Resolving Dimensionality in a Child Assessment Tool: An Application of the Multilevel Bifactor Model

Area of Science:

• Psychometrics
• Educational Measurement
• Developmental Psychology

Background:

• Assessment data often exhibits multidimensionality and hierarchical structures.
• Ignoring these features can compromise the validity of factor analysis results.
• Previous analyses of the Child Observation Record Advantage 1.5 (COR-Adv1.5) showed correlated factors and neglected child nesting within classrooms.

Purpose of the Study:

• To describe and demonstrate the multilevel bifactor model for assessing test dimensionality.
• To address the challenges posed by multidimensionality and hierarchical data in assessment.
• To improve the interpretability of factor scores from complex assessment instruments.

Main Methods:

• Application of the multilevel bifactor model.
• Utilizing the Child Observation Record Advantage 1.5 (COR-Adv1.5) as a case study.
• Analysis of nested data structures (children within classrooms).

Main Results:

• The multilevel bifactor model effectively handles multidimensionality and hierarchical data.
• Demonstrated the model's utility in judging test instrument dimensionality.
• Provided model-based statistics to enhance factor score interpretability.

Conclusions:

• The multilevel bifactor model offers a flexible approach to analyzing complex assessment data.
• This method enhances the validity of factor analysis for test dimensionality.
• Findings inform the appropriate use and interpretation of the COR-Adv1.5 assessment tool.