Estimating Trait, Method, and Error Variance: Generalizing Across 70 Construct Validation Studies

Abstract

The authors examine the construct validation results of 70 published data sets. The analysis shows that, on average, traits account for less than 50% of the variance in construct measures. These findings raise questions about the application of statistical techniques that assume minimal measurement error or do not properly model systematic measurement error.