A Confounding Plans in a 2K Facorial Design

Abstract

When the number of treatments is greater than the available blocks in 2^k factorial experiments confounding becomes necessary to reduce the block size as well as reduce experimental errors. It is also unavoidable when the treatments are greater than the block size. 2^k factorial confounding plans were studied with k = 2, 3, 4 and k > 4. Confounding when k = 2 and 3 is not necessary as the treatment combinations are not much. However, from k > 4, confounding becomes necessary as the treatment combinations are many. From our confounding plans, it is only in 2⁴ that all the main effects are found in one block when confounded with ABCD. The result is not the same with k > 4. Therefore, when many factors are needed is an experiment, the 2⁴ factorial experiment is recommended.