This paper is concerned with designing experiments under the assumption tha
t an analysis strategy is used that considers interactions in addition to m
ain effects. A criterion which averages an approximation to A(s)-efficiency
over lower-dimensional projections of the design is introduced to compare
designs. A columnwise design procedure is used to construct three-level des
igns for six factors in 18 runs. The projection effciencies of these design
s are explored and compared with designs obtained from the L-18 orthogonal
array. Results are also given for designs with 14 and 17 runs.