A Multivariate Probit Model for Learning Trajectories: A Fine-Grained Evaluation of an Educational Intervention

Area of Science:

• Educational Psychology
• Psychometrics
• Learning Sciences

Background:

• Educational technology generates extensive student performance data.
• Longitudinal data analysis is crucial for real-time skill mastery diagnosis.
• Cognitive Diagnosis Models (CDMs) provide a framework for skill assessment.

Purpose of the Study:

• To develop a multivariate latent growth curve model for longitudinal CDMs.
• To address limitations of existing models in high-dimensional skill spaces and modest sample sizes.
• To create a flexible framework for investigating factors influencing student skill acquisition.

Main Methods:

• Proposed a multivariate latent growth curve model for student learning trajectories.
• Developed a lower-dimensional approximation suitable for typical educational studies.
• Integrated covariates to examine factors related to skill acquisition.

Main Results:

• The proposed model effectively describes student learning trajectories over time.
• The method is applicable to educational studies with modest sample sizes.
• Demonstrated the framework's utility in assessing educational intervention effects.

Conclusions:

• The longitudinal CDM framework offers a powerful tool for diagnosing student skill mastery.
• The model facilitates the identification of factors that promote or hinder learning.
• This approach enhances the assessment of educational interventions and research questions.