modeling (MLM) and Structural Equations Modeling (SEM) frameworks struggle with complex dependencies in data.
• Round-robin designs, common in social psychology, present unique relational data structures.
• Existing methods are insufficient for modeling reciprocal dyadic ratings in complex designs.

Purpose of the Study:

• Introduce a general latent variable modeling framework, n-Level Structural Equations Modeling (NL-SEM), for dependent data.
• Extend the capabilities of SEM to accommodate complex multilevel and relational data structures.
• Provide a unified framework for analyzing data with intricate dependencies, such as those found in round-robin designs.

Main Methods:

• Developed the n-Level Structural Equations Modeling (NL-SEM) framework.
• Introduced novel constructs like 'virtual levels' for latent variable specification.
• Utilized the Social Relations Model (SRM) as a case study for NL-SEM.
• Implemented NL-SEM in the xxM software package for empirical application.

Main Results:

• NL-SEM successfully models complex dependencies in round-robin data.
• The framework allows for natural specification of latent variable Social Relations Models (SRMs).
• Empirical application demonstrated that person perceptions are integral to personality.

Conclusions:

• NL-SEM offers a powerful and flexible approach for analyzing complex dependent data structures.
• The framework has significant methodological implications for contextual- and relational-SEMs.
• NL-SEM advances the modeling of intricate relationships in psychological and social science research.