Assessing Construct Validity in Math Achievement: An Application of Multilevel Structural Equation Modeling (MSEM) | JoVE Visualize

Assessing Construct Validity in Math Achievement: An Application of Multilevel Structural Equation Modeling (MSEM)

Georgios D Sideridis^1,2, Ioannis Tsaousis³, Abdullah Al-Sadaawi^4,5

¹Harvard Medical School, Boston Children's Hospital, Boston, MA, United States.

Frontiers in Psychology|September 21, 2018

View abstract on PubMed

Summary

Multilevel modeling reveals distinct math achievement structures at individual and university levels. Older institutions show superior math performance, highlighting the importance of accounting for data hierarchy.

Area of Science:

Educational Psychology
Quantitative Psychology
Higher Education Research

Background:

Understanding math achievement is crucial for educational interventions.
Previous studies often overlook hierarchical data structures in educational research.
The factor structure of math competency requires nuanced investigation.

Purpose of the Study:

To determine the optimal factor structure of math competency.
To model math achievement at both person and university levels.
To investigate the impact of institutional age on math achievement.

Main Methods:

Utilized Multilevel Structural Equation Modeling (MSEM) on data from 2,881 students in Saudi Arabia.
Analyzed math achievement using four factors: numbers/operations, algebra/analysis, geometry/measurement, and statistics/probabilities.
Employed Ryu and West (2009) methodology for model evaluation and Gorsuch (1983) for discriminant validation.

Main Results:

Aggregate data analysis supported both unidimensional and 4-factor models; unidimensional was preferred for parsimony.
Multilevel analysis revealed a unidimensional structure at the university level and a 4-factor structure at the person level.
Older universities demonstrated superior math achievement compared to newer institutions.

Conclusions:

Ignoring multilevel data structures can lead to inaccurate conclusions about math competency.
The optimal factor structure for math achievement differs between individual and institutional levels.
Institutional age is a significant factor influencing student math performance in higher education.