Theorized that measurement errors would decrease (ie. attenuate) the correlation between two tests, thus the validity would also be decreased. He developed a correction for attenuation as a method of estimating the actual correlation between two tests (X and Y). The correction for attenuation provides an estimate of the correlation between perfectly reliable measure of X and Y. In practise, the correction almost always overestimates the actual correlation
between X and Y. Also, this formula assumes that researchers can create a test with no measurement error, which is impossible. This formula allows us to estimate the effects of both raising and lowering the reliability of X and Y on the correlation between X and Y. Disadvantages: There are some problems with translating the theory of correcting for attenuation and the practise of it. |