Reference
Metsämuuronen, Jari. (2022). Attenuation-Corrected Estimators of Reliability. Applied Psychological Measurement, 46(8), 720-737. https://doi.org/10.1177/01466216221108131
Review
Metsämuuronen (2022) explored the conventional issues of reliability estimates in the field of psychological measurement. The primary contention is that widely accepted estimators often present attenuated or deflated results due to the item-score correlation being influenced by numerous sources of mechanical error. The author brings to light that, in specific datasets, traditional alpha estimates can be diminished by significant factors, often ranging from 0.40 to 0.60 units of reliability.
The crux of the article rests on introducing a fresh estimator of correlation, named the attenuation-corrected correlation (RAC). Metsämuuronen posits that the RAC, defined as the observed correlation's proportion with the maximal attainable correlation for a particular item and score, can yield more accurate reliability results. By integrating RAC in place of the ρgX in the established formulas, a series of newly formulated deflation-corrected estimators emerge, including the attenuation-corrected versions of alpha, theta, omega, and maximal reliability.
While Metsämuuronen's contribution offers a promising avenue for refining reliability estimates, it also invites further empirical validation. The empirical examples provided are persuasive, but the scope of its application to diverse datasets remains a point of exploration. Nonetheless, this study is commendable for challenging the status quo and initiating a discussion on enhancing the precision of reliability measures in psychological assessments.
