Memory and Reasoning Abilities Assessed by the Universal Nonverbal Intelligence Test: A Reliable Component Analysis (RCA) Study

Abstract

The Universal Nonverbal Intelligence Test (UNIT) is an important advancement in intelligence assessment, but the equal-weighting method used to compute scores for the memory and reasoning abilities results in some psychometric weaknesses. The equally weighted memory and reasoning scores are highly correlated, leading to a lack of discriminant validity, a potential lack of incremental validity, and poor reliability for their score differences. Reliable component analysis (RCA) was performed on the UNIT normative data to allow for the computation of reliable, uncorrelated memory and reasoning scores. The RCA scores were highly replicable across samples and ages, have good convergent validity, and have greater discriminant validity than the equally weighted scores. Furthermore, the difference between the memory and reasoning scores is much more reliable when using RCA. This results in narrower confidence intervals around difference scores and more powerful tests of the null hypothesis that there is no true difference between memory and reasoning abilities.