Measurement Stability, Effect Size Attenuation, and the Replication Crisis

The replication crisis has produced a number of suggested reasons and remedies. Recent models have suggested effect sizes may be overestimated when the variables contain error variance (i.e., under ‘noisy’ conditions). However, clear action implications of this realization for researchers are needed. One useful action step concerns addressing the stability of the measurements in question. Here I provide a perspective on the role of measurement stability in effect size estimates and suggest the use of stability estimates to adjust effect size reporting and sample size planning.