IGNORING PSYCHOMETRIC PROBLEMS IN THE STUDY OF GROUP DIFFERENCES IN COGNITIVE TEST PERFORMANCE.

Area of Science:

• Psychometrics
• Cognitive Psychology
• Differential Psychology

Background:

• Jensen's method of correlated vectors is a statistical technique used to analyze group differences in psychometric testing.
• Raven's Progressive Matrices (RPM) is a non-verbal test often used to assess fluid intelligence.
• Previous research has explored ethnic group differences in cognitive test performance.

Purpose of the Study:

• To comment on the methodological appropriateness of Jensen's method as applied by te Nijenhuis et al. (2017).
• To highlight psychometric limitations of the method when applied to dichotomous test items and group differences.
• To question the validity of conclusions drawn regarding general intelligence (g) and ethnic group differences.

Main Methods:

• Critique of Jensen's method of correlated vectors.
• Analysis of psychometric properties for dichotomous items.
• Discussion of item statistics (e.g., item-total correlation) and their behavior across ability groups.
• Examination of linear modeling assumptions for non-linear relationships.

Main Results:

• Jensen's method is psychometrically inappropriate for studying group differences on dichotomous items.
• The method's reliance on item statistics that vary across ability groups can lead to misleading results.
• Wicherts (2017) demonstrated that the method can produce inflated correlations unrelated to general intelligence (g).

Conclusions:

• The psychometric issues with Jensen's method undermine the conclusions of te Nijenhuis et al. (2017).
• Serious doubts are cast on the role of general intelligence (g) in the observed ethnic group differences on the RPM test.
• The application of inappropriate statistical methods can lead to erroneous interpretations of cognitive test data.