Researchers studied the validity of the GRE by determining the relationship between different test sections and psychometric characteristics on the 4 chosen populations and the convergent and discriminant validity of each section. Each of the 3 sections (verbal, quantitative, analytical ability) were further divided based on question type into a total 10 subsections. Maximum likelihood (MLH) estimates were used to find invariant variance-covariance matrices for the relationships between question type and abilities.