Psychometric evaluation reveals regression to the mean across diverse psychological scales in US cohorts, indicating statistical artifacts can distort perceived changes in attitude.
Key Points
To determine whether extreme regression to the mean, manifested as a crossover effect, occurs across diverse measures of attitudes, disability, metacognition, and emotional intelligence.
Evaluated pre- and post-test scores from two USA samples across four instruments: the Bipolar Adjective Scale (BAS), Emotional Intelligence Scale (EIS), Attitudes Toward Disabled Persons (ATDP) scale, and Metacognitive Awareness Inventory (MAI).
Calculated test-retest correlations and applied statistical adjustments to correct for regression to the mean.
A crossover effect—in which initial high scorers scored lowest on retest and initial low scorers scored highest—appeared in three of the four evaluated scales and in one of two POSHA–S scores.
Statistical correction for regression to the mean diminished or eliminated the crossover effect in several measures due to lower-than-expected test-retest correlations.