Identification of dispersion effects from unreplicated 2n-k fractional factorial designs

In this article, we present a test for dispersion effects from the unreplicated 2n-k regular fractional factorial designs. The proposed procedure for the identification of dispersion effects uses the log-likelihood ratio based on normal errors. Some practical examples are given to illustrate the applicability of the test. It is shown that the proposed method is a useful and economical means for the identification of dispersion effects at the screening stage of experiments. Comparing the power of our method with the two methods published in the literature, we suggest that our test might be more sensitive for identifying the dispersion effects.