Two Approaches for Correcting Correlation Attenuation Caused by Measurement Error: Implications for Research Practice

Abstract

This study focused on the issue of measurement reliability and its attenuation on correlation between two composites and two seemingly different approaches for correcting the attenuation. As expected, correlation coefficients uncorrected for measurement error are systematically biased downward. For the data conditions examined, the two correction approaches provided not only near identical and unbiased estimates but also near identical confidence intervals for the sampling distribution of the corrected correlation coefficients. The highly comparable results from the two approaches suggest that these two approacheswork equally well for these data. It is pointed out that the confirmatory factor analysis (CFA) modeling approach may be less applicable because of more difficult data conditions at the item level in research practice. The findings point to the importance of reporting measurement reliability information in substantive research. The findings further suggest that, in research practice, correction for attenuation should be considered when information about score reliability is available.