Factorial designs

Abstract

Obtain a 2× 3 design by deleting the last level of the first factor, for example. This gives treatment combinations {(00), (01), (02), (10), (11), (12)}. One can also obtain a 2 × 3 design by collapsing the last two levels of the first factor. This gives treatment combinations {(00), (01), (02), (10), (11), (12), (10), (11), (12)}. The second approach 21.73 In a search design the parameters in the model are divided into two sets. One is a set of unknown parameters, all of which are to be estimated. The second is a set of parameters that is known to have at most v nonzero entries. The aim of a search design is to identify and estimate these v parameters. Any design can be used as a search design, but some layouts have better properties than others. See [461, 897].