MIIVefa: An R Package for a New Type of Exploratory Factor Anaylysis Using Model-Implied Instrumental Variables

Area of Science:

• Psychometrics
• Statistical Software
• Data Analysis

Background:

• Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA) are standard methods for identifying latent variables.
• Existing methods may not fully capture complex data structures, such as those with hypothesized correlated errors in longitudinal data.

Purpose of the Study:

• Introduce the R package MIIVefa, which implements the novel MIIV-EFA algorithm.
• Provide a tool for exploring and identifying underlying factor structures in data.
• Offer an alternative to traditional EFA and CFA with unique capabilities.

Main Methods:

• The MIIV-EFA algorithm is implemented within the MIIVefa R package.
• The algorithm identifies the number of factors and item loadings directly from the data.
• It allows for fixed zero loadings and the inclusion of hypothesized correlated errors.

Main Results:

• The MIIVefa package successfully implements the MIIV-EFA algorithm.
• Simulation and empirical examples demonstrate the package's application.
• The algorithm's ability to determine factor structure from data is illustrated.

Conclusions:

• MIIVefa offers a flexible approach to factor analysis, blending aspects of EFA and CFA.
• The package is valuable for researchers dealing with complex data structures, including longitudinal data.
• The study discusses the benefits and limitations of the MIIV-EFA algorithm and the MIIVefa package.