A generalized item response tree model for psychological assessments

Area of Science:

• Psychometrics
• Statistical Modeling
• Psychological Assessment

Background:

• Item response theory (IRT) models are crucial for analyzing assessment data.
• Existing IRT models may not fully capture complex response structures, such as those found in psychological assessments.
• There is a need for flexible IRT models to accommodate various data patterns and covariates.

Purpose of the Study:

• To introduce a generalized item response tree model with enhanced flexibility.
• To demonstrate the model's utility in psychological assessment applications.
• To provide an accessible estimation method using the R package flirt.

Main Methods:

• Development of a generalized item response tree model with flexible parametric forms, dimensionality, and covariates.
• Application of the model to Likert scale item response data.
• Application of the model to omitted item response data.
• Estimation using the freely available R package flirt.

Main Results:

• The generalized item response tree model successfully accommodates complex response structures.
• The model provides valuable insights into Likert scale responses and omitted item behaviors in psychological assessments.
• The R package flirt facilitates practical implementation and estimation of the proposed model.

Conclusions:

• The generalized item response tree model offers a powerful and flexible tool for psychometric analysis.
• This approach enhances the understanding of response processes in psychological testing.
• The availability of the flirt R package promotes wider adoption and application of this advanced IRT methodology.