Hunter and Schmidt 1990 indicated other sources of artifactual error that may have influence on validity. Between these last sources, imperfect construct measurement was suggested by Mount and Barrick (1995) and Ones, Schmidt, and Viswesvaran (1994) as a relevant artifactual error source in the case of personality measures. This artifact understates the real magnitude of validity. Therefore, if the interest of a researcher is at a theoretical level, validity must be corrected for this artifact. From an applied point of view, this correction is not of interest, because in an applied setting we can never use the theoretical validity at the construct level. Hanter and Schmidt (1990, p.455) provided a formula, based on the average correlation between the scales that measure the same construct, to estimate validity at construct level.