EGAnet: Framework for Dimensionality Estimation in Psychological Data

Exploratory Graph Analysis (EGA) implements a framework for dimensionality and psychometric assessment using network estimation and community detection. The method, developed by Hudson Golino, includes bootstrap stability assessment, configural and metric invariance testing, and analysis for hierarchical and time-series data. It provides network loadings and scores analogous to factor analysis within a network psychometrics paradigm.

Use Cases

• Estimating the number of latent dimensions in psychological survey data based on network estimation methods.
• Assessing the stability and invariance of psychological constructs using the provided bootstrap and permutation approaches.
• Analyzing hierarchical or time-series data structures in psychological assessments using Hierarchical EGA and Dynamic EGA.
• Evaluating item redundancy and local dependence among variables using Unique Variable Analysis.
• Computing network scores for individuals or groups based on network loadings, similar to factor scores.

Strengths

• Implements a full methodological framework for Exploratory Graph Analysis (EGA) as described in the source paper.
• Includes multiple analysis extensions: bootstrap stability, invariance testing, hierarchical, and dynamic (time-series) EGA.
• Provides fit evaluation using the Entropy Fit family of indices.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and dataset scale are unknown, which may limit suitability assessment.

Provenance

Source

Hudson Golino

Collection Method

Likely contains software implementations and example datasets for the EGA methodology.