Assessing Comprehensive Spatial Ability and Specific Attributes Through Higher-Order LLM.

Area of Science:

• Cognitive Psychology
• Educational Measurement
• Human Intelligence

Background:

• Spatial reasoning is crucial for academic success, especially in STEM fields.
• Current assessments often fail to capture general spatial ability, focusing on limited attributes.
• The Cattell-Horn-Carroll (CHC) theory posits spatial reasoning as a broad ability with specific components.

Purpose of the Study:

• To develop and evaluate a mixed spatial test assessing both general spatial ability and specific attributes (mental rotation, object assembly, isometric perception).
• To investigate the relationship between general spatial ability and mastery of specific attributes using higher-order linear logistic models (HO-LLMs).
• To compare different spatial ability classification frameworks within the HO-LLM structure.

Main Methods:

• Development of a novel mixed spatial test incorporating mental rotation, object assembly, and isometric perception subtests.
• Application of higher-order linear logistic models (HO-LLMs) to simultaneously estimate general spatial ability and sub-attributes.
• Comparison of four spatial ability classification frameworks to construct Q-matrices for HO-LLM analysis.

Main Results:

• HO-LLMs demonstrated improved model fit compared to traditional approaches.
• Distinct patterns of attribute mastery were identified, revealing key contributors to general spatial ability.
• The study successfully highlighted specific spatial attributes' influence on overall spatial reasoning.

Conclusions:

• Higher-order linear logistic models offer a more profound and interpretable assessment of spatial ability.
• This approach facilitates tailored educational training by pinpointing individual strengths and weaknesses in spatial reasoning.
• The findings support the use of comprehensive spatial assessments for better prediction of STEM academic outcomes.